We'll make use of it inside the vhost device backend
code. The function itself is generic enough to be put
in the public vhost.h header rather than vhost_internal.h.
Change-Id: I60602c61d8bba665dcf9c6d27af2e910c208a7be
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/448226
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>
spdk_dma_*malloc() is about to be deprecated.
Change-Id: Iacf9f6536ba5baca7b245e639d0d42a89720ba58
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/448173
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>
First of all, this struct was used when stopping
a session and wasn't directly related to any vhost
device despite its name.
Second, the struct contained just a single poller.
Instead of renaming it, we remove it. We can use
that poller pointer directly.
Change-Id: I66ad0826f7e809365c07662e59979b1942243c2e
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/448225
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>
* don't iterate through g_nvme_ctrlrs when it's unnecessary
* fixup a potential deadlock on session stop error
(which can't practically happen unless the SPDK generic
vhost layer is malfunctioning)
* add a FIXME note to wait for pending I/Os before putting
bdev io channels and stopping the vhost pollers.
Change-Id: I576c4771f51e432fbbab244fd1b91668436004bf
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/448224
Reviewed-by: Jim Harris <james.r.harris@intel.com>
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>
The context had to be previously carried around by
particular vhost backend code and now it's embedded
inside the generic vsession struct. This serves mostly
as a cleanup.
Change-Id: I7b6ac2c3cb5d60a035d56affbf42fe5d4697f0f6
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/448223
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>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Nothing actually needs this to be asynchronous. If something
comes up, we can make it asynchronous again.
Change-Id: Icde3af3f8f9efebe75b08471b4afcce3a70da541
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/447114
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Windows Virtio drivers use indirect descriptors without
negotiating their feature flag, which is explicitly
forbidden by the Virtio 1.0 spec. "(2.4.5.3.1 Driver
Requirements: Indirect Descriptors) The driver MUST NOT
set the VIRTQ_DESC_F_INDIRECT flag unless the
VIRTIO_F_INDIRECT_DESC feature was negotiated.".
Violating this rule doesn't cause any issues for SPDK
vhost, but triggers an assert, so we can only run Windows
VMs with non-debug SPDK builds.
This patch removes the assert and allows Windows VMs
to be run with debug versions of SPDK vhost.
Fixes#650
Change-Id: I95f534c33c384a4e1126a8c343c21eb63ec7bcef
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/447803
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>
Introduce EMPTY, PRESENT and REMOVED states for Vhost SCSI targets. This
does not introduce any functional changes but opens a way to stop using
both removed flag and spdk_scsi_dev pointer as indicator if device is
removed or not.
Change-Id: Iecd76ffe9e8121cc1359b1e268eb21679d13598e
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/447070
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>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
rte_vhost has rejected a patch with this feature, so
we implement it using the external rte_vhost msg handling
hooks directly in SPDK.
Change-Id: Ib072fc19b921fe0fa01c7f4892e60430232e3a1c
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/447025
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>
Make the vdev initialization happen before calling
any vdev related functions. This is mostly needed
for an upcomming patch where additional step is
required after initializing the vdev and before
starting rte vhost.
On the other hand, this patch also fixes a technically
possible scenario where rte vhost starts processing
vhost-user messages and calling our ops before the
related vdev was initialized.
Change-Id: I8fbc7e7bc0b364327cfcec60faa74d4f64d6fad8
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/447024
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>
rte_vhost requires all queues to be fully initialized
in order to start I/O processing. This behavior is not
compliant with the vhost-user specification and doesn't
work with QEMU 2.12+, which will only initialize 1 I/O
queue for the SeaBIOS boot. Theoretically, we should
start polling each virtqueue individually after
receiving its SET_VRING_KICK message, but rte_vhost is
not designed to poll individual queues. So we use
a workaround to detect when a vhost session could be
potentially at that SeaBIOS stage and we mark it to
start polling as soon as its first virtqueue gets
initialized. This doesn't hurt any non-QEMU vhost slaves
and allows QEMU 2.12+ to boot correctly. SET_FEATURES
could be sent at any time, but QEMU will send it at
least once on SeaBIOS initialization - whenever
powered-up or rebooted.
Vhost sessions are still mostly started/stopped from
within rte_vhost callbacks, but now there's additional
concept of "forced" polling, in which SPDK starts
sessions manually, while rte_vhost still thinks the
sessions are stopped. This can potentially lead to cases
where a session is "started" twice, or gets destroyed
while it's still being polled (by force). Those cases
also need to be handled within this patch.
Change-Id: I70636d63e27914906ddece59cec34f1dd37ec5cd
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/446086
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>
VHOST_USER_SET_VRING_CALL invalidates the previous
file descriptor that SPDK may be using on a different
thread. The new descriptor is stored inside rte_vhost
internals and is queryable with rte_vhost APIs, but
those APIs are too expensive to be called every tick
or every time we need to use that fd. Hence, we will
now stop the entire vhost session before processing
SET_VRING_CALL msg and restart it right after. SPDK
will query the most recent call descriptor on session
start.
We do not necessarily have to stop the device - just
letting the session know that its callfd has changed
would be enough. That's an area for future optimization.
Change-Id: Idccf56fccd21ad0d3c2307eefee7bf35e350fec6
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/447639
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>
DPDK 19.05+ gives us an ability to pre or post-process
any single vhost-user message. The user can either perform
additional actions upon some generic events, or can
implement handling for brand new message types that
rte_vhost doesn't even know about.
In order to smoothly switch to the upstream rte_vhost
and drop our internal copy, we introduce an SPDK wrapper
function to register SPDK-specific message handlers. For
DPDK 19.05+ this will use the new rte_vhost API to
register those message handlers, and for older DPDKs
this function simply won't do anything - as w assume the
internal rte_copy already contains all the necessary
changes and does not need any "external" hooks.
For now we use the message handlers to stop the vhost
device and wait for any pending DMA ops before letting
rte_vhost to process the SET_MEM_TABLE message and unmap
the current shared memory.
Change-Id: Ic0fefa9174254627cb3fc0ed30ab1e54be4dd654
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/446085
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>
It's disabled by default, so no functionality is changed yet.
The intention is to use the upstream rte_vhost from DPDK,
which - starting from DPDK 19.05 - is finally capable of
running with storage device backends.
SPDK still requires a lot of changes in order to support
that upstream version, but the most fundamental change is
dropping vhost-nvme support. It'll remain usable only with
the internal rte_vhost copy and with the upstream rte_vhost
it simply won't be compiled. This allows us at least to
compile with that upstream rte_vhost, where we can pursue
adding the full integration.
Change-Id: Ic8bc5497c4d77bfef77c57f3d5a1f8681ffb6d1f
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/446082
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>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
This is already done for JSON info dump. In addition, the
spdk_vhost_scsi_dev_get_tgt function might implement additional logic to
no return SCSI targets under removal process.
Change-Id: I21d6f660926091dfd34da553705116926f27b30d
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/446910
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Adapted our custom rte_vhost APIs to the upstream DPDK
version which has independently added similar APIs.
This will potentially allow us to remove our internal
rte_vhost copy.
rte_vhost_set_vhost_vring_last_idx() was renamed to
rte_vhost_set_vring_base() and the last vring indices
have to be acquired with a newly introduced rte_vhost_get_vring_base()
rather than rte_vhost_get_vhost_vring().
This is only a refactor, no functionality is changed.
Change-Id: I1ca2c1216635c117832c9d9c784d5661145c04cd
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/446081
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>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Older versions of QEMU (<= 2.11) expose the VGA BIOS
hole (0xA0000-0xBFFFF) by specifying two separate memory
regions - one before and one after the hole. This results
in the "size" not being a 2MB multiple. But the underlying
memory is still mmaped at a 2MB multiple - so that's what
we should be checking to ensure the memory is hugepage backed.
Fixes#673.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I1644bb6d8a8fb1fd51a548ae7a17da061c18c669
Reviewed-on: https://review.gerrithub.io/c/445764
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>
We assumed io_channel allocation always succeeds, but
that's not true. Doing I/O to any vhost session that
failed to allocate an io_channel would most likely
cause a crash.
We'll now detect io_channel allocation failure and
print a proper error message. The SCSI target for
which the channel allocation failed simply won't be
visible to the vhost master. All I/O to that target
will be rejected.
We should probably report the error to the upper
layer and either prevent the device from starting
or fail the SCSI target hotplug request. But for now
let's just prevent the crash.
Change-Id: I735dfb930d8905f70636a236b4fa94288d0aaf3a
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/444874
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>
We explicitly checked for one of the strings in the
parsed RPC request even though it's required for the
entire request to parse successfully. The extra check
is now removed.
Change-Id: I19c446786e4ac88b88f14e18dc5258f31b1a87f1
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/443317
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Since we no longer use external events and we access
all vhost devices synchronously, we no longer need
to dynamically allocate our RPC request contexts. They
can be put just on the stack.
Change-Id: Ie887607b67451aba4f3404c4b9551e6424335beb
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/440380
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Removed their various usages inside the core vhost code
together with the external events themselves. External
events were completely replaced by spdk_vhost_lock()
and spdk_vhost_dev_find().
Change-Id: I1f9d0268c27a06e2eecab9e7d179b1fd54d4223d
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/440379
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Replaced them with inline code that performs exactly
the same but is shorter and easier to follow. External
events were replaced by spdk_vhost_lock() and
spdk_vhost_dev_find().
Change-Id: Id46a619c592c20a573664b54efc097489e9bb893
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/440378
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Vhost external events no longer do any asynchronous
calls, they only lock the vhost mutex and directly
call the provided function. The mutex encapsulation
isn't worth the additional complexity of splitting
each vdev-handling code into multiple functions, so
we expose low-level APIs that should eventually
replace external events entirely.
Instead of:
```
static int do_something_cb(struct spdk_vhost_dev *vdev, void *arg)
{
struct my_data *ctx = arg;
/* access the vdev and ctx */
free(ctx);
}
struct my_data *ctx = calloc(...);
rc = spdk_vhost_call_external_event("my_vdev", do_something_cb, ctx);
if (rc != 0) { /* err handling */ }
```
We can now do just:
```
spdk_vhost_lock();
vdev = spdk_vhost_dev_find("my_vdev");
if (vdev == NULL) { /* err handling */ }
/* access the vdev any context data */
spdk_vhost_unlock();
```
Change-Id: I06e1e149d6dd006720b021d3bef8d9b7bfaeceaa
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/440377
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Existing specific vhost socket messages for vhost-nvme target
will get some information from backend target before start_session
call, so we should iterate the associated nvme controller by vid
but not session.
Fix issue #628.
Change-Id: Ia400bf33895a0feee0058a870f26b0ff72b7556f
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/442498
Reviewed-by: Ziye Yang <optimistyzy@gmail.com>
Reviewed-by: Liang Yan <liang.z.yan@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
We assumed the second descriptor in an I/O descriptor
chain will always point to a payload buffer, but in case
there is no payload, the second descriptor will point to
a response buffer. The vhost code doesn't provide proper
checks to handle such case, so to avoid various errors
down the stack, we just fail all requests with no
payload.
Change-Id: I6785c2843d6db4fc17e68e03562c2a1530bb469b
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/437187
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: <dstepanov.src@gmail.com>
This ensures that SPDK will detect descriptor chains
that are too long.
The additional check in vhost block stands as an
optimization and makes us fail the corrupted I/O early.
Change-Id: Icceaa0dd938dca96a1872e5ee96bf6a151fdd9e7
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Signed-off-by: Dima Stepanov <dstepanov.src@gmail.com>
Reviewed-on: https://review.gerrithub.io/c/433641
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
SPDK doesn't provide sufficient runtime checks to properly
handle clients with memory sizes that aren't 2MB multiples
and could potentially segfault during I/O processing.
That's why we'll reject such clients now.
Change-Id: I34e85be5b5c6df863371d0ad688f228ed44107ff
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/433640
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Those cases should never occur. Klocwork pointed out
possible dereference based on the returns later in
the functions.
Change-Id: I282a56f3f415f85c38e9c451cbb10bc80fc6176b
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/441546
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>
Although Vhost SCSI code is technically capable
of polling different sessions on different lcores,
the underlying SCSI API won't allow allocating
io_channels on more than one lcore.
That's why we will now let device backends assign
lcores by themselves.
The first Vhost SCSI session will now choose one
core from the available ones, and any subsequent
sessions will stick to the same one.
Change-Id: I616cd195a919960dff68508473cea236abf8d6a3
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/441581
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: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
With all the patches in place, we can finally
enable having more than one simultaneous sessions
to a single vhost device.
This patch adds a unique id to the session structure,
similar to the one in a vhost device and also fills in
the implementation holes in foreach_session().
Vhost-NVMe can support only one session per device
and now has an additional check that prevents it from
starting more than one at a time.
Vhost-SCSI also has the same check now since it needs
additional work on the lcore assignment policy. The
check will be removed once the required work is done.
Change-Id: I13a32c7a0eae808e9bec63a7b8c15ec0bc2e36ed
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439324
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Particular backends will now be responsible for sending
events to vsession->lcore. This was previously done by
the generic vhost layer, but since some backends will
need different lcore assignment policies soon, we need
to give them more power now.
Change-Id: I72cbbccb9d5a5b2358acca6d4b6bb882131937af
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/441580
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
It's sessions that are tied with the lcores now.
This makes the vhost devices accessible by any
thread that only locks the global vhost mutex.
The mechanism used for external device events was
refactored to serve for foreach_session() API.
Additionally, since we don't want to handle cases
where the entire vhost device gets removed while
an asynchronous foreach_session chain is pending,
a new per-vdev counter of pending async operations
was added. We'll fail the device removal request
if there are any pending operations. Eventually
we would like the device removal to be asynchronous,
but that's a todo for later.
The external events are still there, although
they only lock the mutex and call the provided
function now.
Change-Id: I20618f9420a9bc04270373469deaad8fb2049c7c
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439323
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Each Vhost SCSI session will now keep its local
SCSI targets state that can be accessed without the
global vhost mutex.
Hotplug will still add the SCSI target reference to
the device struct, but will also asynchronously tell
each active session to add it's session-local copy.
Hotremove, on the other hand, will now try to
asynchronously remove a SCSI target from all sessions
first and only afterwards it will remove it from the
device struct.
This allows us to safely hotplug and hotremove SCSI
targets into Vhost SCSI devices that have multiple
active sessions.
Each session will still use its management poller to
try to locally remove those targets that were scheduled
for hotremoval and the additional
spdk_vhost_dev_foreach_session() will now also try to
remove each one of them from the entire Vhost SCSI
device after making sure they were already removed from
all sessions.
Change-Id: Idd080b618768c71cd1cd564efeaf930bf79fb578
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439321
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Before we implement the support for multiple sessions
per device, we still need to make a few intermediate
changes that will require a counter of currently polled
sessions. So here it is.
Change-Id: I0a1d928eafa75efa1b5c2e6670a5ceb282c87fa4
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/441734
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
The backend struct will get some new dependencies
soon, so move its definition further in a header
file.
Change-Id: I39c25027312777c7e570b12511dc9c5e9b9023d4
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439322
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: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Returning negative value from a `foreach` callback
will now break the entire chain. This is required
for refactoring spdk_vhost_dev_foreach_session() to
use the same mechanism as external events. Before
we actually do the refactor, we add the only feature
that external events were missing.
Change-Id: I70bda3df99748de51429e329a056c37a3bc7e348
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439444
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>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Partially decoded request need to be free even if
spdk_json_decode_object() fails.
Change-Id: Icd00f835537dbaf197cc4f05930be8c543a534a6
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439716
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@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>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
For the backend block device which can support flush command,
vhost-blk should report such feature to Guest, and leave such
decision to Guest.
Change-Id: I6cd6fd94ed80256ffe268bc1bf2c1dd57a164825
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439605
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ziye Yang <optimistyzy@gmail.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: GangCao <gang.cao@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Both the scsi device and its state will need an
additional per-session copy, so we put those two
in a single struct now.
This also serves as a cleanup.
Change-Id: I01bc11db4070e9f258de15e6c44b6f4fcd1d51f2
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439320
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>
Prepared APIs to operate on a session parameter rather
than a device parameter. Some of those functions are
now ready to support more than one session per device.
Change-Id: Ie4a49cd1d1f8ee1b826952bc87e66aa0cabdd925
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439319
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Vhost SCSI session specific fields were moved from the
device struct into a new session struct. This is the
same change as Vhost Block had.
Most of the functions inside vhost scsi still accept
vhost device as a parameter and then get the session
object internally. Those functions will be refactored
to accept session object directly in a separate patch
since the amount of changes required is too big to be
done here.
Change-Id: I8b87ba1187413d471b463aa7067821928ac0303e
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439318
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Following the same change with SCSI target hotremove,
we'll now allow hotremoving an underlying bdev without
VIRTIO_SCSI_F_HOTPLUG negotiated. The hotremoval will
be still reported through SCSI sense codes.
Change-Id: I5b3dd09bd72456eda745f4225f76603fad999da6
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439317
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Failing the hotremove request will become more tricky once
we allow creating multiple sessions per device, so we try
to get rid of any unnecessary error checks.
VIRTIO_SCSI_F_HOTPLUG tells us just if the host is capable
of receiving hotplug events, but the scsi target can be
hotremoved even without them. The hotremoval will be still
reported through SCSI sense codes. All I/O to a hotremoved
target will be failed with sense key ILLEGAL REQUEST,
asc 0x25, ascq 0x00 (LOGICAL UNIT NOT SUPPORTED).
Change-Id: I2be4e0167eb06804112758a5825cd91745128408
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439316
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This lets us remove the assumption of having only
a single session per device and brings us closer
towards supporting more.
Change-Id: Ibbb7b1ed789ff0690e62c00fb5ed39ce64245028
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/438680
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Different Vhost Block sessions could be technically
polled on different threads, so we move the io_channel
from the device struct into the session struct.
Change-Id: I004cad8b6dc6d198844fca3bb11724e3f176dc9d
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439315
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Prepared APIs to operate on a session parameter rather
than a device parameter. Some of those functions are
now ready to support more than one session per device.
Change-Id: Id55e70ae521039f5acc47e80ab8b0aa043679d95
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/438679
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
With all the core vhost changes in place, we can refactor
the upper layers now. We start with the vhost block since
it's the easiest one.
Vhost Block session specific fields were moved from the
device struct into a new session struct. What's tricky,
is that the blk-specific struct directly contains the
generic session struct. This gives us handy access to
the generic session data from the blk-specific session,
and also allows us to directly upcast generic session
objects.
Most of the functions inside vhost block still accept
vhost device as a parameter and then get the session
object internally. Those functions will be refactored
to accept session object directly in a separate patch
since the amount of changes required is too big to be
done here.
Because of the above, some parts of this patch might
seem overcomplicated. Especially the to_blk_session()
funtion, which checks the device backend inside. The
ultimate goal is to receive session object through
start/stop callbacks - and for that the backend check
does make sense.
Change-Id: If222c31aec16a8cbe2d0cfb98c828e1ac75b91fc
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/438678
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
kill a spdk process with aio bdev storage many times will get error messages:
bdev_aio.c: 244:bdev_aio_group_poll: *ERROR*: epoll_wait error(9): Bad file descriptor on ch=0x152a7f0
vhost.c:1010:_spdk_vhost_event_send: *ERROR*: Timeout waiting for event: stop device.
When spdk process exits, the connfd is closed by rte_vhost_driver_unregister, then
other function such as epoll_create1 in aio bdev may allocate the same fd,
but this fd is closed by vhost_user_read_cb again, so epoll_wait return -1
Signed-off-by: Honghui Wang <wanghonghui@ucloud.cn>
Change-Id: Ic3fd938892f004c18fb38d4594c006c40a01efaa
Reviewed-on: https://review.gerrithub.io/c/439851
Reviewed-by: GangCao <gang.cao@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Tested-by: Changpeng Liu <changpeng.liu@intel.com>
Sessions are allocated internally by the core vhost
library whenever DPDK accepts a new connection, so
the only reasonable way to store additional per-sesion
data is to tell the core vhost library how much extra
memory it needs to allocate. Hence, we add a new field
to the vhost device backend struct.
Change-Id: Id6c8285505b2e610e28e5d985aceb271ed232555
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/437778
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>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Instead of calculating those settings once and storing
them in the device struct, we'll now recalculate them
whenever a device session is created. This lets us
remove 2 fields from the device struct.
Change-Id: I2cb2bdbc570a41ae78c0666490fb1462a00d0b6f
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439081
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>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
dev->mem_table_fds init in vhost_user_set_mem_table
but dev->mem may init later in vhost_user_set_vring_addr,
so if qemu crash or lost conntion after vhost_user_set_mem_table
and before vhost_user_set_vring_addr,
it's hugepage memory is not being freed
Signed-off-by: Honghui Wang <wanghonghui@ucloud.cn>
Change-Id: I782c106078829ff6691ed3265a5d1718493de90c
Reviewed-on: https://review.gerrithub.io/c/440254
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>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Makes the code slightly more readable.
Change-Id: Iebf8fb07bceacf433d4bdad0a30419a3faab7eee
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439370
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
We use those values in various places in SPDK,
so let's define them in a single place now.
Change-Id: Iad9a5745d69166a6e6032370d4e5a0e604914e45
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/439369
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Keep all coalescing variables inside the session struct.
Interrupt coalescing is still configured with the device-
pecific APIs, but those will now transparently propagate
the change to all active connections.
This is the last piece that held struct spdk_vhost_dev
tied with the session's lcore. Now that device
settings aren't actively polled by any sessions, they
only need to be synchronized with the global vhost lock.
This will potentially let us get rid of the vhost external
events API, allowing user to lock the mutex directly,
set coalescing params directly, and transparently let
the internal spdk_vhost_dev_foreach_session() do the
tricky synchronization.
Change-Id: Ifba96d241c736d33376861fa894c738e7d9b5b40
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/437777
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
When device is changed, e.g. the underlying bdev is
hotremoved, all sessions need to be notified. For
instance, Vhost-SCSI would send an additional hotremove
eventq message. That's why we introduce a helper
function to iterate through all active sessions.
Eventually, we may want to poll different sessions
from different lcores, so there will be some kind of
internal cross-lcore message management required
- just like there is one for spdk_vhost_call_external_event_foreach().
For now, though, we can get away with a dumbest
implementation.
We still want to keep this API internal for the time
being. The end-user (RPC) should only modify the
device, and the whole concept of sessions should be
completely encapsulated.
Change-Id: I2e142632c07a23daeac15cabea4cffecf984e455
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/418736
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Session struct will be now allocated inside the
`new_connection` rte_vhost callback. There can be
still only one connection per device, but this
change brings us one step towards supporting more.
Besides the obvious pointer changes, we'll now also
use the session pointer to check if the connection
actually exists. We used to set device vid to -1
when there was no connection but we no longer have
to do that.
Change-Id: I4d062c0b5f093fef132a6a2c9cc29458cbaad414
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/437776
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Discard and Write Zeroes commands was supported with
commit 1f23816b in Linux virtio-blk driver. While
here, also add the support in SPDK vhost target.
Change-Id: I425be2a4961eac04e27ff71151d40c8d799cd37d
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/431723
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
This follows the same trend as the mem_map APIs.
Currently, most of the spdk_vtophys() callers manually
detect physically noncontiguous buffers to split them
into multiple physically contiguous chunks. This patch
is a first step towards encapsulating most of that logic
in a single place - in spdk_vtophys() itself.
This patch doesn't change any functionality on its own,
it only extends the API.
Change-Id: I16faa9dea270c370f2a814cd399f59055b5ccc3d
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/438449
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: wuzhouhui <wuzhouhui@kingsoft.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Each connection is created with the `new_connection`
rte_vhost callback with a unique vid parameter. Storing
the vid inside the device struct was sufficient until
we wanted to have multiple connections per device.
Change-Id: Ic730d3377e1410499bdc163ce961863c530b880d
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/437775
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Grouped a few spdk_vhost_dev struct fields into a new
struct spdk_vhost_session. A session will represent the
connection between SPDK vhost device (vhost-user slave)
and QEMU (vhost-user master).
This essentially serves two purposes. The first is to
allow multiple simultaneous connections to a single
vhost device. Each connection (session) will have access
to the same storage, but will use separate virtqueues,
separate features and possibly different memory. For
Vhost-SCSI, this could be used together with the upcoming
SCSI reservations feature.
The other purpose is to untie devices from lcores and tie
sessions instead. This will potentially allow us to modify
the device struct from any thread, meaning we'll be able
to get rid of the external events API and simplify a lot
of the code that manages vhost - vhost RPC for instance.
Device backends themselves would be responsible for
propagating all device events to each session, but it could
be completely transparent to the upper layers.
Change-Id: I39984cc0a3ae2e76e0817d48fdaa5f43d3339607
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/437774
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
By subsequent patches for iSCSI, spdk_scsi_dev_queue_mgmt_task()
will not be called directly from the function that knows TMF code,
and currently setting TMF code to SCSI task is done in
spdk_scsi_dev_queue_mgmt_task().
Hence after subsequent patches for iSCSI, to hand off TMF code to
SCSI task, any dynamic context will be required.
To avoid the dynamic context, extract setting TMF code from
spdk_scsi_dev_queue_mgmt_task() and put appropriate place for
each call of spdk_scsi_dev_queue_mgmt_task().
Additionally, in spdk_abort_transfer_task_in_task_mgmt_resp(),
ref_task_tag is got from PDU but getting it from SCSI task is
much easier. Hence get ref_task_tag from SCSI task in the callback.
Change-Id: I7add9290598d2df7cfcf1506ec75d74c70c0f236
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/436643
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
VHOST_USER_NVME_IO_CMD is designed to deliver NVMe IO command
header to slave target via socket, this can be used in BIOS
which will not enable Shadow Doorbell Buffer feature, since
we enabled the shadow BAR feature to support some old Guest
kernel without Shadow Doorbell Buffer feature, so the message
isn't required, just remove it.
Change-Id: I72e55f11176af2405c8cc09da404a9f4e5e71526
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/420821
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
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>
For some old Linux Guest kernels, the new NVMe 1.3 feature: shadow
doorbell buffer is not enabled, while here, make a dummy BAR region
inside slave target, when Guest submits a new request, the doorbell
value will be write to the shared memory between Guest and vhost
target, so that the existing vhost target can support both new
Linux Guest kernel(newer than 4.12) and old Guest kernel.
Also, the shared BAR space can be used in future which we can move
ADMIN queue processing into SPDK vhost target, with this feature,
the QEMU driver will become very small and easy for upstreaming.
Change-Id: I9463e9f13421368f43bfe4076facddd119f4552e
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/419157
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
We always need to ensure bvdev->bdev is not NULL before
accessing it. GET_CONFIG handler was standing out in this
matter, causing segfaults when connecting to a vhost block
device that went through a bdev hotremove.
If the bdev has been hotremoved we'll now report
blocksize == 0 and blockcount == 0. From what I checked,
QEMU will still expose a virtio-blk device to the VM, but
linux (because that's what I checked) won't create a block
device for it.
Note: The behavior above can be tricky to investigate as
neither QEMU nor VM doesn't log any messages about it.
We should eventually reject any new connections to a vhost-blk
socket with hotremoved bdev, but that's a task for later.
Fixes#460
Change-Id: I266257f4b35d07f35ba03adf46cb18425f3cf39a
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/417934
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@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>
Gcc reports error "vhost_user.c:665: error: ‘MADV_DONTDUMP’ undeclared
(first use in this function)" when build in CentOS-6, include
<asm/mman.h> fixes it.
Change-Id: I1b19c0cb6424a8c5ad1e1dd7d1c724edeb06e171
Signed-off-by: wuzhouhui <wuzhouhui@kingsoft.com>
Reviewed-on: https://review.gerrithub.io/428912
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
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>
User can use this optional parameter to get the information of specific
vhost controller.
Change-Id: I3911c6c7d4e7b75e82277d1e4690d5e40019aa06
Signed-off-by: wuzhouhui <wuzhouhui@kingsoft.com>
Reviewed-on: https://review.gerrithub.io/425451
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
according to commit:
bdev: add spdk_bdev_queue_io_wait
This patch will make io_wait to support vhost_nvme
Change-Id: I20f7c9f2005a0d4140db9ff2d522fb5e15356cd5
Signed-off-by: Ni Xun <nixun@baidu.com>
Signed-off-by: Li Lin <lilin24@baidu.com>
Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Reviewed-on: https://review.gerrithub.io/425884
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
So we don't need to do extra memory allocation when stop vhost device.
This makes code more clean.
Change-Id: I27a1b446621ce4f452fee62acd634737b4ffe174
Signed-off-by: wuzhouhui <wuzhouhui@kingsoft.com>
Reviewed-on: https://review.gerrithub.io/427336
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
according to commit:
bdev: add spdk_bdev_queue_io_wait
This patch will make io_wait to support vhost_blk
Change-Id: I529001fc74427adda63d0d41901a98229364175d
Signed-off-by: Ni Xun <nixun@baidu.com>
Signed-off-by: Li Lin <lilin24@baidu.com>
Signed-off-by: Zhang Yu <zhangyu31@baidu.com>
Reviewed-on: https://review.gerrithub.io/425479
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Sun Zhenyuan <sunzhenyuan@baidu.com>
Reviewed-by: Xiaodong Liu <xiaodong.liu@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
spdk_rpc_get_vhost_controllers() forgot to free memory of w when ctx
allocation failed. Defering the invoke of spdk_jsonrpc_begin_result(),
so we can handle the failure conveniently.
Change-Id: I223f5b9a918b068922d5a68c8d688331ebf90e42
Signed-off-by: wuzhouhui <wuzhouhui@kingsoft.com>
Reviewed-on: https://review.gerrithub.io/425358
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@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>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
This is in response to a Scan-build issue reported on Clang 6.
Change-Id: I2edc853145762998db818cbbe0e9ca0d9b8c123d
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/424139
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
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>
On hot remove, before closing bdev, put its IO channel.
If we don't do this bdev won't be destroyed until controller
is removed.
This fixes GitHub issue #401.
Signed-off-by: Maciej Szwed <maciej.szwed@intel.com>
Change-Id: I92600c2f7136f56afd6593f663f2d2cf468a37dc
Reviewed-on: https://review.gerrithub.io/422258
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Change-Id: I7716f1748803872cac85e296d60747752ca046f4
Signed-off-by: wuzhouhui <wuzhouhui@kingsoft.com>
Reviewed-on: https://review.gerrithub.io/422273
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>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
QEMU will send SET_VRING_ADDR when perform live migration,
it's not correct to update the memory table while the device
is running.
Change-Id: I899d3a996355ab6aa69835d90da14a86f93240fa
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/420944
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
We need these errno to make outputs more clear and make
spdk_vhost_blk_construct() conforms to it's declaration.
Change-Id: I1936e7393494a0344d97c8a920f3d719dee27002
Signed-off-by: wuzhouhui <wuzhouhui@kingsoft.com>
Reviewed-on: https://review.gerrithub.io/421859
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
irq_delay must be not less than zero.
Change-Id: I22d8a7df453f07a44a32582d8e880949824bf868
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/421685
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This fixes queue handling for QEMU 2.12 which is sending new memory
table but resetting only some queues
Fixes#339
Change-Id: Ic971725261720d7459e49a4f14bc15c2f2a77b1a
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/420372
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Chen Wang <chenx.wang@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
This also save CWD on init so socket path is no longer relative.
Change-Id: I303401fe0340f0bc2ea5e4ba468361f68ea84c3d
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/419067
Reviewed-by: Karol Latecki <karol.latecki@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Pawel Kaminski <pawelx.kaminski@intel.com>
Reviewed-by: Paweł Niedźwiecki <pawelx.niedzwiecki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Print an error message instead. The driver can still do
a manual rescan if it wants to see the new SCSI target.
Change-Id: Ieb76ada8625bf00ad068a791b860e4b08ad5cb83
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/417268
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I6babd4cf990bf19b510db88bdfb0ca81e29d9252
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/414700
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Madhu Pai <mpai@netapp.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
As of NVMe 1.3b, there is only one command set. But pipe
this through the driver per-spec anyway.
Change-Id: I4faf8596f5ce638e5e2a500b424e00ceb6e89edc
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/412102
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
"Dev" was deprecated since v18.01, now we have released v18.04,
so remove the "Dev" support from existing code.
Change-Id: I54c4cf83f78d3b0fdb13e625936c889d7bfaeba9
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/409989
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Fix the memory leaks when construct the vhost nvme dev
using the error number of io queues.
Change-Id: Ie63e048b355d8a3d602e0da415dca279c0515a8b
Signed-off-by: Chen Wang <chenx.wang@intel.com>
Reviewed-on: https://review.gerrithub.io/410547
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Liang Yan <liang.z.yan@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
This clarifies the bdev_io completion callbacks, since they can now
assume that they always have a valid bdev_io. All call sites that
originally passed NULL for bdev_io now set an appropriate status code in
the task and complete it with the new spdk_vhost_nvme_task_complete()
function.
Change-Id: Id74aafb28e83e135bbb0a410ff9766dc1b9ece50
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/410080
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Currently Get/Set features vhost messages use 4096 data buffer, but
it does need this buffer for real usage scenario.
Change-Id: If84f795209d771670449283cef3143f3019baee0
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/409613
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Previously after IO is finished, we will put completion entry and
irq event notice at the same IO completion callback, for performance
consideration, move the irq event routine into IO poller context, this
can be used to implement interrupt coalescing feature in future.
Change-Id: Ic20b50af47b73ffcb91938802e18b316c07a4d11
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/408943
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
For namespace we need to use array of objects. Remove useless check for
bdev != NULL as this is already guarded by ns_active.
As we are here also switch to new JSON named API.
Change-Id: If8ab15c63ea015731829276e06e5cb5a801620d2
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/409025
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Add state_mask to each RPC method and state to RPC server, respectively.
State mask of RPC method is set at registration. State of RPC server
is changed according to the state of the SPDK.
When any RPC method is recieved, if the bit of the RPC server is on in
the state mask of the RPC method, it is allowed. Otherwise, it is
rejected.
When any RPC is rejected by state_mask control, the new error code
is returned to describe the error clearly.
Change-Id: I84e52b8725a286e9329d61c56f498aa2c8664ec1
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/407397
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
__rte_always_inline was added in DPDK 17.05; replace the single use with
a regular 'inline' to restore compatibility with older DPDK versions.
Change-Id: Ia8a0f729cc4c39a9aaab0700f3c827a9766d1dd0
Fixes: e30595fbe3 ("rte_vhost: introduce safe API for GPA translation")
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/409077
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The existing code was using sizeof() on a calculation that already
returned the right size, so this would originally just memset() 4 bytes
instead of the whole region.
The spec requires the doorbell buffer to be exactly one physical memory
page, and our controller emulation chooses MPS so that only 4096-byte
pages are supported, so just zero out the page and drop the calculation
entirely.
Change-Id: I71db1bebf0a4d5dbe55fd411786e19a8d6802c20
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/408730
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
If the user requests an invalid QID or CQID in Create I/O Completion
Queue or Create I/O Submission Queue, we need to return an error of
Invalid Queue Identifier as indicated by the spec.
Change-Id: I7467aa04da9e374bb596731f4e4174967d44cffb
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/408767
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Replace the hard-coded 64 and 16 constants, which are the size of the
submission queue entry and completion queue entry, with equivalent
sizeof expressions.
Change-Id: I5a9d8fc1ab98276312445f0699aae3d86beee705
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/408762
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
These assert() calls were all checking guest-provided data; it should
not be possible for a guest to trigger an assert(), so replace them with
actual runtime error checks (which won't get removed in release builds).
Change-Id: I0692dae0e8f6157fc04e88a3ed8087cce9367fe2
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/408761
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Currently QEMU vhost-nvme driver hasn't been pushed to QEMU community,
for vhost-user socket messages, QEMU can pick the opcode at any time,
QEMU 2.12 already picked 27-30 for other driver, for the purpose to
mitigate rebase work in future, while here, we reserve a bigger value
so that it will not conflict with QEMU for very long time.
Change-Id: Ic404bb14330c4acc484aa9c86983030803a31e77
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/408771
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Now that spdk_vhost_gpa_to_vva() is used in place of
rte_vhost_gpa_to_vva(), we can directly use the void * return value
instead of casting it via uintptr_t.
Change-Id: Ie063439def1eeeebc7a949ff67f14a80bfce6cd7
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/408760
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
One call to spdk_vhost_gpa_to_vva() was missing a return code check.
Change-Id: I3581ade98f7fbf72419f8216718a17de50bac4b7
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/408759
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
QEMU added a protocol feature bit to indicate the slave
target can support GET/SET config messages, while here,
add it to SPDK vhost target so that it can work with
QEMU 2.12.
Change-Id: I41a813ef23fba4d3fdf7bb3e3617a9feb4209509
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/408416
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
DPDK will deprecate the old API soon.
Change-Id: I0522d47d9cc0b80fb0e2ceb9cc47c45ff51a5077
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/408722
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
This patch ensures that all the address range is mapped
when translating addresses from master's addresses
(e.g. QEMU host addressess) to process VAs.
Change-Id: If141670951064a8d2b4b7343bf4cc9ca93fe2e6d
Reported-by: Yongji Xie <xieyongji@baidu.com>
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/408721
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This new rte_vhost_va_from_guest_pa API takes an extra len parameter,
used to specify the size of the range to be mapped.
Effective mapped range is returned via len parameter.
Change-Id: Ib3830e1da9e0cb477d99860a03684c665bb3f6ec
Reported-by: Yongji Xie <xieyongji@baidu.com>
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/408720
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I215bc269dee704e60a167023e2a6c24d3ae1fab0
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/395404
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
task must be non-NULL in blk_unmap_complete_cb - it is retrieved from
child->parent, which is always set to a known good task in the only
call site.
Change-Id: I28dce62d4cb586311dc87f988c01aa0e03665e4d
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/408393
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
If basename was the string "", with length 0, it would have
resulted in accessing index -1 in dev_dirname.
Change-Id: Ib389f8fe220f5335a54f6a155a20fcca35b94e3e
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/408253
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Change-Id: I4bec75ee2965394edb294a163818925e6a26fb0c
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/408239
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I929e2a668189c36a8837ce8cb8731e394bcb6d9a
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/408238
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
This fixes two flagged Klocwork issues.
Change-Id: I98ac136995ebcdc89aa94c76fec095573e102674
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/408237
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Moved it to the DPDK thread, so that we don't stress
SPDK I/O reactors on device start/stop. This is mandatory
if we want to maintain hundreds of simultaneous connections.
This patch also fixes various memory registrations leaks
in cases where further device initiation fails.
Change-Id: I435062108fe96d7e67e2a078a3547acb1f73ad11
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/406960
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
rte_vhost_driver_start() can start a new posix thread
that polls for incoming socket connections and calls
backend->start/stop_device. Soon we're going to put
more work on this thread, so we need to make sure it
doesn't collide with SPDK reactors.
This patch also fixes vdev memory leaks in case the
rte_vhost_driver_start() fails.
Change-Id: I16fdff228176a245c478251b39aa244a49bd124b
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/406959
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Change-Id: Ib77a219e02a5cde69293eb1f7002507cf4930ae3
Fixes: 90c0e24410 ("vhost_user_nvme: add vhost user nvme target to SPDK")
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/407193
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Due to the check on line 153 in t_nvme_dev, Klocwork
thinks nvme can be null. Therefore, we must check that
it isn't null before dereferencing it. We can either
solve this the way I have here, or remove check
that causes to_nvme_dev to return null.
Change-Id: I86d4939664704ff1117a7c1b7dada7e1ae479c6f
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/406992
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
SPDK_NOTICELOG should only be used for significant but non-warning/error
messages, such as notifications of deprecated config file sections.
Change-Id: I885dfe839c9cbc9eceabee234e4d0eb74c6910df
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/405923
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Similar with exist vhost scsi/blk target, this commit introduces
a new target: vhost nvme I/O slave target, QEMU will present an
emulated NVMe controller to VM, the SPDK I/O slave target will
process the I/Os sent from Guest VM.
Users can follow the example configuation file to evaluate this
feature, refer to etc/spdk/vhost.conf.in [VhostNvme].
Change-Id: Ia2a8a3f719573f3268177234812bd28ed0082d5c
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/384213
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This will be used to track time used in pollers - each poller can now
indicate if it found any work to do or not.
For cases where it was obvious and the infrastructure was already in
place, existing pollers have been modified to return 0 or a positive
value to indicate whether work was done. Other pollers have been
modified to return -1 by default, indicating that the poller isn't
indicating anything about whether work was performed. This will allow
us to find un-annotated pollers easily in the future and fix them
incrementally.
Change-Id: Ifebfa56604a38434fac5c76ba7263267574ff199
Signed-off-by: Roman Sudarikov <roman.sudarikov@intel.com>
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/391042
Tested-by: SPDK Automated Test System <sys_sgsw@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>
This allows vhost initialization to be fully handled in the event
framework subsystem.
Change-Id: Ic0ff11f3765cc553b7ca183027209fb6dd131364
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/403226
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
This lets us drop the basename parameter from spdk_vhost_startup(),
which is a step toward making vhost initialization happen as part of the
event subsystem instead of in the spdk_app_start() callback.
Change-Id: I4fe18c4c3f12b706e73ae74500be58f1582110fd
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/403225
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
spdk_vhost_dev_has_feature() is internal function so we can move it's
declaration to header file. This remove function call overhead.
Change-Id: I1704e8279cd6720177047a1ae8818f68982998db
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/400241
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Patch 808a2b05 [1] fixed overflow of avail entries count,
also fixing indirect descriptors support. Without indirect
descriptors, we usually didn't manage to fill entire queue
- there were always a couple of descriptors left. A couple,
but not enough to form a vhost-scsi descriptor. Indirect
descriptors, however, are 1-descriptor long. They have higher
possibility to fill entire avail queue, and that's why they
used to fail more often. But we can safely reenable them now.
[1] 808a2b05 ("vhost: fix overflow of avail entries count")
Change-Id: Idec8568cdfc1255fb578e5d18f5c476a4c034d2d
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/396805
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
The cpuset API is const-aware, so we can safely
return const cpuset here. This will simplify
future usages of this function - for now it remains
unused.
Change-Id: If849ed49056a54035fb7a6eccde84121630f9974
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/399457
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This doesn't really fix anything as rte_vhost
logging functions are extremely poorly designed
and do smp_wmb() by themselves - but only if
logging is enabled...
Let's move the barrier in our vhost.c. This
doesn't change any behavior, but makes the code
more readable. It surely looked weird to update
the ring, log it, and then do smp_wmb().
Change-Id: I90827b14292b60bd5641a48d546e7249a0496a4b
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/398326
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
This feature is not particularly useful for storage.
It forces us to read additional vq memory on each
I/O completion and that's quite expensive.
Quoting VIRTIO 1.0:
```
VIRTIO_F_NOTIFY_ON_EMPTY (24)
If this feature has been negotiated by driver, the device MUST issue an
interrupt if the device runs out of available descriptors on a
virtqueue, even though interrupts are suppressed using the
VIRTQ_AVAIL_F_NO_INTERRUPT flag or the used_event field.
```
Later on:
```
Note: An example of a driver using this feature is the legacy networking
driver: it doesn’t need to know every time a packet is transmitted, but
it does need to free the transmitted packets a finite time after they
are transmitted. It can avoid using a timer if the device interrupts it
when all the packets are transmitted.
```
Change-Id: I7f53293bf811a4cd5ae8e42e18f35042ea6f4ba8
Suggested-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/398325
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
This is required only for event index,
which we don't support for now. The whole
NOTIFY_ON_EMPTY feature is about to be removed
soon. See subsequent patches for details. If
necessary, this should be re-implemented in a
different way.
Change-Id: Idd0c1ab929eef46e692834ffdc62472fa0f6fb74
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/398324
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Drop max vhost initiators limitation in SPDK (64).
We're still limited by rte_vhost limits, but
they're set to 1024 at the moment.
Change-Id: Ia1ad25665d6e798bc22709cdd43b72d60f1f4cf0
Signed-off-by: Slawomir Mrozowicz <slawomirx.mrozowicz@intel.com>
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/389811
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This must've been overlooked in patch 601bcbcf.
Fixes: 601bcbcf66 ("util: extend cpumask to hold more than 64 cpus")
Change-Id: Ia7e06fcacba5cd99770b81542678550e9fbc3ca2
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/399455
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This internal function was now being used in one
place only. Inlined it so there's less code.
Change-Id: Ide52494d0e0f9563ce35ef7bf9166f62734a8309
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/399440
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Added vdev->id field and refactored the hot
vhost external event code operating on
g_spdk_vhost_devices array, so that we will
be able to switch from that array to a linked
list with unlimited capacity more easily.
This patch does not change any functionality
on its own.
We would very willingly reuse vdev->vid here,
but it's available only on devices with active
connections.
Change-Id: Ife911a6ca1531aade374b99ef4502cf77f691fa2
Signed-off-by: Slawomir Mrozowicz <slawomirx.mrozowicz@intel.com>
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/389071
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Doing a manual vhost module fini followed
by an spdk_app_stop is wrong, as the app_stop
will trigger the vhost module fini again.
This will potentially free the g_num_ctrlrs
twice. Doing just spdk_app_stop is enough.
Fixes#238
Fixes: 3fb4bc717b ("vhost: move app fini to the vhost subsystem")
Change-Id: I70f886f6792ffc5c499a8d8cb0b6ab01855f7e9b
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/399129
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
QEMU overrides this value anyway, but other
virtio implementation may not. Setting
blkcfg->num_queues to its target value
won't break anything, and will help out
SPDK virtio-blk driver implementation.
For equivalent field in virtio-scsi, the
Virtio 1.0 spec says:
```
num_queues
is the total number of request virtqueues exposed by the device. The
driver MAY use only one request queue, or it can use more to achieve
better performance.
```
Multiqueue for virtio-blk is oficially still
unsupported, so the blkcfg->num_queues remains
undocumented. Let's implement it adequately
to the virtio-scsi spec.
Change-Id: Ief520385997bb7e745ed17501a972c55955346d2
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/399121
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
New vhost user messages GET_CONFIG/SET_CONFIG can be used for
vhost-blk for the purpose to get configuration parameter such
as: Capacity and block size. This commit enable this feature,
users don't need to append capacity any more when started
QEMU. Also event notifier is added for the purpose to change
capacity of block device while QEMU is running.
Also re-enable the vhost-blk tests.
Change-Id: I06ef697984eeea3abbbd655bdcaccaa3b7aa72d7
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/386546
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tree-wide cleanup of all instances of printf()-style functions where a
format string contains a space before a newline character.
Change-Id: Ib5b5861e97bed9e9d62db03875547e3f771f4769
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/397031
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: <shuhei.matsumoto.xt@hitachi.com>
This field was only required to check
if we can safely upcast vdev object.
We can just as well check vdev->backend
instead. The vdev->type is not needed here.
Change-Id: I525350957406d4299151e0557b9025ca7bea5371
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/396584
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
When a target is removed, spdk_scsi_dev_destruct
removes its luns, consequently closing bdev descriptors
and in vhost case - triggering target hotremove again.
This doesn't really have any negative consequences, as
the second hotremove just fails silently. But let's
cleanup this for sanity.
Change-Id: I47c76814696b49905b9fb00667e9529db3f58c12
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/396575
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: <shuhei.matsumoto.xt@hitachi.com>
Instead of:
* spdk_vhost_scsi_dev_remove(vdev)
* spdk_vhost_blk_dev_remove(vdev)
we now have
* spdk_vhost_dev_remove(vdev)
All the logic is already handled internally. This patch only
changes the API. Also, previous vhost_dev_construct()/remove()
functions have been renamed to vhost_dev_register()/unregister()
because that's what they really do.
Change-Id: I7dd0d77bc5b633bec075e0a71345ddbed62697b4
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/396574
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: <shuhei.matsumoto.xt@hitachi.com>
We already have a vhost subsystem and have to do
its fini anyway, so let's remove application fini.
Technically, any SPDK app can use vhost library
and create vhost devices, so we need to delete them
on shutdown in all cases - in module fini.
Change-Id: I013b45b4e73fd19288b9c10bc995dc24894ecf35
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/396573
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: <shuhei.matsumoto.xt@hitachi.com>
This patch adds support for live migration for vhost-scsi and vhost-blk
backends.
Change-Id: Ibfc8a713dbba14ba8cb38377a71e28fd340b1487
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/394203
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
QEMU always set offset to 0 but for sanity we should take the offset
into account.
Change-Id: I36213cd8fbeb85862b6de59c60bd6bcee7f9d1b2
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/395740
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Some messages like SET_FEATURES, SET_VRING_ADDR etc will change internal
state of VQ or device. To prevent race vs thread polling those queues
stop the device.
Change-Id: I15caf9da0decbaa660e9773c93d45ff148e5e9a8
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/395739
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
See comment in vhost.c file for more details why this fix is needed.
Change-Id: I782119768fb200ad36f9253e704cc08a3be36ae6
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/395514
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Virtio 1.0 says that the effective capacity of vrings
is equal to the virtqueue size. When avail ring was full,
we detected it as empty.
Change-Id: I72f5d45cdc14db6b934ce7780e603a3de3131594
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/396804
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Ensure the used->idx changes are visible to
the guest at the time of interrupt.
Change-Id: I245b91c7fbe2e21600aa94d8314271ea5b987343
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/396929
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Fixes github issue #218.
This patch introduces spdk_cpuset object to store and manipulate
the set of individual CPUs. The main objective of this object is
to replace cpumask declared as uint64_t and extend the limitation
of supported CPUs (lcores) above 64 CPUs.
spdk_cpuset is always allocated dynamically and accessed by opaque
pointer, what makes it easier to extend in the future without
breaking API/ABI.
This patch also extends parsing function allowing to set cpumask
using a list of cpus e.g. "[0-4,10,12]" sets mask of 0,1,2,3,4,10,12
as well as hexadecimal string with and without "0x" prefix.
Change-Id: I475c3ba7fab629021a22e03176e57e400dd24a49
Signed-off-by: Tomasz Kulasek <tomaszx.kulasek@intel.com>
Reviewed-on: https://review.gerrithub.io/390794
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Also move the test plan file to a more appropiate location.
Change-Id: I462058279d4d2dc03790657b82b3cfeb00c8a93a
Signed-off-by: Karol Latecki <karol.latecki@intel.com>
Reviewed-on: https://review.gerrithub.io/394180
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
New added vhost user messages: GET_CONFIG/SET_CONFIG are
used for get/set virtio device's configuration space, this
commit enable the new added vhost messages.
Change-Id: I5c3e3f8fb6ed55e99299323c39658765b1724bb8
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/386545
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
vhost I/O only need to be split on 2MB boundaries if
there is a break in the VM's memtable at that 2MB
boundary.
This should drastically reduce (if not eliminate)
the intermittent test pool failures seen recently.
virtio limits number of segments to 128, but this
2MB splitting could introduce additional segment
breaks which we do not allocate IOVs for. In almost
all cases, there are no memtable breaks except at
low 2MB, so most of the extra segment breaks we are
adding are unnecessary.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I12d85c289ad80c7bb65e3d2030a2405092b19deb
Reviewed-on: https://review.gerrithub.io/396058
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Currently we test up to 512KB I/O sizes, and this I/O
size could fragment to 129 IOVs if split on 4KB
boundaries if the buffer start is not 4KB aligned.
This is likely the cause of a recent slew of failures
in the test pool which coincide with updating to
fio 3.3.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I189157af578e75e025ca8e3420712739e604fca7
Reviewed-on: https://review.gerrithub.io/395872
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
There is no need to keep a lun name anymore - we always
use the bdev name as the lun name so it is not providing any
additional value. This also keeps us from associating
the same bdev with different LUNs on different iSCSI target
nodes or vhost-scsi controllers.
Side effect of this change is:
1) Use "bdev_name" across the APIs to make it more clear
what these names refer to.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I3d42fde22087352ce1d5dc80178bd8c5cac8cb7c
Reviewed-on: https://review.gerrithub.io/390843
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
This `len` field describes the amount of bytes
*written* by the device. It is used to prevent
the driver from reading buffer memory that wasn't
really written to - it could contain either
leftover values from previous requests or even
be completely unitialized.
In a couple of invalid-request error conditions
we still write the VIRTIO_BLK_S_UNSUPP response while
returning used_len == 0, but that's fine. A properly
written driver will likely handle both of these cases
the same way.
Change-Id: I8e9f3a8051561096e2ee755122d534b4ffd94db1
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/393963
Reviewed-by: Pawel Kaminski <pawelx.kaminski@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This `len` field describes the amount of bytes
*written* by the device. It is used to prevent
the driver from reading buffer memory that wasn't
really written to - it could contain either
leftover values from previous requests or even
be completely unitialized.
Change-Id: I67aee5c8b9455fc86cdeb600ea3a051a7737c66f
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/393962
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The same check is done inside
spdk_vhost_vring_desc_to_iov.
Change-Id: Id93a05848ed69b8dd048ac490df2432b4252c6e0
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/394088
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
The SCSI layer no longer needs to know about the parent/subtask
relationship maintained by iSCSI.
Change-Id: Ia6f7c5367c5b656bd7521ed1abb6d0f713a0500b
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/393697
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: <shuhei.matsumoto.xt@hitachi.com>
This patch remove need for additional buffer when translating error code
to string.
Change-Id: Iaa60088b5c450581d3cdddbb425119b17d55a44b
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/386114
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
LUN hotremove callback can be triggered when
it's vhost-scsi controller is not started
yet. In such case it is not necessary to check
against HOTPLUG flag (if the controller doesn't
have any driver connected yet, there might be
no negotiated flags whatsoever).
Change-Id: Ic7c78e3bdf45fa4d5628cd19c4c5ad723699e45f
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/371864
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
The SCSI task transfer_len field is only distinct from length in iSCSI;
vhost should always have length == transfer_len.
Change-Id: I37faec26f0f49c27019a965c1f1f48903ca54b58
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/393203
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
pthread_create() returns error codes via its return value, not errno.
Change-Id: Iac0f2380dd4daf15dc48829a3b9b0a75aedd87d1
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/392984
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Adopt two improvements for iSCSI by Ziye to VHOST.
- iscsi/conn: remove rte_config.h header
- env: export spdk_env_get_last_core function
Change-Id: I8f067093d593c8d483c52587669f8b0b706f497f
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/392910
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
The latest patch for the iSCSI connection is applied to the vhost
too.
RTE_MAX_LCORE in the for loop is removed and the for loop is
replaced by SPDK_ENV_FOREACH_CORE().
When the cpumask is unexpectedly 0, not 0 but the first core is
returned.
Change-Id: I39cfc2219a3532eccc8c0ce59712102b947a76d7
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/392588
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ziye Yang <optimistyzy@gmail.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
The latest change for the portal group is applied to the vhost.
The following comment is quoted from it.
Currently the cpumask must be a subset of the reactor mask.
However, this is different from sched_setaffinity() function
and taskset command of FreeBSD and Linux. The latter will
be familier for more people. Hence the later is adopted.
The following is quoted from the FreeBSD Man Page of taskset:
The CPU affinity is represented as a bitmask, with the lowest
order bit corresponding to the first logical CPU and the
highest order bit corresponding to the last logical CPU.
Not all CPUs may exist on a given system but a mask may specify
more CPUs than are present.
A retrieved mask will reflect only the bits that correspond to
CPUs physically on the system.
If an invalid mask is given (i.e., one that corresponds to no
valid CPUs on the current system) an error is returned.
The masks are typically given in hexadecimal.
Change-Id: Idcd72a12ef52e4ccec8476e7d54fab82867cf936
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/392587
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Ziye Yang <optimistyzy@gmail.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
For normal exit logic, such as Ctrl+C, vhost blk will not
shutdown the backend device, e.g: NVMe controller.
Change-Id: I7fdf8687a2cfa6a8cc6a61428d722debfa9a2180
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/391348
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
get_vhost_controllers was printing only one device of vhost
scsi controller with multiple devices
Change-Id: I392fc393d9e8f3d20c28139724d83b79d800456a
Signed-off-by: Pawel Niedzwiecki <pawelx.niedzwiecki@intel.com>
Reviewed-on: https://review.gerrithub.io/391593
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
We used to sprintf the path from
directory and socket name, while
the complete path was already held
in vdev->path.
Change-Id: I27bc71be52a805b39e083eb4f76947587c36a4ad
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/390411
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Require braces around all conditional statements, e.g.:
if (cond)
statement();
becomes:
if (cond) {
statement();
}
This is the style used through most of the SPDK code, but several
exceptions crept in over time. Add the astyle option to make sure we
are consistent.
Change-Id: I5a71980147fe8dfb471ff42e8bc06db2124a1a7f
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/390914
Reviewed-by: <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Added new keyword 'Target' that will replace 'Dev'.
'Dev' is still usable, but deprecated and will be removed.
Change-Id: Iafd6114ef2cc5b2f8d58497e9fb454a3a237ed16
Signed-off-by: Sebastian Basierski <sebastianx.basierski@intel.com>
Reviewed-on: https://review.gerrithub.io/388568
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Disambiguate the log components from the trace functionality
(include/spdk/trace.h).
The internal spdk_trace_flag structure and related functions will be
renamed in a later commit - this is just a find and replace on
SPDK_TRACE_* and SPDK_LOG_REGISTER_TRACE_FLAG().
Change-Id: I617bd5a9fbe35ffb44ae6020b292658c094a0ad6
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/376421
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Althrough the two functions are empty now, but they are not
vhost_scsi specific.
Change-Id: I331a455374f7dd92fd76cab255a43072baceb0dd
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/389477
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Make this generic and not directly dependent on
the event framework. That way our libraries can
register pollers without adding a dependency.
Change-Id: I7ad7a7ddc131596ca1791a7b0f43dabfda050f5f
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/387690
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Change-Id: I8ea8c59a0c67176c0c0c39abf807afad61ff3828
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/387689
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Change-Id: I864cd8a6c206dfbe62fcb3f72275c1ae51aa4ed7
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/387688
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
DPDK 17.11 removed all rte_config.h includes
from rte_*.h headers, meaning we should either
use gcc param -include rte_config.h (just
like DPDK does), or include this file before
each other rte_*.h include. Since we're using
the latter approach in many places already,
I decided to follow it.
While here, also removed rte_vdev.h dependency
from rte_virtio/virtio_user.c. It's not used
anyway.
Change-Id: I865ee9f828211c03a60fd0446f7a418d5dddd140
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/387653
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
There are two separate abstraction layers:
* vsocket - which represents a unix domain socket
* virtio_net - which represents a vsocket connection
There can be many connections
on the same socket. vsocket
provides an API to enable/disable
particular virtio features on
the fly, but it's the virtio_net
that uses these features.
virtio_net used to rely on
vsocket->features during
feature negotiation, breaking
the layer encapsulation (and
yet causing a deadlock - two
locks were being locked in a
separate order). Now each
virtio_net device has it's own
copy of vsocket features, created
at the time of virtio_net creation.
vsocket->features have to be
still present, as features can be
enabled/disabled while no
virtio_net device has been
created yet.
Fixes#214
Change-Id: Ic4b2dd8cae6050813fc9a420b2ed30bc5ae60393
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/386294
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Virtio spec say that any IRQ requests are only hints. So try to limit
number of interrupts generated by vhost by defining minimum interval
between sending IRQ. Coalescing is disabled by default. Can be enabled
using RPC command 'set_vhost_controller_coalescing'.
Change-Id: I9b96014d004ea0ea022b4498c6b47d30d867091a
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/378130
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
The vhost connection can be closed
concurrently from 2 places:
* the connection thread itself
* rte_vhost_driver_unregister
The connection thread will terminate
the connection if any recv error
occured. The unregister function
will terminate the connection
together with the thread.
However, there is no sychronization
between those two. The connection
thread runs in the background
without any mutex.
The rte_vhost_driver_unregister
now signals the connection thread
to terminate itself and waits
until it's killed.
Change-Id: I012e97ebb8a79edcb2c17c28b2fc7e8041bf92b3
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/383085
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Casting pointers without checking their length could
potentially lead to a crash
Change-Id: I7c61e5818ecfbf32bb363858965503341353c51e
Signed-off-by: Pawel Niedzwiecki <pawelx.niedzwiecki@intel.com>
Reviewed-on: https://review.gerrithub.io/382420
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Each virtqueue will now contain
it's own buffer for I/O tasks.
Some of the task struct fields
are now the same across subsequent
I/Os, so they can be now set
just once - at the vdev start/stop.
This simplifies the code, the
debugging process and also introduces
additional sanity check preventing
vhost from processing two requests
with the same id at the same time.
Change-Id: Idcf388e8bf7c92e5536199c35eb0eb6339c00d84
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/369114
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This enable storing SPDK specific stuff per queue. First use of
this will be event index feature.
Change-Id: Ieca6fa47a6f2e23bec73d2cda8b0ed8b9185bd28
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/376636
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Clarify some error messages
and simplify the code.
Change-Id: I586ea55a1d9fa10142d4a9d469b62f1b83076cd5
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/381925
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Some problems with indirect descriptors shown. Disable it for now to
have time to investigate it.
Change-Id: I8c6cfff94dc2159ab01f72986a8ae7d42c353e8b
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/382301
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Closing the framework has to go through all subsystems
without failure, so return codes are unnecessary.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I53c9b4df12d2357e641130869f398b18637e6ff7
Reviewed-on: https://review.gerrithub.io/381681
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
The task was being aborted before
setting it's free_cb. This
results in i.e. not allowing
vhost device to stop it's pollers,
as it would endlessly wait for
this task to finish.
Change-Id: I0de2d1f21da109db1045e433c07f153316e09997
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/381866
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Every descriptor may now contain
a separate descriptor table.
This increases the maximum number
of concurrent IO in a virtqueue
from about (vq->size / 3) to
(vq->size * SPDK_VHOST_IOVS_MAX).
Knowing that SPDK_VHOST_IOVS_MAX
is 128, and assuming that vq->size
is 128 (current hardcoded limit
for QEMU), this gives us over 16k
concurrent possible iovectors
on the fly
Change-Id: I0853d80f6f90d53f8774231972b430a5bf05460e
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/373703
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Also squashed function has_next_desc
into get_next_desc to simplify the
code.
We can't just mask indexes with
(desc_table_size - 1), since in
indirect descriptors case
desc_table_size might not be a
power of 2.
Change-Id: I8053b0e37c553548d76c7a9cfe6b4dbc11c28cfc
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/373744
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
A single timeout error message is printed on the
screen already, so the user knows about the
timeout. Instead of aborting the app here, wait
endlessly for the event to complete. User
can kill the app any time if he wants to.
Change-Id: Ie1853e27b277d654a8e1099866d120f7a51638ba
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/381075
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
One instance of strerror() was introduced after the spdk_strerror_r()
function was introduced in commit 4d43844f4d. Replace it with the
thread-safe version.
Change-Id: I83c6702f7965044ef5fa3577175f89a13dad833f
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/381601
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This change allows for spdk_app_stop() to be executed on a DPDK lcore
instead outside of DPDK lcore range.
This is preparation for SPDK asynchronous finish, to assure
calling spdk_app_stop() from master core.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: Iedc878346c74296b105f93507b2c858aedfbf16e
Reviewed-on: https://review.gerrithub.io/381535
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Maciej Szwed <maciej.szwed@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Piotr Pelpliński <piotr.pelplinski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
A device without LUNs can't be used at all,
so remove it automatically
Change-Id: Iefea73916c83dfddf0229bbfac5dacb23590d920
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/371299
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
A vdev can change between the time the event
is enqueued and executed on proper reactor.
To detect this case, the event callback
was looking up vdev with the same name and
comparing it with old vdev pointer (from
the enqueue). This is now being simplified,
by looking up the current vdev just by
it's ID.
Change-Id: I33de03809b9d56840dd99caa8ccdf190a631ef31
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/380647
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
This makes the async event body non-blocking
and also fixes potential deadlock scenario:
* call_external_event() enqueues vhost_event
* destroy_device() locks g_spdk_vhost_mutex
* destroy_device() enqueues it's vhost_event
* reactor dequeues events
* cb from call_external_event is dequeued first
* g_spdk_vhost_mutex is locked (deadlock)
* destroy_device() times out, abort()
Change-Id: If8f51ada3ee5c47ba126f74187a40a285c001c77
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/380634
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
replaced "remove_vhost_scsi_controller" and "remove_vhost_blk_controller"
with "remove_vhost_controller".
Change-Id: I6f4b180054c13f25aae992e9be50375d3750a376
Signed-off-by: Pawel Niedzwiecki <pawelx.niedzwiecki@intel.com>
Reviewed-on: https://review.gerrithub.io/377197
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Defer writing RPC response until after the device has been hotremoved.
Change-Id: I052280b205415e4a2ffa1421653cc49b8e3b1445
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/371119
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Devices can be added to a SCSI
vhost controller at any time now.
Change-Id: Ic5dd4b78465d3431479e0389b7ea902e1e25c337
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/371281
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
This simplifies eventq API greatly. Since vhost supports only a single
LUN per device, lun parameter in eventq isn't required for now.
Change-Id: I1c41a3f509aca96f2de8de6563a3cac5c4dd1b8b
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/371118
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Since we are calling eventq_enqueue only from inside of the controller
reactor, such ring is not needed. Eventq events will now be sent
immediately. This simplifies the whole process greatly.
The abstract vhost event layer is being replaced with much simpler and
less error-prone vhost external spdk_event API.
This reverts commit 542b5415 [1] together with vhost scsi device
hot-attach feature. Hot-attach shall be reintroduced in upcoming
patches.
[1] 542b541588 ("vhost_scsi: implemented abstract vhost event layer")
Change-Id: I39a427332d573c4e0be548d1c8caf0cf2858f6a5
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/371117
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Script for testing vhost using lvols as backends for
virtual machines. Option for using nested lvol stores and
nested lvol bdevs also provided.
Change-Id: Iad8a196c97f292881a8b424b4433ec281f12aec6
Signed-off-by: Lukasz <lukaszx.galka@intel.com>
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Signed-off-by: Karol Latecki <karol.latecki@intel.com>
Reviewed-on: https://review.gerrithub.io/365798
Reviewed-by: Maciej Szwed <maciej.szwed@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Used vhost external event API to call RPC commands on
controller's reactor. RPC call get_vhost_scsi_controllers
has been left unchanged as it requires special *foreach*
external event API. See next patches for details.
Change-Id: I1fd77c9af2606bbcea45cd0089015dd890b6e328
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/371279
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Controller creation RPC can't be done on controller's reactor
(as there's obviously no controller yet...), so a special
set of vhost_mutex_lock/unlock() functions to has been added
to synchronise controller creation with the rest of
ctrlr-enumerating code.
Change-Id: Ib6e4b894a85ff1f70ebd047832c5a3568a00f1a7
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/377651
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Continuation of patch 94afad5a [1].
That function could be called from outside of the vhost reactor,
causing data races and possible segfaults.
Now, after this patch, all ctrlr-changing functions take spdk_vhost_dev
parameter, meaning they should be called via external event API [1],
which soon will be the only way of obtaining spdk_vhost_dev pointer.
[1] 94afad5a ("vhost: added external API to call spdk_events on vdev reactor")
Change-Id: I40ea66ad09fb5c433dd897a4e22aedeb423f9b4b
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/371013
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Added spdk_vhost_dev_backend_event_done function
that sets the return code for the callback and
transparently calls sem_post.
Change-Id: Iba27af780cd1753056c1607177c945e13c95c712
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/377585
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
DPDK is now fully synchronized with the rest of vhost.
This patch also removes timed_event API from vhost_internal.h.
It should only be used internally in vhost.c under g_spdk_vhost_mutex.
Also, this patch temporarily gets rid of return codes for
vhost_scsi/blk dpdk callbacks. -1 will be returned if the
call timed out, 0 otherwise. A patch reenabling this
functionality is on the way
Change-Id: Ifd9566ceb55a3c6dbe4bac013cc3b7600c834d17
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/372687
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The rte_malloc_socket() call just above that allocates vq is only
allocating sizeof(*vq), but the memcpy() would have tried to copy
sizeof(*vq) * 2.
This code is under #ifdef RTE_LIBRTE_VHOST_NUMA, so it was not normally
enabled with DPDK 17.05, but it breaks when DPDK 17.08 turns on libnuma
support by default.
Change-Id: I75c0c8666a9147346038d313fb419350988d8187
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/377596
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: John Kariuki <John.K.Kariuki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Some rte_vhost files use #ifdef RTE_LIBRTE_VHOST_NUMA, but they don't
explicitly include rte_config.h, which defines this macro. Instruct
the compiler to pre-include rte_config.h in the same way DPDK's build
system does.
Change-Id: Iddde76b8c3d0956ccd5f481956cede650d858586
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/377595
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: John Kariuki <John.K.Kariuki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This is a private function that is only called in one place, and cpumask
is always provided. Additionally, the code inside the mask == NULL ||
cpumask == NULL check always writes to *cpumask, so it was not possible
to pass NULL as cpumask anyway.
Change-Id: I2cde53dfeb07edc8a8f87f4ff485ec0c5c9ba7a8
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/376685
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The next step towards fixing synchronization issues between RPC and
vhost controller reactor.
The general idea is to hide spdk_vhost_dev_find declaration, so that
vdev-related functions can't be (incorrectly) called from outside of
the vhost controller reactor. The vdev pointer should only be acquired
via spdk_vhost_call_external_event.
Change-Id: I689226c2271f81624b2651bae3b960b7efb1c01b
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/371098
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The next step towards fixing synchronization issues between RPC and
vhost controller reactor.
This patch makes changes to already present vhost timed event API
to conform it to the upcoming external spdk_events API. The timed_event
API shall be removed from headers in future. See next patch for details.
Change-Id: I31b0d7c383b39c32326daa750663ebdeb4edd562
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/377584
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This patch properly closes sockets on shutdown.
Previous rte_vhost_driver_unregister calls were returning
immediately due to inexistent sockets at given paths.
Change-Id: I5e4d64f04fc8b6bef914c2eb4c4d2220be9ca9ec
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/377165
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
In preparation to make them static
Change-Id: I086868aa88fa4f6f9f1c24b15a2e3cd0a04caab1
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/377096
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Removed leftover code.
spdk_vhost_timed_event struct has been stripped from
spdk_event and timeout fields. The first one was not
required since vhost_timed_events are not reusable
anymore. The latter has been removed to prepare
vhost_timed_event struct to be expanded to generic
- also non-timing out - vhost_event struct.
Change-Id: Ifd22f59b2748752f2776700676f07e70b45957ed
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/377095
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Public API has been made smaller before refactoring timed_events
entirely.
Change-Id: Ie03bf5066582b6fd243229a9c8b1ceb77658dc2f
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/372686
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Make use of new rte_vhost API.
vdev->vid is now set via new/destroy_connection(),
new/destroy_device sets vdev->lcore.
lcore != -1 implies vid != -1, hence some lcore checks have been
replaced with vid checks.
Change-Id: I08f4932ee7e6aeb3eba4874f2e94ad407764d9c2
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/372075
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Added new callbacks to notify about socket connection status.
As destroy_device is used for virtqueue processing *pause* as well as
connection close, the user has no distinction between those.
Consider the following scenario:
rte_vhost: received SET_VRING_BASE message,
calling destroy_device() as usual
user: end-user asks to remove the device (together with socket file),
OK, device is not *in use* - that's NOT the behavior we want
calling rte_vhost_driver_unregister() etc.
Instead of changing new_device/destroy_device callbacks and breaking
the ABI, a set of new functions new_connection/destroy_connection
has been added.
Change-Id: I50a8ca4035045892d6c658da7df58c0c97025ec3
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/372074
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
If the IOMMU is enabled, automatically register memory
added by the user through spdk_mem_register().
Change-Id: Ie02c7bf445314da23e2efee9de9c187ed0773a9f
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/375249
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
2 locks are executed in 2 places in opposite orders.
Consider the following scenario, threads A and B:
(A)
* fdset_event_dispatch() start
* pfdentry->busy = 1; (lock #1)
* vhost_user_read_cb() start
* vhost_destroy_device() start
(B)
* rte_vhost_driver_unregister() start
* pthread_mutex_lock(&vsocket->conn_mutex); (lock #2)
* fdset_del()
* endless loop, waiting for pfdentry->busy == 0 (lock #1)
(A)
* vhost_destroy_device() end
* pthread_mutex_lock(&vsocket->conn_mutex); (lock #2)
(mutex already locked - deadlock at this point)
Thread B has locked vsocket->conn_mutex and is in while(1)
loop waiting for given fd to change it's busy flag to 0.
Thread A would have to finish vhost_user_read_cb() in order
to set busy flag back to 0, but that can't happen due to
the vsocket->conn_mutex lock.
This patch defers the fdset_del(), so that it's called outside of
vsocket->conn_mutex.
Change-Id: Ifb5d4699bdafe96a573444c11ad4eae3adc359f5
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/375910
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This reverts commit 6973898164.
This solution was incomplete, see the next patch which properly
fixes the deadlock issue.
Change-Id: Ib3cc609814276f1c48b05280379b8c2849ad831f
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/375909
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Replace SPDK_TRACE_DEBUG with component-specific flags.
Change-Id: Iee7eafab5e6ac8713f247323a18552b5afb0e86a
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/375834
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This matches the name to the behavior and prepares for addition of a new
log macro for "info" log level.
Change-Id: I94ccd49face4309d3368e399528776ab140748c4
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/375833
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This function was unnecessarily exposed in public vhost headers.
Controller-constructing functions now take string cpumask param.
Change-Id: Ie97d218c525b1dfb11cfd7e7e13c1bf702b1a58d
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/373759
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
These headers are needed on systems that have old
kernels without virtio_scsi.h and virtio_types.h.
Move them out of the vhost directory and into include
so that they can be used by an upcoming virtio-scsi
initiator bdev module.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ie6614b8845568729a38665c4c1a366b97978e315
Reviewed-on: https://review.gerrithub.io/374548
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Added mutex to synchronize vdev access from external threads.
This is the next step towards fixing synchronization issues
between vhost reactor, RPC, and DPDK.
Change-Id: Id4b52658e05d02c479618ae0e5b5e58af2f02789
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/371175
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
new/destroy_device implementations in particular backends now take
spdk_vhost_dev pointer.
This patch is a first step towards fixing synchronization issues between
RPC and vhost reactor. See the next patch which introduces generic
vhost mutex in vhost.c.
Change-Id: I448492330734726c21189f71ec7a9a8ed81c8195
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/372073
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Try to do the same as DPDK does: volatile write on used->idx and memory
barier before reading avail->flags.
Change-Id: Ibe4629a8228a02088913593ac9b32de56a60b062
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/373578
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
replaces all references to strerror in the spdk lib directory with
references to the thread safe strerror_r
Change-Id: I80d946cce3299007ee10500b93f7e1c8e503ee41
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/374012
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Added helper function to check if vhost controller supports given
feature.
Change-Id: If20883f5dd6d5fedf7fc253239961fdb55ca51fd
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/373702
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
A single warning message is printed instead.
This is required for VMs which have memory mapped virtual devices. (e.g
NVDIMM in Clear Containers). This memory won't ever be used by guest
kernel for IO processing.
Change-Id: Ie24501acbdb3ba0d67bd29b146a09b222451427d
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/373224
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I9f2debc0a9404f4fc36ab307c73ead95a41cbd75
Reviewed-on: https://review.gerrithub.io/371021
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
A virtio descriptor may specify a buffer that spans
a huge page boundary or even a vhost memory region.
So modify spdk_vhost_vring_desc_to_iov() to take this
into account.
While here, also change spdk_vhost_vring_desc_to_iov
to return int instead of bool. int return with -1 is
most standard to indicate failure.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I71faefce0367dc9e44d70ea4429a28dd64f04c10
Reviewed-on: https://review.gerrithub.io/371756
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Piotr Pelpliński <piotr.pelplinski@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Randomly we loose some interrupts. This stop IO in guest till next
request is issued which is abort
Change-Id: I1a0b6391a324cabb0df17d2704126d161b1868f7
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/370248
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Update exisitng tests with vhost blk information
Add tests for scsi/blk hot-attach, hot-detach and hot-plug
Change-Id: I95dad5986d3795430fea304712a6491dfef712d0
Signed-off-by: Karol Latecki <karol.latecki@intel.com>
Reviewed-on: https://review.gerrithub.io/369168
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
- task_data_setup() - value of 1 is no loger returned
- task_submit() -removed outdated comment
Change-Id: I523a465eba7af93535e2dc3a583abb315950f4b3
Signed-off-by: Pawel Niedzwiecki <pawelx.niedzwiecki@intel.com>
Reviewed-on: https://review.gerrithub.io/371154
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
blk_iovs_setup() error code was missinterpeted and no error returned for
IO that was directed to removed bdev.
Change-Id: Iec5ebef20734cca07f20f068b83b74b189cdcb66
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/370711
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Lukasz Galka <lukaszx.galka@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Implemented abstract vhost event layer on top of virtio eventq.
This allows us to defer virtio-event-related actions to be called from
the vhost poller thread. This is required for hot-attach, for which
io_channels have to be created on the I/O thread.
This adds full support for hot-attaching and hot-detaching vhost scsi
devices. They can be now attached to a controller via RPC at any time.
Change-Id: Icf353bfcf69c83ef16b8fc771ea4c487002094f9
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Signed-off-by: Pawel Kaminski <pawelx.kaminski@intel.com>
Reviewed-on: https://review.gerrithub.io/370016
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
-changed "cpu_mask" to "cpumask"
-changed "name" to "lun_name"
-changed "bdev" to "dev_name"
Change-Id: I52b77fa01384df8e6e725f236f23cb568b35f189
Signed-off-by: Paweł Niedźwiecki <pawelx.niedzwiecki@intel.com>
Reviewed-on: https://review.gerrithub.io/368774
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This enables checking permissions - for example,
spdk_bdev_write will fail if the descriptor was not
created with write permissions.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I68b65a560f471f2e0f71a7f42cfa6689b911110f
Reviewed-on: https://review.gerrithub.io/369493
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Change-Id: I50203e0e02d2201fea5112789c1fd84b9a61417a
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/369283
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Virtio block doesn't know how to handle hot-remove so no way of
informing guest when that happen but at least we don't crash.
Change-Id: I54576bc0693c2e706ef3c45b0de313e9beb2a2c7
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/368789
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This variable was misleading and unnecessary. Replaced all it's
occurences with SPDK_SCSI_DEV_MAX_LUN.
Change-Id: Iaaf9ed5efb60f9a89585121d2fbec7b43c3723e9
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/368115
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Renamed it to avoid misconceptions.
Struct rte_vhost_vring describes a virtqueue, not a vring.
Change-Id: I89b94c8d78052b1a4796a9b74653237e5f8b6f52
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/368720
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
struct spdk_vhost_task was renamed to spdk_vhost_scsi_task in commit
68f0c87e64, but this was rebased without
adjusting the spdk_vhost_task introduced by commit
05fd28399a.
Change-Id: I7046d04d3ce56d62fdfb1166f8b4c5676bc7397b
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/369080
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
We need MP MC ring for tasks. So instead implementing this add per
controller ring to avoid this issue.
Change-Id: I5926ad7866ab28251aeebd163917de2df2b44a22
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/368547
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The function body uses VFIO_IOMMU_MAP_DMA which appears to be a
copy&paste mistake, make it VFIO_IOMMU_UNMAP_DMA because we are doing
the opposite of spdk_iommu_mem_register.
Change-Id: If9f3b188178e71ff71221c77c1680b679fec1df0
Signed-off-by: Fam Zheng <famz@redhat.com>
Reviewed-on: https://review.gerrithub.io/368551
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This will enable asynchronous request handling in a future patch, and it
also removes the need for the RPC handlers to know about request id and
the JSON-RPC rules about notification-only requests.
Change-Id: I25aaa8e48bff8d5594ffcccecb61842b1e31ec3c
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/368225
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
-Add a configuration option: Read-only flag
Vhost block controller can be set in read-only mode.
Option can be enabled in config file or in RPC call:
'construct_vhost_block_controllerr' with '-r' option
Change-Id: I7e58243be00d33bc04120d573fd4ed7775bb9b2c
Signed-off-by: Pawel Niedzwiecki <pawelx.niedzwiecki@intel.com>
Reviewed-on: https://review.gerrithub.io/366086
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Retire the old claim/unclaim semantics in favor of
open/close. Clients must now open a bdev to get
an spdk_bdev_desc, then pass this desc to get an
I/O channel.
This allows multiple clients to open a bdev,
although only one may open a bdev with write
access.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I4d319f1278170124169a8a75fd791e926b3f7171
Reviewed-on: https://review.gerrithub.io/367611
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Vhost devices can now be removed via RPC at any time.
Change-Id: Ib91dd6eb11ae8c777e8b9498d140e89a3fe382db
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/366726
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
This flag is needed for compatibility with older versions of the virtio_blk library.
Change-Id: I8243d51a39971dd37d0c65b29bd976b5b84978aa
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/367612
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Removed yet another direct DPDK dependencies.
Change-Id: I4df3cb8975bb7a00b1316ef7ed449248f28ab7ee
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/366724
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
spdk_vhost_get_tasks() is now fully generic function for getting tasks,
which have to be specifically initialized *later*.
This allows us to remove extra loop inside spdk_vhost_get_tasks().
Change-Id: I9a9155189599cbae1b2eb94f64fba29a3e2d010b
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/366723
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Removed get_scsi_dev and get_scsi_lun functions, moved virtio lun field
parsing into a single function.
Change-Id: I2d1588491e0aee523b053d8ea49fdc6bb7f90613
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/366721
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Vhost scsi tasks aren't complicated enough to have their own files.
Also used spdk_ring instead of rte_membuf for task management.
This patch is a first step towards making vhost task pool
queue-specific, so that it can be single-producer, single-consumer.
Change-Id: I181417f49ccd9a5b8f53d4b96c9fefddca087692
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/366720
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Instead of getting tasks from mempool 1 by 1, prefetch a whole chunk.
Change-Id: I17f61e3de9b081b61af7a2a8568681aafe26ab0a
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/366719
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Removed these functions in preperation to removing task.h/task.c.
Change-Id: I84b3c81505c0b8a7d08e2ca803aada1d20bc518e
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/366718
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Since vhost_user_set_features failure is not handled in any way, a
single error log has been added to at least to let the user know that
something has gone wrong.
Change-Id: Ifcf27320af75ba74347b742643b23e43b7c01149
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/365807
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
As of virtio spec 1.0, driver must not accept flags which the device
did not offer (section 2.2). However, that flag is apparently being
accepted, causing silent exception to be triggered in rte_vhost. This
results in all negotiated features to be rejected in rte_vhost.
This patch is a workaround. Enabling that flag does not enforce us to do
any additional actions.
Change-Id: Iede0c0360de33d438f109442c7a83231e9d4479e
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/365806
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
A vhost_dev had been explicitly converted to vhost_scsi_dev in a
following manner:
vhost_dev* -> void* -> vhost_scsi_dev*
Change-Id: Idfd4307bfea7385d71e67f140554dfc2e86447b5
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/367310
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Implemented eventq. For now only hotremove is supported.
Change-Id: I1ec7aa2137173b5dfb30e9906dd41707e371b7a8
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/364285
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Renamed it in preparation for handling eventq & hotremove in the very
same poller. It would only complicate code to handle all of these in
separate pollers.
This patch does not change any functionality on it's own.
Change-Id: I321e2ced0d6afb3ccb66a13bc25bbf42da20417b
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/367528
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Added optional callback inside scsi lun hotremove, so that higher-level
abstraction can be notified about hotremove.
Change-Id: I5f1bd8160e3d770a484068dd73928bc2b64c876f
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/367309
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
For each virt queue's kickfd and callfd, there are 2 invalid
status: VIRTIO_UNINITIALIZED_EVENTFD and VIRTIO_INVALID_EVENTFD.
Don't set the virt queue to ready status until got the valid
descriptor.
This is safe for polling mode drivers in Guest OS, the backend
vhost process will not post notification to interrupt vector for
PMD mode in Guest, but the interrupt vector still valid.
Change-Id: Icdf1e67f3c4e8da221843eb1383469ca1fba485c
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/365327
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
This is an artifact from a past design. There is no longer
any reason to create an event here - the bdev layer will
correctly queue events and call completions on the correct
thread.
Change-Id: I145dab4046899834c0449ec7380dcbb28215b493
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/364831
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
This removes one dependency on the event framework.
Change-Id: Ib9c8160b1704a15b662a35d5ad1f32c0b745485c
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/365729
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Fixes vhost crash on null lun target.
Added extra check in vhost IO path.
Change-Id: Ia6aa29cce41c2fc74e0a8eec545e6675858a28e3
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/365196
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
