]> git.openfabrics.org - ~shefty/libmlx4.git/commit
Handle new FW requirement for send request prefetching
authorRoland Dreier <rolandd@cisco.com>
Wed, 13 Jun 2007 17:31:16 +0000 (10:31 -0700)
committerRoland Dreier <rolandd@cisco.com>
Wed, 13 Jun 2007 17:31:16 +0000 (10:31 -0700)
commit561da8d10e419ffb333fe6faf05004d9a3670e7a
treeb2f3e824476c9cfc4e79d0510caae96a6e6a2cfa
parent8804cbd650b0cfc685a7cd19dc0cdce552746c90
Handle new FW requirement for send request prefetching

New ConnectX firmware introduces FW command interface revision 2,
which requires that for each QP, a chunk of send queue entries (the
"headroom") is kept marked as invalid, so that the HCA doesn't get
confused if it prefetches entries that haven't been posted yet.  Add
code to libmlx4 to do this.

Also, handle the new kernel ABI that adds the sq_no_prefetch parameter
to the create QP operation.  We just hard-code sq_no_prefetch to 0 and
always provide the full SQ headroom for now.

Based on a patch from Jack Morgenstein <jackm@dev.mellanox.co.il>.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
src/cq.c
src/mlx4-abi.h
src/mlx4.h
src/qp.c
src/verbs.c