The following changes will require nvme_bdev_ctrlr_create() to be
asynchronous. As a preparation, merge nvme_ctrlr_populate_namespaces()
and populate_namespaces_cb() into nvme_bdev_ctrlr_create().
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I623f927da072d746b4bd83314c0f5bcb9cc34ef0
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/7043
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
This will reduce the size of the following patches.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I81d7c90d8676a42aff3ffe73932d5612642a4114
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/7042
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Factor out the operation to destroy qpair into a helper function
bdev_nvme_destroy_qpair().
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I5ac9215407e0b4f4368a68bc58d8e3daac029393
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6783
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Pass nvme_bdev_ctrlr to callback to each spdk_for_each_channel() call
in reset processing.
The following patches will register nvme_bdev_subsystem instead of
nvme_bdev_ctrlr as io_device. Hence we need a different way to pass
nvme_bdev_ctrlr to the completion functions of spdk_for_each_channel().
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: Ief31dbbc145b5850e4e5b2e4c9ad7287128ce870
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6950
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Hold the bdev_io which submits the reset request into the nvme_bdev_ctrlr.
as reset_bio. Then differentiate the callback function between
_bdev_nvme_complete_pending_resets() and _bdev_nvme_abort_pending_resets()
to spdk_for_each_channel() in _bdev_nvme_reset_complete().
The next patch will pass nvme_bdev_ctrlr to for_each_channel() instead.
The following patches will register nvme_bdev_subsystem instead of
nvme_bdev_ctrlr as io_device. Hence we need a different way to pass
nvme_bdev_ctrlr to the completion functions of spdk_for_each_channel().
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: If7a3386815429e8ed44f4e9e5365a21bd97e7fb6
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6913
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
The API spdk_nvme_ctrlr_free_io_qpair() returns immediately if the
passed qpair is NULL, but calling spdk_nvme_ctrlr_free_io_qpair()
with NULL should be avoided.
This patch cleans up the code to ensure that nvme_ch->qpair is NULL if
disconnected and spdk_nvme_ctrlr_free_io_qpair() is called only if
nvme_ch->qpair is not NULL.
Then add a test scenario that two reset requests were submitted
simultaneously and the first reset request failed and then the second
reset request also failed. This verifies the refactoring done in the
next patch.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: Iae461f7f826b0e1a4607a17e528c04a642242d6e
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/7041
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Add support for zoned namespaces in the bdev/nvme module.
Query the namespace type, and if a zoned namespace is found, set the
bdev struct members needed by the bdev_zone.h APIs.
Note that while vbdev_zone_block and bdev_ocssd have zone locks,
bdev/nvme does not need any kind of zone lock, because NVMe ZNS
supports append natively.
The whole point of the zone append command is that you don't need any
locks, even when having multiple writers, since the drive will return
back the Assigned LBA where the blocks were actually written.
If a SPDK application will utilize writes instead of zone appends,
the SPDK application itself will need a per zone lock in order to
make sure that there is only one outstanding write.
Adding zone locks in bdev/nvme would reduce the performance of zone
appends, for no reason at all.
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Change-Id: Ibacdb9b6e839624aba869c9949231343db5914be
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6911
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
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>
Add support in bdev_zone.h for getting the maximum zone append data
transfer size.
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Change-Id: I61203e64d51601232c6578a090fa52975364c1f3
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6910
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
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>
Currently there is possibility of adding a sock to pending_recv
list again if sock->pending_recv is true. Check if flag is false
before adding to the list.
Signed-off-by: Sudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Change-Id: Ie23e1e8dbe1aa5594d9ddea30e7f235e3bf8ddad
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6381
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Ziye Yang <ziye.yang@intel.com>
This seems to be cleaning up the pending_recv list to account for the
missed cases in the previous patches in this series. Now that we're
correctly cleaning up the list, don't do this.
Note that if an EPOLLIN event is received but the application never does
a read/recv, the socket will remain in the pending recv list. The next
poll will get another EPOLLIN event, but the logic already handles that
case.
Additionally, left a TODO for a performance optimization.
Change-Id: I1cdde500a5c76554401a89de766d35b7a486b207
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6746
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>
If there was an EPOLLIN event the socket gets adding to the pending_recv
list. But if the application then does a very large read, it will bypass
the logic that clears the socket from the pending_recv list. Fix this.
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Change-Id: Ia0ba86012f7c6dfd14eb43ba6eeed94dbbce90ce
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6744
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>
from pending_recv list
If the upper layer performs a read/recv, it should still remove the
socket from the pending_recv list.
Change-Id: I32ca8ecccbfe1e53ecc7d6f57343c2727e84b851
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6743
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>
Add a check and only free g_hotplug_probe_ctx once.
This is to fix below issue:
https://github.com/spdk/spdk/issues/1849
Change-Id: I7387495890830badd06a7341d9ac4d149ca1c9f4
Signed-off-by: GangCao <gang.cao@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/7002
Reviewed-by: Jim Harris <james.r.harris@intel.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>
Reviewed-by: <dongx.yi@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
This patch really uses the accelrated engine.
Currently, we only offload the crc32c caculation,
but it can be extended.
Change-Id: If0e4c6a44b6e1e10e03f7eca355bed418d67326b
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6760
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Separate bdev_nvme_reset() and _bdev_nvme_reset() by making
bdev_nvme_reset() call _bdev_nvme_reset_start(), and then
remove the ctx parameter from _bdev_nvme_reset().
This clarifies the next patch and reduces the size of the next patch.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I76b0f2f5b83445845a313203e594dca0be150bc3
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6949
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
We need to pass bio only when the reset request came from the upper
layer. Previously, bdev_nvme_reset() called _bdev_nvme_reset() and
_bdev_nvme_reset() had the second parameter, ctx, to pass bio to
for_each_channel(). However, it was not clean.
Let's separate bdev_nvme_reset() which processes a reset request
came from the upper layer and _bdev_nvme_reset() which proesses
an internal reset request by adding a helper function
_bdev_nvme_reset_start().
The next patch will remove the second parameter, ctx, from
_bdev_nvme_reset() for clarification and further refactorings.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I2bd00e3abb56fe8d540b8af55743ee92ca9cd526
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6948
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
The recent refactoring removed the destruct poller and change the
reset processing to destruct ctrlr after its completion by conditionally
sending message.
But differentiating callback function is difficult if we reset multiple
ctrlrs.
If nvme_bdev_ctrlr->destruct is set, any new reset cannot start. So we can
use an new variable and always execute the callback function.
Add an new variable pending_destruct to struct nvme_bdev_ctrlr, and set
pending_destruct if ctrlr->ref is zero and ctrlr->destruct is true, and
then start destruct ctrlr if ctrlr->destruct_after_reset is set after
clearing pending resets.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I9f34c42a40c5a5da54611e7871aef8c58117a56a
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6714
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
The following patches will cache and use spdk_nvme_ns pointer rather than
nvme_bdev_ns pointer in I/O paths. As a preparation, change a few cases
to get spdk_nvme_ctrlr pointer from spdk_nvme_ns.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: Ief5f01f85f46e2ad822af5b5abb13a14f28a0c6a
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6834
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
This will make the following changes a little easier.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I97ce4fb03d02679f5be152b04c33650bf5164f2a
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6833
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Purpose: To setup an accelerated function callback
for created spdk_nvme_poll_group. In this patch,
we just create the interface. The real usage of this
call back will be provided in the other patch.
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: I0d936aa4eba4dbfcc0137942156b9f2919eb5b78
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6758
Community-CI: Broadcom CI
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: Aleksey Marchuk <alexeymar@mellanox.com>
Previously, not nvme_bdev_ctrlr but spdk_nvme_ctrlr pointer had been
passed to spdk_get_io_channel() in nvme_rpc_io_cmd_bdev_nvme().
There was no unit test for bdev_nvme_get_io_qpair(), and so add
unit test for it to guard us from degradation.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I00fdc14ea6467162c5fb90d04389883d7c409a17
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6920
Community-CI: Broadcom CI
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
OCF now includes on its own only small part of env, that is crucial for
compiling headers for casadm - an OCL management tool. The idea behind
this is to reduce the amount of included stuff when it is not needed.
Signed-off-by: Rafal Stefanowski <rafal.stefanowski@intel.com>
Change-Id: I58fef2be3ccb1cce743093ed973ba54cdbb605e1
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6413
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
This is the only reason the event.h header is included here
and it isn't necessary.
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Change-Id: I4c0a9828cfede1ccd1c9263ad66b354ba167f434
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6919
Community-CI: Broadcom CI
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: <dongx.yi@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Signed-off-by: Jacek Kalwas <jacek.kalwas@intel.com>
Change-Id: I191ad5e3b153fb563256eba1aa695716f66db788
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6377
Community-CI: Broadcom CI
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Purpose: Make the code clean and remove duplicated code.
Change-Id: I6cc8b94af24f89301531bd38d766afb53898e0ed
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6811
Community-CI: Broadcom CI
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: <dongx.yi@intel.com>
The next patch will add nvme_bdev_channel and we will not be able to
io_channel from nvme_ch anymore.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I8ba93e66657d21da58df63f3ad6dde4b155fc174
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6612
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Add mutex per nvme_bdev_ctrlr and replace g_bdev_nvme_mutex by it
for possible cases. These are just drop-in-replacement.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: Ib1d5eebd65e514cfe96ad60a9004d9bace7b4f6b
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6693
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
This will make us easier to maintain the operation to delete
nvme_bdev_ctrlr and its trids. The added unit test cases guard us
from degradation.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I400d4092020e89bacaebc7be045a456b8760ed8d
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6688
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
By adding the second parameter, hotplug, factor out the common part
of remove_cb() and bdev_nvme_delete() into a helper function
_bdev_nvme_delete().
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I5becbc7a7a94e852510e6a2745afb9c533d21713
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6692
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
spdk_bdev_part_base_construct() is deprecated so this patch
moves remaining instances to spdk_bdev_part_base_construct_ext().
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: Iccd4b15933bec41885adfaf4fc4ce2abedd2d6c7
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6627
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
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: Aleksey Marchuk <alexeymar@mellanox.com>
Leverage SO_INCOMING_CPU to get the CPU affinity of connections
(sockets). And allocate the connections to specific poll groups,
which aims to utilize cache locality.
From our test:
6 P4600 NVMe on target,target uses 8 cores, NIC irqs are bound to
these 8 cores, and initiator side uses 24 and 32 cores,
we can get 11%~17% randwrite performance boost for posix, and 8%~12%
for uring.
Change-Id: I011e0a21502c85adcccd4a14fbe9838b43f54976
Signed-off-by: Richael Zhuang <richael.zhuang@arm.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5748
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Ziye Yang <ziye.yang@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
nvme_bdev_ctrlr is valid while populating namespaces of the
nvme_bdev_ctrlr regardless of the result. nvme_bdev_ns holds
nvme_bdev_ctrlr by its pointer, ctrlr. Hence it is not necessary for
nvme_ctrlr_populate_namespaces_done() to use nvme_bdev_ctrlr_get_by_name()
to get nvme_bdev_ctrlr.
To simplify the code, add nvme_bdev_ctrlr parameter to
nvme_ctrlr_populate_namespaces_done() and remove the
nvme_bdev_ctrlr_get_by_name() call from nvme_ctrlr_populate_namespaces_done().
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: Ic2013842fad16ea0de49c18b5b853d147c23a6b2
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6621
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: <dongx.yi@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
bdev_nvme_no_pi_readv() had not been used for bdev_nvme_no_pi_readv() yet.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I9bf815b36ef0daa2f1248f618a9c7e2f1ea62346
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6603
Community-CI: Broadcom CI
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
spdk_bdev_part_base_construct_ext is called by spdk_bdev_part_base_construct.
there's already an example if a base bdev is an abnormal case we need to do
spdk_bdev_part_base_free to close the base bdev.
From constructing a successful part base bdev util finishing vbdev_split_create.
there're except failure cases to exit.
if a bdev is open and when runs into bdev_unregister_unsafe, it's in an EBUSY
state, bdev will not finished from the list. That spdk_io_device_unregister
will no be executed in spdk_bdev_unregister.
That a subsytem with it callback which is app_reactors_stop can't be finished.
Signed-off-by: yidong0635 <dongx.yi@intel.com>
Change-Id: Idc156557f936da5f2618421ede2619f1f2e29c51
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6599
Community-CI: Broadcom CI
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Use the same style compared the code in posix_sock_close.
Thus if we cannot close sock->fd, i.e., we leak the fd,
but we can still free the memory related with uring sock.
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: Id2f0e8a2c7065f100c2b009e76a49b528fd221b6
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6539
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
In nvme_bdev_ctrlr_create, calloc will be called with a zero size
allocation request if the number of namespaces is zero. The behaviour
is implementation defined if the size of the space requested is zero -
calloc will either return a pointer that mustn't be dereferenced, or
NULL. If NULL is returned, the nvme_bdev_ctrlr_create will fail.
Only call calloc if there are a non-zero number of namespaces.
Otherwise, leave the namespaces pointer with a NULL value. All
references to namespaces[] are either known to be safe, or occur
in the context of looping through the namespaces which will be
skipped if the count is zero. The exception to this is in
vbdev_opal_create, where an assert has been added to match
equivalent code in bdev_ocssd_create_bdev.
Tested by running unit tests on a system that returns a null pointer
for a zero size allocation.
Signed-off-by: Nick Connolly <nick.connolly@mayadata.io>
Change-Id: I058b0683fd9b3a20bf90e54db93ca48b9bb4e40e
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6551
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: sunshihao <sunshihao@huawei.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
The statement causes this issue is:
assert(group_impl->num_removed_socks < MAX_EVENTS_PER_POLL);
The call trace is:
The previous solution is:
commitid with: e71e81b631
But with this solution, it will always add the sock
into the removed_socks list even if it is not under polling
context by sock_group_impl_poll_count. So it will exceed the size of
removed_socks array if sock_group_impl_poll_count function will not be
called. And we should not use a large array, because it is just a workaround,
it just hides the bug.
So our current solution is:
1 Remove the code in sock layer, i.e., rollback the commit
e71e81b631. This patch is
not the right fix. The sock->cb_fn's NULL pointer case is
caused by the cb_fn of write operation (if the
spdk_sock_group_remove_sock is inside the cb_fn). And it is not
caused by the epoll related cache issue described in commit
"e7181.." commit, but caused by the following situation:
(1)The socket's cb_fn is set to NULL which is caused by
spdk_sock_group_remove_sock by the socket itself
inside a call back function from a write operation.
(2) And the socket is already in the pending_recv list. It is
not caused by the epoll event issue, e.g., socket A changes Socket B's
cb_fn. By the way, A socket A should never remove a socket B from a polling group.
If it really does it, it should use spdk_thread_sendmsg to make sure
it happens in the next round.
2 Add the code check in each posix, uring implementation module.
If sock->cb_fn is NULL, we will not return the socket to the active socks list.
And this is enough to address the issue.
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: I79187f2f1301c819c46a5c3bdd84372f75534f2f
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6472
Reviewed-by: Xiaodong Liu <xiaodong.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
When there is not enough memory needed to start cache, print
relevant message and inform about possible solutions.
Fixes#1763
Signed-off-by: Rafal Stefanowski <rafal.stefanowski@intel.com>
Change-Id: I911dcb4e86ad7e3f2e2b051678736d4e88869db5
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6078
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
When multipath is supported, nvme_bdev will be got via bdev_subsystem.
To make such change transparent, add a helper function
nvme_bdev_ns_to_bdev() and use it for some cases.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: Ia5ad0b87178bc739ba0a75789f9f26b860241a04
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5801
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: <dongx.yi@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
By the recent changes, the linked list nvme_ns->bdevs has only a
single bdev at the maximum. Hence replace it by the pointer
nvme_ns->bdev, and remove the linked list pointer nvme_bdev->tailq.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: Ib976e15bb128ba8479070b58e5f4c43fb9dcf479
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6230
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
By the last changes, not only standard namespace but also ocssd
namespace has only one nvme_bdev, and standard namespace processes
only the head of nvme_ns->bdevs.
This patch changes the common and standard namespace specific
part to process only the head of nvme_ns->bdevs.
The following patch will replace the linked list nvme_ns->bdevs
by the pointer nvme_ns->bdev.
Add a particular error case that nvme_bdev is failed to create even
if ctrlr has one namespace. If ctrlr has one namespace but the
corresponding bdev is failed to create, nvme_ns->populated should
be false and hence nvme_ns->bdevs should not be accessed. However
the code had not assumed such case.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I5495882fad8c8a012305177179a46d4373ba75f5
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5800
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Remove range instance from struct bdev_ocssd_range, and additionally
remove range pointer from struct bdev_ocssd_create_ctx.
Then remove the definition of struct bdev_ocssd_range.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I5cb38d17ff2ec852878b481b8974983922fe25d2
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6225
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
The range parameter was deprecated from bdev_ocssd_create(), and
hence one ocssd_bdev can be created per one ocssd_ns now.
Hence process only one ocssd_bdev per one ocssd_ns. The following
patches will replace a linked list nvme_ns->bdevs by a pointer
nvme_ns->bdev.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I824d26f4d2f73326238a6426634e28cfd655a04d
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6224
Community-CI: Broadcom CI
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
