The following patches will support data corruption.
For write I/O, data corruption will be injected before submission,
and for read I/O, data corruption will be injected after completion.
To do these cleanly, use switch-case and reorder to process error
injection at submission.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I3b830b4331cb4c7d0794a555957cdcc73902c14f
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/15026
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Krzysztof Karas <krzysztof.karas@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Community-CI: Mellanox Build Bot
If we use error bdev in general use cases, the upper layer may submit
I/O commands other than read, write, unmap, or flush.
However, before this patch, the upper layer could submit only read,
write, unmap, and flush.
To improve the usability of error bdev, pass thorugh I/Os other than
read, write, unmap, or flush.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Ia642b13771f42505055f1372733825153085b805
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/15027
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@nvidia.com>
Reviewed-by: Krzysztof Karas <krzysztof.karas@intel.com>
This will make it easier to add more parameters for error injection.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Ie5b22c31b5ba9d8c256d369213fa8fb4b985fa26
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/15025
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@nvidia.com>
Reviewed-by: Krzysztof Karas <krzysztof.karas@intel.com>
This is a small clean up.
Use custom JSON decoders for the io_type and error_type parameters
in the bdev_error_injection_error RPC.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I528fe4a31fac7eddb8ec33594b90e107d71693be
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/15024
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@nvidia.com>
Reviewed-by: Krzysztof Karas <krzysztof.karas@intel.com>
If during a channel creation, an error happens, due to incorrect parameters
e.g. wrong pool / container name, or some other internal DAOS errors (like
reaching CART context limit), bdev_daos_io_channel_create_cb() signals
about such errors, however, spdk_io_device_register() does not takes them into account.
The device creation succeeds, returning successful RPC response and leaving bdev
in the bdev lists but it's completely unusable and not amendable.
This patch tries to detect it early and return an RPC error on failure.
Signed-off-by: Denis Barakhtanov <denis.barahtanov@croit.io>
Change-Id: I04758e6243566b4e619a1420aa7c01f6041441a6
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/15168
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
For write, verify DIF/DIX before submission and for read, verify
DIF/DIX after successful completion.
As same as the NVMe bdev module and the NULL bdev module, DIF/DIX
verification is done based on the DIF type and DIF insert/strip is
not supported.
In near future, the bdev I/O APIs bring an I/O flag to the underlying
bdev and the malloc bdev module will be able to decide DIF/DIX
verification based on the I/O flag.
One important feature is to setup protection information when
creating a malloc disk. Otherwise, all initial reads will fail
if protection information is enabled.
For users, add some explanation about the dif_type parameter
into doc/jsonrpc.md.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I93757b77c03cade766c872e418bb46d44918bee2
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14985
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
The malloc bdev module supports both of interleaved and separated
metadata in this patch.
Different from the NULL bdev module, opts->block_size is a data block
size and a block size is caculated internally as a sum of
opts->block_size and opts->md_size if opts->md_interleave is true, or
opts->block_size otherwise. This will be more intuitive. Additionally,
opts->md_size accepts only either of 0, 8, 16, 32, 64, or 128.
Protection information (T10 DIF/DIX) will be supported in the
following patches.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Icd9e92c8ea94e30139e416f8c533ab4cf473d2a8
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14984
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Community-CI: Mellanox Build Bot
Define a options structure, malloc_bdev_opts, and use it directly for
the bdev_malloc_create RPC. To do this, bdev_malloc.h includes
bdev_module.h instead of bdev.h to have the definition of the struct
spdk_uuid, and the struct malloc_bdev_opts has a instance of struct
spdk_uuid. Clean up file inclusion together. Furthermore, use
spdk_uuid_copy() to copy uuid from the malloc_bdev_opts to the malloc
disk rather than the = operator, and remove a duplicated size check.
These are helpful to add more parameters for creation.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Ief25f12586c21b1666180ce10cfc6256ede8eba9
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14982
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
If we use a custom decoder for malloc disk's uuid for the
bdev_malloc_create RPC, the code is simplified. Furthermore,
when we add an options structure, we will be able to include
the options structure into struct rpc_construct_malloc directly.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Ib36fa628569f973218f2cc5ce65a51181cd9fb71
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/15125
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
spdk_nvme_probe_poll_async() can only return 0 or -EAGAIN.
But the code currently indicates that other values might be
possible - so change the code to remove that confusion.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I8e7b2e1183f6650043dd751d610d434d626efa7e
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/15110
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Change 'bdev_nvme_compare_trids' to name 'bdev_nvme_check_secondary_trid'
and 'bdev_nvme_compare_namespaces' to 'bdev_nvme_check_secondary_namespace'
which better explain the aim of these function.
RPC bdev_nvme_attach_controller only response "invalid parameter"
if the path is not sucessfully added due to not meeting the conditions
in bdev_nvme_compare_trids. Add more log.
Change-Id: If9ba6ec1d397d49689aa2ed6ab74fbfa96e16afa
Signed-off-by: Richael Zhuang <richael.zhuang@arm.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14251
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
A new `oclass` parameter allow to specify DAOS DFS object class that is
responsible for data redundancy and protection.
Examples of the object classes could be found here: https://github.com/daos-stack/daos/blob/master/src/include/daos_obj_class.h
The default value is OC_SX for the max IOPS.
Signed-off-by: Denis Barakhtanov <denis.barahtanov@croit.io>
Change-Id: Ia48681832458c2266eb7c3bcae0df2055e59e309
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/15006
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Support Aio bdev 'readonly' option in RPC call. The read-only flag
can be dumped from the bdev info. Any writes on a read-only aio bdev
will be fail.
Signed-off-by: Yuhua <yuhua@smartx.com>
Change-Id: I939f72479f8953a3678a8df3083ecce0f96844fb
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14955
Reviewed-by: GangCao <gang.cao@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
In a use case, a custom sock module supports zero copy for read.
The custom sock module wants to keep the recv_buf_size to be sufficiently
large, for example 16MB. However, most upper layers overwrite the recv_buf_size
by a smaller value via spdk_sock_set_recvbuf() later. This is not desirable.
To fix the described issue, change the meaning of impl_opts->recv_buf_size
to be the minimum size, and spdk_sock_set_recvbuf() uses the maximum value
among the requested size, g_spdk_sock_impl_opts->recv_buf_size, and
MIN_SO_RCVBUF_SIZE.
We may have to change the code to initially create a socket. However,
for most cases, the upper layer calls spdk_sock_set_recvbuf() anyway.
Hence this fix will be minimal and enough.
For the use case, it is enough to change recv_buf_size of the posix sock
module. However, the custom sock module may support I/O uring in future.
Hence, change I/O uring sock module together.
Additionally, for consistency, change the meaning of impl_opts->send_buf_size
to be the minimum size, and spdk_sock_set_sendbuf() uses the maximum value
among the requested size, g_spdk_sock_impl_opts->send_buf_size, and
MIN_SO_SNDBUF_SIZE.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I051ba7cb50bc9dcad229e922198b04fe45335219
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14915
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Community-CI: Mellanox Build Bot
If the module stop handler is asynchronous we must wait until it
finishes before unregistering the io_device.
Change-Id: I149b716d9f4b0c1680b3e43b395fc9ec5b90d70c
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14717
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Implement support for full stripe writes without parity calculation.
The size and alignment of write IOs must be a multiple of full stripe
size. To reflect this, the raid bdev's write_unit_size is set
accordingly. We rely on the bdev layer to split larger IOs based on
that.
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Change-Id: I6940280ad870f3bd678fd19346b06ba4bdadd52e
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/7702
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Let the raid modules create their own IO channels by implementing the
get_io_channel callback.
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Change-Id: Id4f6c90721474edd70a6e987c67f8f774737da27
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/7700
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Move device cleanup to spdk_io_device_unregister() callback. This fixes
a case when the device would be freed before its last io channel was
closed, leading to use after free condition.
Repurpose raid_bdev_free() to actually free the bdev.
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Change-Id: Ib667b4d5ac1b34a0f2dda69f6b0775d9363dbfee
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11398
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Depend on bdev layer to send down chunk sized IOs, based on
optimal_io_boundary.
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Change-Id: Iec45f4917117d35c3a9e807940e49091dfcba870
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/7699
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Here we use vfu_tgt library and emulate a virtio-scsi device
as the next use case.
Compared with vhost-user-scsi, the packed ring is supported with this
patch.
Example usage method:
1. scripts/rpc.py bdev_malloc_create -b malloc0 $((512)) 512
2. scripts/rpc.py vfu_virtio_create_scsi_endpoint vfu.0 --cpumask 0x1 --num-io-queues=4 \
--qsize=128 --packed-ring
3. scripts/rpc.py vfu_virtio_scsi_add_target vfu.0 --scsi-target-num=0 --bdev-name malloc0
4. Start QEMU with '-device vfio-user-pci,socket=/spdk/vfu.0'
Change-Id: I8f35d1d21aaec34844d6ddb59dc997a64f141179
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12673
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Here we use vfu-tgt library and emulate a virtio-blk device
as the first use case of vfu-tgt library.
Usage example with QEMU:
1. build/bin/spdk_tgt
2. scripts/rpc.py bdev_malloc_create -b malloc0 $((512)) 512
3. scripts/rpc.py vfu_virtio_create_blk_endpoint vfu.0 --bdev-name malloc0 \
--cpumask=0x1 --num-queues=2 \
--qsize=256 --packed-ring
4. Start QEMU with '-device vfio-user-pci,socket=/spdk/vfu.0'
Change-Id: I45e45360c669584583b0b8a3f83250ab6c48efec
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12315
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Community-CI: Mellanox Build Bot
Previously SPDK use libvfio-user library to provide emulated NVMe
devices to VM, but it's limited to NVMe device type only. Here we
add SPDK vfu_target library abstraction based on libvfio-user which
supports more PCI device types.
We will add virtio-blk and virtio-scsi devices emulation based on
vfu_tgt library in following patches, actually this library can
support NVMe emulation too, due to the fact that the NVMe emulation
is already exist, so we will keep the NVMe emulation which based on
libvfio-user directly as it is.
Change-Id: Ib0ead6c6118fa62308355fe432003dd928a2fae9
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12597
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Fix for the issue 2702.
spdk_bdev plugin crashes while handling xnvme devices.
xnvme_queue_term gets invoked after the bdev_xnvme_free executes and
frees the xnvme structures. xnvme_queue_term references the xnvme->dev
structure, after it is freed, resulting in a segmentation fault.
Implemented a bdev_xnvme_destruct_cb, that frees the xnvme structures
after the xnvme_queue_term invokes and the device is fully unregistered.
Signed-off-by: Krishna Kanth Reddy <krish.reddy@samsung.com>
Change-Id: I9338c84baf4b61ec2e0d324e67bfefcb96485156
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14680
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Identify and properly handle conventional zones (in smr drives) by using
zone type and WP state. Bdevs supporting zoned devices(like uring, nvme and
vbdev_zone_block) now update the zone type information. As a result, the
fio plugin now uses this info instead of hard coding the zone type.
Also adds new WP state(ZONE_STATE_NOT_WP) for handling zones w/o WP.
Signed-off-by: Indraneel M <Indraneel.Mukherjee@wdc.com>
Change-Id: If031e0742d68c55c35e95ddc33d478939bbd52fe
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14572
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Community-CI: Mellanox Build Bot
Fixes#2708 for io_uring_cmd.
This is a minimal approach to enabling polling for bdev_xnvme. Ideally,
more would be done, however, there are several blockers to remove when
doing so. Thus, this commit only enables polling on
io_mechanism=io_uring_cmd.
Signed-off-by: Simon A. F. Lund <simon.lund@samsung.com>
Change-Id: Ifa604b52bb2b924fab4b559fae06f26a3574db42
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14679
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
In prep for upcoming iovec based compression/decompression patches.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I413493f764bead9e56266e488b74f8bca979e225
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14633
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>
In prep for adding both src and dst iovec support for compression.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I704b8d2bd459de03deb7f8ee45d76261910a3727
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13746
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>
A new parameter io_drain_timeout has been added to spdk_bdev
structure. If this value is unset, the bdev reset behavior
does not change.
The io_drain_timeout controls how long a bdev reset must wait for IO
to complete prior to issuing a reset to the underlying device.
If there is no outstanding IO at the end of that period, the reset
is skipped.
Change-Id: I585af427064ce234a4f60afc3d69bc9fc3252432
Signed-off-by: Krzysztof Karas <krzysztof.karas@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14501
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
This is consistent with the use of terms in other parts of SPDK and fits
with the code living under module/
Change-Id: If182f7cf2d160d57443a1b5f24e0065f191b59b2
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13919
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
After a controller was hot-removed, if a reset sequence started to
the controller, spdk_nvme_ctrlr_disconnect() failed and caused core
dump in debug mode.
When implemented, how to cause the failure and how to process the
failure were not clear. Hence assert was added to detect the
failure.
We know how we cause the failure now. Let's handle the failure
appropriately.
If spdk_nvme_ctrlr_disconnect() fails, we are on the nvme_ctrlr->thread.
Hence call bdev_nvme_reset_complete() with failure immediately.
Even if spdk_nvme_ctrlr_disconnect() completes synchronously, the
completion callback is executed asynchronously when polling an adminq.
Hence set the completion callback only if spdk_nvme_ctrlr_disconnect()
succeeds.
Fixes issue #2632
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I11f61853aba9eca2515592f964a291e59def7247
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13892
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Krzysztof Karas <krzysztof.karas@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
By a recent improvement, failure of I/O qpair creation is ignored
if the nvme_ctrlr is being reset or scheduled to reconnect.
However, failure of I/O qpair creation is not ignored if a new I/O
channel is allocated. It is normal to allocate a new I/O channel
when a nvme_ctrlr is being reset or scheduled to reconnect.
Fix this bug by relaxing the condition to ignore the result of
bdev_nvme_create_qpair() to if reconnect_delay_sec is non-zero and
bdev_retry_count is non-zero.
If reconnect_delay_sec is non-zero, reconnect will be tried sooner
or later, and if bdev_retry_count is non-zero, submitted IOs will be
queued until it succeeds.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Icf26e1ea65d292f9b8d24966abe25907d2cc33ec
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14446
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Add gathering of some performance counters and RPC for printing them.
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
Change-Id: I2e77d37fb66459240ff2e241f2b1f77c60f4eef4
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13390
Community-CI: Mellanox Build Bot
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>
config_json callback is used to preserve all
bdevs that were created during application runtime.
xnvme bdev module was missing this callback.
While here, io_mechanism is now saved in the bdev_xnvme
structure for reference in the config_json.
Haven't seen an option to pull this from xnvme itself.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I1a88ad2bb761f589d214fec8f0690c38572824d6
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14116
Reviewed-by: Michal Berger <michal.berger@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Krzysztof Karas <krzysztof.karas@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Signed-off-by: Konrad Sztyber <konrad.sztyber@intel.com>
Change-Id: Ifba581c92269c675b03b67586b1fea0b096f9e95
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14511
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Krzysztof Karas <krzysztof.karas@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
If res = 0, do not pass this value directly to bdev io completion function,
otherwise it will be treated as a successful completion.
Fix issue: 2679
Signed-off-by: jun.ran <ranjunsh@163.com>
Change-Id: Ice525de1bb5b0ada9ac30d3b59e5c731419efe2f
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14364
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Xiaodong Liu <xiaodong.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Two functions have been added to implement bdev io type
SEEK_DATA and SEEK_HOLE. Blob functions to find next
[un]allocated io_unit are used
Signed-off-by: Damiano Cipriani <damiano.cipriani@suse.com>
Change-Id: Ic3be3c0c86bd3010a23ba2681f0f00c62abcaaba
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14362
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 will allow to have context per connection.
And free context when connection closes.
Fixes#2689
Change-Id: Ic4e9adfa3f1bd8574b9ccf75ff42c4f3bd442b26
Signed-off-by: Boris Glimcher <Boris.Glimcher@emc.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14443
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Adding `psk` field to `spdk_nvme_ctrlr_opts`
Adding `psk` parameter to `bdev_nvme_attach_controller` RPC
Change-Id: Ie6f0d8b04ce472e6153934e985c026acded6cdfc
Signed-off-by: Boris Glimcher <Boris.Glimcher@emc.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14046
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
the calculation (offset_blocks + num_blocks - 1) didn't check if num_blocks is 0
which it is in case of flush.
in flush case, offset_blocks was also 0, so we got (-1) and we got big unsigned number.
just replace it with (offset_blocks + num_blocks - (num_blocks > 0)) to solve the issue.
NOTE this is only fixing the wrong math, there might be more issues that outside the scope of this commit
for example, if flush(x, y) called, no bdev implementation really use those values, and all bdevs just flush the whole disk
also, a convention is that fluch(0, 0) should flush it all (but like I said before, all bdevs flush the whole disk anyway)
Change-Id: I7e991653bc3050349dc155365b2c37ecc2d6b24c
Signed-off-by: Amir Haroush <amir.haroush@huawei.com>
Signed-off-by: Shai Fultheim <shai.fultheim@huawei.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13579
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Trim is now also available as a management operation via RPC.
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Change-Id: I05b778a611e9809a14bfed50b01986bb4649a35c
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13379
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>
Adds ability to send trim commands to FTL - only 4MiB aligned requests (both
for offset and length of request) will be processed. During a trim
operation an L2P page (containing 1024 4B entries, 1 per user LBA; which
is where the 4MiB alignment comes from) will be marked as unmapped.
After this point any L2P access to that page will actually set the
entries themselves as FTL_ADDR_INVALID. This is done to make the trim as
fast as possible, since for large requests it's probable that most of
the L2P pages aren't actually in DRAM.
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Change-Id: I4a04ee9498a2a6939af31b06f2e45d2b7cccbf19
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13378
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>
the current timeout parameter of the interface bdev_iscsi_set_opts is
duplicated with the timeout parameter of the JSONRPCClient parameter,
which may cause the iscsi timeout and JSONRPCClient parameters to
overwrite each other.
Signed-off-by: gongwei <gongwei833x@gmail.com>
Change-Id: I96604a7e1a495ac2e99518812297230680df42fd
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14306
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 <smatsumoto@nvidia.com>
These checked parameters are necessary themselves even for single path
configuration.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Ie1eb2f51eeec1dbc634c6bae462a41d4c209d6ac
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12052
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Dong Yi <yidong0635@126.com>
Community-CI: Mellanox Build Bot
Clean up the code by using a custom decoder. Use multipath mode to
follow doc/nvme_multipath.doc.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Ie1f4109dae3a2929dcf933939a9c1b67bece0caf
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12051
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Community-CI: Mellanox Build Bot
Previously, reset just set a long timer to wait until all inflight
I/Os complete.
As already noted as a TODO item, check if any I/O is still inflight
before completing the reset by using a new API
spdk_bdev_get_current_qd().
The RBD bdev module can count outstanding I/Os itself but is not
efficient.
Signed-off-by: liu-darong <liu.darong@xsky.com>
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Iecaf90b06cae8e21198ec3822b978b54f5404d2b
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13945
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Dong Yi <yidong0635@126.com>
Reviewed-by: Xiaodong Liu <xiaodong.liu@intel.com>
Reviewed-by: GangCao <gang.cao@intel.com>
SSL_CTX isn't created in accept(), but when a socket on which accept()
is called is created, so it shouldn't be freed when accept() fails, as
this makes the socket unusable (any subsequent operations using SSL_CTX
would be using freed memory).
This caused the segfaults reported in issue #2681, where the second
connection would crash the application.
Signed-off-by: Konrad Sztyber <konrad.sztyber@intel.com>
Change-Id: I1a01a748c5a34ce3dd0fd3c557b860c0ff314b85
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14355
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Dong Yi <dongx.yi@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Bumped up the priority of the posix and uring sock implementations to
make sure they are selected before SSL, when no impl is explicitly
specified.
Fixes#2681
Signed-off-by: Konrad Sztyber <konrad.sztyber@intel.com>
Change-Id: Ic8e1e2e13f7bce7ccd746f66087e348677df28d9
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14354
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
hexlify and unhexlify utils from vbdev_crypto.h have been moved so that
they could be included and reused outside of vbdev_crypto module.
Signed-off-by: Blachut, Bartosz <bartosz.blachut@intel.com>
Change-Id: Ia074250176907f4803b84024239ecd4e9d8a5fc1
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14191
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Jacek Kalwas <jacek.kalwas@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
`v2c` helper function returns -1 on error, hovewer its return type was
char - which may cause an overflow situation on architectures where char
is interpreted as unsigned char, which was reported by the CI.
The return type has been changed to signed char.
Signed-off-by: Blachut, Bartosz <bartosz.blachut@intel.com>
Change-Id: I84f5784b2de7d681f78c69b5f3646e851e8dee88
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14273
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Jacek Kalwas <jacek.kalwas@intel.com>
The word engine was both used (interchangeably with module) to refer to
the things that plug into the framework and to the framework itself.
This patch eliminates all use of the word engine that meant the
framework. It leaves uses of the word that meant "module".
Change-Id: I6b9b50e2f045ac39f2a74d0152ee8d6269be4bd1
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13918
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
spdk_accel_module_finish
Also move it into the internal header that defines the interface used by
modules.
Change-Id: I3aeb41e643f27a69556099cb8d166f64c9e5d67f
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13917
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Krzysztof Karas <krzysztof.karas@intel.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Writing to unallocated cluster triggers copy-on-write sequence. If
this cluster is backed by zeroes device we can skip the copy part. For
a simple thin provisioned volume copy this shortcut is already
implemented because `blob->parent_id == SPDK_BLOBID_INVALID`. But this
will not work for thin provisioned volumes created from snapshot. In
this case we need to traverse the whole stack of underlying
`spdk_bs_dev` devices for specific cluster to check if it is zeroes
backed.
This patch adds `is_zeroes` operation to `spdk_bs_dev`. For zeroes
device it always returns 'true', for real bdev (`blob_bs_dev`) always
returns false, for another layer of `blob_bs_dev` does lba conversion
and forwards to backing device.
In blobstore's cluster copy flow we check if cluster is backed by
zeroes device and skip copy part if it is.
Signed-off-by: Evgeniy Kochetov <evgeniik@nvidia.com>
Change-Id: I640773ac78f8f466b96e96a34c3a6c3c91f87dab
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13446
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
L2P cache allows for partial storing of L2P in memory, paging in and out
as necessary, lowering the total memory consumption.
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Change-Id: I727fec9d2f0ade4ca73e872d62a2ec10cfdb0a88
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13353
Community-CI: Mellanox Build Bot
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>
Added a new RPC, vmd_rescan, that forces the VMD driver to do a rescan
of all devices behind the VMD. A device that was previously removed via
spdk_vmd_remove_device() will be found again during vmd_rescan.
Signed-off-by: Konrad Sztyber <konrad.sztyber@intel.com>
Change-Id: Ide87eb44c1d6d524234820dc07c78ba5b8bcd3ad
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13958
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tom Nabarro <tom.nabarro@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Added new RPC, vmd_remove_device, that allows users to remove a PCI
device managed by the VMD library simulating a hot-remove.
Signed-off-by: Konrad Sztyber <konrad.sztyber@intel.com>
Change-Id: Ifb84818ce8d147d1d586b52590527e85fe9c10de
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13957
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tom Nabarro <tom.nabarro@intel.com>
The new name is consistent with the naming scheme of
<subsystem>_<action> that all of our other RPCs use.
Signed-off-by: Konrad Sztyber <konrad.sztyber@intel.com>
Change-Id: I2cae7af5715add8eba26501cd192a6ac4884ec69
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13952
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Tom Nabarro <tom.nabarro@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This makes sure that if the app is killed before the subsystems have
been initialized (i.e. before framework_start_init is called), we don't
leak resources. Before this change, we initialized the VMD library and
registered the hotplug poller from the context of the RPC and we only
freed it in subsystem's fini(). However, if subsystems were never
initialized, we never freed them.
Signed-off-by: Konrad Sztyber <konrad.sztyber@intel.com>
Change-Id: I56b746c46b94135ef56a478c5f80194ebe51a942
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13951
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tom Nabarro <tom.nabarro@intel.com>
bdev_nvme_create_bdev_channel_cb() did not initialized the multipath
policy of the newly created channel. 0 was active-passive and hence
multipath policy was always initialized to active-passive.
Fix the bug and add unit tests for verification.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I6e44108740da4b9ff72311ae4b5500558c65c5c3
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14225
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: Aleksey Marchuk <alexeymar@nvidia.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Admin passthrough supported retry and failover as same as I/O by
using the bdev_retry_count. However, doing retry or failover for
admin passthrough may have unexpected side effects and its value
is not clear. The safest way is to limit retry and failover for I/O.
If we need to support retry and failover for admin passthrough,
restore the code and add a new option bdev_admin_retry_count.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I680513a40a80041f6ea6f546c74c672f2a81812d
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14227
Community-CI: Mellanox Build Bot
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.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>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Abort is used usually for error cases. When abort is done, controller
may not be available. At completion, controller availability was checked
prior to aborted by request. Hence if abort is done when controller is
not available, the aborted request is retried.
Fix the case in this patch.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I3e7b356797dd50faed0e5113f6f7a47fea26d9cc
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14098
Community-CI: Mellanox Build Bot
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Michael Haeuptle <michaelhaeuptle@gmail.com>
Reviewed-by: Dong Yi <dongx.yi@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Adds API for fast shutdown - the ability for FTL to skip most
of the metadata persists made during clean shutdown, and relying
on their representation in shared memory instead. This allows for
faster update of SPDK (or just FTL, assuming no metadata changes),
with downtime reduction from 2-5 seconds to 500-1000 ms (for
14TiB+800GiB base and cache drives).
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
Change-Id: I5999d31698a81512db8d5893eabee7b505c80d06
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13348
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>
Enables the use of uring bdev with ZNS devices.
Uses BLKXXXZONE ioctls for implementing the zone operations.
Signed-off-by: Indraneel M <Indraneel.Mukherjee@wdc.com>
Change-Id: I440e316138182e25d89eb7224932e19bef9a005f
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13550
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>
Here no need to define a variable only return.
Signed-off-by: yidong0635 <dongx.yi@intel.com>
Change-Id: I6babb8ef94a9b111d5522f1e782cbd095c7da83a
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14207
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: wanghailiang <hailiangx.e.wang@intel.com>
Reviewed-by: Krzysztof Karas <krzysztof.karas@intel.com>
If we use dd or bdevperf to run the xnvme bdev.
the destruct function of xnvme misses to free
dev while destructing bdev_xnvme.
xnvme_dev_close does free dev, and we can put it
in bdev_xnvme_free.
Meanwhile fixing cleanup for delete_xnvme_bdev.
This rpc for delete_xnvme_bdev doesn't really cleanup
to remove the bdev from the list and free the names.
Fixes issue: #2654
Signed-off-by: yidong0635 <dongx.yi@intel.com>
Change-Id: I1c493cb8130b012d891ba8ee90cd0bfb127207d4
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14177
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: GangCao <gang.cao@intel.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Found while debugging issue #2596, unfortunately this
is not the root cause of that issue.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I27501e283ce7c9bf7a431e8b48842c83f80792c8
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14165
Community-CI: Mellanox Build Bot
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: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Dong Yi <dongx.yi@intel.com>
This commmit introduces a new bdev type backed up by DAOS DFS.
Design wise this bdev is a file named as the bdev itself in the DAOS POSIX
container that uses daos event queue per io channel.
Having an event queue per io channel is showing the best IO throughput.
The implementation uses the independent pool and container connections per
device's channel for the best IO throughput.
The semantic of usage is the same as any other bdev type.
To build SPDK with daos support, daos-devel package has to be installed.
The current supported DAOS version is v2.X, please see the installatoin and
setup guide here: https://docs.daos.io/v2.0/
$ ./configure --with-daos
To run it, the target machine should have daos_agent up and running, as
well as the pool and POSIX container ready to use, please see the
detailed requirements here: https://docs.daos.io/v2.0/admin/hardware/.
To export bdev over tcp:
$ ./nvmf_tgt &
$ ./scripts/rpc.py nvmf_create_transport -t TCP -u 2097152 -i 2097152
$ ./scripts/rpc.py bdev_daos_create daosdev0 <pool-label> <cont-label>
1048576 4096
$ ./scripts/rpc.py nvmf_create_subsystem nqn.2016-06.io.spdk1:cnode1 -a -s
SPDK00000000000001 -d SPDK_Virtual_Controller_1
$ ./scripts/rpc.py nvmf_subsystem_add_ns nqn.2016-06.io.spdk1:cnode1
daosdev0
$ ./scripts/rpc.py nvmf_subsystem_add_listener nqn.2016-06.io.spdk1:cnode1
-t tcp -a <IP> -s 4420
On the initiator side, make sure that `nvme-tcp` module is loaded then
connect drives, for instance:
$ nvme connect-all -t tcp -a 172.31.91.61 -s 4420
$ nvme list
Signed-off-by: Denis Barakhtanov <denis.barahtanov@croit.io>
Change-Id: I51945465122e0fb96de4326db742169419966806
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12260
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>
This allows mapping an nvme_request back to the
nvme_bdev_io.
This requires bumping up the max number of arguments per
tracepoint. 5 was previously chosen as max since it
exactly fit in 64 bytes (1 cacheline) when all
arguments were stored as uint64_t, but now that we
support uint32_t arguments we can afford extra
arguments when some of them are uint32_t. I've
bumped it to 8 so we can avoid having to touch
this value multiple times if we find some cases
where we need 7 or 8 args.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ie2ef5e59d10549860b47542e68c1c34efa63047f
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13995
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jacek Kalwas <jacek.kalwas@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
This will allow us to map spdk_bdev_io events
to nvme_request events coming in a future patch.
Since we pass the nvme_bdev_io to the nvme driver
(not the spdk_bdev_io), we need to add tracepoints
for the nvme_bdev_io so that spdk_trace can
do the spdk_bdev_io->nvme_bdev_io->nvme_request
mapping.
An alternative would have been to pass the spdk_bdev_io
as the cb_arg to the nvme driver, but that change
seemed to invasive, and I think we will find other
uses for the nvme_bdev_io events anyways.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Id7519e689b01875093359f41a1ca2af912061a8b
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13994
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Jacek Kalwas <jacek.kalwas@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
This is just a simple wrapper for now around the
calls to spdk_bdev_io_complete and its
nvme status variant. Upcoming patch will
add an spdk_trace_record to this function as well.
This avoids having to litter spdk_trace_record calls
in too many places.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Id2fb3aeb8b070ad6e09c1dfb9a30a61666a35688
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13993
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Jacek Kalwas <jacek.kalwas@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Community-CI: Mellanox Build Bot
The public interface of lib/accel is now include/spdk/accel.h
Change-Id: Id94f623a494eb1b524b060f4413f633073ea7466
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13916
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
These hold a pointer to the channel which eliminates the need to look it
back up in the completion path.
Change-Id: Ie4fc98d92d6434262e64b9483ef8b3b0591d764a
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13914
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
This expects the full size of the task for each module. This only worked
because the software module returned the right size.
Change-Id: I481cfad8b4bb9c3748301bdacd90e7f44fd2d878
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13913
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
These are 1:1 - they do not need to be separate objects.
Change-Id: I74ab52863f911d9be59ce98e1525302b5bd40846
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13910
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
This change fixes the lower performance issue of the
xNVMe BDEV for libaio and io_uring backends. There is no
drop in the performance for the io_uring passthrough backend.
Signed-off-by: Krishna Kanth Reddy <krish.reddy@samsung.com>
Change-Id: I21de1b49a534cfc642d1873ef623271063da6af8
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/14015
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Dong Yi <dongx.yi@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Note, this change only sets defaults for the ID/KEY,
more specific use cases like NVMe/TCP may set the ID and KEY on a per connection basis.
Also simplify PSK identity string, that isn't NVMe focused.
NVMe libraries using this will need to construct more complicated
identity strings and pass them to the sock layer.
Example:
rpc.py sock_impl_set_options -i ssl --psk-key 4321DEADBEEF1234
rpc.py sock_impl_set_options -i ssl --psk-identity psk.spdk.io
./build/examples/perf --psk-key 4321DEADBEEF1234 --psk-identity psk.spdk.io
./build/examples/hello_sock --psk-key 4321DEADBEEF1234 --psk-identity psk.spdk.io
Change-Id: I1cb5b0b706bdeafbccbc71f8320bc8e2961cbb55
Signed-off-by: Boris Glimcher <Boris.Glimcher@emc.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13759
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
If a nvme_ctrlr is unregistered while I/O path caches are clearing, the
unregistration would fail. This race was not considered for a case that
a nvme_ctrlr is unregsitered when adminq poller started I/O path cache
clearing.
As a bug fix, control ANA log page updating and I/O path cache clearing
separately by adding a new flag io_path_cache_clearing.
Fixes issue #2617
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Id58684caf9b2a10fa68bdb717288ff2bd799c3f7
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13924
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Dong Yi <dongx.yi@intel.com>
Instead of passing each parameter to create a module, just have the user
make one and pass it in. This makes it easier to change the module
definition later.
Change-Id: I3a29f59432a6f0773129d7b210fbc011175b2252
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13909
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Parameter `num_queues` for virtio_scsi PCI device means
maximum number of queues, it SHOULD include the `eventq`
and `controlq`, while for `vhost_user` RPC call, it means
the number of IO queues, so here we use it as `max_queues`
in lib/virtio and add the fixed number queues for `vhost_user`
SCSI device.
Also fix `vhost_fuzz` to get `num_queues` earlier than
negotiate the feature bits.
Change-Id: I41b3da5e4b4dc37127befd414226ea6eafcd9ad0
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13791
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>
VHOST_USER_F_PROTOCOL_FEATURES is used for `vhost_user` transport,
so unmask it for PCI devices.
Change-Id: If84d6c0ee7558886f14647dad07e41530e306206
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13790
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
While running into this function, even the subsystem can't be
destroyed due to error subsystem state, it's better to continue
the execution.
Continue to fix#2590, QEMU is stuck for the failure case, and
nvmf target should process such error because it may support other
normal subsystems at the same time.
Change-Id: Ib05e24996378b52070d2b760519f476f9b2d7e76
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/13839
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Dong Yi <dongx.yi@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
