[ofa-general] OFED 1.3 rc2 kernel panic: trying to enable SA cache

[Jack Morgenstein]

On Sunday 03 February 2008 12:28, Tziporet Koren wrote:
> please look at this issue with mthca
> 
Problem is that a change in sa_query.c (by Roland) was not reflected
in Sean's patches.

Roland's commit:
==========
commit 2aec5c602c6a44e2a3a173339a9ab94549658e4b
Author: Sean Hefty <sean.hefty at intel.com>
Date:   Mon Jun 18 11:03:58 2007 -0700

    IB/sa: Make sure SA queries use default P_Key

    MADs sent to the SA should use the the default P_Key (0x7fff/0xffff).
    There's no requirement that the default P_Key is stored at index 0 in
    the local P_Key table, so add code to the sa_query module to look up
    the index of the default P_Key when creating an address handle for the
    SA (which is done any time the P_Key table might change), and use this
    index for all SA queries.

    Signed-off-by: Sean Hefty <sean.hefty at intel.com>
    Signed-off-by: Roland Dreier <rolandd at cisco.com>
=========
moved initializing the ah in a MAD query from procedure send_mad (in file sa_query.c) to a
new procedure alloc_mad().

However, the informinfo query in patch "kernel_patches/fixes/sean_local_sa_1_notifications.patch"
was not updated as well to reflect the above change.  As a result, the ah field remained zeroed out
in the informinfo query; when, down the stack, mlx4_ib_post_send gets invoked, the Oops occurs.

This oversight probably occurred when we migrated patches from ofed 1.2.5 over to the ofed 1.3 tree.

I guess this is the first time anyone has tried to activate the local sa cache.

I'll post a fix today.

- Jack