As reactors no longer have a thread created with them.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: Ie9e9411c52c215b8cffd894fef6394448ae8167d
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6312
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>
This function is used to eliminate the duplicated code.
Also rename _free_task to _free_task_buffers to make
it easy to understand the code.
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: I2ef9dbbb81c12c229952f5590f616a17dc2fe087
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6128
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Community-CI: Mellanox Build Bot
When the task is allocated from _get_task function,
the task is removed from the tasks_pool in non batched mode.
So in _accel_done function, when the is_draining flag is true
and it is not a batched mode, we should add task back to the pool.
If not, the the memory related resource allocated by the task will
not be freed.
PS: In _check_draining function, the related resource of task
in the tasks_pool will be freed. However, if the task is not
added back, the resource free will not work as expected.
PS: The issue is reported by ASAN.
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: I981117037fba8e111987c771cae65bc06a734a6f
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6097
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Community-CI: Mellanox Build Bot
Previous version just used batching for initial queue depth loading
as an example of how to batch. This version adds a new param -b that
allows the caller to either disable or set a specific batch value
for use during the entire run. For example Q depth 128 with a batch of
64 will always send IO in batches of 64 so in effect once the Q depth
has been met the tool will wait until the Q depth has dropped enough
to send another 64. This will allow for a more accurate measurement
of batching performance vs sending IO one at a time.
Also added clarrifying note on usage the Q depth is per core, not
spread over all cores.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: Ieb1df5f668ae9c7b37b0d789fceab1138044a5e6
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5489
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Community-CI: Mellanox Build Bot
* change how tasks are allocated and freed (simplifcation)
* added helper for getting and freeting a single task
* minor drive-by in chaning function parms for _submit_tasks()
Note that the task pool is used to manage tasks and their data
buffers. It is fully allocated and populated before the first IO
is sent and tasks are never retired, they are re-used so they are
not removed from the list except for error or exit cleanup.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I5fea5ef8c989df6310f15b2c9bb4e8aef9bd3d3b
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5487
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ziye Yang <ziye.yang@intel.com>
First in a series of patches to change how batching is implemented.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: Iaee780cd948beee66a37528fdad6fc6841ea642b
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5486
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: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
All of our Makefiles duplicate huge lists of libraries
in SPDK_LIB_LIST. We have a very precise and accurate
accounting of the library dependencies in
mk/spdk.lib_deps.mk which can be used to generate
the full list if the app specifies the modules and
subsystem libraries it wishes to link.
I did a first pass through all of the existing
Makefiles to take advantage of this new functionality.
There may be more optimizations we can make later but
don't want to hold up this patch for all of them.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Icdaf6f749a6908df2c2ce2db22631a4af4ff3a9e
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5553
Community-CI: Broadcom CI
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
This patch is used to support ABI compatability issue for spdk_app_opts.
Fixes#1484
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: I6fed777fa15b367f7c3706b8f218d86d56493906
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/5330
Community-CI: Broadcom CI
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jacek Kalwas <jacek.kalwas@intel.com>
rc has been assigned to -1 directly, so it's no used here.
Issue:
accel_perf.c:681:7: warning: Although the value
stored to 'rc' is used in the enclosing expression,
the value is never actually read from 'rc'
if ((rc = spdk_app_parse_args(argc, argv, &opts, "o:q:t:yw:P:f:", NULL, parse_args,
^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated.
Signed-off-by: yidong0635 <dongx.yi@intel.com>
Change-Id: I1d0a4447c4655bde97fa18d50b694809fc9010ea
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/4825
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
For performance. Also allocate the accurate number of tasks instead
of just blindly doubling to account for batching.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I4e5accfd97477f908368f16d3fc1d8f47896a4d6
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/3321
Community-CI: Mellanox Build Bot
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
No longer allocate accel_tasks or convert to and from application
tasks and accel tasks.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: Id2d598d79bebcef522c76c125e65b100266ced57
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/3210
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 was sort of a clunky interface requiring a couple of inline
functions in every app that wants to use the accel_fw moving
forward. By having the accel_fw public API accept a callback arg
instead of an accel_task combined with adding a pool of accel_tasks
in the accel_fw engine we can eliminate this.
After changing the parm to a cb_arg, changes were made to all accel_fw
interfaces to put cb_fn and cb_arg as the last parms in public and
private function calls.
Related bdev_malloc changes need to be in this patch in order to pass CI.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I2b75764e534562d91484a094c3352266156d8425
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/3209
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>
Also remove the batching check in accel_perf as all 3 engines
now support it.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: Ib4cf4b148b50df50a4fc7be9e861cc83f355623a
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/3155
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Also one small bug fix w/compare in accel_perf as a result
of changes made in accel_perf sicne base compare was added.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: Id8e67bd9de9cbd006ac148f4a77807cc3e8e662b
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/2958
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Also update accel_perf to support it.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: Ida7b863de02c64239ec4cfbdc3b0235d4e0521f9
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/2954
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Also as a minor drive-by, added verify capability for fill that
wasn't there before, useful in making sure the prep function
was working and really should have been there anyway.
idxd support for prep fill will follow.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: Ib54311c1fb98abd2fb61df6603cf3c5300b71161
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/2952
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
When an engine has not filled out an entry point for an accel_fw
function, call the sw implmentation that's built into the accel_fw
transparently.
Also update the accel_perf app to no longer block attempts to use
capabilities not reported as supported by the selected engine but
instead to print an informational message.
The exception is batch related commands as their implementation
is specific to the engine.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I0b3971d075ced20182f97e95147e8887103e68cd
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/2930
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
Includes a few related changes:
* slight refactor to have common prep function for batch prep
commands in the sw engine
* added support for dualcast batching in accel_perf example app
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: If9f078a0f8eb509d296e60a94331c92a72f8cb10
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/2913
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 patch uses the IDXD batch feature for initial queue depth loading.
As there is a good amount of code needed to support batching on a per
command basis, this starts with only copy. As other commands are
enabled for batching, they will be added to accel_perf. If batching is
not available, the perf tool will simply submit the initial queue depth via
a loop, the same way it did before batching.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I2669936e4da9b31a1d8fa8f0c71e9f5a6fcae412
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/2745
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>
Before the value was hardcoded, provide it as a user option.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I841a3da266b7ab6e7d71503f2872338d07c8bead
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/2139
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Community-CI: Mellanox Build Bot
Dual-cast copies the same source to two separate destination buffers.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I0535720b2b167ee8191e90998e55c2d5ac4ce6ef
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/2130
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
A subsequent patch brought me to realize that I was not checking the
return values from operational calls to the engine. Also, noticed
that on any failure we'd stop submitting IO but still wait until the
timer expired to exist. To match what other perf tools do, now
continue to submit even after an error and run until the clock is
out giving an accurate count of transient failures and successes.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I5e6a0f382f9090e428ee67fbdfb60926e9c85917
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/2879
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Optionally let the user specify what % into the buffers there should
be a miscompare if desired.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I3a5524de5e0f7cb91fcf92e76f937bdf280a515b
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/2109
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
This is more to exercise the various CRC-32C back ends than anything
else but will calc a CRC on the source data and if -y is
specified it will also calculate SW (ISA-L)f defined CRC-32C and
compare every operation against what the accel engine came up
with using whatever back-end was enabled.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I5ecd3c474717bce4e46d75875a784f888052993e
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/2073
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 <shuhei.matsumoto.xt@hitachi.com>
Need to check that a valid workload type was specified and print
usage if not.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: Ib956e28f9049e49f19adcc09e861db42034799b8
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1900
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Community-CI: Mellanox Build Bot
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Usable with both ioat and upcoming idxd modules.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: Ie25743854aa6a05d1b5f580ea64fb5c7d7c722d1
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1720
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
We will be create fine name for each poller but it will need large
effort. Replacing spdk_poller_register by the macro SPDK_POLLER_REGISTER
will provide better name than function address with minimum effort.
Following patches may improve function name for clarification.
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: If862a274c5879065c3f7cb04dcb5ca7844523e68
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1781
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: Paul Luse <paul.e.luse@intel.com>
Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: Maciej Szwed <maciej.szwed@intel.com>
Community-CI: Broadcom CI
Supports SW accel engine, CBDMA accel engine and eventually the DSA
accel engine through common API. Basic functions only are included
here, subsequent patches will add more features as needed. For
example, the verify function is very crude right now.
Tool output was compared to the ioat perf tool w/CBDM and results
are very close although there a small penalty for using the
framework.
For CBDMA, start the app with --wait-for-rpc and run the ioat scan
RPC with desired whitelist. If there are not enough channels
(channels are not a concept in the framework) then SW accel engine
will be used for cores that could not get a channel.
Signed-off-by: paul luse <paul.e.luse@intel.com>
Change-Id: I9c5ab33e74db7bdbc5bfe5d7a08cc7ffaeab625b
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/883
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
