bdevperf: Add hide_metadata option

Add hide_metadata option to the bdevperf tool as a global option.
Verify the option by adding a simple test with three DIF configuration.

Signed-off-by: Shuhei Ma

bdevperf: Add hide_metadata option

Add hide_metadata option to the bdevperf tool as a global option.
Verify the option by adding a simple test with three DIF configuration.

Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I20696a07fe8334473854377c38d2ac7516af494f
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/25145
Community-CI: Community CI Samsung <spdk.community.ci.samsung@gmail.com>
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: Jim Harris <jim.harris@nvidia.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>

show more ...

bdevperf: Get metadata config by not bdev but bdev_desc

This is another preparation to enable no_metadata option to each bdev.

Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I85

bdevperf: Get metadata config by not bdev but bdev_desc

This is another preparation to enable no_metadata option to each bdev.

Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I8514b59dfd225b0e5beb995029fdde561ae6ad9d
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/25144
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@nvidia.com>
Community-CI: Community CI Samsung <spdk.community.ci.samsung@gmail.com>
Reviewed-by: Jim Harris <jim.harris@nvidia.com>

show more ...

bdevperf: Store the result of DIF type check into job structure

md_check is used only in one place but the place is in the main I/O
path. md_check cannot change. If we store md_check into job struct

bdevperf: Store the result of DIF type check into job structure

md_check is used only in one place but the place is in the main I/O
path. md_check cannot change. If we store md_check into job structure,
we can save cycles.

Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Iff501539e04da5a5fece254b1e4698dd0c16b1a0
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/25459
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Community-CI: Community CI Samsung <spdk.community.ci.samsung@gmail.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Jim Harris <jim.harris@nvidia.com>

show more ...

bdevperf: g_main_thread calls bdev_open() instead of job->thread

The following patches will enable no_metadata option to each bdev
and use new APIs to get metadata configuration.

Hence, it will be

bdevperf: g_main_thread calls bdev_open() instead of job->thread

The following patches will enable no_metadata option to each bdev
and use new APIs to get metadata configuration.

Hence, it will be necessary to open bdev before getting metadata
configuration.

As a preparation, move bdev open and close operations from job->thread
to g_main_thread. set timeout should be called on the thread which
called bdev open. Hence, move set timeout operation too. Furthermore,
event notification should be redirected from g_main_thread to
job->thread.

Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I3623e7c0ccc314932f79b78243be9ede0b8927b3
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/25143
Reviewed-by: Jim Harris <jim.harris@nvidia.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Community CI Samsung <spdk.community.ci.samsung@gmail.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>

show more ...

bdevperf: Remove TAILQ_REMOVE which may result in potential memory leak

Only bdevperf_test_done() dequeues jobs from g_bdevperf.jobs list and
frees these. If _bdevperf_construct_job() dequeues job,

bdevperf: Remove TAILQ_REMOVE which may result in potential memory leak

Only bdevperf_test_done() dequeues jobs from g_bdevperf.jobs list and
frees these. If _bdevperf_construct_job() dequeues job, it is leaked.
bdevperf does not have any special handling when g_run_rc is -ENOMEM
before bdevperf_construct_job(). To avoid future potential issues,
remove the TAILQ_REMOVE().

Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I00ef484b99dd8b65f6c527e428d5bfb97df1c508
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/25458
Community-CI: Community CI Samsung <spdk.community.ci.samsung@gmail.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Jim Harris <jim.harris@nvidia.com>

show more ...

bdevperf: Allocate an md buffer for verify op

For the read operation in the verify operation flow,
if the generic bdev layer cannot allocate the buffers, the data
buffers are allocated by bdevperf.

bdevperf: Allocate an md buffer for verify op

For the read operation in the verify operation flow,
if the generic bdev layer cannot allocate the buffers, the data
buffers are allocated by bdevperf.

This patch adds metadata buffers allocation to be consistent with
the existing data buffers allocation.

Change-Id: Id065c6ec80048e600520c45c097ab63302781721
Signed-off-by: Slawomir Ptak <slawomir.ptak@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/24960
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Community-CI: Mellanox Build Bot

show more ...

bdevperf: send RPC error resp. if no jobs started

When bdevperf receives 'perform_tests' RPC call, it prints 'Running
I/O for x seconds' message on the console and is trying to start I/O
jobs. The i

bdevperf: send RPC error resp. if no jobs started

When bdevperf receives 'perform_tests' RPC call, it prints 'Running
I/O for x seconds' message on the console and is trying to start I/O
jobs. The issue here is that the message is printed even if there are
no jobs defined (e.g. no bdevs created) so nothing really happens. In
such case no RPC response was sent, and the RPC call initiating tests
just timeouts.
This is fixed now and bdevperf will print on the console 'Running
I/O for x seconds' message only if there are any jobs started.
If no jobs are defined, the message is not printed, and RPC error
response is generated:
Error code: SPDK_JSONRPC_ERROR_INVALID_PARAMS (-32602)
Error message: "No jobs defined or bdevs created"

Change-Id: Ibac7a961cd79713ff2bb56a2b5d56c80fb6cc98f
Signed-off-by: Marcin Spiewak <marcin.spiewak@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/23902
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Community-CI: Mellanox Build Bot

show more ...

bdevperf: send test results in RPC response

When bdevperf test is triggered with 'bdevperf.py perform_tests' RPC
command, the tool starts test jobs and when they are finished, the
test results are p

bdevperf: send test results in RPC response

When bdevperf test is triggered with 'bdevperf.py perform_tests' RPC
command, the tool starts test jobs and when they are finished, the
test results are printed on the console. This patch adds reporting
of the test results in RPC response. The response consists of two
sections, an object indicating the number of cores on which the tests
were run, and a table with test results. When a bdevperf is started
on 8 cores with a command:
'sudo ./build/examples/bdevperf -m 0xFF -z'
an example response to the command starting tests:
'sudo PYTHONPATH=python examples/bdev/bdevperf/bdevperf.py \
perform_tests -q 16 -o 4096 -t 5 -w write'
would be:
{
"results": [
{
"job": "Malloc0",
"core_mask": "0x1",
"workload": "write",
"status": "finished",
"queue_depth": 16,
"io_size": 4096,
"runtime": 5.000048,
"iops": 1200455.675625514,
"mibps": 4689.279982912164,
"io_failed": 0,
"io_timeout": 0,
"avg_latency_us": 13.252784244696382,
"min_latency_us": 11.130434782608695,
"max_latency_us": 222.6086956521739
},
{
"job": "Malloc1",
"core_mask": "0x2",
"workload": "write",
"status": "finished",
"queue_depth": 16,
"io_size": 4096,
"runtime": 5.000024,
"iops": 1216432.5611237066,
"mibps": 4751.689691889479,
"io_failed": 0,
"io_timeout": 0,
"avg_latency_us": 13.082916284986958,
"min_latency_us": 5.7043478260869565,
"max_latency_us": 90.82434782608695
}
],
"core_count": 2
}

Majority of the information provided in RPC response is the same as
information printed on the console. If the test finishes successfully,
it is indicated by:
"status": "finished"
If the test fails, it is indicated by:
"status": "failed"
If the bdevperf was terminated (SIGINT, SIGTERM sent to the process
or by Ctrl-C in the console window), it will be indicated by:
"status": "terminated"
The "runtime" field carries information about test execution time in
seconds, latencies are presented in microseconds.
If the test fails, "runtime" indicates the time when the test failed.
"core_count" provides number of cores that were actually used during
the test.

Change-Id: I327db7a57bdc1033b6803813d54861a6c970a718
Signed-off-by: Marcin Spiewak <marcin.spiewak@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/23881
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Karol Latecki <karol.latecki@intel.com>

show more ...

bdevperf: statistics calculation in separate function

The statistics calculation is moved out of performance_dump_job()
function to separate function (bdevperf_job_get_stats()). Now we can
reuse the

bdevperf: statistics calculation in separate function

The statistics calculation is moved out of performance_dump_job()
function to separate function (bdevperf_job_get_stats()). Now we can
reuse the statistics in two places: in a function that prints test
results on the console and in a function that provide test results in
RPC response.
ema_period was removed from bdevperf_stats and now is passed as a
variable.

Change-Id: I7d9b59a7d9cf24fd35cdd059a4ae9fe715028423
Signed-off-by: Marcin Spiewak <marcin.spiewak@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/24892
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: Konrad Sztyber <konrad.sztyber@intel.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>

show more ...

bdevperf: split bdevperf_aggregate_stats

This patch splits bdevperf_aggregate_stats into two structures, in
order to have the statistics counters in separate structure. These
can be then used in pla

bdevperf: split bdevperf_aggregate_stats

This patch splits bdevperf_aggregate_stats into two structures, in
order to have the statistics counters in separate structure. These
can be then used in places where current_job member of the
bdevperf_aggregate_stats is not needed.

Change-Id: Ie0ffddddc3b8ccbd4033edec1d617215c380c3fc
Signed-off-by: Marcin Spiewak <marcin.spiewak@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/24891
Community-CI: Mellanox Build Bot
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>

show more ...

bdevperf: allocate data buffers based on bdev's numa id

Signed-off-by: Jim Harris <jim.harris@samsung.com>
Change-Id: Idd96a7925741753531152b2f777bcedd46cae73e
Reviewed-on: https://review.spdk.io/ge

bdevperf: allocate data buffers based on bdev's numa id

Signed-off-by: Jim Harris <jim.harris@samsung.com>
Change-Id: Idd96a7925741753531152b2f777bcedd46cae73e
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/24154
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Reviewed-by: Ben Walker <ben@nvidia.com>

show more ...

bdev/bdevperf: print continual updates

Change-Id: Ib6f0e666205e668ee1490108fb3cdfec8640c53a
Signed-off-by: Anisa Su <anisa.su@samsung.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/24

bdev/bdevperf: print continual updates

Change-Id: Ib6f0e666205e668ee1490108fb3cdfec8640c53a
Signed-off-by: Anisa Su <anisa.su@samsung.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/24418
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>

show more ...

bdevperf: add python/ dir to PYTHONPATH

The other python scripts (e.g. scripts/rpc.py) already add the relative
location of the python directory inside the repo. It's very convenient,
as it doesn't

bdevperf: add python/ dir to PYTHONPATH

The other python scripts (e.g. scripts/rpc.py) already add the relative
location of the python directory inside the repo. It's very convenient,
as it doesn't require users running them from within the repo to adjust
PYTHONPATH by themselves. Do the same for bdevperf.py.

Signed-off-by: Konrad Sztyber <konrad.sztyber@intel.com>
Change-Id: I436248bc73f0a4ca7332af65d8c51a6f574144b4
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/23788
Reviewed-by: Tomasz Zawadzki <tomasz@tzawadzki.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Community-CI: Mellanox Build Bot

show more ...

bdevperf: Use spdk_bdev_get_dif_pi_format()

Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I1ee0aad6bf919adb5497bb9a06282797d0437d27
Reviewed-on: https://review.spdk.io/gerrit/c/

bdevperf: Use spdk_bdev_get_dif_pi_format()

Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I1ee0aad6bf919adb5497bb9a06282797d0437d27
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/23627
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Reviewed-by: Aleksey Marchuk <alexeymar@nvidia.com>
Reviewed-by: Ben Walker <ben@nvidia.com>

show more ...

bdevperf/bdevperf.py: extend RPC timeout if needed

This patch is created to automatically extend
RPC timeout, which by default is set to 60 sec.,
if someone starts a test with longer duration.
E.g.

bdevperf/bdevperf.py: extend RPC timeout if needed

This patch is created to automatically extend
RPC timeout, which by default is set to 60 sec.,
if someone starts a test with longer duration.
E.g. the call like this:
bdevperf.py perform_tests -q 16 -o 4096 -w write -t 70
would normally result in RPC timeout during waiting for
the response, because the scheduled test will take ~70 seconds,
while the RPC timeout is 60. With this change, RPC timeout
will be set to 75 sec (5 sec longer than expected test), to
allow ample time for the test to complete.

Also, if the user explicitly specifies RPC timeout that is
shorter than the test duration, it will also be extended to
allow time for the test to complete.
I.e., for the call:
bdevperf.py -t 10 perform_tests -q 16 -o 4096 -w write -t 30
the RPC timeout will be automatically extended to 35 sec,
even though the user specified 10 sec (which is insufficient
in this case)

Change-Id: I094f2d2e27b1d363f81977394d6c2fe8f39773eb
Signed-off-by: Marcin Spiewak <marcin.spiewak@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/23776
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>

show more ...

Revert "bdevperf: Use spdk_bdev_ext_io_opts for metadata"

This reverts commit 56f2a365eb0e353cdb93c2a2884e5bcff3db1abd.

Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I6488e4

Revert "bdevperf: Use spdk_bdev_ext_io_opts for metadata"

This reverts commit 56f2a365eb0e353cdb93c2a2884e5bcff3db1abd.

Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I6488e44e99cb00061ccf24a982491e6c14bdd492
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/23234
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>

show more ...

Revert "bdevperf: Set dif_check_flags_exclude_mask by negating job->dif_check_flags"

This reverts commit 058f4cfeb17a0d0713852ed34cb6276027ed29c5.

Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@in

Revert "bdevperf: Set dif_check_flags_exclude_mask by negating job->dif_check_flags"

This reverts commit 058f4cfeb17a0d0713852ed34cb6276027ed29c5.

Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I2e1790a11a731e48d0fdc6ce44d6abe8683d4929
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/23233
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>

show more ...

bdevperf/nvme_perf/hello_world: Add fflush(stdout)

In case the app hangs in cleanup/fini, most likely stdout will not get
printed out (buffered) therefore no logs will be visible.
This change should

bdevperf/nvme_perf/hello_world: Add fflush(stdout)

In case the app hangs in cleanup/fini, most likely stdout will not get
printed out (buffered) therefore no logs will be visible.
This change shouldn't harm performance as fflush() is performed after
the test itself.

Change-Id: I7d12ae037de5b686c776de8d66c5cbec5df6a544
Signed-off-by: Wojciech Panfil <wojciech.panfil@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/23022
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jacek Kalwas <jacek.kalwas@intel.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Community-CI: Mellanox Build Bot

show more ...

bdevperf: Set dif_check_flags_exclude_mask by negating job->dif_check_flags

By this patch, we can control DIF check per I/O but the behavior is
backward compatible because we just negate job->dif_ch

bdevperf: Set dif_check_flags_exclude_mask by negating job->dif_check_flags

By this patch, we can control DIF check per I/O but the behavior is
backward compatible because we just negate job->dif_check_flags and set
its result to dif_check_flags_exclude_mask for each I/O.

Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: Ia3037bc4d43b566f0ccc46588d430edcb5f38626
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/22823
Community-CI: Mellanox Build Bot
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>

show more ...

bdevperf: Use spdk_bdev_ext_io_opts for metadata

This is necessary for bdevperf to use dif_check_flags_excluded_mask
of spdk_bdev_ext_io_opts structure.

Signed-off-by: Shuhei Matsumoto <smatsumoto@

bdevperf: Use spdk_bdev_ext_io_opts for metadata

This is necessary for bdevperf to use dif_check_flags_excluded_mask
of spdk_bdev_ext_io_opts structure.

Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I23faa877fdcdaeb4cdf018b1f3b4336c70b659d7
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/21262
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>

show more ...

bdevperf: Use task->iov for non-zcopy read and verify read

To use spdk_bdev_ext_io_opts structure, we have to use iovec
structure for non-zcopy read and verify read.

For verify read, task->verify_b

bdevperf: Use task->iov for non-zcopy read and verify read

To use spdk_bdev_ext_io_opts structure, we have to use iovec
structure for non-zcopy read and verify read.

For verify read, task->verify_buf is allocated and used if job->buf_size
is large but a buffer allocated internally in the generic bdev layer is
used otherwise. For the later case, the iovec passed to the bdev I/O API
must be allocated until the corresponding completion callback is called.
Hence, we use task->iov for verify read.

For generic non-zcopy read, we also use task->iov.

In bdevperf_complete(), we do not have to get iovec using
spdk_bdev_io_get_iovec() because we use task->iov for all cases now.

Signed-off-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Change-Id: I64e148e1f8b4b5b4c6bd435e126af0cc98155778
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/21261
Community-CI: Mellanox Build Bot
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>

show more ...

example/bdevperf: fix rw_percentage for reset flow

(b2cfb6b)bdevperf: fix non expected reads for verify flow

Patch above added check for job type during I/O completion,
that catches unexpected read

example/bdevperf: fix rw_percentage for reset flow

(b2cfb6b)bdevperf: fix non expected reads for verify flow

Patch above added check for job type during I/O completion,
that catches unexpected read operations. This was done by
setting rw_percentage to 0.

That patch did not account for reset job type, which
shares the flow with verify and only adds resets.

Fix this by fallingthrough to verify in job config for
reset job.

Fixes #3320

Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I4e449d533ea3ab2587653493953f0dea18e1c656
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/22519
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Reviewed-by: Shuhei Matsumoto <smatsumoto@nvidia.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Jacek Kalwas <jacek.kalwas@intel.com>

show more ...

bdevperf: fix non expected reads for verify flow

SPDK_BDEV_IO_TYPE_READ can be set within bdevperf_submit_single based
on rw_percentage. However, for verify flow read shall be done on write
completi

bdevperf: fix non expected reads for verify flow

SPDK_BDEV_IO_TYPE_READ can be set within bdevperf_submit_single based
on rw_percentage. However, for verify flow read shall be done on write
completion callback only.

Fix by change rw_percentage to 0. It will prevent setting
SPDK_BDEV_IO_TYPE_READ in bdevperf_submit_single by conditions
already existing there.

Change-Id: Ica135e8aedbca4693c546a3bf8498dd4b3ce11ef
Signed-off-by: Jacek Kalwas <jacek.kalwas@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/22473
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Community-CI: Mellanox Build Bot

show more ...

bdevperf: more verbose logging for data mismatch

With prev logs only the first dword was dumped e.g. `dword expected
0x0 got 0x0` and that is not enough for debugging purposes.

Change-Id: Iaddc6d15

bdevperf: more verbose logging for data mismatch

With prev logs only the first dword was dumped e.g. `dword expected
0x0 got 0x0` and that is not enough for debugging purposes.

Change-Id: Iaddc6d1525cbb83f38f7930ea15c07c1916a0ac4
Signed-off-by: Jacek Kalwas <jacek.kalwas@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/22082
Reviewed-by: Konrad Sztyber <konrad.sztyber@intel.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>

show more ...

bdevperf: remove max io size limit for verify

Allocate a buffer for verify jobs' readbacks, thus removing the max io
size limit of SPDK_BDEV_LARGE_BUF_MAX_SIZE

Change-Id: Ia95342c336edffdd742a228f

bdevperf: remove max io size limit for verify

Allocate a buffer for verify jobs' readbacks, thus removing the max io
size limit of SPDK_BDEV_LARGE_BUF_MAX_SIZE

Change-Id: Ia95342c336edffdd742a228fce4079f399f672cf
Signed-off-by: Olivier Mayer <omayer@kalrayinc.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/21230
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <jim.harris@samsung.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>

show more ...