Patch afe860ae deferred freeing the io_device. However, for nvme, the
io_device context (spdk_nvme_ctrlr) is still being destructed before
io_channels are destroyed, causing segfaults on hotremove.
This patch defers io_device context destruction and fixes nvme
hotremove.
Fixes: afe860aeb1 ("channel: Correctly defer unregisters if channels exist")
Fixes: 5533c3d208 ("util: defer put_io_channel")
Change-Id: I7af699174cac0c6c6a6faa2cc65418c47347eb9a
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/370459
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Unregistering a device should be deferred if I/O
channels still exist. Those I/O channels are likely
undergoing a deferred unregister themselves.
Change-Id: I67186232a58f212b867f6ef894c3d37aae2a4d53
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/370351
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
As specified in the doc of spdk_for_each_channel:
```
Once 'fn' has been called on each channel, 'cpl' will be called
on the thread that spdk_for_each_channel was initially called from.
```
Fixes: ff87d29cc3 ("io_channel: Add mechanism to call a function on
each channel")
Change-Id: Ic60b061ec402672f510d99697943e96ff9a73417
Signed-off-by: Dariusz Stojaczyk <dariuszx.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/369719
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
With vbdevs, there are many more cases where the last
(or only) channel for a given I/O device may be destroyed
in the context of the poller for that I/O device. To
avoid returning to the poller and having it check
for data on a resource that was just released, instead
defer the actual put_io_channel work via
spdk_thread_send_msg().
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: Ic99776bddbe9a305f221f8c094ea97dce2d6df0b
Reviewed-on: https://review.gerrithub.io/368619
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
This is faster than spdk_get_thread(), which iterates over all of the
threads under a lock until it finds the current one.
Change-Id: I957a68438efb5b7d632d2b9b1b567fed7714e02a
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
Reviewed-on: https://review.gerrithub.io/365079
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
For a given I/O device, call a user function on the correct
thread for each channel.
Change-Id: I568a443184ac834c80c5e36b80aa9b6f8bb0ac99
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/362256
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This serves two purposes. First, some older compiler
chains don't support thread local storage. Second,
we're going to need a way to iterate the list of
threads in the future, so keep them in a list.
Change-Id: I80e709f4665afb03cf4bcf0db19ef8ea353acdc1
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/362255
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
When a thread is registered, the user must provide
a function pointer that can pass a message to that thread.
Change-Id: I743b5e0d6e3b5118c0a68d2fcedbccdd6fb237f9
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/362067
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
This will be used in the future to pass a message
to any given thread.
Change-Id: I3be5fe66244e360b7667427647fd8fdede110930
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/362066
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This wasn't used anywhere and we currently believe there
are superior software-only techniques for controlling
quality of service.
Change-Id: Icdadd5870ed0629b338c307d2619bbc242c3e7a3
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/362065
Reviewed-by: Jim Harris <james.r.harris@intel.com>
This is no longer used anywhere. For the places where we previously
used it, we've since found alternate solutions that do not
require it.
Change-Id: I738a80b95ef50348ce1c14969a3812b0a625b3fd
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-on: https://review.gerrithub.io/362064
Tested-by: <sys_sgsw@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Enforce exactly one trailing \n, and fix all of the existing cases.
Change-Id: I6218e4700e90aeb647eaee78089530c79993c8c8
Signed-off-by: Daniel Verkamp <daniel.verkamp@intel.com>
The free function is missed for dev in spdk_io_device_
unregister function.
Change-Id: Id212344dfcde2ae4780c631e3443f530ef25cfd1
Signed-off-by: Ziye Yang <ziye.yang@intel.com>
Some subsystems may wish to create unique I/O channels
which are not shared across all users of the same I/O
device on the same thread.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I3ade3675d57338cf85b6a301285e6f392bd6cd2e
This patch adds a basic framework for creating I/O channels
for I/O devices. An spdk_io_channel represents a one-to-one
mapping between a calling thread (represented by spdk_thread)
and an I/O device that the thread will perform I/O operations
on.
Signed-off-by: Jim Harris <james.r.harris@intel.com>
Change-Id: I658ab7f995cc962f4e2a204e058cdd3ad3fd735d
