[OpenAFS] 1.6.0 causes kernel panic (OS X 10.7.1 32-bit kernel)

Stuart Yarrow s.yarrow@ed.ac.uk
Fri, 2 Sep 2011 11:51:18 +0100

Hi Derrick,

That works!  AFS plays nice with kerberos, mounts successfully.  Many =



On 2 Sep 2011, at 05:22, Derrick Brashear wrote:

> On Thu, Sep 1, 2011 at 10:51 PM, Derrick Brashear <shadow@gmail.com> =
>> a decoded panic (/Library/OpenAFS/Tools/tools/decode-panic) is always
>> more interesting. I decoded it, tho, and...
>> In any case, i have seen this one before, in 10.6, 32 bit.
>> 0x22032e <panic+510>:   mov    0x85fab4,%eax
>> 0x2ccb0f <kernel_trap+1535>:    add    $0x9c,%esp
>> 0x2e2e34 <return_from_trap+212>:        mov    0x8(%esp),%esp
>> 0x4943a7d8 <_rxevent_Post+552>: mov    %ecx,(%eax)
>> 0x4943acf4 <rxevent_Post+100>:  mov    %eax,-0x20(%ebp)
>> 0x4943367c <rxi_ReapConnections+4812>:  add    $0x128,%esp
>> 0x4941a0c8 <rx_StartServer+376>:        mov    -0x4(%ebp),%eax
>> 0x493d87e3 <afs_ResourceInit+395>:      movl   $0x49474948,(%esp)
>> 0x49456e99 <afs_InitSetup+169>: movl   $0x1,0x49472cf4
>> 0x494574ee <afs_DaemonOp+78>:   mov    %eax,-0x34(%ebp)
>> in that case, it went away... when I recompiled with a different
>> Xcode, namely, 3.2.3; For Lion, that's not an option.
>> Fun thing is, providing information to track this to Apple is =
>> a huge adventure, because it's probably not our bug.
>> I'll see what I can come up with.
> Try this. No code changes, so it's still labelled (correctly) 1.6.0.
> I smashed my face against the compiler a bit and it spit this out.
> It's ok on the 32 bit test box (where I couldn't produce the above
> crash with a debug build, but got a different crash).
> /afs/your-file-system.com/user/shadow/OpenAFS-1.6.0-Lion.dmg
> --=20
> Derrick


Stuart Yarrow
Research Student

Institute for Adaptive and Neural Computation
Neuroinformatics and Computational Neuroscience DTC
School of Informatics, University of Edinburgh


The University of Edinburgh is a charitable body, registered in =
Scotland, with registration number SC005336.