Regression impacts ServeRAID II on IBM xSeries 5500

Regression impacts ServeRAID II on IBM xSeries 5500

[David Newall]

Hello Fujita-san,

This is my first time reporting a regression.  I hope I'm doing it right.

According to "git bisect," your patch
10db10d144c0248f285242f79daf6b9de6b00a62 is the first bad commit before
a regression which affects my ServeRAID II controller.  I get segfault
running ipssend, which is the program used to interrogate and configure
the SCSI RAID controller.

    ipssend[5949]: segfault at 20202020 ip b7ece8b1 sp bfce1d1c error 6 in libc-2.9.so[b7e55000+15c000]
      

Here's what git output on the final bisect:

    commit 10db10d144c0248f285242f79daf6b9de6b00a62
    Author: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
    Date:   Fri Aug 29 12:32:18 2008 +0200

        sg: convert the indirect IO path to use the block layer
        
        This patch converts the indirect IO path (including mmap IO and old
        struct sg_header) to use the block layer functions (blk_get_request,
        blk_execute_rq_nowait, blk_rq_map_user, etc) instead of
        scsi_execute_async().
        
        [Jens: fixed compile error with SCSI logging enabled]
        
        Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
        Signed-off-by: Douglas Gilbert <dougg@torque.net>
        Cc: Mike Christie <michaelc@cs.wisc.edu>
        Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
        Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

    :040000 040000 a2b5d80fb0a666b49d4bdfc2de3aa15c9d8b1ec7 84a927a0964a616313deb410a1da60dd22e2dc13 M    drivers
      

My machine and assistance is at your disposal.  Rather than guess what
information you want: what can I tell you?

Regards,

David

Re: Regression impacts ServeRAID II on IBM xSeries 5500

[FUJITA Tomonori]

Sorry for the delay,

On Sat, 18 Jul 2009 16:30:00 +0930
David Newall <davidn@davidnewall.com> wrote:

> Hello Fujita-san,
> 
> This is my first time reporting a regression.  I hope I'm doing it right.
> 
> According to "git bisect," your patch
> 10db10d144c0248f285242f79daf6b9de6b00a62 is the first bad commit before
> a regression which affects my ServeRAID II controller.  I get segfault
> running ipssend, which is the program used to interrogate and configure
> the SCSI RAID controller.
> 
>     ipssend[5949]: segfault at 20202020 ip b7ece8b1 sp bfce1d1c error 6 in libc-2.9.so[b7e55000+15c000]

Can I get 'ipssend' source code? I'm not sure what ipssend does (I
guess, it issues some special commands via sg).


> Here's what git output on the final bisect:
> 
>     commit 10db10d144c0248f285242f79daf6b9de6b00a62
>     Author: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
>     Date:   Fri Aug 29 12:32:18 2008 +0200
> 
>         sg: convert the indirect IO path to use the block layer
>         
>         This patch converts the indirect IO path (including mmap IO and old
>         struct sg_header) to use the block layer functions (blk_get_request,
>         blk_execute_rq_nowait, blk_rq_map_user, etc) instead of
>         scsi_execute_async().
>         
>         [Jens: fixed compile error with SCSI logging enabled]
>         
>         Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
>         Signed-off-by: Douglas Gilbert <dougg@torque.net>
>         Cc: Mike Christie <michaelc@cs.wisc.edu>
>         Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
>         Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

Very sorry about that.


>     :040000 040000 a2b5d80fb0a666b49d4bdfc2de3aa15c9d8b1ec7 84a927a0964a616313deb410a1da60dd22e2dc13 M    drivers
>       
> 
> My machine and assistance is at your disposal.  Rather than guess what
> information you want: what can I tell you?

What kernel version do you use, the latest git tree?

Re: Regression impacts ServeRAID II on IBM xSeries 5500

[David Newall]

FUJITA Tomonori wrote:
> Sorry for the delay,
>   

No problem. I appreciate your time.

> On Sat, 18 Jul 2009 16:30:00 +0930
> David Newall <davidn@davidnewall.com> wrote:
>   
>> According to "git bisect," your patch
>> 10db10d144c0248f285242f79daf6b9de6b00a62 is the first bad commit before
>> a regression which affects my ServeRAID II controller.  I get segfault
>> running ipssend, which is the program used to interrogate and configure
>> the SCSI RAID controller.
>>
>>     ipssend[5949]: segfault at 20202020 ip b7ece8b1 sp bfce1d1c error 6 in libc-2.9.so[b7e55000+15c000]
>>     
>
> Can I get 'ipssend' source code? I'm not sure what ipssend does (I
> guess, it issues some special commands via sg).
>   

It's provided by IBM with their ServeRAID CD, and not available in
source as far as I can discover. The RAID hardware uses Adaptec chips
and I think ipssend was written by them.


>> My machine and assistance is at your disposal.  Rather than guess what
>> information you want: what can I tell you?
>>     
>
> What kernel version do you use, the latest git tree?

I was running Ubuntu's 2.6.28-13-server, but have reverted to their
2.6.27-10-server, that being the latest Ubuntu-provided version on which
ipssend works. The patch which apparently introduced the regression is
for 2.6.27-rc9, which is quite old. The regression was a year ago, and
although I've found quite a few people who have the same problem, I
haven't found any indication of the problem having been previously
reported or bisected, so it's likely the latest version still has the
problem. I am happy to try the latest version of you want.

Re: Regression impacts ServeRAID II on IBM xSeries 5500

[FUJITA Tomonori]

On Sat, 25 Jul 2009 00:59:05 +0930
David Newall <davidn@davidnewall.com> wrote:

> >> My machine and assistance is at your disposal.  Rather than guess what
> >> information you want: what can I tell you?
> >>     
> >
> > What kernel version do you use, the latest git tree?
> 
> I was running Ubuntu's 2.6.28-13-server, but have reverted to their
> 2.6.27-10-server, that being the latest Ubuntu-provided version on which
> ipssend works. The patch which apparently introduced the regression is
> for 2.6.27-rc9, which is quite old. The regression was a year ago, and
> although I've found quite a few people who have the same problem, I
> haven't found any indication of the problem having been previously
> reported or bisected, so it's likely the latest version still has the
> problem. I am happy to try the latest version of you want.

I merged lots of changes to sg in 2.6.28. I fixed several bugs due to
it (the last one was merged in 2.6.31-rc3). Can you try the
2.6.31-rc4?

Re: Regression impacts ServeRAID II on IBM xSeries 5500

[David Newall]

FUJITA Tomonori wrote:
> I merged lots of changes to sg in 2.6.28. I fixed several bugs due to
> it (the last one was merged in 2.6.31-rc3). Can you try the
> 2.6.31-rc4?
>   

Yes, that works. Thank you very much.