Re: [PATCH] fix ia64 breakage with PHYSDEVOP_pirq_eoi_mfn (was Re: [Xen-devel] [PATCH 2/2] linux/x86: use shared page indicating the need for an EOI notification)

[Isaku Yamahata <yamahata@xxxxxxxxxxxxx>]

On Tue, Dec 09, 2008 at 11:06:58AM +0000, Keir Fraser wrote:
> On 09/12/2008 10:54, "Jan Beulich" <jbeulich@xxxxxxxxxx> wrote:
> 
> >>>> Isaku Yamahata <yamahata@xxxxxxxxxxxxx> 09.12.08 11:43 >>>
> >> You are the first person to pass the kernel symbol address
> >> to virt_to_machine() in arch independent code.
> >> Is there any necessity to allocate pirq_needs_eoi statically?
> >> (except it did before)
> > 
> > Perhaps not - avoiding the possible allocation failure (-> BUG()) and the
> > extra indirection were the main reasons I kept it allocated statically.
> > 
> >> If no, can we allocate the pages for them dynamically?
> >> Then the issue will go away.
> > 
> > Indeed.
> 
> Yes please. Just do this and be done. It's a one-off start-of-day allocation
> which, if it fails, means you're screwed anyway.

Here is.

evtchn: allocate pirq_needs_eoi bitmap dynamically.

allocating pirq_needs_eoi statically causes an address conversion
issue between ia64 and x86 because ia64 kernel symbol address
doesn't lay in 1:1 mapping area which can't be apssed to virt_to_machine().
So allocate it dynamically to work around it.

Signed-off-by: Isaku Yamahata <yamahata@xxxxxxxxxxxxx>

diff --git a/drivers/xen/core/evtchn.c b/drivers/xen/core/evtchn.c
--- a/drivers/xen/core/evtchn.c
+++ b/drivers/xen/core/evtchn.c
@@ -31,6 +31,7 @@
  */
 
 #include <linux/module.h>
+#include <linux/bootmem.h>
 #include <linux/irq.h>
 #include <linux/interrupt.h>
 #include <linux/sched.h>
@@ -755,8 +756,7 @@ static struct hw_interrupt_type dynirq_t
 
 /* Bitmap indicating which PIRQs require Xen to be notified on unmask. */
 static int pirq_eoi_does_unmask;
-static DECLARE_BITMAP(pirq_needs_eoi, ALIGN(NR_PIRQS, PAGE_SIZE * 8))
-       __attribute__ ((__section__(".bss.page_aligned"), 
__aligned__(PAGE_SIZE)));
+static unsigned long *pirq_needs_eoi;
 
 static void pirq_unmask_and_notify(unsigned int evtchn, unsigned int irq)
 {
@@ -1041,8 +1041,7 @@ void irq_resume(void)
        if (pirq_eoi_does_unmask) {
                struct physdev_pirq_eoi_gmfn eoi_gmfn;
 
-               eoi_gmfn.gmfn = arbitrary_virt_to_machine(pirq_needs_eoi)
-                       >> PAGE_SHIFT;
+               eoi_gmfn.gmfn = virt_to_machine(pirq_needs_eoi) >> PAGE_SHIFT;
                if (HYPERVISOR_physdev_op(PHYSDEVOP_pirq_eoi_gmfn, &eoi_gmfn))
                        BUG();
        }
@@ -1137,9 +1136,12 @@ void __init xen_init_IRQ(void)
 
        init_evtchn_cpu_bindings();
 
+       pirq_needs_eoi = alloc_bootmem_pages(PAGE_SIZE);
+       memset(pirq_needs_eoi, 0, PAGE_SIZE);
+       if (pirq_needs_eoi == NULL)
+               panic("failed to allocate a page for event channel.");
        BUG_ON(!bitmap_empty(pirq_needs_eoi, PAGE_SIZE * 8));
-       eoi_gmfn.gmfn = arbitrary_virt_to_machine(pirq_needs_eoi)
-               >> PAGE_SHIFT;
+       eoi_gmfn.gmfn = virt_to_machine(pirq_needs_eoi) >> PAGE_SHIFT;
        if (HYPERVISOR_physdev_op(PHYSDEVOP_pirq_eoi_gmfn, &eoi_gmfn) == 0)
                pirq_eoi_does_unmask = 1;
 


-- 
yamahata

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