Xen project Mailing List

[Xen-changelog] Merge

From: Xen patchbot -unstable <patchbot-unstable@xxxxxxxxxxxxxxxxxxx>

Date: Sat, 14 Jan 2006 01:28:10 +0000

Delivery-date: Sat, 14 Jan 2006 01:36:05 +0000

List-id: BK change log <xen-changelog.lists.xensource.com>

# HG changeset patch # User djm@xxxxxxxxxxxxxxx # Node ID f5ab28885c7569a7a25ef6b4321222bce8ee0131 # Parent 2884f14895d29b679ba9bc8747eccbe345e9e573 # Parent 3f702887d4a62931182392c07436dcdba93a77ea Merge diff -r 2884f14895d2 -r f5ab28885c75 docs/src/user.tex --- a/docs/src/user.tex Fri Jan 13 16:38:44 2006 +++ b/docs/src/user.tex Fri Jan 13 20:12:24 2006 @@ -1885,6 +1885,220 @@ \appendix +\chapter{Unmodified (VMX) guest domains in Xen with Intel\textregistered Virtualization Technology (VT)} + +Xen supports guest domains running unmodified Guest operating systems using Virtualization Technology (VT) available on recent Intel Processors. More information about the Intel Virtualization Technology implementing Virtual Machine Extensions (VMX) in the processor is available on the Intel website at \\ + {\small {\tt http://www.intel.com/technology/computing/vptech}} + +\section{Building Xen with VT support} + +The following packages need to be installed in order to build Xen with VT support. Some Linux distributions do not provide these packages by default. + +\begin{tabular}{lp{11.0cm}} +{\bfseries Package} & {\bfseries Description} \\ + +dev86 & The dev86 package provides an assembler and linker for real mode 80x86 instructions. You need to have this package installed in order to build the BIOS code which runs in (virtual) real mode. + +If the dev86 package is not available on the x86\_64 distribution, you can install the i386 version of it. The dev86 rpm package for various distributions can be found at {\scriptsize {\tt http://www.rpmfind.net/linux/rpm2html/search.php?query=dev86\&submit=Search}} \\ + +LibVNCServer & The unmodified guest's VGA display, keyboard, and mouse are virtualized using the vncserver library provided by this package. You can get the sources of libvncserver from {\small {\tt http://sourceforge.net/projects/libvncserver}}. Build and install the sources on the build system to get the libvncserver library. The 0.8pre version of libvncserver is currently working well with Xen.\\ + +SDL-devel, SDL & Simple DirectMedia Layer (SDL) is another way of virtualizing the unmodified guest console. It provides an X window for the guest console. + +If the SDL and SDL-devel packages are not installed by default on the build system, they can be obtained from {\scriptsize {\tt http://www.rpmfind.net/linux/rpm2html/search.php?query=SDL\&submit=Search}} +, {\scriptsize {\tt http://www.rpmfind.net/linux/rpm2html/search.php?query=SDL-devel\&submit=Search}} \\ + +\end{tabular} + +\section{Configuration file for unmodified VMX guests} + +The Xen installation includes a sample configuration file, {\small {\tt /etc/xen/xmexample.vmx}}. There are comments describing all the options. In addition to the common options that are the same as those for paravirtualized guest configurations, VMX guest configurations have the following settings: + +\begin{tabular}{lp{11.0cm}} + +{\bfseries Parameter} & {\bfseries Description} \\ + +kernel & The VMX firmware loader, {\small {\tt /usr/lib/xen/boot/vmxloader}}\\ + +builder & The domain build function. The VMX domain uses the vmx builder.\\ + +acpi & Enable VMX guest ACPI, default=0 (disabled)\\ + +apic & Enable VMX guest APIC, default=0 (disabled)\\ + +vif & Optionally defines MAC address and/or bridge for the network interfaces. Random MACs are assigned if not given. {\small {\tt type=ioemu}} means ioemu is used to virtualize the VMX NIC. If no type is specified, vbd is used, as with paravirtualized guests.\\ + +disk & Defines the disk devices you want the domain to have access to, and what you want them accessible as. If using a physical device as the VMX guest's disk, each disk entry is of the form + +{\small {\tt phy:UNAME,ioemu:DEV,MODE,}} + +where UNAME is the device, DEV is the device name the domain will see, and MODE is r for read-only, w for read-write. ioemu means the disk will use ioemu to virtualize the VMX disk. If not adding ioemu, it uses vbd like paravirtualized guests. + +If using disk image file, its form should be like + +{\small {\tt file:FILEPATH,ioemu:DEV,MODE}} + +If using more than one disk, there should be a comma between each disk entry. For example: + +{\scriptsize {\tt disk = ['file:/var/images/image1.img,ioemu:hda,w', 'file:/var/images/image2.img,ioemu:hdb,w']}}\\ + +cdrom & Disk image for CD-ROM. The default is {\small {\tt /dev/cdrom}} for Domain0. Inside the VMX domain, the CD-ROM will available as device {\small {\tt /dev/hdc}}. The entry can also point to an ISO file.\\ + +boot & Boot from floppy (a), hard disk (c) or CD-ROM (d). For example, to boot from CD-ROM, the entry should be: + +boot='d'\\ + +device\_model & The device emulation tool for VMX guests. This parameter should not be changed.\\ + +sdl & Enable SDL library for graphics, default = 0 (disabled)\\ + +vnc & Enable VNC library for graphics, default = 1 (enabled)\\ + +vncviewer & Enable spawning of the vncviewer (only valid when vnc=1), default = 1 (enabled) + +If vnc=1 and vncviewer=0, user can use vncviewer to manually connect VMX from remote. For example: + +{\small {\tt vncviewer domain0\_IP\_address:VMX\_domain\_id}} \\ + +ne2000 & Enable ne2000, default = 0 (disabled; use pcnet)\\ + +serial & Enable redirection of VMX serial output to pty device\\ + +localtime & Set the real time clock to local time [default=0, that is, set to UTC].\\ + +enable-audio & Enable audio support. This is under development.\\ + +full-screen & Start in full screen. This is under development.\\ + +nographic & Another way to redirect serial output. If enabled, no 'sdl' or 'vnc' can work. Not recommended.\\ + +\end{tabular} + + +\section{Creating virtual disks from scratch} +\subsection{Using physical disks} +If you are using a physical disk or physical disk partition, you need to install a Linux OS on the disk first. Then the boot loader should be installed in the correct place. For example {\small {\tt dev/sda}} for booting from the whole disk, or {\small {\tt /dev/sda1}} for booting from partition 1. + +\subsection{Using disk image files} +You need to create a large empty disk image file first; then, you need to install a Linux OS onto it. There are two methods you can choose. One is directly installing it using a VMX guest while booting from the OS installation CD-ROM. The other is copying an installed OS into it. The boot loader will also need to be installed. + +\subsubsection*{To create the image file:} +The image size should be big enough to accommodate the entire OS. This example assumes the size is 1G (which is probably too small for most OSes). + +{\small {\tt \# dd if=/dev/zero of=hd.img bs=1M count=1 seek=1023}} + +\subsubsection*{To directly install Linux OS into an image file using a VMX guest:} + +Install Xen and create VMX with the original image file with booting from CD-ROM. Then it is just like a normal Linux OS installation. The VMX configuration file should have these two entries before creating: + +{\small {\tt cdrom='/dev/cdrom' +boot='d'}} + +If this method does not succeed, you can choose the following method of copying an installed Linux OS into an image file. + +\subsubsection*{To copy a installed OS into an image file:} +Directly installing is an easier way to make partitions and install an OS in a disk image file. But if you want to create a specific OS in your disk image, then you will most likely want to use this method. + +\begin{enumerate} +\item {\bfseries Install a normal Linux OS on the host machine}\\ +You can choose any way to install Linux, such as using yum to install Red Hat Linux or YAST to install Novell SuSE Linux. The rest of this example assumes the Linux OS is installed in {\small {\tt /var/guestos/}}. + +\item {\bfseries Make the partition table}\\ +The image file will be treated as hard disk, so you should make the partition table in the image file. For example: + +{\scriptsize {\tt \# losetup /dev/loop0 hd.img\\ +\# fdisk -b 512 -C 4096 -H 16 -S 32 /dev/loop0\\ +press 'n' to add new partition\\ +press 'p' to choose primary partition\\ +press '1' to set partition number\\ +press "Enter" keys to choose default value of "First Cylinder" parameter.\\ +press "Enter" keys to choose default value of "Last Cylinder" parameter.\\ +press 'w' to write partition table and exit\\ +\# losetup -d /dev/loop0}} + +\item {\bfseries Make the file system and install grub}\\ +{\scriptsize {\tt \# ln -s /dev/loop0 /dev/loop\\ +\# losetup /dev/loop0 hd.img\\ +\# losetup -o 16384 /dev/loop1 hd.img\\ +\# mkfs.ext3 /dev/loop1\\ +\# mount /dev/loop1 /mnt\\ +\# mkdir -p /mnt/boot/grub\\ +\# cp /boot/grub/stage* /boot/grub/e2fs\_stage1\_5 /mnt/boot/grub\\ +\# umount /mnt\\ +\# grub\\ +grub> device (hd0) /dev/loop\\ +grub> root (hd0,0)\\ +grub> setup (hd0)\\ +grub> quit\\ +\# rm /dev/loop\\ +\# losetup -d /dev/loop0\\ +\# losetup -d /dev/loop1}} + +The {\small {\tt losetup}} option {\small {\tt -o 16384}} skips the partition table in the image file. It is the number of sectors times 512. We need {\small {\tt /dev/loop}} because grub is expecting a disk device \emph{name}, where \emph{name} represents the entire disk and \emph{name1} represents the first partition. + +\item {\bfseries Copy the OS files to the image}\\ +If you have Xen installed, you can easily use {\small {\tt lomount}} instead of {\small {\tt losetup}} and {\small {\tt mount}} when coping files to some partitions. {\small {\tt lomount}} just needs the partition information. + +{\scriptsize {\tt \# lomount -t ext3 -diskimage hd.img -partition 1 /mnt/guest\\ +\# cp -ax /var/guestos/\{root,dev,var,etc,usr,bin,sbin,lib\} /mnt/guest\\ +\# mkdir /mnt/guest/\{proc,sys,home,tmp\}}} + +\item {\bfseries Edit the {\small {\tt /etc/fstab}} of the guest image}\\ +The fstab should look like this: + +{\scriptsize {\tt \# vim /mnt/guest/etc/fstab\\ +/dev/hda1 / ext3 defaults 1 1\\ +none /dev/pts devpts gid=5,mode=620 0 0\\ +none /dev/shm tmpfs defaults 0 0\\ +none /proc proc defaults 0 0\\ +none /sys sysfs efaults 0 0}} + +\item {\bfseries umount the image file}\\ +{\small {\tt \# umount /mnt/guest}} +\end{enumerate} + +Now, the guest OS image {\small {\tt hd.img}} is ready. You can also reference {\small {\tt http://free.oszoo.org}} for quickstart images. But make sure to install the boot loader. + +\subsection{Install Windows into an Image File using a VMX guest} +In order to install a Windows OS, you should keep {\small {\tt acpi=0}} in your VMX configuration file. + +\section{VMX Guests} +\subsection{Editing the Xen VMX config file} +Make a copy of the example VMX configuration file {\small {\tt /etc/xen/xmeaxmple.vmx}} and edit the line that reads + +{\small {\tt disk = [ 'file:/var/images/\emph{guest.img},ioemu:hda,w' ]}} + +replacing \emph{guest.img} with the name of the guest OS image file you just made. + +\subsection{Creating VMX guests} +Simply follow the usual method of creating the guest, using the -f parameter and providing the filename of your VMX configuration file:\\ + +{\small {\tt \# xend start\\ +\# xm create /etc/xen/vmxguest.vmx}} + +In the default configuration, VNC is on and SDL is off. Therefore VNC windows will open when VMX guests are created. If you want to use SDL to create VMX guests, set {\small {\tt sdl=1}} in your VMX configuration file. You can also turn off VNC by setting {\small {\tt vnc=0}}. + +\subsection{Destroy VMX guests} +VMX guests can be destroyed in the same way as can paravirtualized guests. We recommend that you type the command + +{\small {\tt poweroff}} + +in the VMX guest's console first to prevent data loss. Then execute the command + +{\small {\tt xm destroy \emph{vmx\_guest\_id} }} + +at the Domain0 console. + +\subsection{VMX window (X or VNC) Hot Key} +If you are running in the X environment after creating a VMX guest, an X window is created. There are several hot keys for control of the VMX guest that can be used in the window. + +{\bfseries Ctrl+Alt+2} switches from guest VGA window to the control window. Typing {\small {\tt help }} shows the control commands help. For example, 'q' is the command to destroy the VMX guest.\\ +{\bfseries Ctrl+Alt+1} switches back to VMX guest's VGA.\\ +{\bfseries Ctrl+Alt+3} switches to serial port output. It captures serial output from the VMX guest. It works only if the VMX guest was configured to use the serial port. \\ + +\subsection{Save/Restore and Migration} +VMX guests currently cannot be saved and restored, nor migrated. These features are currently under active development. + %% Chapter Glossary of Terms moved to glossary.tex \chapter{Glossary of Terms} diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/Makefile --- a/linux-2.6-xen-sparse/arch/xen/Makefile Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/Makefile Fri Jan 13 20:12:24 2006 @@ -77,8 +77,6 @@ install -m0664 .config $(INSTALL_PATH)/boot/config-$(XINSTALL_NAME)$(INSTALL_SUFFIX) install -m0664 System.map $(INSTALL_PATH)/boot/System.map-$(XINSTALL_NAME)$(INSTALL_SUFFIX) ln -f -s vmlinuz-$(XINSTALL_NAME)$(INSTALL_SUFFIX) $(INSTALL_PATH)/boot/vmlinuz-$(VERSION).$(PATCHLEVEL)$(XENGUEST)$(INSTALL_SUFFIX) - mkdir -p $(INSTALL_PATH)/usr/include/xen/linux - install -m0644 $(srctree)/include/asm-xen/linux-public/*.h $(INSTALL_PATH)/usr/include/xen/linux archclean: @if [ -e arch/xen/arch ]; then $(MAKE) $(clean)=arch/xen/arch; fi; diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/i386/kernel/acpi/boot.c --- a/linux-2.6-xen-sparse/arch/xen/i386/kernel/acpi/boot.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/i386/kernel/acpi/boot.c Fri Jan 13 20:12:24 2006 @@ -39,8 +39,6 @@ #ifdef CONFIG_XEN #include <asm/fixmap.h> #endif - -void (*pm_power_off)(void) = NULL; #ifdef CONFIG_X86_64 diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/i386/kernel/cpu/mtrr/main.c --- a/linux-2.6-xen-sparse/arch/xen/i386/kernel/cpu/mtrr/main.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/i386/kernel/cpu/mtrr/main.c Fri Jan 13 20:12:24 2006 @@ -67,8 +67,11 @@ op.u.add_memtype.pfn = base; op.u.add_memtype.nr_pfns = size; op.u.add_memtype.type = type; - if ((error = HYPERVISOR_dom0_op(&op))) + error = HYPERVISOR_dom0_op(&op); + if (error) { + BUG_ON(error > 0); return error; + } if (increment) ++usage_table[op.u.add_memtype.reg]; @@ -121,8 +124,12 @@ if (--usage_table[reg] < 1) { op.cmd = DOM0_DEL_MEMTYPE; op.u.del_memtype.handle = 0; - op.u.add_memtype.reg = reg; - (void)HYPERVISOR_dom0_op(&op); + op.u.del_memtype.reg = reg; + error = HYPERVISOR_dom0_op(&op); + if (error) { + BUG_ON(error > 0); + goto out; + } } error = reg; out: diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/i386/kernel/entry.S --- a/linux-2.6-xen-sparse/arch/xen/i386/kernel/entry.S Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/i386/kernel/entry.S Fri Jan 13 20:12:24 2006 @@ -76,7 +76,9 @@ DF_MASK = 0x00000400 NT_MASK = 0x00004000 VM_MASK = 0x00020000 - +/* Pseudo-eflags. */ +NMI_MASK = 0x80000000 + /* Offsets into shared_info_t. */ #define evtchn_upcall_pending /* 0 */ #define evtchn_upcall_mask 1 @@ -305,8 +307,8 @@ je ldt_ss # returning to user-space with LDT SS #endif /* XEN */ restore_nocheck: - testl $VM_MASK, EFLAGS(%esp) - jnz resume_vm86 + testl $(VM_MASK|NMI_MASK), EFLAGS(%esp) + jnz hypervisor_iret movb EVENT_MASK(%esp), %al notb %al # %al == ~saved_mask XEN_GET_VCPU_INFO(%esi) @@ -328,11 +330,11 @@ .long 1b,iret_exc .previous -resume_vm86: - XEN_UNBLOCK_EVENTS(%esi) +hypervisor_iret: + andl $~NMI_MASK, EFLAGS(%esp) RESTORE_REGS movl %eax,(%esp) - movl $__HYPERVISOR_switch_vm86,%eax + movl $__HYPERVISOR_iret,%eax int $0x82 ud2 @@ -691,6 +693,15 @@ call do_debug jmp ret_from_exception +ENTRY(nmi) + pushl %eax + SAVE_ALL + xorl %edx,%edx # zero error code + movl %esp,%eax # pt_regs pointer + call do_nmi + orl $NMI_MASK, EFLAGS(%esp) + jmp restore_all + #if 0 /* XEN */ /* * NMI is doubly nasty. It can happen _while_ we're handling diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/i386/kernel/i386_ksyms.c --- a/linux-2.6-xen-sparse/arch/xen/i386/kernel/i386_ksyms.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/i386/kernel/i386_ksyms.c Fri Jan 13 20:12:24 2006 @@ -76,9 +76,7 @@ EXPORT_SYMBOL(iounmap); EXPORT_SYMBOL(kernel_thread); EXPORT_SYMBOL(pm_idle); -#ifdef CONFIG_ACPI_BOOT EXPORT_SYMBOL(pm_power_off); -#endif EXPORT_SYMBOL(get_cmos_time); EXPORT_SYMBOL(cpu_khz); EXPORT_SYMBOL(apm_info); diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/i386/kernel/io_apic.c --- a/linux-2.6-xen-sparse/arch/xen/i386/kernel/io_apic.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/i386/kernel/io_apic.c Fri Jan 13 20:12:24 2006 @@ -622,9 +622,11 @@ try_to_freeze(PF_FREEZE); if (time_after(jiffies, prev_balance_time+balanced_irq_interval)) { + preempt_disable(); do_irq_balance(); prev_balance_time = jiffies; time_remaining = balanced_irq_interval; + preempt_enable(); } } return 0; diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/i386/kernel/traps.c --- a/linux-2.6-xen-sparse/arch/xen/i386/kernel/traps.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/i386/kernel/traps.c Fri Jan 13 20:12:24 2006 @@ -506,18 +506,11 @@ static void io_check_error(unsigned char reason, struct pt_regs * regs) { - unsigned long i; - printk("NMI: IOCK error (debug interrupt?)\n"); show_registers(regs); /* Re-enable the IOCK line, wait for a few seconds */ - reason = (reason & 0xf) | 8; - outb(reason, 0x61); - i = 2000; - while (--i) udelay(1000); - reason &= ~8; - outb(reason, 0x61); + clear_io_check_error(reason); } static void unknown_nmi_error(unsigned char reason, struct pt_regs * regs) @@ -647,12 +640,6 @@ do_trap(3, SIGTRAP, "int3", 1, regs, error_code, NULL); } #endif - -static inline void conditional_sti(struct pt_regs *regs) -{ - if (regs->eflags & (X86_EFLAGS_IF|VM_MASK)) - local_irq_enable(); -} /* * Our handling of the processor debug registers is non-trivial. @@ -686,9 +673,9 @@ if (notify_die(DIE_DEBUG, "debug", regs, condition, error_code, SIGTRAP) == NOTIFY_STOP) return; - /* It's safe to allow irq's after DR6 has been saved */ - conditional_sti(regs); + if (regs->eflags & X86_EFLAGS_IF) + local_irq_enable(); /* Mask out spurious debug traps due to lazy DR7 setting */ if (condition & (DR_TRAP0|DR_TRAP1|DR_TRAP2|DR_TRAP3)) { diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/i386/mm/init.c --- a/linux-2.6-xen-sparse/arch/xen/i386/mm/init.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/i386/mm/init.c Fri Jan 13 20:12:24 2006 @@ -65,7 +65,7 @@ { pud_t *pud; pmd_t *pmd_table; - + #ifdef CONFIG_X86_PAE pmd_table = (pmd_t *) alloc_bootmem_low_pages(PAGE_SIZE); make_lowmem_page_readonly(pmd_table); diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/kernel/reboot.c --- a/linux-2.6-xen-sparse/arch/xen/kernel/reboot.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/kernel/reboot.c Fri Jan 13 20:12:24 2006 @@ -16,6 +16,13 @@ #include <linux/cpu.h> #include <linux/kthread.h> #include <asm-xen/xencons.h> + +#if defined(__i386__) || defined(__x86_64__) +/* + * Power off function, if any + */ +void (*pm_power_off)(void); +#endif #define SHUTDOWN_INVALID -1 #define SHUTDOWN_POWEROFF 0 @@ -283,15 +290,15 @@ const char **vec, unsigned int len) { char *str; - struct xenbus_transaction *xbt; + xenbus_transaction_t xbt; int err; if (shutting_down != SHUTDOWN_INVALID) return; again: - xbt = xenbus_transaction_start(); - if (IS_ERR(xbt)) + err = xenbus_transaction_start(&xbt); + if (err) return; str = (char *)xenbus_read(xbt, "control", "shutdown", NULL); /* Ignore read errors and empty reads. */ @@ -332,12 +339,12 @@ unsigned int len) { char sysrq_key = '\0'; - struct xenbus_transaction *xbt; + xenbus_transaction_t xbt; int err; again: - xbt = xenbus_transaction_start(); - if (IS_ERR(xbt)) + err = xenbus_transaction_start(&xbt); + if (err) return; if (!xenbus_scanf(xbt, "control", "sysrq", "%c", &sysrq_key)) { printk(KERN_ERR "Unable to read sysrq code in " diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/kernel/smpboot.c --- a/linux-2.6-xen-sparse/arch/xen/kernel/smpboot.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/kernel/smpboot.c Fri Jan 13 20:12:24 2006 @@ -298,7 +298,7 @@ return; sprintf(dir, "cpu/%d", cpu); - err = xenbus_scanf(NULL, dir, "availability", "%s", state); + err = xenbus_scanf(XBT_NULL, dir, "availability", "%s", state); if (err != 1) { printk(KERN_ERR "XENBUS: Unable to read cpu state\n"); return; diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/x86_64/kernel/e820.c --- a/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/e820.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/e820.c Fri Jan 13 20:12:24 2006 @@ -526,7 +526,7 @@ unsigned long __init e820_end_of_ram(void) { - unsigned long max_end_pfn; + unsigned long max_end_pfn; if (xen_override_max_pfn == 0) { max_end_pfn = xen_start_info->nr_pages; @@ -612,7 +612,7 @@ { end_user_pfn = memparse(p, from); end_user_pfn >>= PAGE_SHIFT; - xen_override_max_pfn = (unsigned long) end_user_pfn; + xen_override_max_pfn = (unsigned long) end_user_pfn; } /* diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/x86_64/kernel/entry.S --- a/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/entry.S Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/entry.S Fri Jan 13 20:12:24 2006 @@ -57,7 +57,7 @@ #ifndef CONFIG_PREEMPT #define retint_kernel retint_restore_args #endif - + /* * C code is not supposed to know about undefined top of stack. Every time * a C function with an pt_regs argument is called from the SYSCALL based @@ -65,7 +65,7 @@ * RESTORE_TOP_OF_STACK syncs the syscall state after any possible ptregs * manipulation. */ - + /* %rsp:at FRAMEEND */ .macro FIXUP_TOP_OF_STACK tmp movq $__USER_CS,CS(%rsp) @@ -121,19 +121,19 @@ .endm /* - * Must be consistent with the definition in arch_x86_64.h: - * struct switch_to_user { + * Must be consistent with the definition in arch-x86_64.h: + * struct iret_context { * u64 rax, r11, rcx, flags, rip, cs, rflags, rsp, ss; * }; * #define VGCF_IN_SYSCALL (1<<8) */ - .macro SWITCH_TO_USER flag + .macro HYPERVISOR_IRET flag subq $8*4,%rsp # reuse rip, cs, rflags, rsp, ss in the stack movq %rax,(%rsp) movq %r11,1*8(%rsp) movq %rcx,2*8(%rsp) # we saved %rcx upon exceptions movq $\flag,3*8(%rsp) - movq $__HYPERVISOR_switch_to_user,%rax + movq $__HYPERVISOR_iret,%rax syscall .endm @@ -225,7 +225,7 @@ jnz sysret_careful XEN_UNBLOCK_EVENTS(%rsi) RESTORE_ARGS 0,8,0 - SWITCH_TO_USER VGCF_IN_SYSCALL + HYPERVISOR_IRET VGCF_IN_SYSCALL /* Handle reschedules */ /* edx: work, edi: workmask */ @@ -417,7 +417,6 @@ RESTORE_REST jmp int_ret_from_sys_call CFI_ENDPROC - /* * Interrupt entry/exit. @@ -479,7 +478,7 @@ orb $3,1*8(%rsp) iretq user_mode: - SWITCH_TO_USER 0 + HYPERVISOR_IRET 0 /* edi: workmask, edx: work */ retint_careful: @@ -720,6 +719,18 @@ call evtchn_do_upcall jmp error_exit +#ifdef CONFIG_X86_LOCAL_APIC +ENTRY(nmi) + zeroentry do_nmi_callback +ENTRY(do_nmi_callback) + addq $8, %rsp + call do_nmi + RESTORE_REST + XEN_BLOCK_EVENTS(%rsi) + GET_THREAD_INFO(%rcx) + jmp retint_restore_args +#endif + ALIGN restore_all_enable_events: XEN_UNBLOCK_EVENTS(%rsi) # %rsi is already set up... @@ -734,7 +745,7 @@ orb $3,1*8(%rsp) iretq crit_user_mode: - SWITCH_TO_USER 0 + HYPERVISOR_IRET 0 14: XEN_LOCKED_BLOCK_EVENTS(%rsi) XEN_PUT_VCPU_INFO(%rsi) diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/x86_64/kernel/head64.c --- a/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/head64.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/head64.c Fri Jan 13 20:12:24 2006 @@ -68,13 +68,13 @@ static void __init setup_boot_cpu_data(void) { - int dummy, eax; + unsigned int dummy, eax; /* get vendor info */ - cpuid(0, &boot_cpu_data.cpuid_level, - (int *)&boot_cpu_data.x86_vendor_id[0], - (int *)&boot_cpu_data.x86_vendor_id[8], - (int *)&boot_cpu_data.x86_vendor_id[4]); + cpuid(0, (unsigned int *)&boot_cpu_data.cpuid_level, + (unsigned int *)&boot_cpu_data.x86_vendor_id[0], + (unsigned int *)&boot_cpu_data.x86_vendor_id[8], + (unsigned int *)&boot_cpu_data.x86_vendor_id[4]); /* get cpu type */ cpuid(1, &eax, &dummy, &dummy, @@ -109,7 +109,6 @@ if (s != NULL) setup_early_printk(s); #endif - #ifdef CONFIG_DISCONTIGMEM s = strstr(saved_command_line, "numa="); if (s != NULL) diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/x86_64/kernel/io_apic.c --- a/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/io_apic.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/io_apic.c Fri Jan 13 20:12:24 2006 @@ -255,10 +255,8 @@ return 1; } - __setup("noapic", disable_ioapic_setup); __setup("apic", enable_ioapic_setup); - #include <asm/pci-direct.h> #include <linux/pci_ids.h> @@ -1146,6 +1144,7 @@ v = inb(0x4d1) << 8 | inb(0x4d0); printk(KERN_DEBUG "... PIC ELCR: %04x\n", v); } + #endif /* 0 */ #else @@ -1191,6 +1190,7 @@ * Clear the IO-APIC before rebooting: */ clear_IO_APIC(); + #ifndef CONFIG_XEN disconnect_bsp_APIC(); #endif @@ -1202,6 +1202,7 @@ * * by Matt Domsch <Matt_Domsch@xxxxxxxx> Tue Dec 21 12:25:05 CST 1999 */ + #ifndef CONFIG_XEN static void __init setup_ioapic_ids_from_mpc (void) { diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/x86_64/kernel/irq.c --- a/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/irq.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/irq.c Fri Jan 13 20:12:24 2006 @@ -9,18 +9,15 @@ * x86_64-specific irq controller code. (e.g. i8259.c and * io_apic.c.) */ + +#include <linux/kernel_stat.h> +#include <linux/interrupt.h> +#include <linux/seq_file.h> +#include <linux/module.h> #include <asm/uaccess.h> -#include <linux/module.h> -#include <linux/seq_file.h> -#include <linux/interrupt.h> -#include <linux/kernel_stat.h> - -/* - * Interrupt statistics: - */ +#include <asm/io_apic.h> atomic_t irq_err_count; - #ifdef CONFIG_X86_IO_APIC #ifdef APIC_MISMATCH_DEBUG atomic_t irq_mis_count; diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/x86_64/kernel/ldt.c --- a/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/ldt.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/ldt.c Fri Jan 13 20:12:24 2006 @@ -62,6 +62,7 @@ if (reload) { #ifdef CONFIG_SMP cpumask_t mask; + preempt_disable(); #endif make_pages_readonly(pc->ldt, (pc->size * LDT_ENTRY_SIZE) / @@ -201,6 +202,7 @@ struct user_desc ldt_info; error = -EINVAL; + if (bytecount != sizeof(ldt_info)) goto out; error = -EFAULT; diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/x86_64/kernel/setup.c --- a/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/setup.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/setup.c Fri Jan 13 20:12:24 2006 @@ -62,6 +62,7 @@ #include <asm-xen/xen-public/physdev.h> #include "setup_arch_pre.h" #include <asm/hypervisor.h> +#include <asm-xen/xen-public/nmi.h> #define PFN_UP(x) (((x) + PAGE_SIZE-1) >> PAGE_SHIFT) #define PFN_PHYS(x) ((x) << PAGE_SHIFT) #define end_pfn_map end_pfn @@ -304,7 +305,6 @@ } #endif - static __init void parse_cmdline_early (char ** cmdline_p) { char c = ' ', *to = command_line, *from = COMMAND_LINE; @@ -379,6 +379,7 @@ acpi_skip_timer_override = 1; #endif #endif + #ifndef CONFIG_XEN if (!memcmp(from, "nolapic", 7) || !memcmp(from, "disableapic", 11)) @@ -391,7 +392,8 @@ skip_ioapic_setup = 0; ioapic_force = 1; } -#endif +#endif + if (!memcmp(from, "mem=", 4)) parse_memopt(from+4, &from); @@ -588,7 +590,7 @@ HYPERVISOR_vm_assist(VMASST_CMD_enable, VMASST_TYPE_writable_pagetables); - ARCH_SETUP + ARCH_SETUP #else ROOT_DEV = old_decode_dev(ORIG_ROOT_DEV); drive_info = DRIVE_INFO; @@ -612,7 +614,7 @@ init_mm.end_code = (unsigned long) &_etext; init_mm.end_data = (unsigned long) &_edata; #ifdef CONFIG_XEN - init_mm.brk = start_pfn << PAGE_SHIFT; + init_mm.brk = start_pfn << PAGE_SHIFT; #else init_mm.brk = (unsigned long) &_end; @@ -667,7 +669,6 @@ /* reserve ebda region */ reserve_ebda_region(); #endif - #ifdef CONFIG_SMP /* @@ -790,8 +791,6 @@ } - - if ( ! (xen_start_info->flags & SIF_INITDOMAIN)) { acpi_disabled = 1; @@ -835,7 +834,7 @@ * and also for regions reported as reserved by the e820. */ probe_roms(); - e820_reserve_resources(); + e820_reserve_resources(); #endif request_resource(&iomem_resource, &video_ram_resource); diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/x86_64/kernel/setup64.c --- a/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/setup64.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/setup64.c Fri Jan 13 20:12:24 2006 @@ -33,6 +33,7 @@ #ifdef CONFIG_XEN #include <asm/hypervisor.h> #endif + char x86_boot_params[BOOT_PARAM_SIZE] __initdata = {0,}; cpumask_t cpu_initialized __cpuinitdata = CPU_MASK_NONE; @@ -165,7 +166,6 @@ } #endif - void pda_init(int cpu) { struct x8664_pda *pda = &cpu_pda[cpu]; @@ -175,9 +175,10 @@ #ifndef CONFIG_XEN wrmsrl(MSR_GS_BASE, cpu_pda + cpu); #else - HYPERVISOR_set_segment_base(SEGBASE_GS_KERNEL, - (unsigned long)(cpu_pda + cpu)); -#endif + HYPERVISOR_set_segment_base(SEGBASE_GS_KERNEL, + (unsigned long)(cpu_pda + cpu)); +#endif + pda->me = pda; pda->cpunumber = cpu; pda->irqcount = -1; @@ -201,6 +202,7 @@ } switch_pt(); + pda->irqstackptr += IRQSTACKSIZE-64; } diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/x86_64/kernel/smp.c --- a/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/smp.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/smp.c Fri Jan 13 20:12:24 2006 @@ -30,8 +30,9 @@ #include <asm/apicdef.h> #ifdef CONFIG_XEN #include <asm-xen/evtchn.h> - -#else +#endif + +#ifndef CONFIG_XEN /* * Smarter SMP flushing macros. * c/o Linus Torvalds. diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/x86_64/kernel/traps.c --- a/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/traps.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/traps.c Fri Jan 13 20:12:24 2006 @@ -559,9 +559,11 @@ printk("Uhhuh. NMI received. Dazed and confused, but trying to continue\n"); printk("You probably have a hardware problem with your RAM chips\n"); +#if 0 /* XEN */ /* Clear and disable the memory parity error line. */ reason = (reason & 0xf) | 4; outb(reason, 0x61); +#endif /* XEN */ } static void io_check_error(unsigned char reason, struct pt_regs * regs) @@ -569,12 +571,14 @@ printk("NMI: IOCK error (debug interrupt?)\n"); show_registers(regs); +#if 0 /* XEN */ /* Re-enable the IOCK line, wait for a few seconds */ reason = (reason & 0xf) | 8; outb(reason, 0x61); mdelay(2000); reason &= ~8; outb(reason, 0x61); +#endif /* XEN */ } static void unknown_nmi_error(unsigned char reason, struct pt_regs * regs) @@ -890,7 +894,6 @@ asmlinkage void math_state_restore(void) { struct task_struct *me = current; - /* clts(); */ /* 'clts' is done for us by Xen during virtual trap. */ if (!used_math()) diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/x86_64/kernel/x8664_ksyms.c --- a/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/x8664_ksyms.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/x86_64/kernel/x8664_ksyms.c Fri Jan 13 20:12:24 2006 @@ -59,9 +59,7 @@ EXPORT_SYMBOL(probe_irq_mask); EXPORT_SYMBOL(kernel_thread); EXPORT_SYMBOL(pm_idle); -#ifdef CONFIG_ACPI_BOOT EXPORT_SYMBOL(pm_power_off); -#endif EXPORT_SYMBOL(get_cmos_time); EXPORT_SYMBOL(__down_failed); diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/x86_64/mm/fault.c --- a/linux-2.6-xen-sparse/arch/xen/x86_64/mm/fault.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/x86_64/mm/fault.c Fri Jan 13 20:12:24 2006 @@ -153,7 +153,6 @@ pgd = (pgd_t *)per_cpu(cur_pgd, smp_processor_id()); preempt_enable(); pgd += pgd_index(address); - printk("PGD %lx ", pgd_val(*pgd)); if (bad_address(pgd)) goto bad; if (!pgd_present(*pgd)) goto ret; @@ -258,7 +257,6 @@ pgd = (pgd_t *)per_cpu(cur_pgd, smp_processor_id()); preempt_enable(); pgd += pgd_index(address); - pgd_ref = pgd_offset_k(address); if (pgd_none(*pgd_ref)) return -1; diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/arch/xen/x86_64/mm/init.c --- a/linux-2.6-xen-sparse/arch/xen/x86_64/mm/init.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/arch/xen/x86_64/mm/init.c Fri Jan 13 20:12:24 2006 @@ -40,14 +40,14 @@ #include <asm/proto.h> #include <asm/smp.h> +#ifndef Dprintk +#define Dprintk(x...) +#endif + extern unsigned long *contiguous_bitmap; #if defined(CONFIG_SWIOTLB) extern void swiotlb_init(void); -#endif - -#ifndef Dprintk -#define Dprintk(x...) #endif extern char _stext[]; @@ -200,9 +200,9 @@ static inline pud_t *pud_offset_u(unsigned long address) { - pud_t *pud = level3_user_pgt; - - return pud + pud_index(address); + pud_t *pud = level3_user_pgt; + + return pud + pud_index(address); } static void set_pte_phys(unsigned long vaddr, @@ -215,34 +215,27 @@ Dprintk("set_pte_phys %lx to %lx\n", vaddr, phys); - pgd = (user_mode ? pgd_offset_u(vaddr) : pgd_offset_k(vaddr)); - + pgd = (user_mode ? pgd_offset_u(vaddr) : pgd_offset_k(vaddr)); if (pgd_none(*pgd)) { printk("PGD FIXMAP MISSING, it should be setup in head.S!\n"); return; } - - pud = (user_mode ? pud_offset_u(vaddr) : pud_offset(pgd, vaddr)); - + pud = (user_mode ? pud_offset_u(vaddr) : pud_offset(pgd, vaddr)); if (pud_none(*pud)) { pmd = (pmd_t *) spp_getpage(); - - make_page_readonly(pmd); - xen_pmd_pin(__pa(pmd)); + make_page_readonly(pmd); + xen_pmd_pin(__pa(pmd)); set_pud(pud, __pud(__pa(pmd) | _KERNPG_TABLE | _PAGE_USER)); if (pmd != pmd_offset(pud, 0)) { printk("PAGETABLE BUG #01! %p <-> %p\n", pmd, pmd_offset(pud,0)); return; } } - pmd = pmd_offset(pud, vaddr); - if (pmd_none(*pmd)) { pte = (pte_t *) spp_getpage(); - make_page_readonly(pte); - - xen_pte_pin(__pa(pte)); + make_page_readonly(pte); + xen_pte_pin(__pa(pte)); set_pmd(pmd, __pmd(__pa(pte) | _KERNPG_TABLE | _PAGE_USER)); if (pte != pte_offset_kernel(pmd, 0)) { printk("PAGETABLE BUG #02!\n"); @@ -252,11 +245,10 @@ new_pte = pfn_pte(phys >> PAGE_SHIFT, prot); pte = pte_offset_kernel(pmd, vaddr); - if (!pte_none(*pte) && pte_val(*pte) != (pte_val(new_pte) & __supported_pte_mask)) pte_ERROR(*pte); - set_pte(pte, new_pte); + set_pte(pte, new_pte); /* * It's enough to flush this one mapping. @@ -284,11 +276,11 @@ if (pud_none(*pud)) { pmd = (pmd_t *) spp_getpage(); - make_page_readonly(pmd); - xen_pmd_pin(__pa(pmd)); + make_page_readonly(pmd); + xen_pmd_pin(__pa(pmd)); set_pud(pud, __pud(__pa(pmd) | _KERNPG_TABLE | _PAGE_USER)); - + if (pmd != pmd_offset(pud, 0)) { printk("PAGETABLE BUG #01! %p <-> %p\n", pmd, pmd_offset(pud,0)); return; @@ -298,8 +290,8 @@ if (pmd_none(*pmd)) { pte = (pte_t *) spp_getpage(); - make_page_readonly(pte); - xen_pte_pin(__pa(pte)); + make_page_readonly(pte); + xen_pte_pin(__pa(pte)); set_pmd(pmd, __pmd(__pa(pte) | _KERNPG_TABLE | _PAGE_USER)); if (pte != pte_offset_kernel(pmd, 0)) { @@ -311,12 +303,12 @@ new_pte = pfn_pte_ma(phys >> PAGE_SHIFT, prot); pte = pte_offset_kernel(pmd, vaddr); - /* - * Note that the pte page is already RO, thus we want to use - * xen_l1_entry_update(), not set_pte(). - */ - xen_l1_entry_update(pte, - pfn_pte_ma(phys >> PAGE_SHIFT, prot)); + /* + * Note that the pte page is already RO, thus we want to use + * xen_l1_entry_update(), not set_pte(). + */ + xen_l1_entry_update(pte, + pfn_pte_ma(phys >> PAGE_SHIFT, prot)); /* * It's enough to flush this one mapping. @@ -347,7 +339,6 @@ } } - /* * At this point it only supports vsyscall area. */ @@ -360,18 +351,18 @@ return; } - set_pte_phys(address, phys, prot, SET_FIXMAP_USER); + set_pte_phys(address, phys, prot, SET_FIXMAP_USER); } unsigned long __initdata table_start, tables_space; unsigned long get_machine_pfn(unsigned long addr) { - pud_t* pud = pud_offset_k(addr); - pmd_t* pmd = pmd_offset(pud, addr); - pte_t *pte = pte_offset_kernel(pmd, addr); - - return pte_mfn(*pte); + pud_t* pud = pud_offset_k(addr); + pmd_t* pmd = pmd_offset(pud, addr); + pte_t *pte = pte_offset_kernel(pmd, addr); + + return pte_mfn(*pte); } static __init void *alloc_static_page(unsigned long *phys) @@ -411,12 +402,11 @@ static void __init phys_pud_init(pud_t *pud, unsigned long address, unsigned long end) { - long i, j, k; - unsigned long paddr; + long i, j, k; + unsigned long paddr; i = pud_index(address); pud = pud + i; - for (; i < PTRS_PER_PUD; pud++, i++) { unsigned long pmd_phys; pmd_t *pmd; @@ -429,38 +419,37 @@ } pmd = alloc_static_page(&pmd_phys); - early_make_page_readonly(pmd); - xen_pmd_pin(pmd_phys); + early_make_page_readonly(pmd); + xen_pmd_pin(pmd_phys); set_pud(pud, __pud(pmd_phys | _KERNPG_TABLE)); - for (j = 0; j < PTRS_PER_PMD; pmd++, j++) { - unsigned long pte_phys; - pte_t *pte, *pte_save; + unsigned long pte_phys; + pte_t *pte, *pte_save; if (paddr >= end) { for (; j < PTRS_PER_PMD; j++, pmd++) set_pmd(pmd, __pmd(0)); break; } - pte = alloc_static_page(&pte_phys); - pte_save = pte; - for (k = 0; k < PTRS_PER_PTE; pte++, k++, paddr += PTE_SIZE) { - if ((paddr >= end) || - ((paddr >> PAGE_SHIFT) >= - xen_start_info->nr_pages)) { - __set_pte(pte, __pte(0)); - continue; - } - if (make_readonly(paddr)) { - __set_pte(pte, - __pte(paddr | (_KERNPG_TABLE & ~_PAGE_RW))); - continue; - } - __set_pte(pte, __pte(paddr | _KERNPG_TABLE)); - } - pte = pte_save; - early_make_page_readonly(pte); - xen_pte_pin(pte_phys); + pte = alloc_static_page(&pte_phys); + pte_save = pte; + for (k = 0; k < PTRS_PER_PTE; pte++, k++, paddr += PTE_SIZE) { + if ((paddr >= end) || + ((paddr >> PAGE_SHIFT) >= + xen_start_info->nr_pages)) { + __set_pte(pte, __pte(0)); + continue; + } + if (make_readonly(paddr)) { + __set_pte(pte, + __pte(paddr | (_KERNPG_TABLE & ~_PAGE_RW))); + continue; + } + __set_pte(pte, __pte(paddr | _KERNPG_TABLE)); + } + pte = pte_save; + early_make_page_readonly(pte); + xen_pte_pin(pte_phys); set_pmd(pmd, __pmd(pte_phys | _KERNPG_TABLE)); } } @@ -506,7 +495,7 @@ level3_kernel_pgt[pud_index(__START_KERNEL_map)] = __pud(__pa_symbol(level2_kernel_pgt) | _KERNPG_TABLE | _PAGE_USER); - memcpy((void *)level2_kernel_pgt, page, PAGE_SIZE); + memcpy((void *)level2_kernel_pgt, page, PAGE_SIZE); early_make_page_readonly(init_level4_pgt); early_make_page_readonly(init_level4_user_pgt); @@ -618,7 +607,7 @@ void zap_low_mappings(void) { - /* this is not required for Xen */ + /* this is not required for Xen */ #if 0 swap_low_mappings(); #endif @@ -629,11 +618,11 @@ { { unsigned long zones_size[MAX_NR_ZONES] = {0, 0, 0}; - /* unsigned int max_dma; */ - /* max_dma = virt_to_phys((char *)MAX_DMA_ADDRESS) >> PAGE_SHIFT; */ - /* if (end_pfn < max_dma) */ + /* unsigned int max_dma; */ + /* max_dma = virt_to_phys((char *)MAX_DMA_ADDRESS) >> PAGE_SHIFT; */ + /* if (end_pfn < max_dma) */ zones_size[ZONE_DMA] = end_pfn; -#if 0 +#if 0 else { zones_size[ZONE_DMA] = max_dma; zones_size[ZONE_NORMAL] = end_pfn - max_dma; @@ -642,16 +631,16 @@ free_area_init(zones_size); } - set_fixmap(FIX_SHARED_INFO, xen_start_info->shared_info); - HYPERVISOR_shared_info = (shared_info_t *)fix_to_virt(FIX_SHARED_INFO); - - memset(empty_zero_page, 0, sizeof(empty_zero_page)); + set_fixmap(FIX_SHARED_INFO, xen_start_info->shared_info); + HYPERVISOR_shared_info = (shared_info_t *)fix_to_virt(FIX_SHARED_INFO); + + memset(empty_zero_page, 0, sizeof(empty_zero_page)); init_mm.context.pinned = 1; #ifdef CONFIG_XEN_PHYSDEV_ACCESS { int i; - /* Setup mapping of lower 1st MB */ + /* Setup mapping of lower 1st MB */ for (i = 0; i < NR_FIX_ISAMAPS; i++) if (xen_start_info->flags & SIF_PRIVILEGED) set_fixmap(FIX_ISAMAP_BEGIN - i, i * PAGE_SIZE); @@ -701,7 +690,7 @@ static inline int page_is_ram (unsigned long pagenr) { - return 1; + return 1; } static struct kcore_list kcore_mem, kcore_vmalloc, kcore_kernel, kcore_modules, @@ -790,10 +779,10 @@ void free_initmem(void) { #ifdef __DO_LATER__ - /* - * Some pages can be pinned, but some are not. Unpinning such pages - * triggers BUG(). - */ + /* + * Some pages can be pinned, but some are not. Unpinning such pages + * triggers BUG(). + */ unsigned long addr; addr = (unsigned long)(&__init_begin); @@ -801,12 +790,12 @@ ClearPageReserved(virt_to_page(addr)); set_page_count(virt_to_page(addr), 1); memset((void *)(addr & ~(PAGE_SIZE-1)), 0xcc, PAGE_SIZE); - xen_pte_unpin(__pa(addr)); - make_page_writable(__va(__pa(addr))); - /* - * Make pages from __PAGE_OFFSET address as well - */ - make_page_writable((void *)addr); + xen_pte_unpin(__pa(addr)); + make_page_writable(__va(__pa(addr))); + /* + * Make pages from __PAGE_OFFSET address as well + */ + make_page_writable((void *)addr); free_page(addr); totalram_pages++; } @@ -856,7 +845,7 @@ if (pgd_none(*pgd)) return 0; - pud = pud_offset_k(addr); + pud = pud_offset_k(addr); if (pud_none(*pud)) return 0; diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/balloon/balloon.c --- a/linux-2.6-xen-sparse/drivers/xen/balloon/balloon.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/balloon/balloon.c Fri Jan 13 20:12:24 2006 @@ -354,7 +354,7 @@ unsigned long long new_target; int err; - err = xenbus_scanf(NULL, "memory", "target", "%llu", &new_target); + err = xenbus_scanf(XBT_NULL, "memory", "target", "%llu", &new_target); if (err != 1) { /* This is ok (for domain0 at least) - so just return */ return; diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/blkback/blkback.c --- a/linux-2.6-xen-sparse/drivers/xen/blkback/blkback.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/blkback/blkback.c Fri Jan 13 20:12:24 2006 @@ -540,6 +540,9 @@ pending_vaddrs = kmalloc(sizeof(pending_vaddrs[0]) * mmap_pages, GFP_KERNEL); if (!pending_reqs || !pending_grant_handles || !pending_vaddrs) { + kfree(pending_reqs); + kfree(pending_grant_handles); + kfree(pending_vaddrs); printk("%s: out of memory\n", __FUNCTION__); return -1; } diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/blkback/xenbus.c --- a/linux-2.6-xen-sparse/drivers/xen/blkback/xenbus.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/blkback/xenbus.c Fri Jan 13 20:12:24 2006 @@ -145,7 +145,7 @@ if (err) goto fail; - err = xenbus_switch_state(dev, NULL, XenbusStateInitWait); + err = xenbus_switch_state(dev, XBT_NULL, XenbusStateInitWait); if (err) goto fail; @@ -175,7 +175,7 @@ DPRINTK(""); - err = xenbus_scanf(NULL, dev->nodename, "physical-device", "%x:%x", + err = xenbus_scanf(XBT_NULL, dev->nodename, "physical-device", "%x:%x", &major, &minor); if (XENBUS_EXIST_ERR(err)) { /* Since this watch will fire once immediately after it is @@ -197,7 +197,7 @@ return; } - be->mode = xenbus_read(NULL, dev->nodename, "mode", NULL); + be->mode = xenbus_read(XBT_NULL, dev->nodename, "mode", NULL); if (IS_ERR(be->mode)) { err = PTR_ERR(be->mode); be->mode = NULL; @@ -268,7 +268,7 @@ break; case XenbusStateClosing: - xenbus_switch_state(dev, NULL, XenbusStateClosing); + xenbus_switch_state(dev, XBT_NULL, XenbusStateClosing); break; case XenbusStateClosed: @@ -302,7 +302,7 @@ */ static void connect(struct backend_info *be) { - struct xenbus_transaction *xbt; + xenbus_transaction_t xbt; int err; struct xenbus_device *dev = be->dev; @@ -310,10 +310,9 @@ /* Supply the information about the device the frontend needs */ again: - xbt = xenbus_transaction_start(); - - if (IS_ERR(xbt)) { - err = PTR_ERR(xbt); + err = xenbus_transaction_start(&xbt); + + if (err) { xenbus_dev_fatal(dev, err, "starting transaction"); return; } @@ -366,7 +365,7 @@ DPRINTK("%s", dev->otherend); - err = xenbus_gather(NULL, dev->otherend, "ring-ref", "%lu", &ring_ref, + err = xenbus_gather(XBT_NULL, dev->otherend, "ring-ref", "%lu", &ring_ref, "event-channel", "%u", &evtchn, NULL); if (err) { xenbus_dev_fatal(dev, err, diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/blkfront/blkfront.c --- a/linux-2.6-xen-sparse/drivers/xen/blkfront/blkfront.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/blkfront/blkfront.c Fri Jan 13 20:12:24 2006 @@ -91,7 +91,7 @@ struct blkfront_info *info; /* FIXME: Use dynamic device id if this is not set. */ - err = xenbus_scanf(NULL, dev->nodename, + err = xenbus_scanf(XBT_NULL, dev->nodename, "virtual-device", "%i", &vdevice); if (err != 1) { xenbus_dev_fatal(dev, err, "reading virtual-device"); @@ -161,7 +161,7 @@ struct blkfront_info *info) { const char *message = NULL; - struct xenbus_transaction *xbt; + xenbus_transaction_t xbt; int err; /* Create shared ring, alloc event channel. */ @@ -170,8 +170,8 @@ goto out; again: - xbt = xenbus_transaction_start(); - if (IS_ERR(xbt)) { + err = xenbus_transaction_start(&xbt); + if (err) { xenbus_dev_fatal(dev, err, "starting transaction"); goto destroy_blkring; } @@ -319,7 +319,7 @@ DPRINTK("blkfront.c:connect:%s.\n", info->xbdev->otherend); - err = xenbus_gather(NULL, info->xbdev->otherend, + err = xenbus_gather(XBT_NULL, info->xbdev->otherend, "sectors", "%lu", &sectors, "info", "%u", &binfo, "sector-size", "%lu", &sector_size, @@ -338,7 +338,7 @@ return; } - (void)xenbus_switch_state(info->xbdev, NULL, XenbusStateConnected); + (void)xenbus_switch_state(info->xbdev, XBT_NULL, XenbusStateConnected); /* Kick pending requests. */ spin_lock_irq(&blkif_io_lock); @@ -367,7 +367,7 @@ info->mi = NULL; } - xenbus_switch_state(dev, NULL, XenbusStateClosed); + xenbus_switch_state(dev, XBT_NULL, XenbusStateClosed); } @@ -775,7 +775,7 @@ kfree(copy); - (void)xenbus_switch_state(info->xbdev, NULL, XenbusStateConnected); + (void)xenbus_switch_state(info->xbdev, XBT_NULL, XenbusStateConnected); /* Now safe for us to use the shared ring */ spin_lock_irq(&blkif_io_lock); diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/blktap/blktap.c --- a/linux-2.6-xen-sparse/drivers/xen/blktap/blktap.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/blktap/blktap.c Fri Jan 13 20:12:24 2006 @@ -208,7 +208,7 @@ } struct vm_operations_struct blktap_vm_ops = { - nopage: blktap_nopage, + .nopage = blktap_nopage, }; /****************************************************************** @@ -225,7 +225,7 @@ /* Allocate the fe ring. */ sring = (blkif_sring_t *)get_zeroed_page(GFP_KERNEL); if (sring == NULL) - goto fail_nomem; + return -ENOMEM; SetPageReserved(virt_to_page(sring)); @@ -233,9 +233,6 @@ FRONT_RING_INIT(&blktap_ufe_ring, sring, PAGE_SIZE); return 0; - - fail_nomem: - return -ENOMEM; } static int blktap_release(struct inode *inode, struct file *filp) @@ -391,12 +388,12 @@ } static struct file_operations blktap_fops = { - owner: THIS_MODULE, - poll: blktap_poll, - ioctl: blktap_ioctl, - open: blktap_open, - release: blktap_release, - mmap: blktap_mmap, + .owner = THIS_MODULE, + .poll = blktap_poll, + .ioctl = blktap_ioctl, + .open = blktap_open, + .release = blktap_release, + .mmap = blktap_mmap, }; diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/console/console.c --- a/linux-2.6-xen-sparse/drivers/xen/console/console.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/console/console.c Fri Jan 13 20:12:24 2006 @@ -314,39 +314,31 @@ { int sent, sz, work_done = 0; - if (xen_start_info->flags & SIF_INITDOMAIN) { - if (x_char) { + if (x_char) { + if (xen_start_info->flags & SIF_INITDOMAIN) kcons_write_dom0(NULL, &x_char, 1); - x_char = 0; - work_done = 1; - } - - while (wc != wp) { - sz = wp - wc; - if (sz > (wbuf_size - WBUF_MASK(wc))) - sz = wbuf_size - WBUF_MASK(wc); + else + while (x_char) + if (xencons_ring_send(&x_char, 1) == 1) + break; + x_char = 0; + work_done = 1; + } + + while (wc != wp) { + sz = wp - wc; + if (sz > (wbuf_size - WBUF_MASK(wc))) + sz = wbuf_size - WBUF_MASK(wc); + if (xen_start_info->flags & SIF_INITDOMAIN) { kcons_write_dom0(NULL, &wbuf[WBUF_MASK(wc)], sz); wc += sz; - work_done = 1; - } - } else { - while (x_char) { - if (xencons_ring_send(&x_char, 1) == 1) { - x_char = 0; - work_done = 1; - } - } - - while (wc != wp) { - sz = wp - wc; - if (sz > (wbuf_size - WBUF_MASK(wc))) - sz = wbuf_size - WBUF_MASK(wc); + } else { sent = xencons_ring_send(&wbuf[WBUF_MASK(wc)], sz); if (sent == 0) break; wc += sent; - work_done = 1; } + work_done = 1; } if (work_done && (xencons_tty != NULL)) { diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/net_driver_util.c --- a/linux-2.6-xen-sparse/drivers/xen/net_driver_util.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/net_driver_util.c Fri Jan 13 20:12:24 2006 @@ -38,7 +38,7 @@ char *s; int i; char *e; - char *macstr = xenbus_read(NULL, dev->nodename, "mac", NULL); + char *macstr = xenbus_read(XBT_NULL, dev->nodename, "mac", NULL); if (IS_ERR(macstr)) { return PTR_ERR(macstr); } diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/netback/interface.c --- a/linux-2.6-xen-sparse/drivers/xen/netback/interface.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/netback/interface.c Fri Jan 13 20:12:24 2006 @@ -183,7 +183,7 @@ int netif_map(netif_t *netif, unsigned long tx_ring_ref, unsigned long rx_ring_ref, unsigned int evtchn) { - int err; + int err = -ENOMEM; netif_tx_sring_t *txs; netif_rx_sring_t *rxs; evtchn_op_t op = { @@ -199,25 +199,16 @@ if (netif->tx_comms_area == NULL) return -ENOMEM; netif->rx_comms_area = alloc_vm_area(PAGE_SIZE); - if (netif->rx_comms_area == NULL) { - free_vm_area(netif->tx_comms_area); - return -ENOMEM; - } + if (netif->rx_comms_area == NULL) + goto err_rx; err = map_frontend_pages(netif, tx_ring_ref, rx_ring_ref); - if (err) { - free_vm_area(netif->tx_comms_area); - free_vm_area(netif->rx_comms_area); - return err; - } + if (err) + goto err_map; err = HYPERVISOR_event_channel_op(&op); - if (err) { - unmap_frontend_pages(netif); - free_vm_area(netif->tx_comms_area); - free_vm_area(netif->rx_comms_area); - return err; - } + if (err) + goto err_hypervisor; netif->evtchn = op.u.bind_interdomain.local_port; @@ -245,6 +236,13 @@ rtnl_unlock(); return 0; +err_hypervisor: + unmap_frontend_pages(netif); +err_map: + free_vm_area(netif->rx_comms_area); +err_rx: + free_vm_area(netif->tx_comms_area); + return err; } static void free_netif_callback(void *arg) diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/netback/netback.c --- a/linux-2.6-xen-sparse/drivers/xen/netback/netback.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/netback/netback.c Fri Jan 13 20:12:24 2006 @@ -14,6 +14,7 @@ #include <asm-xen/balloon.h> #include <asm-xen/xen-public/memory.h> +/*#define NETBE_DEBUG_INTERRUPT*/ static void netif_idx_release(u16 pending_idx); static void netif_page_release(struct page *page); @@ -727,6 +728,7 @@ return notify; } +#ifdef NETBE_DEBUG_INTERRUPT static irqreturn_t netif_be_dbg(int irq, void *dev_id, struct pt_regs *regs) { struct list_head *ent; @@ -758,6 +760,7 @@ return IRQ_HANDLED; } +#endif static int __init netback_init(void) { @@ -794,6 +797,7 @@ netif_xenbus_init(); +#ifdef NETBE_DEBUG_INTERRUPT (void)bind_virq_to_irqhandler( VIRQ_DEBUG, 0, @@ -801,6 +805,7 @@ SA_SHIRQ, "net-be-dbg", &netif_be_dbg); +#endif return 0; } diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/netback/xenbus.c --- a/linux-2.6-xen-sparse/drivers/xen/netback/xenbus.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/netback/xenbus.c Fri Jan 13 20:12:24 2006 @@ -93,7 +93,7 @@ if (err) goto fail; - err = xenbus_switch_state(dev, NULL, XenbusStateInitWait); + err = xenbus_switch_state(dev, XBT_NULL, XenbusStateInitWait); if (err) { goto fail; } @@ -122,7 +122,7 @@ DPRINTK("netback_hotplug"); - val = xenbus_read(NULL, xdev->nodename, "script", NULL); + val = xenbus_read(XBT_NULL, xdev->nodename, "script", NULL); if (IS_ERR(val)) { int err = PTR_ERR(val); xenbus_dev_fatal(xdev, err, "reading script"); @@ -160,7 +160,7 @@ DPRINTK(""); - err = xenbus_scanf(NULL, dev->nodename, "handle", "%li", &handle); + err = xenbus_scanf(XBT_NULL, dev->nodename, "handle", "%li", &handle); if (XENBUS_EXIST_ERR(err)) { /* Since this watch will fire once immediately after it is registered, we expect this. Ignore it, and wait for the @@ -212,7 +212,7 @@ break; case XenbusStateClosing: - xenbus_switch_state(dev, NULL, XenbusStateClosing); + xenbus_switch_state(dev, XBT_NULL, XenbusStateClosing); break; case XenbusStateClosed: @@ -256,7 +256,7 @@ return; } - xenbus_switch_state(dev, NULL, XenbusStateConnected); + xenbus_switch_state(dev, XBT_NULL, XenbusStateConnected); } @@ -269,7 +269,7 @@ DPRINTK(""); - err = xenbus_gather(NULL, dev->otherend, + err = xenbus_gather(XBT_NULL, dev->otherend, "tx-ring-ref", "%lu", &tx_ring_ref, "rx-ring-ref", "%lu", &rx_ring_ref, "event-channel", "%u", &evtchn, NULL); diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/netfront/netfront.c --- a/linux-2.6-xen-sparse/drivers/xen/netfront/netfront.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/netfront/netfront.c Fri Jan 13 20:12:24 2006 @@ -212,7 +212,7 @@ struct netfront_info *info; unsigned int handle; - err = xenbus_scanf(NULL, dev->nodename, "handle", "%u", &handle); + err = xenbus_scanf(XBT_NULL, dev->nodename, "handle", "%u", &handle); if (err != 1) { xenbus_dev_fatal(dev, err, "reading handle"); return err; @@ -260,7 +260,7 @@ struct netfront_info *info) { const char *message; - struct xenbus_transaction *xbt; + xenbus_transaction_t xbt; int err; err = xen_net_read_mac(dev, info->mac); @@ -275,8 +275,8 @@ goto out; again: - xbt = xenbus_transaction_start(); - if (IS_ERR(xbt)) { + err = xenbus_transaction_start(&xbt); + if (err) { xenbus_dev_fatal(dev, err, "starting transaction"); goto destroy_ring; } @@ -1199,7 +1199,7 @@ close_netdev(info); - xenbus_switch_state(dev, NULL, XenbusStateClosed); + xenbus_switch_state(dev, XBT_NULL, XenbusStateClosed); } @@ -1217,31 +1217,34 @@ static void close_netdev(struct netfront_info *info) +{ + spin_lock_irq(&info->netdev->xmit_lock); + netif_stop_queue(info->netdev); + spin_unlock_irq(&info->netdev->xmit_lock); + +#ifdef CONFIG_PROC_FS + xennet_proc_delif(info->netdev); +#endif + + del_timer_sync(&info->rx_refill_timer); + + unregister_netdev(info->netdev); +} + + +static void netif_disconnect_backend(struct netfront_info *info) { /* Stop old i/f to prevent errors whilst we rebuild the state. */ spin_lock_irq(&info->tx_lock); spin_lock(&info->rx_lock); - netif_stop_queue(info->netdev); - /* info->backend_state = BEST_DISCONNECTED; */ + info->backend_state = BEST_DISCONNECTED; spin_unlock(&info->rx_lock); spin_unlock_irq(&info->tx_lock); - -#ifdef CONFIG_PROC_FS - xennet_proc_delif(info->netdev); -#endif if (info->irq) unbind_from_irqhandler(info->irq, info->netdev); info->evtchn = info->irq = 0; - del_timer_sync(&info->rx_refill_timer); - - unregister_netdev(info->netdev); -} - - -static void netif_disconnect_backend(struct netfront_info *info) -{ end_access(info->tx_ring_ref, info->tx.sring); end_access(info->rx_ring_ref, info->rx.sring); info->tx_ring_ref = GRANT_INVALID_REF; diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/tpmback/xenbus.c --- a/linux-2.6-xen-sparse/drivers/xen/tpmback/xenbus.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/tpmback/xenbus.c Fri Jan 13 20:12:24 2006 @@ -89,7 +89,7 @@ goto fail; } - err = xenbus_switch_state(dev, NULL, XenbusStateInitWait); + err = xenbus_switch_state(dev, XBT_NULL, XenbusStateInitWait); if (err) { goto fail; } @@ -109,7 +109,7 @@ = container_of(watch, struct backend_info, backend_watch); struct xenbus_device *dev = be->dev; - err = xenbus_scanf(NULL, dev->nodename, + err = xenbus_scanf(XBT_NULL, dev->nodename, "instance","%li", &instance); if (XENBUS_EXIST_ERR(err)) { return; @@ -177,7 +177,7 @@ break; case XenbusStateClosing: - xenbus_switch_state(dev, NULL, XenbusStateClosing); + xenbus_switch_state(dev, XBT_NULL, XenbusStateClosing); break; case XenbusStateClosed: @@ -230,15 +230,14 @@ static void connect(struct backend_info *be) { - struct xenbus_transaction *xbt; + xenbus_transaction_t xbt; int err; struct xenbus_device *dev = be->dev; unsigned long ready = 1; again: - xbt = xenbus_transaction_start(); - if (IS_ERR(xbt)) { - err = PTR_ERR(xbt); + err = xenbus_transaction_start(&xbt); + if (err) { xenbus_dev_fatal(be->dev, err, "starting transaction"); return; } @@ -275,7 +274,7 @@ unsigned int evtchn; int err; - err = xenbus_gather(NULL, dev->otherend, + err = xenbus_gather(XBT_NULL, dev->otherend, "ring-ref", "%lu", &ring_ref, "event-channel", "%u", &evtchn, NULL); if (err) { diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/tpmfront/tpmfront.c --- a/linux-2.6-xen-sparse/drivers/xen/tpmfront/tpmfront.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/tpmfront/tpmfront.c Fri Jan 13 20:12:24 2006 @@ -271,7 +271,7 @@ { const char *message = NULL; int err; - struct xenbus_transaction *xbt; + xenbus_transaction_t xbt; err = setup_tpmring(dev, info); if (err) { @@ -280,8 +280,8 @@ } again: - xbt = xenbus_transaction_start(); - if (IS_ERR(xbt)) { + err = xenbus_transaction_start(&xbt); + if (err) { xenbus_dev_fatal(dev, err, "starting transaction"); goto destroy_tpmring; } diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/xenbus/xenbus_client.c --- a/linux-2.6-xen-sparse/drivers/xen/xenbus/xenbus_client.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/xenbus/xenbus_client.c Fri Jan 13 20:12:24 2006 @@ -87,7 +87,7 @@ int xenbus_switch_state(struct xenbus_device *dev, - struct xenbus_transaction *xbt, + xenbus_transaction_t xbt, XenbusState state) { /* We check whether the state is currently set to the given value, and @@ -152,7 +152,7 @@ goto fail; } - if (xenbus_write(NULL, path_buffer, "error", printf_buffer) != 0) { + if (xenbus_write(XBT_NULL, path_buffer, "error", printf_buffer) != 0) { printk("xenbus: failed to write error node for %s (%s)\n", dev->nodename, printf_buffer); goto fail; @@ -187,7 +187,7 @@ _dev_error(dev, err, fmt, ap); va_end(ap); - xenbus_switch_state(dev, NULL, XenbusStateClosing); + xenbus_switch_state(dev, XBT_NULL, XenbusStateClosing); } EXPORT_SYMBOL(xenbus_dev_fatal); @@ -223,7 +223,7 @@ { XenbusState result; - int err = xenbus_gather(NULL, path, "state", "%d", &result, NULL); + int err = xenbus_gather(XBT_NULL, path, "state", "%d", &result, NULL); if (err) result = XenbusStateClosed; diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/xenbus/xenbus_dev.c --- a/linux-2.6-xen-sparse/drivers/xen/xenbus/xenbus_dev.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/xenbus/xenbus_dev.c Fri Jan 13 20:12:24 2006 @@ -47,7 +47,7 @@ struct xenbus_dev_transaction { struct list_head list; - struct xenbus_transaction *handle; + xenbus_transaction_t handle; }; struct xenbus_dev_data { @@ -109,9 +109,8 @@ size_t len, loff_t *ppos) { struct xenbus_dev_data *u = filp->private_data; - struct xenbus_dev_transaction *trans; + struct xenbus_dev_transaction *trans = NULL; void *reply; - int err = 0; if ((len + u->len) > sizeof(u->u.buffer)) return -EINVAL; @@ -135,42 +134,40 @@ case XS_MKDIR: case XS_RM: case XS_SET_PERMS: + if (u->u.msg.type == XS_TRANSACTION_START) { + trans = kmalloc(sizeof(*trans), GFP_KERNEL); + if (!trans) + return -ENOMEM; + } + reply = xenbus_dev_request_and_reply(&u->u.msg); if (IS_ERR(reply)) { - err = PTR_ERR(reply); - } else { - if (u->u.msg.type == XS_TRANSACTION_START) { - trans = kmalloc(sizeof(*trans), GFP_KERNEL); - trans->handle = (struct xenbus_transaction *) - simple_strtoul(reply, NULL, 0); - list_add(&trans->list, &u->transactions); - } else if (u->u.msg.type == XS_TRANSACTION_END) { - list_for_each_entry(trans, &u->transactions, - list) - if ((unsigned long)trans->handle == - (unsigned long)u->u.msg.tx_id) - break; - BUG_ON(&trans->list == &u->transactions); - list_del(&trans->list); - kfree(trans); - } - queue_reply(u, (char *)&u->u.msg, sizeof(u->u.msg)); - queue_reply(u, (char *)reply, u->u.msg.len); - kfree(reply); + kfree(trans); + return PTR_ERR(reply); } + + if (u->u.msg.type == XS_TRANSACTION_START) { + trans->handle = simple_strtoul(reply, NULL, 0); + list_add(&trans->list, &u->transactions); + } else if (u->u.msg.type == XS_TRANSACTION_END) { + list_for_each_entry(trans, &u->transactions, list) + if (trans->handle == u->u.msg.tx_id) + break; + BUG_ON(&trans->list == &u->transactions); + list_del(&trans->list); + kfree(trans); + } + queue_reply(u, (char *)&u->u.msg, sizeof(u->u.msg)); + queue_reply(u, (char *)reply, u->u.msg.len); + kfree(reply); break; default: - err = -EINVAL; - break; + return -EINVAL; } - if (err == 0) { - u->len = 0; - err = len; - } - - return err; + u->len = 0; + return len; } static int xenbus_dev_open(struct inode *inode, struct file *filp) diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/xenbus/xenbus_probe.c --- a/linux-2.6-xen-sparse/drivers/xen/xenbus/xenbus_probe.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/xenbus/xenbus_probe.c Fri Jan 13 20:12:24 2006 @@ -115,7 +115,7 @@ static int read_otherend_details(struct xenbus_device *xendev, char *id_node, char *path_node) { - int err = xenbus_gather(NULL, xendev->nodename, + int err = xenbus_gather(XBT_NULL, xendev->nodename, id_node, "%i", &xendev->otherend_id, path_node, NULL, &xendev->otherend, NULL); @@ -126,7 +126,7 @@ return err; } if (strlen(xendev->otherend) == 0 || - !xenbus_exists(NULL, xendev->otherend, "")) { + !xenbus_exists(XBT_NULL, xendev->otherend, "")) { xenbus_dev_fatal(xendev, -ENOENT, "missing other end from %s", xendev->nodename); kfree(xendev->otherend); @@ -200,14 +200,14 @@ devid = strrchr(nodename, '/') + 1; - err = xenbus_gather(NULL, nodename, "frontend-id", "%i", &domid, + err = xenbus_gather(XBT_NULL, nodename, "frontend-id", "%i", &domid, "frontend", NULL, &frontend, NULL); if (err) return err; if (strlen(frontend) == 0) err = -ERANGE; - if (!err && !xenbus_exists(NULL, frontend, "")) + if (!err && !xenbus_exists(XBT_NULL, frontend, "")) err = -ENOENT; kfree(frontend); @@ -373,7 +373,7 @@ return 0; fail: xenbus_dev_error(dev, err, "xenbus_dev_probe on %s", dev->nodename); - xenbus_switch_state(dev, NULL, XenbusStateClosed); + xenbus_switch_state(dev, XBT_NULL, XenbusStateClosed); return -ENODEV; } @@ -391,7 +391,7 @@ if (drv->remove) drv->remove(dev); - xenbus_switch_state(dev, NULL, XenbusStateClosed); + xenbus_switch_state(dev, XBT_NULL, XenbusStateClosed); return 0; } @@ -470,12 +470,17 @@ DPRINTK("%s", info->nodename); - if (!strncmp(xendev->nodename, info->nodename, len)) { - info->dev = xendev; - get_device(dev); - return 1; - } - return 0; + /* Match the info->nodename path, or any subdirectory of that path. */ + if (strncmp(xendev->nodename, info->nodename, len)) + return 0; + + /* If the node name is longer, ensure it really is a subdirectory. */ + if ((strlen(xendev->nodename) > len) && (xendev->nodename[len] != '/')) + return 0; + + info->dev = xendev; + get_device(dev); + return 1; } static void xenbus_cleanup_devices(const char *path, struct bus_type *bus) @@ -542,14 +547,6 @@ const char *type, const char *nodename) { -#define CHECK_FAIL \ - do { \ - if (err) \ - goto fail; \ - } \ - while (0) \ - - int err; struct xenbus_device *xendev; size_t stringlen; @@ -584,19 +581,18 @@ xendev->dev.release = xenbus_dev_release; err = bus->get_bus_id(xendev->dev.bus_id, xendev->nodename); - CHECK_FAIL; + if (err) + goto fail; /* Register with generic device framework. */ err = device_register(&xendev->dev); - CHECK_FAIL; + if (err) + goto fail; device_create_file(&xendev->dev, &dev_attr_nodename); device_create_file(&xendev->dev, &dev_attr_devtype); return 0; - -#undef CHECK_FAIL - fail: xenbus_dev_free(xendev); return err; @@ -652,7 +648,7 @@ if (!nodename) return -ENOMEM; - dir = xenbus_directory(NULL, nodename, "", &dir_n); + dir = xenbus_directory(XBT_NULL, nodename, "", &dir_n); if (IS_ERR(dir)) { kfree(nodename); return PTR_ERR(dir); @@ -675,7 +671,7 @@ unsigned int dir_n = 0; int i; - dir = xenbus_directory(NULL, bus->root, type, &dir_n); + dir = xenbus_directory(XBT_NULL, bus->root, type, &dir_n); if (IS_ERR(dir)) return PTR_ERR(dir); @@ -694,7 +690,7 @@ char **dir; unsigned int i, dir_n; - dir = xenbus_directory(NULL, bus->root, "", &dir_n); + dir = xenbus_directory(XBT_NULL, bus->root, "", &dir_n); if (IS_ERR(dir)) return PTR_ERR(dir); @@ -740,7 +736,7 @@ if (char_count(node, '/') < 2) return; - exists = xenbus_exists(NULL, node, ""); + exists = xenbus_exists(XBT_NULL, node, ""); if (!exists) { xenbus_cleanup_devices(node, &bus->bus); return; diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/drivers/xen/xenbus/xenbus_xs.c --- a/linux-2.6-xen-sparse/drivers/xen/xenbus/xenbus_xs.c Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/drivers/xen/xenbus/xenbus_xs.c Fri Jan 13 20:12:24 2006 @@ -190,7 +190,7 @@ } /* Send message to xs, get kmalloc'ed reply. ERR_PTR() on error. */ -static void *xs_talkv(struct xenbus_transaction *t, +static void *xs_talkv(xenbus_transaction_t t, enum xsd_sockmsg_type type, const struct kvec *iovec, unsigned int num_vecs, @@ -201,7 +201,7 @@ unsigned int i; int err; - msg.tx_id = (u32)(unsigned long)t; + msg.tx_id = t; msg.req_id = 0; msg.type = type; msg.len = 0; @@ -242,7 +242,7 @@ } /* Simplified version of xs_talkv: single message. */ -static void *xs_single(struct xenbus_transaction *t, +static void *xs_single(xenbus_transaction_t t, enum xsd_sockmsg_type type, const char *string, unsigned int *len) @@ -309,7 +309,7 @@ return ret; } -char **xenbus_directory(struct xenbus_transaction *t, +char **xenbus_directory(xenbus_transaction_t t, const char *dir, const char *node, unsigned int *num) { char *strings, *path; @@ -329,7 +329,7 @@ EXPORT_SYMBOL(xenbus_directory); /* Check if a path exists. Return 1 if it does. */ -int xenbus_exists(struct xenbus_transaction *t, +int xenbus_exists(xenbus_transaction_t t, const char *dir, const char *node) { char **d; @@ -347,7 +347,7 @@ * Returns a kmalloced value: call free() on it after use. * len indicates length in bytes. */ -void *xenbus_read(struct xenbus_transaction *t, +void *xenbus_read(xenbus_transaction_t t, const char *dir, const char *node, unsigned int *len) { char *path; @@ -366,7 +366,7 @@ /* Write the value of a single file. * Returns -err on failure. */ -int xenbus_write(struct xenbus_transaction *t, +int xenbus_write(xenbus_transaction_t t, const char *dir, const char *node, const char *string) { const char *path; @@ -389,7 +389,7 @@ EXPORT_SYMBOL(xenbus_write); /* Create a new directory. */ -int xenbus_mkdir(struct xenbus_transaction *t, +int xenbus_mkdir(xenbus_transaction_t t, const char *dir, const char *node) { char *path; @@ -406,7 +406,7 @@ EXPORT_SYMBOL(xenbus_mkdir); /* Destroy a file or directory (directories must be empty). */ -int xenbus_rm(struct xenbus_transaction *t, const char *dir, const char *node) +int xenbus_rm(xenbus_transaction_t t, const char *dir, const char *node) { char *path; int ret; @@ -424,30 +424,28 @@ /* Start a transaction: changes by others will not be seen during this * transaction, and changes will not be visible to others until end. */ -struct xenbus_transaction *xenbus_transaction_start(void) +int xenbus_transaction_start(xenbus_transaction_t *t) { char *id_str; - unsigned long id; down_read(&xs_state.suspend_mutex); - id_str = xs_single(NULL, XS_TRANSACTION_START, "", NULL); + id_str = xs_single(XBT_NULL, XS_TRANSACTION_START, "", NULL); if (IS_ERR(id_str)) { up_read(&xs_state.suspend_mutex); - return (struct xenbus_transaction *)id_str; - } - - id = simple_strtoul(id_str, NULL, 0); + return PTR_ERR(id_str); + } + + *t = simple_strtoul(id_str, NULL, 0); kfree(id_str); - - return (struct xenbus_transaction *)id; + return 0; } EXPORT_SYMBOL(xenbus_transaction_start); /* End a transaction. * If abandon is true, transaction is discarded instead of committed. */ -int xenbus_transaction_end(struct xenbus_transaction *t, int abort) +int xenbus_transaction_end(xenbus_transaction_t t, int abort) { char abortstr[2]; int err; @@ -466,7 +464,7 @@ EXPORT_SYMBOL(xenbus_transaction_end); /* Single read and scanf: returns -errno or num scanned. */ -int xenbus_scanf(struct xenbus_transaction *t, +int xenbus_scanf(xenbus_transaction_t t, const char *dir, const char *node, const char *fmt, ...) { va_list ap; @@ -489,7 +487,7 @@ EXPORT_SYMBOL(xenbus_scanf); /* Single printf and write: returns -errno or 0. */ -int xenbus_printf(struct xenbus_transaction *t, +int xenbus_printf(xenbus_transaction_t t, const char *dir, const char *node, const char *fmt, ...) { va_list ap; @@ -515,7 +513,7 @@ EXPORT_SYMBOL(xenbus_printf); /* Takes tuples of names, scanf-style args, and void **, NULL terminated. */ -int xenbus_gather(struct xenbus_transaction *t, const char *dir, ...) +int xenbus_gather(xenbus_transaction_t t, const char *dir, ...) { va_list ap; const char *name; @@ -553,7 +551,7 @@ iov[1].iov_base = (void *)token; iov[1].iov_len = strlen(token) + 1; - return xs_error(xs_talkv(NULL, XS_WATCH, iov, + return xs_error(xs_talkv(XBT_NULL, XS_WATCH, iov, ARRAY_SIZE(iov), NULL)); } @@ -566,7 +564,7 @@ iov[1].iov_base = (char *)token; iov[1].iov_len = strlen(token) + 1; - return xs_error(xs_talkv(NULL, XS_UNWATCH, iov, + return xs_error(xs_talkv(XBT_NULL, XS_UNWATCH, iov, ARRAY_SIZE(iov), NULL)); } diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/include/asm-xen/asm-i386/hypercall.h --- a/linux-2.6-xen-sparse/include/asm-xen/asm-i386/hypercall.h Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/include/asm-xen/asm-i386/hypercall.h Fri Jan 13 20:12:24 2006 @@ -32,6 +32,7 @@ #include <asm-xen/xen-public/xen.h> #include <asm-xen/xen-public/sched.h> +#include <asm-xen/xen-public/nmi.h> #define _hypercall0(type, name) \ ({ \ @@ -300,6 +301,14 @@ SHUTDOWN_suspend, srec); } +static inline int +HYPERVISOR_nmi_op( + unsigned long op, + unsigned long arg) +{ + return _hypercall2(int, nmi_op, op, arg); +} + #endif /* __HYPERCALL_H__ */ /* diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/include/asm-xen/asm-i386/mach-xen/setup_arch_post.h --- a/linux-2.6-xen-sparse/include/asm-xen/asm-i386/mach-xen/setup_arch_post.h Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/include/asm-xen/asm-i386/mach-xen/setup_arch_post.h Fri Jan 13 20:12:24 2006 @@ -29,6 +29,7 @@ extern void hypervisor_callback(void); extern void failsafe_callback(void); +extern void nmi(void); static void __init machine_specific_arch_setup(void) { @@ -36,5 +37,7 @@ __KERNEL_CS, (unsigned long)hypervisor_callback, __KERNEL_CS, (unsigned long)failsafe_callback); + HYPERVISOR_nmi_op(XENNMI_register_callback, (unsigned long)&nmi); + machine_specific_modify_cpu_capabilities(&boot_cpu_data); } diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/include/asm-xen/asm-x86_64/hypercall.h --- a/linux-2.6-xen-sparse/include/asm-xen/asm-x86_64/hypercall.h Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/include/asm-xen/asm-x86_64/hypercall.h Fri Jan 13 20:12:24 2006 @@ -287,9 +287,9 @@ } static inline int -HYPERVISOR_switch_to_user(void) -{ - return _hypercall0(int, switch_to_user); +HYPERVISOR_iret(void) +{ + return _hypercall0(int, iret); } static inline int @@ -305,6 +305,14 @@ { return _hypercall3(int, sched_op, SCHEDOP_shutdown, SHUTDOWN_suspend, srec); +} + +static inline int +HYPERVISOR_nmi_op( + unsigned long op, + unsigned long arg) +{ + return _hypercall2(int, nmi_op, op, arg); } #endif /* __HYPERCALL_H__ */ diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/include/asm-xen/asm-x86_64/mach-xen/setup_arch_post.h --- a/linux-2.6-xen-sparse/include/asm-xen/asm-x86_64/mach-xen/setup_arch_post.h Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/include/asm-xen/asm-x86_64/mach-xen/setup_arch_post.h Fri Jan 13 20:12:24 2006 @@ -35,6 +35,7 @@ extern void hypervisor_callback(void); extern void failsafe_callback(void); +extern void nmi(void); static void __init machine_specific_arch_setup(void) { @@ -43,5 +44,9 @@ (unsigned long) failsafe_callback, (unsigned long) system_call); +#ifdef CONFIG_X86_LOCAL_APIC + HYPERVISOR_nmi_op(XENNMI_register_callback, (unsigned long)&nmi); +#endif + machine_specific_modify_cpu_capabilities(&boot_cpu_data); } diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/include/asm-xen/xenbus.h --- a/linux-2.6-xen-sparse/include/asm-xen/xenbus.h Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/include/asm-xen/xenbus.h Fri Jan 13 20:12:24 2006 @@ -36,6 +36,8 @@ #include <asm/semaphore.h> #include <asm-xen/xen-public/io/xenbus.h> #include <asm-xen/xen-public/io/xs_wire.h> + +#define XBT_NULL 0 /* Register callback to watch this node. */ struct xenbus_watch @@ -100,35 +102,35 @@ int xenbus_register_backend(struct xenbus_driver *drv); void xenbus_unregister_driver(struct xenbus_driver *drv); -struct xenbus_transaction; - -char **xenbus_directory(struct xenbus_transaction *t, +typedef u32 xenbus_transaction_t; + +char **xenbus_directory(xenbus_transaction_t t, const char *dir, const char *node, unsigned int *num); -void *xenbus_read(struct xenbus_transaction *t, +void *xenbus_read(xenbus_transaction_t t, const char *dir, const char *node, unsigned int *len); -int xenbus_write(struct xenbus_transaction *t, +int xenbus_write(xenbus_transaction_t t, const char *dir, const char *node, const char *string); -int xenbus_mkdir(struct xenbus_transaction *t, +int xenbus_mkdir(xenbus_transaction_t t, const char *dir, const char *node); -int xenbus_exists(struct xenbus_transaction *t, +int xenbus_exists(xenbus_transaction_t t, const char *dir, const char *node); -int xenbus_rm(struct xenbus_transaction *t, const char *dir, const char *node); -struct xenbus_transaction *xenbus_transaction_start(void); -int xenbus_transaction_end(struct xenbus_transaction *t, int abort); +int xenbus_rm(xenbus_transaction_t t, const char *dir, const char *node); +int xenbus_transaction_start(xenbus_transaction_t *t); +int xenbus_transaction_end(xenbus_transaction_t t, int abort); /* Single read and scanf: returns -errno or num scanned if > 0. */ -int xenbus_scanf(struct xenbus_transaction *t, +int xenbus_scanf(xenbus_transaction_t t, const char *dir, const char *node, const char *fmt, ...) __attribute__((format(scanf, 4, 5))); /* Single printf and write: returns -errno or 0. */ -int xenbus_printf(struct xenbus_transaction *t, +int xenbus_printf(xenbus_transaction_t t, const char *dir, const char *node, const char *fmt, ...) __attribute__((format(printf, 4, 5))); /* Generic read function: NULL-terminated triples of name, * sprintf-style type string, and pointer. Returns 0 or errno.*/ -int xenbus_gather(struct xenbus_transaction *t, const char *dir, ...); +int xenbus_gather(xenbus_transaction_t t, const char *dir, ...); /* notifer routines for when the xenstore comes up */ int register_xenstore_notifier(struct notifier_block *nb); @@ -194,7 +196,7 @@ * XenbusStateClosing, and the error will be saved in the store. */ int xenbus_switch_state(struct xenbus_device *dev, - struct xenbus_transaction *xbt, + xenbus_transaction_t xbt, XenbusState new_state); diff -r 2884f14895d2 -r f5ab28885c75 tools/Makefile --- a/tools/Makefile Fri Jan 13 16:38:44 2006 +++ b/tools/Makefile Fri Jan 13 20:12:24 2006 @@ -12,6 +12,7 @@ SUBDIRS += security SUBDIRS += console SUBDIRS += xenmon +SUBDIRS += guest-headers ifeq ($(VTPM_TOOLS),y) SUBDIRS += vtpm_manager SUBDIRS += vtpm diff -r 2884f14895d2 -r f5ab28885c75 tools/Rules.mk --- a/tools/Rules.mk Fri Jan 13 16:38:44 2006 +++ b/tools/Rules.mk Fri Jan 13 20:12:24 2006 @@ -35,6 +35,8 @@ mk-symlinks: mkdir -p xen ( cd xen && ln -sf ../$(XEN_ROOT)/xen/include/public/*.h . ) + mkdir -p xen/hvm + ( cd xen/hvm && ln -sf ../../$(XEN_ROOT)/xen/include/public/hvm/*.h . ) mkdir -p xen/io ( cd xen/io && ln -sf ../../$(XEN_ROOT)/xen/include/public/io/*.h . ) mkdir -p xen/linux diff -r 2884f14895d2 -r f5ab28885c75 tools/console/client/main.c --- a/tools/console/client/main.c Fri Jan 13 16:38:44 2006 +++ b/tools/console/client/main.c Fri Jan 13 20:12:24 2006 @@ -220,7 +220,7 @@ if (path == NULL) err(ENOMEM, "realloc"); strcat(path, "/console/tty"); - str_pty = xs_read(xs, NULL, path, &len); + str_pty = xs_read(xs, XBT_NULL, path, &len); /* FIXME consoled currently does not assume domain-0 doesn't have a console which is good when we break domain-0 up. To keep us @@ -245,7 +245,7 @@ struct timeval tv = { 0, 500 }; select(0, NULL, NULL, NULL, &tv); /* pause briefly */ - str_pty = xs_read(xs, NULL, path, &len); + str_pty = xs_read(xs, XBT_NULL, path, &len); } if (str_pty == NULL) { diff -r 2884f14895d2 -r f5ab28885c75 tools/console/daemon/io.c --- a/tools/console/daemon/io.c Fri Jan 13 16:38:44 2006 +++ b/tools/console/daemon/io.c Fri Jan 13 20:12:24 2006 @@ -174,7 +174,7 @@ success = asprintf(&path, "%s/limit", dom->conspath) != -1; if (!success) goto out; - data = xs_read(xs, NULL, path, &len); + data = xs_read(xs, XBT_NULL, path, &len); if (data) { dom->buffer.max_capacity = strtoul(data, 0, 0); free(data); @@ -184,7 +184,7 @@ success = asprintf(&path, "%s/tty", dom->conspath) != -1; if (!success) goto out; - success = xs_write(xs, NULL, path, slave, strlen(slave)); + success = xs_write(xs, XBT_NULL, path, slave, strlen(slave)); free(path); if (!success) goto out; @@ -214,7 +214,7 @@ char *p; asprintf(&path, "%s/%s", dir, name); - p = xs_read(xs, NULL, path, NULL); + p = xs_read(xs, XBT_NULL, path, NULL); free(path); if (p == NULL) { ret = ENOENT; diff -r 2884f14895d2 -r f5ab28885c75 tools/debugger/libxendebug/xendebug.c --- a/tools/debugger/libxendebug/xendebug.c Fri Jan 13 16:38:44 2006 +++ b/tools/debugger/libxendebug/xendebug.c Fri Jan 13 20:12:24 2006 @@ -119,8 +119,8 @@ if ( !ctxt->valid[vcpu] ) { - if ( (rc = xc_domain_get_vcpu_context(xc_handle, domid, vcpu, - &ctxt->context[vcpu])) ) + if ( (rc = xc_vcpu_getcontext(xc_handle, domid, vcpu, + &ctxt->context[vcpu])) ) return NULL; ctxt->valid[vcpu] = true; @@ -139,10 +139,10 @@ return -EINVAL; op.interface_version = DOM0_INTERFACE_VERSION; - op.cmd = DOM0_SETDOMAININFO; - op.u.setdomaininfo.domain = ctxt->domid; - op.u.setdomaininfo.vcpu = vcpu; - op.u.setdomaininfo.ctxt = &ctxt->context[vcpu]; + op.cmd = DOM0_SETVCPUCONTEXT; + op.u.setvcpucontext.domain = ctxt->domid; + op.u.setvcpucontext.vcpu = vcpu; + op.u.setvcpucontext.ctxt = &ctxt->context[vcpu]; if ( (rc = mlock(&ctxt->context[vcpu], sizeof(vcpu_guest_context_t))) ) return rc; diff -r 2884f14895d2 -r f5ab28885c75 tools/examples/network-bridge --- a/tools/examples/network-bridge Fri Jan 13 16:38:44 2006 +++ b/tools/examples/network-bridge Fri Jan 13 20:12:24 2006 @@ -68,48 +68,19 @@ vdev="veth${vifnum}" vif0="vif0.${vifnum}" -legacy_mask_to_prefix() { - mask=$1 - first=${mask%%.*} - second=${mask#*.} - third=${second#*.} - fourth=${third#*.} - second=${second%%.*} - third=${third%%.*} - declare -i INT FULLMASK BIT - INT=$((((($first*256)+$second)*256+$third)*256+$fourth)) - FULLMASK=4294967295 - BIT=1 - for bit in `seq 32 -1 0`; do - if test $FULLMASK -eq $INT; then PREFIX=$bit; return; fi - FULLMASK=$(($FULLMASK-$BIT)) - BIT=$((BIT*2)) - done - echo "ERROR converting netmask $mask to prefix" - exit 1 -} - -parse_kernel_ip() { - if egrep 'ip=[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+:' /proc/cmdline; then - kip=`sed -e 's!.*ip=$[0-9]\+\.[0-9]\+\.[0-9]\+\.[0-9]\+$:.*!\1!' /proc/cmdline` - kmask=`sed -e 's!.*ip=[^:]*:[^:]*:[^:]*:$[^:]*$:.*!\1!' /proc/cmdline` - kgate=`sed -e 's!.*ip=[^:]*:[^:]*:$[^:]*$:.*!\1!' /proc/cmdline` - fi -} - +get_ip_info() { + addr_pfx=`ip addr show dev $1 | egrep '^ *inet' | sed -e 's/ *inet //' -e 's/ .*//'` + gateway=`ip route show dev $1 | fgrep default | sed 's/default via //'` +} + do_ifup() { if ! ifup $1 ; then - if [ ${kip} ] ; then - # use the addresses we grocked from /proc/cmdline - if [ -z "${kmask}" ]; then - PREFIX=32 - else - legacy_mask_to_prefix ${kmask} - fi + if [ ${addr_pfx} ] ; then + # use the info from get_ip_info() ip addr flush $1 - ip addr add ${kip}/${PREFIX} dev $1 + ip addr add ${addr_pfx} dev $1 ip link set dev $1 up - [ ${kgate} ] && ip route add default via ${kgate} + [ ${gateway} ] && ip route add default via ${gateway} fi fi } @@ -171,7 +142,7 @@ # link_exists() { - if ip link show "$1" >&/dev/null + if ip link show "$1" >/dev/null 2>/dev/null then return 0 else @@ -231,7 +202,7 @@ } op_start () { - if [ "${bridge}" == "null" ] ; then + if [ "${bridge}" = "null" ] ; then return fi @@ -259,9 +230,8 @@ preiftransfer ${netdev} transfer_addrs ${netdev} ${vdev} if ! ifdown ${netdev}; then - # If ifdown fails, take the IP details from the kernel command - # line. - parse_kernel_ip + # If ifdown fails, remember the IP details. + get_ip_info ${netdev} ip link set ${netdev} down ip addr flush ${netdev} fi @@ -283,13 +253,13 @@ transfer_routes ${netdev} ${bridge} fi - if [ ${antispoof} == 'yes' ] ; then + if [ ${antispoof} = 'yes' ] ; then antispoofing fi } op_stop () { - if [ "${bridge}" == "null" ]; then + if [ "${bridge}" = "null" ]; then return fi if ! link_exists "$bridge"; then @@ -301,7 +271,7 @@ mac=`ip link show ${netdev} | grep 'link\/ether' | sed -e 's/.*ether $..:..:..:..:..:..$.*/\1/'` transfer_addrs ${netdev} ${pdev} if ! ifdown ${netdev}; then - parse_kernel_ip + get_ip_info ${netdev} fi ip link set ${netdev} down arp off ip link set ${netdev} addr fe:ff:ff:ff:ff:ff diff -r 2884f14895d2 -r f5ab28885c75 tools/examples/xen-network-common.sh --- a/tools/examples/xen-network-common.sh Fri Jan 13 16:38:44 2006 +++ b/tools/examples/xen-network-common.sh Fri Jan 13 20:12:24 2006 @@ -42,7 +42,7 @@ { /sbin/ifup ${HWD_CONFIG_0} $1 } -elif ! which ifup >&/dev/null +elif ! which ifup >/dev/null 2>/dev/null then if [ -e /etc/conf.d/net ] then @@ -59,9 +59,18 @@ /etc/init.d/net.$1 stop } else - logger -p "daemon.crit" -- \ - "You don't have ifup and don't seem to be running Gentoo either!" - exit 1 + preiftransfer() + { + true + } + ifup() + { + false + } + ifdown() + { + false + } fi else preiftransfer() diff -r 2884f14895d2 -r f5ab28885c75 tools/examples/xmexample.vmx --- a/tools/examples/xmexample.vmx Fri Jan 13 16:38:44 2006 +++ b/tools/examples/xmexample.vmx Fri Jan 13 20:12:24 2006 @@ -28,10 +28,13 @@ #----------------------------------------------------------------------------- # the number of cpus guest platform has, default=1 -vcpus=1 +#vcpus=1 # enable/disalbe vmx guest ACPI, default=0 (disabled) #acpi=0 + +# enable/disalbe vmx guest APIC, default=0 (disabled) +#apic=0 # List of which CPUS this domain is allowed to use, default Xen picks #cpus = "" # leave to Xen to pick diff -r 2884f14895d2 -r f5ab28885c75 tools/firmware/vmxassist/acpi_madt.c --- a/tools/firmware/vmxassist/acpi_madt.c Fri Jan 13 16:38:44 2006 +++ b/tools/firmware/vmxassist/acpi_madt.c Fri Jan 13 20:12:24 2006 @@ -17,34 +17,34 @@ * this program; if not, write to the Free Software Foundation, Inc., 59 Temple * Place - Suite 330, Boston, MA 02111-1307 USA. */ + #include "../acpi/acpi2_0.h" #include "../acpi/acpi_madt.h" + +#include <xen/hvm/hvm_info_table.h> #define NULL ((void*)0) extern int puts(const char *s); -#define HVM_INFO_PAGE 0x0009F000 -#define HVM_INFO_OFFSET 0x00000800 - -struct hvm_info_table { - char signature[8]; /* "HVM INFO" */ - uint32_t length; - uint8_t checksum; - uint8_t acpi_enabled; - uint8_t pad[2]; - uint32_t nr_vcpus; -}; - static struct hvm_info_table *table = NULL; -static int -checksum_valid(uint8_t *ptr, int len) +static int validate_hvm_info(struct hvm_info_table *t) { - uint8_t sum=0; + char signature[] = "HVM INFO"; + uint8_t *ptr = (uint8_t *)t; + uint8_t sum = 0; int i; - for (i = 0; i < len; i++) + /* strncmp(t->signature, "HVM INFO", 8) */ + for (i = 0; i < 8; i++) { + if (signature[i] != t->signature[i]) { + puts("Bad hvm info signature\n"); + return 0; + } + } + + for (i = 0; i < t->length; i++) sum += ptr[i]; return (sum == 0); @@ -55,24 +55,15 @@ get_hvm_info_table(void) { struct hvm_info_table *t; - char signature[] = "HVM INFO"; int i; if (table != NULL) return table; - t = (struct hvm_info_table *)(HVM_INFO_PAGE + HVM_INFO_OFFSET); + t = (struct hvm_info_table *)HVM_INFO_PADDR; - /* strncmp(t->signature, "HVM INFO", 8) */ - for (i = 0; i < 8; i++) { - if (signature[i] != t->signature[i]) { - puts("Bad hvm info signature\n"); - return NULL; - } - } - - if (!checksum_valid((uint8_t *)t, t->length)) { - puts("Bad hvm info checksum\n"); + if (!validate_hvm_info(t)) { + puts("Bad hvm info table\n"); return NULL; } @@ -126,10 +117,10 @@ return madt; } -static void +static void set_checksum(void *start, int checksum_offset, int len) { - unsigned char sum = 0; + unsigned char sum = 0; unsigned char *ptr; ptr = start; @@ -141,9 +132,9 @@ ptr[checksum_offset] = -sum; } -static int +static int acpi_madt_set_local_apics( - int nr_vcpu, + int nr_vcpu, ACPI_MULTIPLE_APIC_DESCRIPTION_TABLE *madt) { int i; @@ -156,14 +147,14 @@ madt->LocalApic[i].Length = sizeof (ACPI_LOCAL_APIC_STRUCTURE); madt->LocalApic[i].AcpiProcessorId = i; madt->LocalApic[i].ApicId = i; - madt->LocalApic[i].Flags = 1; + madt->LocalApic[i].Flags = 1; } madt->Header.Header.Length = - sizeof(ACPI_MULTIPLE_APIC_DESCRIPTION_TABLE) - + sizeof(ACPI_MULTIPLE_APIC_DESCRIPTION_TABLE) - (MAX_VIRT_CPUS - nr_vcpu)* sizeof(ACPI_LOCAL_APIC_STRUCTURE); - return 0; + return 0; } #define FIELD_OFFSET(TYPE,Field) ((unsigned int)(&(((TYPE *) 0)->Field))) @@ -185,7 +176,7 @@ madt, FIELD_OFFSET(ACPI_TABLE_HEADER, Checksum), madt->Header.Header.Length); - return 0; + return 0; } /* diff -r 2884f14895d2 -r f5ab28885c75 tools/firmware/vmxassist/vm86.h --- a/tools/firmware/vmxassist/vm86.h Fri Jan 13 16:38:44 2006 +++ b/tools/firmware/vmxassist/vm86.h Fri Jan 13 20:12:24 2006 @@ -24,7 +24,7 @@ #include <stdint.h> #endif -#include <xen/vmx_assist.h> +#include <xen/hvm/vmx_assist.h> #define NR_EXCEPTION_HANDLER 32 #define NR_INTERRUPT_HANDLERS 16 diff -r 2884f14895d2 -r f5ab28885c75 tools/ioemu/hw/i8254.c --- a/tools/ioemu/hw/i8254.c Fri Jan 13 16:38:44 2006 +++ b/tools/ioemu/hw/i8254.c Fri Jan 13 20:12:24 2006 @@ -23,7 +23,7 @@ */ #include "vl.h" #include <xenctrl.h> -#include <xen/io/ioreq.h> +#include <xen/hvm/ioreq.h> //#define DEBUG_PIT diff -r 2884f14895d2 -r f5ab28885c75 tools/ioemu/hw/i8259.c --- a/tools/ioemu/hw/i8259.c Fri Jan 13 16:38:44 2006 +++ b/tools/ioemu/hw/i8259.c Fri Jan 13 20:12:24 2006 @@ -23,7 +23,7 @@ */ #include "vl.h" #include <xenctrl.h> -#include <xen/io/ioreq.h> +#include <xen/hvm/ioreq.h> /* debug PIC */ //#define DEBUG_PIC diff -r 2884f14895d2 -r f5ab28885c75 tools/ioemu/hw/i8259_stub.c --- a/tools/ioemu/hw/i8259_stub.c Fri Jan 13 16:38:44 2006 +++ b/tools/ioemu/hw/i8259_stub.c Fri Jan 13 20:12:24 2006 @@ -22,7 +22,7 @@ * THE SOFTWARE. */ #include "xenctrl.h" -#include <xen/io/ioreq.h> +#include <xen/hvm/ioreq.h> #include <stdio.h> #include "cpu.h" #include "cpu-all.h" diff -r 2884f14895d2 -r f5ab28885c75 tools/ioemu/target-i386-dm/helper2.c --- a/tools/ioemu/target-i386-dm/helper2.c Fri Jan 13 16:38:44 2006 +++ b/tools/ioemu/target-i386-dm/helper2.c Fri Jan 13 20:12:24 2006 @@ -48,7 +48,7 @@ #include <sys/ioctl.h> #include <xenctrl.h> -#include <xen/io/ioreq.h> +#include <xen/hvm/ioreq.h> #include <xen/linux/evtchn.h> #include "cpu.h" diff -r 2884f14895d2 -r f5ab28885c75 tools/libxc/xc_core.c --- a/tools/libxc/xc_core.c Fri Jan 13 16:38:44 2006 +++ b/tools/libxc/xc_core.c Fri Jan 13 20:12:24 2006 @@ -55,7 +55,7 @@ } for (i = 0; i < info.max_vcpu_id; i++) - if (xc_domain_get_vcpu_context(xc_handle, domid, + if (xc_vcpu_getcontext(xc_handle, domid, i, &ctxt[nr_vcpus]) == 0) nr_vcpus++; diff -r 2884f14895d2 -r f5ab28885c75 tools/libxc/xc_domain.c --- a/tools/libxc/xc_domain.c Fri Jan 13 16:38:44 2006 +++ b/tools/libxc/xc_domain.c Fri Jan 13 20:12:24 2006 @@ -58,16 +58,16 @@ return do_dom0_op(xc_handle, &op); } -int xc_domain_pincpu(int xc_handle, - uint32_t domid, - int vcpu, - cpumap_t cpumap) -{ - DECLARE_DOM0_OP; - op.cmd = DOM0_PINCPUDOMAIN; - op.u.pincpudomain.domain = (domid_t)domid; - op.u.pincpudomain.vcpu = vcpu; - op.u.pincpudomain.cpumap = cpumap; +int xc_vcpu_setaffinity(int xc_handle, + uint32_t domid, + int vcpu, + cpumap_t cpumap) +{ + DECLARE_DOM0_OP; + op.cmd = DOM0_SETVCPUAFFINITY; + op.u.setvcpuaffinity.domain = (domid_t)domid; + op.u.setvcpuaffinity.vcpu = vcpu; + op.u.setvcpuaffinity.cpumap = cpumap; return do_dom0_op(xc_handle, &op); } @@ -155,7 +155,7 @@ return ret; } -int xc_domain_get_vcpu_context(int xc_handle, +int xc_vcpu_getcontext(int xc_handle, uint32_t domid, uint32_t vcpu, vcpu_guest_context_t *ctxt) @@ -345,10 +345,10 @@ return do_dom0_op(xc_handle, &op); } -int xc_domain_get_vcpu_info(int xc_handle, - uint32_t domid, - uint32_t vcpu, - xc_vcpuinfo_t *info) +int xc_vcpu_getinfo(int xc_handle, + uint32_t domid, + uint32_t vcpu, + xc_vcpuinfo_t *info) { int rc; DECLARE_DOM0_OP; @@ -380,18 +380,18 @@ return do_dom0_op(xc_handle, &op); } -int xc_domain_setinfo(int xc_handle, - uint32_t domid, - uint32_t vcpu, - vcpu_guest_context_t *ctxt) +int xc_vcpu_setcontext(int xc_handle, + uint32_t domid, + uint32_t vcpu, + vcpu_guest_context_t *ctxt) { dom0_op_t op; int rc; - op.cmd = DOM0_SETDOMAININFO; - op.u.setdomaininfo.domain = domid; - op.u.setdomaininfo.vcpu = vcpu; - op.u.setdomaininfo.ctxt = ctxt; + op.cmd = DOM0_SETVCPUCONTEXT; + op.u.setvcpucontext.domain = domid; + op.u.setvcpucontext.vcpu = vcpu; + op.u.setvcpucontext.ctxt = ctxt; if ( (rc = mlock(ctxt, sizeof(*ctxt))) != 0 ) return rc; diff -r 2884f14895d2 -r f5ab28885c75 tools/libxc/xc_ia64_stubs.c --- a/tools/libxc/xc_ia64_stubs.c Fri Jan 13 16:38:44 2006 +++ b/tools/libxc/xc_ia64_stubs.c Fri Jan 13 20:12:24 2006 @@ -5,7 +5,7 @@ #include <stdlib.h> #include <zlib.h> #include "xen/arch-ia64.h" -#include <xen/io/ioreq.h> +#include <xen/hvm/ioreq.h> /* this is a very ugly way of getting FPSR_DEFAULT. struct ia64_fpreg is * mysteriously declared in two places: /usr/include/asm/fpu.h and @@ -23,7 +23,8 @@ } int xc_linux_save(int xc_handle, int io_fd, uint32_t dom, uint32_t max_iters, - uint32_t max_factor, uint32_t flags) + uint32_t max_factor, uint32_t flags /* XCFLAGS_xxx */, + int (*suspend)(int domid)) { PERROR("xc_linux_save not implemented\n"); return -1; @@ -664,7 +665,7 @@ goto error_out; } - if ( xc_domain_get_vcpu_context(xc_handle, domid, 0, ctxt) ){ + if ( xc_vcpu_getcontext(xc_handle, domid, 0, ctxt) ){ PERROR("Could not get vcpu context"); goto error_out; } @@ -688,11 +689,11 @@ memset( &launch_op, 0, sizeof(launch_op) ); - launch_op.u.setdomaininfo.domain = (domid_t)domid; - launch_op.u.setdomaininfo.vcpu = 0; - launch_op.u.setdomaininfo.ctxt = ctxt; - - launch_op.cmd = DOM0_SETDOMAININFO; + launch_op.u.setvcpucontext.domain = (domid_t)domid; + launch_op.u.setvcpucontext.vcpu = 0; + launch_op.u.setvcpucontext.ctxt = ctxt; + + launch_op.cmd = DOM0_SETVCPUCONTEXT; rc = do_dom0_op(xc_handle, &launch_op); return rc; diff -r 2884f14895d2 -r f5ab28885c75 tools/libxc/xc_linux_build.c --- a/tools/libxc/xc_linux_build.c Fri Jan 13 16:38:44 2006 +++ b/tools/libxc/xc_linux_build.c Fri Jan 13 20:12:24 2006 @@ -402,8 +402,11 @@ ctxt->initrd.start = 0; ctxt->initrd.size = 0; } - strncpy((char *)ctxt->cmdline, cmdline, IA64_COMMAND_LINE_SIZE); - ctxt->cmdline[IA64_COMMAND_LINE_SIZE-1] = '\0'; + if ( cmdline != NULL ) + { + strncpy((char *)ctxt->cmdline, cmdline, IA64_COMMAND_LINE_SIZE); + ctxt->cmdline[IA64_COMMAND_LINE_SIZE-1] = '\0'; + } munmap(start_info, PAGE_SIZE); free(page_array); @@ -693,8 +696,11 @@ start_info->mod_start = vinitrd_start; start_info->mod_len = initrd_len; } - strncpy((char *)start_info->cmd_line, cmdline, MAX_GUEST_CMDLINE); - start_info->cmd_line[MAX_GUEST_CMDLINE-1] = '\0'; + if ( cmdline != NULL ) + { + strncpy((char *)start_info->cmd_line, cmdline, MAX_GUEST_CMDLINE); + start_info->cmd_line[MAX_GUEST_CMDLINE-1] = '\0'; + } munmap(start_info, PAGE_SIZE); /* shared_info page starts its life empty. */ @@ -794,7 +800,7 @@ goto error_out; } - if ( xc_domain_get_vcpu_context(xc_handle, domid, 0, ctxt) ) + if ( xc_vcpu_getcontext(xc_handle, domid, 0, ctxt) ) { PERROR("Could not get vcpu context"); goto error_out; @@ -897,11 +903,11 @@ memset( &launch_op, 0, sizeof(launch_op) ); - launch_op.u.setdomaininfo.domain = (domid_t)domid; - launch_op.u.setdomaininfo.vcpu = 0; - launch_op.u.setdomaininfo.ctxt = ctxt; - - launch_op.cmd = DOM0_SETDOMAININFO; + launch_op.u.setvcpucontext.domain = (domid_t)domid; + launch_op.u.setvcpucontext.vcpu = 0; + launch_op.u.setvcpucontext.ctxt = ctxt; + + launch_op.cmd = DOM0_SETVCPUCONTEXT; rc = xc_dom0_op(xc_handle, &launch_op); return rc; diff -r 2884f14895d2 -r f5ab28885c75 tools/libxc/xc_linux_restore.c --- a/tools/libxc/xc_linux_restore.c Fri Jan 13 16:38:44 2006 +++ b/tools/libxc/xc_linux_restore.c Fri Jan 13 20:12:24 2006 @@ -171,7 +171,7 @@ /* Only have to worry about vcpu 0 even for SMP */ - if (xc_domain_get_vcpu_context( xc_handle, dom, 0, &ctxt)) { + if (xc_vcpu_getcontext( xc_handle, dom, 0, &ctxt)) { ERR("Could not get vcpu context"); goto out; } @@ -735,10 +735,10 @@ DPRINTF("Domain ready to be built.\n"); - op.cmd = DOM0_SETDOMAININFO; - op.u.setdomaininfo.domain = (domid_t)dom; - op.u.setdomaininfo.vcpu = 0; - op.u.setdomaininfo.ctxt = &ctxt; + op.cmd = DOM0_SETVCPUCONTEXT; + op.u.setvcpucontext.domain = (domid_t)dom; + op.u.setvcpucontext.vcpu = 0; + op.u.setvcpucontext.ctxt = &ctxt; rc = xc_dom0_op(xc_handle, &op); if (rc != 0) { diff -r 2884f14895d2 -r f5ab28885c75 tools/libxc/xc_linux_save.c --- a/tools/libxc/xc_linux_save.c Fri Jan 13 16:38:44 2006 +++ b/tools/libxc/xc_linux_save.c Fri Jan 13 20:12:24 2006 @@ -357,21 +357,14 @@ } -static int suspend_and_state(int xc_handle, int io_fd, int dom, - xc_dominfo_t *info, +static int suspend_and_state(int (*suspend)(int), int xc_handle, int io_fd, + int dom, xc_dominfo_t *info, vcpu_guest_context_t *ctxt) { int i = 0; - char ans[30]; - - printf("suspend\n"); - fflush(stdout); - if (fgets(ans, sizeof(ans), stdin) == NULL) { - ERR("failed reading suspend reply"); - return -1; - } - if (strncmp(ans, "done\n", 5)) { - ERR("suspend reply incorrect: %s", ans); + + if (!(*suspend)(dom)) { + ERR("Suspend request failed"); return -1; } @@ -382,7 +375,7 @@ return -1; } - if ( xc_domain_get_vcpu_context(xc_handle, dom, 0 /* XXX */, ctxt)) + if ( xc_vcpu_getcontext(xc_handle, dom, 0 /* XXX */, ctxt)) ERR("Could not get vcpu context"); @@ -568,7 +561,7 @@ int xc_linux_save(int xc_handle, int io_fd, uint32_t dom, uint32_t max_iters, - uint32_t max_factor, uint32_t flags) + uint32_t max_factor, uint32_t flags, int (*suspend)(int)) { xc_dominfo_t info; @@ -643,7 +636,7 @@ } /* Only have to worry about vcpu 0 even for SMP */ - if (xc_domain_get_vcpu_context(xc_handle, dom, 0, &ctxt)) { + if (xc_vcpu_getcontext(xc_handle, dom, 0, &ctxt)) { ERR("Could not get vcpu context"); goto out; } @@ -748,7 +741,7 @@ last_iter = 1; - if (suspend_and_state( xc_handle, io_fd, dom, &info, &ctxt)) { + if (suspend_and_state(suspend, xc_handle, io_fd, dom, &info, &ctxt)) { ERR("Domain appears not to have suspended"); goto out; } @@ -1054,7 +1047,8 @@ DPRINTF("Start last iteration\n"); last_iter = 1; - if (suspend_and_state(xc_handle, io_fd, dom, &info, &ctxt)) { + if (suspend_and_state(suspend, xc_handle, io_fd, dom, &info, + &ctxt)) { ERR("Domain appears not to have suspended"); goto out; } @@ -1164,6 +1158,9 @@ if (live_shinfo) munmap(live_shinfo, PAGE_SIZE); + if (live_p2m_frame_list_list) + munmap(live_p2m_frame_list_list, PAGE_SIZE); + if (live_p2m_frame_list) munmap(live_p2m_frame_list, P2M_FLL_ENTRIES * PAGE_SIZE); diff -r 2884f14895d2 -r f5ab28885c75 tools/libxc/xc_pagetab.c --- a/tools/libxc/xc_pagetab.c Fri Jan 13 16:38:44 2006 +++ b/tools/libxc/xc_pagetab.c Fri Jan 13 20:12:24 2006 @@ -74,7 +74,7 @@ #define pt_levels 4 #endif - if (xc_domain_get_vcpu_context(xc_handle, dom, vcpu, &ctx) != 0) { + if (xc_vcpu_getcontext(xc_handle, dom, vcpu, &ctx) != 0) { fprintf(stderr, "failed to retreive vcpu context\n"); goto out; } diff -r 2884f14895d2 -r f5ab28885c75 tools/libxc/xc_ptrace.c --- a/tools/libxc/xc_ptrace.c Fri Jan 13 16:38:44 2006 +++ b/tools/libxc/xc_ptrace.c Fri Jan 13 20:12:24 2006 @@ -33,7 +33,7 @@ if (online) *online = 0; if ( !(regs_valid & (1 << cpu)) ) { - retval = xc_domain_get_vcpu_context(xc_handle, current_domid, + retval = xc_vcpu_getcontext(xc_handle, current_domid, cpu, &ctxt[cpu]); if ( retval ) goto done; @@ -43,8 +43,7 @@ if ( online == NULL ) goto done; - retval = xc_domain_get_vcpu_info(xc_handle, current_domid, - cpu, &info); + retval = xc_vcpu_getinfo(xc_handle, current_domid, cpu, &info); *online = info.online; done: @@ -395,7 +394,7 @@ case PTRACE_SETREGS: SET_XC_REGS(((struct gdb_regs *)data), ctxt[cpu].user_regs); - retval = xc_domain_setinfo(xc_handle, current_domid, cpu, &ctxt[cpu]); + retval = xc_vcpu_setcontext(xc_handle, current_domid, cpu, &ctxt[cpu]); if (retval) goto error_out; break; @@ -405,7 +404,7 @@ * during single-stepping - but that just seems retarded */ ctxt[cpu].user_regs.eflags |= PSL_T; - retval = xc_domain_setinfo(xc_handle, current_domid, cpu, &ctxt[cpu]); + retval = xc_vcpu_setcontext(xc_handle, current_domid, cpu, &ctxt[cpu]); if ( retval ) { perror("dom0 op failed"); @@ -423,8 +422,8 @@ /* Clear trace flag */ if ( ctxt[cpu].user_regs.eflags & PSL_T ) { ctxt[cpu].user_regs.eflags &= ~PSL_T; - retval = xc_domain_setinfo(xc_handle, current_domid, - cpu, &ctxt[cpu]); + retval = xc_vcpu_setcontext(xc_handle, current_domid, + cpu, &ctxt[cpu]); if ( retval ) { perror("dom0 op failed"); goto error_out; diff -r 2884f14895d2 -r f5ab28885c75 tools/libxc/xc_vmx_build.c --- a/tools/libxc/xc_vmx_build.c Fri Jan 13 16:38:44 2006 +++ b/tools/libxc/xc_vmx_build.c Fri Jan 13 20:12:24 2006 @@ -9,7 +9,8 @@ #include <stdlib.h> #include <unistd.h> #include <zlib.h> -#include <xen/io/ioreq.h> +#include <xen/hvm/hvm_info_table.h> +#include <xen/hvm/ioreq.h> #define VMX_LOADER_ENTR_ADDR 0x00100000 @@ -33,18 +34,6 @@ #define E820_MAP_NR_OFFSET 0x000001E8 #define E820_MAP_OFFSET 0x000002D0 -#define HVM_INFO_PAGE 0x0009F000 -#define HVM_INFO_OFFSET 0x00000800 - -struct hvm_info_table { - char signature[8]; /* "HVM INFO" */ - uint32_t length; - uint8_t checksum; - uint8_t acpi_enabled; - uint8_t pad[2]; - uint32_t nr_vcpus; -}; - struct e820entry { uint64_t addr; uint64_t size; @@ -128,7 +117,7 @@ return (*(((unsigned char *)e820_page) + E820_MAP_NR_OFFSET) = nr_map); } -static void +static void set_hvm_info_checksum(struct hvm_info_table *t) { uint8_t *ptr = (uint8_t *)t, sum = 0; @@ -148,14 +137,18 @@ */ static int set_hvm_info(int xc_handle, uint32_t dom, unsigned long *pfn_list, unsigned int vcpus, - unsigned int acpi) + unsigned int acpi, unsigned int apic) { char *va_map; struct hvm_info_table *va_hvm; - va_map = xc_map_foreign_range(xc_handle, dom, - PAGE_SIZE, PROT_READ|PROT_WRITE, - pfn_list[HVM_INFO_PAGE >> PAGE_SHIFT]); + va_map = xc_map_foreign_range( + xc_handle, + dom, + PAGE_SIZE, + PROT_READ|PROT_WRITE, + pfn_list[HVM_INFO_PFN]); + if ( va_map == NULL ) return -1; @@ -164,8 +157,9 @@ strncpy(va_hvm->signature, "HVM INFO", 8); va_hvm->length = sizeof(struct hvm_info_table); va_hvm->acpi_enabled = acpi; + va_hvm->apic_enabled = apic; va_hvm->nr_vcpus = vcpus; - + set_hvm_info_checksum(va_hvm); munmap(va_map, PAGE_SIZE); @@ -307,9 +301,9 @@ vcpu_guest_context_t *ctxt, unsigned long shared_info_frame, unsigned int control_evtchn, - unsigned int lapic, unsigned int vcpus, unsigned int acpi, + unsigned int apic, unsigned int store_evtchn, unsigned long *store_mfn) { @@ -519,20 +513,14 @@ goto error_out; } - if (set_hvm_info(xc_handle, dom, page_array, vcpus, acpi)) { + if ( set_hvm_info(xc_handle, dom, page_array, vcpus, acpi, apic) ) { fprintf(stderr, "Couldn't set hvm info for VMX guest.\n"); goto error_out; } - *store_mfn = page_array[(v_end-2) >> PAGE_SHIFT]; - if ( xc_clear_domain_page(xc_handle, dom, *store_mfn) ) - goto error_out; - - shared_page_frame = (v_end - PAGE_SIZE) >> PAGE_SHIFT; - - if ((e820_page = xc_map_foreign_range( - xc_handle, dom, PAGE_SIZE, PROT_READ|PROT_WRITE, - page_array[E820_MAP_PAGE >> PAGE_SHIFT])) == 0) + if ( (e820_page = xc_map_foreign_range( + xc_handle, dom, PAGE_SIZE, PROT_READ|PROT_WRITE, + page_array[E820_MAP_PAGE >> PAGE_SHIFT])) == 0 ) goto error_out; memset(e820_page, 0, PAGE_SIZE); e820_map_nr = build_e820map(e820_page, v_end); @@ -547,25 +535,29 @@ munmap(e820_page, PAGE_SIZE); /* shared_info page starts its life empty. */ - if ((shared_info = xc_map_foreign_range( - xc_handle, dom, PAGE_SIZE, PROT_READ|PROT_WRITE, - shared_info_frame)) == 0) + if ( (shared_info = xc_map_foreign_range( + xc_handle, dom, PAGE_SIZE, PROT_READ|PROT_WRITE, + shared_info_frame)) == 0 ) goto error_out; memset(shared_info, 0, sizeof(shared_info_t)); /* Mask all upcalls... */ for ( i = 0; i < MAX_VIRT_CPUS; i++ ) shared_info->vcpu_info[i].evtchn_upcall_mask = 1; - munmap(shared_info, PAGE_SIZE); /* Populate the event channel port in the shared page */ - if ((sp = (shared_iopage_t *) xc_map_foreign_range( - xc_handle, dom, PAGE_SIZE, PROT_READ|PROT_WRITE, - page_array[shared_page_frame])) == 0) + shared_page_frame = page_array[(v_end >> PAGE_SHIFT) - 1]; + if ( (sp = (shared_iopage_t *) xc_map_foreign_range( + xc_handle, dom, PAGE_SIZE, PROT_READ|PROT_WRITE, + shared_page_frame)) == 0 ) goto error_out; memset(sp, 0, PAGE_SIZE); sp->sp_global.eport = control_evtchn; munmap(sp, PAGE_SIZE); + + *store_mfn = page_array[(v_end >> PAGE_SHIFT) - 2]; + if ( xc_clear_domain_page(xc_handle, dom, *store_mfn) ) + goto error_out; /* Send the page update requests down to the hypervisor. */ if ( xc_finish_mmu_updates(xc_handle, mmu) ) @@ -588,7 +580,7 @@ ctxt->user_regs.eax = 0; ctxt->user_regs.esp = 0; ctxt->user_regs.ebx = 0; /* startup_32 expects this to be 0 to signal boot cpu */ - ctxt->user_regs.ecx = lapic; + ctxt->user_regs.ecx = 0; ctxt->user_regs.esi = 0; ctxt->user_regs.edi = 0; ctxt->user_regs.ebp = 0; @@ -608,9 +600,9 @@ int memsize, const char *image_name, unsigned int control_evtchn, - unsigned int lapic, unsigned int vcpus, unsigned int acpi, + unsigned int apic, unsigned int store_evtchn, unsigned long *store_mfn) { @@ -659,7 +651,7 @@ goto error_out; } - if ( xc_domain_get_vcpu_context(xc_handle, domid, 0, ctxt) ) + if ( xc_vcpu_getcontext(xc_handle, domid, 0, ctxt) ) { PERROR("Could not get vcpu context"); goto error_out; @@ -674,7 +666,7 @@ if ( setup_guest(xc_handle, domid, memsize, image, image_size, nr_pages, ctxt, op.u.getdomaininfo.shared_info_frame, control_evtchn, - lapic, vcpus, acpi, store_evtchn, store_mfn) < 0) + vcpus, acpi, apic, store_evtchn, store_mfn) < 0) { ERROR("Error constructing guest OS"); goto error_out; @@ -716,11 +708,11 @@ memset( &launch_op, 0, sizeof(launch_op) ); - launch_op.u.setdomaininfo.domain = (domid_t)domid; - launch_op.u.setdomaininfo.vcpu = 0; - launch_op.u.setdomaininfo.ctxt = ctxt; - - launch_op.cmd = DOM0_SETDOMAININFO; + launch_op.u.setvcpucontext.domain = (domid_t)domid; + launch_op.u.setvcpucontext.vcpu = 0; + launch_op.u.setvcpucontext.ctxt = ctxt; + + launch_op.cmd = DOM0_SETVCPUCONTEXT; rc = xc_dom0_op(xc_handle, &launch_op); return rc; diff -r 2884f14895d2 -r f5ab28885c75 tools/libxc/xenctrl.h --- a/tools/libxc/xenctrl.h Fri Jan 13 16:38:44 2006 +++ b/tools/libxc/xenctrl.h Fri Jan 13 20:12:24 2006 @@ -181,10 +181,11 @@ */ int xc_domain_destroy(int xc_handle, uint32_t domid); -int xc_domain_pincpu(int xc_handle, - uint32_t domid, - int vcpu, - cpumap_t cpumap); + +int xc_vcpu_setaffinity(int xc_handle, + uint32_t domid, + int vcpu, + cpumap_t cpumap); /** * This function will return information about one or more domains. It is @@ -208,7 +209,7 @@ /** - * This function will set the vcpu context for the specified domain. + * This function will set the execution context for the specified vcpu. * * @parm xc_handle a handle to an open hypervisor interface * @parm domid the domain to set the vcpu context for @@ -216,10 +217,10 @@ * @parm ctxt pointer to the the cpu context with the values to set * @return the number of domains enumerated or -1 on error */ -int xc_domain_setinfo(int xc_handle, - uint32_t domid, - uint32_t vcpu, - vcpu_guest_context_t *ctxt); +int xc_vcpu_setcontext(int xc_handle, + uint32_t domid, + uint32_t vcpu, + vcpu_guest_context_t *ctxt); /** * This function will return information about one or more domains, using a * single hypercall. The domain information will be stored into the supplied @@ -249,17 +250,16 @@ * domain * @return 0 on success, -1 on failure */ -int xc_domain_get_vcpu_context(int xc_handle, +int xc_vcpu_getcontext(int xc_handle, uint32_t domid, uint32_t vcpu, vcpu_guest_context_t *ctxt); typedef dom0_getvcpuinfo_t xc_vcpuinfo_t; -int xc_domain_get_vcpu_info(int xc_handle, - uint32_t domid, - uint32_t vcpu, - xc_vcpuinfo_t *info); - +int xc_vcpu_getinfo(int xc_handle, + uint32_t domid, + uint32_t vcpu, + xc_vcpuinfo_t *info); int xc_domain_setcpuweight(int xc_handle, uint32_t domid, diff -r 2884f14895d2 -r f5ab28885c75 tools/libxc/xenguest.h --- a/tools/libxc/xenguest.h Fri Jan 13 16:38:44 2006 +++ b/tools/libxc/xenguest.h Fri Jan 13 20:12:24 2006 @@ -21,8 +21,10 @@ * @parm dom the id of the domain * @return 0 on success, -1 on failure */ -int xc_linux_save(int xc_handle, int fd, uint32_t dom, uint32_t max_iters, - uint32_t max_factor, uint32_t flags /* XCFLAGS_xxx */); +int xc_linux_save(int xc_handle, int io_fd, uint32_t dom, uint32_t max_iters, + uint32_t max_factor, uint32_t flags /* XCFLAGS_xxx */, + int (*suspend)(int domid)); + /** * This function will restore a saved domain running Linux. @@ -56,9 +58,9 @@ int memsize, const char *image_name, unsigned int control_evtchn, - unsigned int lapic, unsigned int vcpus, unsigned int acpi, + unsigned int apic, unsigned int store_evtchn, unsigned long *store_mfn); diff -r 2884f14895d2 -r f5ab28885c75 tools/libxc/xg_private.c --- a/tools/libxc/xg_private.c Fri Jan 13 16:38:44 2006 +++ b/tools/libxc/xg_private.c Fri Jan 13 20:12:24 2006 @@ -16,6 +16,9 @@ gzFile kernel_gfd = NULL; char *image = NULL; unsigned int bytes; + + if ( filename == NULL ) + goto out; if ( (kernel_fd = open(filename, O_RDONLY)) < 0 ) { diff -r 2884f14895d2 -r f5ab28885c75 tools/python/xen/lowlevel/xc/xc.c --- a/tools/python/xen/lowlevel/xc/xc.c Fri Jan 13 16:38:44 2006 +++ b/tools/python/xen/lowlevel/xc/xc.c Fri Jan 13 20:12:24 2006 @@ -16,7 +16,7 @@ #include <netdb.h> #include <arpa/inet.h> -#include "xc_private.h" +#include "xenctrl.h" /* Needed for Python versions earlier than 2.3. */ #ifndef PyMODINIT_FUNC @@ -135,9 +135,9 @@ } -static PyObject *pyxc_domain_pincpu(XcObject *self, - PyObject *args, - PyObject *kwds) +static PyObject *pyxc_vcpu_setaffinity(XcObject *self, + PyObject *args, + PyObject *kwds) { uint32_t dom; int vcpu = 0, i; @@ -157,7 +157,7 @@ cpumap |= (cpumap_t)1 << PyInt_AsLong(PyList_GetItem(cpulist, i)); } - if ( xc_domain_pincpu(self->xc_handle, dom, vcpu, cpumap) != 0 ) + if ( xc_vcpu_setaffinity(self->xc_handle, dom, vcpu, cpumap) != 0 ) return PyErr_SetFromErrno(xc_error); Py_INCREF(zero); @@ -297,7 +297,7 @@ &dom, &vcpu) ) return NULL; - rc = xc_domain_get_vcpu_info(self->xc_handle, dom, vcpu, &info); + rc = xc_vcpu_getinfo(self->xc_handle, dom, vcpu, &info); if ( rc < 0 ) return PyErr_SetFromErrno(xc_error); @@ -362,22 +362,23 @@ uint32_t dom; char *image; int control_evtchn, store_evtchn; + int memsize; int vcpus = 1; - int lapic = 0; int acpi = 0; - int memsize; + int apic = 0; unsigned long store_mfn = 0; static char *kwd_list[] = { "dom", "control_evtchn", "store_evtchn", - "memsize", "image", "lapic", "vcpus", "acpi",NULL }; + "memsize", "image", "vcpus", "acpi", "apic", + NULL }; if ( !PyArg_ParseTupleAndKeywords(args, kwds, "iiiisiii", kwd_list, &dom, &control_evtchn, &store_evtchn, - &memsize, &image, &lapic, &vcpus,&acpi) ) + &memsize, &image, &vcpus, &acpi, &apic) ) return NULL; if ( xc_vmx_build(self->xc_handle, dom, memsize, image, control_evtchn, - lapic, vcpus, acpi, store_evtchn, &store_mfn) != 0 ) + vcpus, acpi, apic, store_evtchn, &store_mfn) != 0 ) return PyErr_SetFromErrno(xc_error); return Py_BuildValue("{s:i}", "store_mfn", store_mfn); @@ -889,8 +890,8 @@ " dom [int]: Identifier of domain to be destroyed.\n\n" "Returns: [int] 0 on success; -1 on error.\n" }, - { "domain_pincpu", - (PyCFunction)pyxc_domain_pincpu, + { "vcpu_setaffinity", + (PyCFunction)pyxc_vcpu_setaffinity, METH_VARARGS | METH_KEYWORDS, "\n" "Pin a VCPU to a specified set CPUs.\n" " dom [int]: Identifier of domain to which VCPU belongs.\n" diff -r 2884f14895d2 -r f5ab28885c75 tools/python/xen/lowlevel/xs/xs.c --- a/tools/python/xen/lowlevel/xs/xs.c Fri Jan 13 16:38:44 2006 +++ b/tools/python/xen/lowlevel/xs/xs.c Fri Jan 13 20:12:24 2006 @@ -66,7 +66,7 @@ static int parse_transaction_path(XsHandle *self, PyObject *args, struct xs_handle **xh, - struct xs_transaction_handle **th, + xs_transaction_t *th, char **path); @@ -83,7 +83,7 @@ static PyObject *xspy_read(XsHandle *self, PyObject *args) { struct xs_handle *xh; - struct xs_transaction_handle *th; + xs_transaction_t th; char *path; char *xsval; @@ -120,7 +120,7 @@ { static char *arg_spec = "sss#"; struct xs_handle *xh = xshandle(self); - struct xs_transaction_handle *th; + xs_transaction_t th; char *thstr; char *path; char *data; @@ -132,7 +132,7 @@ if (!PyArg_ParseTuple(args, arg_spec, &thstr, &path, &data, &data_n)) return NULL; - th = (struct xs_transaction_handle *)strtoul(thstr, NULL, 16); + th = strtoul(thstr, NULL, 16); Py_BEGIN_ALLOW_THREADS result = xs_write(xh, th, path, data, data_n); @@ -155,7 +155,7 @@ static PyObject *xspy_ls(XsHandle *self, PyObject *args) { struct xs_handle *xh; - struct xs_transaction_handle *th; + xs_transaction_t th; char *path; char **xsval; @@ -193,7 +193,7 @@ static PyObject *xspy_mkdir(XsHandle *self, PyObject *args) { struct xs_handle *xh; - struct xs_transaction_handle *th; + xs_transaction_t th; char *path; bool result; @@ -221,7 +221,7 @@ static PyObject *xspy_rm(XsHandle *self, PyObject *args) { struct xs_handle *xh; - struct xs_transaction_handle *th; + xs_transaction_t th; char *path; bool result; @@ -256,7 +256,7 @@ unsigned int perms_n = 0; int i; - struct xs_transaction_handle *th; + xs_transaction_t th; char *thstr; if (!xh) @@ -264,7 +264,7 @@ if (!PyArg_ParseTuple(args, arg_spec, &thstr, &path)) return NULL; - th = (struct xs_transaction_handle *)strtoul(thstr, NULL, 16); + th = strtoul(thstr, NULL, 16); Py_BEGIN_ALLOW_THREADS perms = xs_get_permissions(xh, th, path, &perms_n); @@ -312,7 +312,7 @@ int xsperms_n; PyObject *tuple0 = NULL; - struct xs_transaction_handle *th; + xs_transaction_t th; char *thstr; if (!xh) @@ -320,7 +320,7 @@ if (!PyArg_ParseTuple(args, "ssO", &thstr, &path, &perms)) goto exit; - th = (struct xs_transaction_handle *)strtoul(thstr, NULL, 16); + th = strtoul(thstr, NULL, 16); if (!PyList_Check(perms)) { PyErr_SetString(PyExc_RuntimeError, "perms must be a list"); @@ -509,7 +509,7 @@ static PyObject *xspy_transaction_start(XsHandle *self) { struct xs_handle *xh = xshandle(self); - struct xs_transaction_handle *th; + xs_transaction_t th; char thstr[MAX_STRLEN(unsigned long) + 1]; if (!xh) @@ -519,7 +519,7 @@ th = xs_transaction_start(xh); Py_END_ALLOW_THREADS - if (th == NULL) { + if (th == XBT_NULL) { PyErr_SetFromErrno(PyExc_RuntimeError); return NULL; } @@ -547,7 +547,7 @@ struct xs_handle *xh = xshandle(self); bool result; - struct xs_transaction_handle *th; + xs_transaction_t th; char *thstr; if (!xh) @@ -556,7 +556,7 @@ &thstr, &abort)) return NULL; - th = (struct xs_transaction_handle *)strtoul(thstr, NULL, 16); + th = strtoul(thstr, NULL, 16); Py_BEGIN_ALLOW_THREADS result = xs_transaction_end(xh, th, abort); @@ -727,7 +727,7 @@ */ static int parse_transaction_path(XsHandle *self, PyObject *args, struct xs_handle **xh, - struct xs_transaction_handle **th, + xs_transaction_t *th, char **path) { char *thstr; @@ -740,7 +740,7 @@ if (!PyArg_ParseTuple(args, "ss", &thstr, path)) return 0; - *th = (struct xs_transaction_handle *)strtoul(thstr, NULL, 16); + *th = strtoul(thstr, NULL, 16); return 1; } diff -r 2884f14895d2 -r f5ab28885c75 tools/python/xen/xend/XendDomain.py --- a/tools/python/xen/xend/XendDomain.py Fri Jan 13 16:38:44 2006 +++ b/tools/python/xen/xend/XendDomain.py Fri Jan 13 20:12:24 2006 @@ -443,7 +443,7 @@ cpumap = map(lambda x: int(x), cpumap.replace("[", "").replace("]", "").split(",")) try: - return xc.domain_pincpu(dominfo.getDomid(), vcpu, cpumap) + return xc.vcpu_setaffinity(dominfo.getDomid(), vcpu, cpumap) except Exception, ex: raise XendError(str(ex)) diff -r 2884f14895d2 -r f5ab28885c75 tools/python/xen/xend/XendDomainInfo.py --- a/tools/python/xen/xend/XendDomainInfo.py Fri Jan 13 16:38:44 2006 +++ b/tools/python/xen/xend/XendDomainInfo.py Fri Jan 13 20:12:24 2006 @@ -1179,7 +1179,7 @@ for v in range(0, self.info['max_vcpu_id']+1): # pincpu takes a list of ints cpu = [ int( cpus[v % len(cpus)] ) ] - xc.domain_pincpu(self.domid, v, cpu) + xc.vcpu_setaffinity(self.domid, v, cpu) m = self.image.getDomainMemory(self.info['memory'] * 1024) balloon.free(m) diff -r 2884f14895d2 -r f5ab28885c75 tools/python/xen/xend/image.py --- a/tools/python/xen/xend/image.py Fri Jan 13 16:38:44 2006 +++ b/tools/python/xen/xend/image.py Fri Jan 13 20:12:24 2006 @@ -209,13 +209,9 @@ self.dmargs += self.configVNC(imageConfig) - self.lapic = 0 - lapic = sxp.child_value(imageConfig, 'lapic') - if not lapic is None: - self.lapic = int(lapic) - self.acpi = int(sxp.child_value(imageConfig, 'acpi', 0)) - + self.apic = int(sxp.child_value(imageConfig, 'apic', 0)) + def buildDomain(self): # Create an event channel self.device_channel = xc.evtchn_alloc_unbound(dom=self.vm.getDomid(), @@ -229,18 +225,18 @@ log.debug("control_evtchn = %d", self.device_channel) log.debug("store_evtchn = %d", store_evtchn) log.debug("memsize = %d", self.vm.getMemoryTarget() / 1024) - log.debug("lapic = %d", self.lapic) log.debug("vcpus = %d", self.vm.getVCpuCount()) log.debug("acpi = %d", self.acpi) + log.debug("apic = %d", self.apic) return xc.vmx_build(dom = self.vm.getDomid(), image = self.kernel, control_evtchn = self.device_channel, store_evtchn = store_evtchn, memsize = self.vm.getMemoryTarget() / 1024, - lapic = self.lapic, + vcpus = self.vm.getVCpuCount(), acpi = self.acpi, - vcpus = self.vm.getVCpuCount()) + apic = self.apic) # Return a list of cmd line args to the device models based on the # xm config file diff -r 2884f14895d2 -r f5ab28885c75 tools/python/xen/xend/server/DevController.py --- a/tools/python/xen/xend/server/DevController.py Fri Jan 13 16:38:44 2006 +++ b/tools/python/xen/xend/server/DevController.py Fri Jan 13 20:12:24 2006 @@ -32,11 +32,10 @@ HOTPLUG_STATUS_BUSY = "busy" Connected = 1 -Died = 2 -Error = 3 -Missing = 4 -Timeout = 5 -Busy = 6 +Error = 2 +Missing = 3 +Timeout = 4 +Busy = 5 xenbusState = { 'Unknown' : 0, @@ -157,11 +156,6 @@ raise VmError("Device %s (%s) could not be connected. " "Device not found." % (devid, self.deviceClass)) - elif status == Died: - self.destroyDevice(devid) - raise VmError("Device %s (%s) could not be connected. " - "Device has died." % (devid, self.deviceClass)) - elif status == Busy: err = None frontpath = self.frontendPath(devid) @@ -408,20 +402,17 @@ def hotplugStatusCallback(statusPath, ev, result): log.debug("hotplugStatusCallback %s.", statusPath) - try: - status = xstransact.Read(statusPath) - - if status is not None: - if status == HOTPLUG_STATUS_ERROR: - result['status'] = Error - elif status == HOTPLUG_STATUS_BUSY: - result['status'] = Busy - else: - result['status'] = Connected - else: - return 1 - except VmError: - result['status'] = Died + status = xstransact.Read(statusPath) + + if status is not None: + if status == HOTPLUG_STATUS_ERROR: + result['status'] = Error + elif status == HOTPLUG_STATUS_BUSY: + result['status'] = Busy + else: + result['status'] = Connected + else: + return 1 log.debug("hotplugStatusCallback %d.", result['status']) diff -r 2884f14895d2 -r f5ab28885c75 tools/python/xen/xm/create.py --- a/tools/python/xen/xm/create.py Fri Jan 13 16:38:44 2006 +++ b/tools/python/xen/xm/create.py Fri Jan 13 20:12:24 2006 @@ -160,13 +160,13 @@ fn=set_int, default=None, use="CPUS to run the domain on.") -gopts.var('lapic', val='LAPIC', - fn=set_int, default=0, - use="Disable or enable local APIC of VMX domain.") - gopts.var('acpi', val='ACPI', fn=set_int, default=0, use="Disable or enable ACPI of VMX domain.") + +gopts.var('apic', val='APIC', + fn=set_int, default=0, + use="Disable or enable APIC of VMX domain.") gopts.var('vcpus', val='VCPUS', fn=set_int, default=1, @@ -534,8 +534,8 @@ """ args = [ 'device_model', 'vcpus', 'cdrom', 'boot', 'fda', 'fdb', 'localtime', 'serial', 'stdvga', 'isa', 'nographic', 'audio', - 'vnc', 'vncviewer', 'sdl', 'display', 'ne2000', 'lapic', - 'xauthority', 'acpi' ] + 'vnc', 'vncviewer', 'sdl', 'display', 'ne2000', 'acpi', 'apic', + 'xauthority' ] for a in args: if (vals.__dict__[a]): config_image.append([a, vals.__dict__[a]]) diff -r 2884f14895d2 -r f5ab28885c75 tools/python/xen/xm/main.py --- a/tools/python/xen/xm/main.py Fri Jan 13 16:38:44 2006 +++ b/tools/python/xen/xm/main.py Fri Jan 13 20:12:24 2006 @@ -390,7 +390,6 @@ def xm_vcpu_list(args): - print 'Name ID VCPU CPU State Time(s) CPU Affinity' from xen.xend.XendClient import server if args: @@ -400,6 +399,8 @@ dominfo = map( lambda x: server.xend_domain_vcpuinfo(sxp.child_value(x, 'name')), doms) + + print 'Name ID VCPU CPU State Time(s) CPU Affinity' for dom in dominfo: def get_info(n): @@ -625,6 +626,8 @@ server.xend_domain_cpu_sedf_set(dom, *v) def xm_info(args): + arg_check(args, "info", 0) + from xen.xend.XendClient import server info = server.xend_node() @@ -645,9 +648,12 @@ def xm_top(args): + arg_check(args, "top", 0) + os.execvp('xentop', ['xentop']) def xm_dmesg(args): + arg_check(args, "dmesg", 0) gopts = Opts(use="""[-c|--clear] diff -r 2884f14895d2 -r f5ab28885c75 tools/tests/test_x86_emulator.c --- a/tools/tests/test_x86_emulator.c Fri Jan 13 16:38:44 2006 +++ b/tools/tests/test_x86_emulator.c Fri Jan 13 20:12:24 2006 @@ -92,7 +92,7 @@ regs.ecx = 0x12345678; cr2 = (unsigned long)&res; res = 0x7FFFFFFF; - rc = x86_emulate_memop(&regs, cr2, &emulops, 4); + rc = x86_emulate_memop(&regs, cr2, &emulops, X86EMUL_MODE_PROT32); if ( (rc != 0) || (res != 0x92345677) || (regs.eflags != 0xa94) || @@ -110,7 +110,7 @@ regs.ecx = 0x12345678UL; #endif cr2 = (unsigned long)&res; - rc = x86_emulate_memop(&regs, cr2, &emulops, 4); + rc = x86_emulate_memop(&regs, cr2, &emulops, X86EMUL_MODE_PROT32); if ( (rc != 0) || (res != 0x92345677) || (regs.ecx != 0x8000000FUL) || @@ -125,7 +125,7 @@ regs.eax = 0x92345677UL; regs.ecx = 0xAA; cr2 = (unsigned long)&res; - rc = x86_emulate_memop(&regs, cr2, &emulops, 4); + rc = x86_emulate_memop(&regs, cr2, &emulops, X86EMUL_MODE_PROT32); if ( (rc != 0) || (res != 0x923456AA) || (regs.eflags != 0x244) || @@ -141,7 +141,7 @@ regs.eax = 0xAABBCC77UL; regs.ecx = 0xFF; cr2 = (unsigned long)&res; - rc = x86_emulate_memop(&regs, cr2, &emulops, 4); + rc = x86_emulate_memop(&regs, cr2, &emulops, X86EMUL_MODE_PROT32); if ( (rc != 0) || (res != 0x923456AA) || ((regs.eflags&0x240) != 0x200) || @@ -157,7 +157,7 @@ regs.eip = (unsigned long)&instr[0]; regs.ecx = 0x12345678; cr2 = (unsigned long)&res; - rc = x86_emulate_memop(&regs, cr2, &emulops, 4); + rc = x86_emulate_memop(&regs, cr2, &emulops, X86EMUL_MODE_PROT32); if ( (rc != 0) || (res != 0x12345678) || (regs.eflags != 0x200) || @@ -174,7 +174,7 @@ regs.eax = 0x923456AAUL; regs.ecx = 0xDDEEFF00L; cr2 = (unsigned long)&res; - rc = x86_emulate_memop(&regs, cr2, &emulops, 4); + rc = x86_emulate_memop(&regs, cr2, &emulops, X86EMUL_MODE_PROT32); if ( (rc != 0) || (res != 0xDDEEFF00) || (regs.eflags != 0x244) || @@ -193,7 +193,7 @@ regs.edi = (unsigned long)&res + 2; regs.error_code = 0; /* read fault */ cr2 = regs.esi; - rc = x86_emulate_memop(&regs, cr2, &emulops, 4); + rc = x86_emulate_memop(&regs, cr2, &emulops, X86EMUL_MODE_PROT32); if ( (rc != 0) || (res != 0x44554455) || (regs.eflags != 0x200) || @@ -211,7 +211,7 @@ regs.eip = (unsigned long)&instr[0]; regs.edi = (unsigned long)&res; cr2 = regs.edi; - rc = x86_emulate_memop(&regs, cr2, &emulops, 4); + rc = x86_emulate_memop(&regs, cr2, &emulops, X86EMUL_MODE_PROT32); if ( (rc != 0) || (res != 0x2233445D) || ((regs.eflags&0x201) != 0x201) || @@ -229,7 +229,7 @@ regs.eip = (unsigned long)&instr[0]; regs.edi = (unsigned long)cmpxchg8b_res; cr2 = regs.edi; - rc = x86_emulate_memop(&regs, cr2, &emulops, 4); + rc = x86_emulate_memop(&regs, cr2, &emulops, X86EMUL_MODE_PROT32); if ( (rc != 0) || (cmpxchg8b_res[0] != 0x9999AAAA) || (cmpxchg8b_res[1] != 0xCCCCFFFF) || @@ -243,7 +243,7 @@ regs.eip = (unsigned long)&instr[0]; regs.edi = (unsigned long)cmpxchg8b_res; cr2 = regs.edi; - rc = x86_emulate_memop(&regs, cr2, &emulops, 4); + rc = x86_emulate_memop(&regs, cr2, &emulops, X86EMUL_MODE_PROT32); if ( (rc != 0) || (cmpxchg8b_res[0] != 0x9999AAAA) || (cmpxchg8b_res[1] != 0xCCCCFFFF) || @@ -260,7 +260,7 @@ regs.ecx = 0x12345678; cr2 = (unsigned long)&res; res = 0x82; - rc = x86_emulate_memop(&regs, cr2, &emulops, 4); + rc = x86_emulate_memop(&regs, cr2, &emulops, X86EMUL_MODE_PROT32); if ( (rc != 0) || (res != 0x82) || (regs.ecx != 0xFFFFFF82) || @@ -275,7 +275,7 @@ regs.ecx = 0x12345678; cr2 = (unsigned long)&res; res = 0x1234aa82; - rc = x86_emulate_memop(&regs, cr2, &emulops, 4); + rc = x86_emulate_memop(&regs, cr2, &emulops, X86EMUL_MODE_PROT32); if ( (rc != 0) || (res != 0x1234aa82) || (regs.ecx != 0xaa82) || diff -r 2884f14895d2 -r f5ab28885c75 tools/vtpm/Makefile --- a/tools/vtpm/Makefile Fri Jan 13 16:38:44 2006 +++ b/tools/vtpm/Makefile Fri Jan 13 20:12:24 2006 @@ -10,6 +10,8 @@ # Emulator tarball name TPM_EMULATOR_TARFILE = tpm_emulator-0.2b.tar.gz + +GMP_HEADER = /usr/include/gmp.h all: build @@ -55,5 +57,12 @@ patch -p1 <../vtpm.patch build_sub: - $(MAKE) -C $(TPM_EMULATOR_DIR) - $(MAKE) -C $(VTPM_DIR) + if [ -e $(GMP_HEADER) ]; then \ + $(MAKE) -C $(VTPM_DIR); \ + if [ "$(BUILD_EMULATOR)" = "y" ]; then \ + $(MAKE) -C $(TPM_EMULATOR_DIR); \ + fi \ + else \ + echo "*** Unable to build VTPMs. libgmp could not be found."; \ + fi + diff -r 2884f14895d2 -r f5ab28885c75 tools/vtpm/Rules.mk --- a/tools/vtpm/Rules.mk Fri Jan 13 16:38:44 2006 +++ b/tools/vtpm/Rules.mk Fri Jan 13 20:12:24 2006 @@ -33,5 +33,7 @@ -include $(DEP_FILES) +BUILD_EMULATOR = n + # Make sure these are just rules .PHONY : all build install clean diff -r 2884f14895d2 -r f5ab28885c75 tools/vtpm_manager/Makefile --- a/tools/vtpm_manager/Makefile Fri Jan 13 16:38:44 2006 +++ b/tools/vtpm_manager/Makefile Fri Jan 13 20:12:24 2006 @@ -4,13 +4,18 @@ include $(XEN_ROOT)/tools/vtpm_manager/Rules.mk SUBDIRS = crypto tcs util manager +OPENSSL_HEADER = /usr/include/openssl/crypto.h all: build build: - @set -e; for subdir in $(SUBDIRS); do \ - $(MAKE) -C $$subdir $@; \ - done + if [ -e $(OPENSSL_HEADER) ]; then \ + @set -e; for subdir in $(SUBDIRS); do \ + $(MAKE) -C $$subdir $@; \ + done; \ + else \ + echo "*** Cannot build vtpm_manager: OpenSSL developement files missing."; \ + fi install: build @set -e; for subdir in $(SUBDIRS); do \ diff -r 2884f14895d2 -r f5ab28885c75 tools/vtpm_manager/manager/vtsp.c --- a/tools/vtpm_manager/manager/vtsp.c Fri Jan 13 16:38:44 2006 +++ b/tools/vtpm_manager/manager/vtsp.c Fri Jan 13 20:12:24 2006 @@ -144,7 +144,10 @@ if (memcmp (&hm, &(auth->HMAC), sizeof(TPM_DIGEST)) == 0) // 0 indicates equality return (TPM_SUCCESS); else { - VTSP_OIAP( hContext, auth); + // If specified, reconnect the OIAP session. + // NOTE: This only works for TCS's that never have a 0 context. + if (hContext) + VTSP_OIAP( hContext, auth); return (TPM_AUTHFAIL); } } @@ -157,6 +160,10 @@ TPMTRYRETURN( TCSP_OIAP(hContext, &auth->AuthHandle, &auth->NonceEven) ); + + memset(&auth->HMAC, 0, sizeof(TPM_DIGEST)); + auth->fContinueAuthSession = FALSE; + goto egress; abort_egress: @@ -195,6 +202,9 @@ BSG_TPM_NONCE, &nonceOddOSAP); Crypto_HMAC(sharedSecretText, sizeof(sharedSecretText), (BYTE *) usageAuth, TPM_DIGEST_SIZE, (BYTE *) sharedSecret); + + memset(&auth->HMAC, 0, sizeof(TPM_DIGEST)); + auth->fContinueAuthSession = FALSE; goto egress; @@ -287,9 +297,6 @@ srkKeyInfo.parms = (BYTE *) &srkRSAkeyInfo; struct pack_buf_t srkText; - - // GenerateAuth new nonceOdd - Crypto_GetRandom(&auth->NonceOdd, sizeof(TPM_NONCE) ); //These values are accurate for an enc(AuthData). struct pack_buf_t encOwnerAuth, encSrkAuth; @@ -383,9 +390,6 @@ BYTE *paramText; // Digest to make Auth. UINT32 paramTextSize; - // Generate HMAC - Crypto_GetRandom(&auth->NonceOdd, sizeof(TPM_NONCE) ); - paramText = (BYTE *) malloc(sizeof(BYTE) * TCPA_MAX_BUFFER_LENGTH); paramTextSize = BSG_PackList(paramText, 1, @@ -504,9 +508,6 @@ newKeyText.data = flatKey; newKeyText.size = flatKeySize; - // GenerateAuth new nonceOdd - Crypto_GetRandom(&auth->NonceOdd, sizeof(TPM_NONCE) ); - // Generate HMAC paramText = (BYTE *) malloc(sizeof(BYTE) * TCPA_MAX_BUFFER_LENGTH); @@ -587,9 +588,6 @@ // Generate Extra TCS Parameters TPM_HANDLE phKeyHMAC; - // Generate HMAC - Crypto_GetRandom(&auth->NonceOdd, sizeof(TPM_NONCE) ); - paramText = (BYTE *) malloc(sizeof(BYTE) * TCPA_MAX_BUFFER_LENGTH); paramTextSize = BSG_PackList(paramText, 1, @@ -675,9 +673,6 @@ struct pack_buf_t clear_data32; BYTE *clear_data_text; UINT32 clear_data_size; - - // Generate HMAC - Crypto_GetRandom(&auth->NonceOdd, sizeof(TPM_NONCE) ); struct pack_buf_t bound_data32 = {bound_data->size, bound_data->bytes}; @@ -787,6 +782,196 @@ return TPM_SUCCESS; } +TPM_RESULT VTSP_Seal(const TCS_CONTEXT_HANDLE hContext, + const TPM_KEY_HANDLE keyHandle, + const TPM_AUTHDATA *sealDataAuth, + const TPM_PCR_COMPOSITE *pcrComp, + const buffer_t *inData, + TPM_STORED_DATA *sealedData, + const TPM_SECRET *osapSharedSecret, + TCS_AUTH *auth) { + + TPM_RESULT status = TPM_SUCCESS; + TPM_COMMAND_CODE command = TPM_ORD_Seal; + + BYTE *paramText; // Digest to make Auth. + UINT32 paramTextSize; + + // Generate PCR_Info Struct from Comp + TPM_PCR_INFO pcrInfo; + UINT32 pcrInfoSize, flatpcrSize; + BYTE flatpcr[3 + // PCR_Select = 3 1 byte banks + sizeof(UINT16) + // 2 byte UINT16 + sizeof(UINT32) + // PCR_Comp = 4 byte UINT32 + 24 * sizeof(TPM_PCRVALUE) ]; // up to 24 PCRs + + if (pcrComp != NULL) { + //printf("\n\tBinding to PCRs: "); + //for(int i = 0 ; i < pcrComp->select.sizeOfSelect ; i++) + //printf("%2.2x", pcrComp->select.pcrSelect[i]); + + memcpy(&pcrInfo.pcrSelection, &pcrComp->select, sizeof(TPM_PCR_SELECTION)); + + flatpcrSize = BSG_Pack(BSG_TPM_PCR_COMPOSITE, (BYTE *) pcrComp, flatpcr); + Crypto_SHA1Full((BYTE *) flatpcr, flatpcrSize, (BYTE *) &(pcrInfo.digestAtRelease)); + memset(&(pcrInfo.digestAtCreation), 0, sizeof(TPM_DIGEST)); + pcrInfoSize = BSG_Pack(BSG_TPM_PCR_INFO, (BYTE *) &pcrInfo, flatpcr); + } else { + //printf("\n\tBinding to no PCRS."); + pcrInfoSize = 0; + } + + // Calculate encUsageAuth + BYTE XORbuffer[sizeof(TPM_SECRET) + sizeof(TPM_NONCE)]; + UINT32 XORbufferSize = sizeof(XORbuffer); + TPM_DIGEST XORKey; + TPM_ENCAUTH encAuth; + + BSG_PackList( XORbuffer, 2, + BSG_TPM_SECRET, osapSharedSecret, + BSG_TPM_NONCE, &auth->NonceEven ); + + Crypto_SHA1Full(XORbuffer, XORbufferSize, (BYTE *) &XORKey); + + int i; + for (i=0; i < TPM_DIGEST_SIZE; i++) + ((BYTE *) &encAuth)[i] = ((BYTE *) &XORKey)[i] ^ ((BYTE *) sealDataAuth)[i]; + + // Generate Extra TCS Parameters + UINT32 inDataSize = buffer_len(inData); + struct pack_buf_t inData_pack = {inDataSize, inData->bytes}; + struct pack_buf_t pcrInfo_pack = {pcrInfoSize, flatpcr}; + + UINT32 sealedDataSize; + BYTE *flatSealedData=NULL; + + paramText = (BYTE *) malloc(sizeof(BYTE) * TCPA_MAX_BUFFER_LENGTH); + + paramTextSize = BSG_PackList(paramText, 4, + BSG_TPM_COMMAND_CODE, &command, + BSG_TPM_ENCAUTH, &encAuth, + BSG_TPM_SIZE32_DATA, &pcrInfo_pack, + BSG_TPM_SIZE32_DATA, &inData_pack); + + TPMTRYRETURN( GenerateAuth( paramText, paramTextSize, + osapSharedSecret, auth) ); + + // Call TCS + TPMTRYRETURN( TCSP_Seal( hContext, + keyHandle, + encAuth, + pcrInfoSize, + flatpcr, + inDataSize, + inData->bytes, + auth, + &sealedDataSize, + &flatSealedData) ); + + // Unpack/return key structure + BSG_Unpack( BSG_TPM_STORED_DATA, flatSealedData, sealedData ); + + paramTextSize = BSG_PackList(paramText, 3, + BSG_TPM_RESULT, &status, + BSG_TPM_COMMAND_CODE, &command, + BSG_TPM_STORED_DATA, sealedData); + + TPMTRYRETURN( VerifyAuth( paramText, paramTextSize, + osapSharedSecret, auth, + 0) ); + + + goto egress; + + abort_egress: + egress: + + if (flatSealedData) + TCS_FreeMemory( hContext, flatSealedData); + + free(paramText); + return status; +} + + +TPM_RESULT VTSP_Unseal(const TCS_CONTEXT_HANDLE hContext, + const TPM_KEY_HANDLE keyHandle, + const TPM_STORED_DATA *sealedData, + const TPM_AUTHDATA *key_usage_auth, + const TPM_AUTHDATA *data_usage_auth, + buffer_t *outData, + TCS_AUTH *auth, + TCS_AUTH *dataAuth) { + + TPM_RESULT status = TPM_SUCCESS; + TPM_COMMAND_CODE command = TPM_ORD_Unseal; + + BYTE *paramText; // Digest to make Auth. + UINT32 paramTextSize; + + // Generate Extra TCS Parameters + UINT32 sealDataSize, clearDataSize; + BYTE *flatSealedData= (BYTE *) malloc(sizeof(TPM_VERSION) + + 2 * sizeof(UINT32) + + sealedData->sealInfoSize + + sealedData->encDataSize), + *clearData=NULL; + + sealDataSize = BSG_Pack(BSG_TPM_STORED_DATA, sealedData, flatSealedData ); + + paramText = (BYTE *) malloc(sizeof(BYTE) * TCPA_MAX_BUFFER_LENGTH); + + paramTextSize = BSG_PackList(paramText, 2, + BSG_TPM_COMMAND_CODE, &command, + BSG_TPM_STORED_DATA, sealedData); + + TPMTRYRETURN( GenerateAuth( paramText, paramTextSize, + key_usage_auth, auth) ); + + TPMTRYRETURN( GenerateAuth( paramText, paramTextSize, + data_usage_auth, dataAuth) ); + // Call TCS + TPMTRYRETURN( TCSP_Unseal( hContext, + keyHandle, + sealDataSize, + flatSealedData, + auth, + dataAuth, + &clearDataSize, + &clearData) ); + + // Verify Auth + struct pack_buf_t clearData_pack = {clearDataSize, clearData}; + + paramTextSize = BSG_PackList(paramText, 3, + BSG_TPM_RESULT, &status, + BSG_TPM_COMMAND_CODE, &command, + BSG_TPM_SIZE32_DATA, &clearData_pack); + + TPMTRYRETURN( VerifyAuth( paramText, paramTextSize, + key_usage_auth, auth, + hContext) ); + + TPMTRYRETURN( VerifyAuth( paramText, paramTextSize, + data_usage_auth, dataAuth, + hContext) ); + + // Unpack/return key structure + TPMTRYRETURN( buffer_init(outData, clearDataSize, clearData) ); + + goto egress; + + abort_egress: + egress: + + if (flatSealedData) + TCS_FreeMemory( hContext, clearData); + + free(paramText); + return status; +} + + // Function Reaches into unsupported TCS command, beware. TPM_RESULT VTSP_RawTransmit(const TCS_CONTEXT_HANDLE hContext, const buffer_t *inbuf, diff -r 2884f14895d2 -r f5ab28885c75 tools/vtpm_manager/manager/vtsp.h --- a/tools/vtpm_manager/manager/vtsp.h Fri Jan 13 16:38:44 2006 +++ b/tools/vtpm_manager/manager/vtsp.h Fri Jan 13 20:12:24 2006 @@ -100,4 +100,22 @@ const buffer_t *inData, buffer_t *outData); +TPM_RESULT VTSP_Seal(const TCS_CONTEXT_HANDLE hContext, + const TPM_KEY_HANDLE keyHandle, + const TPM_AUTHDATA *sealDataAuth, + const TPM_PCR_COMPOSITE *pcrComp, + const buffer_t *inData, + TPM_STORED_DATA *sealedData, + const TPM_SECRET *osapSharedSecret, + TCS_AUTH *auth); + +TPM_RESULT VTSP_Unseal(const TCS_CONTEXT_HANDLE hContext, + const TPM_KEY_HANDLE keyHandle, + const TPM_STORED_DATA *sealedData, + const TPM_AUTHDATA *key_usage_auth, + const TPM_AUTHDATA *data_usage_auth, + buffer_t *outData, + TCS_AUTH *auth, + TCS_AUTH *dataAuth); + #endif //_VTSP_H_ diff -r 2884f14895d2 -r f5ab28885c75 tools/vtpm_manager/tcs/tcs.c --- a/tools/vtpm_manager/tcs/tcs.c Fri Jan 13 16:38:44 2006 +++ b/tools/vtpm_manager/tcs/tcs.c Fri Jan 13 20:12:24 2006 @@ -636,7 +636,7 @@ TDDL_UINT32 OutLength = TCPA_MAX_BUFFER_LENGTH; // check input params - if (inData == NULL || pubAuth == NULL || SealedDataSize == NULL || *SealedData == NULL) + if (inData == NULL || pubAuth == NULL || SealedDataSize == NULL || SealedData == NULL) return TPM_BAD_PARAMETER; // Convert Byte Input parameter in the input byte stream InBuf diff -r 2884f14895d2 -r f5ab28885c75 tools/xcutils/xc_save.c --- a/tools/xcutils/xc_save.c Fri Jan 13 16:38:44 2006 +++ b/tools/xcutils/xc_save.c Fri Jan 13 20:12:24 2006 @@ -10,9 +10,27 @@ #include <err.h> #include <stdlib.h> #include <stdint.h> +#include <string.h> #include <stdio.h> #include <xenguest.h> + + +/** + * Issue a suspend request through stdout, and receive the acknowledgement + * from stdin. This is handled by XendCheckpoint in the Python layer. + */ +static int suspend(int domid) +{ + char ans[30]; + + printf("suspend\n"); + fflush(stdout); + + return (fgets(ans, sizeof(ans), stdin) != NULL && + !strncmp(ans, "done\n", 5)); +} + int main(int argc, char **argv) @@ -29,5 +47,5 @@ max_f = atoi(argv[5]); flags = atoi(argv[6]); - return xc_linux_save(xc_fd, io_fd, domid, maxit, max_f, flags); + return xc_linux_save(xc_fd, io_fd, domid, maxit, max_f, flags, &suspend); } diff -r 2884f14895d2 -r f5ab28885c75 tools/xenmon/Makefile --- a/tools/xenmon/Makefile Fri Jan 13 16:38:44 2006 +++ b/tools/xenmon/Makefile Fri Jan 13 20:12:24 2006 @@ -13,12 +13,9 @@ INSTALL = install INSTALL_PROG = $(INSTALL) -m0755 INSTALL_DIR = $(INSTALL) -d -m0755 -INSTALL_DATA = $(INSTALL) -m064 +INSTALL_DATA = $(INSTALL) -m0644 -prefix=/usr/local -mandir=$(prefix)/share/man -man1dir=$(mandir)/man1 -sbindir=$(prefix)/sbin +sbindir=/usr/sbin XEN_ROOT=../.. include $(XEN_ROOT)/tools/Rules.mk diff -r 2884f14895d2 -r f5ab28885c75 tools/xenstat/libxenstat/src/xenstat.c --- a/tools/xenstat/libxenstat/src/xenstat.c Fri Jan 13 16:38:44 2006 +++ b/tools/xenstat/libxenstat/src/xenstat.c Fri Jan 13 20:12:24 2006 @@ -705,7 +705,7 @@ snprintf(path, sizeof(path),"/local/domain/%i/name", domain_id); - name = xs_read(handle->xshandle, NULL, path, NULL); + name = xs_read(handle->xshandle, XBT_NULL, path, NULL); if (name == NULL) name = strdup(" "); diff -r 2884f14895d2 -r f5ab28885c75 tools/xenstore/xenstore_client.c --- a/tools/xenstore/xenstore_client.c Fri Jan 13 16:38:44 2006 +++ b/tools/xenstore/xenstore_client.c Fri Jan 13 20:12:24 2006 @@ -66,7 +66,7 @@ #if defined(CLIENT_rm) static int -do_rm(char *path, struct xs_handle *xsh, struct xs_transaction_handle *xth) +do_rm(char *path, struct xs_handle *xsh, xs_transaction_t xth) { if (xs_rm(xsh, xth, path)) { return 0; @@ -81,7 +81,7 @@ static int perform(int optind, int argc, char **argv, struct xs_handle *xsh, - struct xs_transaction_handle *xth, int prefix, int tidy) + xs_transaction_t xth, int prefix, int tidy) { while (optind < argc) { #if defined(CLIENT_read) @@ -179,7 +179,7 @@ main(int argc, char **argv) { struct xs_handle *xsh; - struct xs_transaction_handle *xth; + xs_transaction_t xth; int ret = 0, socket = 0; int prefix = 0; int tidy = 0; @@ -243,7 +243,7 @@ again: xth = xs_transaction_start(xsh); - if (xth == NULL) + if (xth == XBT_NULL) errx(1, "couldn't start transaction"); ret = perform(optind, argc, argv, xsh, xth, prefix, tidy); diff -r 2884f14895d2 -r f5ab28885c75 tools/xenstore/xs.c --- a/tools/xenstore/xs.c Fri Jan 13 16:38:44 2006 +++ b/tools/xenstore/xs.c Fri Jan 13 20:12:24 2006 @@ -292,7 +292,7 @@ } /* Send message to xs, get malloc'ed reply. NULL and set errno on error. */ -static void *xs_talkv(struct xs_handle *h, struct xs_transaction_handle *t, +static void *xs_talkv(struct xs_handle *h, xs_transaction_t t, enum xsd_sockmsg_type type, const struct iovec *iovec, unsigned int num_vecs, @@ -304,7 +304,7 @@ unsigned int i; struct sigaction ignorepipe, oldact; - msg.tx_id = (uint32_t)(unsigned long)t; + msg.tx_id = t; msg.req_id = 0; msg.type = type; msg.len = 0; @@ -368,7 +368,7 @@ } /* Simplified version of xs_talkv: single message. */ -static void *xs_single(struct xs_handle *h, struct xs_transaction_handle *t, +static void *xs_single(struct xs_handle *h, xs_transaction_t t, enum xsd_sockmsg_type type, const char *string, unsigned int *len) @@ -388,7 +388,7 @@ return true; } -char **xs_directory(struct xs_handle *h, struct xs_transaction_handle *t, +char **xs_directory(struct xs_handle *h, xs_transaction_t t, const char *path, unsigned int *num) { char *strings, *p, **ret; @@ -420,7 +420,7 @@ * Returns a malloced value: call free() on it after use. * len indicates length in bytes, not including the nul. */ -void *xs_read(struct xs_handle *h, struct xs_transaction_handle *t, +void *xs_read(struct xs_handle *h, xs_transaction_t t, const char *path, unsigned int *len) { return xs_single(h, t, XS_READ, path, len); @@ -429,7 +429,7 @@ /* Write the value of a single file. * Returns false on failure. */ -bool xs_write(struct xs_handle *h, struct xs_transaction_handle *t, +bool xs_write(struct xs_handle *h, xs_transaction_t t, const char *path, const void *data, unsigned int len) { struct iovec iovec[2]; @@ -446,7 +446,7 @@ /* Create a new directory. * Returns false on failure, or success if it already exists. */ -bool xs_mkdir(struct xs_handle *h, struct xs_transaction_handle *t, +bool xs_mkdir(struct xs_handle *h, xs_transaction_t t, const char *path) { return xs_bool(xs_single(h, t, XS_MKDIR, path, NULL)); @@ -455,7 +455,7 @@ /* Destroy a file or directory (directories must be empty). * Returns false on failure, or success if it doesn't exist. */ -bool xs_rm(struct xs_handle *h, struct xs_transaction_handle *t, +bool xs_rm(struct xs_handle *h, xs_transaction_t t, const char *path) { return xs_bool(xs_single(h, t, XS_RM, path, NULL)); @@ -465,7 +465,7 @@ * Returns malloced array, or NULL: call free() after use. */ struct xs_permissions *xs_get_permissions(struct xs_handle *h, - struct xs_transaction_handle *t, + xs_transaction_t t, const char *path, unsigned int *num) { char *strings; @@ -499,7 +499,7 @@ * Returns false on failure. */ bool xs_set_permissions(struct xs_handle *h, - struct xs_transaction_handle *t, + xs_transaction_t t, const char *path, struct xs_permissions *perms, unsigned int num_perms) @@ -560,7 +560,7 @@ iov[1].iov_base = (void *)token; iov[1].iov_len = strlen(token) + 1; - return xs_bool(xs_talkv(h, NULL, XS_WATCH, iov, + return xs_bool(xs_talkv(h, XBT_NULL, XS_WATCH, iov, ARRAY_SIZE(iov), NULL)); } @@ -627,28 +627,28 @@ iov[1].iov_base = (char *)token; iov[1].iov_len = strlen(token) + 1; - return xs_bool(xs_talkv(h, NULL, XS_UNWATCH, iov, + return xs_bool(xs_talkv(h, XBT_NULL, XS_UNWATCH, iov, ARRAY_SIZE(iov), NULL)); } /* Start a transaction: changes by others will not be seen during this * transaction, and changes will not be visible to others until end. * You can only have one transaction at any time. - * Returns NULL on failure. - */ -struct xs_transaction_handle *xs_transaction_start(struct xs_handle *h) + * Returns XBT_NULL on failure. + */ +xs_transaction_t xs_transaction_start(struct xs_handle *h) { char *id_str; - unsigned long id; - - id_str = xs_single(h, NULL, XS_TRANSACTION_START, "", NULL); + xs_transaction_t id; + + id_str = xs_single(h, XBT_NULL, XS_TRANSACTION_START, "", NULL); if (id_str == NULL) - return NULL; + return XBT_NULL; id = strtoul(id_str, NULL, 0); free(id_str); - return (struct xs_transaction_handle *)id; + return id; } /* End a transaction. @@ -656,7 +656,7 @@ * Returns false on failure, which indicates an error: transactions will * not fail spuriously. */ -bool xs_transaction_end(struct xs_handle *h, struct xs_transaction_handle *t, +bool xs_transaction_end(struct xs_handle *h, xs_transaction_t t, bool abort) { char abortstr[2]; @@ -693,7 +693,7 @@ iov[2].iov_base = eventchn_str; iov[2].iov_len = strlen(eventchn_str) + 1; - return xs_bool(xs_talkv(h, NULL, XS_INTRODUCE, iov, + return xs_bool(xs_talkv(h, XBT_NULL, XS_INTRODUCE, iov, ARRAY_SIZE(iov), NULL)); } @@ -705,7 +705,7 @@ sprintf(domid_str, "%u", domid); - return xs_single(h, NULL, type, domid_str, NULL); + return xs_single(h, XBT_NULL, type, domid_str, NULL); } bool xs_release_domain(struct xs_handle *h, unsigned int domid) @@ -719,7 +719,7 @@ sprintf(domid_str, "%u", domid); - return xs_single(h, NULL, XS_GET_DOMAIN_PATH, domid_str, NULL); + return xs_single(h, XBT_NULL, XS_GET_DOMAIN_PATH, domid_str, NULL); } bool xs_is_domain_introduced(struct xs_handle *h, unsigned int domid) @@ -739,7 +739,7 @@ iov[1].iov_base = data; iov[1].iov_len = len; - return xs_talkv(h, NULL, XS_DEBUG, iov, + return xs_talkv(h, XBT_NULL, XS_DEBUG, iov, ARRAY_SIZE(iov), NULL); } diff -r 2884f14895d2 -r f5ab28885c75 tools/xenstore/xs.h --- a/tools/xenstore/xs.h Fri Jan 13 16:38:44 2006 +++ b/tools/xenstore/xs.h Fri Jan 13 20:12:24 2006 @@ -22,8 +22,10 @@ #include <xs_lib.h> +#define XBT_NULL 0 + struct xs_handle; -struct xs_transaction_handle; +typedef uint32_t xs_transaction_t; /* On failure, these routines set errno. */ @@ -45,45 +47,45 @@ * Returns a malloced array: call free() on it after use. * Num indicates size. */ -char **xs_directory(struct xs_handle *h, struct xs_transaction_handle *t, +char **xs_directory(struct xs_handle *h, xs_transaction_t t, const char *path, unsigned int *num); /* Get the value of a single file, nul terminated. * Returns a malloced value: call free() on it after use. * len indicates length in bytes, not including terminator. */ -void *xs_read(struct xs_handle *h, struct xs_transaction_handle *t, +void *xs_read(struct xs_handle *h, xs_transaction_t t, const char *path, unsigned int *len); /* Write the value of a single file. * Returns false on failure. */ -bool xs_write(struct xs_handle *h, struct xs_transaction_handle *t, +bool xs_write(struct xs_handle *h, xs_transaction_t t, const char *path, const void *data, unsigned int len); /* Create a new directory. * Returns false on failure, or success if it already exists. */ -bool xs_mkdir(struct xs_handle *h, struct xs_transaction_handle *t, +bool xs_mkdir(struct xs_handle *h, xs_transaction_t t, const char *path); /* Destroy a file or directory (and children). - * Returns false on failure, or success if it doesn't exist. + * Returns false on failure, or if it doesn't exist. */ -bool xs_rm(struct xs_handle *h, struct xs_transaction_handle *t, +bool xs_rm(struct xs_handle *h, xs_transaction_t t, const char *path); /* Get permissions of node (first element is owner, first perms is "other"). * Returns malloced array, or NULL: call free() after use. */ struct xs_permissions *xs_get_permissions(struct xs_handle *h, - struct xs_transaction_handle *t, + xs_transaction_t t, const char *path, unsigned int *num); /* Set permissions of node (must be owner). * Returns false on failure. */ -bool xs_set_permissions(struct xs_handle *h, struct xs_transaction_handle *t, +bool xs_set_permissions(struct xs_handle *h, xs_transaction_t t, const char *path, struct xs_permissions *perms, unsigned int num_perms); @@ -113,14 +115,14 @@ * You can only have one transaction at any time. * Returns NULL on failure. */ -struct xs_transaction_handle *xs_transaction_start(struct xs_handle *h); +xs_transaction_t xs_transaction_start(struct xs_handle *h); /* End a transaction. * If abandon is true, transaction is discarded instead of committed. * Returns false on failure: if errno == EAGAIN, you have to restart * transaction. */ -bool xs_transaction_end(struct xs_handle *h, struct xs_transaction_handle *t, +bool xs_transaction_end(struct xs_handle *h, xs_transaction_t t, bool abort); /* Introduce a new domain. diff -r 2884f14895d2 -r f5ab28885c75 tools/xenstore/xs_test.c --- a/tools/xenstore/xs_test.c Fri Jan 13 16:38:44 2006 +++ b/tools/xenstore/xs_test.c Fri Jan 13 20:12:24 2006 @@ -37,12 +37,13 @@ #include <sys/time.h> #include "utils.h" #include "xs_lib.h" +#include "xs.h" #include "list.h" #define XSTEST static struct xs_handle *handles[10] = { NULL }; -static struct xs_transaction_handle *txh[10] = { NULL }; +static xs_transaction_t txh[10] = { XBT_NULL }; static unsigned int timeout_ms = 500; static bool timeout_suppressed = true; @@ -492,7 +493,7 @@ static void do_start(unsigned int handle) { txh[handle] = xs_transaction_start(handles[handle]); - if (txh[handle] == NULL) + if (txh[handle] == XBT_NULL) failed(handle); } @@ -500,7 +501,7 @@ { if (!xs_transaction_end(handles[handle], txh[handle], abort)) failed(handle); - txh[handle] = NULL; + txh[handle] = XBT_NULL; } static void do_introduce(unsigned int handle, @@ -534,7 +535,7 @@ *(uint16_t *)((void *)interface + 36) = atoi(eventchn); if (!xs_introduce_domain(handles[handle], atoi(domid), - atol(mfn), atoi(eventchn), path)) { + atol(mfn), atoi(eventchn))) { failed(handle); munmap(interface, getpagesize()); return; @@ -718,7 +719,7 @@ else if (streq(command, "close")) { xs_daemon_close(handles[handle]); handles[handle] = NULL; - txh[handle] = NULL; + txh[handle] = XBT_NULL; } else if (streq(command, "start")) do_start(handle); else if (streq(command, "commit")) diff -r 2884f14895d2 -r f5ab28885c75 tools/xenstore/xsls.c --- a/tools/xenstore/xsls.c Fri Jan 13 16:38:44 2006 +++ b/tools/xenstore/xsls.c Fri Jan 13 20:12:24 2006 @@ -11,7 +11,7 @@ int i; unsigned int num, len; - e = xs_directory(h, NULL, path, &num); + e = xs_directory(h, XBT_NULL, path, &num); if (e == NULL) err(1, "xs_directory (%s)", path); @@ -22,7 +22,7 @@ sprintf(newpath, "%s%s%s", path, path[strlen(path)-1] == '/' ? "" : "/", e[i]); - val = xs_read(h, NULL, newpath, &len); + val = xs_read(h, XBT_NULL, newpath, &len); if (val == NULL) printf(":\n"); else if ((unsigned)len > (151 - strlen(e[i]))) diff -r 2884f14895d2 -r f5ab28885c75 tools/xentrace/xenctx.c --- a/tools/xentrace/xenctx.c Fri Jan 13 16:38:44 2006 +++ b/tools/xentrace/xenctx.c Fri Jan 13 20:12:24 2006 @@ -380,10 +380,10 @@ exit(-1); } - ret = xc_domain_get_vcpu_context(xc_handle, domid, vcpu, &ctx); + ret = xc_vcpu_getcontext(xc_handle, domid, vcpu, &ctx); if (ret < 0) { xc_domain_unpause(xc_handle, domid); - perror("xc_domain_get_vcpu_context"); + perror("xc_vcpu_getcontext"); exit(-1); } diff -r 2884f14895d2 -r f5ab28885c75 tools/xm-test/lib/XmTestLib/XenDomain.py --- a/tools/xm-test/lib/XmTestLib/XenDomain.py Fri Jan 13 16:38:44 2006 +++ b/tools/xm-test/lib/XmTestLib/XenDomain.py Fri Jan 13 20:12:24 2006 @@ -72,7 +72,8 @@ } VmxDefaults = {"memory" : 64, "vcpus" : 1, - "nics" : 0, + "acpi" : 0, + "apic" : 0, "disk" : ["file:%s/disk.img,ioemu:%s,w" % (getRdPath(), BLOCK_ROOT_DEV)], "kernel" : "/usr/lib/xen/boot/vmxloader", diff -r 2884f14895d2 -r f5ab28885c75 tools/xm-test/tests/block-destroy/06_block-destroy_check_list_pos.py --- a/tools/xm-test/tests/block-destroy/06_block-destroy_check_list_pos.py Fri Jan 13 16:38:44 2006 +++ b/tools/xm-test/tests/block-destroy/06_block-destroy_check_list_pos.py Fri Jan 13 20:12:24 2006 @@ -46,6 +46,8 @@ if not checkXmLongList(domain): FAIL("xm long list does not show that hda1 was attached") +time.sleep(2) + s, o = traceCommand("xm block-detach %s hda1" % domain.getName()) if s != 0: FAIL("block-detach failed") diff -r 2884f14895d2 -r f5ab28885c75 tools/xm-test/tests/network-attach/Makefile.am --- a/tools/xm-test/tests/network-attach/Makefile.am Fri Jan 13 16:38:44 2006 +++ b/tools/xm-test/tests/network-attach/Makefile.am Fri Jan 13 20:12:24 2006 @@ -6,7 +6,7 @@ 03_network_attach_detach_multiple_pos.test \ 04_network_attach_baddomain_neg.test -XFAIL_TESTS = 03_network_attach_detach_multiple_pos.test +XFAIL_TESTS = EXTRA_DIST = $(TESTS) $(XFAIL_TESTS) network_utils.py diff -r 2884f14895d2 -r f5ab28885c75 tools/xm-test/tests/network/02_network_local_ping_pos.py --- a/tools/xm-test/tests/network/02_network_local_ping_pos.py Fri Jan 13 16:38:44 2006 +++ b/tools/xm-test/tests/network/02_network_local_ping_pos.py Fri Jan 13 20:12:24 2006 @@ -50,7 +50,7 @@ console.sendInput("bhs") # Bring up the "lo" interface. - console.runCmd("ifconfig lo up") + console.runCmd("ifconfig lo 127.0.0.1") console.runCmd("ifconfig eth0 inet "+ip+" netmask "+mask+" up") diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/linux-xen/process-linux-xen.c --- a/xen/arch/ia64/linux-xen/process-linux-xen.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/linux-xen/process-linux-xen.c Fri Jan 13 20:12:24 2006 @@ -241,7 +241,7 @@ max_xtp(); local_irq_disable(); - idle_task_exit(); + idle_domain_exit(); ia64_jump_to_sal(&sal_boot_rendez_state[this_cpu]); /* * The above is a point of no-return, the processor is diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/linux-xen/smpboot.c --- a/xen/arch/ia64/linux-xen/smpboot.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/linux-xen/smpboot.c Fri Jan 13 20:12:24 2006 @@ -482,10 +482,8 @@ struct vcpu *v; void *stack; - if ( (idle = do_createdomain(IDLE_DOMAIN_ID, cpu)) == NULL ) - panic("failed 'createdomain' for CPU %d", cpu); - set_bit(_DOMF_idle_domain, &idle->domain_flags); - v = idle->vcpu[0]; + v = idle_vcpu[cpu] = alloc_vcpu(idle_vcpu[0]->domain, cpu, cpu); + BUG_ON(v == NULL); printf ("do_boot_cpu: cpu=%d, domain=%p, vcpu=%p\n", cpu, idle, v); diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/vmx/mmio.c --- a/xen/arch/ia64/vmx/mmio.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/vmx/mmio.c Fri Jan 13 20:12:24 2006 @@ -29,7 +29,7 @@ #include <asm/vmx_vcpu.h> #include <asm/privop.h> #include <asm/types.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> #include <asm/mm.h> #include <asm/vmx.h> diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/vmx/vlsapic.c --- a/xen/arch/ia64/vmx/vlsapic.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/vmx/vlsapic.c Fri Jan 13 20:12:24 2006 @@ -119,7 +119,7 @@ itc_freq = local_cpu_data->itc_freq; vtm->cfg_max_jump=itc_freq*MAX_JUMP_STEP/1000; vtm->cfg_min_grun=itc_freq*MIN_GUEST_RUNNING_TIME/1000; - init_ac_timer(&vtm->vtm_timer, vtm_timer_fn, vcpu, 0); + init_timer(&vtm->vtm_timer, vtm_timer_fn, vcpu, 0); vtm_reset(vcpu); } @@ -163,20 +163,20 @@ local_irq_save(spsr); itv = VCPU(vcpu, itv); if ( ITV_IRQ_MASK(itv) ) - rem_ac_timer(&vtm->vtm_timer); + stop_timer(&vtm->vtm_timer); vtm_interruption_update(vcpu, vtm); local_irq_restore(spsr); } /* - * Update interrupt or hook the vtm ac_timer for fire + * Update interrupt or hook the vtm timer for fire * At this point vtm_timer should be removed if itv is masked. */ /* Interrupt must be disabled at this point */ extern u64 cycle_to_ns(u64 cyle); -#define TIMER_SLOP (50*1000) /* ns */ /* copy from ac_timer.c */ +#define TIMER_SLOP (50*1000) /* ns */ /* copy from timer.c */ void vtm_interruption_update(VCPU *vcpu, vtime_t* vtm) { uint64_t cur_itc,vitm,vitv; @@ -198,7 +198,7 @@ if ( diff_last >= 0 ) { // interrupt already fired. - rem_ac_timer(&vtm->vtm_timer); + stop_timer(&vtm->vtm_timer); } else if ( diff_now >= 0 ) { // ITV is fired. @@ -207,30 +207,30 @@ /* Both last_itc & cur_itc < itm, wait for fire condition */ else { expires = NOW() + cycle_to_ns(0-diff_now) + TIMER_SLOP; - set_ac_timer(&vtm->vtm_timer, expires); + set_timer(&vtm->vtm_timer, expires); } local_irq_restore(spsr); } /* * Action for vtm when the domain is scheduled out. - * Remove the ac_timer for vtm. + * Remove the timer for vtm. */ void vtm_domain_out(VCPU *vcpu) { - if(!is_idle_task(vcpu->domain)) - rem_ac_timer(&vcpu->arch.arch_vmx.vtm.vtm_timer); + if(!is_idle_domain(vcpu->domain)) + stop_timer(&vcpu->arch.arch_vmx.vtm.vtm_timer); } /* * Action for vtm when the domain is scheduled in. - * Fire vtm IRQ or add the ac_timer for vtm. + * Fire vtm IRQ or add the timer for vtm. */ void vtm_domain_in(VCPU *vcpu) { vtime_t *vtm; - if(!is_idle_task(vcpu->domain)) { + if(!is_idle_domain(vcpu->domain)) { vtm=&(vcpu->arch.arch_vmx.vtm); vtm_interruption_update(vcpu, vtm); } diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/vmx/vmx_init.c --- a/xen/arch/ia64/vmx/vmx_init.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/vmx/vmx_init.c Fri Jan 13 20:12:24 2006 @@ -42,7 +42,7 @@ #include <xen/lib.h> #include <asm/vmmu.h> #include <public/arch-ia64.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> #include <asm/vmx_phy_mode.h> #include <asm/processor.h> #include <asm/vmx.h> diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/vmx/vmx_process.c --- a/xen/arch/ia64/vmx/vmx_process.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/vmx/vmx_process.c Fri Jan 13 20:12:24 2006 @@ -231,7 +231,7 @@ struct domain *d = current->domain; struct vcpu *v = current; // FIXME: Will this work properly if doing an RFI??? - if (!is_idle_task(d) ) { // always comes from guest + if (!is_idle_domain(d) ) { // always comes from guest extern void vmx_dorfirfi(void); struct pt_regs *user_regs = vcpu_regs(current); if (local_softirq_pending()) diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/vmx/vmx_support.c --- a/xen/arch/ia64/vmx/vmx_support.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/vmx/vmx_support.c Fri Jan 13 20:12:24 2006 @@ -21,7 +21,7 @@ */ #include <xen/config.h> #include <xen/sched.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> #include <asm/vmx.h> #include <asm/vmx_vcpu.h> diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/xen/domain.c --- a/xen/arch/ia64/xen/domain.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/xen/domain.c Fri Jan 13 20:12:24 2006 @@ -46,7 +46,7 @@ #include <asm/vmx_vcpu.h> #include <asm/vmx_vpd.h> #include <asm/pal.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> #define CONFIG_DOMAIN0_CONTIGUOUS unsigned long dom0_start = -1L; @@ -87,7 +87,6 @@ int cpu = smp_processor_id(); for ( ; ; ) { - printf ("idle%dD\n", cpu); #ifdef IA64 // __IRQ_STAT(cpu, idle_timestamp) = jiffies #else @@ -146,15 +145,26 @@ { struct vcpu *v; + /* Still keep idle vcpu0 static allocated at compilation, due + * to some code from Linux still requires it in early phase. + */ + if (is_idle_domain(d) && !vcpu_id) + return idle_vcpu[0]; + if ((v = alloc_xenheap_pages(KERNEL_STACK_SIZE_ORDER)) == NULL) return NULL; memset(v, 0, sizeof(*v)); - memcpy(&v->arch, &idle0_vcpu.arch, sizeof(v->arch)); - v->arch.privregs = + memcpy(&v->arch, &idle_vcpu[0]->arch, sizeof(v->arch)); + + if (!is_idle_domain(d)) { + v->arch.privregs = alloc_xenheap_pages(get_order(sizeof(mapped_regs_t))); + BUG_ON(v->arch.privregs == NULL); + memset(v->arch.privregs, 0, PAGE_SIZE); + } + printf("arch_vcpu_info=%p\n", v->arch.privregs); - memset(v->arch.privregs, 0, PAGE_SIZE); return v; } @@ -191,6 +201,14 @@ memset(ti, 0, sizeof(struct thread_info)); init_switch_stack(v); + // the following will eventually need to be negotiated dynamically + d->xen_vastart = XEN_START_ADDR; + d->xen_vaend = XEN_END_ADDR; + d->shared_info_va = SHAREDINFO_ADDR; + + if (is_idle_vcpu(v)) + return 0; + d->shared_info = (void *)alloc_xenheap_page(); if (!d->shared_info) { printk("ERROR/HALTING: CAN'T ALLOC PAGE\n"); @@ -200,12 +218,7 @@ if (v == d->vcpu[0]) memset(&d->shared_info->evtchn_mask[0], 0xff, sizeof(d->shared_info->evtchn_mask)); -#if 0 - d->vcpu[0].arch.privregs = - alloc_xenheap_pages(get_order(sizeof(mapped_regs_t))); - printf("arch_vcpu_info=%p\n", d->vcpu[0].arch.privregs); - memset(d->vcpu.arch.privregs, 0, PAGE_SIZE); -#endif + v->vcpu_info = &(d->shared_info->vcpu_info[0]); d->max_pages = (128UL*1024*1024)/PAGE_SIZE; // 128MB default // FIXME @@ -227,28 +240,21 @@ BUG(); v->arch.starting_rid = d->arch.starting_rid; v->arch.ending_rid = d->arch.ending_rid; - // the following will eventually need to be negotiated dynamically - d->xen_vastart = XEN_START_ADDR; - d->xen_vaend = XEN_END_ADDR; - d->shared_info_va = SHAREDINFO_ADDR; d->arch.breakimm = 0x1000; v->arch.breakimm = d->arch.breakimm; d->arch.sys_pgnr = 0; - if (d->domain_id != IDLE_DOMAIN_ID) { - d->arch.mm = xmalloc(struct mm_struct); - if (unlikely(!d->arch.mm)) { - printk("Can't allocate mm_struct for domain %d\n",d->domain_id); - return -ENOMEM; - } - memset(d->arch.mm, 0, sizeof(*d->arch.mm)); - d->arch.mm->pgd = pgd_alloc(d->arch.mm); - if (unlikely(!d->arch.mm->pgd)) { - printk("Can't allocate pgd for domain %d\n",d->domain_id); - return -ENOMEM; - } - } else - d->arch.mm = NULL; + d->arch.mm = xmalloc(struct mm_struct); + if (unlikely(!d->arch.mm)) { + printk("Can't allocate mm_struct for domain %d\n",d->domain_id); + return -ENOMEM; + } + memset(d->arch.mm, 0, sizeof(*d->arch.mm)); + d->arch.mm->pgd = pgd_alloc(d->arch.mm); + if (unlikely(!d->arch.mm->pgd)) { + printk("Can't allocate pgd for domain %d\n",d->domain_id); + return -ENOMEM; + } printf ("arch_do_create_domain: domain=%p\n", d); return 0; @@ -1070,15 +1076,6 @@ vcpu_pend_interrupt(dom0->vcpu[0],irq); } -void vcpu_migrate_cpu(struct vcpu *v, int newcpu) -{ - if ( v->processor == newcpu ) - return; - - set_bit(_VCPUF_cpu_migrated, &v->vcpu_flags); - v->processor = newcpu; -} - void sync_vcpu_execstate(struct vcpu *v) { ia64_save_fpu(v->arch._thread.fph); diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/xen/idle0_task.c --- a/xen/arch/ia64/xen/idle0_task.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/xen/idle0_task.c Fri Jan 13 20:12:24 2006 @@ -11,30 +11,14 @@ .mmlist = LIST_HEAD_INIT(name.mmlist), \ } -#define IDLE0_EXEC_DOMAIN(_ed,_d) \ +#define IDLE_VCPU(_v) \ { \ processor: 0, \ - mm: 0, \ - thread: INIT_THREAD, \ - domain: (_d) \ -} - -#define IDLE0_DOMAIN(_t) \ -{ \ - domain_id: IDLE_DOMAIN_ID, \ - domain_flags:DOMF_idle_domain, \ - refcnt: ATOMIC_INIT(1) \ + domain: 0 \ } struct mm_struct init_mm = INIT_MM(init_mm); EXPORT_SYMBOL(init_mm); - -struct domain idle0_domain = IDLE0_DOMAIN(idle0_domain); -#if 0 -struct vcpu idle0_vcpu = IDLE0_EXEC_DOMAIN(idle0_vcpu, - &idle0_domain); -#endif - /* * Initial task structure. @@ -44,15 +28,12 @@ */ union { struct { - struct domain task; + struct vcpu task; } s; unsigned long stack[KERNEL_STACK_SIZE/sizeof (unsigned long)]; -} init_task_mem asm ("init_task") __attribute__((section(".data.init_task"))); -// = {{ - ; -//.task = IDLE0_EXEC_DOMAIN(init_task_mem.s.task,&idle0_domain), -//}; -//}; +} init_task_mem asm ("init_task") __attribute__((section(".data.init_task"))) = {{ + .task = IDLE_VCPU(init_task_mem.s.task) +}}; EXPORT_SYMBOL(init_task); diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/xen/process.c --- a/xen/arch/ia64/xen/process.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/xen/process.c Fri Jan 13 20:12:24 2006 @@ -65,26 +65,16 @@ extern struct schedule_data schedule_data[NR_CPUS]; -void schedule_tail(struct vcpu *next) -{ - unsigned long rr7; - //printk("current=%lx,shared_info=%lx\n",current,current->vcpu_info); - //printk("next=%lx,shared_info=%lx\n",next,next->vcpu_info); - - // TG: Real HACK FIXME. - // This is currently necessary because when a new domain is started, - // the context_switch function of xen/common/schedule.c(__enter_scheduler) - // never returns. Therefore, the lock must be released. - // schedule_tail is only called when a domain is started. - spin_unlock_irq(&schedule_data[current->processor].schedule_lock); - - /* rr7 will be postponed to last point when resuming back to guest */ - if(VMX_DOMAIN(current)){ - vmx_load_all_rr(current); - }else{ - load_region_regs(current); - vcpu_load_kernel_regs(current); - } +void schedule_tail(struct vcpu *prev) +{ + context_saved(prev); + + if (VMX_DOMAIN(current)) { + vmx_load_all_rr(current); + } else { + load_region_regs(current); + vcpu_load_kernel_regs(current); + } } void tdpfoo(void) { } @@ -252,7 +242,7 @@ struct domain *d = current->domain; struct vcpu *v = current; // FIXME: Will this work properly if doing an RFI??? - if (!is_idle_task(d) && user_mode(regs)) { + if (!is_idle_domain(d) && user_mode(regs)) { //vcpu_poke_timer(v); if (vcpu_deliverable_interrupts(v)) reflect_extint(regs); diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/xen/vcpu.c --- a/xen/arch/ia64/xen/vcpu.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/xen/vcpu.c Fri Jan 13 20:12:24 2006 @@ -1085,7 +1085,7 @@ /* gloss over the wraparound problem for now... we know it exists * but it doesn't matter right now */ - if (is_idle_task(vcpu->domain)) { + if (is_idle_domain(vcpu->domain)) { // printf("****** vcpu_set_next_timer called during idle!!\n"); vcpu_safe_set_itm(s); return; diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/xen/xenmisc.c --- a/xen/arch/ia64/xen/xenmisc.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/xen/xenmisc.c Fri Jan 13 20:12:24 2006 @@ -75,7 +75,7 @@ void raise_actimer_softirq(void) { - raise_softirq(AC_TIMER_SOFTIRQ); + raise_softirq(TIMER_SOFTIRQ); } unsigned long @@ -320,18 +320,15 @@ ia64_set_iva(&ia64_ivt); ia64_set_pta(VHPT_ADDR | (1 << 8) | (VHPT_SIZE_LOG2 << 2) | VHPT_ENABLED); - if (!is_idle_task(current->domain)) { + if (!is_idle_domain(current->domain)) { load_region_regs(current); vcpu_load_kernel_regs(current); if (vcpu_timer_expired(current)) vcpu_pend_timer(current); } if (vcpu_timer_expired(current)) vcpu_pend_timer(current); } -} - -void context_switch_finalise(struct vcpu *next) -{ - /* nothing to do */ + + context_saved(prev); } void continue_running(struct vcpu *same) diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/xen/xensetup.c --- a/xen/arch/ia64/xen/xensetup.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/xen/xensetup.c Fri Jan 13 20:12:24 2006 @@ -27,7 +27,7 @@ char saved_command_line[COMMAND_LINE_SIZE]; -struct vcpu *idle_task[NR_CPUS] = { &idle0_vcpu }; +struct vcpu *idle_vcpu[NR_CPUS]; cpumask_t cpu_present_map; @@ -157,15 +157,11 @@ unsigned long dom0_memory_start, dom0_memory_size; unsigned long dom0_initrd_start, dom0_initrd_size; unsigned long initial_images_start, initial_images_end; + struct domain *idle_domain; running_on_sim = is_platform_hp_ski(); /* Kernel may be relocated by EFI loader */ xen_pstart = ia64_tpa(KERNEL_START); - - /* Must do this early -- e.g., spinlocks rely on get_current(). */ - //set_current(&idle0_vcpu); - ia64_r13 = (void *)&idle0_vcpu; - idle0_vcpu.domain = &idle0_domain; early_setup_arch(&cmdline); @@ -282,18 +278,22 @@ (xenheap_phys_end-__pa(heap_start)) >> 20, (xenheap_phys_end-__pa(heap_start)) >> 10); +printk("About to call scheduler_init()\n"); + scheduler_init(); + idle_vcpu[0] = (struct vcpu*) ia64_r13; + idle_domain = do_createdomain(IDLE_DOMAIN_ID, 0); + BUG_ON(idle_domain == NULL); + late_setup_arch(&cmdline); setup_per_cpu_areas(); mem_init(); -printk("About to call scheduler_init()\n"); - scheduler_init(); local_irq_disable(); init_IRQ (); printk("About to call init_xen_time()\n"); init_xen_time(); /* initialise the time */ -printk("About to call ac_timer_init()\n"); - ac_timer_init(); +printk("About to call timer_init()\n"); + timer_init(); #ifdef CONFIG_XEN_CONSOLE_INPUT /* CONFIG_SERIAL_8250_CONSOLE=n in dom0! */ initialize_keytable(); @@ -309,13 +309,9 @@ } smp_prepare_cpus(max_cpus); - /* We aren't hotplug-capable yet. */ - //BUG_ON(!cpus_empty(cpu_present_map)); for_each_cpu ( i ) cpu_set(i, cpu_present_map); - - //BUG_ON(!local_irq_is_enabled()); /* Enable IRQ to receive IPI (needed for ITC sync). */ local_irq_enable(); @@ -345,12 +341,7 @@ /* Create initial domain 0. */ printk("About to call do_createdomain()\n"); dom0 = do_createdomain(0, 0); - init_task.domain = &idle0_domain; - init_task.processor = 0; -// init_task.mm = &init_mm; - init_task.domain->arch.mm = &init_mm; -// init_task.thread = INIT_THREAD; - //arch_do_createdomain(current); + #ifdef CLONE_DOMAIN0 { int i; @@ -383,8 +374,7 @@ panic("Could not set up DOM0 guest OS\n"); /* PIN domain0 on CPU 0. */ - dom0->vcpu[0]->cpumap=1; - set_bit(_VCPUF_cpu_pinned, &dom0->vcpu[0]->vcpu_flags); + dom0->vcpu[0]->cpu_affinity = cpumask_of_cpu(0); #ifdef CLONE_DOMAIN0 { @@ -433,8 +423,8 @@ local_irq_enable(); - printf("About to call schedulers_start dom0=%p, idle0_dom=%p\n", - dom0, &idle0_domain); + printf("About to call schedulers_start dom0=%p, idle_dom=%p\n", + dom0, &idle_domain); schedulers_start(); domain_unpause_by_systemcontroller(dom0); diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/ia64/xen/xentime.c --- a/xen/arch/ia64/xen/xentime.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/ia64/xen/xentime.c Fri Jan 13 20:12:24 2006 @@ -127,7 +127,7 @@ vcpu_wake(dom0->vcpu[0]); } } - if (!is_idle_task(current->domain)) { + if (!is_idle_domain(current->domain)) { if (vcpu_timer_expired(current)) { vcpu_pend_timer(current); // ensure another timer interrupt happens even if domain doesn't @@ -196,7 +196,7 @@ //#endif /* double check, in case we got hit by a (slow) PMI: */ } while (time_after_eq(ia64_get_itc(), new_itm)); - raise_softirq(AC_TIMER_SOFTIRQ); + raise_softirq(TIMER_SOFTIRQ); return IRQ_HANDLED; } @@ -235,7 +235,7 @@ return 0; } -int reprogram_ac_timer(s_time_t timeout) +int reprogram_timer(s_time_t timeout) { struct vcpu *v = current; s_time_t expire; diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/apic.c --- a/xen/arch/x86/apic.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/apic.c Fri Jan 13 20:12:24 2006 @@ -870,7 +870,7 @@ * returns 1 on success * returns 0 if the timeout value is too small or in the past. */ -int reprogram_ac_timer(s_time_t timeout) +int reprogram_timer(s_time_t timeout) { s_time_t now; s_time_t expire; @@ -931,7 +931,7 @@ { ack_APIC_irq(); perfc_incrc(apic_timer); - raise_softirq(AC_TIMER_SOFTIRQ); + raise_softirq(TIMER_SOFTIRQ); } /* diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/boot/x86_32.S --- a/xen/arch/x86/boot/x86_32.S Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/boot/x86_32.S Fri Jan 13 20:12:24 2006 @@ -100,7 +100,7 @@ 1: stosl /* low mappings cover as much physmem as possible */ add $4,%edi add $(1<<L2_PAGETABLE_SHIFT),%eax - cmp $__HYPERVISOR_VIRT_START+0xe3,%eax + cmp $HYPERVISOR_VIRT_START+0xe3,%eax jne 1b #else /* Initialize low and high mappings of all memory with 4MB pages */ @@ -113,7 +113,7 @@ jne 1b 1: stosl /* low mappings cover as much physmem as possible */ add $(1<<L2_PAGETABLE_SHIFT),%eax - cmp $__HYPERVISOR_VIRT_START+0xe3,%eax + cmp $HYPERVISOR_VIRT_START+0xe3,%eax jne 1b #endif diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/dm/i8259.c --- a/xen/arch/x86/dm/i8259.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/dm/i8259.c Fri Jan 13 20:12:24 2006 @@ -29,7 +29,7 @@ #include <xen/lib.h> #include <xen/errno.h> #include <xen/sched.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> #include <asm/vmx.h> #include <asm/vmx_vpic.h> #include <asm/current.h> diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/dm/vmx_vioapic.c --- a/xen/arch/x86/dm/vmx_vioapic.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/dm/vmx_vioapic.c Fri Jan 13 20:12:24 2006 @@ -37,7 +37,7 @@ #include <xen/lib.h> #include <xen/errno.h> #include <xen/sched.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> #include <asm/vmx.h> #include <asm/vmx_vpic.h> #include <asm/current.h> diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/dom0_ops.c --- a/xen/arch/x86/dom0_ops.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/dom0_ops.c Fri Jan 13 20:12:24 2006 @@ -36,13 +36,13 @@ static void write_msr_for(void *unused) { - if ( ((1 << current->processor) & msr_cpu_mask) ) + if ( ((1 << smp_processor_id()) & msr_cpu_mask) ) (void)wrmsr_user(msr_addr, msr_lo, msr_hi); } static void read_msr_for(void *unused) { - if ( ((1 << current->processor) & msr_cpu_mask) ) + if ( ((1 << smp_processor_id()) & msr_cpu_mask) ) (void)rdmsr_user(msr_addr, msr_lo, msr_hi); } @@ -103,12 +103,27 @@ op->u.add_memtype.nr_pfns, op->u.add_memtype.type, 1); + if (ret > 0) + { + (void)__put_user(0, &u_dom0_op->u.add_memtype.handle); + (void)__put_user(ret, &u_dom0_op->u.add_memtype.reg); + ret = 0; + } } break; case DOM0_DEL_MEMTYPE: { - ret = mtrr_del_page(op->u.del_memtype.reg, 0, 0); + if (op->u.del_memtype.handle == 0 + /* mtrr/main.c otherwise does a lookup */ + && (int)op->u.del_memtype.reg >= 0) + { + ret = mtrr_del_page(op->u.del_memtype.reg, 0, 0); + if (ret > 0) + ret = 0; + } + else + ret = -EINVAL; } break; @@ -179,7 +194,7 @@ memcpy(pi->hw_cap, boot_cpu_data.x86_capability, NCAPINTS*4); ret = 0; if ( copy_to_user(u_dom0_op, op, sizeof(*op)) ) - ret = -EFAULT; + ret = -EFAULT; } break; diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/domain.c --- a/xen/arch/x86/domain.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/domain.c Fri Jan 13 20:12:24 2006 @@ -46,17 +46,16 @@ struct percpu_ctxt { struct vcpu *curr_vcpu; - unsigned int context_not_finalised; unsigned int dirty_segment_mask; } __cacheline_aligned; static struct percpu_ctxt percpu_ctxt[NR_CPUS]; -static void continue_idle_task(struct vcpu *v) +static void continue_idle_domain(struct vcpu *v) { reset_stack_and_jump(idle_loop); } -static void continue_nonidle_task(struct vcpu *v) +static void continue_nonidle_domain(struct vcpu *v) { reset_stack_and_jump(ret_from_intr); } @@ -92,10 +91,9 @@ { struct vcpu *v = current; - ASSERT(is_idle_task(v->domain)); - percpu_ctxt[smp_processor_id()].curr_vcpu = v; - cpu_set(smp_processor_id(), v->domain->cpumask); - v->arch.schedule_tail = continue_idle_task; + ASSERT(is_idle_vcpu(v)); + cpu_set(smp_processor_id(), v->domain->domain_dirty_cpumask); + cpu_set(smp_processor_id(), v->vcpu_dirty_cpumask); reset_stack_and_jump(idle_loop); } @@ -217,14 +215,20 @@ memset(v, 0, sizeof(*v)); - memcpy(&v->arch, &idle0_vcpu.arch, sizeof(v->arch)); + memcpy(&v->arch, &idle_vcpu[0]->arch, sizeof(v->arch)); v->arch.flags = TF_kernel_mode; + + if ( is_idle_domain(d) ) + { + percpu_ctxt[vcpu_id].curr_vcpu = v; + v->arch.schedule_tail = continue_idle_domain; + } if ( (v->vcpu_id = vcpu_id) != 0 ) { v->arch.schedule_tail = d->vcpu[0]->arch.schedule_tail; v->arch.perdomain_ptes = - d->arch.mm_perdomain_pt + (vcpu_id << PDPT_VCPU_SHIFT); + d->arch.mm_perdomain_pt + (vcpu_id << GDT_LDT_VCPU_SHIFT); } return v; @@ -259,32 +263,11 @@ int i; #endif - if ( is_idle_task(d) ) - return 0; - - d->arch.ioport_caps = - rangeset_new(d, "I/O Ports", RANGESETF_prettyprint_hex); - if ( d->arch.ioport_caps == NULL ) - return -ENOMEM; - - if ( (d->shared_info = alloc_xenheap_page()) == NULL ) - return -ENOMEM; - - if ( (rc = ptwr_init(d)) != 0 ) - { - free_xenheap_page(d->shared_info); - return rc; - } - - v->arch.schedule_tail = continue_nonidle_task; - - memset(d->shared_info, 0, PAGE_SIZE); - v->vcpu_info = &d->shared_info->vcpu_info[v->vcpu_id]; - v->cpumap = CPUMAP_RUNANYWHERE; - SHARE_PFN_WITH_DOMAIN(virt_to_page(d->shared_info), d); - pdpt_order = get_order_from_bytes(PDPT_L1_ENTRIES * sizeof(l1_pgentry_t)); d->arch.mm_perdomain_pt = alloc_xenheap_pages(pdpt_order); + if ( d->arch.mm_perdomain_pt == NULL ) + goto fail_nomem; + memset(d->arch.mm_perdomain_pt, 0, PAGE_SIZE << pdpt_order); v->arch.perdomain_ptes = d->arch.mm_perdomain_pt; @@ -297,49 +280,73 @@ */ gdt_l1e = l1e_from_page(virt_to_page(gdt_table), PAGE_HYPERVISOR); for ( vcpuid = 0; vcpuid < MAX_VIRT_CPUS; vcpuid++ ) - d->arch.mm_perdomain_pt[ - (vcpuid << PDPT_VCPU_SHIFT) + FIRST_RESERVED_GDT_PAGE] = gdt_l1e; + d->arch.mm_perdomain_pt[((vcpuid << GDT_LDT_VCPU_SHIFT) + + FIRST_RESERVED_GDT_PAGE)] = gdt_l1e; v->arch.guest_vtable = __linear_l2_table; v->arch.shadow_vtable = __shadow_linear_l2_table; -#ifdef __x86_64__ +#if defined(__i386__) + + mapcache_init(d); + +#else /* __x86_64__ */ + v->arch.guest_vl3table = __linear_l3_table; v->arch.guest_vl4table = __linear_l4_table; d->arch.mm_perdomain_l2 = alloc_xenheap_page(); + d->arch.mm_perdomain_l3 = alloc_xenheap_page(); + if ( (d->arch.mm_perdomain_l2 == NULL) || + (d->arch.mm_perdomain_l3 == NULL) ) + goto fail_nomem; + memset(d->arch.mm_perdomain_l2, 0, PAGE_SIZE); for ( i = 0; i < (1 << pdpt_order); i++ ) d->arch.mm_perdomain_l2[l2_table_offset(PERDOMAIN_VIRT_START)+i] = l2e_from_page(virt_to_page(d->arch.mm_perdomain_pt)+i, __PAGE_HYPERVISOR); - d->arch.mm_perdomain_l3 = alloc_xenheap_page(); memset(d->arch.mm_perdomain_l3, 0, PAGE_SIZE); d->arch.mm_perdomain_l3[l3_table_offset(PERDOMAIN_VIRT_START)] = l3e_from_page(virt_to_page(d->arch.mm_perdomain_l2), __PAGE_HYPERVISOR); -#endif + +#endif /* __x86_64__ */ shadow_lock_init(d); INIT_LIST_HEAD(&d->arch.free_shadow_frames); + if ( !is_idle_domain(d) ) + { + d->arch.ioport_caps = + rangeset_new(d, "I/O Ports", RANGESETF_prettyprint_hex); + if ( d->arch.ioport_caps == NULL ) + goto fail_nomem; + + if ( (d->shared_info = alloc_xenheap_page()) == NULL ) + goto fail_nomem; + + if ( (rc = ptwr_init(d)) != 0 ) + goto fail_nomem; + + memset(d->shared_info, 0, PAGE_SIZE); + v->vcpu_info = &d->shared_info->vcpu_info[v->vcpu_id]; + SHARE_PFN_WITH_DOMAIN(virt_to_page(d->shared_info), d); + + v->arch.schedule_tail = continue_nonidle_domain; + } + return 0; -} - -void vcpu_migrate_cpu(struct vcpu *v, int newcpu) -{ - if ( v->processor == newcpu ) - return; - - set_bit(_VCPUF_cpu_migrated, &v->vcpu_flags); - v->processor = newcpu; - - if ( VMX_DOMAIN(v) ) - { - __vmpclear(virt_to_phys(v->arch.arch_vmx.vmcs)); - v->arch.schedule_tail = arch_vmx_do_relaunch; - } + + fail_nomem: + free_xenheap_page(d->shared_info); +#ifdef __x86_64__ + free_xenheap_page(d->arch.mm_perdomain_l2); + free_xenheap_page(d->arch.mm_perdomain_l3); +#endif + free_xenheap_pages(d->arch.mm_perdomain_pt, pdpt_order); + return -ENOMEM; } /* This is called by arch_final_setup_guest and do_boot_vcpu */ @@ -472,14 +479,6 @@ #ifdef __x86_64__ - -void toggle_guest_mode(struct vcpu *v) -{ - v->arch.flags ^= TF_kernel_mode; - __asm__ __volatile__ ( "swapgs" ); - update_pagetables(v); - write_ptbase(v); -} #define loadsegment(seg,value) ({ \ int __r = 1; \ @@ -650,35 +649,6 @@ percpu_ctxt[smp_processor_id()].dirty_segment_mask = dirty_segment_mask; } -long do_switch_to_user(void) -{ - struct cpu_user_regs *regs = guest_cpu_user_regs(); - struct switch_to_user stu; - struct vcpu *v = current; - - if ( unlikely(copy_from_user(&stu, (void *)regs->rsp, sizeof(stu))) || - unlikely(pagetable_get_paddr(v->arch.guest_table_user) == 0) ) - return -EFAULT; - - toggle_guest_mode(v); - - regs->rip = stu.rip; - regs->cs = stu.cs | 3; /* force guest privilege */ - regs->rflags = (stu.rflags & ~(EF_IOPL|EF_VM)) | EF_IE; - regs->rsp = stu.rsp; - regs->ss = stu.ss | 3; /* force guest privilege */ - - if ( !(stu.flags & VGCF_IN_SYSCALL) ) - { - regs->entry_vector = 0; - regs->r11 = stu.r11; - regs->rcx = stu.rcx; - } - - /* Saved %rax gets written back to regs->rax in entry.S. */ - return stu.rax; -} - #define switch_kernel_stack(_n,_c) ((void)0) #elif defined(__i386__) @@ -705,7 +675,10 @@ struct vcpu *p = percpu_ctxt[cpu].curr_vcpu; struct vcpu *n = current; - if ( !is_idle_task(p->domain) ) + ASSERT(p != n); + ASSERT(cpus_empty(n->vcpu_dirty_cpumask)); + + if ( !is_idle_vcpu(p) ) { memcpy(&p->arch.guest_context.user_regs, stack_regs, @@ -714,7 +687,7 @@ save_segments(p); } - if ( !is_idle_task(n->domain) ) + if ( !is_idle_vcpu(n) ) { memcpy(stack_regs, &n->arch.guest_context.user_regs, @@ -740,7 +713,8 @@ } if ( p->domain != n->domain ) - cpu_set(cpu, n->domain->cpumask); + cpu_set(cpu, n->domain->domain_dirty_cpumask); + cpu_set(cpu, n->vcpu_dirty_cpumask); write_ptbase(n); @@ -753,7 +727,8 @@ } if ( p->domain != n->domain ) - cpu_clear(cpu, p->domain->cpumask); + cpu_clear(cpu, p->domain->domain_dirty_cpumask); + cpu_clear(cpu, p->vcpu_dirty_cpumask); percpu_ctxt[cpu].curr_vcpu = n; } @@ -762,29 +737,32 @@ void context_switch(struct vcpu *prev, struct vcpu *next) { unsigned int cpu = smp_processor_id(); - - ASSERT(!local_irq_is_enabled()); + cpumask_t dirty_mask = next->vcpu_dirty_cpumask; + + ASSERT(local_irq_is_enabled()); + + /* Allow at most one CPU at a time to be dirty. */ + ASSERT(cpus_weight(dirty_mask) <= 1); + if ( unlikely(!cpu_isset(cpu, dirty_mask) && !cpus_empty(dirty_mask)) ) + { + /* Other cpus call __sync_lazy_execstate from flush ipi handler. */ + flush_tlb_mask(dirty_mask); + } + + local_irq_disable(); set_current(next); - if ( (percpu_ctxt[cpu].curr_vcpu != next) && !is_idle_task(next->domain) ) + if ( (percpu_ctxt[cpu].curr_vcpu == next) || is_idle_vcpu(next) ) + { + local_irq_enable(); + } + else { __context_switch(); - percpu_ctxt[cpu].context_not_finalised = 1; - } -} - -void context_switch_finalise(struct vcpu *next) -{ - unsigned int cpu = smp_processor_id(); - - ASSERT(local_irq_is_enabled()); - - if ( percpu_ctxt[cpu].context_not_finalised ) - { - percpu_ctxt[cpu].context_not_finalised = 0; - - BUG_ON(percpu_ctxt[cpu].curr_vcpu != next); + + /* Re-enable interrupts before restoring state which may fault. */ + local_irq_enable(); if ( VMX_DOMAIN(next) ) { @@ -798,6 +776,8 @@ } } + context_saved(prev); + schedule_tail(next); BUG(); } @@ -827,20 +807,11 @@ void sync_vcpu_execstate(struct vcpu *v) { - unsigned int cpu = v->processor; - - if ( !cpu_isset(cpu, v->domain->cpumask) ) - return; - - if ( cpu == smp_processor_id() ) - { + if ( cpu_isset(smp_processor_id(), v->vcpu_dirty_cpumask) ) (void)__sync_lazy_execstate(); - } - else - { - /* Other cpus call __sync_lazy_execstate from flush ipi handler. */ - flush_tlb_mask(cpumask_of_cpu(cpu)); - } + + /* Other cpus call __sync_lazy_execstate from flush ipi handler. */ + flush_tlb_mask(v->vcpu_dirty_cpumask); } unsigned long __hypercall_create_continuation( @@ -966,7 +937,7 @@ struct vcpu *v; unsigned long pfn; - BUG_ON(!cpus_empty(d->cpumask)); + BUG_ON(!cpus_empty(d->domain_dirty_cpumask)); ptwr_destroy(d); diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/domain_build.c --- a/xen/arch/x86/domain_build.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/domain_build.c Fri Jan 13 20:12:24 2006 @@ -366,26 +366,19 @@ l2tab[(LINEAR_PT_VIRT_START >> L2_PAGETABLE_SHIFT)+i] = l2e_from_paddr((u32)l2tab + i*PAGE_SIZE, __PAGE_HYPERVISOR); } - { - unsigned long va; - for (va = PERDOMAIN_VIRT_START; va < PERDOMAIN_VIRT_END; - va += (1 << L2_PAGETABLE_SHIFT)) { - l2tab[va >> L2_PAGETABLE_SHIFT] = - l2e_from_paddr(__pa(d->arch.mm_perdomain_pt) + - (va-PERDOMAIN_VIRT_START), - __PAGE_HYPERVISOR); - } - } v->arch.guest_table = mk_pagetable((unsigned long)l3start); #else l2start = l2tab = (l2_pgentry_t *)mpt_alloc; mpt_alloc += PAGE_SIZE; memcpy(l2tab, &idle_pg_table[0], PAGE_SIZE); l2tab[LINEAR_PT_VIRT_START >> L2_PAGETABLE_SHIFT] = l2e_from_paddr((unsigned long)l2start, __PAGE_HYPERVISOR); - l2tab[PERDOMAIN_VIRT_START >> L2_PAGETABLE_SHIFT] = - l2e_from_paddr(__pa(d->arch.mm_perdomain_pt), __PAGE_HYPERVISOR); v->arch.guest_table = mk_pagetable((unsigned long)l2start); #endif + + for ( i = 0; i < PDPT_L2_ENTRIES; i++ ) + l2tab[l2_linear_offset(PERDOMAIN_VIRT_START) + i] = + l2e_from_page(virt_to_page(d->arch.mm_perdomain_pt) + i, + __PAGE_HYPERVISOR); l2tab += l2_linear_offset(dsi.v_start); mfn = alloc_spfn; diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/io_apic.c --- a/xen/arch/x86/io_apic.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/io_apic.c Fri Jan 13 20:12:24 2006 @@ -1807,3 +1807,47 @@ return 0; } + +void dump_ioapic_irq_info(void) +{ + struct irq_pin_list *entry; + struct IO_APIC_route_entry rte; + unsigned int irq, pin, printed = 0; + unsigned long flags; + + for ( irq = 0; irq < NR_IRQS; irq++ ) + { + entry = &irq_2_pin[irq]; + if ( entry->pin == -1 ) + continue; + + if ( !printed++ ) + printk("IO-APIC interrupt information:\n"); + + printk(" IRQ%3d Vec%3d:\n", irq, irq_to_vector(irq)); + + for ( ; ; ) + { + pin = entry->pin; + + printk(" Apic 0x%02x, Pin %2d: ", entry->apic, pin); + + spin_lock_irqsave(&ioapic_lock, flags); + *(((int *)&rte) + 0) = io_apic_read(entry->apic, 0x10 + 2 * pin); + *(((int *)&rte) + 1) = io_apic_read(entry->apic, 0x11 + 2 * pin); + spin_unlock_irqrestore(&ioapic_lock, flags); + + printk("vector=%u, delivery_mode=%u, dest_mode=%s, " + "delivery_status=%d, polarity=%d, irr=%d, " + "trigger=%s, mask=%d\n", + rte.vector, rte.delivery_mode, + rte.dest_mode ? "logical" : "physical", + rte.delivery_status, rte.polarity, rte.irr, + rte.trigger ? "level" : "edge", rte.mask); + + if ( entry->next == 0 ) + break; + entry = &irq_2_pin[entry->next]; + } + } +} diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/irq.c --- a/xen/arch/x86/irq.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/irq.c Fri Jan 13 20:12:24 2006 @@ -12,6 +12,7 @@ #include <xen/irq.h> #include <xen/perfc.h> #include <xen/sched.h> +#include <xen/keyhandler.h> #include <asm/current.h> #include <asm/smpboot.h> @@ -198,15 +199,21 @@ int pirq_guest_bind(struct vcpu *v, int irq, int will_share) { - unsigned int vector = irq_to_vector(irq); - irq_desc_t *desc = &irq_desc[vector]; + unsigned int vector; + irq_desc_t *desc; irq_guest_action_t *action; unsigned long flags; int rc = 0; cpumask_t cpumask = CPU_MASK_NONE; + if ( (irq < 0) || (irq >= NR_IRQS) ) + return -EINVAL; + + vector = irq_to_vector(irq); if ( vector == 0 ) - return -EBUSY; + return -EINVAL; + + desc = &irq_desc[vector]; spin_lock_irqsave(&desc->lock, flags); @@ -305,3 +312,71 @@ spin_unlock_irqrestore(&desc->lock, flags); return 0; } + +extern void dump_ioapic_irq_info(void); + +static void dump_irqs(unsigned char key) +{ + int i, irq, vector; + irq_desc_t *desc; + irq_guest_action_t *action; + struct domain *d; + unsigned long flags; + + printk("Guest interrupt information:\n"); + + for ( irq = 0; irq < NR_IRQS; irq++ ) + { + vector = irq_to_vector(irq); + if ( vector == 0 ) + continue; + + desc = &irq_desc[vector]; + + spin_lock_irqsave(&desc->lock, flags); + + if ( desc->status & IRQ_GUEST ) + { + action = (irq_guest_action_t *)desc->action; + + printk(" IRQ%3d Vec%3d: type=%-15s status=%08x " + "in-flight=%d domain-list=", + irq, vector, desc->handler->typename, + desc->status, action->in_flight); + + for ( i = 0; i < action->nr_guests; i++ ) + { + d = action->guest[i]; + printk("%u(%c%c%c%c)", + d->domain_id, + (test_bit(d->pirq_to_evtchn[irq], + &d->shared_info->evtchn_pending[0]) ? + 'P' : '-'), + (test_bit(d->pirq_to_evtchn[irq]/BITS_PER_LONG, + &d->shared_info->vcpu_info[0]. + evtchn_pending_sel) ? + 'S' : '-'), + (test_bit(d->pirq_to_evtchn[irq], + &d->shared_info->evtchn_mask[0]) ? + 'M' : '-'), + (test_bit(irq, &d->pirq_mask) ? + 'M' : '-')); + if ( i != action->nr_guests ) + printk(","); + } + + printk("\n"); + } + + spin_unlock_irqrestore(&desc->lock, flags); + } + + dump_ioapic_irq_info(); +} + +static int __init setup_dump_irqs(void) +{ + register_keyhandler('i', dump_irqs, "dump interrupt bindings"); + return 0; +} +__initcall(setup_dump_irqs); diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/mm.c --- a/xen/arch/x86/mm.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/mm.c Fri Jan 13 20:12:24 2006 @@ -297,7 +297,6 @@ #if defined(__x86_64__) /* If in user mode, switch to kernel mode just to read LDT mapping. */ - extern void toggle_guest_mode(struct vcpu *); int user_mode = !(v->arch.flags & TF_kernel_mode); #define TOGGLE_MODE() if ( user_mode ) toggle_guest_mode(v) #elif defined(__i386__) @@ -841,10 +840,11 @@ L2_PAGETABLE_XEN_SLOTS * sizeof(l2_pgentry_t)); pl2e[l2_table_offset(LINEAR_PT_VIRT_START)] = l2e_from_pfn(pfn, __PAGE_HYPERVISOR); - pl2e[l2_table_offset(PERDOMAIN_VIRT_START)] = - l2e_from_page( - virt_to_page(page_get_owner(page)->arch.mm_perdomain_pt), - __PAGE_HYPERVISOR); + for ( i = 0; i < PDPT_L2_ENTRIES; i++ ) + pl2e[l2_table_offset(PERDOMAIN_VIRT_START) + i] = + l2e_from_page( + virt_to_page(page_get_owner(page)->arch.mm_perdomain_pt) + i, + __PAGE_HYPERVISOR); #endif unmap_domain_page(pl2e); @@ -1457,7 +1457,8 @@ * was GDT/LDT) but those circumstances should be * very rare. */ - cpumask_t mask = page_get_owner(page)->cpumask; + cpumask_t mask = + page_get_owner(page)->domain_dirty_cpumask; tlbflush_filter(mask, page->tlbflush_timestamp); if ( unlikely(!cpus_empty(mask)) ) @@ -1619,7 +1620,7 @@ if ( shadow_mode_enabled(d) ) shadow_sync_all(d); if ( deferred_ops & DOP_FLUSH_ALL_TLBS ) - flush_tlb_mask(d->cpumask); + flush_tlb_mask(d->domain_dirty_cpumask); else local_flush_tlb(); } @@ -1691,7 +1692,7 @@ struct domain *d, unsigned long vmask) { unsigned int vcpu_id; - cpumask_t pmask; + cpumask_t pmask = CPU_MASK_NONE; struct vcpu *v; while ( vmask != 0 ) @@ -1700,7 +1701,7 @@ vmask &= ~(1UL << vcpu_id); if ( (vcpu_id < MAX_VIRT_CPUS) && ((v = d->vcpu[vcpu_id]) != NULL) ) - cpu_set(v->processor, pmask); + cpus_or(pmask, pmask, v->vcpu_dirty_cpumask); } return pmask; @@ -1869,7 +1870,6 @@ break; } pmask = vcpumask_to_pcpumask(d, vmask); - cpus_and(pmask, pmask, d->cpumask); if ( op.cmd == MMUEXT_TLB_FLUSH_MULTI ) flush_tlb_mask(pmask); else @@ -1878,11 +1878,11 @@ } case MMUEXT_TLB_FLUSH_ALL: - flush_tlb_mask(d->cpumask); + flush_tlb_mask(d->domain_dirty_cpumask); break; case MMUEXT_INVLPG_ALL: - flush_tlb_one_mask(d->cpumask, op.arg1.linear_addr); + flush_tlb_one_mask(d->domain_dirty_cpumask, op.arg1.linear_addr); break; case MMUEXT_FLUSH_CACHE: @@ -2497,7 +2497,7 @@ l1_pgentry_t val = l1e_from_intpte(val64); struct vcpu *v = current; struct domain *d = v->domain; - unsigned int cpu = v->processor; + unsigned int cpu = smp_processor_id(); unsigned long vmask, bmap_ptr; cpumask_t pmask; int rc = 0; @@ -2548,13 +2548,12 @@ local_flush_tlb(); break; case UVMF_ALL: - flush_tlb_mask(d->cpumask); + flush_tlb_mask(d->domain_dirty_cpumask); break; default: if ( unlikely(get_user(vmask, (unsigned long *)bmap_ptr)) ) rc = -EFAULT; pmask = vcpumask_to_pcpumask(d, vmask); - cpus_and(pmask, pmask, d->cpumask); flush_tlb_mask(pmask); break; } @@ -2569,13 +2568,12 @@ local_flush_tlb_one(va); break; case UVMF_ALL: - flush_tlb_one_mask(d->cpumask, va); + flush_tlb_one_mask(d->domain_dirty_cpumask, va); break; default: if ( unlikely(get_user(vmask, (unsigned long *)bmap_ptr)) ) rc = -EFAULT; pmask = vcpumask_to_pcpumask(d, vmask); - cpus_and(pmask, pmask, d->cpumask); flush_tlb_one_mask(pmask, va); break; } @@ -2972,7 +2970,6 @@ #ifdef CONFIG_X86_64 struct vcpu *v = current; - extern void toggle_guest_mode(struct vcpu *); int user_mode = !(v->arch.flags & TF_kernel_mode); #endif @@ -3002,7 +2999,7 @@ BUG(); } PTWR_PRINTK("[%c] disconnected_l1va at %p is %"PRIpte"\n", - PTWR_PRINT_WHICH, ptep, pte.l1); + PTWR_PRINT_WHICH, ptep, l1e_get_intpte(pte)); l1e_remove_flags(pte, _PAGE_RW); /* Write-protect the p.t. page in the guest page table. */ @@ -3018,20 +3015,33 @@ /* Ensure that there are no stale writable mappings in any TLB. */ /* NB. INVLPG is a serialising instruction: flushes pending updates. */ - flush_tlb_one_mask(d->cpumask, l1va); + flush_tlb_one_mask(d->domain_dirty_cpumask, l1va); PTWR_PRINTK("[%c] disconnected_l1va at %p now %"PRIpte"\n", - PTWR_PRINT_WHICH, ptep, pte.l1); + PTWR_PRINT_WHICH, ptep, l1e_get_intpte(pte)); /* * STEP 2. Validate any modified PTEs. */ - pl1e = d->arch.ptwr[which].pl1e; - modified = revalidate_l1(d, pl1e, d->arch.ptwr[which].page); - unmap_domain_page(pl1e); - perfc_incr_histo(wpt_updates, modified, PT_UPDATES); - ptwr_eip_stat_update(d->arch.ptwr[which].eip, d->domain_id, modified); - d->arch.ptwr[which].prev_nr_updates = modified; + if ( likely(d == current->domain) ) + { + pl1e = map_domain_page(l1e_get_pfn(pte)); + modified = revalidate_l1(d, pl1e, d->arch.ptwr[which].page); + unmap_domain_page(pl1e); + perfc_incr_histo(wpt_updates, modified, PT_UPDATES); + ptwr_eip_stat_update(d->arch.ptwr[which].eip, d->domain_id, modified); + d->arch.ptwr[which].prev_nr_updates = modified; + } + else + { + /* + * Must make a temporary global mapping, since we are running in the + * wrong address space, so no access to our own mapcache. + */ + pl1e = map_domain_page_global(l1e_get_pfn(pte)); + modified = revalidate_l1(d, pl1e, d->arch.ptwr[which].page); + unmap_domain_page_global(pl1e); + } /* * STEP 3. Reattach the L1 p.t. page into the current address space. @@ -3209,7 +3219,7 @@ { unsigned long pfn; struct pfn_info *page; - l1_pgentry_t pte; + l1_pgentry_t *pl1e, pte; l2_pgentry_t *pl2e, l2e; int which, flags; unsigned long l2_idx; @@ -3342,15 +3352,14 @@ if ( which == PTWR_PT_ACTIVE ) { l2e_remove_flags(*pl2e, _PAGE_PRESENT); - flush_tlb_mask(d->cpumask); + flush_tlb_mask(d->domain_dirty_cpumask); } /* Temporarily map the L1 page, and make a copy of it. */ - d->arch.ptwr[which].pl1e = map_domain_page(pfn); - memcpy(d->arch.ptwr[which].page, - d->arch.ptwr[which].pl1e, - L1_PAGETABLE_ENTRIES * sizeof(l1_pgentry_t)); - + pl1e = map_domain_page(pfn); + memcpy(d->arch.ptwr[which].page, pl1e, PAGE_SIZE); + unmap_domain_page(pl1e); + /* Finally, make the p.t. page writable by the guest OS. */ l1e_add_flags(pte, _PAGE_RW); if ( unlikely(__put_user(pte.l1, @@ -3359,7 +3368,6 @@ MEM_LOG("ptwr: Could not update pte at %p", (unsigned long *) &linear_pg_table[l1_linear_offset(addr)]); /* Toss the writable pagetable state and crash. */ - unmap_domain_page(d->arch.ptwr[which].pl1e); d->arch.ptwr[which].l1va = 0; domain_crash(d); return 0; @@ -3369,7 +3377,7 @@ emulate: if ( x86_emulate_memop(guest_cpu_user_regs(), addr, - &ptwr_mem_emulator, BITS_PER_LONG/8) ) + &ptwr_mem_emulator, X86EMUL_MODE_HOST) ) return 0; perfc_incrc(ptwr_emulations); return EXCRET_fault_fixed; diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/nmi.c --- a/xen/arch/x86/nmi.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/nmi.c Fri Jan 13 20:12:24 2006 @@ -23,18 +23,20 @@ #include <xen/sched.h> #include <xen/console.h> #include <xen/smp.h> +#include <xen/keyhandler.h> #include <asm/current.h> #include <asm/mc146818rtc.h> #include <asm/msr.h> #include <asm/mpspec.h> #include <asm/debugger.h> #include <asm/div64.h> +#include <asm/apic.h> unsigned int nmi_watchdog = NMI_NONE; static unsigned int nmi_hz = HZ; static unsigned int nmi_perfctr_msr; /* the MSR to reset in NMI handler */ static unsigned int nmi_p4_cccr_val; -static struct ac_timer nmi_timer[NR_CPUS]; +static struct timer nmi_timer[NR_CPUS]; static unsigned int nmi_timer_ticks[NR_CPUS]; /* @@ -132,7 +134,7 @@ { int cpu = smp_processor_id(); nmi_timer_ticks[cpu]++; - set_ac_timer(&nmi_timer[cpu], NOW() + MILLISECS(1000)); + set_timer(&nmi_timer[cpu], NOW() + MILLISECS(1000)); } static void disable_lapic_nmi_watchdog(void) @@ -308,8 +310,6 @@ void __pminit setup_apic_nmi_watchdog(void) { - int cpu = smp_processor_id(); - if (!nmi_watchdog) return; @@ -344,49 +344,37 @@ lapic_nmi_owner = LAPIC_NMI_WATCHDOG; nmi_active = 1; - - init_ac_timer(&nmi_timer[cpu], nmi_timer_fn, NULL, cpu); } static unsigned int last_irq_sums [NR_CPUS], alert_counter [NR_CPUS]; -static spinlock_t watchdog_lock = SPIN_LOCK_UNLOCKED; -static unsigned int watchdog_disable_count = 1; -static unsigned int watchdog_on; +static atomic_t watchdog_disable_count = ATOMIC_INIT(1); void watchdog_disable(void) { - unsigned long flags; - - spin_lock_irqsave(&watchdog_lock, flags); - - if ( watchdog_disable_count++ == 0 ) - watchdog_on = 0; - - spin_unlock_irqrestore(&watchdog_lock, flags); + atomic_inc(&watchdog_disable_count); } void watchdog_enable(void) { - unsigned int cpu; - unsigned long flags; - - spin_lock_irqsave(&watchdog_lock, flags); - - if ( --watchdog_disable_count == 0 ) + static unsigned long heartbeat_initialised; + unsigned int cpu; + + if ( !atomic_dec_and_test(&watchdog_disable_count) || + test_and_set_bit(0, &heartbeat_initialised) ) + return; + + /* + * Activate periodic heartbeats. We cannot do this earlier during + * setup because the timer infrastructure is not available. + */ + for_each_online_cpu ( cpu ) { - watchdog_on = 1; - /* - * Ensure periodic heartbeats are active. We cannot do this earlier - * during setup because the timer infrastructure is not available. - */ - for_each_online_cpu ( cpu ) - set_ac_timer(&nmi_timer[cpu], NOW()); - } - - spin_unlock_irqrestore(&watchdog_lock, flags); + init_timer(&nmi_timer[cpu], nmi_timer_fn, NULL, cpu); + set_timer(&nmi_timer[cpu], NOW()); + } } void nmi_watchdog_tick(struct cpu_user_regs * regs) @@ -395,7 +383,7 @@ sum = nmi_timer_ticks[cpu]; - if ( (last_irq_sums[cpu] == sum) && watchdog_on ) + if ( (last_irq_sums[cpu] == sum) && !atomic_read(&watchdog_disable_count) ) { /* * Ayiee, looks like this CPU is stuck ... wait a few IRQs (5 seconds) @@ -440,3 +428,29 @@ write_watchdog_counter(NULL); } } + +/* + * For some reason the destination shorthand for self is not valid + * when used with the NMI delivery mode. This is documented in Tables + * 8-3 and 8-4 in IA32 Reference Manual Volume 3. We send the IPI to + * our own APIC ID explicitly which is valid. + */ +static void do_nmi_trigger(unsigned char key) +{ + u32 id = apic_read(APIC_ID); + + printk("Triggering NMI on APIC ID %x\n", id); + + local_irq_disable(); + apic_wait_icr_idle(); + apic_write_around(APIC_ICR2, SET_APIC_DEST_FIELD(id)); + apic_write_around(APIC_ICR, APIC_DM_NMI | APIC_INT_ASSERT); + local_irq_enable(); +} + +static __init int register_nmi_trigger(void) +{ + register_keyhandler('n', do_nmi_trigger, "trigger an NMI"); + return 0; +} +__initcall(register_nmi_trigger); diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/setup.c --- a/xen/arch/x86/setup.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/setup.c Fri Jan 13 20:12:24 2006 @@ -81,6 +81,10 @@ extern void initialize_keytable(void); extern void early_cpu_init(void); +struct tss_struct init_tss[NR_CPUS]; + +struct vcpu *idle_vcpu[NR_CPUS]; + extern unsigned long cpu0_stack[]; struct cpuinfo_x86 boot_cpu_data = { 0, 0, 0, 0, -1, 1, 0, 0, -1 }; @@ -91,8 +95,6 @@ unsigned long mmu_cr4_features = X86_CR4_PSE; #endif EXPORT_SYMBOL(mmu_cr4_features); - -struct vcpu *idle_task[NR_CPUS] = { &idle0_vcpu }; int acpi_disabled; @@ -144,8 +146,8 @@ void __init __start_xen(multiboot_info_t *mbi) { - unsigned long vgdt, gdt_pfn; char *cmdline; + struct domain *idle_domain; unsigned long _initrd_start = 0, _initrd_len = 0; unsigned int initrdidx = 1; module_t *mod = (module_t *)__va(mbi->mods_addr); @@ -163,9 +165,8 @@ if ( (mbi->flags & MBI_CMDLINE) && (mbi->cmdline != 0) ) cmdline_parse(__va(mbi->cmdline)); - /* Must do this early -- e.g., spinlocks rely on get_current(). */ - set_current(&idle0_vcpu); - set_processor_id(0); + set_current((struct vcpu *)0xfffff000); /* debug sanity */ + set_processor_id(0); /* needed early, for smp_processor_id() */ smp_prepare_boot_cpu(); @@ -343,6 +344,12 @@ BUG_ON(sizeof(shared_info_t) > PAGE_SIZE); BUG_ON(sizeof(vcpu_info_t) != 64); + /* __foo are defined in public headers. Check they match internal defs. */ + BUG_ON(__HYPERVISOR_VIRT_START != HYPERVISOR_VIRT_START); +#ifdef HYPERVISOR_VIRT_END + BUG_ON(__HYPERVISOR_VIRT_END != HYPERVISOR_VIRT_END); +#endif + init_frametable(); end_boot_allocator(); @@ -376,6 +383,14 @@ early_cpu_init(); + scheduler_init(); + + idle_domain = do_createdomain(IDLE_DOMAIN_ID, 0); + BUG_ON(idle_domain == NULL); + + set_current(idle_domain->vcpu[0]); + idle_vcpu[0] = current; + paging_init(); /* Unmap the first page of CPU0's stack. */ @@ -388,21 +403,6 @@ sort_exception_tables(); - if ( arch_do_createdomain(current) != 0 ) - BUG(); - - /* - * Map default GDT into its final positions in the idle page table. As - * noted in arch_do_createdomain(), we must map for every possible VCPU#. - */ - vgdt = GDT_VIRT_START(current) + FIRST_RESERVED_GDT_BYTE; - gdt_pfn = virt_to_phys(gdt_table) >> PAGE_SHIFT; - for ( i = 0; i < MAX_VIRT_CPUS; i++ ) - { - map_pages_to_xen(vgdt, gdt_pfn, 1, PAGE_HYPERVISOR); - vgdt += 1 << PDPT_VCPU_VA_SHIFT; - } - find_smp_config(); smp_alloc_memory(); @@ -423,13 +423,11 @@ trap_init(); - ac_timer_init(); + timer_init(); early_time_init(); arch_init_memory(); - - scheduler_init(); identify_cpu(&boot_cpu_data); if ( cpu_has_fxsr ) @@ -480,7 +478,8 @@ schedulers_start(); - watchdog_enable(); + if ( opt_watchdog ) + watchdog_enable(); shadow_mode_init(); diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/shadow.c --- a/xen/arch/x86/shadow.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/shadow.c Fri Jan 13 20:12:24 2006 @@ -469,6 +469,7 @@ { unsigned long smfn; l2_pgentry_t *spl2e; + int i; SH_VVLOG("shadow_l2_table(gpfn=%lx, gmfn=%lx)", gpfn, gmfn); @@ -503,9 +504,11 @@ spl2e[l2_table_offset(SH_LINEAR_PT_VIRT_START)] = l2e_from_pfn(smfn, __PAGE_HYPERVISOR); - spl2e[l2_table_offset(PERDOMAIN_VIRT_START)] = - l2e_from_paddr(__pa(page_get_owner(pfn_to_page(gmfn))->arch.mm_perdomain_pt), - __PAGE_HYPERVISOR); + for ( i = 0; i < PDPT_L2_ENTRIES; i++ ) + spl2e[l2_table_offset(PERDOMAIN_VIRT_START) + i] = + l2e_from_page(virt_to_page(page_get_owner(pfn_to_page(gmfn))-> + arch.mm_perdomain_pt) + i, + __PAGE_HYPERVISOR); if ( shadow_mode_translate(d) ) // NB: not external { @@ -1800,7 +1803,7 @@ } /* Other VCPUs mustn't use the revoked writable mappings. */ - other_vcpus_mask = d->cpumask; + other_vcpus_mask = d->domain_dirty_cpumask; cpu_clear(smp_processor_id(), other_vcpus_mask); flush_tlb_mask(other_vcpus_mask); @@ -2150,8 +2153,8 @@ if ( max_mode & (SHM_enable | SHM_external) ) { if ( likely(v->arch.guest_vtable != NULL) ) - unmap_domain_page(v->arch.guest_vtable); - v->arch.guest_vtable = map_domain_page(gmfn); + unmap_domain_page_global(v->arch.guest_vtable); + v->arch.guest_vtable = map_domain_page_global(gmfn); } /* @@ -2187,8 +2190,8 @@ ) { if ( v->arch.shadow_vtable ) - unmap_domain_page(v->arch.shadow_vtable); - v->arch.shadow_vtable = map_domain_page(smfn); + unmap_domain_page_global(v->arch.shadow_vtable); + v->arch.shadow_vtable = map_domain_page_global(smfn); } #if CONFIG_PAGING_LEVELS == 2 @@ -2204,8 +2207,8 @@ if ( unlikely(!(hl2mfn = __shadow_status(d, gpfn, PGT_hl2_shadow))) ) hl2mfn = shadow_hl2_table(d, gpfn, gmfn, smfn); if ( v->arch.hl2_vtable ) - unmap_domain_page(v->arch.hl2_vtable); - v->arch.hl2_vtable = map_domain_page(hl2mfn); + unmap_domain_page_global(v->arch.hl2_vtable); + v->arch.hl2_vtable = map_domain_page_global(hl2mfn); } /* diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/shadow32.c --- a/xen/arch/x86/shadow32.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/shadow32.c Fri Jan 13 20:12:24 2006 @@ -726,6 +726,7 @@ l2_pgentry_t *mpl2e; struct pfn_info *mmfn_info; struct domain *d = v->domain; + int i; ASSERT(pagetable_get_paddr(v->arch.monitor_table) == 0); @@ -733,16 +734,17 @@ ASSERT(mmfn_info != NULL); mmfn = page_to_pfn(mmfn_info); - mpl2e = (l2_pgentry_t *)map_domain_page(mmfn); + mpl2e = (l2_pgentry_t *)map_domain_page_global(mmfn); memset(mpl2e, 0, PAGE_SIZE); memcpy(&mpl2e[DOMAIN_ENTRIES_PER_L2_PAGETABLE], &idle_pg_table[DOMAIN_ENTRIES_PER_L2_PAGETABLE], HYPERVISOR_ENTRIES_PER_L2_PAGETABLE * sizeof(l2_pgentry_t)); - mpl2e[l2_table_offset(PERDOMAIN_VIRT_START)] = - l2e_from_paddr(__pa(d->arch.mm_perdomain_pt), - __PAGE_HYPERVISOR); + for ( i = 0; i < PDPT_L2_ENTRIES; i++ ) + mpl2e[l2_table_offset(PERDOMAIN_VIRT_START) + i] = + l2e_from_page(virt_to_page(d->arch.mm_perdomain_pt) + i, + __PAGE_HYPERVISOR); // map the phys_to_machine map into the Read-Only MPT space for this domain mpl2e[l2_table_offset(RO_MPT_VIRT_START)] = @@ -794,7 +796,7 @@ * Then free monitor_table. */ mfn = pagetable_get_pfn(v->arch.monitor_table); - unmap_domain_page(v->arch.monitor_vtable); + unmap_domain_page_global(v->arch.monitor_vtable); free_domheap_page(pfn_to_page(mfn)); v->arch.monitor_table = mk_pagetable(0); @@ -929,7 +931,7 @@ if ( v->arch.guest_vtable && (v->arch.guest_vtable != __linear_l2_table) ) { - unmap_domain_page(v->arch.guest_vtable); + unmap_domain_page_global(v->arch.guest_vtable); } if ( (mode & (SHM_translate | SHM_external)) == SHM_translate ) v->arch.guest_vtable = __linear_l2_table; @@ -942,7 +944,7 @@ if ( v->arch.shadow_vtable && (v->arch.shadow_vtable != __shadow_linear_l2_table) ) { - unmap_domain_page(v->arch.shadow_vtable); + unmap_domain_page_global(v->arch.shadow_vtable); } if ( !(mode & SHM_external) ) v->arch.shadow_vtable = __shadow_linear_l2_table; @@ -955,7 +957,7 @@ if ( v->arch.hl2_vtable && (v->arch.hl2_vtable != __linear_hl2_table) ) { - unmap_domain_page(v->arch.hl2_vtable); + unmap_domain_page_global(v->arch.hl2_vtable); } if ( (mode & (SHM_translate | SHM_external)) == SHM_translate ) v->arch.hl2_vtable = __linear_hl2_table; @@ -1508,6 +1510,7 @@ { unsigned long smfn; l2_pgentry_t *spl2e; + int i; SH_VVLOG("shadow_l2_table(gpfn=%lx, gmfn=%lx)", gpfn, gmfn); @@ -1542,9 +1545,11 @@ spl2e[l2_table_offset(SH_LINEAR_PT_VIRT_START)] = l2e_from_pfn(smfn, __PAGE_HYPERVISOR); - spl2e[l2_table_offset(PERDOMAIN_VIRT_START)] = - l2e_from_paddr(__pa(page_get_owner(pfn_to_page(gmfn))->arch.mm_perdomain_pt), - __PAGE_HYPERVISOR); + for ( i = 0; i < PDPT_L2_ENTRIES; i++ ) + spl2e[l2_table_offset(PERDOMAIN_VIRT_START) + i] = + l2e_from_page(virt_to_page(page_get_owner(pfn_to_page(gmfn))-> + arch.mm_perdomain_pt) + i, + __PAGE_HYPERVISOR); if ( shadow_mode_translate(d) ) // NB: not external { @@ -2586,7 +2591,7 @@ } /* Other VCPUs mustn't use the revoked writable mappings. */ - other_vcpus_mask = d->cpumask; + other_vcpus_mask = d->domain_dirty_cpumask; cpu_clear(smp_processor_id(), other_vcpus_mask); flush_tlb_mask(other_vcpus_mask); @@ -2906,8 +2911,8 @@ if ( max_mode & (SHM_enable | SHM_external) ) { if ( likely(v->arch.guest_vtable != NULL) ) - unmap_domain_page(v->arch.guest_vtable); - v->arch.guest_vtable = map_domain_page(gmfn); + unmap_domain_page_global(v->arch.guest_vtable); + v->arch.guest_vtable = map_domain_page_global(gmfn); } /* @@ -2932,8 +2937,8 @@ if ( max_mode == SHM_external ) { if ( v->arch.shadow_vtable ) - unmap_domain_page(v->arch.shadow_vtable); - v->arch.shadow_vtable = map_domain_page(smfn); + unmap_domain_page_global(v->arch.shadow_vtable); + v->arch.shadow_vtable = map_domain_page_global(smfn); } /* @@ -2948,8 +2953,8 @@ if ( unlikely(!(hl2mfn = __shadow_status(d, gpfn, PGT_hl2_shadow))) ) hl2mfn = shadow_hl2_table(d, gpfn, gmfn, smfn); if ( v->arch.hl2_vtable ) - unmap_domain_page(v->arch.hl2_vtable); - v->arch.hl2_vtable = map_domain_page(hl2mfn); + unmap_domain_page_global(v->arch.hl2_vtable); + v->arch.hl2_vtable = map_domain_page_global(hl2mfn); } /* diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/shadow_public.c --- a/xen/arch/x86/shadow_public.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/shadow_public.c Fri Jan 13 20:12:24 2006 @@ -151,6 +151,8 @@ for (i = 0; i < L1_PAGETABLE_ENTRIES; i++) put_page_from_l1e(pl1e[i], d); + + unmap_domain_page(pl1e); } /* @@ -254,6 +256,7 @@ pae_l3 = map_domain_page(pagetable_get_pfn(d->arch.phys_table)); for (i = 0; i < PDP_ENTRIES; i++) l3[i] = l3e_from_pfn(l3e_get_pfn(pae_l3[i]), __PAGE_HYPERVISOR); + unmap_domain_page(pae_l3); unmap_domain_page(l4); unmap_domain_page(l3); @@ -275,7 +278,7 @@ ASSERT( mmfn_info ); mmfn = page_to_pfn(mmfn_info); - mpl4e = (l4_pgentry_t *) map_domain_page(mmfn); + mpl4e = (l4_pgentry_t *) map_domain_page_global(mmfn); memcpy(mpl4e, &idle_pg_table[0], PAGE_SIZE); mpl4e[l4_table_offset(PERDOMAIN_VIRT_START)] = l4e_from_paddr(__pa(d->arch.mm_perdomain_l3), __PAGE_HYPERVISOR); @@ -298,7 +301,7 @@ * free monitor_table. */ mfn = pagetable_get_pfn(v->arch.monitor_table); - unmap_domain_page(v->arch.monitor_vtable); + unmap_domain_page_global(v->arch.monitor_vtable); free_domheap_page(pfn_to_page(mfn)); v->arch.monitor_table = mk_pagetable(0); @@ -325,6 +328,7 @@ l2_pgentry_t *mpl2e; struct pfn_info *mmfn_info; struct domain *d = v->domain; + int i; ASSERT(pagetable_get_paddr(v->arch.monitor_table) == 0); @@ -332,16 +336,17 @@ ASSERT(mmfn_info != NULL); mmfn = page_to_pfn(mmfn_info); - mpl2e = (l2_pgentry_t *)map_domain_page(mmfn); + mpl2e = (l2_pgentry_t *)map_domain_page_global(mmfn); memset(mpl2e, 0, PAGE_SIZE); memcpy(&mpl2e[DOMAIN_ENTRIES_PER_L2_PAGETABLE], &idle_pg_table[DOMAIN_ENTRIES_PER_L2_PAGETABLE], HYPERVISOR_ENTRIES_PER_L2_PAGETABLE * sizeof(l2_pgentry_t)); - mpl2e[l2_table_offset(PERDOMAIN_VIRT_START)] = - l2e_from_paddr(__pa(d->arch.mm_perdomain_pt), - __PAGE_HYPERVISOR); + for ( i = 0; i < PDPT_L2_ENTRIES; i++ ) + mpl2e[l2_table_offset(PERDOMAIN_VIRT_START) + i] = + l2e_from_page(virt_to_page(d->arch.mm_perdomain_pt) + i, + __PAGE_HYPERVISOR); // map the phys_to_machine map into the Read-Only MPT space for this domain mpl2e[l2_table_offset(RO_MPT_VIRT_START)] = @@ -393,7 +398,7 @@ * Then free monitor_table. */ mfn = pagetable_get_pfn(v->arch.monitor_table); - unmap_domain_page(v->arch.monitor_vtable); + unmap_domain_page_global(v->arch.monitor_vtable); free_domheap_page(pfn_to_page(mfn)); v->arch.monitor_table = mk_pagetable(0); @@ -977,7 +982,7 @@ if ( v->arch.guest_vtable && (v->arch.guest_vtable != __linear_l2_table) ) { - unmap_domain_page(v->arch.guest_vtable); + unmap_domain_page_global(v->arch.guest_vtable); } if ( (mode & (SHM_translate | SHM_external)) == SHM_translate ) v->arch.guest_vtable = __linear_l2_table; @@ -990,7 +995,7 @@ if ( v->arch.shadow_vtable && (v->arch.shadow_vtable != __shadow_linear_l2_table) ) { - unmap_domain_page(v->arch.shadow_vtable); + unmap_domain_page_global(v->arch.shadow_vtable); } if ( !(mode & SHM_external) && d->arch.ops->guest_paging_levels == 2) v->arch.shadow_vtable = __shadow_linear_l2_table; @@ -1004,7 +1009,7 @@ if ( v->arch.hl2_vtable && (v->arch.hl2_vtable != __linear_hl2_table) ) { - unmap_domain_page(v->arch.hl2_vtable); + unmap_domain_page_global(v->arch.hl2_vtable); } if ( (mode & (SHM_translate | SHM_external)) == SHM_translate ) v->arch.hl2_vtable = __linear_hl2_table; diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/smpboot.c --- a/xen/arch/x86/smpboot.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/smpboot.c Fri Jan 13 20:12:24 2006 @@ -435,7 +435,7 @@ extern void percpu_traps_init(void); - set_current(idle_task[cpu]); + set_current(idle_vcpu[cpu]); set_processor_id(cpu); percpu_traps_init(); @@ -761,7 +761,6 @@ * Returns zero if CPU booted OK, else error code from wakeup_secondary_cpu. */ { - struct domain *idle; struct vcpu *v; unsigned long boot_error; int timeout, cpu; @@ -770,14 +769,10 @@ cpu = ++cpucount; - if ( (idle = do_createdomain(IDLE_DOMAIN_ID, cpu)) == NULL ) - panic("failed 'createdomain' for CPU %d", cpu); - - v = idle_task[cpu] = idle->vcpu[0]; - - set_bit(_DOMF_idle_domain, &idle->domain_flags); - - v->arch.monitor_table = mk_pagetable(__pa(idle_pg_table)); + v = idle_vcpu[cpu] = alloc_vcpu(idle_vcpu[0]->domain, cpu, cpu); + BUG_ON(v == NULL); + + v->arch.monitor_table = mk_pagetable(__pa(idle_pg_table)); /* start_eip had better be page-aligned! */ start_eip = setup_trampoline(); diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/time.c --- a/xen/arch/x86/time.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/time.c Fri Jan 13 20:12:24 2006 @@ -17,7 +17,7 @@ #include <xen/config.h> #include <xen/init.h> #include <xen/time.h> -#include <xen/ac_timer.h> +#include <xen/timer.h> #include <xen/smp.h> #include <xen/irq.h> #include <xen/softirq.h> @@ -56,7 +56,7 @@ s_time_t stime_local_stamp; s_time_t stime_master_stamp; struct time_scale tsc_scale; - struct ac_timer calibration_timer; + struct timer calibration_timer; } __cacheline_aligned; static struct cpu_time cpu_time[NR_CPUS]; @@ -163,7 +163,7 @@ /* Rough hack to allow accurate timers to sort-of-work with no APIC. */ if ( !cpu_has_apic ) - raise_softirq(AC_TIMER_SOFTIRQ); + raise_softirq(TIMER_SOFTIRQ); if ( using_pit ) pit_overflow(); @@ -342,7 +342,7 @@ /* Protected by platform_timer_lock. */ static u64 hpet_counter64, hpet_overflow_period; static u32 hpet_stamp; -static struct ac_timer hpet_overflow_timer; +static struct timer hpet_overflow_timer; static void hpet_overflow(void *unused) { @@ -354,7 +354,7 @@ hpet_stamp = counter; spin_unlock_irq(&platform_timer_lock); - set_ac_timer(&hpet_overflow_timer, NOW() + hpet_overflow_period); + set_timer(&hpet_overflow_timer, NOW() + hpet_overflow_period); } static u64 read_hpet_count(void) @@ -430,7 +430,7 @@ (void)do_div(hpet_overflow_period, (u32)hpet_rate); } - init_ac_timer(&hpet_overflow_timer, hpet_overflow, NULL, 0); + init_timer(&hpet_overflow_timer, hpet_overflow, NULL, 0); hpet_overflow(NULL); platform_timer_stamp = hpet_counter64; @@ -459,7 +459,7 @@ /* Protected by platform_timer_lock. */ static u64 cyclone_counter64; static u32 cyclone_stamp; -static struct ac_timer cyclone_overflow_timer; +static struct timer cyclone_overflow_timer; static volatile u32 *cyclone_timer; /* Cyclone MPMC0 register */ static void cyclone_overflow(void *unused) @@ -472,7 +472,7 @@ cyclone_stamp = counter; spin_unlock_irq(&platform_timer_lock); - set_ac_timer(&cyclone_overflow_timer, NOW() + MILLISECS(20000)); + set_timer(&cyclone_overflow_timer, NOW() + MILLISECS(20000)); } static u64 read_cyclone_count(void) @@ -510,7 +510,7 @@ read_platform_count = read_cyclone_count; - init_ac_timer(&cyclone_overflow_timer, cyclone_overflow, NULL, 0); + init_timer(&cyclone_overflow_timer, cyclone_overflow, NULL, 0); cyclone_overflow(NULL); platform_timer_stamp = cyclone_counter64; set_time_scale(&platform_timer_scale, CYCLONE_TIMER_FREQ); @@ -876,7 +876,7 @@ cpu_time[cpu].stime_master_stamp = curr_master_stime; out: - set_ac_timer(&cpu_time[cpu].calibration_timer, NOW() + EPOCH); + set_timer(&cpu_time[cpu].calibration_timer, NOW() + EPOCH); if ( cpu == 0 ) platform_time_calibration(); @@ -896,9 +896,9 @@ cpu_time[cpu].stime_master_stamp = now; cpu_time[cpu].stime_local_stamp = now; - init_ac_timer(&cpu_time[cpu].calibration_timer, + init_timer(&cpu_time[cpu].calibration_timer, local_time_calibration, NULL, cpu); - set_ac_timer(&cpu_time[cpu].calibration_timer, NOW() + EPOCH); + set_timer(&cpu_time[cpu].calibration_timer, NOW() + EPOCH); } /* Late init function (after all CPUs are booted). */ diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/traps.c --- a/xen/arch/x86/traps.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/traps.c Fri Jan 13 20:12:24 2006 @@ -130,9 +130,19 @@ static void show_guest_stack(struct cpu_user_regs *regs) { int i; - unsigned long *stack = (unsigned long *)regs->esp, addr; - - printk("Guest stack trace from "__OP"sp=%p:\n ", stack); + unsigned long *stack, addr; + + if ( VM86_MODE(regs) ) + { + stack = (unsigned long *)((regs->ss << 4) + (regs->esp & 0xffff)); + printk("Guest stack trace from ss:sp = %04x:%04x (VM86)\n ", + regs->ss, (uint16_t)(regs->esp & 0xffff)); + } + else + { + stack = (unsigned long *)regs->esp; + printk("Guest stack trace from "__OP"sp=%p:\n ", stack); + } for ( i = 0; i < (debug_stack_lines*stack_words_per_line); i++ ) { @@ -427,7 +437,7 @@ tb->flags |= TBF_INTERRUPT; } -static int handle_perdomain_mapping_fault( +static int handle_gdt_ldt_mapping_fault( unsigned long offset, struct cpu_user_regs *regs) { extern int map_ldt_shadow_page(unsigned int); @@ -437,14 +447,14 @@ int ret; /* Which vcpu's area did we fault in, and is it in the ldt sub-area? */ - unsigned int is_ldt_area = (offset >> (PDPT_VCPU_VA_SHIFT-1)) & 1; - unsigned int vcpu_area = (offset >> PDPT_VCPU_VA_SHIFT); + unsigned int is_ldt_area = (offset >> (GDT_LDT_VCPU_VA_SHIFT-1)) & 1; + unsigned int vcpu_area = (offset >> GDT_LDT_VCPU_VA_SHIFT); /* Should never fault in another vcpu's area. */ BUG_ON(vcpu_area != current->vcpu_id); /* Byte offset within the gdt/ldt sub-area. */ - offset &= (1UL << (PDPT_VCPU_VA_SHIFT-1)) - 1UL; + offset &= (1UL << (GDT_LDT_VCPU_VA_SHIFT-1)) - 1UL; if ( likely(is_ldt_area) ) { @@ -490,9 +500,9 @@ { if ( shadow_mode_external(d) && GUEST_CONTEXT(v, regs) ) return shadow_fault(addr, regs); - if ( (addr >= PERDOMAIN_VIRT_START) && (addr < PERDOMAIN_VIRT_END) ) - return handle_perdomain_mapping_fault( - addr - PERDOMAIN_VIRT_START, regs); + if ( (addr >= GDT_LDT_VIRT_START) && (addr < GDT_LDT_VIRT_END) ) + return handle_gdt_ldt_mapping_fault( + addr - GDT_LDT_VIRT_START, regs); } else if ( unlikely(shadow_mode_enabled(d)) ) { @@ -596,7 +606,6 @@ u16 x; #if defined(__x86_64__) /* If in user mode, switch to kernel mode just to read I/O bitmap. */ - extern void toggle_guest_mode(struct vcpu *); int user_mode = !(v->arch.flags & TF_kernel_mode); #define TOGGLE_MODE() if ( user_mode ) toggle_guest_mode(v) #elif defined(__i386__) @@ -964,16 +973,26 @@ case 0x30: /* WRMSR */ /* Ignore the instruction if unprivileged. */ if ( !IS_PRIV(v->domain) ) - DPRINTK("Non-priv domain attempted WRMSR(%p,%08lx,%08lx).\n", - _p(regs->ecx), (long)regs->eax, (long)regs->edx); + { + u32 l, h; + if ( (rdmsr_user(regs->ecx, l, h) != 0) || + (regs->ecx != MSR_EFER) || + (regs->eax != l) || (regs->edx != h) ) + DPRINTK("Non-priv domain attempted WRMSR %p from " + "%08x:%08x to %08lx:%08lx.\n", + _p(regs->ecx), h, l, (long)regs->edx, (long)regs->eax); + } else if ( wrmsr_user(regs->ecx, regs->eax, regs->edx) ) goto fail; break; case 0x32: /* RDMSR */ if ( !IS_PRIV(v->domain) ) - DPRINTK("Non-priv domain attempted RDMSR(%p,%08lx,%08lx).\n", - _p(regs->ecx), (long)regs->eax, (long)regs->edx); + { + if ( regs->ecx != MSR_EFER ) + DPRINTK("Non-priv domain attempted RDMSR %p.\n", + _p(regs->ecx)); + } /* Everyone can read the MSR space. */ if ( rdmsr_user(regs->ecx, regs->eax, regs->edx) ) goto fail; @@ -1080,26 +1099,23 @@ return 0; } - -/* Defer dom0 notification to softirq context (unsafe in NMI context). */ -static unsigned long nmi_dom0_softirq_reason; -#define NMI_DOM0_PARITY_ERR 0 -#define NMI_DOM0_IO_ERR 1 -#define NMI_DOM0_UNKNOWN 2 - -static void nmi_dom0_softirq(void) -{ - if ( dom0 == NULL ) +static void nmi_softirq(void) +{ + /* Only used to defer wakeup of dom0,vcpu0 to a safe (non-NMI) context. */ + evtchn_notify(dom0->vcpu[0]); +} + +static void nmi_dom0_report(unsigned int reason_idx) +{ + struct domain *d; + + if ( (d = dom0) == NULL ) return; - if ( test_and_clear_bit(NMI_DOM0_PARITY_ERR, &nmi_dom0_softirq_reason) ) - send_guest_virq(dom0->vcpu[0], VIRQ_PARITY_ERR); - - if ( test_and_clear_bit(NMI_DOM0_IO_ERR, &nmi_dom0_softirq_reason) ) - send_guest_virq(dom0->vcpu[0], VIRQ_IO_ERR); - - if ( test_and_clear_bit(NMI_DOM0_UNKNOWN, &nmi_dom0_softirq_reason) ) - send_guest_virq(dom0->vcpu[0], VIRQ_NMI); + set_bit(reason_idx, &d->shared_info->arch.nmi_reason); + + if ( test_and_set_bit(_VCPUF_nmi_pending, &d->vcpu[0]->vcpu_flags) ) + raise_softirq(NMI_SOFTIRQ); /* not safe to wake up a vcpu here */ } asmlinkage void mem_parity_error(struct cpu_user_regs *regs) @@ -1107,8 +1123,7 @@ switch ( opt_nmi[0] ) { case 'd': /* 'dom0' */ - set_bit(NMI_DOM0_PARITY_ERR, &nmi_dom0_softirq_reason); - raise_softirq(NMI_DOM0_SOFTIRQ); + nmi_dom0_report(_XEN_NMIREASON_parity_error); case 'i': /* 'ignore' */ break; default: /* 'fatal' */ @@ -1127,8 +1142,7 @@ switch ( opt_nmi[0] ) { case 'd': /* 'dom0' */ - set_bit(NMI_DOM0_IO_ERR, &nmi_dom0_softirq_reason); - raise_softirq(NMI_DOM0_SOFTIRQ); + nmi_dom0_report(_XEN_NMIREASON_io_error); case 'i': /* 'ignore' */ break; default: /* 'fatal' */ @@ -1147,8 +1161,7 @@ switch ( opt_nmi[0] ) { case 'd': /* 'dom0' */ - set_bit(NMI_DOM0_UNKNOWN, &nmi_dom0_softirq_reason); - raise_softirq(NMI_DOM0_SOFTIRQ); + nmi_dom0_report(_XEN_NMIREASON_unknown); case 'i': /* 'ignore' */ break; default: /* 'fatal' */ @@ -1347,7 +1360,7 @@ cpu_init(); - open_softirq(NMI_DOM0_SOFTIRQ, nmi_dom0_softirq); + open_softirq(NMI_SOFTIRQ, nmi_softirq); } diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/vmx.c --- a/xen/arch/x86/vmx.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/vmx.c Fri Jan 13 20:12:24 2006 @@ -42,7 +42,7 @@ #include <asm/shadow_64.h> #endif #include <public/sched.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> #include <asm/vmx_vpic.h> #include <asm/vmx_vlapic.h> @@ -53,7 +53,7 @@ integer_param("vmx_debug", opt_vmx_debug_level); static unsigned long trace_values[NR_CPUS][4]; -#define TRACE_VMEXIT(index,value) trace_values[current->processor][index]=value +#define TRACE_VMEXIT(index,value) trace_values[smp_processor_id()][index]=value static int vmx_switch_on; @@ -65,11 +65,6 @@ { struct domain *d = v->domain; struct vcpu *vc; - - d->arch.vmx_platform.lapic_enable = v->arch.guest_context.user_regs.ecx; - v->arch.guest_context.user_regs.ecx = 0; - VMX_DBG_LOG(DBG_LEVEL_VLAPIC, "lapic enable is %d.\n", - d->arch.vmx_platform.lapic_enable); /* Initialize monitor page table */ for_each_vcpu(d, vc) @@ -95,7 +90,7 @@ void vmx_relinquish_resources(struct vcpu *v) { struct vmx_virpit *vpit; - + if ( !VMX_DOMAIN(v) ) return; @@ -103,19 +98,18 @@ /* unmap IO shared page */ struct domain *d = v->domain; if ( d->arch.vmx_platform.shared_page_va ) - unmap_domain_page((void *)d->arch.vmx_platform.shared_page_va); + unmap_domain_page_global( + (void *)d->arch.vmx_platform.shared_page_va); } destroy_vmcs(&v->arch.arch_vmx); free_monitor_pagetable(v); vpit = &v->domain->arch.vmx_platform.vmx_pit; - if ( active_ac_timer(&(vpit->pit_timer)) ) - rem_ac_timer(&vpit->pit_timer); - if ( active_ac_timer(&v->arch.arch_vmx.hlt_timer) ) - rem_ac_timer(&v->arch.arch_vmx.hlt_timer); + kill_timer(&vpit->pit_timer); + kill_timer(&v->arch.arch_vmx.hlt_timer); if ( vmx_apic_support(v->domain) && (VLAPIC(v) != NULL) ) { - rem_ac_timer(&VLAPIC(v)->vlapic_timer); + kill_timer(&VLAPIC(v)->vlapic_timer); xfree(VLAPIC(v)); } } @@ -1604,7 +1598,7 @@ next_wakeup = next_pit; } if ( next_wakeup != - 1 ) - set_ac_timer(&current->arch.arch_vmx.hlt_timer, next_wakeup); + set_timer(&current->arch.arch_vmx.hlt_timer, next_wakeup); do_block(); } @@ -1955,9 +1949,12 @@ asmlinkage void trace_vmentry (void) { - TRACE_5D(TRC_VMENTRY,trace_values[current->processor][0], - trace_values[current->processor][1],trace_values[current->processor][2], - trace_values[current->processor][3],trace_values[current->processor][4]); + TRACE_5D(TRC_VMENTRY, + trace_values[smp_processor_id()][0], + trace_values[smp_processor_id()][1], + trace_values[smp_processor_id()][2], + trace_values[smp_processor_id()][3], + trace_values[smp_processor_id()][4]); TRACE_VMEXIT(0,9); TRACE_VMEXIT(1,9); TRACE_VMEXIT(2,9); diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/vmx_intercept.c --- a/xen/arch/x86/vmx_intercept.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/vmx_intercept.c Fri Jan 13 20:12:24 2006 @@ -24,7 +24,7 @@ #include <asm/vmx_vpit.h> #include <asm/vmx_intercept.h> #include <asm/vmx_vlapic.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> #include <xen/lib.h> #include <xen/sched.h> #include <asm/current.h> @@ -356,19 +356,19 @@ vpit->pending_intr_nr++; if ( test_bit(_VCPUF_running, &v->vcpu_flags) ) { vpit->scheduled += vpit->period; - set_ac_timer(&vpit->pit_timer, vpit->scheduled); + set_timer(&vpit->pit_timer, vpit->scheduled); } } void pickup_deactive_ticks(struct vmx_virpit *vpit) { - if ( !active_ac_timer(&(vpit->pit_timer)) ) { + if ( !active_timer(&(vpit->pit_timer)) ) { /* pick up missed timer tick */ missed_ticks(vpit); vpit->scheduled += vpit->period; - set_ac_timer(&vpit->pit_timer, vpit->scheduled); + set_timer(&vpit->pit_timer, vpit->scheduled); } } @@ -385,14 +385,14 @@ /* load init count*/ if (p->state == STATE_IORESP_HOOK) { /* set up actimer, handle re-init */ - if ( active_ac_timer(&(vpit->pit_timer)) ) { + if ( active_timer(&(vpit->pit_timer)) ) { VMX_DBG_LOG(DBG_LEVEL_1, "VMX_PIT: guest reset PIT with channel %lx!\n", (unsigned long) ((p->u.data >> 24) & 0x3) ); - rem_ac_timer(&(vpit->pit_timer)); + stop_timer(&(vpit->pit_timer)); reinit = 1; } else { - init_ac_timer(&vpit->pit_timer, pit_timer_fn, v, v->processor); + init_timer(&vpit->pit_timer, pit_timer_fn, v, v->processor); } /* init count for this channel */ @@ -431,7 +431,7 @@ } vpit->scheduled = NOW() + vpit->period; - set_ac_timer(&vpit->pit_timer, vpit->scheduled); + set_timer(&vpit->pit_timer, vpit->scheduled); /*restore the state*/ p->state = STATE_IORESP_READY; diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/vmx_io.c --- a/xen/arch/x86/vmx_io.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/vmx_io.c Fri Jan 13 20:12:24 2006 @@ -37,7 +37,7 @@ #include <asm/shadow.h> #include <asm/vmx_vpic.h> #include <asm/vmx_vlapic.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> #ifdef CONFIG_VMX #if defined (__i386__) @@ -819,7 +819,7 @@ if ( !vpit->first_injected ) { vpit->pending_intr_nr = 0; vpit->scheduled = NOW() + vpit->period; - set_ac_timer(&vpit->pit_timer, vpit->scheduled); + set_timer(&vpit->pit_timer, vpit->scheduled); vpit->first_injected = 1; } else { vpit->pending_intr_nr--; diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/vmx_platform.c --- a/xen/arch/x86/vmx_platform.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/vmx_platform.c Fri Jan 13 20:12:24 2006 @@ -27,7 +27,7 @@ #include <xen/trace.h> #include <asm/vmx.h> #include <asm/vmx_platform.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> #include <xen/lib.h> #include <xen/sched.h> diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/vmx_vlapic.c --- a/xen/arch/x86/vmx_vlapic.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/vmx_vlapic.c Fri Jan 13 20:12:24 2006 @@ -32,7 +32,7 @@ #include <xen/lib.h> #include <xen/sched.h> #include <asm/current.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> #ifdef CONFIG_VMX @@ -62,7 +62,7 @@ int vmx_apic_support(struct domain *d) { - return d->arch.vmx_platform.lapic_enable; + return d->arch.vmx_platform.apic_enabled; } s_time_t get_apictime_scheduled(struct vcpu *v) @@ -391,7 +391,7 @@ (262144 / get_apic_bus_scale()) * vlapic->timer_divide_counter; vlapic->vlapic_timer.expires = cur + offset; - set_ac_timer(&(vlapic->vlapic_timer), vlapic->vlapic_timer.expires ); + set_timer(&(vlapic->vlapic_timer), vlapic->vlapic_timer.expires ); VMX_DBG_LOG(DBG_LEVEL_VLAPIC, "vlapic_begin_timer: " "bus_scale %x now %08x%08x expire %08x%08x " @@ -739,7 +739,7 @@ case APIC_TMICT: if (vlapic_timer_active(vlapic)) - rem_ac_timer(&(vlapic->vlapic_timer)); + stop_timer(&(vlapic->vlapic_timer)); vlapic->timer_initial = val; vlapic->timer_current = val; @@ -846,7 +846,7 @@ vlapic->timer_current = vlapic->timer_initial; offset = vlapic->timer_current * (262144/get_apic_bus_scale()) * vlapic->timer_divide_counter; vlapic->vlapic_timer.expires = NOW() + offset; - set_ac_timer(&(vlapic->vlapic_timer), vlapic->vlapic_timer.expires); + set_timer(&(vlapic->vlapic_timer), vlapic->vlapic_timer.expires); }else { vlapic->timer_current = 0; } @@ -986,7 +986,7 @@ vmx_vioapic_add_lapic(vlapic, v); - init_ac_timer(&vlapic->vlapic_timer, + init_timer(&vlapic->vlapic_timer, vlapic_timer_fn, vlapic, v->processor); #ifdef VLAPIC_NO_BIOS diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/vmx_vmcs.c --- a/xen/arch/x86/vmx_vmcs.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/vmx_vmcs.c Fri Jan 13 20:12:24 2006 @@ -32,7 +32,7 @@ #include <asm/flushtlb.h> #include <xen/event.h> #include <xen/kernel.h> -#include <public/io/ioreq.h> +#include <public/hvm/hvm_info_table.h> #if CONFIG_PAGING_LEVELS >= 4 #include <asm/shadow_64.h> #endif @@ -193,7 +193,7 @@ domain_crash_synchronous(); } - p = map_domain_page(mpfn); + p = map_domain_page_global(mpfn); if (p == NULL) { printk("Can not map io request shared page for VMX domain.\n"); domain_crash_synchronous(); @@ -206,35 +206,55 @@ &d->shared_info->evtchn_mask[0]); } -#define VCPU_NR_PAGE 0x0009F000 -#define VCPU_NR_OFFSET 0x00000800 -#define VCPU_MAGIC 0x76637075 /* "vcpu" */ - -static void vmx_set_vcpu_nr(struct domain *d) +static int validate_hvm_info(struct hvm_info_table *t) +{ + char signature[] = "HVM INFO"; + uint8_t *ptr = (uint8_t *)t; + uint8_t sum = 0; + int i; + + /* strncmp(t->signature, "HVM INFO", 8) */ + for ( i = 0; i < 8; i++ ) { + if ( signature[i] != t->signature[i] ) { + printk("Bad hvm info signature\n"); + return 0; + } + } + + for ( i = 0; i < t->length; i++ ) + sum += ptr[i]; + + return (sum == 0); +} + +static void vmx_get_hvm_info(struct domain *d) { unsigned char *p; unsigned long mpfn; - unsigned int *vcpus; - - mpfn = get_mfn_from_pfn(VCPU_NR_PAGE >> PAGE_SHIFT); - if (mpfn == INVALID_MFN) { - printk("Can not get vcpu number page mfn for VMX domain.\n"); + struct hvm_info_table *t; + + mpfn = get_mfn_from_pfn(HVM_INFO_PFN); + if ( mpfn == INVALID_MFN ) { + printk("Can not get hvm info page mfn for VMX domain.\n"); domain_crash_synchronous(); } p = map_domain_page(mpfn); - if (p == NULL) { - printk("Can not map vcpu number page for VMX domain.\n"); - domain_crash_synchronous(); - } - - vcpus = (unsigned int *)(p + VCPU_NR_OFFSET); - if (vcpus[0] != VCPU_MAGIC) { - printk("Bad vcpus magic, set vcpu number to 1 by default.\n"); - d->arch.vmx_platform.nr_vcpu = 1; - } - - d->arch.vmx_platform.nr_vcpu = vcpus[1]; + if ( p == NULL ) { + printk("Can not map hvm info page for VMX domain.\n"); + domain_crash_synchronous(); + } + + t = (struct hvm_info_table *)(p + HVM_INFO_OFFSET); + + if ( validate_hvm_info(t) ) { + d->arch.vmx_platform.nr_vcpus = t->nr_vcpus; + d->arch.vmx_platform.apic_enabled = t->apic_enabled; + } else { + printk("Bad hvm info table\n"); + d->arch.vmx_platform.nr_vcpus = 1; + d->arch.vmx_platform.apic_enabled = 0; + } unmap_domain_page(p); } @@ -244,10 +264,10 @@ struct vmx_platform *platform; vmx_map_io_shared_page(d); - vmx_set_vcpu_nr(d); + vmx_get_hvm_info(d); platform = &d->arch.vmx_platform; - pic_init(&platform->vmx_pic, pic_irq_request, + pic_init(&platform->vmx_pic, pic_irq_request, &platform->interrupt_request); register_pic_io_hook(); @@ -321,7 +341,7 @@ vlapic_init(v); vmx_set_host_env(v); - init_ac_timer(&v->arch.arch_vmx.hlt_timer, hlt_timer_fn, v, v->processor); + init_timer(&v->arch.arch_vmx.hlt_timer, hlt_timer_fn, v, v->processor); error |= __vmwrite(GUEST_LDTR_SELECTOR, 0); error |= __vmwrite(GUEST_LDTR_BASE, 0); @@ -335,6 +355,8 @@ __vmwrite(HOST_RSP, (unsigned long)get_stack_bottom()); v->arch.schedule_tail = arch_vmx_do_resume; + v->arch.arch_vmx.launch_cpu = smp_processor_id(); + /* init guest tsc to start from 0 */ rdtscll(host_tsc); v->arch.arch_vmx.tsc_offset = 0 - host_tsc; @@ -617,11 +639,21 @@ void arch_vmx_do_resume(struct vcpu *v) { - u64 vmcs_phys_ptr = (u64) virt_to_phys(v->arch.arch_vmx.vmcs); - - load_vmcs(&v->arch.arch_vmx, vmcs_phys_ptr); - vmx_do_resume(v); - reset_stack_and_jump(vmx_asm_do_resume); + if ( v->arch.arch_vmx.launch_cpu == smp_processor_id() ) + { + load_vmcs(&v->arch.arch_vmx, virt_to_phys(v->arch.arch_vmx.vmcs)); + vmx_do_resume(v); + reset_stack_and_jump(vmx_asm_do_resume); + } + else + { + __vmpclear(virt_to_phys(v->arch.arch_vmx.vmcs)); + load_vmcs(&v->arch.arch_vmx, virt_to_phys(v->arch.arch_vmx.vmcs)); + vmx_do_resume(v); + vmx_set_host_env(v); + v->arch.arch_vmx.launch_cpu = smp_processor_id(); + reset_stack_and_jump(vmx_asm_do_relaunch); + } } void arch_vmx_do_launch(struct vcpu *v) @@ -641,18 +673,6 @@ } vmx_do_launch(v); reset_stack_and_jump(vmx_asm_do_launch); -} - -void arch_vmx_do_relaunch(struct vcpu *v) -{ - u64 vmcs_phys_ptr = (u64) virt_to_phys(v->arch.arch_vmx.vmcs); - - load_vmcs(&v->arch.arch_vmx, vmcs_phys_ptr); - vmx_do_resume(v); - vmx_set_host_env(v); - v->arch.schedule_tail = arch_vmx_do_resume; - - reset_stack_and_jump(vmx_asm_do_relaunch); } #endif /* CONFIG_VMX */ diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/x86_32/asm-offsets.c --- a/xen/arch/x86/x86_32/asm-offsets.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/x86_32/asm-offsets.c Fri Jan 13 20:12:24 2006 @@ -65,6 +65,10 @@ arch.guest_context.kernel_ss); OFFSET(VCPU_kernel_sp, struct vcpu, arch.guest_context.kernel_sp); + OFFSET(VCPU_flags, struct vcpu, vcpu_flags); + OFFSET(VCPU_nmi_addr, struct vcpu, nmi_addr); + DEFINE(_VCPUF_nmi_pending, _VCPUF_nmi_pending); + DEFINE(_VCPUF_nmi_masked, _VCPUF_nmi_masked); BLANK(); OFFSET(VCPUINFO_upcall_pending, vcpu_info_t, evtchn_upcall_pending); diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/x86_32/domain_page.c --- a/xen/arch/x86/x86_32/domain_page.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/x86_32/domain_page.c Fri Jan 13 20:12:24 2006 @@ -1,14 +1,9 @@ /****************************************************************************** * domain_page.h * - * Allow temporary mapping of domain pages. Based on ideas from the - * Linux PKMAP code -- the copyrights and credits are retained below. - */ - -/* - * (C) 1999 Andrea Arcangeli, SuSE GmbH, andrea@xxxxxxx - * Gerhard Wichert, Siemens AG, Gerhard.Wichert@xxxxxxxxxxxxxx * - * Copyright (C) 1999 Ingo Molnar <mingo@xxxxxxxxxx> + * Allow temporary mapping of domain pages. + * + * Copyright (c) 2003-2006, Keir Fraser <keir@xxxxxxxxxxxxx> */ #include <xen/config.h> @@ -20,80 +15,203 @@ #include <asm/flushtlb.h> #include <asm/hardirq.h> -#define MAPCACHE_ORDER 10 -#define MAPCACHE_ENTRIES (1 << MAPCACHE_ORDER) - -l1_pgentry_t *mapcache; -static unsigned int map_idx, epoch, shadow_epoch[NR_CPUS]; -static spinlock_t map_lock = SPIN_LOCK_UNLOCKED; - -/* Use a spare PTE bit to mark entries ready for recycling. */ -#define READY_FOR_TLB_FLUSH (1<<10) - -static void flush_all_ready_maps(void) -{ - l1_pgentry_t *cache = mapcache; - unsigned int i; - - for ( i = 0; i < MAPCACHE_ENTRIES; i++ ) - if ( (l1e_get_flags(cache[i]) & READY_FOR_TLB_FLUSH) ) - cache[i] = l1e_empty(); -} - -void *map_domain_pages(unsigned long pfn, unsigned int order) +void *map_domain_page(unsigned long pfn) { unsigned long va; - unsigned int idx, i, flags, cpu = smp_processor_id(); - l1_pgentry_t *cache = mapcache; -#ifndef NDEBUG - unsigned int flush_count = 0; -#endif + unsigned int idx, i, vcpu = current->vcpu_id; + struct domain *d; + struct mapcache *cache; + struct vcpu_maphash_entry *hashent; ASSERT(!in_irq()); + perfc_incrc(map_domain_page_count); - spin_lock(&map_lock); + /* If we are the idle domain, ensure that we run on our own page tables. */ + d = current->domain; + if ( unlikely(is_idle_domain(d)) ) + __sync_lazy_execstate(); + + cache = &d->arch.mapcache; + + hashent = &cache->vcpu_maphash[vcpu].hash[MAPHASH_HASHFN(pfn)]; + if ( hashent->pfn == pfn ) + { + idx = hashent->idx; + hashent->refcnt++; + ASSERT(hashent->refcnt != 0); + ASSERT(l1e_get_pfn(cache->l1tab[idx]) == pfn); + goto out; + } + + spin_lock(&cache->lock); /* Has some other CPU caused a wrap? We must flush if so. */ - if ( epoch != shadow_epoch[cpu] ) - { + if ( unlikely(cache->epoch != cache->shadow_epoch[vcpu]) ) + { + cache->shadow_epoch[vcpu] = cache->epoch; + if ( NEED_FLUSH(tlbflush_time[smp_processor_id()], + cache->tlbflush_timestamp) ) + { + perfc_incrc(domain_page_tlb_flush); + local_flush_tlb(); + } + } + + idx = find_next_zero_bit(cache->inuse, MAPCACHE_ENTRIES, cache->cursor); + if ( unlikely(idx >= MAPCACHE_ENTRIES) ) + { + /* /First/, clean the garbage map and update the inuse list. */ + for ( i = 0; i < ARRAY_SIZE(cache->garbage); i++ ) + { + unsigned long x = xchg(&cache->garbage[i], 0); + cache->inuse[i] &= ~x; + } + + /* /Second/, flush TLBs. */ perfc_incrc(domain_page_tlb_flush); local_flush_tlb(); - shadow_epoch[cpu] = epoch; - } - - do { - idx = map_idx = (map_idx + 1) & (MAPCACHE_ENTRIES - 1); - if ( unlikely(idx == 0) ) - { - ASSERT(flush_count++ == 0); - flush_all_ready_maps(); - perfc_incrc(domain_page_tlb_flush); - local_flush_tlb(); - shadow_epoch[cpu] = ++epoch; - } - - flags = 0; - for ( i = 0; i < (1U << order); i++ ) - flags |= l1e_get_flags(cache[idx+i]); - } - while ( flags & _PAGE_PRESENT ); - - for ( i = 0; i < (1U << order); i++ ) - cache[idx+i] = l1e_from_pfn(pfn+i, __PAGE_HYPERVISOR); - - spin_unlock(&map_lock); - + cache->shadow_epoch[vcpu] = ++cache->epoch; + cache->tlbflush_timestamp = tlbflush_current_time(); + + idx = find_first_zero_bit(cache->inuse, MAPCACHE_ENTRIES); + ASSERT(idx < MAPCACHE_ENTRIES); + } + + set_bit(idx, cache->inuse); + cache->cursor = idx + 1; + + spin_unlock(&cache->lock); + + cache->l1tab[idx] = l1e_from_pfn(pfn, __PAGE_HYPERVISOR); + + out: va = MAPCACHE_VIRT_START + (idx << PAGE_SHIFT); return (void *)va; } -void unmap_domain_pages(void *va, unsigned int order) -{ - unsigned int idx, i; +void unmap_domain_page(void *va) +{ + unsigned int idx; + struct mapcache *cache = &current->domain->arch.mapcache; + unsigned long pfn; + struct vcpu_maphash_entry *hashent; + + ASSERT(!in_irq()); + ASSERT((void *)MAPCACHE_VIRT_START <= va); ASSERT(va < (void *)MAPCACHE_VIRT_END); + idx = ((unsigned long)va - MAPCACHE_VIRT_START) >> PAGE_SHIFT; - for ( i = 0; i < (1U << order); i++ ) - l1e_add_flags(mapcache[idx+i], READY_FOR_TLB_FLUSH); -} + pfn = l1e_get_pfn(cache->l1tab[idx]); + hashent = &cache->vcpu_maphash[current->vcpu_id].hash[MAPHASH_HASHFN(pfn)]; + + if ( hashent->idx == idx ) + { + ASSERT(hashent->pfn == pfn); + ASSERT(hashent->refcnt != 0); + hashent->refcnt--; + } + else if ( hashent->refcnt == 0 ) + { + if ( hashent->idx != MAPHASHENT_NOTINUSE ) + { + /* /First/, zap the PTE. */ + ASSERT(l1e_get_pfn(cache->l1tab[hashent->idx]) == hashent->pfn); + cache->l1tab[hashent->idx] = l1e_empty(); + /* /Second/, mark as garbage. */ + set_bit(hashent->idx, cache->garbage); + } + + /* Add newly-freed mapping to the maphash. */ + hashent->pfn = pfn; + hashent->idx = idx; + } + else + { + /* /First/, zap the PTE. */ + cache->l1tab[idx] = l1e_empty(); + /* /Second/, mark as garbage. */ + set_bit(idx, cache->garbage); + } +} + +void mapcache_init(struct domain *d) +{ + unsigned int i, j; + + d->arch.mapcache.l1tab = d->arch.mm_perdomain_pt + + (GDT_LDT_MBYTES << (20 - PAGE_SHIFT)); + spin_lock_init(&d->arch.mapcache.lock); + + /* Mark all maphash entries as not in use. */ + for ( i = 0; i < MAX_VIRT_CPUS; i++ ) + for ( j = 0; j < MAPHASH_ENTRIES; j++ ) + d->arch.mapcache.vcpu_maphash[i].hash[j].idx = + MAPHASHENT_NOTINUSE; +} + +#define GLOBALMAP_BITS (IOREMAP_MBYTES << (20 - PAGE_SHIFT)) +static unsigned long inuse[BITS_TO_LONGS(GLOBALMAP_BITS)]; +static unsigned long garbage[BITS_TO_LONGS(GLOBALMAP_BITS)]; +static unsigned int inuse_cursor; +static spinlock_t globalmap_lock = SPIN_LOCK_UNLOCKED; + +void *map_domain_page_global(unsigned long pfn) +{ + l2_pgentry_t *pl2e; + l1_pgentry_t *pl1e; + unsigned int idx, i; + unsigned long va; + + ASSERT(!in_irq() && local_irq_is_enabled()); + + spin_lock(&globalmap_lock); + + idx = find_next_zero_bit(inuse, GLOBALMAP_BITS, inuse_cursor); + va = IOREMAP_VIRT_START + (idx << PAGE_SHIFT); + if ( unlikely(va >= FIXADDR_START) ) + { + /* /First/, clean the garbage map and update the inuse list. */ + for ( i = 0; i < ARRAY_SIZE(garbage); i++ ) + { + unsigned long x = xchg(&garbage[i], 0); + inuse[i] &= ~x; + } + + /* /Second/, flush all TLBs to get rid of stale garbage mappings. */ + flush_tlb_all(); + + idx = find_first_zero_bit(inuse, GLOBALMAP_BITS); + va = IOREMAP_VIRT_START + (idx << PAGE_SHIFT); + ASSERT(va < FIXADDR_START); + } + + set_bit(idx, inuse); + inuse_cursor = idx + 1; + + spin_unlock(&globalmap_lock); + + pl2e = virt_to_xen_l2e(va); + pl1e = l2e_to_l1e(*pl2e) + l1_table_offset(va); + *pl1e = l1e_from_pfn(pfn, __PAGE_HYPERVISOR); + + return (void *)va; +} + +void unmap_domain_page_global(void *va) +{ + unsigned long __va = (unsigned long)va; + l2_pgentry_t *pl2e; + l1_pgentry_t *pl1e; + unsigned int idx; + + /* /First/, we zap the PTE. */ + pl2e = virt_to_xen_l2e(__va); + pl1e = l2e_to_l1e(*pl2e) + l1_table_offset(__va); + *pl1e = l1e_empty(); + + /* /Second/, we add to the garbage map. */ + idx = (__va - IOREMAP_VIRT_START) >> PAGE_SHIFT; + set_bit(idx, garbage); +} diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/x86_32/entry.S --- a/xen/arch/x86/x86_32/entry.S Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/x86_32/entry.S Fri Jan 13 20:12:24 2006 @@ -326,7 +326,9 @@ shl $IRQSTAT_shift,%eax test %ecx,irq_stat(%eax,1) jnz process_softirqs -/*test_guest_events:*/ + btr $_VCPUF_nmi_pending,VCPU_flags(%ebx) + jc process_nmi +test_guest_events: movl VCPU_vcpu_info(%ebx),%eax testb $0xFF,VCPUINFO_upcall_mask(%eax) jnz restore_all_guest @@ -348,7 +350,24 @@ sti call do_softirq jmp test_all_events - + + ALIGN +process_nmi: + movl VCPU_nmi_addr(%ebx),%eax + test %eax,%eax + jz test_all_events + bts $_VCPUF_nmi_masked,VCPU_flags(%ebx) + jc 1f + sti + leal VCPU_trap_bounce(%ebx),%edx + movl %eax,TRAPBOUNCE_eip(%edx) + movw $FLAT_KERNEL_CS,TRAPBOUNCE_cs(%edx) + movw $TBF_INTERRUPT,TRAPBOUNCE_flags(%edx) + call create_bounce_frame + jmp test_all_events +1: bts $_VCPUF_nmi_pending,VCPU_flags(%ebx) + jmp test_guest_events + /* CREATE A BASIC EXCEPTION FRAME ON GUEST OS (RING-1) STACK: */ /* {EIP, CS, EFLAGS, [ESP, SS]} */ /* %edx == trap_bounce, %ebx == struct vcpu */ @@ -620,9 +639,7 @@ jne defer_nmi continue_nmi: - movl $(__HYPERVISOR_DS),%edx - movl %edx,%ds - movl %edx,%es + SET_XEN_SEGMENTS(d) movl %esp,%edx pushl %edx call do_nmi @@ -659,42 +676,6 @@ GET_GUEST_REGS(%ecx) movl %eax,UREGS_eax(%ecx) jmp do_sched_op - -do_switch_vm86: - # Reset the stack pointer - GET_GUEST_REGS(%ecx) - movl %ecx,%esp - - # GS:ESI == Ring-1 stack activation - movl UREGS_esp(%esp),%esi -VFLT1: mov UREGS_ss(%esp),%gs - - # ES:EDI == Ring-0 stack activation - leal UREGS_eip(%esp),%edi - - # Restore the hypercall-number-clobbered EAX on our stack frame -VFLT2: movl %gs:(%esi),%eax - movl %eax,UREGS_eax(%esp) - addl $4,%esi - - # Copy the VM86 activation from the ring-1 stack to the ring-0 stack - movl $(UREGS_user_sizeof-UREGS_eip)/4,%ecx -VFLT3: movl %gs:(%esi),%eax - stosl - addl $4,%esi - loop VFLT3 - - # Fix up EFLAGS: IOPL=0, IF=1, VM=1 - andl $~X86_EFLAGS_IOPL,UREGS_eflags(%esp) - orl $X86_EFLAGS_IF|X86_EFLAGS_VM,UREGS_eflags(%esp) - - jmp test_all_events - -.section __ex_table,"a" - .long VFLT1,domain_crash_synchronous - .long VFLT2,domain_crash_synchronous - .long VFLT3,domain_crash_synchronous -.previous .data @@ -744,11 +725,12 @@ .long do_grant_table_op /* 20 */ .long do_vm_assist .long do_update_va_mapping_otherdomain - .long do_switch_vm86 + .long do_iret .long do_vcpu_op .long do_ni_hypercall /* 25 */ .long do_mmuext_op - .long do_acm_op /* 27 */ + .long do_acm_op + .long do_nmi_op .rept NR_hypercalls-((.-hypercall_table)/4) .long do_ni_hypercall .endr @@ -777,11 +759,12 @@ .byte 3 /* do_grant_table_op */ /* 20 */ .byte 2 /* do_vm_assist */ .byte 5 /* do_update_va_mapping_otherdomain */ - .byte 0 /* do_switch_vm86 */ + .byte 0 /* do_iret */ .byte 3 /* do_vcpu_op */ .byte 0 /* do_ni_hypercall */ /* 25 */ .byte 4 /* do_mmuext_op */ .byte 1 /* do_acm_op */ + .byte 2 /* do_nmi_op */ .rept NR_hypercalls-(.-hypercall_args_table) .byte 0 /* do_ni_hypercall */ .endr diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/x86_32/mm.c --- a/xen/arch/x86/x86_32/mm.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/x86_32/mm.c Fri Jan 13 20:12:24 2006 @@ -29,8 +29,6 @@ #include <asm/fixmap.h> #include <public/memory.h> -extern l1_pgentry_t *mapcache; - unsigned int PAGE_HYPERVISOR = __PAGE_HYPERVISOR; unsigned int PAGE_HYPERVISOR_NOCACHE = __PAGE_HYPERVISOR_NOCACHE; @@ -68,7 +66,7 @@ void *ioremap_pt; unsigned long v; struct pfn_info *pg; - int i, mapcache_order; + int i; #ifdef CONFIG_X86_PAE printk("PAE enabled, limit: %d GB\n", MACHPHYS_MBYTES); @@ -76,7 +74,7 @@ printk("PAE disabled.\n"); #endif - idle0_vcpu.arch.monitor_table = mk_pagetable(__pa(idle_pg_table)); + idle_vcpu[0]->arch.monitor_table = mk_pagetable(__pa(idle_pg_table)); if ( cpu_has_pge ) { @@ -121,14 +119,12 @@ l2e_from_page(virt_to_page(ioremap_pt), __PAGE_HYPERVISOR); } - /* Set up mapping cache for domain pages. */ - mapcache_order = get_order_from_bytes( - MAPCACHE_MBYTES << (20 - PAGETABLE_ORDER)); - mapcache = alloc_xenheap_pages(mapcache_order); - memset(mapcache, 0, PAGE_SIZE << mapcache_order); - for ( i = 0; i < (MAPCACHE_MBYTES >> (L2_PAGETABLE_SHIFT - 20)); i++ ) - idle_pg_table_l2[l2_linear_offset(MAPCACHE_VIRT_START) + i] = - l2e_from_page(virt_to_page(mapcache) + i, __PAGE_HYPERVISOR); + /* Install per-domain mappings for idle domain. */ + for ( i = 0; i < PDPT_L2_ENTRIES; i++ ) + idle_pg_table_l2[l2_linear_offset(PERDOMAIN_VIRT_START) + i] = + l2e_from_page(virt_to_page(idle_vcpu[0]->domain-> + arch.mm_perdomain_pt) + i, + __PAGE_HYPERVISOR); } void __init zap_low_mappings(l2_pgentry_t *base) diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/x86_32/traps.c --- a/xen/arch/x86/x86_32/traps.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/x86_32/traps.c Fri Jan 13 20:12:24 2006 @@ -157,6 +157,64 @@ __asm__ __volatile__ ( "hlt" ); } +asmlinkage unsigned long do_iret(void) +{ + struct cpu_user_regs *regs = guest_cpu_user_regs(); + u32 eflags; + + /* Check worst-case stack frame for overlap with Xen protected area. */ + if ( unlikely(!access_ok(regs->esp, 40)) ) + domain_crash_synchronous(); + + /* Pop and restore EAX (clobbered by hypercall). */ + if ( unlikely(__copy_from_user(&regs->eax, (void __user *)regs->esp, 4)) ) + domain_crash_synchronous(); + regs->esp += 4; + + /* Pop and restore CS and EIP. */ + if ( unlikely(__copy_from_user(&regs->eip, (void __user *)regs->esp, 8)) ) + domain_crash_synchronous(); + regs->esp += 8; + + /* + * Pop, fix up and restore EFLAGS. We fix up in a local staging area + * to avoid firing the BUG_ON(IOPL) check in arch_getdomaininfo_ctxt. + */ + if ( unlikely(__copy_from_user(&eflags, (void __user *)regs->esp, 4)) ) + domain_crash_synchronous(); + regs->esp += 4; + regs->eflags = (eflags & ~X86_EFLAGS_IOPL) | X86_EFLAGS_IF; + + if ( VM86_MODE(regs) ) + { + /* Return to VM86 mode: pop and restore ESP,SS,ES,DS,FS and GS. */ + if ( __copy_from_user(&regs->esp, (void __user *)regs->esp, 24) ) + domain_crash_synchronous(); + } + else if ( unlikely(RING_0(regs)) ) + { + domain_crash_synchronous(); + } + else if ( !RING_1(regs) ) + { + /* Return to ring 2/3: pop and restore ESP and SS. */ + if ( __copy_from_user(&regs->esp, (void __user *)regs->esp, 8) ) + domain_crash_synchronous(); + } + + /* No longer in NMI context. */ + clear_bit(_VCPUF_nmi_masked, &current->vcpu_flags); + + /* Restore upcall mask from saved value. */ + current->vcpu_info->evtchn_upcall_mask = regs->saved_upcall_mask; + + /* + * The hypercall exit path will overwrite EAX with this return + * value. + */ + return regs->eax; +} + BUILD_SMP_INTERRUPT(deferred_nmi, TRAP_deferred_nmi) asmlinkage void smp_deferred_nmi(struct cpu_user_regs regs) { diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/x86_64/asm-offsets.c --- a/xen/arch/x86/x86_64/asm-offsets.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/x86_64/asm-offsets.c Fri Jan 13 20:12:24 2006 @@ -65,6 +65,10 @@ arch.guest_context.syscall_callback_eip); OFFSET(VCPU_kernel_sp, struct vcpu, arch.guest_context.kernel_sp); + OFFSET(VCPU_flags, struct vcpu, vcpu_flags); + OFFSET(VCPU_nmi_addr, struct vcpu, nmi_addr); + DEFINE(_VCPUF_nmi_pending, _VCPUF_nmi_pending); + DEFINE(_VCPUF_nmi_masked, _VCPUF_nmi_masked); BLANK(); OFFSET(VCPUINFO_upcall_pending, vcpu_info_t, evtchn_upcall_pending); diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/x86_64/entry.S --- a/xen/arch/x86/x86_64/entry.S Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/x86_64/entry.S Fri Jan 13 20:12:24 2006 @@ -171,7 +171,9 @@ leaq irq_stat(%rip),%rcx testl $~0,(%rcx,%rax,1) jnz process_softirqs -/*test_guest_events:*/ + btr $_VCPUF_nmi_pending,VCPU_flags(%rbx) + jc process_nmi +test_guest_events: movq VCPU_vcpu_info(%rbx),%rax testb $0xFF,VCPUINFO_upcall_mask(%rax) jnz restore_all_guest @@ -322,6 +324,23 @@ call do_softirq jmp test_all_events + ALIGN +/* %rbx: struct vcpu */ +process_nmi: + movq VCPU_nmi_addr(%rbx),%rax + test %rax,%rax + jz test_all_events + bts $_VCPUF_nmi_masked,VCPU_flags(%rbx) + jc 1f + sti + leaq VCPU_trap_bounce(%rbx),%rdx + movq %rax,TRAPBOUNCE_eip(%rdx) + movw $(TBF_INTERRUPT|TBF_SLOW_IRET),TRAPBOUNCE_flags(%rdx) + call create_bounce_frame + jmp test_all_events +1: bts $_VCPUF_nmi_pending,VCPU_flags(%rbx) + jmp test_guest_events + /* CREATE A BASIC EXCEPTION FRAME ON GUEST OS STACK: */ /* { RCX, R11, [DS-GS,] [CR2,] [ERRCODE,] RIP, CS, RFLAGS, RSP, SS } */ /* %rdx: trap_bounce, %rbx: struct vcpu */ @@ -339,6 +358,9 @@ 1: /* In kernel context already: push new frame at existing %rsp. */ movq UREGS_rsp+8(%rsp),%rsi andb $0xfc,UREGS_cs+8(%rsp) # Indicate kernel context to guest. + testw $(TBF_SLOW_IRET),TRAPBOUNCE_flags(%rdx) + jz 2f + orb $0x01,UREGS_cs+8(%rsp) 2: andq $~0xf,%rsi # Stack frames are 16-byte aligned. movq $HYPERVISOR_VIRT_START,%rax cmpq %rax,%rsi @@ -569,7 +591,7 @@ SAVE_ALL movq %rsp,%rdi call do_nmi - jmp restore_all_xen + jmp ret_from_intr do_arch_sched_op: # Ensure we return success even if we return via schedule_tail() @@ -626,11 +648,12 @@ .quad do_grant_table_op /* 20 */ .quad do_vm_assist .quad do_update_va_mapping_otherdomain - .quad do_switch_to_user + .quad do_iret .quad do_vcpu_op .quad do_set_segment_base /* 25 */ .quad do_mmuext_op .quad do_acm_op + .quad do_nmi_op .rept NR_hypercalls-((.-hypercall_table)/4) .quad do_ni_hypercall .endr @@ -659,11 +682,12 @@ .byte 3 /* do_grant_table_op */ /* 20 */ .byte 2 /* do_vm_assist */ .byte 4 /* do_update_va_mapping_otherdomain */ - .byte 0 /* do_switch_to_user */ + .byte 0 /* do_iret */ .byte 3 /* do_vcpu_op */ .byte 2 /* do_set_segment_base */ /* 25 */ .byte 4 /* do_mmuext_op */ .byte 1 /* do_acm_op */ + .byte 2 /* do_nmi_op */ .rept NR_hypercalls-(.-hypercall_args_table) .byte 0 /* do_ni_hypercall */ .endr diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/x86_64/mm.c --- a/xen/arch/x86/x86_64/mm.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/x86_64/mm.c Fri Jan 13 20:12:24 2006 @@ -80,7 +80,7 @@ l2_pgentry_t *l2_ro_mpt; struct pfn_info *pg; - idle0_vcpu.arch.monitor_table = mk_pagetable(__pa(idle_pg_table)); + idle_vcpu[0]->arch.monitor_table = mk_pagetable(__pa(idle_pg_table)); /* Create user-accessible L2 directory to map the MPT for guests. */ l3_ro_mpt = alloc_xenheap_page(); @@ -119,6 +119,12 @@ /* Set up linear page table mapping. */ idle_pg_table[l4_table_offset(LINEAR_PT_VIRT_START)] = l4e_from_paddr(__pa(idle_pg_table), __PAGE_HYPERVISOR); + + /* Install per-domain mappings for idle domain. */ + idle_pg_table[l4_table_offset(PERDOMAIN_VIRT_START)] = + l4e_from_page( + virt_to_page(idle_vcpu[0]->domain->arch.mm_perdomain_l3), + __PAGE_HYPERVISOR); } void __init zap_low_mappings(void) diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/x86_64/traps.c --- a/xen/arch/x86/x86_64/traps.c Fri Jan 13 16:38:44 2006 +++ b/xen/arch/x86/x86_64/traps.c Fri Jan 13 20:12:24 2006 @@ -12,6 +12,7 @@ #include <asm/current.h> #include <asm/flushtlb.h> #include <asm/msr.h> +#include <asm/shadow.h> #include <asm/vmx.h> void show_registers(struct cpu_user_regs *regs) @@ -113,6 +114,52 @@ __asm__ __volatile__ ( "hlt" ); } +void toggle_guest_mode(struct vcpu *v) +{ + v->arch.flags ^= TF_kernel_mode; + __asm__ __volatile__ ( "swapgs" ); + update_pagetables(v); + write_ptbase(v); +} + +long do_iret(void) +{ + struct cpu_user_regs *regs = guest_cpu_user_regs(); + struct iret_context iret_saved; + struct vcpu *v = current; + + if ( unlikely(copy_from_user(&iret_saved, (void *)regs->rsp, + sizeof(iret_saved))) ) + domain_crash_synchronous(); + + /* Returning to user mode? */ + if ( (iret_saved.cs & 3) == 3 ) + { + if ( unlikely(pagetable_get_paddr(v->arch.guest_table_user) == 0) ) + return -EFAULT; + toggle_guest_mode(v); + } + + regs->rip = iret_saved.rip; + regs->cs = iret_saved.cs | 3; /* force guest privilege */ + regs->rflags = (iret_saved.rflags & ~(EF_IOPL|EF_VM)) | EF_IE; + regs->rsp = iret_saved.rsp; + regs->ss = iret_saved.ss | 3; /* force guest privilege */ + + if ( !(iret_saved.flags & VGCF_IN_SYSCALL) ) + { + regs->entry_vector = 0; + regs->r11 = iret_saved.r11; + regs->rcx = iret_saved.rcx; + } + + /* No longer in NMI context. */ + clear_bit(_VCPUF_nmi_masked, &current->vcpu_flags); + + /* Saved %rax gets written back to regs->rax in entry.S. */ + return iret_saved.rax; +} + asmlinkage void syscall_enter(void); void __init percpu_traps_init(void) { diff -r 2884f14895d2 -r f5ab28885c75 xen/common/bitmap.c --- a/xen/common/bitmap.c Fri Jan 13 16:38:44 2006 +++ b/xen/common/bitmap.c Fri Jan 13 20:12:24 2006 @@ -282,6 +282,111 @@ #endif EXPORT_SYMBOL(__bitmap_weight); +/* + * Bitmap printing & parsing functions: first version by Bill Irwin, + * second version by Paul Jackson, third by Joe Korty. + */ + +#define CHUNKSZ 32 +#define nbits_to_hold_value(val) fls(val) +#define roundup_power2(val,modulus) (((val) + (modulus) - 1) & ~((modulus) - 1)) +#define unhex(c) (isdigit(c) ? (c - '0') : (toupper(c) - 'A' + 10)) +#define BASEDEC 10 /* fancier cpuset lists input in decimal */ + +/** + * bitmap_scnprintf - convert bitmap to an ASCII hex string. + * @buf: byte buffer into which string is placed + * @buflen: reserved size of @buf, in bytes + * @maskp: pointer to bitmap to convert + * @nmaskbits: size of bitmap, in bits + * + * Exactly @nmaskbits bits are displayed. Hex digits are grouped into + * comma-separated sets of eight digits per set. + */ +int bitmap_scnprintf(char *buf, unsigned int buflen, + const unsigned long *maskp, int nmaskbits) +{ + int i, word, bit, len = 0; + unsigned long val; + const char *sep = ""; + int chunksz; + u32 chunkmask; + + chunksz = nmaskbits & (CHUNKSZ - 1); + if (chunksz == 0) + chunksz = CHUNKSZ; + + i = roundup_power2(nmaskbits, CHUNKSZ) - CHUNKSZ; + for (; i >= 0; i -= CHUNKSZ) { + chunkmask = ((1ULL << chunksz) - 1); + word = i / BITS_PER_LONG; + bit = i % BITS_PER_LONG; + val = (maskp[word] >> bit) & chunkmask; + len += scnprintf(buf+len, buflen-len, "%s%0*lx", sep, + (chunksz+3)/4, val); + chunksz = CHUNKSZ; + sep = ","; + } + return len; +} +EXPORT_SYMBOL(bitmap_scnprintf); + +/* + * bscnl_emit(buf, buflen, rbot, rtop, bp) + * + * Helper routine for bitmap_scnlistprintf(). Write decimal number + * or range to buf, suppressing output past buf+buflen, with optional + * comma-prefix. Return len of what would be written to buf, if it + * all fit. + */ +static inline int bscnl_emit(char *buf, int buflen, int rbot, int rtop, int len) +{ + if (len > 0) + len += scnprintf(buf + len, buflen - len, ","); + if (rbot == rtop) + len += scnprintf(buf + len, buflen - len, "%d", rbot); + else + len += scnprintf(buf + len, buflen - len, "%d-%d", rbot, rtop); + return len; +} + +/** + * bitmap_scnlistprintf - convert bitmap to list format ASCII string + * @buf: byte buffer into which string is placed + * @buflen: reserved size of @buf, in bytes + * @maskp: pointer to bitmap to convert + * @nmaskbits: size of bitmap, in bits + * + * Output format is a comma-separated list of decimal numbers and + * ranges. Consecutively set bits are shown as two hyphen-separated + * decimal numbers, the smallest and largest bit numbers set in + * the range. Output format is compatible with the format + * accepted as input by bitmap_parselist(). + * + * The return value is the number of characters which would be + * generated for the given input, excluding the trailing '\0', as + * per ISO C99. + */ +int bitmap_scnlistprintf(char *buf, unsigned int buflen, + const unsigned long *maskp, int nmaskbits) +{ + int len = 0; + /* current bit is 'cur', most recently seen range is [rbot, rtop] */ + int cur, rbot, rtop; + + rbot = cur = find_first_bit(maskp, nmaskbits); + while (cur < nmaskbits) { + rtop = cur; + cur = find_next_bit(maskp, nmaskbits, cur+1); + if (cur >= nmaskbits || cur > rtop + 1) { + len = bscnl_emit(buf, buflen, rbot, rtop, len); + rbot = cur; + } + } + return len; +} +EXPORT_SYMBOL(bitmap_scnlistprintf); + /** * bitmap_find_free_region - find a contiguous aligned mem region * @bitmap: an array of unsigned longs corresponding to the bitmap diff -r 2884f14895d2 -r f5ab28885c75 xen/common/dom0_ops.c --- a/xen/common/dom0_ops.c Fri Jan 13 16:38:44 2006 +++ b/xen/common/dom0_ops.c Fri Jan 13 20:12:24 2006 @@ -110,13 +110,13 @@ switch ( op->cmd ) { - case DOM0_SETDOMAININFO: - { - struct domain *d = find_domain_by_id(op->u.setdomaininfo.domain); + case DOM0_SETVCPUCONTEXT: + { + struct domain *d = find_domain_by_id(op->u.setvcpucontext.domain); ret = -ESRCH; if ( d != NULL ) { - ret = set_info_guest(d, &op->u.setdomaininfo); + ret = set_info_guest(d, &op->u.setvcpucontext); put_domain(d); } } @@ -284,11 +284,12 @@ } break; - case DOM0_PINCPUDOMAIN: - { - domid_t dom = op->u.pincpudomain.domain; + case DOM0_SETVCPUAFFINITY: + { + domid_t dom = op->u.setvcpuaffinity.domain; struct domain *d = find_domain_by_id(dom); struct vcpu *v; + cpumask_t new_affinity; if ( d == NULL ) { @@ -296,15 +297,15 @@ break; } - if ( (op->u.pincpudomain.vcpu >= MAX_VIRT_CPUS) || - !d->vcpu[op->u.pincpudomain.vcpu] ) + if ( (op->u.setvcpuaffinity.vcpu >= MAX_VIRT_CPUS) || + !d->vcpu[op->u.setvcpuaffinity.vcpu] ) { ret = -EINVAL; put_domain(d); break; } - v = d->vcpu[op->u.pincpudomain.vcpu]; + v = d->vcpu[op->u.setvcpuaffinity.vcpu]; if ( v == NULL ) { ret = -ESRCH; @@ -319,22 +320,13 @@ break; } - v->cpumap = op->u.pincpudomain.cpumap; - - if ( v->cpumap == CPUMAP_RUNANYWHERE ) - { - clear_bit(_VCPUF_cpu_pinned, &v->vcpu_flags); - } - else - { - /* pick a new cpu from the usable map */ - int new_cpu; - new_cpu = (int)find_first_set_bit(v->cpumap) % num_online_cpus(); - vcpu_pause(v); - vcpu_migrate_cpu(v, new_cpu); - set_bit(_VCPUF_cpu_pinned, &v->vcpu_flags); - vcpu_unpause(v); - } + new_affinity = v->cpu_affinity; + memcpy(cpus_addr(new_affinity), + &op->u.setvcpuaffinity.cpumap, + min((int)(BITS_TO_LONGS(NR_CPUS) * sizeof(long)), + (int)sizeof(op->u.setvcpuaffinity.cpumap))); + + ret = vcpu_set_affinity(v, &new_affinity); put_domain(d); } @@ -506,7 +498,11 @@ op->u.getvcpuinfo.running = test_bit(_VCPUF_running, &v->vcpu_flags); op->u.getvcpuinfo.cpu_time = v->cpu_time; op->u.getvcpuinfo.cpu = v->processor; - op->u.getvcpuinfo.cpumap = v->cpumap; + op->u.getvcpuinfo.cpumap = 0; + memcpy(&op->u.getvcpuinfo.cpumap, + cpus_addr(v->cpu_affinity), + min((int)(BITS_TO_LONGS(NR_CPUS) * sizeof(long)), + (int)sizeof(op->u.getvcpuinfo.cpumap))); ret = 0; if ( copy_to_user(u_dom0_op, op, sizeof(*op)) ) diff -r 2884f14895d2 -r f5ab28885c75 xen/common/domain.c --- a/xen/common/domain.c Fri Jan 13 16:38:44 2006 +++ b/xen/common/domain.c Fri Jan 13 20:12:24 2006 @@ -46,12 +46,10 @@ INIT_LIST_HEAD(&d->page_list); INIT_LIST_HEAD(&d->xenpage_list); - if ( d->domain_id == IDLE_DOMAIN_ID ) - set_bit(_DOMF_idle_domain, &d->domain_flags); - else + if ( !is_idle_domain(d) ) set_bit(_DOMF_ctrl_pause, &d->domain_flags); - if ( !is_idle_task(d) && + if ( !is_idle_domain(d) && ((evtchn_init(d) != 0) || (grant_table_create(d) != 0)) ) goto fail1; @@ -68,7 +66,7 @@ (arch_do_createdomain(v) != 0) ) goto fail3; - if ( !is_idle_task(d) ) + if ( !is_idle_domain(d) ) { write_lock(&domlist_lock); pd = &domain_list; /* NB. domain_list maintained in order of dom_id. */ @@ -173,20 +171,23 @@ BUG_ON(d == NULL); BUG_ON(d == current->domain); - BUG_ON(!test_bit(_DOMF_shuttingdown, &d->domain_flags)); - BUG_ON(test_bit(_DOMF_shutdown, &d->domain_flags)); + + LOCK_BIGLOCK(d); /* Make sure that every vcpu is descheduled before we finalise. */ for_each_vcpu ( d, v ) vcpu_sleep_sync(v); - BUG_ON(!cpus_empty(d->cpumask)); + BUG_ON(!cpus_empty(d->domain_dirty_cpumask)); sync_pagetable_state(d); - set_bit(_DOMF_shutdown, &d->domain_flags); - clear_bit(_DOMF_shuttingdown, &d->domain_flags); - - send_guest_virq(dom0->vcpu[0], VIRQ_DOM_EXC); + /* Don't set DOMF_shutdown until execution contexts are sync'ed. */ + if ( !test_and_set_bit(_DOMF_shutdown, &d->domain_flags) ) + send_guest_virq(dom0->vcpu[0], VIRQ_DOM_EXC); + + UNLOCK_BIGLOCK(d); + + put_domain(d); } static __init int domain_shutdown_finaliser_init(void) @@ -222,16 +223,17 @@ /* Mark the domain as shutting down. */ d->shutdown_code = reason; - if ( !test_and_set_bit(_DOMF_shuttingdown, &d->domain_flags) ) - { - /* This vcpu won the race to finalise the shutdown. */ - domain_shuttingdown[smp_processor_id()] = d; - raise_softirq(DOMAIN_SHUTDOWN_FINALISE_SOFTIRQ); - } /* Put every vcpu to sleep, but don't wait (avoids inter-vcpu deadlock). */ for_each_vcpu ( d, v ) + { + atomic_inc(&v->pausecnt); vcpu_sleep_nosync(v); + } + + get_knownalive_domain(d); + domain_shuttingdown[smp_processor_id()] = d; + raise_softirq(DOMAIN_SHUTDOWN_FINALISE_SOFTIRQ); } @@ -357,11 +359,11 @@ * of domains other than domain 0. ie. the domains that are being built by * the userspace dom0 domain builder. */ -int set_info_guest(struct domain *d, dom0_setdomaininfo_t *setdomaininfo) +int set_info_guest(struct domain *d, dom0_setvcpucontext_t *setvcpucontext) { int rc = 0; struct vcpu_guest_context *c = NULL; - unsigned long vcpu = setdomaininfo->vcpu; + unsigned long vcpu = setvcpucontext->vcpu; struct vcpu *v; if ( (vcpu >= MAX_VIRT_CPUS) || ((v = d->vcpu[vcpu]) == NULL) ) @@ -374,7 +376,7 @@ return -ENOMEM; rc = -EFAULT; - if ( copy_from_user(c, setdomaininfo->ctxt, sizeof(*c)) == 0 ) + if ( copy_from_user(c, setvcpucontext->ctxt, sizeof(*c)) == 0 ) rc = arch_set_info_guest(v, c); xfree(c); diff -r 2884f14895d2 -r f5ab28885c75 xen/common/grant_table.c --- a/xen/common/grant_table.c Fri Jan 13 16:38:44 2006 +++ b/xen/common/grant_table.c Fri Jan 13 20:12:24 2006 @@ -312,8 +312,6 @@ if ( !act->pin ) clear_bit(_GTF_reading, &sha->flags); - spin_unlock(&rd->grant_table->lock); - unlock_out: spin_unlock(&rd->grant_table->lock); (void)__put_user(rc, &uop->status); @@ -471,7 +469,7 @@ for ( i = 0; i < count; i++ ) (void)__gnttab_unmap_grant_ref(&uop[i]); - flush_tlb_mask(current->domain->cpumask); + flush_tlb_mask(current->domain->domain_dirty_cpumask); return 0; } diff -r 2884f14895d2 -r f5ab28885c75 xen/common/kernel.c --- a/xen/common/kernel.c Fri Jan 13 16:38:44 2006 +++ b/xen/common/kernel.c Fri Jan 13 20:12:24 2006 @@ -11,6 +11,7 @@ #include <xen/compile.h> #include <xen/sched.h> #include <asm/current.h> +#include <public/nmi.h> #include <public/version.h> void cmdline_parse(char *cmdline) @@ -146,6 +147,43 @@ } return -ENOSYS; +} + +long do_nmi_op(unsigned int cmd, void *arg) +{ + struct vcpu *v = current; + struct domain *d = current->domain; + long rc = 0; + + switch ( cmd ) + { + case XENNMI_register_callback: + if ( (d->domain_id != 0) || (v->vcpu_id != 0) ) + { + rc = -EINVAL; + } + else + { + v->nmi_addr = (unsigned long)arg; +#ifdef CONFIG_X86 + /* + * If no handler was registered we can 'lose the NMI edge'. + * Re-assert it now. + */ + if ( d->shared_info->arch.nmi_reason != 0 ) + set_bit(_VCPUF_nmi_pending, &v->vcpu_flags); +#endif + } + break; + case XENNMI_unregister_callback: + v->nmi_addr = 0; + break; + default: + rc = -ENOSYS; + break; + } + + return rc; } long do_vm_assist(unsigned int cmd, unsigned int type) diff -r 2884f14895d2 -r f5ab28885c75 xen/common/keyhandler.c --- a/xen/common/keyhandler.c Fri Jan 13 16:38:44 2006 +++ b/xen/common/keyhandler.c Fri Jan 13 20:12:24 2006 @@ -97,13 +97,22 @@ machine_restart(NULL); } -static void do_task_queues(unsigned char key) +static void cpuset_print(char *set, int size, cpumask_t mask) +{ + *set++ = '{'; + set += cpulist_scnprintf(set, size-2, mask); + *set++ = '}'; + *set++ = '\0'; +} + +static void dump_domains(unsigned char key) { struct domain *d; struct vcpu *v; s_time_t now = NOW(); - - printk("'%c' pressed -> dumping task queues (now=0x%X:%08X)\n", key, + char cpuset[100]; + + printk("'%c' pressed -> dumping domain info (now=0x%X:%08X)\n", key, (u32)(now>>32), (u32)now); read_lock(&domlist_lock); @@ -111,9 +120,11 @@ for_each_domain ( d ) { printk("General information for domain %u:\n", d->domain_id); - printk(" flags=%lx refcnt=%d nr_pages=%d xenheap_pages=%d\n", + cpuset_print(cpuset, sizeof(cpuset), d->domain_dirty_cpumask); + printk(" flags=%lx refcnt=%d nr_pages=%d xenheap_pages=%d " + "dirty_cpus=%s\n", d->domain_flags, atomic_read(&d->refcnt), - d->tot_pages, d->xenheap_pages); + d->tot_pages, d->xenheap_pages, cpuset); printk(" handle=%02x%02x%02x%02x-%02x%02x-%02x%02x-" "%02x%02x-%02x%02x%02x%02x%02x%02x\n", d->handle[ 0], d->handle[ 1], d->handle[ 2], d->handle[ 3], @@ -129,12 +140,16 @@ d->domain_id); for_each_vcpu ( d, v ) { printk(" VCPU%d: CPU%d [has=%c] flags=%lx " - "upcall_pend = %02x, upcall_mask = %02x\n", + "upcall_pend = %02x, upcall_mask = %02x ", v->vcpu_id, v->processor, test_bit(_VCPUF_running, &v->vcpu_flags) ? 'T':'F', v->vcpu_flags, v->vcpu_info->evtchn_upcall_pending, v->vcpu_info->evtchn_upcall_mask); + cpuset_print(cpuset, sizeof(cpuset), v->vcpu_dirty_cpumask); + printk("dirty_cpus=%s ", cpuset); + cpuset_print(cpuset, sizeof(cpuset), v->cpu_affinity); + printk("cpu_affinity=%s\n", cpuset); printk(" Notifying guest (virq %d, port %d, stat %d/%d/%d)\n", VIRQ_DEBUG, v->virq_to_evtchn[VIRQ_DEBUG], test_bit(v->virq_to_evtchn[VIRQ_DEBUG], @@ -170,6 +185,27 @@ bit. */ } +void do_nmi_stats(unsigned char key) +{ + int i; + struct domain *d; + struct vcpu *v; + printk("CPU\tNMI\n"); + for_each_cpu(i) + printk("%3d\t%3d\n", i, nmi_count(i)); + + if ((d = dom0) == NULL) + return; + if ((v = d->vcpu[0]) == NULL) + return; + if (v->vcpu_flags & (VCPUF_nmi_pending|VCPUF_nmi_masked)) + printk("dom0 vpu0: NMI %s%s\n", + v->vcpu_flags & VCPUF_nmi_pending ? "pending " : "", + v->vcpu_flags & VCPUF_nmi_masked ? "masked " : ""); + else + printk("dom0 vcpu0: NMI neither pending nor masked\n"); +} + #ifndef NDEBUG void debugtrace_key(unsigned char key) { @@ -193,11 +229,12 @@ register_keyhandler( 'L', reset_sched_histo, "reset sched latency histogram"); register_keyhandler( - 'q', do_task_queues, "dump task queues + guest state"); + 'q', dump_domains, "dump domain (and guest debug) info"); register_keyhandler( 'r', dump_runq, "dump run queues"); register_irq_keyhandler( 'R', halt_machine, "reboot machine"); + register_keyhandler('N', do_nmi_stats, "NMI statistics"); #ifndef NDEBUG register_keyhandler( diff -r 2884f14895d2 -r f5ab28885c75 xen/common/memory.c --- a/xen/common/memory.c Fri Jan 13 16:38:44 2006 +++ b/xen/common/memory.c Fri Jan 13 20:12:24 2006 @@ -38,10 +38,7 @@ if ( (extent_order != 0) && !multipage_allocation_permitted(current->domain) ) - { - DPRINTK("Only I/O-capable domains may allocate multi-page extents.\n"); return 0; - } for ( i = 0; i < nr_extents; i++ ) { diff -r 2884f14895d2 -r f5ab28885c75 xen/common/page_alloc.c --- a/xen/common/page_alloc.c Fri Jan 13 16:38:44 2006 +++ b/xen/common/page_alloc.c Fri Jan 13 20:12:24 2006 @@ -615,7 +615,7 @@ shadow_drop_references(d, &pg[i]); ASSERT((pg[i].u.inuse.type_info & PGT_count_mask) == 0); pg[i].tlbflush_timestamp = tlbflush_current_time(); - pg[i].u.free.cpumask = d->cpumask; + pg[i].u.free.cpumask = d->domain_dirty_cpumask; list_del(&pg[i].list); } diff -r 2884f14895d2 -r f5ab28885c75 xen/common/sched_bvt.c --- a/xen/common/sched_bvt.c Fri Jan 13 16:38:44 2006 +++ b/xen/common/sched_bvt.c Fri Jan 13 20:12:24 2006 @@ -20,7 +20,7 @@ #include <xen/delay.h> #include <xen/event.h> #include <xen/time.h> -#include <xen/ac_timer.h> +#include <xen/timer.h> #include <xen/perfc.h> #include <xen/sched-if.h> #include <xen/softirq.h> @@ -31,7 +31,8 @@ struct list_head run_list; /* runqueue list pointers */ u32 avt; /* actual virtual time */ u32 evt; /* effective virtual time */ - struct vcpu *vcpu; + int migrated; /* migrated to a new CPU */ + struct vcpu *vcpu; struct bvt_dom_info *inf; }; @@ -44,9 +45,9 @@ limits*/ s32 warp_value; /* virtual time warp */ s_time_t warpl; /* warp limit */ - struct ac_timer warp_timer; /* deals with warpl */ + struct timer warp_timer; /* deals with warpl */ s_time_t warpu; /* unwarp time requirement */ - struct ac_timer unwarp_timer; /* deals with warpu */ + struct timer unwarp_timer; /* deals with warpu */ struct bvt_vcpu_info vcpu_inf[MAX_VIRT_CPUS]; }; @@ -97,9 +98,9 @@ static void warp_timer_fn(void *data) { struct bvt_dom_info *inf = data; - unsigned int cpu = inf->domain->vcpu[0]->processor; - - spin_lock_irq(&schedule_data[cpu].schedule_lock); + struct vcpu *v = inf->domain->vcpu[0]; + + vcpu_schedule_lock_irq(v); inf->warp = 0; @@ -107,28 +108,28 @@ if ( inf->warpu == 0 ) { inf->warpback = 0; - cpu_raise_softirq(cpu, SCHEDULE_SOFTIRQ); - } - - set_ac_timer(&inf->unwarp_timer, NOW() + inf->warpu); - - spin_unlock_irq(&schedule_data[cpu].schedule_lock); + cpu_raise_softirq(v->processor, SCHEDULE_SOFTIRQ); + } + + set_timer(&inf->unwarp_timer, NOW() + inf->warpu); + + vcpu_schedule_unlock_irq(v); } static void unwarp_timer_fn(void *data) { struct bvt_dom_info *inf = data; - unsigned int cpu = inf->domain->vcpu[0]->processor; - - spin_lock_irq(&schedule_data[cpu].schedule_lock); + struct vcpu *v = inf->domain->vcpu[0]; + + vcpu_schedule_lock_irq(v); if ( inf->warpback ) { inf->warp = 1; - cpu_raise_softirq(cpu, SCHEDULE_SOFTIRQ); + cpu_raise_softirq(v->processor, SCHEDULE_SOFTIRQ); } - spin_unlock_irq(&schedule_data[cpu].schedule_lock); + vcpu_schedule_unlock_irq(v); } static inline u32 calc_avt(struct vcpu *d, s_time_t now) @@ -167,6 +168,7 @@ static int bvt_alloc_task(struct vcpu *v) { struct domain *d = v->domain; + struct bvt_dom_info *inf; if ( (d->sched_priv == NULL) ) { @@ -175,32 +177,12 @@ memset(d->sched_priv, 0, sizeof(struct bvt_dom_info)); } - v->sched_priv = &BVT_INFO(d)->vcpu_inf[v->vcpu_id]; - - BVT_INFO(d)->vcpu_inf[v->vcpu_id].inf = BVT_INFO(d); - BVT_INFO(d)->vcpu_inf[v->vcpu_id].vcpu = v; - - return 0; -} - -/* - * Add and remove a domain - */ -static void bvt_add_task(struct vcpu *v) -{ - struct bvt_dom_info *inf = BVT_INFO(v->domain); - struct bvt_vcpu_info *einf = EBVT_INFO(v); - ASSERT(inf != NULL); - ASSERT(v != NULL); - - /* Allocate per-CPU context if this is the first domain to be added. */ - if ( CPU_INFO(v->processor) == NULL ) - { - schedule_data[v->processor].sched_priv = xmalloc(struct bvt_cpu_info); - BUG_ON(CPU_INFO(v->processor) == NULL); - INIT_LIST_HEAD(RUNQUEUE(v->processor)); - CPU_SVT(v->processor) = 0; - } + inf = BVT_INFO(d); + + v->sched_priv = &inf->vcpu_inf[v->vcpu_id]; + + inf->vcpu_inf[v->vcpu_id].inf = BVT_INFO(d); + inf->vcpu_inf[v->vcpu_id].vcpu = v; if ( v->vcpu_id == 0 ) { @@ -213,13 +195,30 @@ inf->warpl = MILLISECS(2000); inf->warpu = MILLISECS(1000); /* Initialise the warp timers. */ - init_ac_timer(&inf->warp_timer, warp_timer_fn, inf, v->processor); - init_ac_timer(&inf->unwarp_timer, unwarp_timer_fn, inf, v->processor); - } - - einf->vcpu = v; - - if ( is_idle_task(v->domain) ) + init_timer(&inf->warp_timer, warp_timer_fn, inf, v->processor); + init_timer(&inf->unwarp_timer, unwarp_timer_fn, inf, v->processor); + } + + return 0; +} + +/* + * Add and remove a domain + */ +static void bvt_add_task(struct vcpu *v) +{ + struct bvt_vcpu_info *einf = EBVT_INFO(v); + + /* Allocate per-CPU context if this is the first domain to be added. */ + if ( CPU_INFO(v->processor) == NULL ) + { + schedule_data[v->processor].sched_priv = xmalloc(struct bvt_cpu_info); + BUG_ON(CPU_INFO(v->processor) == NULL); + INIT_LIST_HEAD(RUNQUEUE(v->processor)); + CPU_SVT(v->processor) = 0; + } + + if ( is_idle_vcpu(v) ) { einf->avt = einf->evt = ~0U; BUG_ON(__task_on_runqueue(v)); @@ -250,9 +249,11 @@ /* Set the BVT parameters. AVT should always be updated if CPU migration ocurred.*/ - if ( einf->avt < CPU_SVT(cpu) || - unlikely(test_bit(_VCPUF_cpu_migrated, &v->vcpu_flags)) ) + if ( (einf->avt < CPU_SVT(cpu)) || einf->migrated ) + { einf->avt = CPU_SVT(cpu); + einf->migrated = 0; + } /* Deal with warping here. */ einf->evt = calc_evt(v, einf->avt); @@ -265,29 +266,51 @@ ((einf->evt - curr_evt) / BVT_INFO(curr->domain)->mcu_advance) + ctx_allow; - if ( is_idle_task(curr->domain) || (einf->evt <= curr_evt) ) + if ( is_idle_vcpu(curr) || (einf->evt <= curr_evt) ) cpu_raise_softirq(cpu, SCHEDULE_SOFTIRQ); else if ( schedule_data[cpu].s_timer.expires > r_time ) - set_ac_timer(&schedule_data[cpu].s_timer, r_time); + set_timer(&schedule_data[cpu].s_timer, r_time); } static void bvt_sleep(struct vcpu *v) { - if ( test_bit(_VCPUF_running, &v->vcpu_flags) ) + if ( schedule_data[v->processor].curr == v ) cpu_raise_softirq(v->processor, SCHEDULE_SOFTIRQ); else if ( __task_on_runqueue(v) ) __del_from_runqueue(v); } + +static int bvt_set_affinity(struct vcpu *v, cpumask_t *affinity) +{ + if ( v == current ) + return cpu_isset(v->processor, *affinity) ? 0 : -EBUSY; + + vcpu_pause(v); + v->cpu_affinity = *affinity; + v->processor = first_cpu(v->cpu_affinity); + EBVT_INFO(v)->migrated = 1; + vcpu_unpause(v); + + return 0; +} + + /** * bvt_free_task - free BVT private structures for a task * @d: task */ static void bvt_free_task(struct domain *d) { - ASSERT(d->sched_priv != NULL); - xfree(d->sched_priv); + struct bvt_dom_info *inf = BVT_INFO(d); + + ASSERT(inf != NULL); + + kill_timer(&inf->warp_timer); + kill_timer(&inf->unwarp_timer); + + xfree(inf); } /* Control the scheduler. */ @@ -336,10 +359,10 @@ inf->warpu = MILLISECS(warpu); /* If the unwarp timer set up it needs to be removed */ - rem_ac_timer(&inf->unwarp_timer); + stop_timer(&inf->unwarp_timer); /* If we stop warping the warp timer needs to be removed */ if ( !warpback ) - rem_ac_timer(&inf->warp_timer); + stop_timer(&inf->warp_timer); } else if ( cmd->direction == SCHED_INFO_GET ) { @@ -380,17 +403,17 @@ ASSERT(prev_einf != NULL); ASSERT(__task_on_runqueue(prev)); - if ( likely(!is_idle_task(prev->domain)) ) + if ( likely(!is_idle_vcpu(prev)) ) { prev_einf->avt = calc_avt(prev, now); prev_einf->evt = calc_evt(prev, prev_einf->avt); if(prev_inf->warpback && prev_inf->warpl > 0) - rem_ac_timer(&prev_inf->warp_timer); + stop_timer(&prev_inf->warp_timer); __del_from_runqueue(prev); - if ( domain_runnable(prev) ) + if ( vcpu_runnable(prev) ) __add_to_runqueue_tail(prev); } @@ -436,7 +459,7 @@ } if ( next_einf->inf->warp && next_einf->inf->warpl > 0 ) - set_ac_timer(&next_einf->inf->warp_timer, now + next_einf->inf->warpl); + set_timer(&next_einf->inf->warp_timer, now + next_einf->inf->warpl); /* Extract the domain pointers from the dom infos */ next = next_einf->vcpu; @@ -471,13 +494,13 @@ } /* work out time for next run through scheduler */ - if ( is_idle_task(next->domain) ) + if ( is_idle_vcpu(next) ) { r_time = ctx_allow; goto sched_done; } - if ( (next_prime == NULL) || is_idle_task(next_prime->domain) ) + if ( (next_prime == NULL) || is_idle_vcpu(next_prime) ) { /* We have only one runnable task besides the idle task. */ r_time = 10 * ctx_allow; /* RN: random constant */ @@ -557,6 +580,7 @@ .dump_cpu_state = bvt_dump_cpu_state, .sleep = bvt_sleep, .wake = bvt_wake, + .set_affinity = bvt_set_affinity }; /* diff -r 2884f14895d2 -r f5ab28885c75 xen/common/sched_sedf.c --- a/xen/common/sched_sedf.c Fri Jan 13 16:38:44 2006 +++ b/xen/common/sched_sedf.c Fri Jan 13 20:12:24 2006 @@ -9,7 +9,7 @@ #include <xen/sched.h> #include <xen/sched-if.h> #include <public/sched_ctl.h> -#include <xen/ac_timer.h> +#include <xen/timer.h> #include <xen/softirq.h> #include <xen/time.h> @@ -325,21 +325,29 @@ list_insert_sort(RUNQ(d->processor), LIST(d), runq_comp); } + /* Allocates memory for per domain private scheduling data*/ -static int sedf_alloc_task(struct vcpu *d) { - PRINT(2,"sedf_alloc_task was called, domain-id %i.%i\n",d->domain->domain_id, - d->vcpu_id); - if (d->domain->sched_priv == NULL) { - if ((d->domain->sched_priv = - xmalloc(struct sedf_dom_info)) == NULL ) +static int sedf_alloc_task(struct vcpu *d) +{ + PRINT(2, "sedf_alloc_task was called, domain-id %i.%i\n", + d->domain->domain_id, d->vcpu_id); + + if ( d->domain->sched_priv == NULL ) + { + d->domain->sched_priv = xmalloc(struct sedf_dom_info); + if ( d->domain->sched_priv == NULL ) return -1; memset(d->domain->sched_priv, 0, sizeof(struct sedf_dom_info)); } - if ((d->sched_priv = xmalloc(struct sedf_vcpu_info)) == NULL ) + + if ( (d->sched_priv = xmalloc(struct sedf_vcpu_info)) == NULL ) return -1; + memset(d->sched_priv, 0, sizeof(struct sedf_vcpu_info)); + return 0; } + /* Setup the sedf_dom_info */ static void sedf_add_task(struct vcpu *d) @@ -363,14 +371,17 @@ INIT_LIST_HEAD(EXTRAQ(d->processor,EXTRA_UTIL_Q)); } - if (d->domain->domain_id==0) { + if ( d->domain->domain_id == 0 ) + { /*set dom0 to something useful to boot the machine*/ inf->period = MILLISECS(20); inf->slice = MILLISECS(15); inf->latency = 0; inf->deadl_abs = 0; inf->status = EXTRA_AWARE | SEDF_ASLEEP; - } else { + } + else + { /*other domains run in best effort mode*/ inf->period = WEIGHT_PERIOD; inf->slice = 0; @@ -379,14 +390,18 @@ inf->status = EXTRA_AWARE | SEDF_ASLEEP; inf->extraweight = 1; } + inf->period_orig = inf->period; inf->slice_orig = inf->slice; INIT_LIST_HEAD(&(inf->list)); INIT_LIST_HEAD(&(inf->extralist[EXTRA_PEN_Q])); INIT_LIST_HEAD(&(inf->extralist[EXTRA_UTIL_Q])); - if (!is_idle_task(d->domain)) { + if ( !is_idle_vcpu(d) ) + { extraq_check(d); - } else { + } + else + { EDOM_INFO(d)->deadl_abs = 0; EDOM_INFO(d)->status &= ~SEDF_ASLEEP; } @@ -396,19 +411,28 @@ static void sedf_free_task(struct domain *d) { int i; + PRINT(2,"sedf_free_task was called, domain-id %i\n",d->domain_id); + ASSERT(d->sched_priv != NULL); xfree(d->sched_priv); - for (i = 0; i < MAX_VIRT_CPUS; i++) - if ( d->vcpu[i] ) { + for ( i = 0; i < MAX_VIRT_CPUS; i++ ) + { + if ( d->vcpu[i] ) + { ASSERT(d->vcpu[i]->sched_priv != NULL); xfree(d->vcpu[i]->sched_priv); } -} - -/* handles the rescheduling, bookkeeping of domains running in their realtime-time :)*/ -static inline void desched_edf_dom (s_time_t now, struct vcpu* d) { + } +} + +/* + * Handles the rescheduling & bookkeeping of domains running in their + * guaranteed timeslice. + */ +static void desched_edf_dom(s_time_t now, struct vcpu* d) +{ struct sedf_vcpu_info* inf = EDOM_INFO(d); /*current domain is running in real time mode*/ @@ -418,27 +442,30 @@ /*scheduling decisions, which don't remove the running domain from the runq*/ - if ((inf->cputime < inf->slice) && sedf_runnable(d)) + if ( (inf->cputime < inf->slice) && sedf_runnable(d) ) return; __del_from_queue(d); /*manage bookkeeping (i.e. calculate next deadline, memorize overun-time of slice) of finished domains*/ - if (inf->cputime >= inf->slice) { + if ( inf->cputime >= inf->slice ) + { inf->cputime -= inf->slice; - if (inf->period < inf->period_orig) { + if ( inf->period < inf->period_orig ) + { /*this domain runs in latency scaling or burst mode*/ #if (UNBLOCK == UNBLOCK_BURST) /*if we are runnig in burst scaling wait for two periods before scaling periods up again*/ - if (now - inf->unblock_abs >= 2 * inf->period) + if ( (now - inf->unblock_abs) >= (2 * inf->period) ) #endif { inf->period *= 2; inf->slice *= 2; - if ((inf->period > inf->period_orig) || - (inf->slice > inf->slice_orig)) { + if ( (inf->period > inf->period_orig) || + (inf->slice > inf->slice_orig) ) + { /*reset slice & period*/ inf->period = inf->period_orig; inf->slice = inf->slice_orig; @@ -450,36 +477,46 @@ } /*add a runnable domain to the waitqueue*/ - if (sedf_runnable(d)) + if ( sedf_runnable(d) ) + { __add_to_waitqueue_sort(d); - else { + } + else + { /*we have a blocked realtime task -> remove it from exqs too*/ #if (EXTRA > EXTRA_OFF) #if (EXTRA == EXTRA_BLOCK_WEIGHT) - if (extraq_on(d, EXTRA_PEN_Q)) extraq_del(d, EXTRA_PEN_Q); -#endif - if (extraq_on(d, EXTRA_UTIL_Q)) extraq_del(d, EXTRA_UTIL_Q); -#endif - } + if ( extraq_on(d, EXTRA_PEN_Q) ) + extraq_del(d, EXTRA_PEN_Q); +#endif + if ( extraq_on(d, EXTRA_UTIL_Q) ) + extraq_del(d, EXTRA_UTIL_Q); +#endif + } + ASSERT(EQ(sedf_runnable(d), __task_on_queue(d))); ASSERT(IMPLY(extraq_on(d, EXTRA_UTIL_Q) || extraq_on(d, EXTRA_PEN_Q), sedf_runnable(d))); } + /* Update all elements on the queues */ -static inline void update_queues(s_time_t now, struct list_head* runq, - struct list_head* waitq) { - struct list_head *cur,*tmp; +static void update_queues( + s_time_t now, struct list_head *runq, struct list_head *waitq) +{ + struct list_head *cur, *tmp; struct sedf_vcpu_info *curinf; PRINT(3,"Updating waitq..\n"); + /*check for the first elements of the waitqueue, whether their next period has already started*/ list_for_each_safe(cur, tmp, waitq) { curinf = list_entry(cur, struct sedf_vcpu_info, list); PRINT(4,"\tLooking @ dom %i.%i\n", curinf->vcpu->domain->domain_id, curinf->vcpu->vcpu_id); - if (PERIOD_BEGIN(curinf) <= now) { + if ( PERIOD_BEGIN(curinf) <= now ) + { __del_from_queue(curinf->vcpu); __add_to_runqueue_sort(curinf->vcpu); } @@ -488,13 +525,16 @@ } PRINT(3,"Updating runq..\n"); + /*process the runq, find domains that are on the runqueue which shouldn't be there*/ list_for_each_safe(cur, tmp, runq) { curinf = list_entry(cur,struct sedf_vcpu_info,list); PRINT(4,"\tLooking @ dom %i.%i\n", curinf->vcpu->domain->domain_id, curinf->vcpu->vcpu_id); - if (unlikely(curinf->slice == 0)) { + + if ( unlikely(curinf->slice == 0) ) + { /*ignore domains with empty slice*/ PRINT(4,"\tUpdating zero-slice domain %i.%i\n", curinf->vcpu->domain->domain_id, @@ -504,7 +544,8 @@ /*move them to their next period*/ curinf->deadl_abs += curinf->period; /*ensure that the start of the next period is in the future*/ - if (unlikely(PERIOD_BEGIN(curinf) < now)) { + if ( unlikely(PERIOD_BEGIN(curinf) < now) ) + { curinf->deadl_abs += (DIV_UP(now - PERIOD_BEGIN(curinf), curinf->period)) * curinf->period; @@ -513,8 +554,10 @@ __add_to_waitqueue_sort(curinf->vcpu); continue; } - if (unlikely((curinf->deadl_abs < now) || - (curinf->cputime > curinf->slice))) { + + if ( unlikely((curinf->deadl_abs < now) || + (curinf->cputime > curinf->slice)) ) + { /*we missed the deadline or the slice was already finished... might hapen because of dom_adj.*/ @@ -550,6 +593,7 @@ PRINT(3,"done updating the queues\n"); } + #if (EXTRA > EXTRA_OFF) /* removes a domain from the head of the according extraQ and requeues it at a specified position: @@ -557,9 +601,10 @@ weighted ext.: insert in sorted list by score if the domain is blocked / has regained its short-block-loss time it is not put on any queue */ -static inline void desched_extra_dom(s_time_t now, struct vcpu* d) { +static void desched_extra_dom(s_time_t now, struct vcpu* d) +{ struct sedf_vcpu_info *inf = EDOM_INFO(d); - int i = extra_get_cur_q(inf); + int i = extra_get_cur_q(inf); #if (EXTRA == EXTRA_SLICE_WEIGHT || EXTRA == EXTRA_BLOCK_WEIGHT) unsigned long oldscore; @@ -575,14 +620,15 @@ extraq_del(d, i); #if (EXTRA == EXTRA_ROUNDR) - if (sedf_runnable(d) && (inf->status & EXTRA_AWARE)) + if ( sedf_runnable(d) && (inf->status & EXTRA_AWARE) ) /*add to the tail if it is runnable => round-robin*/ extraq_add_tail(d, EXTRA_UTIL_Q); #elif (EXTRA == EXTRA_SLICE_WEIGHT || EXTRA == EXTRA_BLOCK_WEIGHT) /*update the score*/ - oldscore = inf->score[i]; + oldscore = inf->score[i]; #if (EXTRA == EXTRA_BLOCK_WEIGHT) - if (i == EXTRA_PEN_Q) { + if ( i == EXTRA_PEN_Q ) + { /*domain was running in L0 extraq*/ /*reduce block lost, probably more sophistication here!*/ /*inf->short_block_lost_tot -= EXTRA_QUANTUM;*/ @@ -605,12 +651,13 @@ inf->score[EXTRA_PEN_Q] = (inf->period << 10) / inf->short_block_lost_tot; oldscore = 0; - } else + } + else #endif { /*domain was running in L1 extraq => score is inverse of utilization and is used somewhat incremental!*/ - if (!inf->extraweight) + if ( !inf->extraweight ) /*NB: use fixed point arithmetic with 10 bits*/ inf->score[EXTRA_UTIL_Q] = (inf->period << 10) / inf->slice; @@ -619,24 +666,32 @@ full (ie 100%) utilization is equivalent to 128 extraweight*/ inf->score[EXTRA_UTIL_Q] = (1<<17) / inf->extraweight; } + check_extra_queues: /* Adding a runnable domain to the right queue and removing blocked ones*/ - if (sedf_runnable(d)) { + if ( sedf_runnable(d) ) + { /*add according to score: weighted round robin*/ if (((inf->status & EXTRA_AWARE) && (i == EXTRA_UTIL_Q)) || ((inf->status & EXTRA_WANT_PEN_Q) && (i == EXTRA_PEN_Q))) extraq_add_sort_update(d, i, oldscore); } - else { + else + { /*remove this blocked domain from the waitq!*/ __del_from_queue(d); #if (EXTRA == EXTRA_BLOCK_WEIGHT) /*make sure that we remove a blocked domain from the other extraq too*/ - if (i == EXTRA_PEN_Q) { - if (extraq_on(d, EXTRA_UTIL_Q)) extraq_del(d, EXTRA_UTIL_Q); - } else { - if (extraq_on(d, EXTRA_PEN_Q)) extraq_del(d, EXTRA_PEN_Q); + if ( i == EXTRA_PEN_Q ) + { + if ( extraq_on(d, EXTRA_UTIL_Q) ) + extraq_del(d, EXTRA_UTIL_Q); + } + else + { + if ( extraq_on(d, EXTRA_PEN_Q) ) + extraq_del(d, EXTRA_PEN_Q); } #endif } @@ -647,16 +702,21 @@ } #endif -static inline struct task_slice sedf_do_extra_schedule (s_time_t now, - s_time_t end_xt, struct list_head *extraq[], int cpu) { + +static struct task_slice sedf_do_extra_schedule( + s_time_t now, s_time_t end_xt, struct list_head *extraq[], int cpu) +{ struct task_slice ret; struct sedf_vcpu_info *runinf; ASSERT(end_xt > now); + /* Enough time left to use for extratime? */ - if (end_xt - now < EXTRA_QUANTUM) + if ( end_xt - now < EXTRA_QUANTUM ) goto return_idle; + #if (EXTRA == EXTRA_BLOCK_WEIGHT) - if (!list_empty(extraq[EXTRA_PEN_Q])) { + if ( !list_empty(extraq[EXTRA_PEN_Q]) ) + { /*we still have elements on the level 0 extraq => let those run first!*/ runinf = list_entry(extraq[EXTRA_PEN_Q]->next, @@ -667,9 +727,12 @@ #ifdef SEDF_STATS runinf->pen_extra_slices++; #endif - } else -#endif - if (!list_empty(extraq[EXTRA_UTIL_Q])) { + } + else +#endif + { + if ( !list_empty(extraq[EXTRA_UTIL_Q]) ) + { /*use elements from the normal extraqueue*/ runinf = list_entry(extraq[EXTRA_UTIL_Q]->next, struct sedf_vcpu_info, @@ -680,6 +743,7 @@ } else goto return_idle; + } ASSERT(ret.time > 0); ASSERT(sedf_runnable(ret.task)); @@ -692,6 +756,8 @@ ASSERT(sedf_runnable(ret.task)); return ret; } + + /* Main scheduling function Reasons for calling this function are: -timeslice for the current period used up @@ -699,7 +765,7 @@ -and various others ;) in general: determine which domain to run next*/ static struct task_slice sedf_do_schedule(s_time_t now) { - int cpu = current->processor; + int cpu = smp_processor_id(); struct list_head *runq = RUNQ(cpu); struct list_head *waitq = WAITQ(cpu); #if (EXTRA > EXTRA_OFF) @@ -711,20 +777,21 @@ struct task_slice ret; /*idle tasks don't need any of the following stuf*/ - if (is_idle_task(current->domain)) + if ( is_idle_vcpu(current) ) goto check_waitq; /* create local state of the status of the domain, in order to avoid inconsistent state during scheduling decisions, because data for - domain_runnable is not protected by the scheduling lock!*/ - if(!domain_runnable(current)) + vcpu_runnable is not protected by the scheduling lock!*/ + if ( !vcpu_runnable(current) ) inf->status |= SEDF_ASLEEP; - if (inf->status & SEDF_ASLEEP) + if ( inf->status & SEDF_ASLEEP ) inf->block_abs = now; #if (EXTRA > EXTRA_OFF) - if (unlikely(extra_runs(inf))) { + if ( unlikely(extra_runs(inf)) ) + { /*special treatment of domains running in extra time*/ desched_extra_dom(now, current); } @@ -739,10 +806,12 @@ /*now simply pick the first domain from the runqueue, which has the earliest deadline, because the list is sorted*/ - if (!list_empty(runq)) { + if ( !list_empty(runq) ) + { runinf = list_entry(runq->next,struct sedf_vcpu_info,list); ret.task = runinf->vcpu; - if (!list_empty(waitq)) { + if ( !list_empty(waitq) ) + { waitinf = list_entry(waitq->next, struct sedf_vcpu_info,list); /*rerun scheduler, when scheduled domain reaches it's @@ -751,14 +820,16 @@ ret.time = MIN(now + runinf->slice - runinf->cputime, PERIOD_BEGIN(waitinf)) - now; } - else { + else + { ret.time = runinf->slice - runinf->cputime; } CHECK(ret.time > 0); goto sched_done; } - if (!list_empty(waitq)) { + if ( !list_empty(waitq) ) + { waitinf = list_entry(waitq->next,struct sedf_vcpu_info, list); /*we could not find any suitable domain => look for domains that are aware of extratime*/ @@ -771,7 +842,8 @@ #endif CHECK(ret.time > 0); } - else { + else + { /*this could probably never happen, but one never knows...*/ /*it can... imagine a second CPU, which is pure scifi ATM, but one never knows ;)*/ @@ -782,11 +854,13 @@ sched_done: /*TODO: Do something USEFUL when this happens and find out, why it still can happen!!!*/ - if (ret.time<0) { + if ( ret.time < 0) + { printk("Ouch! We are seriously BEHIND schedule! %"PRIi64"\n", ret.time); ret.time = EXTRA_QUANTUM; } + EDOM_INFO(ret.task)->sched_start_abs = now; CHECK(ret.time > 0); ASSERT(sedf_runnable(ret.task)); @@ -794,30 +868,36 @@ return ret; } -static void sedf_sleep(struct vcpu *d) { - PRINT(2,"sedf_sleep was called, domain-id %i.%i\n",d->domain->domain_id, d->vcpu_id); - - if (is_idle_task(d->domain)) + +static void sedf_sleep(struct vcpu *d) +{ + PRINT(2,"sedf_sleep was called, domain-id %i.%i\n", + d->domain->domain_id, d->vcpu_id); + + if ( is_idle_vcpu(d) ) return; EDOM_INFO(d)->status |= SEDF_ASLEEP; - if ( test_bit(_VCPUF_running, &d->vcpu_flags) ) { + if ( schedule_data[d->processor].curr == d ) + { cpu_raise_softirq(d->processor, SCHEDULE_SOFTIRQ); } - else { + else + { if ( __task_on_queue(d) ) __del_from_queue(d); #if (EXTRA > EXTRA_OFF) - if (extraq_on(d, EXTRA_UTIL_Q)) + if ( extraq_on(d, EXTRA_UTIL_Q) ) extraq_del(d, EXTRA_UTIL_Q); #endif #if (EXTRA == EXTRA_BLOCK_WEIGHT) - if (extraq_on(d, EXTRA_PEN_Q)) + if ( extraq_on(d, EXTRA_PEN_Q) ) extraq_del(d, EXTRA_PEN_Q); #endif } } + /* This function wakes up a domain, i.e. moves them into the waitqueue * things to mention are: admission control is taking place nowhere at @@ -890,17 +970,21 @@ * -either behaviour can lead to missed deadlines in other domains as * opposed to approaches 1,2a,2b */ -static inline void unblock_short_vcons -(struct sedf_vcpu_info* inf, s_time_t now) { +#if (UNBLOCK <= UNBLOCK_SHORT_RESUME) +static void unblock_short_vcons(struct sedf_vcpu_info* inf, s_time_t now) +{ inf->deadl_abs += inf->period; inf->cputime = 0; } - -static inline void unblock_short_cons(struct sedf_vcpu_info* inf, s_time_t now) +#endif + +#if (UNBLOCK == UNBLOCK_SHORT_RESUME) +static void unblock_short_cons(struct sedf_vcpu_info* inf, s_time_t now) { /*treat blocked time as consumed by the domain*/ inf->cputime += now - inf->block_abs; - if (inf->cputime + EXTRA_QUANTUM > inf->slice) { + if ( (inf->cputime + EXTRA_QUANTUM) > inf->slice ) + { /*we don't have a reasonable amount of time in our slice left :( => start in next period!*/ unblock_short_vcons(inf, now); @@ -910,8 +994,11 @@ inf->short_cont++; #endif } -static inline void unblock_short_extra_support (struct sedf_vcpu_info* inf, - s_time_t now) { +#endif + +static void unblock_short_extra_support( + struct sedf_vcpu_info* inf, s_time_t now) +{ /*this unblocking scheme tries to support the domain, by assigning it a priority in extratime distribution according to the loss of time in this slice due to blocking*/ @@ -919,26 +1006,29 @@ /*no more realtime execution in this period!*/ inf->deadl_abs += inf->period; - if (likely(inf->block_abs)) { + if ( likely(inf->block_abs) ) + { //treat blocked time as consumed by the domain*/ /*inf->cputime += now - inf->block_abs;*/ /*penalty is time the domain would have had if it continued to run */ pen = (inf->slice - inf->cputime); - if (pen < 0) pen = 0; + if ( pen < 0 ) + pen = 0; /*accumulate all penalties over the periods*/ /*inf->short_block_lost_tot += pen;*/ /*set penalty to the current value*/ inf->short_block_lost_tot = pen; /*not sure which one is better.. but seems to work well...*/ - if (inf->short_block_lost_tot) { + if ( inf->short_block_lost_tot ) + { inf->score[0] = (inf->period << 10) / inf->short_block_lost_tot; #ifdef SEDF_STATS inf->pen_extra_blocks++; #endif - if (extraq_on(inf->vcpu, EXTRA_PEN_Q)) + if ( extraq_on(inf->vcpu, EXTRA_PEN_Q) ) /*remove domain for possible resorting!*/ extraq_del(inf->vcpu, EXTRA_PEN_Q); else @@ -951,36 +1041,53 @@ extraq_add_sort_update(inf->vcpu, EXTRA_PEN_Q, 0); } } + /*give it a fresh slice in the next period!*/ inf->cputime = 0; } -static inline void unblock_long_vcons(struct sedf_vcpu_info* inf, s_time_t now) + + +#if (UNBLOCK == UNBLOCK_ISOCHRONOUS_EDF) +static void unblock_long_vcons(struct sedf_vcpu_info* inf, s_time_t now) { /* align to next future period */ inf->deadl_abs += (DIV_UP(now - inf->deadl_abs, inf->period) +1) * inf->period; inf->cputime = 0; } - -static inline void unblock_long_cons_a (struct sedf_vcpu_info* inf, - s_time_t now) { +#endif + + +#if 0 +static void unblock_long_cons_a (struct sedf_vcpu_info* inf, s_time_t now) +{ /*treat the time the domain was blocked in the - CURRENT period as consumed by the domain*/ + CURRENT period as consumed by the domain*/ inf->cputime = (now - inf->deadl_abs) % inf->period; - if (inf->cputime + EXTRA_QUANTUM > inf->slice) { + if ( (inf->cputime + EXTRA_QUANTUM) > inf->slice ) + { /*we don't have a reasonable amount of time in our slice left :( => start in next period!*/ unblock_long_vcons(inf, now); } } -static inline void unblock_long_cons_b(struct sedf_vcpu_info* inf,s_time_t now) { +#endif + + +static void unblock_long_cons_b(struct sedf_vcpu_info* inf,s_time_t now) +{ /*Conservative 2b*/ /*Treat the unblocking time as a start of a new period */ inf->deadl_abs = now + inf->period; inf->cputime = 0; } -static inline void unblock_long_cons_c(struct sedf_vcpu_info* inf,s_time_t now) { - if (likely(inf->latency)) { + + +#if (UNBLOCK == UNBLOCK_ATROPOS) +static void unblock_long_cons_c(struct sedf_vcpu_info* inf,s_time_t now) +{ + if ( likely(inf->latency) ) + { /*scale the slice and period accordingly to the latency hint*/ /*reduce period temporarily to the latency hint*/ inf->period = inf->latency; @@ -993,18 +1100,24 @@ inf->deadl_abs = now + inf->period; inf->cputime = 0; } - else { + else + { /*we don't have a latency hint.. use some other technique*/ unblock_long_cons_b(inf, now); } } +#endif + + +#if (UNBLOCK == UNBLOCK_BURST) /*a new idea of dealing with short blocks: burst period scaling*/ -static inline void unblock_short_burst(struct sedf_vcpu_info* inf, s_time_t now) +static void unblock_short_burst(struct sedf_vcpu_info* inf, s_time_t now) { /*treat blocked time as consumed by the domain*/ inf->cputime += now - inf->block_abs; - if (inf->cputime + EXTRA_QUANTUM <= inf->slice) { + if ( (inf->cputime + EXTRA_QUANTUM) <= inf->slice ) + { /*if we can still use some time in the current slice then use it!*/ #ifdef SEDF_STATS @@ -1012,10 +1125,12 @@ inf->short_cont++; #endif } - else { + else + { /*we don't have a reasonable amount of time in our slice left => switch to burst mode*/ - if (likely(inf->unblock_abs)) { + if ( likely(inf->unblock_abs) ) + { /*set the period-length to the current blocking interval, possible enhancements: average over last blocking intervals, user-specified minimum,...*/ @@ -1030,17 +1145,23 @@ /*set new (shorter) deadline*/ inf->deadl_abs += inf->period; } - else { + else + { /*in case we haven't unblocked before start in next period!*/ inf->cputime=0; inf->deadl_abs += inf->period; } } + inf->unblock_abs = now; } -static inline void unblock_long_burst(struct sedf_vcpu_info* inf, s_time_t now) { - if (unlikely(inf->latency && (inf->period > inf->latency))) { + + +static void unblock_long_burst(struct sedf_vcpu_info* inf, s_time_t now) +{ + if ( unlikely(inf->latency && (inf->period > inf->latency)) ) + { /*scale the slice and period accordingly to the latency hint*/ inf->period = inf->latency; /*check for overflows on multiplication*/ @@ -1052,23 +1173,28 @@ inf->deadl_abs = now + inf->period; inf->cputime = 0; } - else { + else + { /*we don't have a latency hint.. or we are currently in "burst mode": use some other technique NB: this should be in fact the normal way of operation, when we are in sync with the device!*/ unblock_long_cons_b(inf, now); } + inf->unblock_abs = now; } +#endif /* UNBLOCK == UNBLOCK_BURST */ + #define DOMAIN_EDF 1 #define DOMAIN_EXTRA_PEN 2 #define DOMAIN_EXTRA_UTIL 3 #define DOMAIN_IDLE 4 -static inline int get_run_type(struct vcpu* d) { +static inline int get_run_type(struct vcpu* d) +{ struct sedf_vcpu_info* inf = EDOM_INFO(d); - if (is_idle_task(d->domain)) + if (is_idle_vcpu(d)) return DOMAIN_IDLE; if (inf->status & EXTRA_RUN_PEN) return DOMAIN_EXTRA_PEN; @@ -1076,6 +1202,8 @@ return DOMAIN_EXTRA_UTIL; return DOMAIN_EDF; } + + /*Compares two domains in the relation of whether the one is allowed to interrupt the others execution. It returns true (!=0) if a switch to the other domain is good. @@ -1085,8 +1213,10 @@ In the same class priorities are assigned as following: EDF: early deadline > late deadline L0 extra-time: lower score > higher score*/ -static inline int should_switch(struct vcpu* cur, - struct vcpu* other, s_time_t now) { +static inline int should_switch(struct vcpu *cur, + struct vcpu *other, + s_time_t now) +{ struct sedf_vcpu_info *cur_inf, *other_inf; cur_inf = EDOM_INFO(cur); other_inf = EDOM_INFO(other); @@ -1119,41 +1249,51 @@ } return 1; } -void sedf_wake(struct vcpu *d) { + +void sedf_wake(struct vcpu *d) +{ s_time_t now = NOW(); struct sedf_vcpu_info* inf = EDOM_INFO(d); PRINT(3, "sedf_wake was called, domain-id %i.%i\n",d->domain->domain_id, d->vcpu_id); - if (unlikely(is_idle_task(d->domain))) + if ( unlikely(is_idle_vcpu(d)) ) return; - if ( unlikely(__task_on_queue(d)) ) { + if ( unlikely(__task_on_queue(d)) ) + { PRINT(3,"\tdomain %i.%i is already in some queue\n", d->domain->domain_id, d->vcpu_id); return; } + ASSERT(!sedf_runnable(d)); inf->status &= ~SEDF_ASLEEP; ASSERT(!extraq_on(d, EXTRA_UTIL_Q)); ASSERT(!extraq_on(d, EXTRA_PEN_Q)); - if (unlikely(inf->deadl_abs == 0)) + if ( unlikely(inf->deadl_abs == 0) ) + { /*initial setup of the deadline*/ inf->deadl_abs = now + inf->slice; + } - PRINT(3,"waking up domain %i.%i (deadl= %"PRIu64" period= %"PRIu64" "\ - "now= %"PRIu64")\n", d->domain->domain_id, d->vcpu_id, inf->deadl_abs, - inf->period, now); + PRINT(3, "waking up domain %i.%i (deadl= %"PRIu64" period= %"PRIu64 + "now= %"PRIu64")\n", + d->domain->domain_id, d->vcpu_id, inf->deadl_abs, inf->period, now); + #ifdef SEDF_STATS inf->block_tot++; #endif - if (unlikely(now < PERIOD_BEGIN(inf))) { + + if ( unlikely(now < PERIOD_BEGIN(inf)) ) + { PRINT(4,"extratime unblock\n"); /* unblocking in extra-time! */ #if (EXTRA == EXTRA_BLOCK_WEIGHT) - if (inf->status & EXTRA_WANT_PEN_Q) { + if ( inf->status & EXTRA_WANT_PEN_Q ) + { /*we have a domain that wants compensation for block penalty and did just block in its compensation time. Give it another @@ -1163,8 +1303,10 @@ #endif extraq_check_add_unblocked(d, 0); } - else { - if (now < inf->deadl_abs) { + else + { + if ( now < inf->deadl_abs ) + { PRINT(4,"short unblocking\n"); /*short blocking*/ #ifdef SEDF_STATS @@ -1182,7 +1324,8 @@ extraq_check_add_unblocked(d, 1); } - else { + else + { PRINT(4,"long unblocking\n"); /*long unblocking*/ #ifdef SEDF_STATS @@ -1197,7 +1340,6 @@ unblock_long_cons_c(inf, now); #elif (UNBLOCK == UNBLOCK_SHORT_RESUME) unblock_long_cons_b(inf, now); - /*unblock_short_cons_c(inf, now);*/ #elif (UNBLOCK == UNBLOCK_BURST) unblock_long_burst(inf, now); #endif @@ -1205,26 +1347,33 @@ extraq_check_add_unblocked(d, 1); } } - PRINT(3,"woke up domain %i.%i (deadl= %"PRIu64" period= %"PRIu64" "\ - "now= %"PRIu64")\n", d->domain->domain_id, d->vcpu_id, inf->deadl_abs, + + PRINT(3, "woke up domain %i.%i (deadl= %"PRIu64" period= %"PRIu64 + "now= %"PRIu64")\n", + d->domain->domain_id, d->vcpu_id, inf->deadl_abs, inf->period, now); - if (PERIOD_BEGIN(inf) > now) { + + if ( PERIOD_BEGIN(inf) > now ) + { __add_to_waitqueue_sort(d); PRINT(3,"added to waitq\n"); } - else { + else + { __add_to_runqueue_sort(d); PRINT(3,"added to runq\n"); } #ifdef SEDF_STATS /*do some statistics here...*/ - if (inf->block_abs != 0) { + if ( inf->block_abs != 0 ) + { inf->block_time_tot += now - inf->block_abs; inf->penalty_time_tot += PERIOD_BEGIN(inf) + inf->cputime - inf->block_abs; } #endif + /*sanity check: make sure each extra-aware domain IS on the util-q!*/ ASSERT(IMPLY(inf->status & EXTRA_AWARE, extraq_on(d, EXTRA_UTIL_Q))); ASSERT(__task_on_queue(d)); @@ -1234,27 +1383,48 @@ ASSERT(d->processor >= 0); ASSERT(d->processor < NR_CPUS); ASSERT(schedule_data[d->processor].curr); - if (should_switch(schedule_data[d->processor].curr, d, now)) + + if ( should_switch(schedule_data[d->processor].curr, d, now) ) cpu_raise_softirq(d->processor, SCHEDULE_SOFTIRQ); } -/*Print a lot of use-{full, less} information about a domains in the system*/ -static void sedf_dump_domain(struct vcpu *d) { + +static int sedf_set_affinity(struct vcpu *v, cpumask_t *affinity) +{ + if ( v == current ) + return cpu_isset(v->processor, *affinity) ? 0 : -EBUSY; + + vcpu_pause(v); + v->cpu_affinity = *affinity; + v->processor = first_cpu(v->cpu_affinity); + vcpu_unpause(v); + + return 0; +} + + +/* Print a lot of useful information about a domains in the system */ +static void sedf_dump_domain(struct vcpu *d) +{ printk("%i.%i has=%c ", d->domain->domain_id, d->vcpu_id, test_bit(_VCPUF_running, &d->vcpu_flags) ? 'T':'F'); - printk("p=%"PRIu64" sl=%"PRIu64" ddl=%"PRIu64" w=%hu c=%"PRIu64" sc=%i xtr(%s)=%"PRIu64" ew=%hu", + printk("p=%"PRIu64" sl=%"PRIu64" ddl=%"PRIu64" w=%hu c=%"PRIu64 + " sc=%i xtr(%s)=%"PRIu64" ew=%hu", EDOM_INFO(d)->period, EDOM_INFO(d)->slice, EDOM_INFO(d)->deadl_abs, - EDOM_INFO(d)->weight, d->cpu_time, EDOM_INFO(d)->score[EXTRA_UTIL_Q], + EDOM_INFO(d)->weight, d->cpu_time, + EDOM_INFO(d)->score[EXTRA_UTIL_Q], (EDOM_INFO(d)->status & EXTRA_AWARE) ? "yes" : "no", EDOM_INFO(d)->extra_time_tot, EDOM_INFO(d)->extraweight); - if (d->cpu_time !=0) + + if ( d->cpu_time != 0 ) printf(" (%"PRIu64"%%)", (EDOM_INFO(d)->extra_time_tot * 100) / d->cpu_time); + #ifdef SEDF_STATS - if (EDOM_INFO(d)->block_time_tot!=0) + if ( EDOM_INFO(d)->block_time_tot != 0 ) printf(" pen=%"PRIu64"%%", (EDOM_INFO(d)->penalty_time_tot * 100) / EDOM_INFO(d)->block_time_tot); - if (EDOM_INFO(d)->block_tot!=0) + if ( EDOM_INFO(d)->block_tot != 0 ) printf("\n blks=%u sh=%u (%u%%) (shc=%u (%u%%) shex=%i "\ "shexsl=%i) l=%u (%u%%) avg: b=%"PRIu64" p=%"PRIu64"", EDOM_INFO(d)->block_tot, EDOM_INFO(d)->short_block_tot, @@ -1271,7 +1441,8 @@ printf("\n"); } -/*dumps all domains on hte specified cpu*/ + +/* dumps all domains on hte specified cpu */ static void sedf_dump_cpu_state(int i) { struct list_head *list, *queue, *tmp; @@ -1284,7 +1455,8 @@ queue = RUNQ(i); printk("RUNQ rq %lx n: %lx, p: %lx\n", (unsigned long)queue, (unsigned long) queue->next, (unsigned long) queue->prev); - list_for_each_safe ( list, tmp, queue ) { + list_for_each_safe ( list, tmp, queue ) + { printk("%3d: ",loop++); d_inf = list_entry(list, struct sedf_vcpu_info, list); sedf_dump_domain(d_inf->vcpu); @@ -1293,7 +1465,8 @@ queue = WAITQ(i); loop = 0; printk("\nWAITQ rq %lx n: %lx, p: %lx\n", (unsigned long)queue, (unsigned long) queue->next, (unsigned long) queue->prev); - list_for_each_safe ( list, tmp, queue ) { + list_for_each_safe ( list, tmp, queue ) + { printk("%3d: ",loop++); d_inf = list_entry(list, struct sedf_vcpu_info, list); sedf_dump_domain(d_inf->vcpu); @@ -1303,7 +1476,8 @@ printk("\nEXTRAQ (penalty) rq %lx n: %lx, p: %lx\n", (unsigned long)queue, (unsigned long) queue->next, (unsigned long) queue->prev); - list_for_each_safe ( list, tmp, queue ) { + list_for_each_safe ( list, tmp, queue ) + { d_inf = list_entry(list, struct sedf_vcpu_info, extralist[EXTRA_PEN_Q]); printk("%3d: ",loop++); @@ -1314,7 +1488,8 @@ printk("\nEXTRAQ (utilization) rq %lx n: %lx, p: %lx\n", (unsigned long)queue, (unsigned long) queue->next, (unsigned long) queue->prev); - list_for_each_safe ( list, tmp, queue ) { + list_for_each_safe ( list, tmp, queue ) + { d_inf = list_entry(list, struct sedf_vcpu_info, extralist[EXTRA_UTIL_Q]); printk("%3d: ",loop++); @@ -1323,69 +1498,93 @@ loop = 0; printk("\nnot on Q\n"); - for_each_domain(d) + + for_each_domain ( d ) + { for_each_vcpu(d, ed) - { - if (!__task_on_queue(ed) && (ed->processor == i)) { - printk("%3d: ",loop++); - sedf_dump_domain(ed); - } - } -} -/*Adjusts periods and slices of the domains accordingly to their weights*/ -static inline int sedf_adjust_weights(struct sched_adjdom_cmd *cmd) { + { + if ( !__task_on_queue(ed) && (ed->processor == i) ) + { + printk("%3d: ",loop++); + sedf_dump_domain(ed); + } + } + } +} + + +/* Adjusts periods and slices of the domains accordingly to their weights. */ +static int sedf_adjust_weights(struct sched_adjdom_cmd *cmd) +{ struct vcpu *p; struct domain *d; int sumw[NR_CPUS]; s_time_t sumt[NR_CPUS]; int cpu; - for (cpu=0; cpu < NR_CPUS; cpu++) { + for ( cpu = 0; cpu < NR_CPUS; cpu++ ) + { sumw[cpu] = 0; sumt[cpu] = 0; } - /*sum up all weights*/ - for_each_domain(d) - for_each_vcpu(d, p) { - if (EDOM_INFO(p)->weight) - sumw[p->processor] += EDOM_INFO(p)->weight; - else { - /*don't modify domains who don't have a weight, but sum - up the time they need, projected to a WEIGHT_PERIOD, - so that this time is not given to the weight-driven - domains*/ - /*check for overflows*/ - ASSERT((WEIGHT_PERIOD < ULONG_MAX) - && (EDOM_INFO(p)->slice_orig < ULONG_MAX)); - sumt[p->processor] += - (WEIGHT_PERIOD * EDOM_INFO(p)->slice_orig) / - EDOM_INFO(p)->period_orig; - } - } - /*adjust all slices (and periods) to the new weight*/ - for_each_domain(d) - for_each_vcpu(d, p) { - if (EDOM_INFO(p)->weight) { - EDOM_INFO(p)->period_orig = - EDOM_INFO(p)->period = WEIGHT_PERIOD; - EDOM_INFO(p)->slice_orig = - EDOM_INFO(p)->slice = - (EDOM_INFO(p)->weight * - (WEIGHT_PERIOD - WEIGHT_SAFETY - sumt[p->processor])) / - sumw[p->processor]; - } - } + + /* sum up all weights */ + for_each_domain( d ) + { + for_each_vcpu( d, p ) + { + if ( EDOM_INFO(p)->weight ) + { + sumw[p->processor] += EDOM_INFO(p)->weight; + } + else + { + /*don't modify domains who don't have a weight, but sum + up the time they need, projected to a WEIGHT_PERIOD, + so that this time is not given to the weight-driven + domains*/ + /*check for overflows*/ + ASSERT((WEIGHT_PERIOD < ULONG_MAX) + && (EDOM_INFO(p)->slice_orig < ULONG_MAX)); + sumt[p->processor] += + (WEIGHT_PERIOD * EDOM_INFO(p)->slice_orig) / + EDOM_INFO(p)->period_orig; + } + } + } + + /* adjust all slices (and periods) to the new weight */ + for_each_domain( d ) + { + for_each_vcpu ( d, p ) + { + if ( EDOM_INFO(p)->weight ) + { + EDOM_INFO(p)->period_orig = + EDOM_INFO(p)->period = WEIGHT_PERIOD; + EDOM_INFO(p)->slice_orig = + EDOM_INFO(p)->slice = + (EDOM_INFO(p)->weight * + (WEIGHT_PERIOD - WEIGHT_SAFETY - sumt[p->processor])) / + sumw[p->processor]; + } + } + } + return 0; } + /* set or fetch domain scheduling parameters */ -static int sedf_adjdom(struct domain *p, struct sched_adjdom_cmd *cmd) { +static int sedf_adjdom(struct domain *p, struct sched_adjdom_cmd *cmd) +{ struct vcpu *v; PRINT(2,"sedf_adjdom was called, domain-id %i new period %"PRIu64" "\ "new slice %"PRIu64"\nlatency %"PRIu64" extra:%s\n", p->domain_id, cmd->u.sedf.period, cmd->u.sedf.slice, cmd->u.sedf.latency, (cmd->u.sedf.extratime)?"yes":"no"); + if ( cmd->direction == SCHED_INFO_PUT ) { /*check for sane parameters*/ @@ -1458,6 +1657,7 @@ .sleep = sedf_sleep, .wake = sedf_wake, .adjdom = sedf_adjdom, + .set_affinity = sedf_set_affinity }; /* diff -r 2884f14895d2 -r f5ab28885c75 xen/common/schedule.c --- a/xen/common/schedule.c Fri Jan 13 16:38:44 2006 +++ b/xen/common/schedule.c Fri Jan 13 20:12:24 2006 @@ -21,7 +21,7 @@ #include <xen/delay.h> #include <xen/event.h> #include <xen/time.h> -#include <xen/ac_timer.h> +#include <xen/timer.h> #include <xen/perfc.h> #include <xen/sched-if.h> #include <xen/softirq.h> @@ -71,7 +71,7 @@ : (typeof(ops.fn(__VA_ARGS__)))0 ) /* Per-CPU periodic timer sends an event to the currently-executing domain. */ -static struct ac_timer t_timer[NR_CPUS]; +static struct timer t_timer[NR_CPUS]; void free_domain(struct domain *d) { @@ -100,7 +100,9 @@ v->vcpu_id = vcpu_id; v->processor = cpu_id; atomic_set(&v->pausecnt, 0); - v->cpumap = CPUMAP_RUNANYWHERE; + + v->cpu_affinity = is_idle_domain(d) ? + cpumask_of_cpu(cpu_id) : CPU_MASK_ALL; d->vcpu[vcpu_id] = v; @@ -138,12 +140,10 @@ */ void sched_add_domain(struct vcpu *v) { - struct domain *d = v->domain; - /* Initialise the per-domain timer. */ - init_ac_timer(&v->timer, dom_timer_fn, v, v->processor); - - if ( is_idle_task(d) ) + init_timer(&v->timer, dom_timer_fn, v, v->processor); + + if ( is_idle_vcpu(v) ) { schedule_data[v->processor].curr = v; schedule_data[v->processor].idle = v; @@ -151,12 +151,12 @@ } SCHED_OP(add_task, v); - TRACE_2D(TRC_SCHED_DOM_ADD, d->domain_id, v->vcpu_id); + TRACE_2D(TRC_SCHED_DOM_ADD, v->domain->domain_id, v->vcpu_id); } void sched_rem_domain(struct vcpu *v) { - rem_ac_timer(&v->timer); + kill_timer(&v->timer); SCHED_OP(rem_task, v); TRACE_2D(TRC_SCHED_DOM_REM, v->domain->domain_id, v->vcpu_id); } @@ -165,26 +165,19 @@ { unsigned long flags; - spin_lock_irqsave(&schedule_data[v->processor].schedule_lock, flags); - if ( likely(!domain_runnable(v)) ) + vcpu_schedule_lock_irqsave(v, flags); + if ( likely(!vcpu_runnable(v)) ) SCHED_OP(sleep, v); - spin_unlock_irqrestore(&schedule_data[v->processor].schedule_lock, flags); + vcpu_schedule_unlock_irqrestore(v, flags); TRACE_2D(TRC_SCHED_SLEEP, v->domain->domain_id, v->vcpu_id); -} +} void vcpu_sleep_sync(struct vcpu *v) { vcpu_sleep_nosync(v); - /* - * We can be sure that the VCPU is finally descheduled after the running - * flag is cleared and the scheduler lock is released. We also check that - * the domain continues to be unrunnable, in case someone else wakes it. - */ - while ( !domain_runnable(v) && - (test_bit(_VCPUF_running, &v->vcpu_flags) || - spin_is_locked(&schedule_data[v->processor].schedule_lock)) ) + while ( !vcpu_runnable(v) && test_bit(_VCPUF_running, &v->vcpu_flags) ) cpu_relax(); sync_vcpu_execstate(v); @@ -194,16 +187,26 @@ { unsigned long flags; - spin_lock_irqsave(&schedule_data[v->processor].schedule_lock, flags); - if ( likely(domain_runnable(v)) ) + vcpu_schedule_lock_irqsave(v, flags); + if ( likely(vcpu_runnable(v)) ) { SCHED_OP(wake, v); v->wokenup = NOW(); } - clear_bit(_VCPUF_cpu_migrated, &v->vcpu_flags); - spin_unlock_irqrestore(&schedule_data[v->processor].schedule_lock, flags); + vcpu_schedule_unlock_irqrestore(v, flags); TRACE_2D(TRC_SCHED_WAKE, v->domain->domain_id, v->vcpu_id); +} + +int vcpu_set_affinity(struct vcpu *v, cpumask_t *affinity) +{ + cpumask_t online_affinity; + + cpus_and(online_affinity, *affinity, cpu_online_map); + if ( cpus_empty(online_affinity) ) + return -EINVAL; + + return SCHED_OP(set_affinity, v, affinity); } /* Block the currently-executing domain until a pertinent event occurs. */ @@ -275,9 +278,9 @@ struct vcpu *v = current; if ( timeout == 0 ) - rem_ac_timer(&v->timer); + stop_timer(&v->timer); else - set_ac_timer(&v->timer, timeout); + set_timer(&v->timer, timeout); return 0; } @@ -304,63 +307,42 @@ { struct domain *d; struct vcpu *v; - int cpu; -#if NR_CPUS <=32 - unsigned long have_lock; - #else - unsigned long long have_lock; -#endif - int succ; - - #define __set_cpu_bit(cpu, data) data |= ((typeof(data))1)<<cpu - #define __get_cpu_bit(cpu, data) (data & ((typeof(data))1)<<cpu) - #define __clear_cpu_bits(data) data = ((typeof(data))0) - if ( cmd->sched_id != ops.sched_id ) - return -EINVAL; - - if ( cmd->direction != SCHED_INFO_PUT && cmd->direction != SCHED_INFO_GET ) + if ( (cmd->sched_id != ops.sched_id) || + ((cmd->direction != SCHED_INFO_PUT) && + (cmd->direction != SCHED_INFO_GET)) ) return -EINVAL; d = find_domain_by_id(cmd->domain); if ( d == NULL ) return -ESRCH; - /* acquire locks on all CPUs on which vcpus of this domain run */ - do { - succ = 0; - __clear_cpu_bits(have_lock); - for_each_vcpu(d, v) { - cpu = v->processor; - if (!__get_cpu_bit(cpu, have_lock)) { - /* if we don't have a lock on this CPU: acquire it*/ - if (spin_trylock(&schedule_data[cpu].schedule_lock)) { - /*we have this lock!*/ - __set_cpu_bit(cpu, have_lock); - succ = 1; - } else { - /*we didn,t get this lock -> free all other locks too!*/ - for (cpu = 0; cpu < NR_CPUS; cpu++) - if (__get_cpu_bit(cpu, have_lock)) - spin_unlock(&schedule_data[cpu].schedule_lock); - /* and start from the beginning! */ - succ = 0; - /* leave the "for_each_domain_loop" */ - break; - } - } - } - } while ( !succ ); + /* + * Most VCPUs we can simply pause. If we are adjusting this VCPU then + * we acquire the local schedule_lock to guard against concurrent updates. + */ + for_each_vcpu ( d, v ) + { + if ( v == current ) + vcpu_schedule_lock_irq(v); + else + vcpu_pause(v); + } SCHED_OP(adjdom, d, cmd); - for (cpu = 0; cpu < NR_CPUS; cpu++) - if (__get_cpu_bit(cpu, have_lock)) - spin_unlock(&schedule_data[cpu].schedule_lock); - __clear_cpu_bits(have_lock); - TRACE_1D(TRC_SCHED_ADJDOM, d->domain_id); + + for_each_vcpu ( d, v ) + { + if ( v == current ) + vcpu_schedule_unlock_irq(v); + else + vcpu_unpause(v); + } + put_domain(d); + return 0; } @@ -371,22 +353,20 @@ */ static void __enter_scheduler(void) { - struct vcpu *prev = current, *next = NULL; - int cpu = prev->processor; - s_time_t now; + struct vcpu *prev = current, *next = NULL; + int cpu = smp_processor_id(); + s_time_t now = NOW(); struct task_slice next_slice; s32 r_time; /* time for new dom to run */ + ASSERT(!in_irq()); + perfc_incrc(sched_run); + + spin_lock_irq(&schedule_data[cpu].schedule_lock); + + stop_timer(&schedule_data[cpu].s_timer); - spin_lock_irq(&schedule_data[cpu].schedule_lock); - - now = NOW(); - - rem_ac_timer(&schedule_data[cpu].s_timer); - - ASSERT(!in_irq()); - prev->cpu_time += now - prev->lastschd; /* get policy-specific decision on scheduling... */ @@ -394,12 +374,12 @@ r_time = next_slice.time; next = next_slice.task; - + schedule_data[cpu].curr = next; next->lastschd = now; - set_ac_timer(&schedule_data[cpu].s_timer, now + r_time); + set_timer(&schedule_data[cpu].s_timer, now + r_time); if ( unlikely(prev == next) ) { @@ -411,11 +391,6 @@ prev->domain->domain_id, now - prev->lastschd); TRACE_3D(TRC_SCHED_SWITCH_INFNEXT, next->domain->domain_id, now - next->wokenup, r_time); - - clear_bit(_VCPUF_running, &prev->vcpu_flags); - set_bit(_VCPUF_running, &next->vcpu_flags); - - perfc_incrc(sched_ctx); /* * Logic of wokenup field in domain struct: @@ -425,10 +400,10 @@ * also set here then a preempted runnable domain will get a screwed up * "waiting time" value next time it is scheduled. */ - prev->wokenup = NOW(); + prev->wokenup = now; #if defined(WAKE_HISTO) - if ( !is_idle_task(next->domain) && next->wokenup ) + if ( !is_idle_vcpu(next) && next->wokenup ) { ulong diff = (ulong)(now - next->wokenup); diff /= (ulong)MILLISECS(1); @@ -438,7 +413,7 @@ next->wokenup = (s_time_t)0; #elif defined(BLOCKTIME_HISTO) prev->lastdeschd = now; - if ( !is_idle_task(next->domain) ) + if ( !is_idle_vcpu(next) ) { ulong diff = (ulong)((now - next->lastdeschd) / MILLISECS(10)); if (diff <= BUCKETS-2) schedule_data[cpu].hist[diff]++; @@ -446,10 +421,16 @@ } #endif + set_bit(_VCPUF_running, &next->vcpu_flags); + + spin_unlock_irq(&schedule_data[cpu].schedule_lock); + + perfc_incrc(sched_ctx); + prev->sleep_tick = schedule_data[cpu].tick; /* Ensure that the domain has an up-to-date time base. */ - if ( !is_idle_task(next->domain) ) + if ( !is_idle_vcpu(next) ) { update_dom_time(next); if ( next->sleep_tick != schedule_data[cpu].tick ) @@ -461,17 +442,6 @@ next->domain->domain_id, next->vcpu_id); context_switch(prev, next); - - spin_unlock_irq(&schedule_data[cpu].schedule_lock); - - context_switch_finalise(next); -} - -/* No locking needed -- pointer comparison is safe :-) */ -int idle_cpu(int cpu) -{ - struct vcpu *p = schedule_data[cpu].curr; - return p == idle_task[cpu]; } @@ -492,12 +462,12 @@ /* Periodic tick timer: send timer event to current domain */ static void t_timer_fn(void *unused) { - struct vcpu *v = current; - unsigned int cpu = v->processor; + struct vcpu *v = current; + unsigned int cpu = smp_processor_id(); schedule_data[cpu].tick++; - if ( !is_idle_task(v->domain) ) + if ( !is_idle_vcpu(v) ) { update_dom_time(v); send_guest_virq(v, VIRQ_TIMER); @@ -505,7 +475,7 @@ page_scrub_schedule_work(); - set_ac_timer(&t_timer[cpu], NOW() + MILLISECS(10)); + set_timer(&t_timer[cpu], NOW() + MILLISECS(10)); } /* Domain timer function, sends a virtual timer interrupt to domain */ @@ -527,12 +497,9 @@ for ( i = 0; i < NR_CPUS; i++ ) { spin_lock_init(&schedule_data[i].schedule_lock); - init_ac_timer(&schedule_data[i].s_timer, s_timer_fn, NULL, i); - init_ac_timer(&t_timer[i], t_timer_fn, NULL, i); - } - - schedule_data[0].curr = idle_task[0]; - schedule_data[0].idle = idle_task[0]; + init_timer(&schedule_data[i].s_timer, s_timer_fn, NULL, i); + init_timer(&t_timer[i], t_timer_fn, NULL, i); + } for ( i = 0; schedulers[i] != NULL; i++ ) { @@ -546,10 +513,16 @@ printk("Using scheduler: %s (%s)\n", ops.name, ops.opt_name); - rc = SCHED_OP(alloc_task, idle_task[0]); - BUG_ON(rc < 0); - - sched_add_domain(idle_task[0]); + if ( idle_vcpu[0] != NULL ) + { + schedule_data[0].curr = idle_vcpu[0]; + schedule_data[0].idle = idle_vcpu[0]; + + rc = SCHED_OP(alloc_task, idle_vcpu[0]); + BUG_ON(rc < 0); + + sched_add_domain(idle_vcpu[0]); + } } /* diff -r 2884f14895d2 -r f5ab28885c75 xen/common/vsprintf.c --- a/xen/common/vsprintf.c Fri Jan 13 16:38:44 2006 +++ b/xen/common/vsprintf.c Fri Jan 13 20:12:24 2006 @@ -12,11 +12,15 @@ /* * Fri Jul 13 2001 Crutcher Dunnavant <crutcher+kernel@xxxxxxxxxxxxxx> * - changed to provide snprintf and vsnprintf functions + * So Feb 1 16:51:32 CET 2004 Juergen Quade <quade@xxxxxxx> + * - scnprintf and vscnprintf */ #include <stdarg.h> #include <xen/ctype.h> #include <xen/lib.h> +#include <asm/div64.h> +#include <asm/page.h> /** * simple_strtoul - convert a string to an unsigned long @@ -33,11 +37,14 @@ if (*cp == '0') { base = 8; cp++; - if ((*cp == 'x') && isxdigit(cp[1])) { + if ((toupper(*cp) == 'X') && isxdigit(cp[1])) { cp++; base = 16; } } + } else if (base == 16) { + if (cp[0] == '0' && toupper(cp[1]) == 'X') + cp += 2; } while (isxdigit(*cp) && (value = isdigit(*cp) ? *cp-'0' : toupper(*cp)-'A'+10) < base) { @@ -49,6 +56,8 @@ return result; } +EXPORT_SYMBOL(simple_strtoul); + /** * simple_strtol - convert a string to a signed long * @cp: The start of the string @@ -61,6 +70,8 @@ return -simple_strtoul(cp+1,endp,base); return simple_strtoul(cp,endp,base); } + +EXPORT_SYMBOL(simple_strtol); /** * simple_strtoull - convert a string to an unsigned long long @@ -77,11 +88,14 @@ if (*cp == '0') { base = 8; cp++; - if ((*cp == 'x') && isxdigit(cp[1])) { + if ((toupper(*cp) == 'X') && isxdigit(cp[1])) { cp++; base = 16; } } + } else if (base == 16) { + if (cp[0] == '0' && toupper(cp[1]) == 'X') + cp += 2; } while (isxdigit(*cp) && (value = isdigit(*cp) ? *cp-'0' : (islower(*cp) ? toupper(*cp) : *cp)-'A'+10) < base) { @@ -92,6 +106,8 @@ *endp = (char *)cp; return result; } + +EXPORT_SYMBOL(simple_strtoull); /** * simple_strtoll - convert a string to a signed long long @@ -123,25 +139,25 @@ #define SPECIAL 32 /* 0x */ #define LARGE 64 /* use 'ABCDEF' instead of 'abcdef' */ -static char * number(char * buf, char * end, long long num, int base, int size, int precision, int type) +static char * number(char * buf, char * end, unsigned long long num, int base, int size, int precision, int type) { char c,sign,tmp[66]; const char *digits; - const char small_digits[] = "0123456789abcdefghijklmnopqrstuvwxyz"; - const char large_digits[] = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"; + static const char small_digits[] = "0123456789abcdefghijklmnopqrstuvwxyz"; + static const char large_digits[] = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"; int i; digits = (type & LARGE) ? large_digits : small_digits; if (type & LEFT) type &= ~ZEROPAD; if (base < 2 || base > 36) - return buf; + return NULL; c = (type & ZEROPAD) ? '0' : ' '; sign = 0; if (type & SIGN) { - if (num < 0) { + if ((signed long long) num < 0) { sign = '-'; - num = -num; + num = - (signed long long) num; size--; } else if (type & PLUS) { sign = '+'; @@ -160,6 +176,9 @@ i = 0; if (num == 0) tmp[i++]='0'; + else while (num != 0) + tmp[i++] = digits[do_div(num,base)]; +#if 0 else { /* XXX KAF: force unsigned mod and div. */ @@ -167,6 +186,7 @@ unsigned int base2=(unsigned int)base; while (num2 != 0) { tmp[i++] = digits[num2%base2]; num2 /= base2; } } +#endif if (i > precision) precision = i; size -= precision; @@ -222,14 +242,22 @@ } /** -* vsnprintf - Format a string and place it in a buffer -* @buf: The buffer to place the result into -* @size: The size of the buffer, including the trailing null space -* @fmt: The format string to use -* @args: Arguments for the format string -* -* Call this function if you are already dealing with a va_list. -* You probably want snprintf instead. + * vsnprintf - Format a string and place it in a buffer + * @buf: The buffer to place the result into + * @size: The size of the buffer, including the trailing null space + * @fmt: The format string to use + * @args: Arguments for the format string + * + * The return value is the number of characters which would + * be generated for the given input, excluding the trailing + * '\0', as per ISO C99. If you want to have the exact + * number of characters written into @buf as return value + * (not including the trailing '\0'), use vscnprintf. If the + * return is greater than or equal to @size, the resulting + * string is truncated. + * + * Call this function if you are already dealing with a va_list. + * You probably want snprintf instead. */ int vsnprintf(char *buf, size_t size, const char *fmt, va_list args) { @@ -247,6 +275,9 @@ int qualifier; /* 'h', 'l', or 'L' for integer fields */ /* 'z' support added 23/7/1999 S.H. */ /* 'z' changed to 'Z' --davidm 1/25/99 */ + + /* Reject out-of-range values early */ + BUG_ON((int)size < 0); str = buf; end = buf + size - 1; @@ -307,17 +338,14 @@ /* get the conversion qualifier */ qualifier = -1; - if (*fmt == 'h' || *fmt == 'l' || *fmt == 'L' || *fmt =='Z') { + if (*fmt == 'h' || *fmt == 'l' || *fmt == 'L' || + *fmt =='Z' || *fmt == 'z') { qualifier = *fmt; ++fmt; if (qualifier == 'l' && *fmt == 'l') { qualifier = 'L'; ++fmt; } - } - if (*fmt == 'q') { - qualifier = 'L'; - ++fmt; } /* default base */ @@ -345,7 +373,7 @@ case 's': s = va_arg(args, char *); - if (!s) + if ((unsigned long)s < PAGE_SIZE) s = "<NULL>"; len = strnlen(s, precision); @@ -386,7 +414,7 @@ if (qualifier == 'l') { long * ip = va_arg(args, long *); *ip = (str - buf); - } else if (qualifier == 'Z') { + } else if (qualifier == 'Z' || qualifier == 'z') { size_t * ip = va_arg(args, size_t *); *ip = (str - buf); } else { @@ -437,7 +465,7 @@ num = va_arg(args, unsigned long); if (flags & SIGN) num = (signed long) num; - } else if (qualifier == 'Z') { + } else if (qualifier == 'Z' || qualifier == 'z') { num = va_arg(args, size_t); } else if (qualifier == 'h') { num = (unsigned short) va_arg(args, int); @@ -463,12 +491,43 @@ return str-buf; } +EXPORT_SYMBOL(vsnprintf); + +/** + * vscnprintf - Format a string and place it in a buffer + * @buf: The buffer to place the result into + * @size: The size of the buffer, including the trailing null space + * @fmt: The format string to use + * @args: Arguments for the format string + * + * The return value is the number of characters which have been written into + * the @buf not including the trailing '\0'. If @size is <= 0 the function + * returns 0. + * + * Call this function if you are already dealing with a va_list. + * You probably want scnprintf instead. + */ +int vscnprintf(char *buf, size_t size, const char *fmt, va_list args) +{ + int i; + + i=vsnprintf(buf,size,fmt,args); + return (i >= size) ? (size - 1) : i; +} + +EXPORT_SYMBOL(vscnprintf); + /** * snprintf - Format a string and place it in a buffer * @buf: The buffer to place the result into * @size: The size of the buffer, including the trailing null space * @fmt: The format string to use * @...: Arguments for the format string + * + * The return value is the number of characters which would be + * generated for the given input, excluding the trailing null, + * as per ISO C99. If the return is greater than or equal to + * @size, the resulting string is truncated. */ int snprintf(char * buf, size_t size, const char *fmt, ...) { @@ -481,26 +540,61 @@ return i; } +EXPORT_SYMBOL(snprintf); + +/** + * scnprintf - Format a string and place it in a buffer + * @buf: The buffer to place the result into + * @size: The size of the buffer, including the trailing null space + * @fmt: The format string to use + * @...: Arguments for the format string + * + * The return value is the number of characters written into @buf not including + * the trailing '\0'. If @size is <= 0 the function returns 0. If the return is + * greater than or equal to @size, the resulting string is truncated. + */ + +int scnprintf(char * buf, size_t size, const char *fmt, ...) +{ + va_list args; + int i; + + va_start(args, fmt); + i = vsnprintf(buf, size, fmt, args); + va_end(args); + return (i >= size) ? (size - 1) : i; +} +EXPORT_SYMBOL(scnprintf); + /** * vsprintf - Format a string and place it in a buffer * @buf: The buffer to place the result into * @fmt: The format string to use * @args: Arguments for the format string * + * The function returns the number of characters written + * into @buf. Use vsnprintf or vscnprintf in order to avoid + * buffer overflows. + * * Call this function if you are already dealing with a va_list. * You probably want sprintf instead. */ int vsprintf(char *buf, const char *fmt, va_list args) { - return vsnprintf(buf, 0xFFFFFFFFUL, fmt, args); -} - + return vsnprintf(buf, INT_MAX, fmt, args); +} + +EXPORT_SYMBOL(vsprintf); /** * sprintf - Format a string and place it in a buffer * @buf: The buffer to place the result into * @fmt: The format string to use * @...: Arguments for the format string + * + * The function returns the number of characters written + * into @buf. Use snprintf or scnprintf in order to avoid + * buffer overflows. */ int sprintf(char * buf, const char *fmt, ...) { @@ -508,11 +602,12 @@ int i; va_start(args, fmt); - i=vsprintf(buf,fmt,args); + i=vsnprintf(buf, INT_MAX, fmt, args); va_end(args); return i; } +EXPORT_SYMBOL(sprintf); /* * Local variables: diff -r 2884f14895d2 -r f5ab28885c75 xen/common/xmalloc.c --- a/xen/common/xmalloc.c Fri Jan 13 16:38:44 2006 +++ b/xen/common/xmalloc.c Fri Jan 13 20:12:24 2006 @@ -30,7 +30,7 @@ #include <xen/config.h> #include <xen/mm.h> #include <xen/spinlock.h> -#include <xen/ac_timer.h> +#include <xen/timer.h> #include <xen/cache.h> #include <xen/prefetch.h> diff -r 2884f14895d2 -r f5ab28885c75 xen/drivers/char/console.c --- a/xen/drivers/char/console.c Fri Jan 13 16:38:44 2006 +++ b/xen/drivers/char/console.c Fri Jan 13 20:12:24 2006 @@ -311,7 +311,7 @@ long guest_console_write(char *buffer, int count) { - char kbuf[128]; + char kbuf[128], *kptr; int kcount; while ( count > 0 ) @@ -333,6 +333,9 @@ kbuf[kcount] = '\0'; serial_puts(sercon_handle, kbuf); + + for ( kptr = kbuf; *kptr != '\0'; kptr++ ) + putchar_console(*kptr); buffer += kcount; count -= kcount; diff -r 2884f14895d2 -r f5ab28885c75 xen/drivers/char/ns16550.c --- a/xen/drivers/char/ns16550.c Fri Jan 13 16:38:44 2006 +++ b/xen/drivers/char/ns16550.c Fri Jan 13 20:12:24 2006 @@ -33,7 +33,7 @@ /* UART with IRQ line: interrupt-driven I/O. */ struct irqaction irqaction; /* UART with no IRQ line: periodically-polled I/O. */ - struct ac_timer timer; + struct timer timer; unsigned int timeout_ms; } ns16550_com[2] = { { 0 } }; @@ -138,7 +138,7 @@ if ( ns_read_reg(uart, LSR) & LSR_THRE ) serial_tx_interrupt(port, regs); - set_ac_timer(&uart->timer, NOW() + MILLISECS(uart->timeout_ms)); + set_timer(&uart->timer, NOW() + MILLISECS(uart->timeout_ms)); } static int ns16550_tx_empty(struct serial_port *port) @@ -214,8 +214,8 @@ bits = uart->data_bits + uart->stop_bits + !!uart->parity; uart->timeout_ms = max_t( unsigned int, 1, (bits * port->tx_fifo_size * 1000) / uart->baud); - init_ac_timer(&uart->timer, ns16550_poll, port, 0); - set_ac_timer(&uart->timer, NOW() + MILLISECS(uart->timeout_ms)); + init_timer(&uart->timer, ns16550_poll, port, 0); + set_timer(&uart->timer, NOW() + MILLISECS(uart->timeout_ms)); } else { diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-ia64/config.h --- a/xen/include/asm-ia64/config.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-ia64/config.h Fri Jan 13 20:12:24 2006 @@ -141,10 +141,6 @@ #undef alloc_task_struct #define get_thread_info(v) alloc_thread_info(v) -// initial task has a different name in Xen -//#define idle0_task init_task -#define idle0_vcpu init_task - // avoid redefining task_t in asm/thread_info.h #define task_t struct domain @@ -160,7 +156,7 @@ #define platform_outl __ia64_outl // FIXME: This just overrides a use in a typedef (not allowed in ia64, -// or maybe just in older gcc's?) used in ac_timer.c but should be OK +// or maybe just in older gcc's?) used in timer.c but should be OK // (and indeed is probably required!) elsewhere #undef __cacheline_aligned #undef ____cacheline_aligned diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-ia64/vmx.h --- a/xen/include/asm-ia64/vmx.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-ia64/vmx.h Fri Jan 13 20:12:24 2006 @@ -23,7 +23,7 @@ #define _ASM_IA64_VT_H #define RR7_SWITCH_SHIFT 12 /* 4k enough */ -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> extern void identify_vmx_feature(void); extern unsigned int vmx_enabled; diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-ia64/vtm.h --- a/xen/include/asm-ia64/vtm.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-ia64/vtm.h Fri Jan 13 20:12:24 2006 @@ -23,7 +23,7 @@ #ifndef _VTM_H_ #define _VTM_H_ -#include <xen/ac_timer.h> +#include <xen/timer.h> #include <xen/types.h> #define MAX_JUMP_STEP (5000) /* 500ms, max jump step */ @@ -46,7 +46,7 @@ uint64_t cfg_max_jump; // max jump within one time suspendsion uint64_t cfg_min_grun; // min guest running time since last jump // uint64_t latest_read_itc; // latest guest read ITC - struct ac_timer vtm_timer; + struct timer vtm_timer; // int triggered; diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-x86/config.h --- a/xen/include/asm-x86/config.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-x86/config.h Fri Jan 13 20:12:24 2006 @@ -148,7 +148,8 @@ #define SH_LINEAR_PT_VIRT_END (SH_LINEAR_PT_VIRT_START + PML4_ENTRY_BYTES) /* Slot 260: per-domain mappings. */ #define PERDOMAIN_VIRT_START (PML4_ADDR(260)) -#define PERDOMAIN_VIRT_END (PERDOMAIN_VIRT_START + PML4_ENTRY_BYTES) +#define PERDOMAIN_VIRT_END (PERDOMAIN_VIRT_START + (PERDOMAIN_MBYTES<<20)) +#define PERDOMAIN_MBYTES ((unsigned long)GDT_LDT_MBYTES) /* Slot 261: machine-to-phys conversion table (16GB). */ #define RDWR_MPT_VIRT_START (PML4_ADDR(261)) #define RDWR_MPT_VIRT_END (RDWR_MPT_VIRT_START + (16UL<<30)) @@ -195,8 +196,7 @@ * ------ ------ * I/O remapping area ( 4MB) * Direct-map (1:1) area [Xen code/data/heap] (12MB) - * map_domain_page cache ( 4MB) - * Per-domain mappings ( 4MB) + * Per-domain mappings (inc. 4MB map_domain_page cache) ( 4MB) * Shadow linear pagetable ( 4MB) ( 8MB) * Guest linear pagetable ( 4MB) ( 8MB) * Machine-to-physical translation table [writable] ( 4MB) (16MB) @@ -209,7 +209,7 @@ #define IOREMAP_MBYTES 4 #define DIRECTMAP_MBYTES 12 #define MAPCACHE_MBYTES 4 -#define PERDOMAIN_MBYTES 4 +#define PERDOMAIN_MBYTES 8 #ifdef CONFIG_X86_PAE # define LINEARPT_MBYTES 8 @@ -227,7 +227,7 @@ #define DIRECTMAP_VIRT_START (DIRECTMAP_VIRT_END - (DIRECTMAP_MBYTES<<20)) #define MAPCACHE_VIRT_END DIRECTMAP_VIRT_START #define MAPCACHE_VIRT_START (MAPCACHE_VIRT_END - (MAPCACHE_MBYTES<<20)) -#define PERDOMAIN_VIRT_END MAPCACHE_VIRT_START +#define PERDOMAIN_VIRT_END DIRECTMAP_VIRT_START #define PERDOMAIN_VIRT_START (PERDOMAIN_VIRT_END - (PERDOMAIN_MBYTES<<20)) #define SH_LINEAR_PT_VIRT_END PERDOMAIN_VIRT_START #define SH_LINEAR_PT_VIRT_START (SH_LINEAR_PT_VIRT_END - (LINEARPT_MBYTES<<20)) @@ -248,12 +248,10 @@ #ifdef CONFIG_X86_PAE /* Hypervisor owns top 168MB of virtual address space. */ -# define __HYPERVISOR_VIRT_START 0xF5800000 -# define HYPERVISOR_VIRT_START (0xF5800000UL) +#define HYPERVISOR_VIRT_START mk_unsigned_long(0xF5800000) #else /* Hypervisor owns top 64MB of virtual address space. */ -# define __HYPERVISOR_VIRT_START 0xFC000000 -# define HYPERVISOR_VIRT_START (0xFC000000UL) +#define HYPERVISOR_VIRT_START mk_unsigned_long(0xFC000000) #endif #define L2_PAGETABLE_FIRST_XEN_SLOT \ @@ -284,14 +282,21 @@ extern unsigned long xenheap_phys_end; /* user-configurable */ #endif -#define GDT_VIRT_START(ed) \ - (PERDOMAIN_VIRT_START + ((ed)->vcpu_id << PDPT_VCPU_VA_SHIFT)) -#define LDT_VIRT_START(ed) \ - (GDT_VIRT_START(ed) + (64*1024)) - -#define PDPT_VCPU_SHIFT 5 -#define PDPT_VCPU_VA_SHIFT (PDPT_VCPU_SHIFT + PAGE_SHIFT) -#define PDPT_L1_ENTRIES (MAX_VIRT_CPUS << PDPT_VCPU_SHIFT) +/* GDT/LDT shadow mapping area. The first per-domain-mapping sub-area. */ +#define GDT_LDT_VCPU_SHIFT 5 +#define GDT_LDT_VCPU_VA_SHIFT (GDT_LDT_VCPU_SHIFT + PAGE_SHIFT) +#define GDT_LDT_MBYTES (MAX_VIRT_CPUS >> (20-GDT_LDT_VCPU_VA_SHIFT)) +#define GDT_LDT_VIRT_START PERDOMAIN_VIRT_START +#define GDT_LDT_VIRT_END (GDT_LDT_VIRT_START + (GDT_LDT_MBYTES << 20)) + +/* The address of a particular VCPU's GDT or LDT. */ +#define GDT_VIRT_START(v) \ + (PERDOMAIN_VIRT_START + ((v)->vcpu_id << GDT_LDT_VCPU_VA_SHIFT)) +#define LDT_VIRT_START(v) \ + (GDT_VIRT_START(v) + (64*1024)) + +#define PDPT_L1_ENTRIES \ + ((PERDOMAIN_VIRT_END - PERDOMAIN_VIRT_START) >> PAGE_SHIFT) #define PDPT_L2_ENTRIES \ ((PDPT_L1_ENTRIES + (1 << PAGETABLE_ORDER) - 1) >> PAGETABLE_ORDER) diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-x86/domain.h --- a/xen/include/asm-x86/domain.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-x86/domain.h Fri Jan 13 20:12:24 2006 @@ -13,12 +13,55 @@ unsigned long eip; }; +#define MAPHASH_ENTRIES 8 +#define MAPHASH_HASHFN(pfn) ((pfn) & (MAPHASH_ENTRIES-1)) +#define MAPHASHENT_NOTINUSE ((u16)~0U) +struct vcpu_maphash { + struct vcpu_maphash_entry { + unsigned long pfn; + uint16_t idx; + uint16_t refcnt; + } hash[MAPHASH_ENTRIES]; +} __cacheline_aligned; + +#define MAPCACHE_ORDER 10 +#define MAPCACHE_ENTRIES (1 << MAPCACHE_ORDER) +struct mapcache { + /* The PTEs that provide the mappings, and a cursor into the array. */ + l1_pgentry_t *l1tab; + unsigned int cursor; + + /* Protects map_domain_page(). */ + spinlock_t lock; + + /* Garbage mappings are flushed from TLBs in batches called 'epochs'. */ + unsigned int epoch, shadow_epoch[MAX_VIRT_CPUS]; + u32 tlbflush_timestamp; + + /* Which mappings are in use, and which are garbage to reap next epoch? */ + unsigned long inuse[BITS_TO_LONGS(MAPCACHE_ENTRIES)]; + unsigned long garbage[BITS_TO_LONGS(MAPCACHE_ENTRIES)]; + + /* Lock-free per-VCPU hash of recently-used mappings. */ + struct vcpu_maphash vcpu_maphash[MAX_VIRT_CPUS]; +}; + +extern void mapcache_init(struct domain *); + +/* x86/64: toggle guest between kernel and user modes. */ +extern void toggle_guest_mode(struct vcpu *); + struct arch_domain { l1_pgentry_t *mm_perdomain_pt; #ifdef CONFIG_X86_64 l2_pgentry_t *mm_perdomain_l2; l3_pgentry_t *mm_perdomain_l3; +#endif + +#ifdef CONFIG_X86_32 + /* map_domain_page() mapping cache. */ + struct mapcache mapcache; #endif /* Writable pagetables. */ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-x86/mm.h --- a/xen/include/asm-x86/mm.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-x86/mm.h Fri Jan 13 20:12:24 2006 @@ -309,16 +309,13 @@ unsigned long l1va; /* Copy of the p.t. page, taken before guest is given write access. */ l1_pgentry_t *page; - /* A temporary Xen mapping of the actual p.t. page. */ - l1_pgentry_t *pl1e; /* Index in L2 page table where this L1 p.t. is always hooked. */ unsigned int l2_idx; /* NB. Only used for PTWR_PT_ACTIVE. */ /* Info about last ptwr update batch. */ unsigned int prev_nr_updates; - /* Exec domain which created writable mapping. */ + /* VCPU which created writable mapping. */ struct vcpu *vcpu; - /* EIP of the address which took the original write fault - used for stats collection only */ + /* EIP of the original write fault (stats collection only). */ unsigned long eip; }; @@ -336,11 +333,13 @@ int revalidate_l1(struct domain *, l1_pgentry_t *, l1_pgentry_t *); void cleanup_writable_pagetable(struct domain *d); -#define sync_pagetable_state(d) \ - do { \ - LOCK_BIGLOCK(d); \ - cleanup_writable_pagetable(d); \ - UNLOCK_BIGLOCK(d); \ +#define sync_pagetable_state(d) \ + do { \ + LOCK_BIGLOCK(d); \ + /* Avoid racing with ptwr_destroy(). */ \ + if ( !test_bit(_DOMF_dying, &(d)->domain_flags) ) \ + cleanup_writable_pagetable(d); \ + UNLOCK_BIGLOCK(d); \ } while ( 0 ) int audit_adjust_pgtables(struct domain *d, int dir, int noisy); diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-x86/nmi.h --- a/xen/include/asm-x86/nmi.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-x86/nmi.h Fri Jan 13 20:12:24 2006 @@ -1,6 +1,8 @@ #ifndef ASM_NMI_H #define ASM_NMI_H + +#include <public/nmi.h> struct cpu_user_regs; diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-x86/processor.h --- a/xen/include/asm-x86/processor.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-x86/processor.h Fri Jan 13 20:12:24 2006 @@ -123,6 +123,7 @@ #define TBF_EXCEPTION_ERRCODE 2 #define TBF_INTERRUPT 8 #define TBF_FAILSAFE 16 +#define TBF_SLOW_IRET 32 /* 'arch_vcpu' flags values */ #define _TF_kernel_mode 0 @@ -190,7 +191,7 @@ #ifdef CONFIG_X86_HT extern void detect_ht(struct cpuinfo_x86 *c); #else -static inline void detect_ht(struct cpuinfo_x86 *c) {} +static always_inline void detect_ht(struct cpuinfo_x86 *c) {} #endif /* @@ -209,7 +210,7 @@ /* * CPUID functions returning a single datum */ -static inline unsigned int cpuid_eax(unsigned int op) +static always_inline unsigned int cpuid_eax(unsigned int op) { unsigned int eax; @@ -219,7 +220,7 @@ : "bx", "cx", "dx"); return eax; } -static inline unsigned int cpuid_ebx(unsigned int op) +static always_inline unsigned int cpuid_ebx(unsigned int op) { unsigned int eax, ebx; @@ -229,7 +230,7 @@ : "cx", "dx" ); return ebx; } -static inline unsigned int cpuid_ecx(unsigned int op) +static always_inline unsigned int cpuid_ecx(unsigned int op) { unsigned int eax, ecx; @@ -239,7 +240,7 @@ : "bx", "dx" ); return ecx; } -static inline unsigned int cpuid_edx(unsigned int op) +static always_inline unsigned int cpuid_edx(unsigned int op) { unsigned int eax, edx; @@ -281,7 +282,7 @@ */ extern unsigned long mmu_cr4_features; -static inline void set_in_cr4 (unsigned long mask) +static always_inline void set_in_cr4 (unsigned long mask) { unsigned long dummy; mmu_cr4_features |= mask; @@ -292,7 +293,7 @@ : "=&r" (dummy) : "irg" (mask) ); } -static inline void clear_in_cr4 (unsigned long mask) +static always_inline void clear_in_cr4 (unsigned long mask) { unsigned long dummy; mmu_cr4_features &= ~mask; @@ -334,7 +335,7 @@ outb((data), 0x23); \ } while (0) -static inline void __monitor(const void *eax, unsigned long ecx, +static always_inline void __monitor(const void *eax, unsigned long ecx, unsigned long edx) { /* "monitor %eax,%ecx,%edx;" */ @@ -343,7 +344,7 @@ : :"a" (eax), "c" (ecx), "d"(edx)); } -static inline void __mwait(unsigned long eax, unsigned long ecx) +static always_inline void __mwait(unsigned long eax, unsigned long ecx) { /* "mwait %eax,%ecx;" */ asm volatile( @@ -460,7 +461,7 @@ }; /* REP NOP (PAUSE) is a good thing to insert into busy-wait loops. */ -static inline void rep_nop(void) +static always_inline void rep_nop(void) { __asm__ __volatile__ ( "rep;nop" : : : "memory" ); } @@ -471,7 +472,7 @@ #ifdef CONFIG_MPENTIUMIII #define ARCH_HAS_PREFETCH -extern inline void prefetch(const void *x) +extern always_inline void prefetch(const void *x) { __asm__ __volatile__ ("prefetchnta (%0)" : : "r"(x)); } @@ -482,12 +483,12 @@ #define ARCH_HAS_PREFETCHW #define ARCH_HAS_SPINLOCK_PREFETCH -extern inline void prefetch(const void *x) +extern always_inline void prefetch(const void *x) { __asm__ __volatile__ ("prefetch (%0)" : : "r"(x)); } -extern inline void prefetchw(const void *x) +extern always_inline void prefetchw(const void *x) { __asm__ __volatile__ ("prefetchw (%0)" : : "r"(x)); } diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-x86/shadow.h --- a/xen/include/asm-x86/shadow.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-x86/shadow.h Fri Jan 13 20:12:24 2006 @@ -341,10 +341,10 @@ #if SHADOW_VERBOSE_DEBUG #define SH_LOG(_f, _a...) \ printk("DOM%uP%u: SH_LOG(%d): " _f "\n", \ - current->domain->domain_id , current->processor, __LINE__ , ## _a ) + current->domain->domain_id , smp_processor_id(), __LINE__ , ## _a ) #define SH_VLOG(_f, _a...) \ printk("DOM%uP%u: SH_VLOG(%d): " _f "\n", \ - current->domain->domain_id, current->processor, __LINE__ , ## _a ) + current->domain->domain_id, smp_processor_id(), __LINE__ , ## _a ) #else #define SH_LOG(_f, _a...) ((void)0) #define SH_VLOG(_f, _a...) ((void)0) @@ -353,7 +353,7 @@ #if SHADOW_VVERBOSE_DEBUG #define SH_VVLOG(_f, _a...) \ printk("DOM%uP%u: SH_VVLOG(%d): " _f "\n", \ - current->domain->domain_id, current->processor, __LINE__ , ## _a ) + current->domain->domain_id, smp_processor_id(), __LINE__ , ## _a ) #else #define SH_VVLOG(_f, _a...) ((void)0) #endif @@ -361,7 +361,7 @@ #if SHADOW_VVVERBOSE_DEBUG #define SH_VVVLOG(_f, _a...) \ printk("DOM%uP%u: SH_VVVLOG(%d): " _f "\n", \ - current->domain->domain_id, current->processor, __LINE__ , ## _a ) + current->domain->domain_id, smp_processor_id(), __LINE__ , ## _a ) #else #define SH_VVVLOG(_f, _a...) ((void)0) #endif @@ -369,7 +369,7 @@ #if FULLSHADOW_DEBUG #define FSH_LOG(_f, _a...) \ printk("DOM%uP%u: FSH_LOG(%d): " _f "\n", \ - current->domain->domain_id, current->processor, __LINE__ , ## _a ) + current->domain->domain_id, smp_processor_id(), __LINE__ , ## _a ) #else #define FSH_LOG(_f, _a...) ((void)0) #endif @@ -591,7 +591,7 @@ if ( need_flush ) { perfc_incrc(update_hl2e_invlpg); - flush_tlb_one_mask(v->domain->cpumask, + flush_tlb_one_mask(v->domain->domain_dirty_cpumask, &linear_pg_table[l1_linear_offset(va)]); } } diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-x86/vmx.h --- a/xen/include/asm-x86/vmx.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-x86/vmx.h Fri Jan 13 20:12:24 2006 @@ -26,7 +26,7 @@ #include <asm/vmx_vmcs.h> #include <asm/i387.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> extern int hvm_enabled; @@ -38,7 +38,6 @@ extern void arch_vmx_do_launch(struct vcpu *); extern void arch_vmx_do_resume(struct vcpu *); -extern void arch_vmx_do_relaunch(struct vcpu *); extern unsigned int cpu_rev; @@ -506,7 +505,7 @@ static inline unsigned int vmx_get_vcpu_nr(struct domain *d) { - return d->arch.vmx_platform.nr_vcpu; + return d->arch.vmx_platform.nr_vcpus; } static inline shared_iopage_t *get_sp(struct domain *d) diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-x86/vmx_intercept.h --- a/xen/include/asm-x86/vmx_intercept.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-x86/vmx_intercept.h Fri Jan 13 20:12:24 2006 @@ -6,7 +6,7 @@ #include <xen/lib.h> #include <xen/time.h> #include <xen/errno.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> #define MAX_IO_HANDLER 8 diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-x86/vmx_platform.h --- a/xen/include/asm-x86/vmx_platform.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-x86/vmx_platform.h Fri Jan 13 20:12:24 2006 @@ -33,10 +33,10 @@ (((size_reg) << 24) | ((index) << 16) | ((seg) << 8) | (flag)) #define operand_size(operand) \ - ((operand >> 24) & 0xFF) + ((operand >> 24) & 0xFF) #define operand_index(operand) \ - ((operand >> 16) & 0xFF) + ((operand >> 16) & 0xFF) /* for instruction.operand[].size */ #define BYTE 1 @@ -81,13 +81,13 @@ struct vmx_platform { unsigned long shared_page_va; - unsigned int nr_vcpu; - unsigned int lapic_enable; + unsigned int nr_vcpus; + unsigned int apic_enabled; struct vmx_virpit vmx_pit; struct vmx_io_handler vmx_io_handler; struct vmx_virpic vmx_pic; - struct vmx_vioapic vmx_vioapic; + struct vmx_vioapic vmx_vioapic; unsigned char round_info[256]; spinlock_t round_robin_lock; int interrupt_request; diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-x86/vmx_vlapic.h --- a/xen/include/asm-x86/vmx_vlapic.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-x86/vmx_vlapic.h Fri Jan 13 20:12:24 2006 @@ -21,7 +21,7 @@ #define VMX_VLAPIC_H #include <asm/msr.h> -#include <public/io/ioreq.h> +#include <public/hvm/ioreq.h> #if defined(__i386__) || defined(__x86_64__) static inline int __fls(uint32_t word) @@ -187,7 +187,7 @@ uint32_t timer_current; uint32_t timer_divconf; uint32_t timer_divide_counter; - struct ac_timer vlapic_timer; + struct timer vlapic_timer; int intr_pending_count[MAX_VECTOR]; s_time_t timer_current_update; uint32_t icr_high; @@ -216,7 +216,7 @@ static inline int vlapic_timer_active(struct vlapic *vlapic) { - return active_ac_timer(&(vlapic->vlapic_timer)); + return active_timer(&(vlapic->vlapic_timer)); } int vlapic_find_highest_irr(struct vlapic *vlapic); diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-x86/vmx_vmcs.h --- a/xen/include/asm-x86/vmx_vmcs.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-x86/vmx_vmcs.h Fri Jan 13 20:12:24 2006 @@ -23,7 +23,7 @@ #include <asm/vmx_cpu.h> #include <asm/vmx_platform.h> #include <asm/vmx_vlapic.h> -#include <public/vmx_assist.h> +#include <public/hvm/vmx_assist.h> extern int start_vmx(void); extern void stop_vmx(void); @@ -86,7 +86,8 @@ #define PC_DEBUG_PORT 0x80 struct arch_vmx_struct { - struct vmcs_struct *vmcs; /* VMCS pointer in virtual */ + struct vmcs_struct *vmcs; /* VMCS pointer in virtual. */ + unsigned int launch_cpu; /* VMCS is valid on this CPU. */ unsigned long flags; /* VMCS flags */ unsigned long cpu_cr0; /* copy of guest CR0 */ unsigned long cpu_shadow_cr0; /* copy of guest read shadow CR0 */ @@ -99,7 +100,7 @@ void *io_bitmap_a, *io_bitmap_b; struct vlapic *vlapic; u64 tsc_offset; - struct ac_timer hlt_timer; /* hlt ins emulation wakeup timer */ + struct timer hlt_timer; /* hlt ins emulation wakeup timer */ }; #define vmx_schedule_tail(next) \ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-x86/vmx_vpit.h --- a/xen/include/asm-x86/vmx_vpit.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-x86/vmx_vpit.h Fri Jan 13 20:12:24 2006 @@ -6,7 +6,7 @@ #include <xen/lib.h> #include <xen/time.h> #include <xen/errno.h> -#include <xen/ac_timer.h> +#include <xen/timer.h> #include <asm/vmx_vmcs.h> #include <asm/vmx_vpic.h> @@ -23,7 +23,7 @@ u64 inject_point; /* the time inject virt intr */ u64 shift; /* save the value of offset - drift */ s_time_t scheduled; /* scheduled timer interrupt */ - struct ac_timer pit_timer; /* periodic timer for mode 2*/ + struct timer pit_timer; /* periodic timer for mode 2*/ unsigned int channel; /* the pit channel, counter 0~2 */ unsigned int pending_intr_nr; /* the couner for pending timer interrupts */ u32 period; /* pit frequency in ns */ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/asm-x86/x86_emulate.h --- a/xen/include/asm-x86/x86_emulate.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/asm-x86/x86_emulate.h Fri Jan 13 20:12:24 2006 @@ -18,10 +18,11 @@ * special treatment or emulation (*_emulated). * * The emulator assumes that an instruction accesses only one 'emulated memory' - * location, and that this is one of its data operands. Instruction fetches and + * location, that this location is the given linear faulting address (cr2), and + * that this is one of the instruction's data operands. Instruction fetches and * stack operations are assumed never to access emulated memory. The emulator * automatically deduces which operand of a string-move operation is accessing - * emulated memory, and requires that the other operand accesses normal memory. + * emulated memory, and assumes that the other operand accesses normal memory. * * NOTES: * 1. The emulator isn't very smart about emulated vs. standard memory. @@ -36,6 +37,7 @@ * then immediately bail. * 3. Valid access sizes are 1, 2, 4 and 8 bytes. On x86/32 systems only * cmpxchg8b_emulated need support 8-byte accesses. + * 4. The emulator cannot handle 64-bit mode emulation on an x86/32 system. */ /* Access completed successfully: continue emulation as normal. */ #define X86EMUL_CONTINUE 0 @@ -141,22 +143,27 @@ struct cpu_user_regs; -/* Current execution mode, passed to the emulator. */ -#define X86EMUL_MODE_REAL 0 -#define X86EMUL_MODE_PROT16 2 -#define X86EMUL_MODE_PROT32 4 -#define X86EMUL_MODE_PROT64 8 +/* Execution mode, passed to the emulator. */ +#define X86EMUL_MODE_REAL 0 /* Real mode. */ +#define X86EMUL_MODE_PROT16 2 /* 16-bit protected mode. */ +#define X86EMUL_MODE_PROT32 4 /* 32-bit protected mode. */ +#define X86EMUL_MODE_PROT64 8 /* 64-bit (long) mode. */ + +/* Host execution mode. */ +#if defined(__i386__) +#define X86EMUL_MODE_HOST X86EMUL_MODE_PROT32 +#elif defined(__x86_64__) +#define X86EMUL_MODE_HOST X86EMUL_MODE_PROT64 +#endif /* * x86_emulate_memop: Emulate an instruction that faulted attempting to * read/write a 'special' memory area. * @regs: Register state at time of fault. - * @cr2: Linear faulting address. + * @cr2: Linear faulting address within an emulated/special memory area. * @ops: Interface to access special memory. - * @mode: Current execution mode, represented by the default size of memory - * addresses, in bytes. Valid values are 2, 4 and 8 (x86/64 only). - * Alternatively use the appropriate X86EMUL_MODE value (which also - * includes a value for emulating real mode). + * @mode: Emulated execution mode, represented by an X86EMUL_MODE value. + * Returns -1 on failure, 0 on success. */ extern int x86_emulate_memop( diff -r 2884f14895d2 -r f5ab28885c75 xen/include/public/arch-ia64.h --- a/xen/include/public/arch-ia64.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/public/arch-ia64.h Fri Jan 13 20:12:24 2006 @@ -9,7 +9,7 @@ /* Maximum number of virtual CPUs in multi-processor guests. */ /* WARNING: before changing this, check that shared_info fits on a page */ -#define MAX_VIRT_CPUS 1 +#define MAX_VIRT_CPUS 4 #ifndef __ASSEMBLY__ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/public/arch-x86_32.h --- a/xen/include/public/arch-x86_32.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/public/arch-x86_32.h Fri Jan 13 20:12:24 2006 @@ -49,10 +49,15 @@ * machine->physical mapping table starts at this address, read-only. */ #ifdef CONFIG_X86_PAE -# define HYPERVISOR_VIRT_START (0xF5800000UL) +#define __HYPERVISOR_VIRT_START 0xF5800000 #else -# define HYPERVISOR_VIRT_START (0xFC000000UL) +#define __HYPERVISOR_VIRT_START 0xFC000000 #endif + +#ifndef HYPERVISOR_VIRT_START +#define HYPERVISOR_VIRT_START mk_unsigned_long(__HYPERVISOR_VIRT_START) +#endif + #ifndef machine_to_phys_mapping #define machine_to_phys_mapping ((unsigned long *)HYPERVISOR_VIRT_START) #endif @@ -130,6 +135,7 @@ unsigned long max_pfn; /* max pfn that appears in table */ /* Frame containing list of mfns containing list of mfns containing p2m. */ unsigned long pfn_to_mfn_frame_list_list; + unsigned long nmi_reason; } arch_shared_info_t; typedef struct { @@ -137,7 +143,7 @@ unsigned long pad[5]; /* sizeof(vcpu_info_t) == 64 */ } arch_vcpu_info_t; -#endif +#endif /* !__ASSEMBLY__ */ #endif diff -r 2884f14895d2 -r f5ab28885c75 xen/include/public/arch-x86_64.h --- a/xen/include/public/arch-x86_64.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/public/arch-x86_64.h Fri Jan 13 20:12:24 2006 @@ -59,9 +59,12 @@ /* And the trap vector is... */ #define TRAP_INSTR "syscall" +#define __HYPERVISOR_VIRT_START 0xFFFF800000000000 +#define __HYPERVISOR_VIRT_END 0xFFFF880000000000 + #ifndef HYPERVISOR_VIRT_START -#define HYPERVISOR_VIRT_START (0xFFFF800000000000UL) -#define HYPERVISOR_VIRT_END (0xFFFF880000000000UL) +#define HYPERVISOR_VIRT_START mk_unsigned_long(__HYPERVISOR_VIRT_START) +#define HYPERVISOR_VIRT_END mk_unsigned_long(__HYPERVISOR_VIRT_END) #endif /* Maximum number of virtual CPUs in multi-processor guests. */ @@ -85,11 +88,20 @@ #define SEGBASE_GS_USER_SEL 3 /* Set user %gs specified in base[15:0] */ /* - * int HYPERVISOR_switch_to_user(void) + * int HYPERVISOR_iret(void) * All arguments are on the kernel stack, in the following format. * Never returns if successful. Current kernel context is lost. + * The saved CS is mapped as follows: + * RING0 -> RING3 kernel mode. + * RING1 -> RING3 kernel mode. + * RING2 -> RING3 kernel mode. + * RING3 -> RING3 user mode. + * However RING0 indicates that the guest kernel should return to iteself + * directly with + * orb $3,1*8(%rsp) + * iretq * If flags contains VGCF_IN_SYSCALL: - * Restore RAX, RIP, RFLAGS, RSP. + * Restore RAX, RIP, RFLAGS, RSP. * Discard R11, RCX, CS, SS. * Otherwise: * Restore RAX, R11, RCX, CS:RIP, RFLAGS, SS:RSP. @@ -97,10 +109,19 @@ */ /* Guest exited in SYSCALL context? Return to guest with SYSRET? */ #define VGCF_IN_SYSCALL (1<<8) +struct iret_context { + /* Top of stack (%rsp at point of hypercall). */ + uint64_t rax, r11, rcx, flags, rip, cs, rflags, rsp, ss; + /* Bottom of iret stack frame. */ +}; +/* + * For compatibility with HYPERVISOR_switch_to_user which is the old + * name for HYPERVISOR_iret. + */ struct switch_to_user { /* Top of stack (%rsp at point of hypercall). */ uint64_t rax, r11, rcx, flags, rip, cs, rflags, rsp, ss; - /* Bottom of switch_to_user stack frame. */ + /* Bottom of iret stack frame. */ }; /* @@ -199,6 +220,7 @@ unsigned long max_pfn; /* max pfn that appears in table */ /* Frame containing list of mfns containing list of mfns containing p2m. */ unsigned long pfn_to_mfn_frame_list_list; + unsigned long nmi_reason; } arch_shared_info_t; typedef struct { diff -r 2884f14895d2 -r f5ab28885c75 xen/include/public/dom0_ops.h --- a/xen/include/public/dom0_ops.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/public/dom0_ops.h Fri Jan 13 20:12:24 2006 @@ -94,14 +94,14 @@ xen_domain_handle_t handle; } dom0_getdomaininfo_t; -#define DOM0_SETDOMAININFO 13 +#define DOM0_SETVCPUCONTEXT 13 typedef struct { /* IN variables. */ domid_t domain; uint32_t vcpu; /* IN/OUT parameters */ vcpu_guest_context_t *ctxt; -} dom0_setdomaininfo_t; +} dom0_setvcpucontext_t; #define DOM0_MSR 15 typedef struct { @@ -163,13 +163,13 @@ /* * Set which physical cpus a vcpu can execute on. */ -#define DOM0_PINCPUDOMAIN 20 +#define DOM0_SETVCPUAFFINITY 20 typedef struct { /* IN variables. */ domid_t domain; uint32_t vcpu; cpumap_t cpumap; -} dom0_pincpudomain_t; +} dom0_setvcpuaffinity_t; /* Get trace buffers machine base address */ #define DOM0_TBUFCONTROL 21 @@ -436,13 +436,13 @@ dom0_getmemlist_t getmemlist; dom0_schedctl_t schedctl; dom0_adjustdom_t adjustdom; - dom0_setdomaininfo_t setdomaininfo; + dom0_setvcpucontext_t setvcpucontext; dom0_getdomaininfo_t getdomaininfo; dom0_getpageframeinfo_t getpageframeinfo; dom0_msr_t msr; dom0_settime_t settime; dom0_readconsole_t readconsole; - dom0_pincpudomain_t pincpudomain; + dom0_setvcpuaffinity_t setvcpuaffinity; dom0_tbufcontrol_t tbufcontrol; dom0_physinfo_t physinfo; dom0_sched_id_t sched_id; diff -r 2884f14895d2 -r f5ab28885c75 xen/include/public/xen.h --- a/xen/include/public/xen.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/public/xen.h Fri Jan 13 20:12:24 2006 @@ -53,12 +53,14 @@ #define __HYPERVISOR_grant_table_op 20 #define __HYPERVISOR_vm_assist 21 #define __HYPERVISOR_update_va_mapping_otherdomain 22 -#define __HYPERVISOR_switch_vm86 23 /* x86/32 only */ -#define __HYPERVISOR_switch_to_user 23 /* x86/64 only */ +#define __HYPERVISOR_iret 23 /* x86 only */ +#define __HYPERVISOR_switch_vm86 23 /* x86/32 only (obsolete name) */ +#define __HYPERVISOR_switch_to_user 23 /* x86/64 only (obsolete name) */ #define __HYPERVISOR_vcpu_op 24 #define __HYPERVISOR_set_segment_base 25 /* x86/64 only */ #define __HYPERVISOR_mmuext_op 26 #define __HYPERVISOR_acm_op 27 +#define __HYPERVISOR_nmi_op 28 /* * VIRTUAL INTERRUPTS @@ -69,10 +71,7 @@ #define VIRQ_DEBUG 1 /* Request guest to dump debug info. */ #define VIRQ_CONSOLE 2 /* (DOM0) Bytes received on emergency console. */ #define VIRQ_DOM_EXC 3 /* (DOM0) Exceptional event for some domain. */ -#define VIRQ_PARITY_ERR 4 /* (DOM0) NMI parity error (port 0x61, bit 7). */ -#define VIRQ_IO_ERR 5 /* (DOM0) NMI I/O error (port 0x61, bit 6). */ #define VIRQ_DEBUGGER 6 /* (DOM0) A domain has paused for debugging. */ -#define VIRQ_NMI 7 /* (DOM0) Unknown NMI (not from ISA port 0x61).*/ #define NR_VIRQS 8 /* @@ -426,6 +425,15 @@ typedef uint8_t xen_domain_handle_t[16]; +/* Turn a plain number into a C unsigned long constant. */ +#define __mk_unsigned_long(x) x ## UL +#define mk_unsigned_long(x) __mk_unsigned_long(x) + +#else /* __ASSEMBLY__ */ + +/* In assembly code we cannot use C numeric constant suffixes. */ +#define mk_unsigned_long(x) x + #endif /* !__ASSEMBLY__ */ #endif /* __XEN_PUBLIC_XEN_H__ */ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/xen/bitmap.h --- a/xen/include/xen/bitmap.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/xen/bitmap.h Fri Jan 13 20:12:24 2006 @@ -41,6 +41,8 @@ * bitmap_weight(src, nbits) Hamming Weight: number set bits * bitmap_shift_right(dst, src, n, nbits) *dst = *src >> n * bitmap_shift_left(dst, src, n, nbits) *dst = *src << n + * bitmap_scnprintf(buf, len, src, nbits) Print bitmap src to buf + * bitmap_scnlistprintf(buf, len, src, nbits) Print bitmap src as list to buf */ /* @@ -93,6 +95,10 @@ const unsigned long *bitmap2, int bits); extern int __bitmap_weight(const unsigned long *bitmap, int bits); +extern int bitmap_scnprintf(char *buf, unsigned int len, + const unsigned long *src, int nbits); +extern int bitmap_scnlistprintf(char *buf, unsigned int len, + const unsigned long *src, int nbits); extern int bitmap_find_free_region(unsigned long *bitmap, int bits, int order); extern void bitmap_release_region(unsigned long *bitmap, int pos, int order); extern int bitmap_allocate_region(unsigned long *bitmap, int pos, int order); diff -r 2884f14895d2 -r f5ab28885c75 xen/include/xen/config.h --- a/xen/include/xen/config.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/xen/config.h Fri Jan 13 20:12:24 2006 @@ -43,4 +43,13 @@ #define __STR(...) #__VA_ARGS__ #define STR(...) __STR(__VA_ARGS__) +#ifndef __ASSEMBLY__ +/* Turn a plain number into a C unsigned long constant. */ +#define __mk_unsigned_long(x) x ## UL +#define mk_unsigned_long(x) __mk_unsigned_long(x) +#else /* __ASSEMBLY__ */ +/* In assembly code we cannot use C numeric constant suffixes. */ +#define mk_unsigned_long(x) x +#endif /* !__ASSEMBLY__ */ + #endif /* __XEN_CONFIG_H__ */ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/xen/cpumask.h --- a/xen/include/xen/cpumask.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/xen/cpumask.h Fri Jan 13 20:12:24 2006 @@ -8,8 +8,8 @@ * See detailed comments in the file xen/bitmap.h describing the * data type on which these cpumasks are based. * - * For details of cpumask_scnprintf() and cpumask_parse(), - * see bitmap_scnprintf() and bitmap_parse() in lib/bitmap.c. + * For details of cpumask_scnprintf() and cpulist_scnprintf(), + * see bitmap_scnprintf() and bitmap_scnlistprintf() in lib/bitmap.c. * * The available cpumask operations are: * @@ -36,8 +36,8 @@ * void cpus_shift_right(dst, src, n) Shift right * void cpus_shift_left(dst, src, n) Shift left * - * int first_cpu(mask) Number lowest set bit, or >= NR_CPUS - * int next_cpu(cpu, mask) Next cpu past 'cpu', or >= NR_CPUS + * int first_cpu(mask) Number lowest set bit, or NR_CPUS + * int next_cpu(cpu, mask) Next cpu past 'cpu', or NR_CPUS * * cpumask_t cpumask_of_cpu(cpu) Return cpumask with bit 'cpu' set * CPU_MASK_ALL Initializer - all bits set @@ -45,7 +45,7 @@ * unsigned long *cpus_addr(mask) Array of unsigned long's in mask * * int cpumask_scnprintf(buf, len, mask) Format cpumask for printing - * int cpumask_parse(ubuf, ulen, mask) Parse ascii string as cpumask + * int cpulist_scnprintf(buf, len, mask) Format cpumask as list for printing * * for_each_cpu_mask(cpu, mask) for-loop cpu over mask * @@ -207,13 +207,13 @@ #define first_cpu(src) __first_cpu(&(src), NR_CPUS) static inline int __first_cpu(const cpumask_t *srcp, int nbits) { - return find_first_bit(srcp->bits, nbits); + return min_t(int, nbits, find_first_bit(srcp->bits, nbits)); } #define next_cpu(n, src) __next_cpu((n), &(src), NR_CPUS) static inline int __next_cpu(int n, const cpumask_t *srcp, int nbits) { - return find_next_bit(srcp->bits, nbits, n+1); + return min_t(int, nbits, find_next_bit(srcp->bits, nbits, n+1)); } #define cpumask_of_cpu(cpu) \ @@ -259,7 +259,6 @@ #define cpus_addr(src) ((src).bits) -/* #define cpumask_scnprintf(buf, len, src) \ __cpumask_scnprintf((buf), (len), &(src), NR_CPUS) static inline int __cpumask_scnprintf(char *buf, int len, @@ -268,14 +267,13 @@ return bitmap_scnprintf(buf, len, srcp->bits, nbits); } -#define cpumask_parse(ubuf, ulen, src) \ - __cpumask_parse((ubuf), (ulen), &(src), NR_CPUS) -static inline int __cpumask_parse(const char __user *buf, int len, - cpumask_t *dstp, int nbits) -{ - return bitmap_parse(buf, len, dstp->bits, nbits); -} -*/ +#define cpulist_scnprintf(buf, len, src) \ + __cpulist_scnprintf((buf), (len), &(src), NR_CPUS) +static inline int __cpulist_scnprintf(char *buf, int len, + const cpumask_t *srcp, int nbits) +{ + return bitmap_scnlistprintf(buf, len, srcp->bits, nbits); +} #if NR_CPUS > 1 #define for_each_cpu_mask(cpu, mask) \ @@ -368,7 +366,7 @@ for_each_cpu_mask(cpu, (mask)) \ if (cpu_online(cpu)) \ break; \ - min_t(int, NR_CPUS, cpu); \ + cpu; \ }) #define for_each_cpu(cpu) for_each_cpu_mask((cpu), cpu_possible_map) diff -r 2884f14895d2 -r f5ab28885c75 xen/include/xen/domain.h --- a/xen/include/xen/domain.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/xen/domain.h Fri Jan 13 20:12:24 2006 @@ -18,8 +18,6 @@ extern int arch_set_info_guest( struct vcpu *v, struct vcpu_guest_context *c); -extern void vcpu_migrate_cpu(struct vcpu *v, int newcpu); - extern void free_perdomain_pt(struct domain *d); extern void domain_relinquish_resources(struct domain *d); diff -r 2884f14895d2 -r f5ab28885c75 xen/include/xen/domain_page.h --- a/xen/include/xen/domain_page.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/xen/domain_page.h Fri Jan 13 20:12:24 2006 @@ -2,6 +2,8 @@ * domain_page.h * * Allow temporary mapping of domain page frames into Xen space. + * + * Copyright (c) 2003-2006, Keir Fraser <keir@xxxxxxxxxxxxx> */ #ifndef __XEN_DOMAIN_PAGE_H__ @@ -10,22 +12,27 @@ #include <xen/config.h> #include <xen/mm.h> -#define map_domain_page(pfn) map_domain_pages(pfn,0) -#define unmap_domain_page(va) unmap_domain_pages(va,0) - #ifdef CONFIG_DOMAIN_PAGE /* - * Maps a given range of page frames, returning the mapped virtual address. The - * pages are now accessible until a corresponding call to unmap_domain_page(). + * Map a given page frame, returning the mapped virtual address. The page is + * then accessible within the current VCPU until a corresponding unmap call. */ -extern void *map_domain_pages(unsigned long pfn, unsigned int order); +extern void *map_domain_page(unsigned long pfn); /* - * Pass a VA within the first page of a range previously mapped with - * map_omain_pages(). Those pages will then be removed from the mapping lists. + * Pass a VA within a page previously mapped in the context of the + * currently-executing VCPU via a call to map_domain_pages(). */ -extern void unmap_domain_pages(void *va, unsigned int order); +extern void unmap_domain_page(void *va); + +/* + * Similar to the above calls, except the mapping is accessible in all + * address spaces (not just within the VCPU that created the mapping). Global + * mappings can also be unmapped from any context. + */ +extern void *map_domain_page_global(unsigned long pfn); +extern void unmap_domain_page_global(void *va); #define DMCACHE_ENTRY_VALID 1U #define DMCACHE_ENTRY_HELD 2U @@ -87,8 +94,11 @@ #else /* !CONFIG_DOMAIN_PAGE */ -#define map_domain_pages(pfn,order) phys_to_virt((pfn)<<PAGE_SHIFT) -#define unmap_domain_pages(va,order) ((void)((void)(va),(void)(order))) +#define map_domain_page(pfn) phys_to_virt((pfn)<<PAGE_SHIFT) +#define unmap_domain_page(va) ((void)(va)) + +#define map_domain_page_global(pfn) phys_to_virt((pfn)<<PAGE_SHIFT) +#define unmap_domain_page_global(va) ((void)(va)) struct domain_mmap_cache { }; diff -r 2884f14895d2 -r f5ab28885c75 xen/include/xen/lib.h --- a/xen/include/xen/lib.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/xen/lib.h Fri Jan 13 20:12:24 2006 @@ -53,10 +53,16 @@ /* vsprintf.c */ extern int sprintf(char * buf, const char * fmt, ...) __attribute__ ((format (printf, 2, 3))); -extern int vsprintf(char *buf, const char *, va_list); +extern int vsprintf(char *buf, const char *, va_list) + __attribute__ ((format (printf, 2, 0))); extern int snprintf(char * buf, size_t size, const char * fmt, ...) __attribute__ ((format (printf, 3, 4))); -extern int vsnprintf(char *buf, size_t size, const char *fmt, va_list args); +extern int vsnprintf(char *buf, size_t size, const char *fmt, va_list args) + __attribute__ ((format (printf, 3, 0))); +extern int scnprintf(char * buf, size_t size, const char * fmt, ...) + __attribute__ ((format (printf, 3, 4))); +extern int vscnprintf(char *buf, size_t size, const char *fmt, va_list args) + __attribute__ ((format (printf, 3, 0))); long simple_strtol( const char *cp,char **endp, unsigned int base); diff -r 2884f14895d2 -r f5ab28885c75 xen/include/xen/perfc_defn.h --- a/xen/include/xen/perfc_defn.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/xen/perfc_defn.h Fri Jan 13 20:12:24 2006 @@ -32,7 +32,7 @@ PERFCOUNTER_CPU(irq_time, "cycles spent in irq handler") PERFCOUNTER_CPU(apic_timer, "apic timer interrupts") -PERFCOUNTER_CPU(ac_timer_max, "ac_timer max error (ns)") +PERFCOUNTER_CPU(timer_max, "timer max error (ns)") PERFCOUNTER_CPU(sched_irq, "sched: timer") PERFCOUNTER_CPU(sched_run, "sched: runs through scheduler") PERFCOUNTER_CPU(sched_ctx, "sched: context switches") diff -r 2884f14895d2 -r f5ab28885c75 xen/include/xen/sched-if.h --- a/xen/include/xen/sched-if.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/xen/sched-if.h Fri Jan 13 20:12:24 2006 @@ -13,19 +13,50 @@ struct schedule_data { spinlock_t schedule_lock; /* spinlock protecting curr */ - struct vcpu *curr; /* current task */ - struct vcpu *idle; /* idle task for this cpu */ + struct vcpu *curr; /* current task */ + struct vcpu *idle; /* idle task for this cpu */ void *sched_priv; - struct ac_timer s_timer; /* scheduling timer */ + struct timer s_timer; /* scheduling timer */ unsigned long tick; /* current periodic 'tick' */ #ifdef BUCKETS u32 hist[BUCKETS]; /* for scheduler latency histogram */ #endif } __cacheline_aligned; +extern struct schedule_data schedule_data[]; + +static inline void vcpu_schedule_lock(struct vcpu *v) +{ + unsigned int cpu; + + for ( ; ; ) + { + cpu = v->processor; + spin_lock(&schedule_data[cpu].schedule_lock); + if ( likely(v->processor == cpu) ) + break; + spin_unlock(&schedule_data[cpu].schedule_lock); + } +} + +#define vcpu_schedule_lock_irq(v) \ + do { local_irq_disable(); vcpu_schedule_lock(v); } while ( 0 ) +#define vcpu_schedule_lock_irqsave(v, flags) \ + do { local_irq_save(flags); vcpu_schedule_lock(v); } while ( 0 ) + +static inline void vcpu_schedule_unlock(struct vcpu *v) +{ + spin_unlock(&schedule_data[v->processor].schedule_lock); +} + +#define vcpu_schedule_unlock_irq(v) \ + do { vcpu_schedule_unlock(v); local_irq_enable(); } while ( 0 ) +#define vcpu_schedule_unlock_irqrestore(v, flags) \ + do { vcpu_schedule_unlock(v); local_irq_restore(flags); } while ( 0 ) + struct task_slice { struct vcpu *task; - s_time_t time; + s_time_t time; }; struct scheduler { @@ -39,6 +70,7 @@ void (*rem_task) (struct vcpu *); void (*sleep) (struct vcpu *); void (*wake) (struct vcpu *); + int (*set_affinity) (struct vcpu *, cpumask_t *); struct task_slice (*do_schedule) (s_time_t); int (*control) (struct sched_ctl_cmd *); int (*adjdom) (struct domain *, @@ -47,6 +79,4 @@ void (*dump_cpu_state) (int); }; -extern struct schedule_data schedule_data[]; - #endif /* __XEN_SCHED_IF_H__ */ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/xen/sched.h --- a/xen/include/xen/sched.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/xen/sched.h Fri Jan 13 20:12:24 2006 @@ -9,7 +9,7 @@ #include <public/xen.h> #include <public/dom0_ops.h> #include <xen/time.h> -#include <xen/ac_timer.h> +#include <xen/timer.h> #include <xen/grant_table.h> #include <xen/rangeset.h> #include <asm/domain.h> @@ -51,8 +51,6 @@ int evtchn_init(struct domain *d); void evtchn_destroy(struct domain *d); -#define CPUMAP_RUNANYWHERE 0xFFFFFFFF - struct vcpu { int vcpu_id; @@ -65,7 +63,7 @@ struct vcpu *next_in_list; - struct ac_timer timer; /* one-shot timer for timeout values */ + struct timer timer; /* one-shot timer for timeout values */ unsigned long sleep_tick; /* tick at which this vcpu started sleep */ s_time_t lastschd; /* time this domain was last scheduled */ @@ -80,7 +78,13 @@ atomic_t pausecnt; - cpumap_t cpumap; /* which cpus this domain can run on */ + /* Bitmask of CPUs on which this VCPU may run. */ + cpumask_t cpu_affinity; + + unsigned long nmi_addr; /* NMI callback address. */ + + /* Bitmask of CPUs which are holding onto this VCPU's state. */ + cpumask_t vcpu_dirty_cpumask; struct arch_vcpu arch; }; @@ -141,7 +145,7 @@ struct vcpu *vcpu[MAX_VIRT_CPUS]; /* Bitmask of CPUs which are holding onto this domain's state. */ - cpumask_t cpumask; + cpumask_t domain_dirty_cpumask; struct arch_domain arch; @@ -170,12 +174,10 @@ char *xen_section_string; }; -extern struct domain idle0_domain; -extern struct vcpu idle0_vcpu; - -extern struct vcpu *idle_task[NR_CPUS]; +extern struct vcpu *idle_vcpu[NR_CPUS]; #define IDLE_DOMAIN_ID (0x7FFFU) -#define is_idle_task(_d) (test_bit(_DOMF_idle_domain, &(_d)->domain_flags)) +#define is_idle_domain(d) ((d)->domain_id == IDLE_DOMAIN_ID) +#define is_idle_vcpu(v) (is_idle_domain((v)->domain)) struct vcpu *alloc_vcpu( struct domain *d, unsigned int vcpu_id, unsigned int cpu_id); @@ -223,7 +225,7 @@ unsigned long image_start, unsigned long image_len, unsigned long initrd_start, unsigned long initrd_len, char *cmdline); -extern int set_info_guest(struct domain *d, dom0_setdomaininfo_t *); +extern int set_info_guest(struct domain *d, dom0_setvcpucontext_t *); struct domain *find_domain_by_id(domid_t dom); extern void domain_destruct(struct domain *d); @@ -269,36 +271,27 @@ extern void sync_vcpu_execstate(struct vcpu *v); /* - * Called by the scheduler to switch to another VCPU. On entry, although - * VCPUF_running is no longer asserted for @prev, its context is still running - * on the local CPU and is not committed to memory. The local scheduler lock - * is therefore still held, and interrupts are disabled, because the local CPU - * is in an inconsistent state. - * - * The callee must ensure that the local CPU is no longer running in @prev's - * context, and that the context is saved to memory, before returning. - * Alternatively, if implementing lazy context switching, it suffices to ensure - * that invoking sync_vcpu_execstate() will switch and commit @prev's state. + * Called by the scheduler to switch to another VCPU. This function must + * call context_saved(@prev) when the local CPU is no longer running in + * @prev's context, and that context is saved to memory. Alternatively, if + * implementing lazy context switching, it suffices to ensure that invoking + * sync_vcpu_execstate() will switch and commit @prev's state. */ extern void context_switch( struct vcpu *prev, struct vcpu *next); /* - * On some architectures (notably x86) it is not possible to entirely load - * @next's context with interrupts disabled. These may implement a function to - * finalise loading the new context after interrupts are re-enabled. This - * function is not given @prev and is not permitted to access it. - */ -extern void context_switch_finalise( - struct vcpu *next); + * As described above, context_switch() must call this function when the + * local CPU is no longer running in @prev's context, and @prev's context is + * saved to memory. Alternatively, if implementing lazy context switching, + * ensure that invoking sync_vcpu_execstate() will switch and commit @prev. + */ +#define context_saved(prev) (clear_bit(_VCPUF_running, &(prev)->vcpu_flags)) /* Called by the scheduler to continue running the current VCPU. */ extern void continue_running( struct vcpu *same); - -/* Is CPU 'cpu' idle right now? */ -int idle_cpu(int cpu); void startup_cpu_idle_loop(void); @@ -364,51 +357,44 @@ /* Currently running on a CPU? */ #define _VCPUF_running 3 #define VCPUF_running (1UL<<_VCPUF_running) - /* Disables auto-migration between CPUs. */ -#define _VCPUF_cpu_pinned 4 -#define VCPUF_cpu_pinned (1UL<<_VCPUF_cpu_pinned) - /* Domain migrated between CPUs. */ -#define _VCPUF_cpu_migrated 5 -#define VCPUF_cpu_migrated (1UL<<_VCPUF_cpu_migrated) /* Initialization completed. */ -#define _VCPUF_initialised 6 +#define _VCPUF_initialised 4 #define VCPUF_initialised (1UL<<_VCPUF_initialised) /* VCPU is not-runnable */ -#define _VCPUF_down 7 +#define _VCPUF_down 5 #define VCPUF_down (1UL<<_VCPUF_down) + /* NMI callback pending for this VCPU? */ +#define _VCPUF_nmi_pending 8 +#define VCPUF_nmi_pending (1UL<<_VCPUF_nmi_pending) + /* Avoid NMI reentry by allowing NMIs to be masked for short periods. */ +#define _VCPUF_nmi_masked 9 +#define VCPUF_nmi_masked (1UL<<_VCPUF_nmi_masked) /* * Per-domain flags (domain_flags). */ - /* Is this one of the per-CPU idle domains? */ -#define _DOMF_idle_domain 0 -#define DOMF_idle_domain (1UL<<_DOMF_idle_domain) /* Is this domain privileged? */ -#define _DOMF_privileged 1 +#define _DOMF_privileged 0 #define DOMF_privileged (1UL<<_DOMF_privileged) /* Guest shut itself down for some reason. */ -#define _DOMF_shutdown 2 +#define _DOMF_shutdown 1 #define DOMF_shutdown (1UL<<_DOMF_shutdown) - /* Guest is in process of shutting itself down (becomes DOMF_shutdown). */ -#define _DOMF_shuttingdown 3 -#define DOMF_shuttingdown (1UL<<_DOMF_shuttingdown) /* Death rattle. */ -#define _DOMF_dying 4 +#define _DOMF_dying 2 #define DOMF_dying (1UL<<_DOMF_dying) /* Domain is paused by controller software. */ -#define _DOMF_ctrl_pause 5 +#define _DOMF_ctrl_pause 3 #define DOMF_ctrl_pause (1UL<<_DOMF_ctrl_pause) /* Domain is being debugged by controller software. */ -#define _DOMF_debugging 6 +#define _DOMF_debugging 4 #define DOMF_debugging (1UL<<_DOMF_debugging) -static inline int domain_runnable(struct vcpu *v) +static inline int vcpu_runnable(struct vcpu *v) { return ( (atomic_read(&v->pausecnt) == 0) && !(v->vcpu_flags & (VCPUF_blocked|VCPUF_down)) && - !(v->domain->domain_flags & - (DOMF_shutdown|DOMF_shuttingdown|DOMF_ctrl_pause)) ); + !(v->domain->domain_flags & (DOMF_shutdown|DOMF_ctrl_pause)) ); } void vcpu_pause(struct vcpu *v); @@ -419,6 +405,8 @@ void domain_unpause_by_systemcontroller(struct domain *d); void cpu_init(void); +int vcpu_set_affinity(struct vcpu *v, cpumask_t *affinity); + static inline void vcpu_unblock(struct vcpu *v) { if ( test_and_clear_bit(_VCPUF_blocked, &v->vcpu_flags) ) diff -r 2884f14895d2 -r f5ab28885c75 xen/include/xen/softirq.h --- a/xen/include/xen/softirq.h Fri Jan 13 16:38:44 2006 +++ b/xen/include/xen/softirq.h Fri Jan 13 20:12:24 2006 @@ -2,11 +2,11 @@ #define __XEN_SOFTIRQ_H__ /* Common softirqs come first in the following list. */ -#define AC_TIMER_SOFTIRQ 0 +#define TIMER_SOFTIRQ 0 #define SCHEDULE_SOFTIRQ 1 #define NEW_TLBFLUSH_CLOCK_PERIOD_SOFTIRQ 2 #define KEYPRESS_SOFTIRQ 3 -#define NMI_DOM0_SOFTIRQ 4 +#define NMI_SOFTIRQ 4 #define PAGE_SCRUB_SOFTIRQ 5 #define DOMAIN_SHUTDOWN_FINALISE_SOFTIRQ 6 #define NR_SOFTIRQS 7 diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/include/asm-xen/asm-i386/mach-xen/mach_traps.h --- /dev/null Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/include/asm-xen/asm-i386/mach-xen/mach_traps.h Fri Jan 13 20:12:24 2006 @@ -0,0 +1,33 @@ +/* + * include/asm-xen/asm-i386/mach-xen/mach_traps.h + * + * Machine specific NMI handling for Xen + */ +#ifndef _MACH_TRAPS_H +#define _MACH_TRAPS_H + +#include <linux/bitops.h> +#include <asm-xen/xen-public/nmi.h> + +static inline void clear_mem_error(unsigned char reason) {} +static inline void clear_io_check_error(unsigned char reason) {} + +static inline unsigned char get_nmi_reason(void) +{ + shared_info_t *s = HYPERVISOR_shared_info; + unsigned char reason = 0; + + /* construct a value which looks like it came from + * port 0x61. + */ + if (test_bit(_XEN_NMIREASON_io_error, &s->arch.nmi_reason)) + reason |= 0x40; + if (test_bit(_XEN_NMIREASON_parity_error, &s->arch.nmi_reason)) + reason |= 0x80; + + return reason; +} + +static inline void reassert_nmi(void) {} + +#endif /* !_MACH_TRAPS_H */ diff -r 2884f14895d2 -r f5ab28885c75 linux-2.6-xen-sparse/include/asm-xen/asm-x86_64/nmi.h --- /dev/null Fri Jan 13 16:38:44 2006 +++ b/linux-2.6-xen-sparse/include/asm-xen/asm-x86_64/nmi.h Fri Jan 13 20:12:24 2006 @@ -0,0 +1,75 @@ +/* + * linux/include/asm-i386/nmi.h + */ +#ifndef ASM_NMI_H +#define ASM_NMI_H + +#include <linux/pm.h> + +#include <asm-xen/xen-public/nmi.h> + +struct pt_regs; + +typedef int (*nmi_callback_t)(struct pt_regs * regs, int cpu); + +/** + * set_nmi_callback + * + * Set a handler for an NMI. Only one handler may be + * set. Return 1 if the NMI was handled. + */ +void set_nmi_callback(nmi_callback_t callback); + +/** + * unset_nmi_callback + * + * Remove the handler previously set. + */ +void unset_nmi_callback(void); + +#ifdef CONFIG_PM + +/** Replace the PM callback routine for NMI. */ +struct pm_dev * set_nmi_pm_callback(pm_callback callback); + +/** Unset the PM callback routine back to the default. */ +void unset_nmi_pm_callback(struct pm_dev * dev); + +#else + +static inline struct pm_dev * set_nmi_pm_callback(pm_callback callback) +{ + return 0; +} + +static inline void unset_nmi_pm_callback(struct pm_dev * dev) +{ +} + +#endif /* CONFIG_PM */ + +extern void default_do_nmi(struct pt_regs *); +extern void die_nmi(char *str, struct pt_regs *regs); + +static inline unsigned char get_nmi_reason(void) +{ + shared_info_t *s = HYPERVISOR_shared_info; + unsigned char reason = 0; + + /* construct a value which looks like it came from + * port 0x61. + */ + if (test_bit(_XEN_NMIREASON_io_error, &s->arch.nmi_reason)) + reason |= 0x40; + if (test_bit(_XEN_NMIREASON_parity_error, &s->arch.nmi_reason)) + reason |= 0x80; + + return reason; +} + +extern int panic_on_timeout; +extern int unknown_nmi_panic; + +extern int check_nmi_watchdog(void); + +#endif /* ASM_NMI_H */ diff -r 2884f14895d2 -r f5ab28885c75 patches/linux-2.6.12/i386-mach-io-check-nmi.patch --- /dev/null Fri Jan 13 16:38:44 2006 +++ b/patches/linux-2.6.12/i386-mach-io-check-nmi.patch Fri Jan 13 20:12:24 2006 @@ -0,0 +1,43 @@ +--- ref-linux-2.6.12/arch/i386/kernel/traps.c 2005-12-19 09:23:44.000000000 +0000 ++++ linux-2.6.12-xen0/arch/i386/kernel/traps.c 2006-01-05 15:51:52.000000000 +0000 +@@ -521,18 +521,11 @@ + + static void io_check_error(unsigned char reason, struct pt_regs * regs) + { +- unsigned long i; +- + printk("NMI: IOCK error (debug interrupt?)\n"); + show_registers(regs); + + /* Re-enable the IOCK line, wait for a few seconds */ +- reason = (reason & 0xf) | 8; +- outb(reason, 0x61); +- i = 2000; +- while (--i) udelay(1000); +- reason &= ~8; +- outb(reason, 0x61); ++ clear_io_check_error(reason); + } + + static void unknown_nmi_error(unsigned char reason, struct pt_regs * regs) +--- ref-linux-2.6.12/include/asm-i386/mach-default/mach_traps.h 2005-06-17 20:48:29.000000000 +0100 ++++ linux-2.6.12-xen0/include/asm-i386/mach-default/mach_traps.h 2006-01-05 15:52:33.000000000 +0000 +@@ -15,6 +15,18 @@ + outb(reason, 0x61); + } + ++static inline void clear_io_check_error(unsigned char reason) ++{ ++ unsigned long i; ++ ++ reason = (reason & 0xf) | 8; ++ outb(reason, 0x61); ++ i = 2000; ++ while (--i) udelay(1000); ++ reason &= ~8; ++ outb(reason, 0x61); ++} ++ + static inline unsigned char get_nmi_reason(void) + { + return inb(0x61); diff -r 2884f14895d2 -r f5ab28885c75 tools/guest-headers/Makefile --- /dev/null Fri Jan 13 16:38:44 2006 +++ b/tools/guest-headers/Makefile Fri Jan 13 20:12:24 2006 @@ -0,0 +1,13 @@ + +XEN_ROOT=../.. +linuxsparsetree = $(XEN_ROOT)/linux-2.6-xen-sparse + +all: + +check: + +install: + mkdir -p $(DESTDIR)/usr/include/xen/linux + install -m0644 $(linuxsparsetree)/include/asm-xen/linux-public/*.h $(DESTDIR)/usr/include/xen/linux + +clean: diff -r 2884f14895d2 -r f5ab28885c75 xen/common/timer.c --- /dev/null Fri Jan 13 16:38:44 2006 +++ b/xen/common/timer.c Fri Jan 13 20:12:24 2006 @@ -0,0 +1,308 @@ +/****************************************************************************** + * timer.c + * + * Copyright (c) 2002-2003 Rolf Neugebauer + * Copyright (c) 2002-2005 K A Fraser + */ + +#include <xen/config.h> +#include <xen/init.h> +#include <xen/types.h> +#include <xen/errno.h> +#include <xen/sched.h> +#include <xen/lib.h> +#include <xen/smp.h> +#include <xen/perfc.h> +#include <xen/time.h> +#include <xen/softirq.h> +#include <xen/timer.h> +#include <xen/keyhandler.h> +#include <asm/system.h> +#include <asm/desc.h> + +/* + * We pull handlers off the timer list this far in future, + * rather than reprogramming the time hardware. + */ +#define TIMER_SLOP (50*1000) /* ns */ + +struct timers { + spinlock_t lock; + struct timer **heap; + struct timer *running; +} __cacheline_aligned; + +struct timers timers[NR_CPUS]; + +extern int reprogram_timer(s_time_t timeout); + +/**************************************************************************** + * HEAP OPERATIONS. + */ + +#define GET_HEAP_SIZE(_h) ((int)(((u16 *)(_h))[0])) +#define SET_HEAP_SIZE(_h,_v) (((u16 *)(_h))[0] = (u16)(_v)) + +#define GET_HEAP_LIMIT(_h) ((int)(((u16 *)(_h))[1])) +#define SET_HEAP_LIMIT(_h,_v) (((u16 *)(_h))[1] = (u16)(_v)) + +/* Sink down element @pos of @heap. */ +static void down_heap(struct timer **heap, int pos) +{ + int sz = GET_HEAP_SIZE(heap), nxt; + struct timer *t = heap[pos]; + + while ( (nxt = (pos << 1)) <= sz ) + { + if ( ((nxt+1) <= sz) && (heap[nxt+1]->expires < heap[nxt]->expires) ) + nxt++; + if ( heap[nxt]->expires > t->expires ) + break; + heap[pos] = heap[nxt]; + heap[pos]->heap_offset = pos; + pos = nxt; + } + + heap[pos] = t; + t->heap_offset = pos; +} + +/* Float element @pos up @heap. */ +static void up_heap(struct timer **heap, int pos) +{ + struct timer *t = heap[pos]; + + while ( (pos > 1) && (t->expires < heap[pos>>1]->expires) ) + { + heap[pos] = heap[pos>>1]; + heap[pos]->heap_offset = pos; + pos >>= 1; + } + + heap[pos] = t; + t->heap_offset = pos; +} + + +/* Delete @t from @heap. Return TRUE if new top of heap. */ +static int remove_entry(struct timer **heap, struct timer *t) +{ + int sz = GET_HEAP_SIZE(heap); + int pos = t->heap_offset; + + t->heap_offset = 0; + + if ( unlikely(pos == sz) ) + { + SET_HEAP_SIZE(heap, sz-1); + goto out; + } + + heap[pos] = heap[sz]; + heap[pos]->heap_offset = pos; + + SET_HEAP_SIZE(heap, --sz); + + if ( (pos > 1) && (heap[pos]->expires < heap[pos>>1]->expires) ) + up_heap(heap, pos); + else + down_heap(heap, pos); + + out: + return (pos == 1); +} + + +/* Add new entry @t to @heap. Return TRUE if new top of heap. */ +static int add_entry(struct timer ***pheap, struct timer *t) +{ + struct timer **heap = *pheap; + int sz = GET_HEAP_SIZE(heap); + + /* Copy the heap if it is full. */ + if ( unlikely(sz == GET_HEAP_LIMIT(heap)) ) + { + /* old_limit == (2^n)-1; new_limit == (2^(n+4))-1 */ + int old_limit = GET_HEAP_LIMIT(heap); + int new_limit = ((old_limit + 1) << 4) - 1; + heap = xmalloc_array(struct timer *, new_limit + 1); + BUG_ON(heap == NULL); + memcpy(heap, *pheap, (old_limit + 1) * sizeof(*heap)); + SET_HEAP_LIMIT(heap, new_limit); + if ( old_limit != 0 ) + xfree(*pheap); + *pheap = heap; + } + + SET_HEAP_SIZE(heap, ++sz); + heap[sz] = t; + t->heap_offset = sz; + up_heap(heap, sz); + return (t->heap_offset == 1); +} + + +/**************************************************************************** + * TIMER OPERATIONS. + */ + +static inline void __add_timer(struct timer *timer) +{ + int cpu = timer->cpu; + if ( add_entry(&timers[cpu].heap, timer) ) + cpu_raise_softirq(cpu, TIMER_SOFTIRQ); +} + + +static inline void __stop_timer(struct timer *timer) +{ + int cpu = timer->cpu; + if ( remove_entry(timers[cpu].heap, timer) ) + cpu_raise_softirq(cpu, TIMER_SOFTIRQ); +} + + +void set_timer(struct timer *timer, s_time_t expires) +{ + int cpu = timer->cpu; + unsigned long flags; + + spin_lock_irqsave(&timers[cpu].lock, flags); + if ( active_timer(timer) ) + __stop_timer(timer); + timer->expires = expires; + if ( likely(!timer->killed) ) + __add_timer(timer); + spin_unlock_irqrestore(&timers[cpu].lock, flags); +} + + +void stop_timer(struct timer *timer) +{ + int cpu = timer->cpu; + unsigned long flags; + + spin_lock_irqsave(&timers[cpu].lock, flags); + if ( active_timer(timer) ) + __stop_timer(timer); + spin_unlock_irqrestore(&timers[cpu].lock, flags); +} + + +void kill_timer(struct timer *timer) +{ + int cpu = timer->cpu; + unsigned long flags; + + BUG_ON(timers[cpu].running == timer); + + spin_lock_irqsave(&timers[cpu].lock, flags); + if ( active_timer(timer) ) + __stop_timer(timer); + timer->killed = 1; + spin_unlock_irqrestore(&timers[cpu].lock, flags); + + for_each_online_cpu ( cpu ) + while ( timers[cpu].running == timer ) + cpu_relax(); +} + + +static void timer_softirq_action(void) +{ + int cpu = smp_processor_id(); + struct timer *t, **heap; + s_time_t now; + void (*fn)(void *); + void *data; + + spin_lock_irq(&timers[cpu].lock); + + do { + heap = timers[cpu].heap; + now = NOW(); + + while ( (GET_HEAP_SIZE(heap) != 0) && + ((t = heap[1])->expires < (now + TIMER_SLOP)) ) + { + remove_entry(heap, t); + + timers[cpu].running = t; + + fn = t->function; + data = t->data; + + spin_unlock_irq(&timers[cpu].lock); + (*fn)(data); + spin_lock_irq(&timers[cpu].lock); + + /* Heap may have grown while the lock was released. */ + heap = timers[cpu].heap; + } + + timers[cpu].running = NULL; + } + while ( !reprogram_timer(GET_HEAP_SIZE(heap) ? heap[1]->expires : 0) ); + + spin_unlock_irq(&timers[cpu].lock); +} + + +static void dump_timerq(unsigned char key) +{ + struct timer *t; + unsigned long flags; + s_time_t now = NOW(); + int i, j; + + printk("Dumping timer queues: NOW=0x%08X%08X\n", + (u32)(now>>32), (u32)now); + + for_each_online_cpu( i ) + { + printk("CPU[%02d] ", i); + spin_lock_irqsave(&timers[i].lock, flags); + for ( j = 1; j <= GET_HEAP_SIZE(timers[i].heap); j++ ) + { + t = timers[i].heap[j]; + printk (" %d : %p ex=0x%08X%08X %p\n", + j, t, (u32)(t->expires>>32), (u32)t->expires, t->data); + } + spin_unlock_irqrestore(&timers[i].lock, flags); + printk("\n"); + } +} + + +void __init timer_init(void) +{ + static struct timer *dummy_heap; + int i; + + open_softirq(TIMER_SOFTIRQ, timer_softirq_action); + + /* + * All CPUs initially share an empty dummy heap. Only those CPUs that + * are brought online will be dynamically allocated their own heap. + */ + SET_HEAP_SIZE(&dummy_heap, 0); + SET_HEAP_LIMIT(&dummy_heap, 0); + + for ( i = 0; i < NR_CPUS; i++ ) + { + spin_lock_init(&timers[i].lock); + timers[i].heap = &dummy_heap; + } + + register_keyhandler('a', dump_timerq, "dump timer queues"); +} + +/* + * Local variables: + * mode: C + * c-set-style: "BSD" + * c-basic-offset: 4 + * tab-width: 4 + * indent-tabs-mode: nil + * End: + */ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/public/hvm/hvm_info_table.h --- /dev/null Fri Jan 13 16:38:44 2006 +++ b/xen/include/public/hvm/hvm_info_table.h Fri Jan 13 20:12:24 2006 @@ -0,0 +1,24 @@ +/****************************************************************************** + * hvm/hvm_info_table.h + * + * HVM parameter and information table, written into guest memory map. + */ + +#ifndef __XEN_PUBLIC_HVM_HVM_INFO_TABLE_H__ +#define __XEN_PUBLIC_HVM_HVM_INFO_TABLE_H__ + +#define HVM_INFO_PFN 0x09F +#define HVM_INFO_OFFSET 0x800 +#define HVM_INFO_PADDR ((HVM_INFO_PFN << 12) + HVM_INFO_OFFSET) + +struct hvm_info_table { + char signature[8]; /* "HVM INFO" */ + uint32_t length; + uint8_t checksum; + uint8_t acpi_enabled; + uint8_t apic_enabled; + uint8_t pad[1]; + uint32_t nr_vcpus; +}; + +#endif /* __XEN_PUBLIC_HVM_HVM_INFO_TABLE_H__ */ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/public/hvm/ioreq.h --- /dev/null Fri Jan 13 16:38:44 2006 +++ b/xen/include/public/hvm/ioreq.h Fri Jan 13 20:12:24 2006 @@ -0,0 +1,90 @@ +/* + * ioreq.h: I/O request definitions for device models + * Copyright (c) 2004, Intel Corporation. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., 59 Temple + * Place - Suite 330, Boston, MA 02111-1307 USA. + * + */ + +#ifndef _IOREQ_H_ +#define _IOREQ_H_ + +#define IOREQ_READ 1 +#define IOREQ_WRITE 0 + +#define STATE_INVALID 0 +#define STATE_IOREQ_READY 1 +#define STATE_IOREQ_INPROCESS 2 +#define STATE_IORESP_READY 3 +#define STATE_IORESP_HOOK 4 + +#define IOREQ_TYPE_PIO 0 /* pio */ +#define IOREQ_TYPE_COPY 1 /* mmio ops */ +#define IOREQ_TYPE_AND 2 +#define IOREQ_TYPE_OR 3 +#define IOREQ_TYPE_XOR 4 + +/* + * VMExit dispatcher should cooperate with instruction decoder to + * prepare this structure and notify service OS and DM by sending + * virq + */ +typedef struct { + uint64_t addr; /* physical address */ + uint64_t size; /* size in bytes */ + uint64_t count; /* for rep prefixes */ + union { + uint64_t data; /* data */ + void *pdata; /* pointer to data */ + } u; + uint8_t state:4; + uint8_t pdata_valid:1; /* if 1, use pdata above */ + uint8_t dir:1; /* 1=read, 0=write */ + uint8_t df:1; + uint8_t type; /* I/O type */ +} ioreq_t; + +#define MAX_VECTOR 256 +#define BITS_PER_BYTE 8 +#define INTR_LEN (MAX_VECTOR/(BITS_PER_BYTE * sizeof(uint64_t))) +#define INTR_LEN_32 (MAX_VECTOR/(BITS_PER_BYTE * sizeof(uint32_t))) + +typedef struct { + uint16_t pic_elcr; + uint16_t pic_irr; + uint16_t pic_last_irr; + uint16_t pic_clear_irr; + int eport; /* Event channel port */ +} global_iodata_t; + +typedef struct { + ioreq_t vp_ioreq; +} vcpu_iodata_t; + +typedef struct { + global_iodata_t sp_global; + vcpu_iodata_t vcpu_iodata[1]; +} shared_iopage_t; + +#endif /* _IOREQ_H_ */ + +/* + * Local variables: + * mode: C + * c-set-style: "BSD" + * c-basic-offset: 4 + * tab-width: 4 + * indent-tabs-mode: nil + * End: + */ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/public/hvm/vmx_assist.h --- /dev/null Fri Jan 13 16:38:44 2006 +++ b/xen/include/public/hvm/vmx_assist.h Fri Jan 13 20:12:24 2006 @@ -0,0 +1,97 @@ +/* + * vmx_assist.h: Context definitions for the VMXASSIST world switch. + * + * Leendert van Doorn, leendert@xxxxxxxxxxxxxx + * Copyright (c) 2005, International Business Machines Corporation. + */ + +#ifndef _VMX_ASSIST_H_ +#define _VMX_ASSIST_H_ + +#define VMXASSIST_BASE 0xD0000 +#define VMXASSIST_MAGIC 0x17101966 +#define VMXASSIST_MAGIC_OFFSET (VMXASSIST_BASE+8) + +#define VMXASSIST_NEW_CONTEXT (VMXASSIST_BASE + 12) +#define VMXASSIST_OLD_CONTEXT (VMXASSIST_NEW_CONTEXT + 4) + +#ifndef __ASSEMBLY__ + +union vmcs_arbytes { + struct arbyte_fields { + unsigned int seg_type : 4, + s : 1, + dpl : 2, + p : 1, + reserved0 : 4, + avl : 1, + reserved1 : 1, + default_ops_size: 1, + g : 1, + null_bit : 1, + reserved2 : 15; + } fields; + unsigned int bytes; +}; + +/* + * World switch state + */ +typedef struct vmx_assist_context { + uint32_t eip; /* execution pointer */ + uint32_t esp; /* stack pointer */ + uint32_t eflags; /* flags register */ + uint32_t cr0; + uint32_t cr3; /* page table directory */ + uint32_t cr4; + uint32_t idtr_limit; /* idt */ + uint32_t idtr_base; + uint32_t gdtr_limit; /* gdt */ + uint32_t gdtr_base; + uint32_t cs_sel; /* cs selector */ + uint32_t cs_limit; + uint32_t cs_base; + union vmcs_arbytes cs_arbytes; + uint32_t ds_sel; /* ds selector */ + uint32_t ds_limit; + uint32_t ds_base; + union vmcs_arbytes ds_arbytes; + uint32_t es_sel; /* es selector */ + uint32_t es_limit; + uint32_t es_base; + union vmcs_arbytes es_arbytes; + uint32_t ss_sel; /* ss selector */ + uint32_t ss_limit; + uint32_t ss_base; + union vmcs_arbytes ss_arbytes; + uint32_t fs_sel; /* fs selector */ + uint32_t fs_limit; + uint32_t fs_base; + union vmcs_arbytes fs_arbytes; + uint32_t gs_sel; /* gs selector */ + uint32_t gs_limit; + uint32_t gs_base; + union vmcs_arbytes gs_arbytes; + uint32_t tr_sel; /* task selector */ + uint32_t tr_limit; + uint32_t tr_base; + union vmcs_arbytes tr_arbytes; + uint32_t ldtr_sel; /* ldtr selector */ + uint32_t ldtr_limit; + uint32_t ldtr_base; + union vmcs_arbytes ldtr_arbytes; +} vmx_assist_context_t; + +#endif /* __ASSEMBLY__ */ + +#endif /* _VMX_ASSIST_H_ */ + +/* + * Local variables: + * mode: C + * c-set-style: "BSD" + * c-basic-offset: 4 + * tab-width: 4 + * indent-tabs-mode: nil + * End: + */ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/public/nmi.h --- /dev/null Fri Jan 13 16:38:44 2006 +++ b/xen/include/public/nmi.h Fri Jan 13 20:12:24 2006 @@ -0,0 +1,54 @@ +/****************************************************************************** + * nmi.h + * + * NMI callback registration and reason codes. + * + * Copyright (c) 2005, Keir Fraser <keir@xxxxxxxxxxxxx> + */ + +#ifndef __XEN_PUBLIC_NMI_H__ +#define __XEN_PUBLIC_NMI_H__ + +/* + * NMI reason codes: + * Currently these are x86-specific, stored in arch_shared_info.nmi_reason. + */ + /* I/O-check error reported via ISA port 0x61, bit 6. */ +#define _XEN_NMIREASON_io_error 0 +#define XEN_NMIREASON_io_error (1UL << _XEN_NMIREASON_io_error) + /* Parity error reported via ISA port 0x61, bit 7. */ +#define _XEN_NMIREASON_parity_error 1 +#define XEN_NMIREASON_parity_error (1UL << _XEN_NMIREASON_parity_error) + /* Unknown hardware-generated NMI. */ +#define _XEN_NMIREASON_unknown 2 +#define XEN_NMIREASON_unknown (1UL << _XEN_NMIREASON_unknown) + +/* + * long nmi_op(unsigned int cmd, void *arg) + * NB. All ops return zero on success, else a negative error code. + */ + +/* + * Register NMI callback for this (calling) VCPU. Currently this only makes + * sense for domain 0, vcpu 0. All other callers will be returned EINVAL. + * arg == address of callback function. + */ +#define XENNMI_register_callback 0 + +/* + * Deregister NMI callback for this (calling) VCPU. + * arg == NULL. + */ +#define XENNMI_unregister_callback 1 + +#endif /* __XEN_PUBLIC_NMI_H__ */ + +/* + * Local variables: + * mode: C + * c-set-style: "BSD" + * c-basic-offset: 4 + * tab-width: 4 + * indent-tabs-mode: nil + * End: + */ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/xen/timer.h --- /dev/null Fri Jan 13 16:38:44 2006 +++ b/xen/include/xen/timer.h Fri Jan 13 20:12:24 2006 @@ -0,0 +1,90 @@ +/****************************************************************************** + * timer.h + * + * Copyright (c) 2002-2003 Rolf Neugebauer + * Copyright (c) 2002-2005 K A Fraser + */ + +#ifndef _TIMER_H_ +#define _TIMER_H_ + +#include <xen/spinlock.h> +#include <xen/time.h> +#include <xen/string.h> + +struct timer { + /* System time expiry value (nanoseconds since boot). */ + s_time_t expires; + /* CPU on which this timer will be installed and executed. */ + unsigned int cpu; + /* On expiry, '(*function)(data)' will be executed in softirq context. */ + void (*function)(void *); + void *data; + /* Timer-heap offset. */ + unsigned int heap_offset; + /* Has this timer been killed (cannot be activated)? */ + int killed; +}; + +/* + * All functions below can be called for any CPU from any CPU in any context. + */ + +/* Returns TRUE if the given timer is on a timer list. */ +static __inline__ int active_timer(struct timer *timer) +{ + return (timer->heap_offset != 0); +} + +/* + * It initialises the static fields of the timer structure. + * It can be called multiple times to reinitialise a single (inactive) timer. + */ +static __inline__ void init_timer( + struct timer *timer, + void (*function)(void *), + void *data, + unsigned int cpu) +{ + memset(timer, 0, sizeof(*timer)); + timer->function = function; + timer->data = data; + timer->cpu = cpu; +} + +/* + * Set the expiry time and activate a timer (which must previously have been + * initialised by init_timer). + */ +extern void set_timer(struct timer *timer, s_time_t expires); + +/* + * Deactivate a timer (which must previously have been initialised by + * init_timer). This function has no effect if the timer is not currently + * active. + */ +extern void stop_timer(struct timer *timer); + +/* + * Deactivate a timer and prevent it from being re-set (future calls to + * set_timer will silently fail). When this function returns it is guaranteed + * that the timer callback handler is not running on any CPU. + */ +extern void kill_timer(struct timer *timer); + +/* + * Initialisation. Must be called before any other timer function. + */ +extern void timer_init(void); + +#endif /* _TIMER_H_ */ + +/* + * Local variables: + * mode: C + * c-set-style: "BSD" + * c-basic-offset: 4 + * tab-width: 4 + * indent-tabs-mode: nil + * End: + */ diff -r 2884f14895d2 -r f5ab28885c75 xen/arch/x86/idle0_task.c --- a/xen/arch/x86/idle0_task.c Fri Jan 13 16:38:44 2006 +++ /dev/null Fri Jan 13 20:12:24 2006 @@ -1,27 +0,0 @@ - -#include <xen/config.h> -#include <xen/sched.h> -#include <asm/desc.h> - -struct domain idle0_domain = { - domain_id: IDLE_DOMAIN_ID, - domain_flags:DOMF_idle_domain, - refcnt: ATOMIC_INIT(1) -}; - -struct vcpu idle0_vcpu = { - processor: 0, - domain: &idle0_domain -}; - -struct tss_struct init_tss[NR_CPUS]; - -/* - * Local variables: - * mode: C - * c-set-style: "BSD" - * c-basic-offset: 4 - * tab-width: 4 - * indent-tabs-mode: nil - * End: - */ diff -r 2884f14895d2 -r f5ab28885c75 xen/common/ac_timer.c --- a/xen/common/ac_timer.c Fri Jan 13 16:38:44 2006 +++ /dev/null Fri Jan 13 20:12:24 2006 @@ -1,286 +0,0 @@ -/****************************************************************************** - * ac_timer.c - * - * Copyright (c) 2002-2003 Rolf Neugebauer - * Copyright (c) 2002-2005 K A Fraser - */ - -#include <xen/config.h> -#include <xen/init.h> -#include <xen/types.h> -#include <xen/errno.h> -#include <xen/sched.h> -#include <xen/lib.h> -#include <xen/smp.h> -#include <xen/perfc.h> -#include <xen/time.h> -#include <xen/softirq.h> -#include <xen/ac_timer.h> -#include <xen/keyhandler.h> -#include <asm/system.h> -#include <asm/desc.h> - -/* - * We pull handlers off the timer list this far in future, - * rather than reprogramming the time hardware. - */ -#define TIMER_SLOP (50*1000) /* ns */ - -struct ac_timers { - spinlock_t lock; - struct ac_timer **heap; - unsigned int softirqs; -} __cacheline_aligned; - -struct ac_timers ac_timers[NR_CPUS]; - -extern int reprogram_ac_timer(s_time_t timeout); - -/**************************************************************************** - * HEAP OPERATIONS. - */ - -#define GET_HEAP_SIZE(_h) ((int)(((u16 *)(_h))[0])) -#define SET_HEAP_SIZE(_h,_v) (((u16 *)(_h))[0] = (u16)(_v)) - -#define GET_HEAP_LIMIT(_h) ((int)(((u16 *)(_h))[1])) -#define SET_HEAP_LIMIT(_h,_v) (((u16 *)(_h))[1] = (u16)(_v)) - -/* Sink down element @pos of @heap. */ -static void down_heap(struct ac_timer **heap, int pos) -{ - int sz = GET_HEAP_SIZE(heap), nxt; - struct ac_timer *t = heap[pos]; - - while ( (nxt = (pos << 1)) <= sz ) - { - if ( ((nxt+1) <= sz) && (heap[nxt+1]->expires < heap[nxt]->expires) ) - nxt++; - if ( heap[nxt]->expires > t->expires ) - break; - heap[pos] = heap[nxt]; - heap[pos]->heap_offset = pos; - pos = nxt; - } - - heap[pos] = t; - t->heap_offset = pos; -} - -/* Float element @pos up @heap. */ -static void up_heap(struct ac_timer **heap, int pos) -{ - struct ac_timer *t = heap[pos]; - - while ( (pos > 1) && (t->expires < heap[pos>>1]->expires) ) - { - heap[pos] = heap[pos>>1]; - heap[pos]->heap_offset = pos; - pos >>= 1; - } - - heap[pos] = t; - t->heap_offset = pos; -} - - -/* Delete @t from @heap. Return TRUE if new top of heap. */ -static int remove_entry(struct ac_timer **heap, struct ac_timer *t) -{ - int sz = GET_HEAP_SIZE(heap); - int pos = t->heap_offset; - - t->heap_offset = 0; - - if ( unlikely(pos == sz) ) - { - SET_HEAP_SIZE(heap, sz-1); - goto out; - } - - heap[pos] = heap[sz]; - heap[pos]->heap_offset = pos; - - SET_HEAP_SIZE(heap, --sz); - - if ( (pos > 1) && (heap[pos]->expires < heap[pos>>1]->expires) ) - up_heap(heap, pos); - else - down_heap(heap, pos); - - out: - return (pos == 1); -} - - -/* Add new entry @t to @heap. Return TRUE if new top of heap. */ -static int add_entry(struct ac_timer ***pheap, struct ac_timer *t) -{ - struct ac_timer **heap = *pheap; - int sz = GET_HEAP_SIZE(heap); - - /* Copy the heap if it is full. */ - if ( unlikely(sz == GET_HEAP_LIMIT(heap)) ) - { - /* old_limit == (2^n)-1; new_limit == (2^(n+4))-1 */ - int old_limit = GET_HEAP_LIMIT(heap); - int new_limit = ((old_limit + 1) << 4) - 1; - heap = xmalloc_array(struct ac_timer *, new_limit + 1); - BUG_ON(heap == NULL); - memcpy(heap, *pheap, (old_limit + 1) * sizeof(*heap)); - SET_HEAP_LIMIT(heap, new_limit); - if ( old_limit != 0 ) - xfree(*pheap); - *pheap = heap; - } - - SET_HEAP_SIZE(heap, ++sz); - heap[sz] = t; - t->heap_offset = sz; - up_heap(heap, sz); - return (t->heap_offset == 1); -} - - -/**************************************************************************** - * TIMER OPERATIONS. - */ - -static inline void __add_ac_timer(struct ac_timer *timer) -{ - int cpu = timer->cpu; - if ( add_entry(&ac_timers[cpu].heap, timer) ) - cpu_raise_softirq(cpu, AC_TIMER_SOFTIRQ); -} - - -static inline void __rem_ac_timer(struct ac_timer *timer) -{ - int cpu = timer->cpu; - if ( remove_entry(ac_timers[cpu].heap, timer) ) - cpu_raise_softirq(cpu, AC_TIMER_SOFTIRQ); -} - - -void set_ac_timer(struct ac_timer *timer, s_time_t expires) -{ - int cpu = timer->cpu; - unsigned long flags; - - spin_lock_irqsave(&ac_timers[cpu].lock, flags); - ASSERT(timer != NULL); - if ( active_ac_timer(timer) ) - __rem_ac_timer(timer); - timer->expires = expires; - __add_ac_timer(timer); - spin_unlock_irqrestore(&ac_timers[cpu].lock, flags); -} - - -void rem_ac_timer(struct ac_timer *timer) -{ - int cpu = timer->cpu; - unsigned long flags; - - spin_lock_irqsave(&ac_timers[cpu].lock, flags); - ASSERT(timer != NULL); - if ( active_ac_timer(timer) ) - __rem_ac_timer(timer); - spin_unlock_irqrestore(&ac_timers[cpu].lock, flags); -} - - -static void ac_timer_softirq_action(void) -{ - int cpu = smp_processor_id(); - struct ac_timer *t, **heap; - s_time_t now; - void (*fn)(void *); - - spin_lock_irq(&ac_timers[cpu].lock); - - do { - heap = ac_timers[cpu].heap; - now = NOW(); - - while ( (GET_HEAP_SIZE(heap) != 0) && - ((t = heap[1])->expires < (now + TIMER_SLOP)) ) - { - remove_entry(heap, t); - - if ( (fn = t->function) != NULL ) - { - void *data = t->data; - spin_unlock_irq(&ac_timers[cpu].lock); - (*fn)(data); - spin_lock_irq(&ac_timers[cpu].lock); - } - - /* Heap may have grown while the lock was released. */ - heap = ac_timers[cpu].heap; - } - } - while ( !reprogram_ac_timer(GET_HEAP_SIZE(heap) ? heap[1]->expires : 0) ); - - spin_unlock_irq(&ac_timers[cpu].lock); -} - - -static void dump_timerq(unsigned char key) -{ - struct ac_timer *t; - unsigned long flags; - s_time_t now = NOW(); - int i, j; - - printk("Dumping ac_timer queues: NOW=0x%08X%08X\n", - (u32)(now>>32), (u32)now); - - for_each_online_cpu( i ) - { - printk("CPU[%02d] ", i); - spin_lock_irqsave(&ac_timers[i].lock, flags); - for ( j = 1; j <= GET_HEAP_SIZE(ac_timers[i].heap); j++ ) - { - t = ac_timers[i].heap[j]; - printk (" %d : %p ex=0x%08X%08X %p\n", - j, t, (u32)(t->expires>>32), (u32)t->expires, t->data); - } - spin_unlock_irqrestore(&ac_timers[i].lock, flags); - printk("\n"); - } -} - - -void __init ac_timer_init(void) -{ - static struct ac_timer *dummy_heap; - int i; - - open_softirq(AC_TIMER_SOFTIRQ, ac_timer_softirq_action); - - /* - * All CPUs initially share an empty dummy heap. Only those CPUs that - * are brought online will be dynamically allocated their own heap. - */ - SET_HEAP_SIZE(&dummy_heap, 0); - SET_HEAP_LIMIT(&dummy_heap, 0); - - for ( i = 0; i < NR_CPUS; i++ ) - { - spin_lock_init(&ac_timers[i].lock); - ac_timers[i].heap = &dummy_heap; - } - - register_keyhandler('a', dump_timerq, "dump ac_timer queues"); -} - -/* - * Local variables: - * mode: C - * c-set-style: "BSD" - * c-basic-offset: 4 - * tab-width: 4 - * indent-tabs-mode: nil - * End: - */ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/public/io/ioreq.h --- a/xen/include/public/io/ioreq.h Fri Jan 13 16:38:44 2006 +++ /dev/null Fri Jan 13 20:12:24 2006 @@ -1,91 +0,0 @@ -/* - * ioreq.h: I/O request definitions for device models - * Copyright (c) 2004, Intel Corporation. - * - * This program is free software; you can redistribute it and/or modify it - * under the terms and conditions of the GNU General Public License, - * version 2, as published by the Free Software Foundation. - * - * This program is distributed in the hope it will be useful, but WITHOUT - * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or - * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for - * more details. - * - * You should have received a copy of the GNU General Public License along with - * this program; if not, write to the Free Software Foundation, Inc., 59 Temple - * Place - Suite 330, Boston, MA 02111-1307 USA. - * - */ - -#ifndef _IOREQ_H_ -#define _IOREQ_H_ - -#define IOREQ_READ 1 -#define IOREQ_WRITE 0 - -#define STATE_INVALID 0 -#define STATE_IOREQ_READY 1 -#define STATE_IOREQ_INPROCESS 2 -#define STATE_IORESP_READY 3 -#define STATE_IORESP_HOOK 4 - -#define IOREQ_TYPE_PIO 0 /* pio */ -#define IOREQ_TYPE_COPY 1 /* mmio ops */ -#define IOREQ_TYPE_AND 2 -#define IOREQ_TYPE_OR 3 -#define IOREQ_TYPE_XOR 4 - -/* - * VMExit dispatcher should cooperate with instruction decoder to - * prepare this structure and notify service OS and DM by sending - * virq - */ -typedef struct { - uint64_t addr; /* physical address */ - uint64_t size; /* size in bytes */ - uint64_t count; /* for rep prefixes */ - union { - uint64_t data; /* data */ - void *pdata; /* pointer to data */ - } u; - uint8_t state:4; - uint8_t pdata_valid:1; /* if 1, use pdata above */ - uint8_t dir:1; /* 1=read, 0=write */ - uint8_t df:1; - uint8_t type; /* I/O type */ -} ioreq_t; - -#define MAX_VECTOR 256 -#define BITS_PER_BYTE 8 -#define INTR_LEN (MAX_VECTOR/(BITS_PER_BYTE * sizeof(uint64_t))) -#define INTR_LEN_32 (MAX_VECTOR/(BITS_PER_BYTE * sizeof(uint32_t))) - -typedef struct { - uint16_t pic_elcr; - uint16_t pic_irr; - uint16_t pic_last_irr; - uint16_t pic_clear_irr; - int eport; /* Event channel port */ -} global_iodata_t; - -typedef struct { - ioreq_t vp_ioreq; - unsigned long vp_intr[INTR_LEN]; -} vcpu_iodata_t; - -typedef struct { - global_iodata_t sp_global; - vcpu_iodata_t vcpu_iodata[1]; -} shared_iopage_t; - -#endif /* _IOREQ_H_ */ - -/* - * Local variables: - * mode: C - * c-set-style: "BSD" - * c-basic-offset: 4 - * tab-width: 4 - * indent-tabs-mode: nil - * End: - */ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/public/vmx_assist.h --- a/xen/include/public/vmx_assist.h Fri Jan 13 16:38:44 2006 +++ /dev/null Fri Jan 13 20:12:24 2006 @@ -1,97 +0,0 @@ -/* - * vmx_assist.h: Context definitions for the VMXASSIST world switch. - * - * Leendert van Doorn, leendert@xxxxxxxxxxxxxx - * Copyright (c) 2005, International Business Machines Corporation. - */ - -#ifndef _VMX_ASSIST_H_ -#define _VMX_ASSIST_H_ - -#define VMXASSIST_BASE 0xD0000 -#define VMXASSIST_MAGIC 0x17101966 -#define VMXASSIST_MAGIC_OFFSET (VMXASSIST_BASE+8) - -#define VMXASSIST_NEW_CONTEXT (VMXASSIST_BASE + 12) -#define VMXASSIST_OLD_CONTEXT (VMXASSIST_NEW_CONTEXT + 4) - -#ifndef __ASSEMBLY__ - -union vmcs_arbytes { - struct arbyte_fields { - unsigned int seg_type : 4, - s : 1, - dpl : 2, - p : 1, - reserved0 : 4, - avl : 1, - reserved1 : 1, - default_ops_size: 1, - g : 1, - null_bit : 1, - reserved2 : 15; - } fields; - unsigned int bytes; -}; - -/* - * World switch state - */ -typedef struct vmx_assist_context { - uint32_t eip; /* execution pointer */ - uint32_t esp; /* stack pointer */ - uint32_t eflags; /* flags register */ - uint32_t cr0; - uint32_t cr3; /* page table directory */ - uint32_t cr4; - uint32_t idtr_limit; /* idt */ - uint32_t idtr_base; - uint32_t gdtr_limit; /* gdt */ - uint32_t gdtr_base; - uint32_t cs_sel; /* cs selector */ - uint32_t cs_limit; - uint32_t cs_base; - union vmcs_arbytes cs_arbytes; - uint32_t ds_sel; /* ds selector */ - uint32_t ds_limit; - uint32_t ds_base; - union vmcs_arbytes ds_arbytes; - uint32_t es_sel; /* es selector */ - uint32_t es_limit; - uint32_t es_base; - union vmcs_arbytes es_arbytes; - uint32_t ss_sel; /* ss selector */ - uint32_t ss_limit; - uint32_t ss_base; - union vmcs_arbytes ss_arbytes; - uint32_t fs_sel; /* fs selector */ - uint32_t fs_limit; - uint32_t fs_base; - union vmcs_arbytes fs_arbytes; - uint32_t gs_sel; /* gs selector */ - uint32_t gs_limit; - uint32_t gs_base; - union vmcs_arbytes gs_arbytes; - uint32_t tr_sel; /* task selector */ - uint32_t tr_limit; - uint32_t tr_base; - union vmcs_arbytes tr_arbytes; - uint32_t ldtr_sel; /* ldtr selector */ - uint32_t ldtr_limit; - uint32_t ldtr_base; - union vmcs_arbytes ldtr_arbytes; -} vmx_assist_context_t; - -#endif /* __ASSEMBLY__ */ - -#endif /* _VMX_ASSIST_H_ */ - -/* - * Local variables: - * mode: C - * c-set-style: "BSD" - * c-basic-offset: 4 - * tab-width: 4 - * indent-tabs-mode: nil - * End: - */ diff -r 2884f14895d2 -r f5ab28885c75 xen/include/xen/ac_timer.h --- a/xen/include/xen/ac_timer.h Fri Jan 13 16:38:44 2006 +++ /dev/null Fri Jan 13 20:12:24 2006 @@ -1,81 +0,0 @@ -/****************************************************************************** - * ac_timer.h - * - * Copyright (c) 2002-2003 Rolf Neugebauer - * Copyright (c) 2002-2005 K A Fraser - */ - -#ifndef _AC_TIMER_H_ -#define _AC_TIMER_H_ - -#include <xen/spinlock.h> -#include <xen/time.h> -#include <xen/string.h> - -struct ac_timer { - /* System time expiry value (nanoseconds since boot). */ - s_time_t expires; - /* CPU on which this timer will be installed and executed. */ - unsigned int cpu; - /* On expiry, '(*function)(data)' will be executed in softirq context. */ - void (*function)(void *); - void *data; - /* Timer-heap offset. */ - unsigned int heap_offset; -}; - -/* - * All functions below can be called for any CPU from any CPU in any context. - */ - -/* Returns TRUE if the given timer is on a timer list. */ -static __inline__ int active_ac_timer(struct ac_timer *timer) -{ - return (timer->heap_offset != 0); -} - -/* - * It initialises the static fields of the ac_timer structure. - * It can be called multiple times to reinitialise a single (inactive) timer. - */ -static __inline__ void init_ac_timer( - struct ac_timer *timer, - void (*function)(void *), - void *data, - unsigned int cpu) -{ - memset(timer, 0, sizeof(*timer)); - timer->function = function; - timer->data = data; - timer->cpu = cpu; -} - -/* - * Set the expiry time and activate a timer (which must previously have been - * initialised by init_ac_timer). - */ -extern void set_ac_timer(struct ac_timer *timer, s_time_t expires); - -/* - * Deactivate a timer (which must previously have been initialised by - * init_ac_timer). This function has no effect if the timer is not currently - * active. - */ -extern void rem_ac_timer(struct ac_timer *timer); - -/* - * Initialisation. Must be called before any other ac_timer function. - */ -extern void ac_timer_init(void); - -#endif /* _AC_TIMER_H_ */ - -/* - * Local variables: - * mode: C - * c-set-style: "BSD" - * c-basic-offset: 4 - * tab-width: 4 - * indent-tabs-mode: nil - * End: - */ _______________________________________________ Xen-changelog mailing list Xen-changelog@xxxxxxxxxxxxxxxxxxx http://lists.xensource.com/xen-changelog

©2013 Xen Project, A Linux Foundation Collaborative Project. All Rights Reserved.
Linux Foundation is a registered trademark of The Linux Foundation.
Xen Project is a trademark of The Linux Foundation.