]> err.no Git - linux-2.6/commit
IB/ipath: fix null deref during rdma ops
authorBryan O'Sullivan <bos@pathscale.com>
Tue, 23 May 2006 18:32:37 +0000 (11:32 -0700)
committerRoland Dreier <rolandd@cisco.com>
Tue, 23 May 2006 20:29:35 +0000 (13:29 -0700)
commit3977026462314dfbb237adf6a964d0f683b8e45d
treed9fbd218ac4a742371468a67db1a738f52af5a83
parent41c75a19bf4a0102f49763a686fb7e39780349f3
IB/ipath: fix null deref during rdma ops

The problem was that node A's sending thread, which handles sending RDMA
read response data, would write the trigger word, the last packet would
be sent, node B would send a new RDMA read request, node A's interrupt
handler would initialize s_rdma_sge, then node A's sending thread would
update s_rdma_sge.  This didn't happen very often naturally but was more
frequent with 1 byte RDMA reads.  Rather than adding more locking or
increasing the QP structure size and copying sge data, I modified the
copy routine to update the pointers before writing the trigger word to
avoid the update race.

Signed-off-by: Ralph Campbell <ralphc@pathscale.com>
Signed-off-by: Bryan O'Sullivan <bos@pathscale.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
drivers/infiniband/hw/ipath/ipath_layer.c