pm/linux-next cause network stop working with qemu-kvm/e1000

pm/linux-next cause network stop working with qemu-kvm/e1000

[Yinghai Lu]

c13085e519e8984fede41fa3d6a5502523b10996 is the first bad commit
commit c13085e519e8984fede41fa3d6a5502523b10996
Author: Bob Moore <robert.moore@intel.com>
Date:   Fri Mar 8 09:19:38 2013 +0000

    ACPICA: Resource Mgr: Prevent infinite loops in resource walks

    Add checks for zero-length resource descriptors in all code that
    loops through a resource descriptor list. This prevents possible
    infinite loops because the length is used to increment the traveral
    pointer and detect the end-of-descriptor.

    Signed-off-by: Bob Moore <robert.moore@intel.com>
    Signed-off-by: Lv Zheng <lv.zheng@intel.com>
    Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

:040000 040000 84f29d6ee654ffe4c935ce9023208019b2ef09a5
c8a6613be9a5a130f69e654c5b560a8adb2174db M	drivers


long delay after
calling  ip_auto_config+0x0/0xea4 @ 1

also happen on one 8 sockets nehalem-ex.

If I revert that commit, network works again.

Thanks

Yinghai

Re: pm/linux-next cause network stop working with qemu-kvm/e1000

[Yinghai Lu]

[-- Attachment #1: Type: text/plain, Size: 1168 bytes --]

On Fri, Mar 22, 2013 at 5:27 PM, Yinghai Lu <yinghai@kernel.org> wrote:
> c13085e519e8984fede41fa3d6a5502523b10996 is the first bad commit
> commit c13085e519e8984fede41fa3d6a5502523b10996
> Author: Bob Moore <robert.moore@intel.com>
> Date:   Fri Mar 8 09:19:38 2013 +0000
>
>     ACPICA: Resource Mgr: Prevent infinite loops in resource walks
>
>     Add checks for zero-length resource descriptors in all code that
>     loops through a resource descriptor list. This prevents possible
>     infinite loops because the length is used to increment the traveral
>     pointer and detect the end-of-descriptor.
>
>     Signed-off-by: Bob Moore <robert.moore@intel.com>
>     Signed-off-by: Lv Zheng <lv.zheng@intel.com>
>     Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
>
> :040000 040000 84f29d6ee654ffe4c935ce9023208019b2ef09a5
> c8a6613be9a5a130f69e654c5b560a8adb2174db M      drivers
>
>
> long delay after
> calling  ip_auto_config+0x0/0xea4 @ 1
>
> also happen on one 8 sockets nehalem-ex.
>
> If I revert that commit, network works again.

Attached patch fix the problem. Please fold it into offending patch in
pm/linux-next.

Thanks

Yinghai

[-- Attachment #2: fix_x.patch --]
[-- Type: application/octet-stream, Size: 542 bytes --]

---
 drivers/acpi/pci_link.c |    1 +
 1 file changed, 1 insertion(+)

Index: linux-2.6/drivers/acpi/pci_link.c
===================================================================
--- linux-2.6.orig/drivers/acpi/pci_link.c
+++ linux-2.6/drivers/acpi/pci_link.c
@@ -354,6 +354,7 @@ static int acpi_pci_link_set(struct acpi
 
 	}
 	resource->end.type = ACPI_RESOURCE_TYPE_END_TAG;
+	resource->end.length = sizeof(struct acpi_resource);
 
 	/* Attempt to set the resource */
 	status = acpi_set_current_resources(link->device->handle, &buffer);

Re: pm/linux-next cause network stop working with qemu-kvm/e1000

[Rafael J. Wysocki]

On Saturday, March 23, 2013 01:38:43 AM Yinghai Lu wrote:
> On Fri, Mar 22, 2013 at 5:27 PM, Yinghai Lu <yinghai@kernel.org> wrote:
> > c13085e519e8984fede41fa3d6a5502523b10996 is the first bad commit
> > commit c13085e519e8984fede41fa3d6a5502523b10996
> > Author: Bob Moore <robert.moore@intel.com>
> > Date:   Fri Mar 8 09:19:38 2013 +0000
> >
> >     ACPICA: Resource Mgr: Prevent infinite loops in resource walks
> >
> >     Add checks for zero-length resource descriptors in all code that
> >     loops through a resource descriptor list. This prevents possible
> >     infinite loops because the length is used to increment the traveral
> >     pointer and detect the end-of-descriptor.
> >
> >     Signed-off-by: Bob Moore <robert.moore@intel.com>
> >     Signed-off-by: Lv Zheng <lv.zheng@intel.com>
> >     Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> >
> > :040000 040000 84f29d6ee654ffe4c935ce9023208019b2ef09a5
> > c8a6613be9a5a130f69e654c5b560a8adb2174db M      drivers
> >
> >
> > long delay after
> > calling  ip_auto_config+0x0/0xea4 @ 1
> >
> > also happen on one 8 sockets nehalem-ex.
> >
> > If I revert that commit, network works again.
> 
> Attached patch fix the problem. Please fold it into offending patch in
> pm/linux-next.

I'd actually prefer to apply it as a separate patch, so can you please post
it with a proper changelog?

Rafael


-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.

Re: pm/linux-next cause network stop working with qemu-kvm/e1000

[Yinghai Lu]

On Sat, Mar 23, 2013 at 5:30 AM, Rafael J. Wysocki <rjw@sisk.pl> wrote:

>>
>> Attached patch fix the problem. Please fold it into offending patch in
>> pm/linux-next.
>
> I'd actually prefer to apply it as a separate patch, so can you please post
> it with a proper changelog?

but that will break bisecting, right?

Yinghai

Re: pm/linux-next cause network stop working with qemu-kvm/e1000

[Rafael J. Wysocki]

On Saturday, March 23, 2013 09:18:55 AM Yinghai Lu wrote:
> On Sat, Mar 23, 2013 at 5:30 AM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> 
> >>
> >> Attached patch fix the problem. Please fold it into offending patch in
> >> pm/linux-next.
> >
> > I'd actually prefer to apply it as a separate patch, so can you please post
> > it with a proper changelog?
> 
> but that will break bisecting, right?

We need to keep ACPICA changes and Linux-specific modifications separate
for process-related reasons.

Thanks,
Rafael


-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.