[openib-general] Announce: preview RPMs for FC-4 and RHEL-4 available

Doug Ledford dledford at redhat.com
Thu Nov 17 17:09:23 PST 2005


Thomas Moschny wrote:
> On Thursday 17 November 2005 15:14, Doug Ledford wrote:
> 
>>Thomas Moschny wrote:
>>
>>>Unfortunately, we got an kernel-oops on ia64 (rhel4) ...
>>>The boot log is attached.
>>
>>I think I know what this is. [...]
>>The attached patch should be able to be dropped into the existing srpm
>>in place of the patch with the same name and a rebuild should then solve
>>the problem, although in the process of creating this patch I had to
>>move it from the 2700 section of the patch list down to the 10002
>>position because it touches things added after the infiniband code.
> 
> 
> The patch seems to work here, thanks. The machines are up now, and at least 
> IPoIB is working.
> 
> There seems to be a (minor?) problem with opensm -o, it aborts:
> 
> -------------------------------------------------
> OpenSM Rev:openib-1.1.0
> Command Line Arguments:
>  Run Once
>  Log File: /var/log/osm.log
> -------------------------------------------------
> OpenSM Rev:openib-1.1.0
> 
> Using default guid 0xxxxxxxxxxxxxxx
> Entering MASTER state
> 
> SUBNET UP
> 
> Exiting SM
> 
> *** glibc detected *** double free or corruption (!prev): 0x6000000000067970 
> ***
> Aborted

There is actually an init script for opensm that can be enabled on one 
machine in the subnet (I suppose you could do more if you assigned 
priorities to the machines).  It seems to run fine, but issues this same 
message on shutdown.  So, at least on x86_64, that much is similar, 
opensm issues this warning on shutdown.

> Subsequent runs of opensm hang in flush_cpu_workqueue or 
> rwsem_down_failed_common.

However, I don't see this on x86_64.


-- 
Doug Ledford <dledford at redhat.com>
http://people.redhat.com/dledford




More information about the general mailing list