From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:33721) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1eRpmL-0007Io-Oh for qemu-devel@nongnu.org; Wed, 20 Dec 2017 20:36:10 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1eRpmK-0001VH-RM for qemu-devel@nongnu.org; Wed, 20 Dec 2017 20:36:09 -0500 From: Eric Blake Date: Wed, 20 Dec 2017 19:35:58 -0600 Message-Id: <20171221013600.17648-3-eblake@redhat.com> In-Reply-To: <20171221013600.17648-1-eblake@redhat.com> References: <20171221013600.17648-1-eblake@redhat.com> Subject: [Qemu-devel] [PULL 2/4] nbd/server: Optimize final chunk of sparse read List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: Paolo Bonzini , "open list:Network Block Dev..." If we are careful to handle 0-length read requests correctly, we can optimize our sparse read to send the NBD_REPLY_FLAG_DONE bit on our last OFFSET_DATA or OFFSET_HOLE chunk rather than needing a separate chunk. Signed-off-by: Eric Blake Message-Id: <20171107030912.23930-3-eblake@redhat.com> Reviewed-by: Vladimir Sementsov-Ogievskiy --- nbd/server.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) diff --git a/nbd/server.c b/nbd/server.c index be7310cb41..e443b3cf5c 100644 --- a/nbd/server.c +++ b/nbd/server.c @@ -1339,12 +1339,14 @@ static int coroutine_fn nbd_co_send_sparse_read(NBDClient *client, offset + progress, size - progress, &pnum, NULL, NULL); + bool final; if (status < 0) { error_setg_errno(errp, -status, "unable to check for holes"); return status; } assert(pnum && pnum <= size - progress); + final = progress + pnum == size; if (status & BDRV_BLOCK_ZERO) { NBDStructuredReadHole chunk; struct iovec iov[] = { @@ -1353,7 +1355,8 @@ static int coroutine_fn nbd_co_send_sparse_read(NBDClient *client, trace_nbd_co_send_structured_read_hole(handle, offset + progress, pnum); - set_be_chunk(&chunk.h, 0, NBD_REPLY_TYPE_OFFSET_HOLE, + set_be_chunk(&chunk.h, final ? NBD_REPLY_FLAG_DONE : 0, + NBD_REPLY_TYPE_OFFSET_HOLE, handle, sizeof(chunk) - sizeof(chunk.h)); stq_be_p(&chunk.offset, offset + progress); stl_be_p(&chunk.length, pnum); @@ -1366,7 +1369,7 @@ static int coroutine_fn nbd_co_send_sparse_read(NBDClient *client, break; } ret = nbd_co_send_structured_read(client, handle, offset + progress, - data + progress, pnum, false, + data + progress, pnum, final, errp); } @@ -1375,9 +1378,6 @@ static int coroutine_fn nbd_co_send_sparse_read(NBDClient *client, } progress += pnum; } - if (!ret) { - ret = nbd_co_send_structured_done(client, handle, errp); - } return ret; } @@ -1542,7 +1542,8 @@ static coroutine_fn void nbd_trip(void *opaque) } } - if (client->structured_reply && !(request.flags & NBD_CMD_FLAG_DF)) { + if (client->structured_reply && !(request.flags & NBD_CMD_FLAG_DF) && + request.len) { ret = nbd_co_send_sparse_read(req->client, request.handle, request.from, req->data, request.len, &local_err); -- 2.14.3