[ofa-general] libipath crash

[Robert Pearson]

Roland,

I may have found the wrong crash. I'll try to reproduce and send to qlogic.

Thanks,

Bob

-----Original Message-----
From: Roland Dreier [mailto:rdreier at cisco.com] 
Sent: Saturday, November 24, 2007 10:54 PM
To: Robert Pearson
Cc: openib-general at openib.org; 'Arthur Jones'
Subject: Re: [ofa-general] libipath crash

 > [root at client1 src]# uname -a
 > Linux client1 2.6.18-8.1.15.el5 #1 SMP Mon Oct 22 08:32:28 EDT 2007
x86_64 x86_64 x86_64 GNU/Linux

OK... your test case doesn't crash on my system and I don't have much
interest in trying to figure out what pile of backport patches OFED
has added on top of the already old kernel sources it started with,
but maybe someone at Qlogic does.

On the other hand, it's pretty odd that your stack trace seems to be
coming from the create QP path:

 > Nov 24 12:56:35 client1 kernel:  [<ffffffff8819ae1f>]
:ib_ipath:ipath_create_qp+0x171/0x5e2
 > Nov 24 12:56:35 client1 kernel:  [<ffffffff88426a34>]
:ib_uverbs:ib_uverbs_create_qp+0x251/0x467

but the test case you sent does only create CQ.  Are you positive that
the code you sent is sufficient to trigger the crash you see?

 - R.