[ xcat-Bugs-3538093 ] Linux chroot uname -r not returning rootimg level

[SourceForge.net <no...@so...>]

Bugs item #3538093, was opened at 2012-06-26 05:33
Message generated for change (Comment added) made by bcroswell
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=1006945&aid=3538093&group_id=208749

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: Infiniband
>Group: 2.7.5
Status: Open
Resolution: Remind
Priority: 5
Private: No
Submitted By: Mark Perez (markprez)
Assigned to: jjh (jjhua)
Summary: Linux chroot uname -r not returning rootimg level

Initial Comment:
On Rh6.2 we were running the procedure to add Melanox OFED to our diskless image. The EMS was running
kernel version 2.6.32-220.el6.x86_64 and the image had been updated to 2.7.1-2.6.32_220.13.1 and the
mellanox support was added for this version.
There are several ofed rpms that are kernel version dependent, for example:
./RPMS/kernel-mft-2.7.1-2.6.32_220.13.1.el6.x86_64.x86_64.rpm
./RPMS/kernel-mft-2.7.1-2.6.32_220.el6.x86_64.x86_64.rpm
The mlnxofedinstall script uses "uname -r" to determine which version of the rpms to install. uname -r returned the EMS level and not the rootimg level. 
dmesg shows the messages:
mlx4_ib: disagrees about version of symbol mlx4_find_cached_vlan
mlx4_ib: Unknown symbol mlx4_find_cached_vlan
mlx4_ib: disagrees about version of symbol mlx4_buf_write_mtt
mlx4_ib: Unknown symbol mlx4_buf_write_mtt
mlx4_ib: Unknown symbol mlx4_cq_ignore_overrun
*
Workaround:
If needed, add the kernel support for ofed and follow the setup instructions to install the ofed rpms. Change the mlnxofedinstall script line:
my $kernel = `uname -r`;
to assign the kernel level:
my $kernel = "2.7.1-2.6.32_220.13.1"





----------------------------------------------------------------------

>Comment By: Brian  Croswell (bcroswell)
Date: 2012-09-14 05:39

Message:
I am cleaning up all open xCAT 2.74 bugs  and putting them in new
releases.

Jie Hua,   Mark,      
Did we fix  this IB issue  in xCAT 2.7.4  with documentation or work
around.
Please update the xCAT bug to reference any new information.  


----------------------------------------------------------------------

Comment By: jjh (jjhua)
Date: 2012-08-28 18:58

Message:
Hi Bruce,

In 07/02/2012, I provided a workaround using  a temporary  uname in
genimage, and the code was checked into  trunk Revision 13224  and Revision
13222 ,  xCAT 2.7 Revision 13223 and Revision 13221 .  

In 07/13/2012,  Mellanox Technical Support gave some responses, and it
looks like they make it as a feature request.  The details is as
following:
-----
Dear ibmtc, 

Mellanox support case # 00062438 has been updated with the following
comment: 
/-------------------------------------------------------------
Hello Mark,

I would like to check this issue and get back to you, i think that we are a
bit restricted with changing the code to support it. 
This is more as feature request. 

Idan
-------------------------------------------------------------/

Thank You,

Mellanox Technical Support

You can access all your closed/Open cases, just go to
http://mellanox.com/content/pages.php?pg=customer_support

-------

Currently, there is not a new response from the Mellanox.  I will add some
comments to the Mellanox Case # 00062438 . 

Thanks


----------------------------------------------------------------------

Comment By: Bruce (bp-sawyers)
Date: 2012-08-28 10:01

Message:
Jie Hua,

Have you heard back from mellanox about your 2 suggested solutions?  I'm
changing the resolution to Remind, since you are waiting for them.

----------------------------------------------------------------------

Comment By: jjh (jjhua)
Date: 2012-06-28 00:27

Message:
Hi Bruce, 

Guang Cheng and I raised 2 solutions for Mellonax. We are waiting for their
response.

Thanks.

----------------------------------------------------------------------

Comment By: Bruce (bp-sawyers)
Date: 2012-06-27 06:18

Message:
Jie Hua,

Is there an alternate way that the mellanox script could check what kernel
is being used in the genimage context?  It would be good to have a
suggested fix to give to mellanox.

----------------------------------------------------------------------

Comment By: jjh (jjhua)
Date: 2012-06-27 02:32

Message:
The mlnxofedinstall script  is from the Mellonax ofed iso. It is invoked
during genimage. I will contact with Jim about this problem. Maybe we need
Mellonax to change this script, and not use 'uname' in this script to
support diskless.  

Thanks.

----------------------------------------------------------------------

Comment By: Brian  Croswell (bcroswell)
Date: 2012-06-26 07:06

Message:
Jie Hua,      Can you look at this xCAT bug  that was opened for Melanox 
IB implementation  with RH6.2  Linux .  
Mark I changed the Group to xCAT 2.7.4   since we made xCAT 2.7.3  GA last
week.  If you need an Efix  we can work with Jie Hua . 

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=1006945&aid=3538093&group_id=208749