* [PATCH V3] nvmet: use shared workqueue for file-ns
@ 2018-06-26 22:32 Chaitanya Kulkarni
2018-06-27 8:04 ` Sagi Grimberg
2018-06-29 9:07 ` Christoph Hellwig
0 siblings, 2 replies; 3+ messages in thread
From: Chaitanya Kulkarni @ 2018-06-26 22:32 UTC (permalink / raw)
This patch removes the per-ns workqueue and adds a global workqueue
which is shared across all the namespaces of multiple subsystems to
offload the I/O submission for buffered I/Os when namespace is
file-backed.
Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni at wdc.com>
---
Changes since V2 :-
1. Change the worqeueue name identifier to "nvmet-buffered-io-wq"
along with global workqueue variable name from "file_wq" ->
"buffered_io_wq".
2. Please note that this patch is generated on the top of flush all
patche series "[PATCH V4 0/2] nvmet: implement flush all support " :-
http://lists.infradead.org/pipermail/linux-nvme/2018-June/018967.html
http://lists.infradead.org/pipermail/linux-nvme/2018-June/018965.html
http://lists.infradead.org/pipermail/linux-nvme/2018-June/018966.html
Changes since V1 :-
1. Don't use per subsys based workqueue, make it global.
---
drivers/nvme/target/core.c | 8 ++++++++
drivers/nvme/target/io-cmd-file.c | 17 +++--------------
drivers/nvme/target/nvmet.h | 3 ++-
3 files changed, 13 insertions(+), 15 deletions(-)
diff --git a/drivers/nvme/target/core.c b/drivers/nvme/target/core.c
index 4fe474c632f2..5d27c46363b8 100644
--- a/drivers/nvme/target/core.c
+++ b/drivers/nvme/target/core.c
@@ -18,6 +18,7 @@
#include "nvmet.h"
+struct workqueue_struct *buffered_io_wq;
static const struct nvmet_fabrics_ops *nvmet_transports[NVMF_TRTYPE_MAX];
static DEFINE_IDA(cntlid_ida);
@@ -1199,6 +1200,12 @@ static int __init nvmet_init(void)
{
int error;
+ buffered_io_wq = alloc_workqueue("nvmet-buffered-io-wq",
+ WQ_MEM_RECLAIM, 0);
+ if (!buffered_io_wq) {
+ error = -ENOMEM;
+ goto out;
+ }
error = nvmet_init_discovery();
if (error)
goto out;
@@ -1219,6 +1226,7 @@ static void __exit nvmet_exit(void)
nvmet_exit_configfs();
nvmet_exit_discovery();
ida_destroy(&cntlid_ida);
+ destroy_workqueue(buffered_io_wq);
BUILD_BUG_ON(sizeof(struct nvmf_disc_rsp_page_entry) != 1024);
BUILD_BUG_ON(sizeof(struct nvmf_disc_rsp_page_hdr) != 1024);
diff --git a/drivers/nvme/target/io-cmd-file.c b/drivers/nvme/target/io-cmd-file.c
index d1c79a3c4f9b..ef8942f0d3d8 100644
--- a/drivers/nvme/target/io-cmd-file.c
+++ b/drivers/nvme/target/io-cmd-file.c
@@ -16,11 +16,8 @@
void nvmet_file_ns_disable(struct nvmet_ns *ns)
{
if (ns->file) {
- if (ns->buffered_io) {
- flush_workqueue(ns->file_wq);
- destroy_workqueue(ns->file_wq);
- ns->file_wq = NULL;
- }
+ if (ns->buffered_io)
+ flush_workqueue(buffered_io_wq);
mempool_destroy(ns->bvec_pool);
ns->bvec_pool = NULL;
kmem_cache_destroy(ns->bvec_cache);
@@ -70,14 +67,6 @@ int nvmet_file_ns_enable(struct nvmet_ns *ns)
goto err;
}
- if (ns->buffered_io) {
- ns->file_wq = alloc_workqueue("nvmet-file", WQ_MEM_RECLAIM, 0);
- if (!ns->file_wq) {
- ret = -ENOMEM;
- goto err;
- }
- }
-
return ret;
err:
ns->size = 0;
@@ -215,7 +204,7 @@ static void nvmet_file_buffered_io_work(struct work_struct *w)
static void nvmet_file_execute_rw_buffered_io(struct nvmet_req *req)
{
INIT_WORK(&req->work, nvmet_file_buffered_io_work);
- queue_work(req->ns->file_wq, &req->work);
+ queue_work(buffered_io_wq, &req->work);
}
void nvmet_file_flush_work(struct work_struct *w)
diff --git a/drivers/nvme/target/nvmet.h b/drivers/nvme/target/nvmet.h
index 48928b1acded..a949f6604152 100644
--- a/drivers/nvme/target/nvmet.h
+++ b/drivers/nvme/target/nvmet.h
@@ -66,7 +66,6 @@ struct nvmet_ns {
uuid_t uuid;
bool buffered_io;
- struct workqueue_struct *file_wq;
bool enabled;
struct nvmet_subsys *subsys;
const char *device_path;
@@ -272,6 +271,8 @@ struct nvmet_req {
const struct nvmet_fabrics_ops *ops;
};
+extern struct workqueue_struct *buffered_io_wq;
+
static inline void nvmet_set_status(struct nvmet_req *req, u16 status)
{
req->rsp->status = cpu_to_le16(status << 1);
--
2.17.0
^ permalink raw reply related [flat|nested] 3+ messages in thread
* [PATCH V3] nvmet: use shared workqueue for file-ns
2018-06-26 22:32 [PATCH V3] nvmet: use shared workqueue for file-ns Chaitanya Kulkarni
@ 2018-06-27 8:04 ` Sagi Grimberg
2018-06-29 9:07 ` Christoph Hellwig
1 sibling, 0 replies; 3+ messages in thread
From: Sagi Grimberg @ 2018-06-27 8:04 UTC (permalink / raw)
Reviewed-by: Sagi Grimberg <sagi at grimberg.me>
^ permalink raw reply [flat|nested] 3+ messages in thread
* [PATCH V3] nvmet: use shared workqueue for file-ns
2018-06-26 22:32 [PATCH V3] nvmet: use shared workqueue for file-ns Chaitanya Kulkarni
2018-06-27 8:04 ` Sagi Grimberg
@ 2018-06-29 9:07 ` Christoph Hellwig
1 sibling, 0 replies; 3+ messages in thread
From: Christoph Hellwig @ 2018-06-29 9:07 UTC (permalink / raw)
Thanks, I've folded this into the original commit for the file based
namespaces.
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2018-06-29 9:07 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-06-26 22:32 [PATCH V3] nvmet: use shared workqueue for file-ns Chaitanya Kulkarni
2018-06-27 8:04 ` Sagi Grimberg
2018-06-29 9:07 ` Christoph Hellwig
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.