On 14/06/11 09:46, Jan Beulich wrote:
>>>> On 13.06.11 at 19:02, Andrew Cooper <andrew.cooper3@xxxxxxxxxx> wrote:
>> nmi_shootdown_cpus is part of the kexec path, coming from a panic, and
>> as such can be called both with interrupts enabled or disabled. We
>> really dont want to accidentally set IF.
> Can interrupts really be enabled when entering this function?
It is unlikely, but I believe there are a few codepaths which could
enter this function with interrupts enabled. Either way, the last thing
we need is the usual occurrence of leaving this function with interrupts
>> Therefore, use save/restore in preference to disable/enable.
> I.e. wouldn't just removing the stray local_irq_enable() suffice?
I was trying to go for minimal changes to the Linux code. I would agree
that a single disable right at the top of the crash path might be a
>> Signed-off-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
>> diff -r 076c3034c8c7 -r 1c3d2e4d06fe xen/arch/x86/crash.c
>> --- a/xen/arch/x86/crash.c Mon Jun 13 17:45:43 2011 +0100
>> +++ b/xen/arch/x86/crash.c Mon Jun 13 17:45:43 2011 +0100
>> @@ -55,9 +55,9 @@ static int crash_nmi_callback(struct cpu
>> static void nmi_shootdown_cpus(void)
>> - unsigned long msecs;
>> + unsigned long msecs, flags;
>> - local_irq_disable();
>> + local_irq_save(flags);
>> crashing_cpu = smp_processor_id();
>> local_irq_count(crashing_cpu) = 0;
>> @@ -80,7 +80,7 @@ static void nmi_shootdown_cpus(void)
>> - local_irq_enable();
>> + local_irq_restore(flags);
>> void machine_crash_shutdown(void)
>> Xen-devel mailing list
Andrew Cooper - Dom0 Kernel Engineer, Citrix XenServer
T: +44 (0)1223 225 900, http://www.citrix.com
Xen-devel mailing list