RDMA QP is attempted to recover after IBV_EVENT_QP_FATAL event
is received from IBV asynchronous event API.
RDMA QP is put into ERROR state and is not processing any inbound
requests. The outstanding requests are only allowed to COMPLETED
and FREE states, no outbound transfers are performed.
IBV_EVENT_QP_LAST_WQE_REACHED or IBV_EVENT_SQ_DRAINED event is
expected to follow IBV_EVENT_QP_FATAL, giving a go to draining of
all outstanding requests and freeing the associated resources.
The requests executed by block layer are gracefully allowed to
complete, but no outbound transfers are made.
Note, outstanding requests can not be reliably completed through
polling the CQ, as WC's with failure status might not have all
the fields valid. The failed WC's are dropped and the outstanding
requests are fetched from the appropriate state's linked list.
QP recovery is triggered when there is no more outstanding requests.
If QP recovery is completed succesfully, the RDMA QP is put back into
ACTIVE state, the QP disconnect is triggered otherwise.
Change-Id: I45ee7feea067f80ccc6402518990014d691fbda3
Signed-off-by: Philipp Skadorov <philipp.skadorov@wdc.com>
Reviewed-on: https://review.gerrithub.io/416879
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
From now on remove any SPDK artifacts just use:
./scripts/setup.sh cleanup
Fixes#302
Change-Id: I5ebe522fecfcb8a867a96ab10bacda6083c8c224
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/419575
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
This function always return 0, so change the type of function
to void
Change-Id: I715cf82c74619a2707b8e5a453710a992489f2c8
Signed-off-by: Ziye Yang <optimistyzy@gmail.com>
Reviewed-on: https://review.gerrithub.io/419045
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: GangCao <gang.cao@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Move logical volume tasting test plan into script.
Change-Id: I2da9ab9348a01df86489c81b6debea4f564cc318
Signed-off-by: Pawel Kaminski <pawelx.kaminski@intel.com>
Reviewed-on: https://review.gerrithub.io/419383
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>
Reviewed-by: Karol Latecki <karol.latecki@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
This patch prevents to copy cluster data when there is not
backing blob to improve cluster allocation performance
in thin provisioned blobs.
Change-Id: Ie766d2e5274daa74c2b13b2198a20205e3417467
Signed-off-by: Tomasz Kulasek <tomaszx.kulasek@intel.com>
Reviewed-on: https://review.gerrithub.io/417938
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Before this patch when serializing extents,
unallocated clusters were treated as separate lba.
This caused metadata to grow without need.
Change-Id: I5d66466dda5f5e6d4d53f4ed5bd0bac18c74be96
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-on: https://review.gerrithub.io/419180
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Maciej Szwed <maciej.szwed@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Information regarding decouple parent lvol bdev was missing
in documentation.
Change-Id: Ib1611efedc472cc37133a57aed41f6c2c0abbe1f
Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Reviewed-on: https://review.gerrithub.io/419342
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>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
When hot removal of a LUN is started, callback is called for each
iSCSI connection which accesses the LUN. Callback checks all transfer
tasks complete and then close the LUN.
If the connection clears all transfer tasks before getting all responses
to them from the initiator, the initiator continues to retry data write.
Hence the connection have to wait until all transfer tasks complete.
Change-Id: Iad9063673cfedbd78758890d55a4254512e4fca4
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/417199
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Use a descriptor got by opening the LUN to allocate IO channel of the LUN.
This requires opening the LUN is done before allocating IO channel of the
LUN.
Use the descriptor to free IO channel of the LUN too.
Additionally, assert is added to the close LUN function to check if
IO channel is freed before the last close LUN function is called.
Change-Id: Iafb2f9ce790fff25801ea45b3286f3e26943807b
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/417807
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
To remove a LUN safely for iSCSI multiple connections, each connection
must return I/O channel after checking completion of all tasks.
Open/close mechanism provides a way to register callback to do it.
Registered callback will be called after checking completion of
outstanding tasks.
Making the close LUN function public is necessary to support hot removal.
Change-Id: I06d50d016b0b7aba0d081da226f5b2e0c911629e
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/417198
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
It appears that we can get caught in a loop when shutting down a
subsystem if there is a qpair that has been moved to the deactivating
state due to the target shutting down early. These qpairs don't have a
state cb and won't ever be destroyed or removed from the global list.
Change-Id: I4f9ed774a94e0e2c7ff7bfa3af1776b38a787035
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/419269
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
DPDK has an option "huge-unlink" which can be used to
remove mmapped files after EAL initialization, SPDK
doesn't enable the option by default, ethier DPDK,
while here, export a new parameter which can let
user to decide enable it or not.
Fix issues #349 and #350.
Change-Id: Ic516b9f48f7b1c7c51712cc7bb7475ed904ff24b
Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/419156
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Root cause: we did not check the status of the bdev_io
in bdevperf_verify_write_complete function.
So the fix is that: if the status is failed, we should
not do the read check again, but just complete this I/O.
Change-Id: I3dc796abe47c3f2e7eeb2b6eb0ebc0338122479d
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Reviewed-on: https://review.gerrithub.io/419338
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Chunyang Hui <chunyang.hui@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
When spdk_put_io_channel is called, if its the last reference,
we defer actual destruction of the channel, so that code
in the same context which may be referring to the channel
doesn't crash.
But it is possible that an io_channel for that same io_device
could be requested before the deferred message is processed.
This would result in a second io_channel being created for
that device on the same thread.
To avoid this case, don't immediately remove the channel from
the list when the last reference is put. When the deferred
message is processed, if additional references were allocated
in the meantime, don't destroy the channel.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Change-Id: Idb8d4705fda0eb9c338e4960430e04edbe537e05
Reviewed-on: https://review.gerrithub.io/418878
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: GangCao <gang.cao@intel.com>
Reviewed-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Change-Id: I50a525ac46ac011af81625fa0f095b1f3d981aba
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/414891
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Change-Id: I276efc2a42641749cefb30708d6f5c2e5e969fa8
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/414890
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Raid module:
============
- SPDK raid bdev module is a new bdev module which is
responsible for striping various NVMe devices and expose the raid bdev
to bdev layer which would enhance the performance and capacity.
- It can support theoretically 256 base devices (currently it is being
tested max upto 8 base devices)
- Multiple strip sizes like 32KB, 64KB, 128KB, 256KB, 512KB etc is
supported. Most of the current testing is focused on 64KB strip size.
- New RPC commands like "create raid bdev", "destroy raid bdev" and "get raid bdevs"
are introduced to configure raid bdev dynamically in a running
SPDK system.
- Currently raid bdev configuration parameters are persisted in the
current SPDK configuration file for across reboot support. DDF will be
introduced later.
High level testing done:
=======================
- Raid bdev is created with 8 base NVMe devices via configuration
file and is exposed to initiator via existing methods. Initiator is
able to see a single NVMe namespace with capacity equal to sum of the
minimum capacities of 8 devices. Initiator was able to run raw
read/write workload, file system workload etc (tested with XFS file
system workload).
- Multiple raid bdevs are also created and exposed to initiator and
tested with file system and other workloads for read/write IO.
- LVS / LVOL are created over raid bdev and exposed to initiator.
Testing was done for raw read/write workloads and XFS file system
workloads.
- RPC testing is done where on the running SPDK system raid bdevs
are created out of NVMe base devices. These raid bdevs (and LVOLs
over raid bdevs) are then exposed to initiator and IO workload was
tested for raw read/write and XFS file system workload.
- RPC testing is done for delete raid bdevs where all raid bdevs
are deleted in running SPDK system.
- RPC testing is done for get raid bdevs where existing list of
raid bdev names is printed (it can be all raid bdevs or only
online or only configuring or only offline).
- RPC testing is done where raid bdevs and underlying NVMe devices
relationship was returned in JSON RPC commands
Change-Id: I10ae1266f8f2cca3c106e4df8c1c0993ddf435d8
Signed-off-by: Kunal Sablok <kunal.sablok@intel.com>
Reviewed-on: https://review.gerrithub.io/410484
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>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
If command is completed manually (e.g.: driver decide to fail this
command from some reason) not all NOTICELOG informations about this
command are valid. To not get confused who completed the command (driver
or a device) show addtional information when command is completed
manually (using NOTICELOG like nvme_qpair_print_command and
nvme_qpair_print_completion do).
Change-Id: I7211bd165aec142ad11e806fa3031371375edd2f
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/411293
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Considering the I/O size is variant from small in byte to
large in megabyte, need to consider the compensation of less
allowed bytes in next timeslice if the current timeslice
sends more bytes down.
Change-Id: I885f0bb21001bd90879aa8622e2b34e3bf78cf6e
Signed-off-by: GangCao <gang.cao@intel.com>
Reviewed-on: https://review.gerrithub.io/417829
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
When multiple LUNs are attached and spdk_clear_all_transfer_tasks() is called with
a LUN, all tasks whose target is the LUN are removed and there will be room to start
queued tasks for other LUNs. However currently no migration is done.
Add UT code to verify the fix too.
Change-Id: I082d370ab86a46e5b4a74a16293a572fae663add
Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-on: https://review.gerrithub.io/418765
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Seth Howell <seth.howell5141@gmail.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ziye Yang <optimistyzy@gmail.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
This should speed things up a little by avoiding
unnecessary compilation of dpdk in the githook
Change-Id: I214d8acc44ca041b6afb1b15dcb8555aa8a0d8f0
Signed-off-by: John Meneghini <johnm@netapp.com>
Signed-off-by: Ed Rodriguez <ed.rodriguez@netapp.com>
Reviewed-on: https://review.gerrithub.io/419218
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>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Add some known Samsung controllers to those requiring the quirk,
NVME_QUIRK_DELAY_BEFORE_CHK_RDY. Addresses an issue for those
who may not have later firmware that corrects the problem.
Correspondingly, extend the delay from 2 secs to 2.5 secs.
Change-Id: Iee773905a2a49711775042c061f6c347e0da85e9
Signed-off-by: Lance Hartmann <lance.hartmann@oracle.com>
Reviewed-on: https://review.gerrithub.io/419273
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>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Following the PI operation inside nvme/perf, add 2 PI
options into fio engine option.
SPDK ioengine will check whether PI information should
be generated or verified.
Running with PI setting, following settings are required:
1. In fio configure file, add PRACT like "pi_act=0", and
set PRCHK by flags(GUARD|REFTAG|APPTAG), like "pi_chk=GUARD".
2. Format device namespace with proper PI setting
3. In fio configure file, blocksize should be set as the
sum of data and metadata.
Sequential Write and following Sequential Read can be used
to evaluate it quickly.
Change-Id: I54e202bb8510cd987772117dff241aec714c8fb9
Signed-off-by: Xiaodong Liu <xiaodong.liu@intel.com>
Reviewed-on: https://review.gerrithub.io/414753
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: John Kariuki <John.K.Kariuki@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Following script allows user to run tests which
follow test cases used in SPDK Vhost Benchamrk tests.
Change-Id: I4e69b13d86165fcf37c678b095c2a11817476465
Signed-off-by: Karol Latecki <karol.latecki@intel.com>
Reviewed-on: https://review.gerrithub.io/416527
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Seth Howell <seth.howell5141@gmail.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Catch expcetions thrown by rpc.client properly.
Solved issue with errors not being displayed properly
when using python2.
Change-Id: I7f0823b2fda7e0e0b6e957de769ae5e10315b446
Signed-off-by: Karol Latecki <karol.latecki@intel.com>
Reviewed-on: https://review.gerrithub.io/419225
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Pawel Kaminski <pawelx.kaminski@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Change-Id: Iae6feaf9f51f32e20376a539e46700f07783419e
Signed-off-by: Karol Latecki <karol.latecki@intel.com>
Signed-off-by: Pawel Kaminski <pawelx.kaminski@intel.com>
Reviewed-on: https://review.gerrithub.io/417683
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Maciej Szwed <maciej.szwed@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This also save CWD on init so socket path is no longer relative.
Change-Id: I303401fe0340f0bc2ea5e4ba468361f68ea84c3d
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/419067
Reviewed-by: Karol Latecki <karol.latecki@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Pawel Kaminski <pawelx.kaminski@intel.com>
Reviewed-by: Paweł Niedźwiecki <pawelx.niedzwiecki@intel.com>
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>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Change-Id: I16010ccae613d707aa7e4f34fc694c467a589fa7
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/418682
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Some parts of documentation like parameters or CPU mask format are
common to all targets so add this common userguide page.
Change-Id: I17410abada008f1fde144a13b4bd938c23902b71
Signed-off-by: Pawel Wodkowski <pawelx.wodkowski@intel.com>
Reviewed-on: https://review.gerrithub.io/417718
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
These parameters include read and write latency ticks, current queue
depth, and the period at which the queue depth is beign polled.
Change-Id: I7c57177e3cd1629b957a16dd8f4fa992ef53ae3d
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/418117
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ziye Yang <optimistyzy@gmail.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
This change includes a function to enable this feature on a per-bdev
basis. The new information stored in the bdev includes the following:
measured_queue_depth: The aggregate of the outstanding oparations from
each channel associated with this bdev.
period: The period at which this bdev's measured_queue_depth is being
updated.
With this information, one could calculate the average queue depth and
the disk utilization of the device
Change-Id: Ie0623ee4796e33b125504fb0965d5ef348cbff7d
Signed-off-by: Seth Howell <seth.howell@intel.com>
Reviewed-on: https://review.gerrithub.io/418102
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
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 the event that something went wrong earluer, we shouldn't
try and touch memory that was never allocated.
This is at least a partial fix for issue 363.
Change-Id: I5174c845fb89f6d386877758fdc99178704ff73c
Signed-off-by: Paul Luse <paul.e.luse@intel.com>
Reviewed-on: https://review.gerrithub.io/419039
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
The spdk_dma_zmalloc guarantee about physical memory contiguity
is about to be removed soon. For hardware rings that require
physical memory or IOVA contiguity we will now enforce hugepage
alignment and size restrictions to make sure they occupy only
a single hugepage.
Change-Id: Iebaf1e7b701d676be1f04a9189201c5d89dad395
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/418547
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Chandler-Test-Pool: SPDK Automated Test System <sys_sgsw@intel.com>
Requests that are being put into IBV context are lost when
IBV QP breaks and its SQ drains.
In order to track NVMf/RDMA requests, RDMA QP has been
reworked to track requests at any state with queues of
requests for each state.
This allowed to get rid of a few intermediate queues and
request counters.
A couple of states has been added to track outbound requests
with and without data. They will be used by QP recovery for
freeing resources assigned to outstanding requests.
Change-Id: Ie84207325c38e5bb2c247cd6dcddb82dfad0d503
Signed-off-by: Philipp Skadorov <philipp.skadorov@wdc.com>
Reviewed-on: https://review.gerrithub.io/416878
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
We didn't handle vq alloc or init failure, because
queues are initialized all at once on device init and
if one vq fails, all of them are to be destroyed.
This behavior is really unintuitive, and with the
latest changes we have a possible segfault scenario.
(We could spdk_dma_free() a buffer that failed to
allocate).
It is now required that the queue allocation function
cleans up after itself.
Change-Id: I6cd1d30c710eb9266288905ab982db363f972a1d
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/419001
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
The spdk_dma_zmalloc guarantee about physical memory contiguity
is about to be removed soon. For hardware rings that require
physical memory or IOVA contiguity we will now enforce hugepage
alignment and size restrictions to make sure they occupy only
a single hugepage.
Change-Id: I8f44ad6f33d60f01403cc3db693497e4f722e528
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/418612
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
