Rather than open-coding our own read/write-all functions, we
can make use of the recently-added qio code. It slightly
changes the error message in one of the iotests.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20170905191114.5959-4-eblake@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Add nbd_co_request, to remove code duplications in
nbd_client_co_{pwrite,pread,...} functions. Also this is
needed for further refactoring.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20170804151440.320927-8-vsementsov@virtuozzo.com>
[eblake: make nbd_co_request a wrapper, rather than merging two
existing functions]
Signed-off-by: Eric Blake <eblake@redhat.com>
Rename nbd_recv_coroutines_enter_all to nbd_recv_coroutines_wake_all,
as it most probably just adds all recv coroutines into co_queue_wakeup,
rather than directly enter them.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20170804151440.320927-9-vsementsov@virtuozzo.com>
[eblake: tweak commit message]
Signed-off-by: Eric Blake <eblake@redhat.com>
Use int variable for nbd_co_send_request return value (as
nbd_co_send_request returns int).
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20170804151440.320927-6-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
The following segfault is encountered if the NBD server closes the UNIX
domain socket immediately after negotiation:
Program terminated with signal SIGSEGV, Segmentation fault.
#0 aio_co_schedule (ctx=0x0, co=0xd3c0ff2ef0) at util/async.c:441
441 QSLIST_INSERT_HEAD_ATOMIC(&ctx->scheduled_coroutines,
(gdb) bt
#0 0x000000d3c01a50f8 in aio_co_schedule (ctx=0x0, co=0xd3c0ff2ef0) at util/async.c:441
#1 0x000000d3c012fa90 in nbd_coroutine_end (bs=bs@entry=0xd3c0fec650, request=<optimized out>) at block/nbd-client.c:207
#2 0x000000d3c012fb58 in nbd_client_co_preadv (bs=0xd3c0fec650, offset=0, bytes=<optimized out>, qiov=0x7ffc10a91b20, flags=0) at block/nbd-client.c:237
#3 0x000000d3c0128e63 in bdrv_driver_preadv (bs=bs@entry=0xd3c0fec650, offset=offset@entry=0, bytes=bytes@entry=512, qiov=qiov@entry=0x7ffc10a91b20, flags=0) at block/io.c:836
#4 0x000000d3c012c3e0 in bdrv_aligned_preadv (child=child@entry=0xd3c0ff51d0, req=req@entry=0x7f31885d6e90, offset=offset@entry=0, bytes=bytes@entry=512, align=align@entry=1, qiov=qiov@entry=0x7ffc10a91b20, f
+lags=0) at block/io.c:1086
#5 0x000000d3c012c6b8 in bdrv_co_preadv (child=0xd3c0ff51d0, offset=offset@entry=0, bytes=bytes@entry=512, qiov=qiov@entry=0x7ffc10a91b20, flags=flags@entry=0) at block/io.c:1182
#6 0x000000d3c011cc17 in blk_co_preadv (blk=0xd3c0ff4f80, offset=0, bytes=512, qiov=0x7ffc10a91b20, flags=0) at block/block-backend.c:1032
#7 0x000000d3c011ccec in blk_read_entry (opaque=0x7ffc10a91b40) at block/block-backend.c:1079
#8 0x000000d3c01bbb96 in coroutine_trampoline (i0=<optimized out>, i1=<optimized out>) at util/coroutine-ucontext.c:79
#9 0x00007f3196cb8600 in __start_context () at /lib64/libc.so.6
The problem is that nbd_client_init() uses
nbd_client_attach_aio_context() -> aio_co_schedule(new_context,
client->read_reply_co). Execution of read_reply_co is deferred to a BH
which doesn't run until later.
In the mean time blk_co_preadv() can be called and nbd_coroutine_end()
calls aio_wake() on read_reply_co. At this point in time
read_reply_co's ctx isn't set because it has never been entered yet.
This patch simplifies the nbd_co_send_request() ->
nbd_co_receive_reply() -> nbd_coroutine_end() lifecycle to just
nbd_co_send_request() -> nbd_co_receive_reply(). The request is "ended"
if an error occurs at any point. Callers no longer have to invoke
nbd_coroutine_end().
This cleanup also eliminates the segfault because we don't call
aio_co_schedule() to wake up s->read_reply_co if sending the request
failed. It is only necessary to wake up s->read_reply_co if a reply was
received.
Note this only happens with UNIX domain sockets on Linux. It doesn't
seem possible to reproduce this with TCP sockets.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20170829122745.14309-2-stefanha@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
The following scenario leads to an assertion failure in
qio_channel_yield():
1. Request coroutine calls qio_channel_yield() successfully when sending
would block on the socket. It is now yielded.
2. nbd_read_reply_entry() calls nbd_recv_coroutines_enter_all() because
nbd_receive_reply() failed.
3. Request coroutine is entered and returns from qio_channel_yield().
Note that the socket fd handler has not fired yet so
ioc->write_coroutine is still set.
4. Request coroutine attempts to send the request body with nbd_rwv()
but the socket would still block. qio_channel_yield() is called
again and assert(!ioc->write_coroutine) is hit.
The problem is that nbd_read_reply_entry() does not distinguish between
request coroutines that are waiting to receive a reply and those that
are not.
This patch adds a per-request bool receiving flag so
nbd_read_reply_entry() can avoid spurious aio_wake() calls.
Reported-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20170822125113.5025-1-stefanha@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Tested-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
travis builds fail at HEAD at rc3 master with
block/nbd-client.c: In function ‘nbd_read_reply_entry’:
block/nbd-client.c:110:8: error: ‘ret’ may be used uninitialized in this function [-Werror=uninitialized]
fix it by initializing 'ret' to 0
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
When we switched NBD to use coroutines for qemu 2.9 (in particular,
commit a12a712a), we introduced a regression: if a server sends us
garbage (such as a corrupted magic number), we quit the read loop
but do not stop sending further queued commands, resulting in the
client hanging when it never reads the response to those additional
commands. In qemu 2.8, we properly detected that the server is no
longer reliable, and cancelled all existing pending commands with
EIO, then tore down the socket so that all further command attempts
get EPIPE.
Restore the proper behavior of quitting (almost) all communication
with a broken server: Once we know we are out of sync or otherwise
can't trust the server, we must assume that any further incoming
data is unreliable and therefore end all pending commands with EIO,
and quit trying to send any further commands. As an exception, we
still (try to) send NBD_CMD_DISC to let the server know we are going
away (in part, because it is easier to do that than to further
refactor nbd_teardown_connection, and in part because it is the
only command where we do not have to wait for a reply).
Based on a patch by Vladimir Sementsov-Ogievskiy.
A malicious server can be created with the following hack,
followed by setting NBD_SERVER_DEBUG to a non-zero value in the
environment when running qemu-nbd:
| --- a/nbd/server.c
| +++ b/nbd/server.c
| @@ -919,6 +919,17 @@ static int nbd_send_reply(QIOChannel *ioc, NBDReply *reply, Error **errp)
| stl_be_p(buf + 4, reply->error);
| stq_be_p(buf + 8, reply->handle);
|
| + static int debug;
| + static int count;
| + if (!count++) {
| + const char *str = getenv("NBD_SERVER_DEBUG");
| + if (str) {
| + debug = atoi(str);
| + }
| + }
| + if (debug && !(count % debug)) {
| + buf[0] = 0;
| + }
| return nbd_write(ioc, buf, sizeof(buf), errp);
| }
Reported-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20170814213426.24681-1-eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
The upstream NBD Protocol has defined a new extension to allow
the server to advertise block sizes to the client, as well as
a way for the client to inform the server whether it intends to
obey block sizes.
When using the block layer as the client, we will obey block
sizes; but when used as 'qemu-nbd -c' to hand off to the
kernel nbd module as the client, we are still waiting for the
kernel to implement a way for us to learn if it will honor
block sizes (perhaps by an addition to sysfs, rather than an
ioctl), as well as any way to tell the kernel what additional
block sizes to obey (NBD_SET_BLKSIZE appears to be accurate
for the minimum size, but preferred and maximum sizes would
probably be new ioctl()s), so until then, we need to make our
request for block sizes conditional.
When using ioctl(NBD_SET_BLKSIZE) to hand off to the kernel,
use the minimum block size as the sector size if it is larger
than 512, which also has the nice effect of cooperating with
(non-qemu) servers that don't do read-modify-write when
exposing a block device with 4k sectors; it might also allow
us to visit a file larger than 2T on a 32-bit kernel.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20170707203049.534-10-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The NBD Protocol is introducing some additional information
about exports, such as minimum request size and alignment, as
well as an advertised maximum request size. It will be easier
to feed this information back to the block layer if we gather
all the information into a struct, rather than adding yet more
pointer parameters during negotiation.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <20170707203049.534-2-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When attaching the NBD QIOChannel to an AioContext, the TLS channel should
be used, not the underlying socket channel. This is because, trivially,
the TLS channel will be the one that we read/write to and thus the one
that will get the qio_channel_yield() call.
Fixes: ff82911cd3
Cc: qemu-stable@nongnu.org
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Tested-by: Daniel P. Berrange <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Change the 'int count' parameter in *pwrite_zeros, *pdiscard related
functions (and some others) to 'int bytes', as they both refer to bytes.
This helps with code legibility.
Signed-off-by: Manos Pitsidianakis <el13635@mail.ntua.gr>
Message-id: 20170609101808.13506-1-el13635@mail.ntua.gr
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Rename
nbd_wr_syncv -> nbd_rwv
read_sync -> nbd_read
read_sync_eof -> nbd_read_eof
write_sync -> nbd_write
drop_sync -> nbd_drop
1. nbd_ prefix
read_sync and write_sync are already shared, so it is good to have a
namespace prefix. drop_sync will be shared, and read_sync_eof is
related to read_sync, so let's rename them all.
2. _sync suffix
_sync is related to the fact that nbd_wr_syncv doesn't return if a
write to socket returns EAGAIN. The first implementation of
nbd_wr_syncv (was wr_sync in 7a5ca8648b) just loops while getting
EAGAIN, the current implementation yields in this case.
Why we want to get rid of it:
- it is normal for r/w functions to be synchronous, so having an
additional suffix for it looks redundant (contrariwise, we have
_aio suffix for async functions)
- _sync suffix in block layer is used when function does flush (so
using it for other thing is confusing a bit)
- keep function names short after adding nbd_ prefix
3. for nbd_wr_syncv let's use more common notation 'rw'
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20170602150150.258222-2-vsementsov@virtuozzo.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
NBD is not thread safe, because it accesses s->in_flight without
a CoMutex. Fixing this will be required for multiqueue.
CoQueue doesn't have spurious wakeups but, when another coroutine can
run between qemu_co_queue_next's wakeup and qemu_co_queue_wait's
re-locking of the mutex, the wait condition can become false and
a loop is necessary.
In fact, it turns out that the loop is necessary even without this
multi-threaded scenario. A particular sequence of coroutine wakeups
is happening ~80% of the time when starting a guest with qcow2 image
served over NBD (i.e. qemu-nbd --format=raw, and QEMU's -drive option
has -format=qcow2). This patch fixes that issue too.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Move to modern errp scheme from just LOGging errors.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20170526110913.89098-1-vsementsov@virtuozzo.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Will be used in following patch to provide actual error message in
some cases.
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20170516094533.6160-4-vsementsov@virtuozzo.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
After the switch to reading replies in a coroutine, nothing is
reentering pending receive coroutines if the connection hangs.
Move nbd_recv_coroutines_enter_all to the reply read coroutine,
which is the place where hangups are detected. nbd_teardown_connection
can simply wait for the reply read coroutine to detect the hangup
and clean up after itself.
This wouldn't be enough though because nbd_receive_reply returns 0
(rather than -EPIPE or similar) when reading from a hung connection.
Fix the return value check in nbd_read_reply_entry.
This fixes qemu-iotests 083.
Reported-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 20170314111157.14464-1-pbonzini@redhat.com
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
All that CoQueue needs in order to become thread-safe is help
from an external mutex. Add this to the API.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Message-id: 20170213181244.16297-6-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
In the client, read the reply headers from a coroutine, switching the
read side between the "read header" coroutine and the I/O coroutine that
reads the body of the reply.
In the server, if the server can read more requests it will create a new
"read request" coroutine as soon as a request has been read. Otherwise,
the new coroutine is created in nbd_request_put.
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Daniel P. Berrange <berrange@redhat.com>
Message-id: 20170213135235.12274-8-pbonzini@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
The new AioPollFn io_poll() argument to aio_set_fd_handler() and
aio_set_event_handler() is used in the next patch.
Keep this code change separate due to the number of files it touches.
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 20161201192652.9509-3-stefanha@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Commit fa778fff wired up support to send the NBD_CMD_WRITE_ZEROES,
but forgot to inform the block layer that FUA unmapping of zeroes is
supported. Without BDRV_REQ_MAY_UNMAP listed as a supported flag,
the block layer will always insist on the NBD layer passing
NBD_CMD_FLAG_NO_HOLE, resulting in the server always allocating
things even when it was desired to let the server punch holes.
Similarly, failing to set BDRV_REQ_FUA means that the client may
send unnecessary NBD_CMD_FLUSH when it could have instead used the
NBD_CMD_FLAG_FUA bit.
CC: qemu-stable@nongnu.org
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1479413642-22463-2-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
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 a hole.
The generic block code takes care of falling back to the obvious
write of lots of zeroes if we return -ENOTSUP because the server
does not have WRITE_ZEROES.
Ideally, since NBD_CMD_WRITE_ZEROES does not involve any data
over the wire, we want to support transactions that are much
larger than the normal 32M limit imposed on NBD_CMD_WRITE. But
the server may still have a limit smaller than UINT_MAX, so
until experimental NBD protocol additions for advertising various
command sizes is finalized (see [1], [2]), for now we just stick to
the same limits as normal writes.
[1] https://github.com/yoe/nbd/blob/extension-info/doc/proto.md
[2] https://sourceforge.net/p/nbd/mailman/message/35081223/
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1476469998-28592-17-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Our coding convention prefers CamelCase names, and we already
have other existing structs with NBDFoo naming. Let's be
consistent, before later patches add even more structs.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1476469998-28592-6-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
It's better to use consistent capitalization of the namespace
used for NBD functions; we have more instances of NBD* than
Nbd*.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1476469998-28592-5-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Current upstream NBD documents that requests have a 16-bit flags,
followed by a 16-bit type integer; although older versions mentioned
only a 32-bit field with masking to find flags. Since the protocol
is in network order (big-endian over the wire), the ABI is unchanged;
but dealing with the flags as a separate field rather than masking
will make it easier to add support for upcoming NBD extensions that
increase the number of both flags and commands.
Improve some comments in nbd.h based on the current upstream
NBD protocol (https://github.com/yoe/nbd/blob/master/doc/proto.md),
and touch some nearby code to keep checkpatch.pl happy.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1476469998-28592-3-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
NBD is using the CoMutex in a way that wasn't anticipated. For example, if there are
N(N=26, MAX_NBD_REQUESTS=16) nbd write requests, so we will invoke nbd_client_co_pwritev
N times.
----------------------------------------------------------------------------------------
time request Actions
1 1 in_flight=1, Coroutine=C1
2 2 in_flight=2, Coroutine=C2
...
15 15 in_flight=15, Coroutine=C15
16 16 in_flight=16, Coroutine=C16, free_sema->holder=C16, mutex->locked=true
17 17 in_flight=16, Coroutine=C17, queue C17 into free_sema->queue
18 18 in_flight=16, Coroutine=C18, queue C18 into free_sema->queue
...
26 N in_flight=16, Coroutine=C26, queue C26 into free_sema->queue
----------------------------------------------------------------------------------------
Once nbd client recieves request No.16' reply, we will re-enter C16. It's ok, because
it's equal to 'free_sema->holder'.
----------------------------------------------------------------------------------------
time request Actions
27 16 in_flight=15, Coroutine=C16, free_sema->holder=C16, mutex->locked=false
----------------------------------------------------------------------------------------
Then nbd_coroutine_end invokes qemu_co_mutex_unlock what will pop coroutines from
free_sema->queue's head and enter C17. More free_sema->holder is C17 now.
----------------------------------------------------------------------------------------
time request Actions
28 17 in_flight=16, Coroutine=C17, free_sema->holder=C17, mutex->locked=true
----------------------------------------------------------------------------------------
In above scenario, we only recieves request No.16' reply. As time goes by, nbd client will
almostly recieves replies from requests 1 to 15 rather than request 17 who owns C17. In this
case, we will encounter assert "mutex->holder == self" failed since Kevin's commit 0e438cdc
"coroutine: Let CoMutex remember who holds it". For example, if nbd client recieves request
No.15' reply, qemu will stop unexpectedly:
----------------------------------------------------------------------------------------
time request Actions
29 15(most case) in_flight=15, Coroutine=C15, free_sema->holder=C17, mutex->locked=false
----------------------------------------------------------------------------------------
Per Paolo's suggestion "The simplest fix is to change it to CoQueue, which is like a condition
variable", this patch replaces CoMutex with CoQueue.
Cc: Wen Congyang <wency@cn.fujitsu.com>
Reported-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Changlong Xie <xiecl.fnst@cn.fujitsu.com>
Message-Id: <1476267508-19499-1-git-send-email-xiecl.fnst@cn.fujitsu.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The NBD protocol doesn't have any notion of sectors, so it is
a fairly easy conversion to use byte-based read and write.
Signed-off-by: Eric Blake <eblake@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1468624988-423-19-git-send-email-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Another step towards killing off sector-based block APIs.
While at it, call directly into nbd-client.c instead of having
a pointless trivial wrapper in nbd.c.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 1468624988-423-14-git-send-email-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Now that NBD relies on the block layer to fragment things, we no
longer need to track an offset argument for which fragment of
a request we are actually servicing.
While at it, use true and false instead of 0 and 1 for a bool
parameter.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1468607524-19021-6-git-send-email-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Now that the block layer will honor max_transfer, we can simplify
our code to rely on that guarantee.
The readv code can call directly into nbd-client, just as the
writev code has done since commit 52a4650.
Interestingly enough, while qemu-io 'w 0 40m' splits into a 32M
and 8M transaction, 'w -z 0 40m' splits into two 16M and an 8M,
because the block layer caps the bounce buffer for writing zeroes
at 16M. When we later introduce support for NBD_CMD_WRITE_ZEROES,
we can get a full 32M zero write (or larger, if the client and
server negotiate that write zeroes can use a larger size than
ordinary writes).
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1468607524-19021-5-git-send-email-eblake@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
In practice the entry argument is always known at creation time, and
it is confusing that sometimes qemu_coroutine_enter is used with a
non-NULL argument to re-enter a coroutine (this happens in
block/sheepdog.c and tests/test-coroutine.c). So pass the opaque value
at creation time, for consistency with e.g. aio_bh_new.
Mostly done with the following semantic patch:
@ entry1 @
expression entry, arg, co;
@@
- co = qemu_coroutine_create(entry);
+ co = qemu_coroutine_create(entry, arg);
...
- qemu_coroutine_enter(co, arg);
+ qemu_coroutine_enter(co);
@ entry2 @
expression entry, arg;
identifier co;
@@
- Coroutine *co = qemu_coroutine_create(entry);
+ Coroutine *co = qemu_coroutine_create(entry, arg);
...
- qemu_coroutine_enter(co, arg);
+ qemu_coroutine_enter(co);
@ entry3 @
expression entry, arg;
@@
- qemu_coroutine_enter(qemu_coroutine_create(entry), arg);
+ qemu_coroutine_enter(qemu_coroutine_create(entry, arg));
@ reentry @
expression co;
@@
- qemu_coroutine_enter(co, NULL);
+ qemu_coroutine_enter(co);
except for the aforementioned few places where the semantic patch
stumbled (as expected) and for test_co_queue, which would otherwise
produce an uninitialized variable warning.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The NBD layer was breaking up request at a limit of 2040 sectors
(just under 1M) to cater to old qemu-nbd. But the server limit
was raised to 32M in commit 2d8214885 to match the kernel, more
than three years ago; and the upstream NBD Protocol is proposing
documentation that without any explicit communication to state
otherwise, a client should be able to safely assume that a 32M
transaction will work. It is time to rely on the larger sizing,
and any downstream distro that cares about maximum
interoperability to older qemu-nbd servers can just tweak the
value of #define NBD_MAX_SECTORS.
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: qemu-stable@nongnu.org
Reviewed-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Now that the block layer honors per-bds FUA support, we don't
have to duplicate the fallback flush at the NBD layer. The
static function nbd_co_writev_flags() is no longer needed, and
the driver can just directly use nbd_client_co_writev().
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Pre-patch, .supported_write_flags lives at the driver level, which
means we are blindly declaring that all block devices using a
given driver will either equally support FUA, or that we need a
fallback at the block layer. But there are drivers where FUA
support is a per-block decision: the NBD block driver is dependent
on the remote server advertising NBD_FLAG_SEND_FUA (and has
fallback code to duplicate the flush that the block layer would do
if NBD had not set .supported_write_flags); and the iscsi block
driver is dependent on the mode sense bits advertised by the
underlying device (and is currently silently ignoring FUA requests
if the underlying device does not support FUA).
The fix is to make supported flags as a per-BDS option, set during
.bdrv_open(). This patch moves the variable and fixes NBD and iscsi
to set it only conditionally; later patches will then further
simplify the NBD driver to quit duplicating work done at the block
layer, as well as tackle the fact that SCSI does not support FUA
semantics on WRITESAME(10/16) but only on WRITE(10/16).
Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
The NBD protocol does not clearly document what will happen
if a client sends NBD_CMD_FLAG_FUA on NBD_CMD_FLUSH.
Historically, both the qemu and upstream NBD servers silently
ignored that flag, but that feels a bit risky. Meanwhile, the
qemu NBD client unconditionally sends the flag (without even
bothering to check whether the caller cares; at least with
NBD_CMD_WRITE the client only sends FUA if requested by a
higher layer).
There is ongoing discussion on the NBD list to fix the
protocol documentation to require that the server MUST ignore
the flag (unless the kernel folks can better explain what FUA
means for a flush), but until those doc improvements land, the
current nbd.git master was recently changed to reject the flag
with EINVAL (see nbd commit ab22e082), which now makes it
impossible for a qemu client to use FLUSH with an upstream NBD
server.
We should not send FUA with flush unless the upstream protocol
documents what it will do, and even then, it should be something
that the caller can opt into, rather than being unconditional.
Signed-off-by: Eric Blake <eblake@redhat.com>
Message-Id: <1459526902-32561-1-git-send-email-eblake@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The NBD server already used to send a FUA flag when the writethrough
mode was set. This code was a remnant from the times where protocol
drivers actually had to implement writethrough modes. Since nowadays the
block layer sends flushes in writethrough mode and non-root nodes are
always writeback, this was mostly dead code - only mostly because if NBD
was configured to be used without a format, we sent _both_ FUA and an
explicit flush afterwards, which makes the code not technically dead,
but useless overhead.
This patch changes the code so that the block layer's FUA flag is
recognised and translated into a NBD FUA flag. The additional flush is
avoided now.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
This modifies the NBD driver so that it is possible to request
use of TLS. This is done by providing the 'tls-creds' parameter
with the ID of a previously created QCryptoTLSCreds object.
For example
$QEMU -object tls-creds-x509,id=tls0,endpoint=client,\
dir=/home/berrange/security/qemutls \
-drive driver=nbd,host=localhost,port=9000,tls-creds=tls0
The client will drop the connection if the NBD server does not
provide TLS.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <1455129674-17255-15-git-send-email-berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This extends the NBD protocol handling code so that it is capable
of negotiating TLS support during the connection setup. This involves
requesting the STARTTLS protocol option before any other NBD options.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <1455129674-17255-14-git-send-email-berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Now that all callers are converted to use I/O channels for
initial connection setup, it is possible to switch the core
NBD protocol handling core over to use QIOChannel APIs for
actual sockets I/O.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <1455129674-17255-7-git-send-email-berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This converts the NBD block driver client to use the QIOChannelSocket
class for initial connection setup. The NbdClientSession struct has
two pointers, one to the master QIOChannelSocket providing the raw
data channel, and one to a QIOChannel which is the current channel
used for I/O. Initially the two point to the same object, but when
TLS support is added, they will point to different objects.
The qemu-img & qemu-io tools now need to use MODULE_INIT_QOM to
ensure the QIOChannel object classes are registered. The qemu-nbd
tool already did this.
In this initial conversion though, all I/O is still actually done
using the raw POSIX sockets APIs.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Message-Id: <1455129674-17255-4-git-send-email-berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Clean up includes so that osdep.h is included first and headers
which it implies are not included manually.
This commit was created with scripts/clean-includes.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
All callers pass in false, and the real external ones will switch to
true in coming patches.
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Message-Id: <1424887718-10800-13-git-send-email-mreitz@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
When we tested the VM migartion between different hosts with NBD
devices, we found if we sent a cancel command after the drive_mirror
was just started, a coroutine re-enter error would occur. The stack
was as follow:
(gdb) bt
00) 0x00007fdfc744d885 in raise () from /lib64/libc.so.6
01) 0x00007fdfc744ee61 in abort () from /lib64/libc.so.6
02) 0x00007fdfca467cc5 in qemu_coroutine_enter (co=0x7fdfcaedb400, opaque=0x0)
at qemu-coroutine.c:118
03) 0x00007fdfca467f6c in qemu_co_queue_run_restart (co=0x7fdfcaedb400) at
qemu-coroutine-lock.c:59
04) 0x00007fdfca467be5 in coroutine_swap (from=0x7fdfcaf3c4e8,
to=0x7fdfcaedb400) at qemu-coroutine.c:96
05) 0x00007fdfca467cea in qemu_coroutine_enter (co=0x7fdfcaedb400, opaque=0x0)
at qemu-coroutine.c:123
06) 0x00007fdfca467f6c in qemu_co_queue_run_restart (co=0x7fdfcaedbdc0) at
qemu-coroutine-lock.c:59
07) 0x00007fdfca467be5 in coroutine_swap (from=0x7fdfcaf3c4e8,
to=0x7fdfcaedbdc0) at qemu-coroutine.c:96
08) 0x00007fdfca467cea in qemu_coroutine_enter (co=0x7fdfcaedbdc0, opaque=0x0)
at qemu-coroutine.c:123
09) 0x00007fdfca4a1fa4 in nbd_recv_coroutines_enter_all (s=0x7fdfcaef7dd0) at
block/nbd-client.c:41
10) 0x00007fdfca4a1ff9 in nbd_teardown_connection (client=0x7fdfcaef7dd0) at
block/nbd-client.c:50
11) 0x00007fdfca4a20f0 in nbd_reply_ready (opaque=0x7fdfcaef7dd0) at
block/nbd-client.c:92
12) 0x00007fdfca45ed80 in aio_dispatch (ctx=0x7fdfcae15e90) at aio-posix.c:144
13) 0x00007fdfca45ef1b in aio_poll (ctx=0x7fdfcae15e90, blocking=false) at
aio-posix.c:222
14) 0x00007fdfca448c34 in aio_ctx_dispatch (source=0x7fdfcae15e90, callback=0x0,
user_data=0x0) at async.c:212
15) 0x00007fdfc8f2f69a in g_main_context_dispatch () from
/usr/lib64/libglib-2.0.so.0
16) 0x00007fdfca45c391 in glib_pollfds_poll () at main-loop.c:190
17) 0x00007fdfca45c489 in os_host_main_loop_wait (timeout=1483677098) at
main-loop.c:235
18) 0x00007fdfca45c57b in main_loop_wait (nonblocking=0) at main-loop.c:484
19) 0x00007fdfca25f403 in main_loop () at vl.c:2249
20) 0x00007fdfca266fc2 in main (argc=42, argv=0x7ffff517d638,
envp=0x7ffff517d790) at vl.c:4814
We find the nbd_recv_coroutines_enter_all function (triggered by a cancel
command or a network connection breaking down) will enter a coroutine which
is waiting for the sending lock. If the lock is still held by another coroutine,
the entering coroutine will be added into the co_queue again. Latter, when the
lock is released, a coroutine re-enter error will occur.
This bug can be fixed simply by delaying the setting of recv_coroutine as
suggested by paolo. After applying this patch, we have tested the cancel
operation in mirror phase looply for more than 5 hous and everything is fine.
Without this patch, a coroutine re-enter error will occur in 5 minutes.
Signed-off-by: Bn Wu <wu.wubin@huawei.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1423552846-3896-1-git-send-email-wu.wubin@huawei.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Before this patch, the "opaque" pointer in an NBD BDS points to a
BDRVNBDState, which contains an NbdClientSession object, which in turn
contains a pointer to the BDS. This pointer may become invalid due to
bdrv_swap(), so drop it, and instead pass the BDS directly to the
nbd-client.c functions which then retrieve the NbdClientSession object
from there.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-id: 1423256778-3340-2-git-send-email-mreitz@redhat.com
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
This patch makes use of the Error object for nbd_receive_negotiate() so
that errors during negotiation look nicer.
Furthermore, this patch adds an additional error message if the received
magic was wrong, but would be correct for the other protocol version,
respectively: So if an export name was specified, but the NBD server
magic corresponds to an old handshake, this condition is explicitly
signaled to the user, and vice versa.
As these messages are now part of the "Could not open image" error
message, additional filtering has to be employed in iotest 083, which
this patch does as well.
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Drop the assumption that we're using the main AioContext. Convert
qemu_aio_set_fd_handler() calls to aio_set_fd_handler().
The .bdrv_detach/attach_aio_context() interfaces also need to be
implemented to move the socket fd handler from the old to the new
AioContext.
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
nbd_receive_reply() is called by the event loop whenever data is
available or the socket has been closed by the remote side.
This patch closes the socket when an error occurs to prevent the
nbd_receive_reply() handler from being called indefinitely after the
connection has failed.
Note that we were already correctly returning EIO for pending requests
but leaving the nbd_receive_reply() handler registered resulted in high
CPU consumption and a flood of error messages.
Reuse nbd_teardown_connection() to close the socket.
Reported-by: Zhifeng Cai <bluewindow@h3c.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Make sure all pending coroutines are finished when closing the session.
Signed-off-by: Marc-André Lureau <marcandre.lureau@gmail.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Eric Blake