In current implementation, lib/vhost assumes that it always runs
together with lib/event, and would call lib/event's functions in
vhost.c. This is not necessary and make program unable to
create/destroy vhost module without init/fini the whole spdk env. It
would cause problems when program runs with vhost and other spdk
components together.
In this patch, we remove the dependency of lib/vhost on lib/event by
adding a global vairable g_vhost_core_mask so that it could handle
core mask by itself.
Signed-off-by: Kyle Zhang <kyle@smartx.com>
Change-Id: I38ceb92ac39b6980955346fda41e968aaead863d
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1204
Community-CI: Mellanox Build Bot
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
This is executed on hot path for I/O. Inline it.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I6ded0dd3c15e81954350a9eb372156e9d38f87e4
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1725
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Add the packed ring operations in spdk vhost.
Change-Id: I25c9701aadbb283f8f3ecccf599a863395d1c7f0
Signed-off-by: Jin Yu <jin.yu@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/758
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Add the packed ring support in spdk vhost.
Negotiate packed ring feature when start device.
Change-Id: Idef50a1426b6e38d789d8c6982a3ed7594e32cf5
Signed-off-by: Jin Yu <jin.yu@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/672
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Extract _spdk_thread_exit() from spdk_thread_exit() and
_spdk_thread_poll() calls _spdk_thread_exit() if the thread is in
the exiting state. spdk_thread_exit() changes to move the state to
the exiting state. The spdk_thread_poll() loop will end after the
thread moves to the exited state because the caller of
spdk_thread_poll() will check if the thread is in the exited state,
and break the loop if true.
If the user does not call spdk_thread_exit() explicitly, the reactor
has to terminate all existing threads at its shutdown. In this case,
multiple threads may have some dependency to release I/O channels or
unregister pollers. So the reactor has the large two loops, the first
loop calls spdk_thread_exit() on all threads, the second loop calls
spdk_thread_destroy() if exited or spdk_thread_poll() otherwise for
each thread until all threads are destroyed.
Besides, change the return value of spdk_thread_exit() to return
always 0. Keep it for ABI compatibility. Change ERRLOG to INFOLOG
for _spdk_thread_exit() because it is called repeatedly now. Remove
the check of I/O reference count from _spdk_thread_exit() because
_free_thread() cannot free I/O channel. Refine the unit test
accordingly.
Fixes issue #1288.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: Iee5fb984a96bfac53110fe991dd994ded31dffa4
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1423
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Not context passed to vhost_dev_for_each_session() but struct
spdk_vhost_session had been passed to the callback to
vhost_dev_for_each_session() by mistake.
This patch fixes the bug. Besides, rename ctx by ev_ctx to avoid
potential future degradation.
Fixes issue #1306.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I8ceed4e1bb7c0c27fb75516527e3bad91a054b02
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1432
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Karol Latecki <karol.latecki@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
There's no reason not to publish those. Especially if
they're needed in other public headers.
Change-Id: I7dfc6922fcc0dfc46822ad8a16a375f997b98e84
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1041
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
By sending message to the thread with which controller is associated,
we can simplify vhost_dev_foreach_session(). We can iterate
sessions list and we do not have to differentiate if session is
started or not.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I59767a5788c190545a81976e75871609da703f45
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1147
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Currently each controller is associated with one of the poll groups.
For each controller, all sessions are associated with the poll
groups with which the corresponding controller is associated.
Vhost poll group does not have any polling loop but its usage was
very complex.
Association of controller with poll group is done based on the
specified cpumask, and poll group is created per CPU core.
This is as same as association of thread with CPU core.
So in this patch, replace poll group per session by thread per
controller.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: Ifa1e136caae11959f7b097b06a22910bc2169b30
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1146
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
The next patch will create a SPDK thread in vhost_dev_register()
and exit the SPDK thread in vhost_dev_unregister().
As a preparation, simplify error paths in vhost_dev_register()
by changing some gotos to return, moving free after the out label,
and moving insertion after succeeding vhost_register_unix_socket().
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: Id1a2a50b4ba5732f91598a326a08de5c652fd136
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1145
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Replace g_tmp_cpuset by tmp_cpuset local in vhost_get_poll_group()
without any side effect.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: Ibb01d2c77d6d82d7cc81df722e7848b357d4e9e7
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/840
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Use the payload to construct iovs so that split desc
or packed desc can both work.
Change-Id: Ib2c60b6eb53d0d615594e97b3d3ae1a365908b6c
Signed-off-by: Jin Yu <jin.yu@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1000
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Following the recent effort and the last patch, allocate g_tmp_cpuset
statically in lib/vhost/vhost.c.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I5ed6179f9d5942cd868be008c3be6c35d755b455
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/478579
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Following the recent effort, allocate struct spdk_cpuset statically
for struct spdk_vhost_dev. In vhost_dev_register(), a dynamically
allocated cpumask had been set to spdk_vhost_dev, but change it
to spdk_cpuset_copy(). So use local cpuset instance in vhost_dev_register()
accordingly.
To reduce the size of patch, this patch doesn't include the change
for g_tmp_cpuset. This will be done by the next patch.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: Ic97753d1f470cbfd9ae7fc7f2af8ced5a31c8477
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/478578
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
This patch is for the vhost-blk live recovery feature
which can make spdk recover from crash or killing.
The relate rte_xx functions are in the the shared memory
protocol patches which have been merged in DPDK 19.11.
Change-Id: Ia0ac99f8ba0bd66dc9f525f2c72bd1de141ec596
Signed-off-by: Li Lin <lilin24@baidu.com>
Signed-off-by: Ni Xun <nixun@baidu.com>
Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Signed-off-by: Jin Yu <jin.yu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471235
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Add the protocol_features in vdev. There are two features
would be used in vhost device one is the virtio_features
the other is the vhost-user protocol_features. For different
vhost device, the supported features are different so we can
separate them.
Another reason is that I tested the VHOST_USER_PROTOCOL_F_
INFLIGHT_SHMFD in vhost-scsi with QEMU(version:4.0) and found
that Qemu can not boot up. After investigating found that inflight
flag is negotiated but the Qemu doesn't support this feature
and in DPDK function it is handled as an error and disconnect
with Qemu. It's a bug in DPDK and will fix it.
Change-Id: I72e418cb1885bf7dcbd0285d9cec1ad6af0665de
Signed-off-by: Jin Yu <jin.yu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/478814
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: GangCao <gang.cao@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
This will enable us to make features dynamic, dependent on underlying
backend device, and to remove usage of rte_vhost functions in
vhost-blk implementation which will improve encapsulation
and enable us to write tests that use vhost-blk functions directly.
Dynamic features are used in vhost_blk, but backend structure is
assumed to be static, so we had to call
rte_vhost_driver_enable_features() after registering it with
some features initially disabled.
This patch moves feature fields to vdev structure where it
can be set dynamically.
Change-Id: Icd76bdd76a3d67ec74e0ac992d8da639beead593
Signed-off-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/470460
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Community-CI: SPDK CI Jenkins <sys_sgci@intel.com>
This patch introduces indirection layer for session management
functions that makes it possible to switch their underlying implementation
if we want to (in unit tests for example).
Change-Id: I563c97bc65d55cc42fecbd1b7eb6679e394784a2
Signed-off-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/470459
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
This change will allow us to move other DPDK specific functions
to rte_vhost_compat.c, such as session callbacks which are
the only consumers of these memory management functions.
Change-Id: Ieb7b3f08ddf2e7cf04ecf18e8af4ad04124ccfea
Signed-off-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/470458
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Move FLOOR_2MB() and CEIL_2MB() definitions to spdk_internal/memory.h
because that is where common memory functions are located.
Change-Id: I0d366686f86520e5564be07254d98a579faa3650
Signed-off-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471713
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Make them accept a struct rte_vhost_memory * param
instead of a full spdk session object.
Change-Id: Ibe12f29146d2e971ef58fd5f7366cb059123224d
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/470457
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: John Kariuki <John.K.Kariuki@intel.com>
Change type of `vhost_stop_device_cb()` and `vhost_destroy_connection_cb()`
to return response code instead of "void".
While DPDK callbacks `stop_device()` and `destroy_connection()`
do not have response code, it does make sense to have them in
our VHOST wrappers because those actions can fail.
Practical benefit we get by adopting this change is that we can
now use high level `vhost_stop_device_cb()` and `vhost_destroy_connection_cb()`
in unittests and check if they succeeded or not.
Change-Id: I2cd1886728b1edce7946e87db7ca0ac435e83a41
Signed-off-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471712
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
This will allow us to write some more interesting unit tests
because we can now mock high level callbacks instead of
low level DPDK API and also, in future, we won't have to deal with sem_wait()
in our mocked implementation since it's a DPDK specific thing.
Change-Id: I9ed5cff216e750685c00e718025ff1802fbe32c8
Signed-off-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/470456
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
rte_vhost_compat.c will now not only handle vhost-user
messages over the unix domain socket, but also setup
that unix domain socket with rte_vhost's APIs.
What was previously called vhost_dev_install_rte_compat_hooks()
is now called vhost_register_unix_socket() and is responsible
for creating the entire unix domain socket.
This enables us to write more advanced unit tests for vhost.
Instead of mocking low-level rte_vhost APIs, we could
now potentially mock vhost_register_unix_socket() and
create vhost devices and sessions without any actual
unix domain sockets involved.
Change-Id: Ifb18b92b37915c3f683b6d4fcdcc9259a3770561
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Signed-off-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/470455
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
We used to call rte_vhost_driver_start() under
spdk_call_unaffinitized() because that function could
spawn a new pthread and we didn't want to to be pinned
to the one single cpu of the current SPDK reactor.
New DPDK versions (>= 19.05) already unaffinitize the
pthread by themselves, so our spdk_call_unaffinitized()
was only required for the legacy, internal rte_vhost fork
in SPDK. To clean up SPDK code, move the un-affinitization
down to the rte_vhost fork.
Change-Id: I53836517e9ec2ff366b509f00e1403845e3c3172
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/466746
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
We'll start using the same code in even more places soon,
so put in a function.
Change-Id: Iee2e091009b14e9d8b56ec8f0d4a86094f7c9727
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Signed-off-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/467229
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Threads were assigned to sessions inside
vhost_session_send_event() so far, but even the doxygen
comments say that sessions would be assigned to the thread
which called vhost_session_start_done(). Currently, Vhost
uses only vhost_session_send_event() to schedule starting
the session on some thread, so the code ends up working.
We're about to remove vhost_session_send_event(), so move
the thread (poll group) assignment to start_done().
While here, publish the vhost_poll_group struct definition
via vhost_internal.h. As a replacement for
vhost_session_send_event() we would like to use
spdk_thread_send_msg() which a requires a thread object -
one of the struct fields inside vhost_poll_group.
The code for starting a session could look as follows:
pg = vhost_get_poll_group(cpumask);
spdk_thread_send_msg(pg->thread, cb);
...
cb:
// start_pollers
vhost_session_start_done(0);
Change-Id: I563f61509674768c1dea0b03767e9f39a9fb0069
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/467228
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
We used to call a dpdk function to do it, but using
a function for something that simple doesn't make sense.
The function also does its internal queue lookup by vid
and queue number, which could potentially fail, return an
error and technically require SPDK to handle it.
The function makes some sense for vhost-net applications
which don't touch vrings directly but rely on rte_vhost's
API for enqueueing/dequeuing mbufs. SPDK touches DPDK's
rings directly for the entire I/O handling, so it might
just as well for initialization.
This serves as cleanup.
Change-Id: Ifb44fa22ea5fc3633aa85f075aa1a5cd02f5423c
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/466745
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Change the way we increase poll group reference counts
for round-robin scheduling.
So far we used to increase them whenever someone called
vhost_get_poll_group() and this worked fine for Vhost-Block
which picks a new poll group for each session. Vhost-SCSI,
however, picks only one poll group for all sessions on
a vhost device. This means that some threads will have
multiple Vhost-SCSI pollers but will still appear to the
vhost scheduler as if they had only one.
To fix it, increase poll group refcnt only when sessions
are really being started - in vhost_session_start_done().
Change-Id: I60f0d2101239e5a91138a5afd30c51dc1ccf7c2e
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/466733
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Currently vhost_dev_foreach_session() accepts a single
callback function for both iterating through all active
sessions and for signaling the end of iteration (called
last time with vsession param == NULL). Now that the
final signal has completely different semantics and is
called on a specific thread, it makes sense to put it in
a separate function.
While here, remove the one-line description of
spdk_vhost_session_fn typepef. It wasn't helpful anyway.
Change-Id: I56b97180110874a813e666f964bb51c39a8ce6bb
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/466732
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Historically the callbacks from vhost_dev_foreach_session()
could be called with vdev argument == NULL, which would
mean that device was removed after enqueuing the event
and before consuming it. Now we keep track of pending
asynchronous operations on each vhost device and don't
allow removing it if there are any unconsumed events,
so the the vdev == NULL checks are redundant. Remove them.
Change-Id: I7aa3785080d20ed06e008c081d3f37a949228f5a
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/466729
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Remove them all at once. spdk_ prefix should be
only applied to publicly exported functions.
Change-Id: Ib6d2bd0954ec5cb7c8cf253d79b9d3cd8aa0eeef
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/466728
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
We currently don't have any way to differentiate different
sessions e.g. in error messages. Whenever there's an error
in some session, we just print the device name.
We now introduce vsession->name with the following format:
<device name>s<dpdk connection id>
Note that it's still impossible to know exactly which
qemu process corresponds to which session in spdk, but
there's not much we could do in that matter right now.
In spdk we don't even have the accepted connection fd.
Change-Id: I666aa60c5e36bf3d56f68133042af2afc8cc5e85
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/466039
Reviewed-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
We've recently switched from manually calling eventfd_write()
to rte_vhost_vring_call(), which besides writing to the
eventfd, always calls a full memory barrier in the upstream
rte_vhost lib. With upstream rte_vhost we're actually
calling two memory barriers on I/O completion - one in
spdk code, one inside rte_vhost_vring_call().
The spdk barrier was only required for our internal rte_vhost
lib, whose rte_vhost_vring_call() implementation (that we
wrote) did not have such membarrier inside. So now we'll
add this membarrier there, and remove the same barrier
from spdk code.
This doesn't change any code flow for the internal rte_vhost
lib, but optimizes I/O path for the upstream version.
Change-Id: I68738d7feb9159f718b0e60ac7eed1fafd4836b9
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/466037
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Vitaliy Mysak <vitaliy.mysak@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
We used to allocate a ctx whenever new event had to
be sent, but since all events in foreach_session are
always called in a chain, we could allocate one ctx
at the start and then re-initialize it before sending
each msg.
Change-Id: Ie5477b07242f0c6eb6dc2160055a829da8ba5d11
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/459167
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
foreach_session() is used to e.g. close a bdev, where
for each session we close any io_channels and then,
on the final "finish" call, close the bdev descriptor.
The vhost init thread is the one that called
spdk_vhost_init() and also the same one that calls
all management APIs. One of those is for hotplugging
LUNs to vhost scsi targets, which practically results
in opening bdev descriptors.
By always scheduling that final foreach_session()
callback to the init thread, we end up with calling
spdk_bdev_close() always on the same thread which
called spdk_bdev_open(), which is actually a bdev
layer requirement.
Change-Id: I2338e15c63f93ef37dd4412dd677dee40d272ec2
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/459166
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
We used to call potentially-asynchronous foreach_session()
in vdev initialization path and that was perfectly
fine because at that time there were no sessions created
and foreach_session() was always finishing synchronously.
We're about to refactor it to be always asynchronous, and
for this coalescing case it could complicate the init
error path. Once asynchronous thread msg is sent, we would
need to wait for it to complete and we just don't want to
do that. We want error handling to be simple.
Since we know there are no sessions at the time of vdev
creation, we just add a new function for setting coalescing
params just for vdev (and not for its sessions) and we
use that function in vdev init code.
Change-Id: I44d204d03b5040525e4871693678d4b4a0204e63
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/459196
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Put it next to other functions in this call chain.
Change-Id: Ic621855b028f9bd110cdcda86b3a182369ec5e90
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/459165
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Put it next to other functions in this call chain.
Change-Id: Ieafd91c6cfefec134594aec8671eb4efdac15dfe
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/459164
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
spdk_ prefix should be only used on public API functions.
Change-Id: I663b107bd6b1c92c2c6263f2ec7c763d9812e7fe
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/459163
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Despite its name, this function is defined as static
and is only used in one place, so inline it.
Change-Id: I4e217b3baae9b735761f5497f06b681a118860e9
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/459162
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
The semaphore was a part of struct spdk_vhost_session_fn_ctx
so far, but since there's only one pthread waiting on that
semaphore and hence only one event using it, we could just
use a single global sem_t. Same thing with response code
for those callbacks - there's only one needed.
Going a step further, the function complete_session_event()
was removed - it would only operate on global variables now,
and its signature wouldn't make much sense after this
refactor, so it's been inlined.
This serves as cleanup.
Change-Id: I63ef41d7e1564fff5e785de101d887bc1014aad9
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/459160
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Enforce spdk_vhost_fini() to be called on the same
thread which called spdk_vhost_init(). We'll also use
the newly added g_vhost_init_thread for other purposes
later on.
Change-Id: I99aebeda2d8ddaf42554aa422c32ed935634595f
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/459159
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
With all the pieces in place we can finally remove
the legacy cross thread messages from vhost.
We replace spdk_vhost_allocate_reactor() with
spdk_vhost_get_poll_group(). The returned poll_group
has to be passed to spdk_vhost_session_send_event(),
where it will be assigned to the session. After the
session it started, that poll group will be used for
all the internal vhost cross-thread messaging.
Change-Id: I17f13d3cc6e2b64e4b614c3ceb1eddb31056669b
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/452207
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
rte_vhost_vring_call() from upstream DPDK can read some
unitialized memory and crash if it's called on invalid
queue ids. The implementation in our internal rte_vhost
fork ends up wiritng to a random descriptor number, which
doesn't cause any crashes but is a bug nevertheless.
To fix it, just check if the queue is initialized before
interrupting it during the session start. It's not a hot
I/O path and there's no performance impact.
Change-Id: I830c1be98ef00d4ece9a6bd88cf79b9dfe29d2a9
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/457247
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
The memory API has been refactored. It is not possible anymore to
register a memory region more than once. This has been introduced in
this patch: https://review.gerrithub.io/426085
In case of vhost with vvu transport, it often happens that two
consequtive vhost memory regions are mapped to virtual addresses that
lie within the same 2MB address range. This means that the vhost memory
regions may not be 2MB-aligned in the process virtual address space. As
a result, the `FLOOR_2MB()` of those addresses gives the same address.
Thus, we end up trying to register the same 2MB memory range twice.
This issue does not appear in case of AF_UNIX transport. Vhost memory
regions in case of AF_UNIX transport are hugepage backed. Therefore, the
mmapped virtual addresses of those memory regions are always
2MB-aligned. On the contrary, in case of vvu transport, the vhost memory
regions are segments of the PCI memory address space of the
virtio-vhost-user PCI device. This MMIO space is mapped in its entirety
by the DPDK vfio interface along with the other PCI BARs. Ultimately,
the vhost memory regions correspond to offsets in this mmapped PCI
memory region and thus there is no warranty that the mmapped virtual
addresses are 2MB-aligned.
This issue is fixed by skipping the already-registered 2MB memory
regions.
Change-Id: I62c9c257e6f172c894cd3454d2cbeee1986e6189
Signed-off-by: Nikos Dragazis <ndragazis@arrikto.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/441057
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
vring notification mechanism is transport-specific. At present, vhost
dataplane code in `lib/vhost/vhost.c` triggers guest notifications with
`eventfd_write()` system call. But this is an AF_UNIX specific
notification mechanism. This patch replaces `eventfd_write()` with the
existing generic `rte_vhost_vring_call()` function that is part of
DPDK's librte_vhost public API.
`rte_vhost_vring_call()` takes a vring_idx as an argument to associate
the `struct spdk_vhost_virtqueue` instance with the relevant `struct
vhost_virtqueue` instance. We introduce a new `vring_idx` field in
`struct spdk_vhost_virtqueue` to enable this association. This field is
initialized in `start_device()`. In addition, a stub for
`rte_vhost_vring_call()` is added in the vhost unit test file.
SPDK's internal `rte_vhost` copy will not be updated in order to support
the virtio-vhost-user transport. However, an `rte_vhost_vring_call()`
function is introduced in SPDK's `rte_vhost` in order to have a solid
API. This function is just a wrapper of `eventfd_write()`.
Change-Id: Ic93e25cd3f06e92f04766521bc850f1ee80b8ec8
Signed-off-by: Nikos Dragazis <ndragazis@arrikto.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/454373
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
