888def105f
nvme_bdev_ctrlr->reset_bio is accessed only by the thread which called spdk_for_each_channel() and the callback to spdk_for_each_channel() is called after unwinding stack via message. Hence bdev_nvme_reset() can call _bdev_nvme_reset() rather than _bdev_nvme_reset_start() and store bio into nvme_bdev_ctrlr->reset_bio after _bdev_nvme_reset() returns zero. Then inline _bdev_nvme_reset_start() into _bdev_nvme_reset() because _bdev_nvme_reset_start() has only a single caller now. The following patches will introduce subsystem and bdev_nvme_reset() will reset all controllers of a subsystem sequentially. These further clean-ups will be helpful for such enhancement. Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Change-Id: I25abaa5be511c967ce20c92e4caa7dfeb3e09dd1 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/7240 Community-CI: Broadcom CI Reviewed-by: Changpeng Liu <changpeng.liu@intel.com> Reviewed-by: Aleksey Marchuk <alexeymar@mellanox.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> |
||
|---|---|---|
| .. | ||
| bdev_nvme_cuse_rpc.c | ||
| bdev_nvme_rpc.c | ||
| bdev_nvme.c | ||
| bdev_nvme.h | ||
| bdev_ocssd_rpc.c | ||
| bdev_ocssd.c | ||
| bdev_ocssd.h | ||
| common.c | ||
| common.h | ||
| Makefile | ||
| nvme_rpc.c | ||
| vbdev_opal_rpc.c | ||
| vbdev_opal.c | ||
| vbdev_opal.h | ||