Re: 1.8.x AIX support

On Tue, 16 May 2023 17:21:58 +0000
Ben Huntsman <[email protected]> wrote:

> Hi there!
>    Here is the backtrace with a debug build:
> 
> bash-4.2# dbx /opt/openafs/libexec/openafs/vlserver core
> Type 'help' for help.
> [using memory image in core]
> reading symbolic information ...
> 
> IOT/Abort trap in pthread_kill at 0xd054cb34 ($t2)
> 0xd054cb34 (pthread_kill+0xb4) 80410014            lwz   r2,0x14(r1)
> (dbx) where all
> Thread $t1
> _sigsetmask(??, ??, ??) at 0xd054b288
> _p_sigaction(??, ??, ??) at 0xd054be68
> raise.sigaction(??, ??, ??) at 0xd0120d50
> signal(??, ??) at 0xd021f6f4
> SetupLogSoftSignals(), line 469 in "serverLog.c"
> main(argc = 0, argv = (nil)), line 399 in "vlserver.c"
> Thread $t2
> pthread_kill(??, ??) at 0xd054cb34
> _p_raise(??) at 0xd054bf84
> raise.raise(??) at 0xd0121020
> abort() at 0xd017ca64
> opr_AssertionFailed(file = (nil), line = 0), line 29 in "assert.c"
> signalHandler(arg = (nil)), line 73 in "softsig.c"
> 
> 
> Hopefully that sheds more light on the situation?
> 
> Thank you!
> 
> -Ben
> 
> ________________________________
> From: Cheyenne Wills <[email protected]>
> Sent: Tuesday, May 16, 2023 6:04 AM
> To: Ben Huntsman <[email protected]>
> Cc: [email protected] <[email protected]>
> Subject: Re: [OpenAFS-devel] 1.8.x AIX support
> 
> On Tue, 16 May 2023 04:46:40 +0000
> Ben Huntsman <[email protected]> wrote:
> 
> > Hi there-
> >    I see that the three AIX issues are being pulled up to 1.8.x.  I
> > just tried them out and found that we also need this one:
> >
> > 15106
> >
> >    We get a kernel panic on AIX as soon as afsd is started without
> > 15106 applied.
> >
> >    I pulled in 15106 and it compiles, but then I have another
> > problem that is that many of the servers coredump immediately:
> >
> > # /opt/openafs/libexec/openafs/vlserver
> > IOT/Abort trap(coredump)
> > # dbx /opt/openafs/libexec/openafs/vlserver core
> > Type 'help' for help.
> > [using memory image in core]
> > reading symbolic information ...warning: no source compiled with -g
> >
> >
> > IOT/Abort trap in pthread_kill at 0xd054cb34 ($t2)
> > 0xd054cb34 (pthread_kill+0xb4) 80410014            lwz   r2,0x14(r1)
> > (dbx) where all
> > Thread $t1
> > _sigsetmask(??, ??, ??) at 0xd054b288
> > _p_sigaction(??, ??, ??) at 0xd054be68
> > raise.sigaction(??, ??, ??) at 0xd0120d50
> > signal(??, ??) at 0xd021f6f4
> > .() at 0x100708e4
> > .() at 0x100019fc
> > Thread $t2
> > pthread_kill(??, ??) at 0xd054cb34
> > _p_raise(??) at 0xd054bf84
> > raise.raise(??) at 0xd0121020
> > abort() at 0xd017ca64
> > .() at 0x10005f8c
> > .() at 0x10071670
> >
> >
> > Are we missing another patch as well?  Anyone have an idea what it
> > might be?
> >
> > Thanks much!
> >
> > -Ben
> >
> >
> 
> Can you try doing a build with --enable-debug as a configure option.
> That should provide information for the backtrace.
> 
> --
> Cheyenne Wills
> [email protected]


Looks like gerrit 14705 (from master) is needed as well
-- 
Cheyenne Wills
[email protected]