This makes it easier to debug failures.
Change-Id: Ic1099ec08308d5e50dd9fe03798dc0873579f2cc
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/423961
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: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
These don't do anything yet.
Change-Id: I718c6fb19c059a39bc5cb360c3da47ec05bfa416
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/423767
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: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: Jim Harris <james.r.harris@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>
When spdk_for_each_channel() was called on an io_device with no
channels, the completion function would be called immediately in the
context of the spdk_for_each_channel() caller. This is different than
the normal completion when channels are present, which is always called
asynchronously via spdk_thread_send_msg(), causing problems for callers
that did not expect to get the completion callback before
for_each_channel returned.
In particular, this causes problems with the set_bdev_qos_limit_iops RPC
method, which holds a mutex around the spdk_for_each_channel() call and
also attempts to lock that mutex again within the completion callback.
(This is GitHub issue #344.)
To avoid this problem and make the completion call consistent between
no-channels and channels-present cases, always send the completion via
spdk_thread_send_msg().
Change-Id: I9e5ec3592462c7b2ed682e665fe62ae3a5b5cc59
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/417068
Tested-by: 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>
The bdev iscsi initiator had a bug which resulted in
an io_device getting unregistered twice. An error message
got printed but it didn't immediately result in any kind
of failure detected by any tests. Really this should be
considered a fatal error, so add an assert() so it's found
more readily should these kinds of bugs occur again in the
future.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I30df86ba2e6a3559e00f054c7dde14713d6ac80f
Reviewed-on: https://review.gerrithub.io/415375
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
This will return the number of currently allocated threads.
Modify the bdev_io caching code to use this new API since these
caches are really per-thread, not per-core. SPDK does not support
dynamic threading yet, but once it does, we will want callers to be
using functions from the thread API - not counting the number of cores
allocated to the application.
spdk_env_get_core_count may still be useful as a helper function, so
it is still kept and not deprecated. For example, app.c uses it to
print the number of cores allocated to the application. bdevperf should
eventually be modified to use spdk_thread_get_count, but holding off on
that for now until spdk_event_allocate() uses threads instead of a reactor
lcore to specify where the event should be executed.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I5a30e3e825e6821da87d3927a2443768dfd740f4
Reviewed-on: https://review.gerrithub.io/414709
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Madhu Pai <mpai@netapp.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Change-Id: I6babd4cf990bf19b510db88bdfb0ca81e29d9252
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/414700
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Madhu Pai <mpai@netapp.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
This makes more sense as a first class library.
Change-Id: Ibd5c578f8708bd8c7d83fe1629e97c9a3316b56b
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/414698
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Reviewed-by: Madhu Pai <mpai@netapp.com>
