github-mirror/linux
1
0
Fork 0
mirror of https://github.com/torvalds/linux.git synced 2026-11-07 06:17:02 +01:00
Code Issues Actions Packages Projects Releases Wiki Activity Graph

nvme-rdma: parallelize I/O queue allocation and startup

Browse Source 
Refactor nvme rdma I/O queue setup to use async API, combining
allocation and startup into a single parallel operation per queue. This
reduces connection and reconnection setup time when there are delays in
establishing connections, which is especially important for
high-core-count hosts.

Key changes:
- Use async API to facilitate parallel calls for io queue setup.
- Add nvme_rdma_setup_ctx for propagating errors from async workers.
- Remove nvme_rdma_alloc_io_queues() and nvme_rdma_start_io_queues();
  their logic is folded into nvme_rdma_setup_io_queues() and
  nvme_rdma_configure_io_queues().
- Move queue count negotiation (nvme_set_queue_count,
  nvmf_set_io_queues) from the removed nvme_rdma_alloc_io_queues()
  into nvme_rdma_configure_io_queues().

Testing on a 64-core host with 64 IO-queues shows
nvme-rdma connection time reduced from ~1.4s to 416ms.

Signed-off-by: Surabhi Gogte <sgogte@purestorage.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Keith Busch <kbusch@kernel.org>
This commit is contained in:
Surabhi Gogte 2026-06-26 22:15:51 -06:00 committed by Keith Busch
parent f4254b18d4
commit 2a8513091d
1 changed files with 76 additions and 46 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

122
drivers/nvme/host/rdma.c
View File

@ -16,6 +16,7 @@
#include <linux/types.h> #include <linux/types.h>
#include <linux/list.h> #include <linux/list.h>
#include <linux/mutex.h> #include <linux/mutex.h>
#include <linux/async.h>
#include <linux/scatterlist.h> #include <linux/scatterlist.h>
#include <linux/nvme.h> #include <linux/nvme.h>
#include <linux/unaligned.h> #include <linux/unaligned.h>
@ -100,6 +101,11 @@ struct nvme_rdma_queue {
struct mutex queue_lock; struct mutex queue_lock;
}; };
struct nvme_rdma_setup_ctx {
struct nvme_rdma_queue *queue;
int *err;
};
struct nvme_rdma_ctrl { struct nvme_rdma_ctrl {
/* read only in the hot path */ /* read only in the hot path */
struct nvme_rdma_queue *queues; struct nvme_rdma_queue *queues;
@ -690,60 +696,68 @@ static int nvme_rdma_start_queue(struct nvme_rdma_ctrl *ctrl, int idx)
return ret; return ret;
} }
static int nvme_rdma_start_io_queues(struct nvme_rdma_ctrl *ctrl, static void nvme_rdma_setup_queue_async(void *data, async_cookie_t cookie)
int first, int last)
{ {
int i, ret = 0; struct nvme_rdma_setup_ctx *ctx = data;
struct nvme_rdma_queue *queue;
int ret;
for (i = first; i < last; i++) { queue = ctx->queue;
ret = nvme_rdma_start_queue(ctrl, i); ret = nvme_rdma_alloc_queue(queue);
if (ret) if (ret)
goto out_stop_queues; goto out_err;
}
return 0; ret = nvme_rdma_start_queue(queue->ctrl, nvme_rdma_queue_idx(queue));
if (ret)
goto out_err;
out_stop_queues: return;
for (i--; i >= first; i--) out_err:
nvme_rdma_stop_queue(&ctrl->queues[i]); WRITE_ONCE(*ctx->err, ret);
return ret;
} }
static int nvme_rdma_alloc_io_queues(struct nvme_rdma_ctrl *ctrl) static int nvme_rdma_setup_io_queues(struct nvme_rdma_ctrl *ctrl,
unsigned int first, unsigned int last, size_t queue_size)
{ {
struct nvmf_ctrl_options *opts = ctrl->ctrl.opts; ASYNC_DOMAIN_EXCLUSIVE(queue_domain);
unsigned int nr_io_queues; struct nvme_rdma_setup_ctx *ctxs;
int i, ret; int nr_queues = last - first;
int err = 0, i, ret;
nr_io_queues = nvmf_nr_io_queues(opts); ctxs = kmalloc_objs(*ctxs, nr_queues);
ret = nvme_set_queue_count(&ctrl->ctrl, &nr_io_queues); if (!ctxs)
if (ret)
return ret;
if (nr_io_queues == 0) {
dev_err(ctrl->ctrl.device,
"unable to set any I/O queues\n");
return -ENOMEM; return -ENOMEM;
for (i = 0; i < nr_queues; i++) {
struct nvme_rdma_queue *queue = &ctrl->queues[first + i];
queue->ctrl = ctrl;
queue->queue_size = queue_size;
ctxs[i].queue = queue;
ctxs[i].err = &err;
async_schedule_domain(nvme_rdma_setup_queue_async, &ctxs[i],
&queue_domain);
} }
ctrl->ctrl.queue_count = nr_io_queues + 1; async_synchronize_full_domain(&queue_domain);
dev_info(ctrl->ctrl.device, kfree(ctxs);
"creating %d I/O queues.\n", nr_io_queues);
nvmf_set_io_queues(opts, nr_io_queues, ctrl->io_queues); ret = READ_ONCE(err);
for (i = 1; i < ctrl->ctrl.queue_count; i++) { if (ret)
ctrl->queues[i].ctrl = ctrl; goto out_free_queues;
ctrl->queues[i].queue_size = ctrl->ctrl.sqsize + 1;
ret = nvme_rdma_alloc_queue(&ctrl->queues[i]);
if (ret)
goto out_free_queues;
}
return 0; return 0;
out_free_queues: out_free_queues:
for (i--; i >= 1; i--) for (i = 0; i < nr_queues; i++) {
nvme_rdma_free_queue(&ctrl->queues[i]); struct nvme_rdma_queue *queue =
&ctrl->queues[first + i];
if (test_bit(NVME_RDMA_Q_LIVE, &queue->flags))
nvme_rdma_stop_queue(queue);
if (test_bit(NVME_RDMA_Q_ALLOCATED, &queue->flags))
nvme_rdma_free_queue(queue);
}
return ret; return ret;
} }
@ -862,12 +876,23 @@ static int nvme_rdma_configure_admin_queue(struct nvme_rdma_ctrl *ctrl,
static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new) static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new)
{ {
unsigned int nr_io_queues;
int ret, nr_queues; int ret, nr_queues;
ret = nvme_rdma_alloc_io_queues(ctrl); nr_io_queues = nvmf_nr_io_queues(ctrl->ctrl.opts);
ret = nvme_set_queue_count(&ctrl->ctrl, &nr_io_queues);
if (ret) if (ret)
return ret; return ret;
if (nr_io_queues == 0) {
dev_err(ctrl->ctrl.device, "unable to set any I/O queues\n");
return -ENOMEM;
}
ctrl->ctrl.queue_count = nr_io_queues + 1;
dev_info(ctrl->ctrl.device, "creating %d I/O queues.\n", nr_io_queues);
nvmf_set_io_queues(ctrl->ctrl.opts, nr_io_queues, ctrl->io_queues);
if (new) { if (new) {
ret = nvme_rdma_alloc_tag_set(&ctrl->ctrl); ret = nvme_rdma_alloc_tag_set(&ctrl->ctrl);
if (ret) if (ret)
@ -880,7 +905,9 @@ static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new)
* queue number might have changed. * queue number might have changed.
*/ */
nr_queues = min(ctrl->tag_set.nr_hw_queues + 1, ctrl->ctrl.queue_count); nr_queues = min(ctrl->tag_set.nr_hw_queues + 1, ctrl->ctrl.queue_count);
ret = nvme_rdma_start_io_queues(ctrl, 1, nr_queues); ret = nvme_rdma_setup_io_queues(ctrl, 1, nr_queues,
ctrl->ctrl.sqsize + 1);
if (ret) if (ret)
goto out_cleanup_tagset; goto out_cleanup_tagset;
@ -904,12 +931,15 @@ static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new)
/* /*
* If the number of queues has increased (reconnect case) * If the number of queues has increased (reconnect case)
* start all new queues now. * setup all new queues now.
*/ */
ret = nvme_rdma_start_io_queues(ctrl, nr_queues, if (ctrl->tag_set.nr_hw_queues + 1 > nr_queues) {
ctrl->tag_set.nr_hw_queues + 1); ret = nvme_rdma_setup_io_queues(ctrl, nr_queues,
if (ret) ctrl->tag_set.nr_hw_queues + 1,
goto out_wait_freeze_timed_out; ctrl->ctrl.sqsize + 1);
if (ret)
goto out_wait_freeze_timed_out;
}
return 0; return 0;