RDMA READs can cause cmds to be submitted to the target
layer in a different order than they were received
from the host. Normally this is fine, but not for
fused commands.
So track fused commands as they reach
nvmf_rdma_request_process(). If we find a pair of
sequential commands that don't have valid FUSED settings
(i.e. NONE/SECOND, FIRST/NONE, FIRST/FIRST), we mark
the requests as "fused_failed" and will later fail them
just before they would be normally sent to the target
layer.
When we do find a pair of valid fused commands (FIRST
followed by SECOND), we will wait until both are
READY_TO_EXECUTE, and then submit them to the target
layer consecutively.
This fixes issue #2428 for RDMA transport.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I01ebf90e17761499fb6601456811f442dc2a2950
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12018
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
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>
To execute a callback function for each registered bdev or unclaimed
bdev, add new public APIs, spdk_for_each_bdev() and
spdk_for_each_bdev_leaf().
These functions are safe for race conditions by opening before and
closing after executing the provided callback function.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I59b702ffec7b4fc5e9779de5a3a75d44922b829b
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12088
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Bdev open/close will be done for each bdev when traversing the bdev
list. This patch is a preparation.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I2486bd823953fe020ed6106844877e1cf49d8a0d
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12126
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
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>
spdk_io_device_unregister() send message to call its callback. So to
make the following patches easier, consolidate g_bdev_mgr.mutex unlocks
to the end of spdk_bdev_close().
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Ib3b5c72be06e764918da30d7aa9fbc2ccd33956e
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12125
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
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>
Bdev open/close will be done for each bdev when traversing the bdev
list. This patch is a preparation.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I4e4fe6f1248176631a74c09585c931b21eb49d2b
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12124
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
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>
That will allow to pass ext opts to bdev layer
Since in ext API metadata is passed as part of ext IO opts
structure and ext opts can be NULL (e.g. upper layed used
regular API), in that case we use *blocks_with_md API
Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com>
Change-Id: I1bfb3fcb11bf42e100ecc7e4058087f12086db3a
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11048
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
memory domains
If bdev doesn't support any memory domain then allocate
internal bounce buffer, pull data for write operation before
IO submission, push data to memory domain once IO completes
for read operation.
Update test tool, add simple pull/push functions
implementation.
Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com>
Change-Id: Ie9b94463e6a818bcd606fbb898fb0d6e0b5d5027
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/10069
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
To unregister a bdev more correctly, we had to call
spdk_bdev_open_ext(), spdk_bdev_desc_get_bdev(), spdk_bdev_unregister(),
and then spdk_bdev_close(). This was correct but complicated.
Hence add a new public API spdk_bdev_unregister_by_name() which does
the whole correct sequence of bdev unregistration.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I9068d4ac49dca944436e0ba587308fd356dfef75
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12065
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
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>
The process of matching qpair to poll group is split into
two distinct parts that occur on different threads.
See spdk_nvmf_tgt_new_qpair().
This results in a race condition for TCP between spdk_sock_map_lookup()
and spdk_sock_map_insert(), which are called in spdk_nvmf_get_optimal_poll_group()
and spdk_nvmf_poll_group_add() respectively.
Fixes#2113
This patch picks a hint from nvmf_tcp for next poll group,
which is then passed down to spdk_sock_map_lookup().
When matching placement_id exists, but does not have
a poll group assigned - the hint will be used.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I4abde2bc9c39225c9f5dd7c3654fa2639bb0a27f
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/10271
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>
The rdma buffer for stripping DIF metadata is added. CPU strips the DIF
metadata and copies it to the rdma buffer, improving the rdma write
bandwith. The network bandwidth during 4KB random read test is increased
from 79 Gbps to 99 Gbps, the IOPS is increased from 2075K to 2637K.
Fixes issue #2418
Signed-off-by: Chunsong Feng <fengchunsong@huawei.com>
Change-Id: If1c31256f0390f31d396812fa33cd650bf52b336
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11861
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.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>
Currently the idxd driver requires VFIO so avoid unexpected errors
if someone tries without it (with UIO).
Temp workaround for issue #2316
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I430cd2193bc8dbd6939af7d0ca799832e7a73213
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11816
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
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: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
In the current behavior the iovcnt is lowered before sending it to
the next BDEV in the stack - however if the returned value is ENOMEM
(due to eg. not enough bdev requests in the pool), the request needs
to be returned to its original state, as it would be resubmitted with
skipped iov entries.
Signed-off-by: Kozlowski Mateusz <mateusz.kozlowski@intel.com>
Change-Id: I7240510a2ec04594b248f7347e86ac11ecfd26a0
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11976
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
When iovs are copied from bounce or to bounce, the bounce is usually
alloced from data_buf_pool for better performance, and is multi iovs
instead of a single buffer. Therefore, block-aligned bounce are
supported.
Signed-off-by: Chunsong Feng <fengchunsong@huawei.com>
Change-Id: If56b21d9e46c73d4c956c227bec33ddd0ab9745b
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11860
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: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Extract the code for DIF from nvmf_rdma_fill_wr_sgl() into
nvmf_rdma_fill_wr_sgl_with_dif().
Then clean up nvmf_rdma_request_fill_iovs() and
nvmf_rdma_request_fill_iovs_multi_sgl().
Additionally, this patch has a bug fix. nvmf_rdma_fill_wr_sgl_with_dif()
returned false if spdk_rdma_get_translation() failed. However, the
type of return value of nvmf_rdma_fill_wr_sgl_with_dif() is not bool
but int. The boolean false is 0 in integer. Hence in this case,
nvmf_rdma_fill_wr_sgl_with_dif() returned 0 even if it failed.
Change nvmf_rdma_fill_wr_sgl_with_dif() to return rc as is if
spdk_rdma_get_translation() returns non-zero rc.
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I71cc186458bfe8863964ab68e2d014c495312cd3
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11965
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: fengchunsong <fengchunsong@huawei.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
To help grep, use a standard sqid:%d style format for identifying queue
IDs.
Signed-off-by: John Levon <john.levon@nutanix.com>
Change-Id: Ib82c81939f85f9beb333a4db10d006524522a1d9
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11822
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Thanos Makatos <thanos.makatos@nutanix.com>
As a general utility function, move it up with the others.
Signed-off-by: John Levon <john.levon@nutanix.com>
Change-Id: I32881c01afd9819c889730d7c09163c95fbb827e
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11790
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Thanos Makatos <thanos.makatos@nutanix.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
For each queue, track its doorbell location individually, rather than
needlessly recalculating it every time we look up the doorbell value.
This will also greatly simplify shadow doorbell support.
Co-authored-by: Andreas Economides <andreas.economides@nutanix.com>
Signed-off-by: John Levon <john.levon@nutanix.com>
Change-Id: I6882d2f92ee2f2b2b90c54ee14e5f6b41ecca85d
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11789
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Separate nvme_rdma_process_event() into nvme_rdma_process_event_start()
and nvme_rdma_process_event_poll().
Use nvme_rdma_process_event_start() and nvme_rdma_process_event_poll()
in nvme_rdma_process_event() to ensure compatibility.
Change-Id: Idc960fab2540efec612dcf22f156acabd2e2874e
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/10594
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
It will be convenient for the following patches to return
negated errno directly.
Change-Id: Ic80181b2ee449946dd60ad0c97a325fd48b92231
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/10990
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
nvme_rdma_poll_events() gets the cm_channel pointer itself.
Before calling nvme_rdma_process_event(), we checks the
rctrlr is valid.
Hence we do not have to pass the cm_channel pointer to
nvme_rdma_process_event() via a parameter.
This simplifies the code and makes the following patches a little easier.
Change-Id: I03f095833469c5b64592264d63a592106d49e13b
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11167
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: Aleksey Marchuk <alexeymar@mellanox.com>
Replace nvme_fabric_qpair_connect() by nvme_fabric_qpair_connect_async()
and nvme_fabric_qpair_connect_poll().
The following is a detail.
Define state of the nvme_rdma_qpair and each rqpair holds it.
Initialize rqpair->state by INVALID at nvme_rdma_ctrlr_create_qpair().
_nvme_rdma_ctrlr_connect_qpair() sets rqpair->state to
FABRIC_CONNECT_SEND instead of calling nvme_fabric_qpair_connect().
Then the new function nvme_rdma_ctrlr_connect_qpair_poll() calls
nvme_fabric_qpair_connect_async() at FABRIC_CONNECT_SEND and
nvme_fabric_qpair_connect_poll() until it returns 0 at FABRIC_CONNECT_POLL.
nvme_rdma_qpair_process_completions() or
nvme_rdma_poll_group_process_completions() calls
nvme_rdma_ctrlr_connect_qpair_poll() if qpair->state is CONECTING.
This patter follows the TCP transport.
Change-Id: I411f4fa8071cb5ea27581f3820eba9b02c731e4c
Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11334
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
At this time only spdk_sock_map_insert() allocates
entries for the sock_map. Next patch will introduce it
to lookup too. So now this part is refactored out
to separate function.
This patch should not introduce any functional change.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I46ac88aedebffe0cbc1f4616dc1fcfaf7f950b05
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/10726
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
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>
spdk_sock_map_insert() allows for allocating a sock_map
entry, without assigning any sock_group.
This is useful for cases where placement_id determined
by the component using spdk_sock_map_*. See PLACEMENT_MARK mode.
Placement_id's are allocated first, then an empty one is found
using spdk_sock_map_find_free().
Since the above is a valid use case, then entry in sock_map
can exist without a group assigned. spdk_sock_map_lookup() has
to handle such cases, rather than trigger an assert.
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: Ia717c38fef5e71fe44471ea12f61a5548463f0cf
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/10725
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: wanghailiang <hailiangx.e.wang@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Dong Yi <dongx.yi@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
As we now only support a single WQ, there's no need for a teble of
them and no need to assert that the stride from WQ to WQ is the
same as the WQ struct size.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I205f36aae22070f532653726dd75249bbafbe3ef
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12081
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
This patch fixes the issue with custom nvme transport. It is possible
to register custom nvme transport with arbitrary name but it is not
usable because 'spdk_nvme_trid_populate_transport' call in probe
function will always set trstring to 'CUSTOM' and transport lookup
will fail.
Signed-off-by: Evgeniy Kochetov <evgeniik@nvidia.com>
Change-Id: I83fd24dd8732ac0a21e22435e0acff20ab0e7521
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/9557
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>
First in a series of patches that will enable multiple engines
to exist at once and choose the best one based on their priorities
and capabilites, the public API will no longer be needed.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: Ia87b83aa2263745a94a822a160b6e97bb2e0dc19
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11948
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
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>
In the compression operation we may have SGL input
if user's buffer is fragmented or less than chunk_size.
If the backing device doesn't support SGL input then
we should copy user's buffers into decomp_buffer
(including paddings if any).
In the decompression operation, if the backing device
doesn't support SGL output, we use a single output buffer
which is pointing to decomp_buffer. Once the operation
completes, we should copy the result into user's buffers.
Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com>
Change-Id: Ic7fddd38374bb6898256633eacd192dbaf36541a
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11970
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
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: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
R2Ts can cause cmds to be submitted to the target
layer in a different order than they were received
from the host. Normally this is fine, but not for
fused commands.
So track fused commands as they reach
nvmf_tcp_req_process(). If we find a pair of sequential
commands that don't have valid FUSED settings (i.e.
NONE/SECOND, FIRST/NONE, FIRST/FIRST), we mark the
requests as "fused_failed" and will later fail them
just before they would be normally sent to the target
layer.
When we do find a pair of valid fused commands (FIRST
followed by SECOND), we will wait until both are
READY_TO_EXECUTE, and then submit them to the target
layer consecutively.
This fixes issue #2428 for TCP transport.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I8a9e13690ecb16429df68ae41b16b439a0913e4e
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/12017
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
This opption allows the bdev_get_bdevs RPC to block until a bdev with
specified name appears. It can be useful, when a bdev is created
asynchronously and the exact moment at which it appears is not known.
For instance, with a discovery service, a bdev is created when a
namespace on a remote NVMeoF target is added, but it's not possible to
specify when that happens exactly.
Signed-off-by: Konrad Sztyber <konrad.sztyber@intel.com>
Change-Id: I6c1f974fba445376ca9d45aac2639202547410cc
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11960
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: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Some targets report they support log page offset,
but then fail GET_LOG_PAGE commands that specify
a non-zero offset, or report the wrong number
of discovery entries when reading more than the
discovery log page header but not the entire log
page.
So just revert to reading the entire discovery log
page, after we've read the header to know how big the
log page will be. This means that when we read the
log page initially (without the individual entries),
we need to save off the genctr, since it will get
overwritten when we read the log page again. We can
just store this in the discovery context, and compare
it to the genctr that we read with the whole log page.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I34929253312fed9924db58904a051f3979283730
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11478
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
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: Shuhei Matsumoto <smatsumoto@nvidia.com>
The driver always creates a single group containing all of the engines
and a single work queue.
Change-Id: I83f170f966abbd141304c49bd75ffe4608f5ad03
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11533
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
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 <smatsumoto@nvidia.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
It is not used for anything.
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Change-Id: I1d967b2d0e404756f7ceda98ddc4ee9017ec83f7
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11489
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
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 <smatsumoto@nvidia.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
It turns out that this can stay on the stack.
Change-Id: I961366307dae5ec7413a86271cd1dfb370b8f9f3
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11488
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
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 <smatsumoto@nvidia.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
initialization
We can make the structs do all of the offset math for us.
Change-Id: Ibe6d86c2abc58655c1354f1eb31091c95cfb283c
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11487
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
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 <smatsumoto@nvidia.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
These aren't ever accessed in the main I/O path, so we can read them in
whenever we need them and make the code a lot simpler.
Change-Id: Icfdbfe9f2d9db13f4d0d28b2b4103cd0c443bcf4
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11485
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
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 <smatsumoto@nvidia.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Assume any given WQ we are allocated can submit the full device allowed
queue depth.
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Change-Id: I044e1f70031ea83ae722ed285b84c06b3e5efb27
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11486
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
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 <smatsumoto@nvidia.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
This is no longer needed.
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Change-Id: I08c788ca0451e739804b568d613c1e52e071c61f
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11794
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>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Community-CI: Mellanox Build Bot
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
In vfio-user transport, whenever one IO is completed, it will trigger
an interrupt to guest machine. This cost quite some overhead. This patch
adds an adaptive irq feature to reduce interrupt overhead and boost
performance.
Signed-off-by: Rui Chang <rui.chang@arm.com>
Change-Id: I585be072231a934fa2e4fdf2439405de95151381
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11840
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
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>
responder_resources parameter of rdma cm tells remote
side how many outstaing RDMA_READ of atomic operations
local side can handle.
Previously it was adjusted on queue depth but that was
not correct since these parameters do not depend on
each other. Even with qdepth=1 remote side may send
several RDMA_READ operations per 1 IO request.
With this change we report responder_resources
equal to the maximum supported by RDMA device.
Linux kernel nvme rdma driver reports this value
in the same way.
Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com>
Change-Id: I77e5c2ead6269da44c32a75a9188429f50d32ae4
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11698
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Fill is sent in as a uint8, we need to populate the full uint64
input with the uint8 pattern or we'll get a miscompare. This is
how idxd was doing it, instead of adding the same code to ioat just
move it up a layer.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: Ia4aab1c6230f35ab88bb8a0e3b8e16dbd93007c7
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11947
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: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Functions _start_writev_request and _write_decompress_done
have very similar decomp_iov configuration code, the only
difference is whether to fill gaps with zeroes or with
decompressed data. Move this code to a common function,
that will reduce amount of changes in the next patch
Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com>
Change-Id: I14509a17a12156b25ceab85a98b4dbd6fb11c732
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11969
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
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>
Add NVME_CTRLR_ERRLOGs to nvme_ctrlr_process_init().
The main goal is to help with debugging #2201 issue.
Change-Id: I1ae6a9b30d6124dfe25eb7912402c37d476b0d4c
Signed-off-by: Krzysztof Karas <krzysztof.karas@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/10627
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Rename ->doorbells to ->bar0_doorbells. This will help avoid confusion
later with shadow doorbells.
Signed-off-by: John Levon <john.levon@nutanix.com>
Change-Id: Id432938cfeb3033e79dc6e1b491dad964227687a
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11788
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Thanos Makatos <thanos.makatos@nutanix.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
