Xen project Mailing List

Hmm, that's pretty much impossible unless either of two assumptions
doesn't hold: Processes don't share (leaf) page table pages, and within a
process any (leaf) page table page can be used only in exactly one place.
Unless that's wrong, _pin_lock() guarantees that each lock gets acquired
exactly once.

Since the locks' addresses aren't being printed, it's hard to judge whether
what you got is an indication of a real problem. Otoh, if there was such
a problem, I'd expect quite a few people would already have reported
deadlocks... So, did the machine indeed hang after printing those
messages?

Jan

>>> Boris Derzhavets <bderzhavets@xxxxxxxxx> 23.10.09 18:05 >>>
Dmesg log attached

=============================================
[ INFO: possible recursive locking detected ]
2.6.31.4 #1
---------------------------------------------
init/1 is trying to acquire lock:
(__pte_lockptr(page)){+.+...}, at: [<ffffffff81025090>] _pin_lock+0xb0/0x124

but task is already holding lock:
(__pte_lockptr(page)){+.+...}, at: [<ffffffff81025090>] _pin_lock+0xb0/0x124

other info that might help us debug this:
4 locks held by init/1:
#0: (&mm->mmap_sem){++++++}, at: [<ffffffff8104193f>] dup_mm+0xc7/0x3a2
#1: (&mm->mmap_sem/1){+.+.+.}, at: [<ffffffff81041954>] dup_mm+0xdc/0x3a2
#2: (&mm->page_table_lock){+.+...}, at: [<ffffffff8102501b>] _pin_lock+0x3b/0x124
#3: (__pte_lockptr(page)){+.+...}, at: [<ffffffff81025090>] _pin_lock+0xb0/0x124

stack backtrace:
Pid: 1, comm: init Not tainted 2.6.31.4 #1
Call Trace:
[<ffffffff81072a6d>] __lock_acquire+0xb70/0xbfa
[<ffffffff81072beb>] lock_acquire+0xf4/0x152
[<ffffffff81025090>] ? _pin_lock+0xb0/0x124
[<ffffffff81025090>] ? _pin_lock+0xb0/0x124
[<ffffffff814bc8bb>] _spin_lock+0x45/0x8e
[<ffffffff81025090>] ? _pin_lock+0xb0/0x124
[<ffffffff81025090>] _pin_lock+0xb0/0x124
[<ffffffff81025928>] mm_pin+0x33/0x61
[<ffffffff810259a3>] arch_dup_mmap+0x4d/0x6a
[<ffffffff81041b5d>] dup_mm+0x2e5/0x3a2
[<ffffffff81042823>] copy_process+0xba2/0x13c5
[<ffffffff810431b9>] do_fork+0x173/0x335
[<ffffffff8110b3a9>] ? do_vfs_ioctl+0x4aa/0x506
[<ffffffff8100a353>] ? sysret_check+0x41/0xc0
[<ffffffff81008358>] sys_clone+0x3b/0x51
[<ffffffff814bbebb>] ? trace_hardirqs_on_thunk+0x3a/0x3f
[<ffffffff8100a7b3>] stub_clone+0x13/0x20
[<ffffffff8100a308>] ? system_call_fastpath+0x16/0x1b

Boris

--- On Fri, 10/23/09, Andrew Lyon <andrew.lyon@xxxxxxxxx> wrote:

From: Andrew Lyon <andrew.lyon@xxxxxxxxx>
Subject: Re: [Xen-users] rebased opensuse 2.6.31 xen patches
To: "Marco Nenciarini" <mnencia@xxxxxxxxxxxxxx>, "Jan Beulich" <jbeulich@xxxxxxxxxx>, "xen-users List"
<xen-users@xxxxxxxxxxxxxxxxxxx>
Date: Friday, October 23, 2009, 7:28 AM

On Fri, Oct 23, 2009 at 12:03 PM, Marco Nenciarini
<mnencia@xxxxxxxxxxxxxx> wrote:
> Andrew Lyon ha scritto:
>>
>> On Fri, Oct 23, 2009 at 11:27 AM, Marco Nenciarini
>>
>> Thanks, does this fix a problem you have encountered? If so I would
>> like to see an example of the oops or warning you receive due to this
>> error.
>>
>> Andy
>
> Without that patch it does not compile at all.
>
>
> scripts/kconfig/conf -s arch/x86/Kconfig
> Using
> /home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen
> as source for kernel
> GEN
> /home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/build_amd64_xen_amd64/Makefile
> CHK    include/linux/version.h
> UPD    include/linux/version.h
> CHK    include/linux/utsrelease.h
> UPD    include/linux/utsrelease.h
> SYMLINK include/asm -> include/asm-x86
> Updating
> /home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/build_amd64_xen_amd64/scripts/Makefile.xen
> CC kernel/bounds.s
> GEN    include/linux/bounds.h
> CC arch/x86/kernel/asm-offsets.s
> In file included from
> /home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/include/linux/spinlock.h:88,
>    from
> /home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/include/linux/seqlock.h:29,
>    from
> /home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/include/linux/time.h:8,
>    from
> /home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/include/linux/stat.h:60,
>    from
> /home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/include/linux/module.h:10,
>    from
> /home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/include/linux/crypto.h:21,
>    from
> /home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/arch/x86/kernel/asm-offsets_64.c:7,
>    from
> /home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/arch/x86/kernel/asm-offsets.c:4:
>
/home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/arch/x86/include/mach-xen/asm/spinlock.h:134:
> error: expected identifier or â(â before â:â token
>
/home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/arch/x86/include/mach-xen/asm/spinlock.h:138:
> error: expected identifier or â(â before â}â token
>
/home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/arch/x86/include/mach-xen/asm/spinlock.h:138:
> error: expected identifier or â(â before âwhileâ
>
/home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/arch/x86/include/mach-xen/asm/spinlock.h:
> In function â__ticket_spin_lockâ:
>
/home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/arch/x86/include/mach-xen/asm/spinlock.h:217:
> error: expected â:â or â)â before â;â token
>
/home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/arch/x86/include/mach-xen/asm/spinlock.h:217:
> warning: unused variable âtmpâ
>
/home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/arch/x86/include/mach-xen/asm/spinlock.h:227:
> error: expected âwhileâ before âstaticâ
> /home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/arch/x86/kernel/asm-offsets.c:5:
> error: expected declaration or statement at end of input
>
/home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/arch/x86/include/mach-xen/asm/spinlock.h:215:
> warning: unused variable âfreeâ
>
/home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/arch/x86/include/mach-xen/asm/spinlock.h:214:
> warning: unused variable âcountâ
>
/home/mnencia/deb/kernel/lab/linux-2.6-2.6.31/debian/build/source_amd64_xen/arch/x86/include/mach-xen/asm/spinlock.h:214:
> warning: unused variable âtokenâ
> make[5]: *** [arch/x86/kernel/asm-offsets.s] Error 1
> make[4]: *** [prepare0] Error 2
>
> Marco
>
> --
> ---------------------------------------------------------------------
> | Marco Nenciarini | Debian/GNU Linux Developer - Plug Member |
> | mnencia@xxxxxxxxxxxxxx | http://www.prato.linux.it/~mnencia    |
> ---------------------------------------------------------------------
> Key fingerprint = FED9 69C7 9E67 21F5 7D95 5270 6864 730D F095 E5E4
>
>

Weird, I can compile both 32 and 64 bit domU and dom0 without any
errors like that, looking at the code I'd expect to hit that error
straight away.

I've CC'ed Jan and attached your patch to this message, he may want to
merge this upstream (or point out that I'm wrong again).

Andy

-----Inline Attachment Follows-----

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users