[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Qemu-devel] [PULL 22/30] nbd: Implement NBD_CMD_WRITE_ZEROES on server
From: |
Paolo Bonzini |
Subject: |
[Qemu-devel] [PULL 22/30] nbd: Implement NBD_CMD_WRITE_ZEROES on server |
Date: |
Tue, 1 Nov 2016 17:29:35 +0100 |
From: Eric Blake <address@hidden>
Upstream NBD protocol recently added the ability to efficiently
write zeroes without having to send the zeroes over the wire,
along with a flag to control whether the client wants to allow
a hole.
Note that when it comes to requiring full allocation, vs.
permitting optimizations, the NBD spec intentionally picked a
different sense for the flag; the rules in qemu are:
MAY_UNMAP == 0: must write zeroes
MAY_UNMAP == 1: may use holes if reads will see zeroes
while in NBD, the rules are:
FLAG_NO_HOLE == 1: must write zeroes
FLAG_NO_HOLE == 0: may use holes if reads will see zeroes
In all cases, the 'may use holes' scenario is optional (the
server need not use a hole, and must not use a hole if
subsequent reads would not see zeroes).
Signed-off-by: Eric Blake <address@hidden>
Message-Id: <address@hidden>
Signed-off-by: Paolo Bonzini <address@hidden>
---
include/block/nbd.h | 8 ++++++--
nbd/server.c | 42 ++++++++++++++++++++++++++++++++++++++++--
2 files changed, 46 insertions(+), 4 deletions(-)
diff --git a/include/block/nbd.h b/include/block/nbd.h
index eea7ef0..3e373f0 100644
--- a/include/block/nbd.h
+++ b/include/block/nbd.h
@@ -71,6 +71,7 @@ typedef struct NBDReply NBDReply;
#define NBD_FLAG_SEND_FUA (1 << 3) /* Send FUA (Force Unit
Access) */
#define NBD_FLAG_ROTATIONAL (1 << 4) /* Use elevator algorithm -
rotational media */
#define NBD_FLAG_SEND_TRIM (1 << 5) /* Send TRIM (discard) */
+#define NBD_FLAG_SEND_WRITE_ZEROES (1 << 6) /* Send WRITE_ZEROES */
/* New-style handshake (global) flags, sent from server to client, and
control what will happen during handshake phase. */
@@ -96,7 +97,8 @@ typedef struct NBDReply NBDReply;
#define NBD_REP_ERR_SHUTDOWN NBD_REP_ERR(7) /* Server shutting down */
/* Request flags, sent from client to server during transmission phase */
-#define NBD_CMD_FLAG_FUA (1 << 0)
+#define NBD_CMD_FLAG_FUA (1 << 0) /* 'force unit access' during write */
+#define NBD_CMD_FLAG_NO_HOLE (1 << 1) /* don't punch hole on zero run */
/* Supported request types */
enum {
@@ -104,7 +106,9 @@ enum {
NBD_CMD_WRITE = 1,
NBD_CMD_DISC = 2,
NBD_CMD_FLUSH = 3,
- NBD_CMD_TRIM = 4
+ NBD_CMD_TRIM = 4,
+ /* 5 reserved for failed experiment NBD_CMD_CACHE */
+ NBD_CMD_WRITE_ZEROES = 6,
};
#define NBD_DEFAULT_PORT 10809
diff --git a/nbd/server.c b/nbd/server.c
index 0b50caa..5b76261 100644
--- a/nbd/server.c
+++ b/nbd/server.c
@@ -616,7 +616,8 @@ static coroutine_fn int nbd_negotiate(NBDClientNewData
*data)
char buf[8 + 8 + 8 + 128];
int rc;
const uint16_t myflags = (NBD_FLAG_HAS_FLAGS | NBD_FLAG_SEND_TRIM |
- NBD_FLAG_SEND_FLUSH | NBD_FLAG_SEND_FUA);
+ NBD_FLAG_SEND_FLUSH | NBD_FLAG_SEND_FUA |
+ NBD_FLAG_SEND_WRITE_ZEROES);
bool oldStyle;
size_t len;
@@ -1146,11 +1147,17 @@ static ssize_t nbd_co_receive_request(NBDRequestData
*req,
rc = request->type == NBD_CMD_WRITE ? -ENOSPC : -EINVAL;
goto out;
}
- if (request->flags & ~NBD_CMD_FLAG_FUA) {
+ if (request->flags & ~(NBD_CMD_FLAG_FUA | NBD_CMD_FLAG_NO_HOLE)) {
LOG("unsupported flags (got 0x%x)", request->flags);
rc = -EINVAL;
goto out;
}
+ if (request->type != NBD_CMD_WRITE_ZEROES &&
+ (request->flags & NBD_CMD_FLAG_NO_HOLE)) {
+ LOG("unexpected flags (got 0x%x)", request->flags);
+ rc = -EINVAL;
+ goto out;
+ }
rc = 0;
@@ -1255,6 +1262,37 @@ static void nbd_trip(void *opaque)
}
break;
+ case NBD_CMD_WRITE_ZEROES:
+ TRACE("Request type is WRITE_ZEROES");
+
+ if (exp->nbdflags & NBD_FLAG_READ_ONLY) {
+ TRACE("Server is read-only, return error");
+ reply.error = EROFS;
+ goto error_reply;
+ }
+
+ TRACE("Writing to device");
+
+ flags = 0;
+ if (request.flags & NBD_CMD_FLAG_FUA) {
+ flags |= BDRV_REQ_FUA;
+ }
+ if (!(request.flags & NBD_CMD_FLAG_NO_HOLE)) {
+ flags |= BDRV_REQ_MAY_UNMAP;
+ }
+ ret = blk_pwrite_zeroes(exp->blk, request.from + exp->dev_offset,
+ request.len, flags);
+ if (ret < 0) {
+ LOG("writing to file failed");
+ reply.error = -ret;
+ goto error_reply;
+ }
+
+ if (nbd_co_send_reply(req, &reply, 0) < 0) {
+ goto out;
+ }
+ break;
+
case NBD_CMD_DISC:
/* unreachable, thanks to special case in nbd_co_receive_request() */
abort();
--
2.7.4
- [Qemu-devel] [PULL 11/30] nbd: Rename NbdClientSession to NBDClientSession, (continued)
- [Qemu-devel] [PULL 11/30] nbd: Rename NbdClientSession to NBDClientSession, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 14/30] nbd: Send message along with server NBD_REP_ERR errors, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 15/30] nbd: Share common option-sending code in client, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 18/30] nbd: Less allocation during NBD_OPT_LIST, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 17/30] nbd: Let client skip portions of server reply, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 16/30] nbd: Let server know when client gives up negotiation, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 12/30] nbd: Rename struct nbd_request and nbd_reply, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 19/30] nbd: Support shorter handshake, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 20/30] nbd: Refactor conversion to errno to silence checkpatch, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 21/30] nbd: Improve server handling of shutdown requests, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 22/30] nbd: Implement NBD_CMD_WRITE_ZEROES on server,
Paolo Bonzini <=
- [Qemu-devel] [PULL 23/30] nbd: Implement NBD_CMD_WRITE_ZEROES on client, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 24/30] qemu-char: do not forward events through the mux until QEMU has started, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 26/30] x86: add AVX512_4VNNIW and AVX512_4FMAPS features, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 25/30] slirp: fix CharDriver breakage, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 28/30] vl: exit qemu on guest panic if -no-shutdown is not set, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 27/30] checkpatch: allow spaces before parenthesis for 'coroutine_fn', Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 30/30] main-loop: Suppress I/O thread warning under qtest, Paolo Bonzini, 2016/11/01
- [Qemu-devel] [PULL 29/30] docs/rcu.txt: Fix minor typo, Paolo Bonzini, 2016/11/01
- Re: [Qemu-devel] [PULL v2 00/30] Misc patches for 2016-10-31, no-reply, 2016/11/01
- Re: [Qemu-devel] [PULL v2 00/30] Misc patches for 2016-10-31, Peter Maydell, 2016/11/01