As with previous performance optimization on Treaddir handling;
reduce the overall latency, i.e. overall time spent on processing
a Twalk request by reducing the amount of thread hops between the
9p server's main thread and fs worker thread(s).
In fact this patch even reduces the thread hops for Twalk handling
to its theoritical minimum of exactly 2 thread hops:
main thread -> fs worker thread -> main thread
This is achieved by doing all the required fs driver tasks altogether
in a single v9fs_co_run_in_worker({ ... }); code block.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <1a6701674afc4f08d40396e3aa2631e18a4dbb33.1622821729.git.qemu_oss@crudebyte.com>
There is no longer a user of root_qid, so drop it.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <6896dd161d3257db6b0513842a14f87ca191fdf6.1622821729.git.qemu_oss@crudebyte.com>
As we are actually only comparing the filesystem ID (i.e. device number
and inode number pair) let's use the POSIX stat buffer instead of QIDs,
because resolving QIDs requires to be done on 9p server's main thread
only as it might mutate the server state if inode remapping is enabled.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <26aa465ff9cc9c07e053331554a02fdae3994417.1622821729.git.qemu_oss@crudebyte.com>
There is only one user of fid_to_qid() which is v9fs_walk(). Let's
open-code fid_to_qid() directly within v9fs_walk(), because
fid_to_qid() hides the POSIX stat buffer which we are going to need
in the subsequent patch.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <e9a4c9c7a0792ed4db6578d105a0823ea05bc324.1622821729.git.qemu_oss@crudebyte.com>
We already capture the QID of the exported 9p root path, i.e. to
prevent client access outside the defined, exported filesystem's tree.
This is currently checked by comparing the root QID with another FID's
QID.
The problem with the latter is that resolving a QID of any given 9p path
can only be done on 9p server's main thread, that's because it might
mutate the server's state if inode remapping is enabled.
For that reason also capture the POSIX stat info of the root path for
being able to identify on any (e.g. worker) thread whether an
arbitrary given path is identical to the export root.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <eb07d6c2e9925788454cfe33d3802e4ffb23ea9a.1622821729.git.qemu_oss@crudebyte.com>
There is only one user of not_same_qid() which is v9fs_walk() and the
latter is using it for comparing a client supplied path with the 9p
export root path, for the sole purpose to prevent a Twalk request
from escaping from the exported 9p tree via "..".
However for that specific purpose the implementation of not_same_qid()
is wrong; if mtime of the 9p export root path changed between Tattach
and Twalk then not_same_qid() returns true when actually comparing
against the export root path.
To fix for the actual semantic being used, only compare QID path
members, but do not compare version or type members.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <ca0abae4a899d81c6e87f683732d6c1f56915232.1622821729.git.qemu_oss@crudebyte.com>
There is only one comparison between nwnames and P9_MAXWELEM required.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <E1liKiz-0006BC-Ja@lizzy.crudebyte.com>
To lower the entry level for new developers, add a link to the 9p
developer docs (i.e. qemu wiki) to MAINTAINERS and to the beginning of
9p source files, that is to: https://wiki.qemu.org/Documentation/9p
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Acked-by: Greg Kurz <groug@kaod.org>
Message-Id: <E1leeDf-0008GZ-9q@lizzy.crudebyte.com>
Commit e50caf4a5c ("tracing: convert documentation to rST")
converted docs/devel/tracing.txt to docs/devel/tracing.rst.
We still have several references to the old file, so let's fix them
with the following command:
sed -i s/tracing.txt/tracing.rst/ $(git grep -l docs/devel/tracing.txt)
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20210517151702.109066-2-sgarzare@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Replaced a call to qemu_mutex_lock and its respective call to
qemu_mutex_unlock and used QEMU_LOCK_GUARD macro in their place.
This simplifies the code by removing the call required to unlock
and also eliminates goto paths.
Signed-off-by: Mahmoud Mandour <ma.mandourr@gmail.com>
Acked-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <20210311031538.5325-9-ma.mandourr@gmail.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
There are 23 files that include the "sysemu/qtest.h",
but they do not use any qtest functions.
Signed-off-by: Chen Qun <kuhn.chenqun@huawei.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20210226081414.205946-1-kuhn.chenqun@huawei.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Use QSLIST instead of open-coding for a slightly improved readability.
No behavioral change.
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <20210122143514.215780-1-groug@kaod.org>
Signed-off-by: Greg Kurz <groug@kaod.org>
If a fid was actually re-opened by v9fs_reopen_fid(), we re-traverse the
fid list from the head in case some other request created a fid that
needs to be marked unreclaimable as well (i.e. the client opened a new
handle on the path that is being unlinked). This is suboptimal since
most if not all fids that require it have likely been taken care of
already.
This is mostly the result of new fids being added to the head of the
list. Since the list is now a QSIMPLEQ, add new fids at the end instead
to avoid the need to rewind. Take a reference on the fid to ensure it
doesn't go away during v9fs_reopen_fid() and that it can be safely
passed to QSIMPLEQ_NEXT() afterwards. Since the associated put_fid()
can also yield, same is done with the next fid. So the logic here is
to get a reference on a fid and only put it back during the next
iteration after we could get a reference on the next fid.
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <20210121181510.1459390-1-groug@kaod.org>
Signed-off-by: Greg Kurz <groug@kaod.org>
The fid_list is currently open-coded. This doesn't seem to serve any
purpose that cannot be met with QEMU's generic lists. Let's go for a
QSIMPLEQ : this will allow to add new fids at the end of the list and
to improve the logic in v9fs_mark_fids_unreclaim().
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <20210118142300.801516-3-groug@kaod.org>
Signed-off-by: Greg Kurz <groug@kaod.org>
This can only be 0 or 1.
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <20210118142300.801516-2-groug@kaod.org>
Signed-off-by: Greg Kurz <groug@kaod.org>
This should always successfully write exactly two 32-bit integers.
Make it clear with an assert(), like v9fs_receive_status() and
v9fs_receive_response() already do when unmarshalling the same
header.
Fixes: Coverity CID 1438968
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <161035859647.1221144.4691749806675653934.stgit@bahia.lan>
Signed-off-by: Greg Kurz <groug@kaod.org>
Depending on the client activity, the server can be asked to open a huge
number of file descriptors and eventually hit RLIMIT_NOFILE. This is
currently mitigated using a reclaim logic : the server closes the file
descriptors of idle fids, based on the assumption that it will be able
to re-open them later. This assumption doesn't hold of course if the
client requests the file to be unlinked. In this case, we loop on the
entire fid list and mark all related fids as unreclaimable (the reclaim
logic will just ignore them) and, of course, we open or re-open their
file descriptors if needed since we're about to unlink the file.
This is the purpose of v9fs_mark_fids_unreclaim(). Since the actual
opening of a file can cause the coroutine to yield, another client
request could possibly add a new fid that we may want to mark as
non-reclaimable as well. The loop is thus restarted if the re-open
request was actually transmitted to the backend. This is achieved
by keeping a reference on the first fid (head) before traversing
the list.
This is wrong in several ways:
- a potential clunk request from the client could tear the first
fid down and cause the reference to be stale. This leads to a
use-after-free error that can be detected with ASAN, using a
custom 9p client
- fids are added at the head of the list : restarting from the
previous head will always miss fids added by a some other
potential request
All these problems could be avoided if fids were being added at the
end of the list. This can be achieved with a QSIMPLEQ, but this is
probably too much change for a bug fix. For now let's keep it
simple and just restart the loop from the current head.
Fixes: CVE-2021-20181
Buglink: https://bugs.launchpad.net/qemu/+bug/1911666
Reported-by: Zero Day Initiative <zdi-disclosures@trendmicro.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
Message-Id: <161064025265.1838153.15185571283519390907.stgit@bahia.lan>
Signed-off-by: Greg Kurz <groug@kaod.org>
Commit b2c00bce54 ("meson: convert hw/9pfs, cleanup") introduced
CONFIG_9PFS (probably a wrong conflict resolution). This config is
not used anywhere. Backends depend on CONFIG_FSDEV_9P which itself
depends on CONFIG_VIRTFS.
Remove the invalid CONFIG_9PFS and use CONFIG_FSDEV_9P instead, to
fix the './configure --without-default-devices --enable-xen' build:
/usr/bin/ld: libcommon.fa.p/hw_xen_xen-legacy-backend.c.o: in function `xen_be_register_common':
hw/xen/xen-legacy-backend.c:754: undefined reference to `xen_9pfs_ops'
/usr/bin/ld: libcommon.fa.p/fsdev_qemu-fsdev.c.o:(.data.rel+0x8): undefined reference to `local_ops'
/usr/bin/ld: libcommon.fa.p/fsdev_qemu-fsdev.c.o:(.data.rel+0x20): undefined reference to `synth_ops'
/usr/bin/ld: libcommon.fa.p/fsdev_qemu-fsdev.c.o:(.data.rel+0x38): undefined reference to `proxy_ops'
collect2: error: ld returned 1 exit status
Fixes: b2c00bce54 ("meson: convert hw/9pfs, cleanup")
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Acked-by: Greg Kurz <groug@kaod.org>
Tested-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Acked-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <20201104115706.3101190-3-philmd@redhat.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Fix code style. Space required before the open parenthesis '('.
Signed-off-by: Xinhao Zhang <zhangxinhao1@huawei.com>
Signed-off-by: Kai Deng <dengkai1@huawei.com>
Reported-by: Euler Robot <euler.robot@huawei.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <20201030043515.1030223-3-zhangxinhao1@huawei.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Fix code style. Open braces for struct should go on the same line.
Signed-off-by: Xinhao Zhang <zhangxinhao1@huawei.com>
Signed-off-by: Kai Deng <dengkai1@huawei.com>
Reported-by: Euler Robot <euler.robot@huawei.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <20201030043515.1030223-2-zhangxinhao1@huawei.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Don't trigger any performance warning if we're just running test cases,
because tests intentionally run for edge cases.
So far performance warnings were suppressed for the 'synth' fs driver
backend only. This patch suppresses them for all 9p fs driver backends.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <a2d2ff2163f8853ea782a7a1d4e6f2afd7c29ffe.1603106145.git.qemu_oss@crudebyte.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Previous patch introduced a performance warning being logged on host
side if client connected with an 'msize' <= 8192. Disable this
performance warning for the synth driver to prevent that warning from
being printed whenever the 9pfs (qtest) test cases are running.
Introduce a new export flag V9FS_NO_PERF_WARN for that purpose, which
might also be used to disable such warnings from the CLI in future.
We could have also prevented the warning by simply raising P9_MAX_SIZE
in virtio-9p-test.c to any value larger than 8192, however in the
context of test cases it makes sense running for edge cases, which
includes the lowest 'msize' value supported by the server which is
4096, hence we want to preserve an msize of 4096 for the test client.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <E1kEyDy-0006nN-5A@lizzy.crudebyte.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
It is essential to choose a reasonable high value for 'msize' to avoid
severely degraded file I/O performance. This parameter can only be
chosen on client/guest side, and a Linux client defaults to an 'msize'
of only 8192 if the user did not explicitly specify a value for 'msize',
which results in very poor file I/O performance.
Unfortunately many users are not aware that they should specify an
appropriate value for 'msize' to avoid severe performance issues, so
log a performance warning (with a QEMU wiki link explaining this issue
in detail) on host side in that case to make it more clear.
Currently a client cannot automatically pick a reasonable value for
'msize', because a good value for 'msize' depends on the file I/O
potential of the underlying storage on host side, i.e. a feature
invisible to the client, and even then a user would still need to trade
off between performance profit and additional RAM costs, i.e. with
growing 'msize' (RAM occupation), performance still increases, but
performance delta will shrink continuously.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <e6fc84845c95816ad5baecb0abd6bfefdcf7ec9f.1599144062.git.qemu_oss@crudebyte.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Some typedefs and macros are defined after the type check macros.
This makes it difficult to automatically replace their
definitions with OBJECT_DECLARE_TYPE.
Patch generated using:
$ ./scripts/codeconverter/converter.py -i \
--pattern=QOMStructTypedefSplit $(git grep -l '' -- '*.[ch]')
which will split "typdef struct { ... } TypedefName"
declarations.
Followed by:
$ ./scripts/codeconverter/converter.py -i --pattern=MoveSymbols \
$(git grep -l '' -- '*.[ch]')
which will:
- move the typedefs and #defines above the type check macros
- add missing #include "qom/object.h" lines if necessary
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20200831210740.126168-9-ehabkost@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20200831210740.126168-10-ehabkost@redhat.com>
Message-Id: <20200831210740.126168-11-ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
hw/Makefile.objs is gone so there is more code that can be removed.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Meson doesn't enjoy the same flexibility we have with Make in choosing
the include path. In particular the tracing headers are using
$(build_root)/$(<D).
In order to keep the include directives unchanged,
the simplest solution is to generate headers with patterns like
"trace/trace-audio.h" and place forwarding headers in the source tree
such that for example "audio/trace.h" includes "trace/trace-audio.h".
This patch is too ugly to be applied to the Makefiles now. It's only
a way to separate the changes to the tracing header files from the
Meson rewrite of the tracing logic.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
As we just fixed a severe performance issue with Treaddir request
handling, clarify this overall issue as a comment on
v9fs_co_run_in_worker() with the intention to hopefully prevent
such performance mistakes in future (and fixing other yet
outstanding ones).
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <4d34d332e1aaa8a2cf8dc0b5da4fd7727f2a86e8.1596012787.git.qemu_oss@crudebyte.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Previous patch suggests that it might make sense to use a different mutex
type now while handling readdir requests, depending on the precise
protocol variant, as v9fs_do_readdir_with_stat() (used by 9P2000.u) uses
a CoMutex to avoid deadlocks that might happen with QemuMutex otherwise,
whereas do_readdir_many() (used by 9P2000.L) should better use a
QemuMutex, as the precise behaviour of a failed CoMutex lock on fs driver
side would not be clear.
And to avoid the wrong lock type being used, be now strict and error out
if a 9P2000.L client sends a Tread on a directory, and likeweise error out
if a 9P2000.u client sends a Treaddir request.
This patch is just intended as transitional measure, as currently 9P2000.u
vs. 9P2000.L implementations currently differ where the main logic of
fetching directory entries is located at (9P2000.u still being more top
half focused, while 9P2000.L already being bottom half focused in regards
to fetching directory entries that is).
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <9a2ddc347e533b0d801866afd9dfac853d2d4106.1596012787.git.qemu_oss@crudebyte.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Make top half really top half and bottom half really bottom half:
Each T_readdir request handling is hopping between threads (main
I/O thread and background I/O driver threads) several times for
every individual directory entry, which sums up to huge latencies
for handling just a single T_readdir request.
Instead of doing that, collect now all required directory entries
(including all potentially required stat buffers for each entry) in
one rush on a background I/O thread from fs driver by calling the
previously added function v9fs_co_readdir_many() instead of
v9fs_co_readdir(), then assemble the entire resulting network
response message for the readdir request on main I/O thread. The
fs driver is still aborting the directory entry retrieval loop
(on the background I/O thread inside of v9fs_co_readdir_many())
as soon as it would exceed the client's requested maximum R_readdir
response size. So this will not introduce a performance penalty on
another end.
Also: No longer seek initial directory position in v9fs_readdir(),
as this is now handled (more consistently) by
v9fs_co_readdir_many() instead.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <c7c3d1cf4e86611538cef44897842819d9359d7a.1596012787.git.qemu_oss@crudebyte.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
The newly added function v9fs_co_readdir_many() retrieves multiple
directory entries with a single fs driver request. It is intended to
replace uses of v9fs_co_readdir(), the latter only retrieves a
single directory entry per fs driver request instead.
The reason for this planned replacement is that for every fs driver
request the coroutine is dispatched from main I/O thread to a
background I/O thread and eventually dispatched back to main I/O
thread. Hopping between threads adds latency. So if a 9pfs Treaddir
request reads a large amount of directory entries, this currently
sums up to huge latencies of several hundred ms or even more. So
using v9fs_co_readdir_many() instead of v9fs_co_readdir() will
provide significant performance improvements.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <73dc827a12ef577ae7e644dcf34a5c0e443ab42f.1596012787.git.qemu_oss@crudebyte.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
The implementation of v9fs_co_readdir() has two parts: the outer
part is executed by main I/O thread, whereas the inner part is
executed by fs driver on a background I/O thread.
Move the inner part to its own new, private function do_readdir(),
so it can be shared by another upcoming new function.
This is just a preparatory patch for the subsequent patch, with the
purpose to avoid the next patch to clutter the overall diff.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <a426ee06e77584fa2d8253ce5d8bea519eb3ffd4.1596012787.git.qemu_oss@crudebyte.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Rename function v9fs_readdir_data_size() -> v9fs_readdir_response_size()
and make it callable from other units. So far this function is only
used by 9p.c, however subsequent patches require the function to be
callable from another 9pfs unit. And as we're at it; also make it clear
for what this function is used for.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <3668ebc7d5b929a0e4f1357457060d96f50f76f4.1596012787.git.qemu_oss@crudebyte.com>
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
If we want to check error after errp-function call, we need to
introduce local_err and then propagate it to errp. Instead, use
the ERRP_GUARD() macro, benefits are:
1. No need of explicit error_propagate call
2. No need of explicit local_err variable: use errp directly
3. ERRP_GUARD() leaves errp as is if it's not NULL or
&error_fatal, this means that we don't break error_abort
(we'll abort on error_set, not on error_propagate)
If we want to add some info to errp (by error_prepend() or
error_append_hint()), we must use the ERRP_GUARD() macro.
Otherwise, this info will not be added when errp == &error_fatal
(the program will exit prior to the error_append_hint() or
error_prepend() call). Fix such a case in
v9fs_device_realize_common().
This commit is generated by command
sed -n '/^virtio-9p$/,/^$/{s/^F: //p}' MAINTAINERS | \
xargs git ls-files | grep '\.[hc]$' | \
xargs spatch \
--sp-file scripts/coccinelle/errp-guard.cocci \
--macro-file scripts/cocci-macro-file.h \
--in-place --no-show-diff --max-width 80
Reported-by: Kevin Wolf <kwolf@redhat.com>
Reported-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Acked-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
[Commit message tweaked]
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20200707165037.1026246-7-armbru@redhat.com>
[ERRP_AUTO_PROPAGATE() renamed to ERRP_GUARD(), and
auto-propagated-errp.cocci to errp-guard.cocci. Commit message
tweaked again.]
Receiving the error in a local variable only to free it is less clear
(and also less efficient) than passing NULL. Clean up.
Cc: Daniel P. Berrange <berrange@redhat.com>
Cc: Jerome Forissier <jerome@forissier.org>
CC: Greg Kurz <groug@kaod.org>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <20200630090351.1247703-4-armbru@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
The max order allowed by the protocol is 9. Increase the max order
supported by QEMU to 9 to increase performance.
Signed-off-by: Stefano Stabellini <stefano.stabellini@xilinx.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <20200521192627.15259-3-sstabellini@kernel.org>
Signed-off-by: Greg Kurz <groug@kaod.org>
Instead of truncating replies, which is problematic, wait until the
client reads more data and frees bytes on the reply ring.
Do that by calling qemu_coroutine_yield(). The corresponding
qemu_coroutine_enter_if_inactive() is called from xen_9pfs_bh upon
receiving the next notification from the client.
We need to be careful to avoid races in case xen_9pfs_bh and the
coroutine are both active at the same time. In xen_9pfs_bh, wait until
either the critical section is over (ring->co == NULL) or until the
coroutine becomes inactive (qemu_coroutine_yield() was called) before
continuing. Then, simply wake up the coroutine if it is inactive.
Signed-off-by: Stefano Stabellini <stefano.stabellini@xilinx.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <20200521192627.15259-2-sstabellini@kernel.org>
Signed-off-by: Greg Kurz <groug@kaod.org>
Locking was introduced in QEMU 2.7 to address the deprecation of
readdir_r(3) in glibc 2.24. It turns out that the frontend code is
the worst place to handle a critical section with a pthread mutex:
the code runs in a coroutine on behalf of the QEMU mainloop and then
yields control, waiting for the fsdev backend to process the request
in a worker thread. If the client resends another readdir request for
the same fid before the previous one finally unlocked the mutex, we're
deadlocked.
This never bit us because the linux client serializes readdir requests
for the same fid, but it is quite easy to demonstrate with a custom
client.
A good solution could be to narrow the critical section in the worker
thread code and to return a copy of the dirent to the frontend, but
this causes quite some changes in both 9p.c and codir.c. So, instead
of that, in order for people to easily backport the fix to older QEMU
versions, let's simply use a CoMutex since all the users for this
sit in coroutines.
Fixes: 7cde47d4a8 ("9p: add locking to V9fsDir")
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <158981894794.109297.3530035833368944254.stgit@bahia.lan>
Signed-off-by: Greg Kurz <groug@kaod.org>
linux/limits.h should be included for the XATTR_SIZE_MAX definition used
by v9fs_xattrcreate.
Fixes: 3b79ef2cf4 ("9pfs: limit xattr size in xattrcreate")
Signed-off-by: Dan Robertson <dan@dlrobertson.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <20200515203015.7090-2-dan@dlrobertson.com>
Signed-off-by: Greg Kurz <groug@kaod.org>
Devices may have component devices and buses.
Device realization may fail. Realization is recursive: a device's
realize() method realizes its components, and device_set_realized()
realizes its buses (which should in turn realize the devices on that
bus, except bus_set_realized() doesn't implement that, yet).
When realization of a component or bus fails, we need to roll back:
unrealize everything we realized so far. If any of these unrealizes
failed, the device would be left in an inconsistent state. Must not
happen.
device_set_realized() lets it happen: it ignores errors in the roll
back code starting at label child_realize_fail.
Since realization is recursive, unrealization must be recursive, too.
But how could a partly failed unrealize be rolled back? We'd have to
re-realize, which can fail. This design is fundamentally broken.
device_set_realized() does not roll back at all. Instead, it keeps
unrealizing, ignoring further errors.
It can screw up even for a device with no buses: if the lone
dc->unrealize() fails, it still unregisters vmstate, and calls
listeners' unrealize() callback.
bus_set_realized() does not roll back either. Instead, it stops
unrealizing.
Fortunately, no unrealize method can fail, as we'll see below.
To fix the design error, drop parameter @errp from all the unrealize
methods.
Any unrealize method that uses @errp now needs an update. This leads
us to unrealize() methods that can fail. Merely passing it to another
unrealize method cannot cause failure, though. Here are the ones that
do other things with @errp:
* virtio_serial_device_unrealize()
Fails when qbus_set_hotplug_handler() fails, but still does all the
other work. On failure, the device would stay realized with its
resources completely gone. Oops. Can't happen, because
qbus_set_hotplug_handler() can't actually fail here. Pass
&error_abort to qbus_set_hotplug_handler() instead.
* hw/ppc/spapr_drc.c's unrealize()
Fails when object_property_del() fails, but all the other work is
already done. On failure, the device would stay realized with its
vmstate registration gone. Oops. Can't happen, because
object_property_del() can't actually fail here. Pass &error_abort
to object_property_del() instead.
* spapr_phb_unrealize()
Fails and bails out when remove_drcs() fails, but other work is
already done. On failure, the device would stay realized with some
of its resources gone. Oops. remove_drcs() fails only when
chassis_from_bus()'s object_property_get_uint() fails, and it can't
here. Pass &error_abort to remove_drcs() instead.
Therefore, no unrealize method can fail before this patch.
device_set_realized()'s recursive unrealization via bus uses
object_property_set_bool(). Can't drop @errp there, so pass
&error_abort.
We similarly unrealize with object_property_set_bool() elsewhere,
always ignoring errors. Pass &error_abort instead.
Several unrealize methods no longer handle errors from other unrealize
methods: virtio_9p_device_unrealize(),
virtio_input_device_unrealize(), scsi_qdev_unrealize(), ...
Much of the deleted error handling looks wrong anyway.
One unrealize methods no longer ignore such errors:
usb_ehci_pci_exit().
Several realize methods no longer ignore errors when rolling back:
v9fs_device_realize_common(), pci_qdev_unrealize(),
spapr_phb_realize(), usb_qdev_realize(), vfio_ccw_realize(),
virtio_device_realize().
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20200505152926.18877-17-armbru@redhat.com>
If delivery of some 9pfs response fails for some reason, log the
error message by mentioning the 9P protocol reply type, not by
client's request type. The latter could be misleading that the
error occurred already when handling the request input.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Acked-by: Stefano Stabellini <sstabellini@kernel.org>
Message-Id: <ad0e5a9b6abde52502aa40b30661d29aebe1590a.1589132512.git.qemu_oss@crudebyte.com>
Signed-off-by: Greg Kurz <groug@kaod.org>
QEMU's local 9pfs server passes through O_NOATIME from the client. If
the QEMU process doesn't have permissions to use O_NOATIME (namely, it
does not own the file nor have the CAP_FOWNER capability), the open will
fail. This causes issues when from the client's point of view, it
believes it has permissions to use O_NOATIME (e.g., a process running as
root in the virtual machine). Additionally, overlayfs on Linux opens
files on the lower layer using O_NOATIME, so in this case a 9pfs mount
can't be used as a lower layer for overlayfs (cf.
dabfe19719/vmtest/onoatimehack.c
and https://github.com/NixOS/nixpkgs/issues/54509).
Luckily, O_NOATIME is effectively a hint, and is often ignored by, e.g.,
network filesystems. open(2) notes that O_NOATIME "may not be effective
on all filesystems. One example is NFS, where the server maintains the
access time." This means that we can honor it when possible but fall
back to ignoring it.
Acked-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Signed-off-by: Omar Sandoval <osandov@fb.com>
Message-Id: <e9bee604e8df528584693a4ec474ded6295ce8ad.1587149256.git.osandov@fb.com>
Signed-off-by: Greg Kurz <groug@kaod.org>
The common fsdev options are set by qemu_fsdev_add() before it calls
the backend specific option parsing code. In the case of "proxy" this
means "writeout" or "readonly" were simply ignored. This has been
broken from the beginning.
Reported-by: Stéphane Graber <stgraber@ubuntu.com>
Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-Id: <158349633705.1237488.8895481990204796135.stgit@bahia.lan>
This will provide the following virtual files by the 9pfs
synth driver:
- /ReadDirDir/ReadDirFile99
- /ReadDirDir/ReadDirFile98
...
- /ReadDirDir/ReadDirFile1
- /ReadDirDir/ReadDirFile0
This virtual directory and its virtual 100 files will be
used by the upcoming 9pfs readdir tests.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <5408c28c8de25dd575b745cef63bf785305ccef2.1579567020.git.qemu_oss@crudebyte.com>
Signed-off-by: Greg Kurz <groug@kaod.org>
A good 9p client sends T_readdir with "count" parameter that's sufficiently
smaller than client's initially negotiated msize (maximum message size).
We perform a check for that though to avoid the server to be interrupted
with a "Failed to encode VirtFS reply type 41" transport error message by
bad clients. This count value constraint uses msize - 11, because 11 is the
header size of R_readdir.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <3990d3891e8ae2074709b56449e96ab4b4b93b7d.1579567020.git.qemu_oss@crudebyte.com>
[groug: added comment ]
Signed-off-by: Greg Kurz <groug@kaod.org>
A client establishes a session by sending a Tversion request along with a
'msize' parameter which client uses to suggest server a maximum message
size ever to be used for communication (for both requests and replies)
between client and server during that session. If client suggests a 'msize'
smaller than 4096 then deny session by server immediately with an error
response (Rlerror for "9P2000.L" clients or Rerror for "9P2000.u" clients)
instead of replying with Rversion.
So far any msize submitted by client with Tversion was simply accepted by
server without any check. Introduction of some minimum msize makes sense,
because e.g. a msize < 7 would not allow any subsequent 9p operation at
all, because 7 is the size of the header section common by all 9p message
types.
A substantial higher value of 4096 was chosen though to prevent potential
issues with some message types. E.g. Rreadlink may yield up to a size of
PATH_MAX which is usually 4096, and like almost all 9p message types,
Rreadlink is not allowed to be truncated by the 9p protocol. This chosen
size also prevents a similar issue with Rreaddir responses (provided client
always sends adequate 'count' parameter with Treaddir), because even though
directory entries retrieval may be split up over several T/Rreaddir
messages; a Rreaddir response must not truncate individual directory entries
though. So msize should be large enough to return at least one directory
entry with the longest possible file name supported by host. Most file
systems support a max. file name length of 255. Largest known file name
lenght limit would be currently ReiserFS with max. 4032 bytes, which is
also covered by this min. msize value because 4032 + 35 < 4096.
Furthermore 4096 is already the minimum msize of the Linux kernel's 9pfs
client.
Signed-off-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <8ceecb7fb9fdbeabbe55c04339349a36929fb8e3.1579567019.git.qemu_oss@crudebyte.com>
Signed-off-by: Greg Kurz <groug@kaod.org>
