Re: in 2.6.23-rc3-git7 in do_cciss_intr

[Randy Dunlap <randy.dunlap@oracle.com>]

Miller, Mike (OS Dev) wrote:
> 
>> -----Original Message-----
>> From: James Bottomley [mailto:James.Bottomley@HansenPartnership.com]
>> Sent: Friday, August 22, 2008 12:02 PM
>> To: Randy Dunlap
>> Cc: Miller, Mike (OS Dev); lkml; scsi; akpm
>> Subject: Re: in 2.6.23-rc3-git7 in do_cciss_intr
>>
>> On Fri, 2008-08-22 at 09:49 -0700, Randy Dunlap wrote:
>>> James Bottomley wrote:
>>>> On Fri, 2008-08-22 at 15:48 +0000, Miller, Mike (OS Dev) wrote:
>>>>>> -----Original Message-----
>>>>>> From: Randy Dunlap [mailto:randy.dunlap@oracle.com]
>>>>>> Sent: Thursday, August 21, 2008 7:27 PM
>>>>>> To: Miller, Mike (OS Dev)
>>>>>> Cc: lkml; scsi; akpm
>>>>>> Subject: Re: in 2.6.23-rc3-git7 in do_cciss_intr
>>>>>>
>>>>>> On Thu, 21 Aug 2008 16:25:24 +0000 Miller, Mike (OS Dev) wrote:
>>>>>>
>>>>>>>>> Randy,
>>>>>>>>> We know of a race condition in cciss_init_one. It's fixed
>>>>>>>> in 2.6.26 I believe. Here's the patch:
>> http://groups.google.com/group/linux.kernel/browse_thread/thread/7
>>>>>> b3
>>>>>>>> 9f
>> 2b77622ab03/4f5f45c008655ca1?hl=en&lnk=gst&q=cciss#4f5f45c008655ca
>>>>>>>>> 1
>>>>>>>> Mike,
>>>>>>>> Sorry, but my fingers have typoed the $subject.  My bad.
>>>>>>>> Kernel is 2.6.27-rc3-git7 (from above):
>>>>>>>>
>>>>>>>>>>>> Modules linked in: cciss(+) ehci_hcd ohci_hcd uhci_hcd
>>>>>>>>>>>> Pid: 0, comm: swapper Not tainted 2.6.27-rc3-git7 #1
>>>>>>>>>>>> RIP: 0010:[<ffffffffa001bb68>]  [<ffffffffa001bb68>]
>>>>>>>>>>>> do_cciss_intr+0x627/0xa6c [cciss]
>>>>>>> Hmmmmm, let me know what happens from your retest. I'll
>>>>>> look at this
>>>>>>> as soon as I finish what I'm doing now. We trying to spin
>>>>>> for our test
>>>>>>> teams but I have something hopelessly broken. :(
>>>>>> It didn't BUG in the retest.  That just means that it's more
>>>>>> difficult to find/fix, right?
>>>>> Yup.
>>>> Randy,
>>>>
>>>> If you can't reproduce it, could you use the debug information or
>>>> gdb to tell us what line in the source code this:
>>>>
>>>> do_cciss_intr+0x627
>>>>
>>>> corresponds to?  That might help isolating the problem.
>>>
>>> Sure, here's an attempt at that.  Please let me know if you want it
>>> differently or some other info.
>>> (gdb) x/20i do_cciss_intr+0x627
>>> 0x3b68 <do_cciss_intr+1575>:    mov    %rdx,0x248(%rax)
>>> 0x3b6f <do_cciss_intr+1582>:    mov    0x248(%rbx),%rdx
>>> 0x3b76 <do_cciss_intr+1589>:    mov    %rax,0x240(%rdx)
>>> 0x3b7d <do_cciss_intr+1596>:    jmp    0x3b8b <do_cciss_intr+1610>
>>> 0x3b7f <do_cciss_intr+1598>:    movq   $0x0,0x100c0(%r12)
>>> 0x3b8b <do_cciss_intr+1610>:    mov    0x234(%rbx),%eax
>>> 0x3b91 <do_cciss_intr+1616>:    test   %eax,%eax
>>> 0x3b93 <do_cciss_intr+1618>:    jne    0x3f27 <do_cciss_intr+2534>
>>> 0x3b99 <do_cciss_intr+1624>:    mov    0x250(%rbx),%r14
>>> 0x3ba0 <do_cciss_intr+1631>:    movl   $0x0,0xcc(%r14)
>>> 0x3bab <do_cciss_intr+1642>:    mov    0x228(%rbx),%r8
>>> 0x3bb2 <do_cciss_intr+1649>:    mov    0x2(%r8),%dx
>>> 0x3bb7 <do_cciss_intr+1654>:    test   %dx,%dx
>>> 0x3bba <do_cciss_intr+1657>:    je     0x3f0e <do_cciss_intr+2509>
>>>
>>>
>>> $ addr2line -e cciss.o -f  do_cciss_intr+0x627 SA5_fifo_full
>>> /home/rdunlap/linsrc/linux-2.6.27-rc3-git7/drivers/block/cciss.h:206
>> OK ...that's confusing.  It seems to be saying that
>> ctrlr_info_t * was NULL.  However, I can't see a way of
>> getting into the fifo_full callback from do_cciss_intr ..
>> especially not with an NULL host.
>>
>> James
> 
> That is weird. Even if we could get there fifo_full doesn't do anything but wait for a bit.

Hi,

This just happened again.  This time it's on 2.6.27-rc5-git3.

~Randy