| commit | 99722fe4d5a634707ced8d8f42b883b87a86b3c5 | |
| author | Chuck Lever <chuck.lever@oracle.com> | |
| Mon, 7 May 2018 19:28:25 +0000 (7 15:28 -0400) | ||
| committer | J. Bruce Fields <bfields@redhat.com> | |
| Fri, 11 May 2018 19:48:57 +0000 (11 15:48 -0400) | ||
| tree | bc89587516be20f4c18330e63c28f087d8749896 | treesnapshot (tar.gz zip) |
| parent | 3abb03facee06ea052be6e3a435f6dbb4e54fc04 | commitdiff |
svcrdma: Persistently allocate and DMA-map Send buffers
While sending each RPC Reply, svc_rdma_sendto allocates and DMA-
maps a separate buffer where the RPC/RDMA transport header is
constructed. The buffer is unmapped and released in the Send
completion handler. This is significant per-RPC overhead,
especially for small RPCs.
Instead, allocate and DMA-map a buffer, and cache it in each
svc_rdma_send_ctxt. This buffer and its mapping can be re-used
for each RPC, saving the cost of memory allocation and DMA
mapping.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
While sending each RPC Reply, svc_rdma_sendto allocates and DMA-
maps a separate buffer where the RPC/RDMA transport header is
constructed. The buffer is unmapped and released in the Send
completion handler. This is significant per-RPC overhead,
especially for small RPCs.
Instead, allocate and DMA-map a buffer, and cache it in each
svc_rdma_send_ctxt. This buffer and its mapping can be re-used
for each RPC, saving the cost of memory allocation and DMA
mapping.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>