Failed is not a final state for either fabric or pcie controllers. We
have historically not allowed resets in the failed state, but we should.
Instead of checking for the failed state, we should check for the
removed state. If the controller is removed, then we cannot even attempt
a reset.
Change-Id: I2c1a3d85db84f84cd1895cbfaf16575c8b496155
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471415
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
We no longer need the private function with a public wrapper.
Change-Id: I0d24dfb282461174729d3eb649c78ac27e42fc8d
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471552
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>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Also, adds a field to the generic qpair for future use in other
transports.
Change-Id: Ie5a66e7f5ebfec1131155fc07e3c671be814fb9b
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471414
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
We shouldn't always fail the whole controller if we get a failure on an
individual qpair.
Change-Id: Id0c90af83e5231593a895be66e7a7de48939e240
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471660
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The way these two functions were separated previously represented a
pretty sserious bug when doing a controller reset.
If there were any outstanding requests in the rqpair, they would get
overwritten during the call to nvme_rdma_qpair_register_reqs and the
application would never get a completion for the higher level requests.
The only thing that we need to do in this function is assign the proper
lkeys.
Change-Id: I304c70646daf9b563cd00badba7141e5e8653aad
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471659
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
check_enabled had a couple bugs in it that made it unfriendly for enabling
I/O qpairs after a reset.
1. It was calling nvme_qpair_abort_queued_requests before setting the
enabled flag to true. For applications that submit new I/O in the
completion callback for old I/O, this means you enter an infinite loop
of submitting requests, and then immediately completing them. SO
instead, wait for the qpair to reset, then just submit those requests to
the lower layer.
2. It didn't check whether we were already in the middle of calling it,
so we could reenter function calls like
nvme_qpair_abort_queued_requests.
Also, now that we have a coherent state machine for qpairs, we can limit
the enabling to a specific state in that state machine.
Change-Id: Ie0b74819a6b16839965bced47c33dec967f725a8
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/470256
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
As the todo states later on in the function, the discovery controller
should really be initialized through traditional methods, but it was
hacked in. For now, enable the admin qpair to get past the non-standard
nature of this controller.
Change-Id: I2cbf1cd47d7249ae3d12bcfc2e8d21e8fb98df7e
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471779
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
This function is identical between the two transports.
Change-Id: If50b781259f224eb2c21de7da14564e6ce487650
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471778
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ziye Yang <ziye.yang@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
These will form the base of a little state machine for managing the nvme
qpair structure.
Change-Id: If6f6df38cc17221ac8fcb7d8c0d7e2e808897a99
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/470534
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
The driver has historically waited until we have to do a listen
before enabling the admin qpair. That is a very PCIe-centric mindset.
For fabric controllers, a lot of the early initialization operations such
as get_cc and set_cc are handled through the admin qpair so it should be
enabled before we begin the initialization process.
As a side effect of this cahnge, the internal API
nvme_ctrlr_enable_admin_qpair has been removed. It would have turned
into a one-liner.
Change-Id: Icd162657d01a85c227a3f20c295d0208e07ce44d
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471743
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>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
This wasn't being done in the previous case which meant that I/O qpairs
were not being moved to the connecting state when connecting for the
first time. However, to prepare the way for a coherent state machine for
nvme qpairs, we need to ensure that all qpairs go through the same
states.
Change-Id: I3cfe799a003acd926b24c107ab1461a96239c1bb
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471753
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
These functions are functionally equivalent. Just unify the way they
wait for completions so that they are completely identical and we can
merge them into a common function.
Change-Id: Id5d734b6ae613b3ac828d89853d986cdadfb211a
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471936
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>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Leaving these on the stack outstanding list can cause unnecessary
buildup. If we fail to post the request to ibv, then the upper layer
request will be freed immediately for reuse, but we will keep that
request in the outstanding queue at the RDMA layer.
Change-Id: Ib422dc9fcb50344ce7c01749f3e20ea9310fd5cb
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/470255
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
This reverts commit bc4e31d6b2.
This change was accidentally merged after it was decided to go with a
different architecture.
Change-Id: Ifc9d8b08bd1fcbc4ace8dd6fb4bd0014330916ed
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/471144
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>
When doing a reset on an NVMe-oF target with active I/O qpairs, we need
to be able to submit fabrics commands on them in order to perform a reset.
Currently, resetting a fabric controller with any I/O qpairs active will
cause the reset to hang indefinitely.
Change-Id: Ic972a301390a4dd64adabedfe01aa4e5253e40b0
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469935
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
The remove callback is a built in way of alerting the user application
that we have removed a controller. Once we fail a controller, we never
move it back out of that state so it is in essence removed.
Change-Id: Iaad6bef0994e9ddd5a424f6b83502f9191b2de49
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469637
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@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>
My recent changes that introduced batching to queued request
resubmission also introduced a regression that can lead to reordering
requests before submitting them to the drive. This change prevents that.
We wait until inside the internal _nvme_qpair_submit_request function to
check for queued entries to avoid queueing a request that has children.
If a request that has children gets queued, when we process completions
and resubmit the parent, it will result in the children being submitted.
Since we only account for the number of requests we completed in the
last iteration, some of the child requests may be requeued out of order,
or worse, none of the child requests will end up being submitted to the
transport and they will all be queued behind previously queued requests.
Change-Id: I58e1c458c25fbf3f9f75364f05b1076b166a6212
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/470890
Reviewed-by: Ziye Yang <ziye.yang@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>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Making this structure available from the ctrlr allows us to call the
remove callback when the controller is failed/removed on transports
other than pcie.
Change-Id: I2c66dfef12b039c0d6daf7df83da745757818006
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469636
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>
This paves the way for doing multiple reconnect attempts before failing
the controller.
Change-Id: I1ff4ee6d41a5ffb47dd186d76793d670287c4783
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469934
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: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: John Kariuki <John.K.Kariuki@intel.com>
By moving the contents of spdk_nvme_ctrlr_reset to a new internal
function, I am paving the way for providing two reset paths. One, which
can be used by the user as an external API function and which provides
the same legacy behavior. Specifically, that it will always fail the
ctrlr after an attempted reset, and a second, internal path, which will
be used by the qpair reconnect code which will defer failing the qpair
to the qpair code.
Change-Id: I9ec9df55c1fecc2f00476c175bcf988207c31257
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469933
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
If we are to have multiple reconnect attempts, we have to control
whetehr the controller is placed in the failed state from outside the
reset function itself. This will allow us to fail the controller only
after all of our retries are exhausted.
Change-Id: Ia82e10325272f25b2b8527336dc3bc507c93b401
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469932
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: Alexey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Community-CI: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
Use the standard API function to fail the controller in all cases.
This patch, and the several following patches are aimed at creating a
mechanism for reporting up to the application layer that a controller is
failed and or removed. To do this, I use the reset_cb to inform the
upper layer that the controller is failed.
This also requires changes to how we handle a controller reset to
pave the way for doing optional reset retries in the libraries.
Change-Id: I06dfce08326c23472a1caa8f6efbac2fd1a720f2
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469635
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 were already passing up from each transport the number of completions
done during the transport specific call. So just use that return code
and batch all of the submissions together at one time in the generic
code.
This change and subsequent moves of code from the transport layer to the
genric layer are aimed at making reset handling at the generic NVMe
layer simpler.
Change-Id: I028aea86d76352363ffffe661deec2215bc9c450
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469757
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.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>
This will be handled by nvme_qpair_submit_request when it receives
-EAGAIN from _nvme_qpair_submit_request.
Change-Id: I5e76aae170c981df0cadaadcd5da1163c715006f
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/470407
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>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
This patch series is aimed at preserving the order of qpair entries
when resubmitting queued requests. The hope is that we will make the API
fool proof and future proof against ever reordering any queued requests.
Change-Id: Ib20d61d3abaed637c9c305b75081947630190fd4
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/470062
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>
Reviewed-by: Alexey Marchuk <alexeymar@mellanox.com>
1. Log level change to info when checking support
2. Delete new lines
3. Enlarge the timeout seconds to 10min for revert
TPer as it sometimes need 6-7min for this operation.
Change-Id: I1b7e32917bd99c859f1515b07f2530669418f0db
Signed-off-by: Chunyang Hui <chunyang.hui@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/468915
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: Paul Luse <paul.e.luse@intel.com>
The tailq and the requests all belong to the generic layer, might as
well put the queueing code there for better encapsulation.
Change-Id: Id5f08f798121b50a21044cfc61856999c50ca227
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469758
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>
This was being ignored, and can cause some problems when trying to reset
a defunt controller over a fabric.
Change-Id: I32c11a0e2df0e140e20f870fe0fb5b9045a567b3
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469638
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>
Previously we would just sit forever. preventing us from properly
attempting reconnects and timing out.
Change-Id: Id7386ab95cf75fd9ac972b44afa2719aad412f49
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469021
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>
This enables us to create a single file descriptor and a single event
channel to poll for completions. With that accomplished, we can easily
poll for events on the admin qpair each time we check it for
completions.
Change-Id: I8b901252510744a956bef12594d1e045715e002e
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/467549
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>
This prevents us from failing a reset and then trying to double put the
rqpair->cq which ends up causing seg faults.
Change-Id: If3e14a3d039b4b19cc587a7482157f4b23f8ee32
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469609
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: Alexey Marchuk <alexeymar@mellanox.com>
In most places, we are passing NVME_TIMEOUT_INFINITE as the
timeout_in_ms argument to nvme_ctrlr_set_state, presumably in an attempt
to specify an infinite timeout. However, nvme_ctrlr_set_state only
checked against 0 when setting the actual timeout, and we didn't have
any logic to check for overflow so we just ended up setting random
timeout_tsc values which changes the behavior of the
nvme_ctrlr_process_init function in several places.
So, change NVME_TIMEOUT_INFINITE to 0, and add some integer overflow
checking to nvme_ctrlr_set_state.
Change-Id: Ic9d0cc57ed153df30c3b20313c3742072a5f992d
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469485
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: Alexey Marchuk <alexeymar@mellanox.com>
Change-Id: I9fb7a998f7c13ce53cba630a895e8e11cf5f4a1c
Signed-off-by: Benjamin Saunders <bsaunders@google.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/467559
Reviewed-by: Paul Luse <paul.e.luse@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>
SPDK_ERRLOG lists the function name, so remove old references that
assume it doesn't and reprint the function name.
Change-Id: I69da6ca0a25bf0eda07d8dad52bcfadf964ac715
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/469487
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>
Weighted Round Robin can be enabled for users, and users
can allocate different priority IO queues for different
purpose. For now we will enable this feature in the
NVMe driver first, following patches will enable this
feature in bdev layer.
Change-Id: I0f799236ca04eb85ef3c9f972ed63ff2718563ba
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/466852
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>
By splitting all cm_event handling into a single function, we can create
a single point of contact for cm_events, whether we want to process them
synchronously or asynchronously.
Change-Id: I053a850358605115362f424de55e66806a769320
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/467546
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
This is paving the way for additional changes to enable polling for
cm_events in the initiator.
For now, just present the same blocking API on top of the now polled
file descriptor. Later, we will change this API to be more useful.
Change-Id: I174dac028720f95c30100f6dc2ed49b5bb2a7e40
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/467545
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
spdk_pci_device_claim() could create a file on the
filesystem that couldn't be deleted programatically.
It could only be overwritten - e.g. by another spdk
instance - but this didn't really work if that
another instance had less privileges and hence no
access to the previous file.
This is exactly the case we're seeing on our CI when
running SPDK as non-root. In general it's a good idea
not to leave any leftover files, so now we'll delete
the pci claim file when the spdk process exits.
spdk_pci_device_claim() used to return a file descriptor
that could be simply closed to "un-claim" the device.
It'll now return only a return code. The fd will be
stored inside spdk_pci_device and will be closed either
when user calls the newly introduced spdk_pci_device_unclaim(),
or when the device is detached.
We'll still need to clean up those files somewhere in
our test scripts (probably ./setup.sh cleanup) to
clean up after crashed processes or so - but we don't
necessarily want to run such scripts inside the autotest
whenever a non-root spdk is about to be started.
Change-Id: I797e079417bb56491013cc5b92f0f0d14f451d18
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/467107
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>
If the CPU reorders the eventidx read before the shadow doorbell
write, it is indeterminate whether the controller will read the
updated shadow doorbell without an MMIO write. See
https://lkml.org/lkml/2018/8/14/1031 for details.
Signed-off-by: Benjamin Saunders <bsaunders@google.com>
Change-Id: I5aa08fdd5b32c7b81e8048ca6efe546318d80b5c
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/468188
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 reverts commit 6129e78d26.
When the initiator sends the discovery log page, if the log page
exceeds the size of its data buffer, it will break it up into
multiple log page commands with appropriate offsets. However,
supporting offsets in log pages is an optional feature in NVMe
and reported by the EDLP bit in the identify data.
This commit changed the discovery process to no longer send an
identify command prior to doing the discovery log page command,
so the values in the identify data are always 0. If the discovery
log page exceeds the size of the data buffer (4k), it will then
fail to send the second log page with an offset because it
believes the controller does not support the feature.
Revert this change to fix it. An identify should always be sent
as part of the discovery process. A test case is included in a
follow up patch the demonstrates the bug.
Reported-by: Zahra Khatami <zahra.k.khatami@oracle.com>
Reported-by: Akshay Shah <akshay.shah@oracle.com>
Change-Id: Iefd512a7521e0fea90541b3eb547671cfa816ea6
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/466819
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Since we use pdu->data_iovcnt to
build the iov in nvme_tcp_build_iovs, so
send out pdu has the maximal iov number
equals to: 2 + pdu->data_iovcnt,
so we change the comparison.
This makes sure that we can handle all the data
owned by one pdu.
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: I2b9258cc5716d706c0fa38af609726c439708768
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/467207
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: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
This will be consistent with TCP and RDMA transport, and we will use
ctrlr->flags in nvme_ctrlr_init_cap() in next patch, the flags will
be cleared to 0 for now.
Change-Id: Ic360cd0c00d60c77452d19cdc1e7a32a5fc34df0
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/466678
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Purpose: Prepare the further optimnization in the
target side whening receving pdu headers, we expect
to use zero copy.
Change-Id: Iae7f9106844736d7160d39d0af1f5941084422ec
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/465380
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
Currently we *always* wait 2 seconds before starting
controller initialization during attach. This
works around an issue where some older Intel NVMe SSDs
could not handle MMIO writes too soon after a PCIe
FLR (which would be triggered when VFIO was enabled).
After further discussion with Intel experts, we know
the SSD models that exhibit this issue. So we can
quirk this so that only the older SSDs incur the extra
delay.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ieb408c24f6afd5bd5147d1c87239aa20f2d13511
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/466064
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
WRITE_TO_RO_PAGE was incorrect and misleading. This
0x82 NVMe status code indicates a write to a read-only
range of LBAs. So modify the constant name and
associated usages to use WRITE_TO_RO_RANGE instead.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I993dbebb5acc2e685a0e99aa14084942ef79d659
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/465083
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
spdk_nvme_detach() will do the normal shutdown notification for
most cases, and it will take some time e.g. 2 seconds to finish
the process for PCIe based controllers. If users' environment
has several drives, each drive will call spdk_nvme_detach() one
by one, and the shutdown process may take very long time.
Since users know exactly what they would like to do for the next
step, so here we provide an option to users, users can enable it
to skip the shutdown notification process so that they can have
very quick shutdown process, and when starting next time, the
controller can be enabled again.
Change-Id: Ie7f87115d57776729fab4cdac489cae6dc13511b
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/463949
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
SPDK_NVME_DEFAULT_RETRY_COUNT is the default value for each controller, so
we can move it out from public header file, and change the value if users
provide a new one.
"NvmeRetryCount" was deprecated for a long time, so we removed the support
for this configuration option as well.
Change-Id: I187251cc1e5342abb4fce96727d06631b7c16a01
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/464489
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
We have defined NVMe controller initialization 'transport_retry_count' option, so
global 'spdk_nvme_retry_count' can be removed, we will remove the variable with
PCIe transport first, and make the retry count can be configured via RPC.
Change-Id: I4d54f78c8da2180d536635587e7291f44a57c4fb
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/464472
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>
Skip token header length which varies for short,
medium and long atom.
Fix Issue #898
Change-Id: I2351193e5a43608495f3d816ff4e5932399a6312
Signed-off-by: Chunyang Hui <chunyang.hui@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/464502
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>
Purpose:
1 Do not caculated the psh_len every time.
2 Small fix, for ch_valid_bypes, and psh_valid_bytes,
we do not need to use uin32_t.
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: I9b643da4b0ebabdfe50f30e9e0a738fe95beb159
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/464253
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
The value of tcp_req->req was asserted after it was already
dereferenced. This patch fixes that.
Change-Id: I5eb01e88be09d41fb8e632c49d5a7ccf2315788f
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/462508
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Admin can enable user and add user to locking range.
Then the user can lock/unlock his range.
Change-Id: Ifc5a8cf5c6b5febeb59c86333981f0cf5b938500
Signed-off-by: Chunyang Hui <chunyang.hui@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/460891
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
There are many req leaks when a controller failure
occurs during submitting IO. It must free all of
the children before freeing the parent req.
If a part of the child req has been sent to the back end
and a part of the child req fails, removes the failed req
from the parent req and the parent req must be retained,
freeing the parent req after all of the submitted reqs return.
Change-Id: Ieb5423fd19c9bb0420f154b3cfc17918c2b80748
Signed-off-by: Huiming Xie <xiehuiming@huawei.com>
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/461734
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Existing children split functions defined in nvme_ns_cmd.c can
also be used in nvme_qpair.c to free children requests with error
paths.
Change-Id: I640b32884424709da67ee89ff780c2de45acc54c
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/461372
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
This change attempts to address the Trello request to decode I/O errors in
NVMe hello_world example.
See https://trello.com/c/MzJJw7hM/2-decode-io-errors-in-nvme-helloworld-example
As part of this change, spdk_nvme_cpl_get_status_string was declared
in nvme.h, and spdk_nvme_qpair_print_command and
spdk_nvme_qpair_print_completion were renamed and added to nvme.h,
allowing all three to used "externally."
To test the failing paths, two compile time defines were added to force a
write or read error (bad LBA) respectively.
As the example does a read after write, if the write fails, the example fails.
Signed-off-by: James Bergsten <jamesx.bergsten@intel.com>
Change-Id: Ib94b4a02495eb40966e3f49517a5bdf64485538a
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/457076
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
The __sync builtin based implementation generates full memory
barriers on some non-x86 platforms. Replace it with C11 atomic
builtins can make:
·arm and ppc from full barrier to half barrier
·x86 code same as before
Signed-off-by: Richael Zhuang <richael.zhuang@arm.com>
Change-Id: Ib6624ef8e45af497b9eced6ecfa7710bcc88a733
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/461590
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
nvme_transport_qpair_process_completions calls nvme_rdma_qpair_process_completions
There are some cases return -1 due to failure of "CQ errors".
Handle CQ polling failures by marking the controller as failed.
That a completion with an error will be treated as controller failed.
Requests will be aborted after retry counter exceeded. Otherwise, code will keep on
reporting errors without recovery.
This is to fix issue #850.
Change-Id: I0b324232310e107bf7fd5722aca54d402a19b14d
Signed-off-by: yidong0635 <dongx.yi@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/460569
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
nvme_tcp_qpair_process_completions returns -1 on socket I/O
error. Unless the caller checks this return value (which
spdk_nvme_wait_for_completion_robust_lock currently doesn't),
on connection loss or any other fatal connection
error spdk_nvme_wait_for_completion will never exit the completion
check loop.
Change-Id: I92bb349beb071db312e6c31b84db2a7b51ec486c
Signed-off-by: Andrey Kuzmin <akuzmin@jetstreamsoft.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/460657
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>
SPDK NVMe-oF initiator driver could not transfer IO whose size is
more than 128KiB even if NVMe-oF target allows IO whose size is
more than 128KiB both for RDMA and TCP transport.
Some use cases need to transfer IO larger than 128KiB.
For RDMA transport, max_mr_size by ibv_query_device of RDMA devices
indicates the maximum size of a single memory region and is independent
from the actual I/O size, and is very likely to be larger than 2 MiB
which is the granularity we currently register memory regions.
Actually some RDMA NICs return UINT64_MAX for max_mr_size by ibv_query_device.
Hence use UINT32_MAX and let the generic layer use the controller data
to moderate this value.
On the other hand, for TCP transport, there is no limit for maximum IO
size and hence use UINT32_MAX.
Besides, for RDMA transport, max_sges should be the minimum of
max_sge got by querying RDMA devices and NVME_RDMA_MAX_SGL_DESCRIPTORS.
Hence do this change together in this patch.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: Idc813afd3e525bf5f370c0fcd2623f9c146a5528
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/459218
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Seth Howell <seth.howell5141@gmail.com>
Reviewed-by: Ziye Yang <ziye.yang@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Previously comparing the transport supported value and the target value
was done in RDMA transport layer. However this comparison should be
done in the generic layer like the maximum IO transfer size. Hence
change the comparison to do in the generic layer in this patch.
Besides, for MSDBD, the value 0 indicates no limit but we had handled
this as maximum number of SGS entries was 0 by mistake. This patch fixes
the bug together.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I54365cf114169b10180ec2c659f9c7302672674c
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/459574
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Using naming rules consistent with other related libraries is helpful
to ensure the quality as verified by this patch series.
This patch changes a few parts to use iov and iovcnt for SGL operations.
Besides, name of an array points to the head of the array and is
constant. So copying name of array to an another pointer is not
necessary and can be removed.
Change-Id: I2324f28126b3088098c1c767cf6c060f22c175c3
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455629
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Maciej Szwed <maciej.szwed@intel.com>
Reviewed-by: Ziye Yang <ziye.yang@intel.com>
NVMe/TCP target had assumed the size of each iovec was io_unit_size.
Using nvme_tcp_pdu_set_data_buf() instead removes the assumption
and supports any alignment transparently.
Hence this patch moves nvme_tcp_pdu_set_data_buf() to
include/spdk_internal/nvme_tcp.h and replaces the current code to use it.
Besides, this patch simplifies spdk_nvmf_tcp_calc_c2h_data_pdu_num()
because sum of iov_len of iovecs is equal to the variable length now.
We cannot separate code movement (lib/nvme/nvme_tcp.c to include/
spdk_internal/nvme_tcp.h) and code replacement (lib/nvmf/tcp.c)
because moved functions are static and compiler give warning if
they are not referenced in lib/nvmf/tcp.c.
The next patch will add UT code.
Change-Id: Iaece5639c6d9a41bd35ee4eb2b75220682dcecd1
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455625
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ziye Yang <ziye.yang@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Compilation Warning on fedora30.
In file included from nvme_ut.c:42:
/home/vagrant/spdk_repo/spdk/test/common/lib/test_env.c:517:17:
warning: The left operand of '>' is a garbage value
if (a1->domain > a2->domain) {
~~~~~~~~~~ ^
This is related to issue #822.
Change-Id: I2b61e821130b89af04db3c475e81d2e91a380a90
Signed-off-by: Hailiang Wang <hailiangx.e.wang@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/459923
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
nvme_tcp_pdu_set_data_buf() has been used to process C2H and H2C for
NVMe/TCP initiator.
In this case, NVMe/TCP cuts out the part of the input data buffer
and transfers the part, and repeats these cut and transfers until
the whole data buffer is transferred. NVMe/TCP uses two SGLs, and
use one to parse from the offset datao to datao + datal and another
to append from the offset 0 to datal.
However, the current nvme_tcp_pdu_set_data_buf() had used
data_length as not data length of this transfer but total length
of the whole transfers by mistake.
Recently DIF library updated to properly handle very similar
cases, and so this patch takes DIF library as a reference and
corrects the implementation.
The next patch will add UT code to verify the bug will be fixed.
The code size is pretty large and so UT code is separated.
Change-Id: Ibeed4de182b8b8740566e874e2757280dc21f9e8
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455623
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: Ziye Yang <ziye.yang@intel.com>
This patch is the first patch of the patch series.
The purpose of this patch series is to correct the bug of
nvme_tcp_pdu_set_data_buf() when the multiple iovecs array is
passed, to share nvme_tcp_pdu_set_data_buf() between NVMe/TCP
initiator and target, and utilize nvme_tcp_pdu_set_data_buf()
not only for C2H and H2C but also in-capsule data in NVMe/TCP
target.
This patch is necessary to satisfy the second requirement, to
share nvme_tcp_pdu_set_data_buf() between NVMe/TCP initiator and target
because struct nvme_tcp_req and struct spdk_nvmf_tcp_req are different.
Four variables, iov, iovcnt, data_offset, and data_len are common,
and hence this patch changes the parameters of nvme_tcp_pdu_set_data_buf()
to accept them.
The bug is fixed in the next patch and tested in after the next patch.
Change-Id: Ifabd9a2227b25f4820738656e804d05dc3f874a5
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455622
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: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Ziye Yang <ziye.yang@intel.com>
spdk_dma_*malloc() is about to be deprecated.
Change-Id: I6c308ee546c28c479ceb903bc1749bf5209dc6fe
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/448172
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: <uma.willpower@gmail.com>
The completion cb of outstanding_tr may submit new requeset to
the outstanding_tr list of the qpair, it's an endless loop.
We only abort the remaining outstanding trackers.
Fix#819
Change-Id: I342f52f4d1836f8ef620ef9e3add0b1986727282
Signed-off-by: JinYu <jin.yu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/457755
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Adds fields to structure spdk_nvme_io_qpair_opts.
These fields allow specifying the locations of memory buffers used
for the submission and/or completion queues.
By default, vaddr is set to NULL meaning SPDK will allocate the memory to be used.
If vaddr is NULL then paddr must be set to 0.
If vaddr is non-NULL, and paddr is zero, SPDK derives the physical
address for the NVMe device, in this case the memory must be registered.
If a paddr value is non-zero, SPDK uses the vaddr and paddr as passed.
SPDK assumes that the memory passed is both virtually and physically
contiguous.
If these fields are used, SPDK will NOT impose any restriction
on the number of elements in the queues.
The buffer sizes are in number of bytes, and are used to confirm
that the buffers are large enough to contain the appropriate queue.
These fields are only used by PCIe attached NVMe devices. They
are presently ignored for other transports.
Signed-off-by: James Bergsten <jamesx.bergsten@intel.com>
Change-Id: Ibfab3939eefe48109335f43a1167082dd4865e7c
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/454074
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
According to the TP 8000 spec in Page 26:
Maximum Number of Outstanding R2T (MAXR2T): Specifies the maximum
number of outstanding R2T PDUs for a command at any point in time
on the connection.
This patch makes the current host driver implementation support one r2t.
We cleanup the code to do the right advertising to the target in the
icreq and avoid attempts to deal with multiple rt2s.
Reported-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: If06ad2e8bde31c2fd7e1c3739f651fb64040e3a9
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455750
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: Or Gerlitz <gerlitz.or@gmail.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
The current nvme_tcp_qpair_disconnect behaviour
is not exactly correct, we do not re-initialize
the state of some data structures of the tqpair.
And this caused the coredump.
Purpose: Fixes#808.
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: I4d2cad8fc0712dbebfc2f3e52373cbe3b9908bf7
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456755
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>
The length should be no larger than the remaining_size.
For example, The remaining_size(firstly, assigned by payload_size) is 128KB,
and user's sgl length is 1MB. Since we already split the I/O, so we should
not use the original length(1MB), but use the remaining_size.
Fix issue reported by: https://github.com/spdk/spdk/issues/808
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: I0a7d0f2282c8ad0e253d8de7091b6c5b87018e9a
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456760
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Previously, if the return value of nvme_tcp_qpair_process_send_queue
is not zero, we directly return but not continue receiving the pdu.
But this is wrong, we should only handle the case when the
return value is negative.
Reported-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: I83453733f5a3e3350a0461b4cb0bc409fde32fea
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455899
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
We will need to put the recently completed nvme_request
object on the qpair's STAILQ. We don't reference any
real data from the nvme_request in the completion path
since we've already stashed the cb_fn and cb_arg in
the nvme_tracker. But we will need to reference the
STAILQ_ENTRY to put it back in the qpair's STAILQ, so
prefetch that cacheline.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Id76122afe4150c84a61fbe38bc874f10d606b3b3
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456673
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
There's no need to set this every time we allocate
a request.
While here, fix a typo near where we needed to modify
the unit test to remove the qpair assertion.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I8af41a6c483415950f625d1ed2ef46088b75a622
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456270
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Request may be submitted several times via nvme_qpair_submit_request
function, such as request in queued_req queue being re-submitted.
With enabling timeout feature, nvme_qpair_submit_request compares
request->submit_tick to zero to check if this is the first submission
for this request. If true, record submit_tick for this reuqest.
So request->submit_tick needs to be set zero in allocation.
Change-Id: Ie3f420aa337802c5ad3962c3fdcd680dec1ccdcb
Signed-off-by: lorneli <lorneli@163.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456328
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>
This is a small optimization.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ib593908d3aeb17aac55be06b8e3be42e28a23061
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456268
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>
Using AVX512 or AVX2 ends up being a small pessimization.
I think AVX works better for copies when there are
multiple cachelines to copy. I see a 2-3% improvement
in high IOPs benchmarks when reverting to SSE.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I3d70a1e359e98cec2a9da41ccf9af2de9baa5868
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456247
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Profiling showed these weren't getting inlined - so add
the inline keyword to make sure it happens. This helps
improve performance a bit.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ia86edccc9163258efdcddcce6989a71fb180caf6
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456099
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
At 10M IO/s, we see a lot of CPU cycles wasted getting
the next tracker into cache. If we only get one
completion at a time, this is unavoidable, but when
there are multiple completions pending, we can prefetch
the second tracker while processing the completion for
the first.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I9de702bee3719e4494eec6f05b09be3672f1e0ac
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456097
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Prior merge contained all of the code EXCEPT for the user-callable function.
Signed-off-by: James Bergsten <jamesx.bergsten@intel.com>
Change-Id: I1cb7105ab85ffae8ed4f600261fed86c9c778893
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456282
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>
Purpose: Make the variable definition consistent
with the same variable in the target side.
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: Ibc4ff92b6346f0a1ad803dcb79d041289f5648b2
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455807
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ie32a1bb144c239b923b5cbb9e608a7dfc9c05208
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/456076
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Maciej Szwed <maciej.szwed@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
In Fedora release 28, plug in nvme device and run setup.sh,
the uevent is like this:
UDEV [1060.112118] add /devices/virtual/vfio/81 (vfio)
ACTION=add
DEVNAME=/dev/vfio/81
DEVPATH=/devices/virtual/vfio/81
MAJOR=509
MINOR=1
SEQNUM=8544
SUBSYSTEM=vfio
USEC_INITIALIZED=1060111894
UDEV [1060.122089] bind /devices/pci0000:d7/0000:d7:00.0/0000:d8:00.0 (pci)
ACTION=bind
DEVPATH=/devices/pci0000:d7/0000:d7:00.0/0000:d8:00.0
DRIVER=vfio-pci
ID_MODEL_FROM_DATABASE=PCIe Data Center SSD (DC P3700 SSD [2.5" SFF])
ID_PCI_CLASS_FROM_DATABASE=Mass storage controller
ID_PCI_INTERFACE_FROM_DATABASE=NVM Express
ID_PCI_SUBCLASS_FROM_DATABASE=Non-Volatile memory controller
ID_VENDOR_FROM_DATABASE=Intel Corporation
MODALIAS=pci:v00008086d00000953sv00008086sd00003703bc01sc08i02
PCI_CLASS=10802
PCI_ID=8086:0953
PCI_SLOT_NAME=0000:d8:00.0
PCI_SUBSYS_ID=8086:3703
SEQNUM=8545
SUBSYSTEM=pci
USEC_INITIALIZED=1060121805
Have tested several kernel versions such as v3.10, v4.10, v4.15, v4.19.
We didn't see an event which is like this:
ACTION=add
DRIVER=vfio-pci
Change-Id: I7299a2fb4d634edaa6bab3412ee8f363f66aae6f
Signed-off-by: JinYu <jin.yu@intel.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/452053
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>
This is the same intention as the patch for iSCSI in this series.
This change will be helpful to extract common part into a specific
helper library if necessary in future.
Change-Id: I1ce36b424ff2afb85f998149f4ef0d7153290802
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455621
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Recently DIF library refined SGL create operation by unifying
size and used count into unused count. This patch applies the
good practice in DIF library to create SGL in NVMe/TCP.
The next patch refines names of related function and variables
to be consistent in NVMe/TCP.
Change-Id: I1e73310c0e3650ede53672d76071a6c37dba82c1
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455473
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
This avoids dereferencing the request to get the qpair
in cases where we already know the qpair. Adding a new
variant instead of just modifying nvme_free_request()
since there are 72 calls to this function and I don't
want to change all of them.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ifd6fd964e546bcd71ff180fd71d5bf5cbab79d4f
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455287
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
In some cases we have the qpair already when calling
this function. So pass the qpair to avoid having
to get it from the request. This shows about a 3%
performance improvement for high IOPs single core
tests.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I22fcca560492f4e7cf5ffedd252e41a027d0dd79
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/455286
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Currently the nvme driver will always log any
request completed with error status. Some
applications may not want this behavior. So provide
an option to disable it at the controller level.
When this option is enabled, any failed requests
from queues associated with that controller
(including the admin queue) will not log the
failed request.
Of course the application will still receive
the failed status code and can decide to do its
own logging there.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ia093fcd23cf321a820fd53183ee7e2dac4f9d378
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/454081
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
This will (finally) enable resets for fabrics
controllers.
Move some of the work previously done in enable_admin_queue
up to this new disconnect/connect logic.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I6239f0c0f36192db921d33f2322b1874b9382a01
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/453939
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
This will be used on the adminq, and set while the
qpair is connecting. It allows the qpair_process_completions
routine to know that it should still try to process completions,
even if the controller is resetting.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I377b9c934295eb5f45f03efd90c2a268defb4bd4
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/453938
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
