[Xen-devel] RE: Kernel BUG at arch/x86/mm/tlb.c:61

[MaoXiaoyun <tinnycloud@xxxxxxxxxxx>]

 
> Date: Fri, 15 Apr 2011 14:22:29 -0700
> From: jeremy@xxxxxxxx
> To: tinnycloud@xxxxxxxxxxx
> CC: giamteckchoon@xxxxxxxxx; xen-devel@xxxxxxxxxxxxxxxxxxx; konrad.wilk@xxxxxxxxxx
> Subject: Re: Kernel BUG at arch/x86/mm/tlb.c:61
>
> On 04/15/2011 05:23 AM, MaoXiaoyun wrote:
> > Hi：
> >
> > Could the crash related to this patch ?
> > http://git.kernel.org/?p=linux/kernel/git/jeremy/xen.git;a=commitdiff;h=45bfd7bfc6cf32f8e60bb91b32349f0b5090eea3
> >
> > Since now TLB state change to TLBSTATE_OK(mmu_context.h:40) is before
> > cpumask_clear_cpu(line 49).
> > Could it possible that right after execute line 40 of mmu_context.h,
> > CPU revice IPI from other CPU to
> > flush the mm, and when in interrupt, find the TLB state happened to be
> > TLBSTATE_OK. Which conflicts.
>
> Does reverting it help?
>
> J
 
Very likely.
 
Previously in 17 machines test, one to three machines will fail in 10hours, very easily.
 
But after reverting, we have 29machines involved the test, 28 successfuly rung 2 days, 1 fail after 28 hours.
Unfortunately I can't tell wether the failed one related to this bug, since I got no log in messages.  And
the machine was reboot by someone before I can see something from serial port.
 
But in my opinion the fail points to another bug, which I happened to confront before. 
 
Before, one of my develop machine(2.6.32.36kernel+xen4.0.1) completely stop response,
including serial console. There is no abnormal message in serial port,  looks like Xen runs in deadlock. 
Well, it is rarely happen, since I only met once till now. 
 
Now I am trying to figure out what might cause the deadlock, we never met this before.
I don't have clear thoughts on how to dig it out, but  I think this bug exists in Xen.
since if dom0 hangs, xen should work,  and serial output will response.
 If so, the bug may be introduced between 4.0.0 and 4.0.1.
 
What do you think,  thanks.

> >
> > Thanks.
> >
> > arch/x86/include/asm/mmu_context.h
> >
> > 33 static inline void switch_mm(struct mm_struct *prev, struct
> > mm_struct *next,
> > 34 <+++<+++<+++ struct task_struct *tsk)
> > 35 {
> > 36 <+++unsigned cpu = smp_processor_id();
> > 37
> > 38 <+++if (likely(prev != next)) {
> > 39 #ifdef CONFIG_SMP
> > 40 <+++<+++percpu_write(cpu_tlbstate.state, TLBSTATE_OK);
> > 41 <+++<+++percpu_write(cpu_tlbstate.active_mm, next);
> > 42 #endif
> > 43 <+++<+++cpumask_set_cpu(cpu, mm_cpumask(next));
> > 44
> > 45 <+++<+++/* Re-load page tables */
> > 46 <+++<+++load_cr3(next->pgd);
> > 47
> > 48 <+++<+++/* stop flush ipis for the previous mm */
> > 49 <+++<+++cpumask_clear_cpu(cpu, mm_cpumask(prev));
&g t; >
> >
>

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