Skip to contentSkip to conversation

nvme: enable retries for authentication commands

PATCH Superseded [email protected] 4 participants 7 messages Jan 25, 2024

Subsystems (3)

v1 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 nvme: enable retries for authentication commands

From: Hannes Reinecke <[email protected]>

Authentication commands are normal NVMe commands, and as such can return a status where NVME_SC_DNR is not set, indicating that the command should be retried.
Rather than checking NVME_SC_DNR manually for each command completion this patch adds a flag 'retry' to __nvme_submit_sync_cmd(), causing the REQ_FAILFAST_DRIVER option to be cleared and the command to be retried via the normal mechanism.

---
 drivers/nvme/host/auth.c    |  2 +-
 drivers/nvme/host/core.c    | 10 ++++++----
 drivers/nvme/host/fabrics.c | 10 +++++-----
 drivers/nvme/host/nvme.h    |  2 +-
 4 files changed, 13 insertions(+), 11 deletions(-)
drivers/nvme/host/auth.c +1 -1
index 35a460cfb4bf..c3ab7c86065e 100644
@@ -82,7 +82,7 @@ static int nvme_auth_submit(struct nvme_ctrl *ctrl, int qid,
 
 	ret = __nvme_submit_sync_cmd(q, &cmd, NULL, data, data_len,
 				     qid == 0 ? NVME_QID_ANY : qid,
-				     0, flags);
+				     0, flags, true);
 	if (ret > 0)
 		dev_warn(ctrl->device,
 			"qid %d auth_send failed with status %d\n", qid, ret);
drivers/nvme/host/core.c +6 -4
index 376efb310eb2..dade4fcb88d2 100644
@@ -1049,7 +1049,7 @@ EXPORT_SYMBOL_NS_GPL(nvme_execute_rq, NVME_TARGET_PASSTHRU);
  */
 int __nvme_submit_sync_cmd(struct request_queue *q, struct nvme_command *cmd,
 		union nvme_result *result, void *buffer, unsigned bufflen,
-		int qid, int at_head, blk_mq_req_flags_t flags)
+		int qid, int at_head, blk_mq_req_flags_t flags, bool retry)
 {
 	struct request *req;
 	int ret;
@@ -1063,6 +1063,8 @@ int __nvme_submit_sync_cmd(struct request_queue *q, struct nvme_command *cmd,
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 	nvme_init_request(req, cmd);
+	if (retry)
+		req->cmd_flags &= ~REQ_FAILFAST_DRIVER;
 
 	if (buffer && bufflen) {
 		ret = blk_rq_map_kern(q, req, buffer, bufflen, GFP_KERNEL);
@@ -1083,7 +1085,7 @@ int nvme_submit_sync_cmd(struct request_queue *q, struct nvme_command *cmd,
 		void *buffer, unsigned bufflen)
 {
 	return __nvme_submit_sync_cmd(q, cmd, NULL, buffer, bufflen,
-			NVME_QID_ANY, 0, 0);
+			NVME_QID_ANY, 0, 0, false);
 }
 EXPORT_SYMBOL_GPL(nvme_submit_sync_cmd);
 
@@ -1547,7 +1549,7 @@ static int nvme_features(struct nvme_ctrl *dev, u8 op, unsigned int fid,
 	c.features.dword11 = cpu_to_le32(dword11);
 
 	ret = __nvme_submit_sync_cmd(dev->admin_q, &c, &res,
-			buffer, buflen, NVME_QID_ANY, 0, 0);
+			buffer, buflen, NVME_QID_ANY, 0, 0, false);
 	if (ret >= 0 && result)
 		*result = le32_to_cpu(res.u32);
 	return ret;
@@ -2149,7 +2151,7 @@ static int nvme_sec_submit(void *data, u16 spsp, u8 secp, void *buffer, size_t l
 	cmd.common.cdw11 = cpu_to_le32(len);
 
 	return __nvme_submit_sync_cmd(ctrl->admin_q, &cmd, NULL, buffer, len,
-			NVME_QID_ANY, 1, 0);
+			NVME_QID_ANY, 1, 0, false);
 }
 
 static void nvme_configure_opal(struct nvme_ctrl *ctrl, bool was_suspended)
drivers/nvme/host/fabrics.c +5 -5
index 8140d8166c3e..66799d59fcd1 100644
@@ -180,7 +180,7 @@ int nvmf_reg_read32(struct nvme_ctrl *ctrl, u32 off, u32 *val)
 	cmd.prop_get.offset = cpu_to_le32(off);
 
 	ret = __nvme_submit_sync_cmd(ctrl->fabrics_q, &cmd, &res, NULL, 0,
-			NVME_QID_ANY, 0, 0);
+			NVME_QID_ANY, 0, 0, false);
 
 	if (ret >= 0)
 		*val = le64_to_cpu(res.u64);
@@ -226,7 +226,7 @@ int nvmf_reg_read64(struct nvme_ctrl *ctrl, u32 off, u64 *val)
 	cmd.prop_get.offset = cpu_to_le32(off);
 
 	ret = __nvme_submit_sync_cmd(ctrl->fabrics_q, &cmd, &res, NULL, 0,
-			NVME_QID_ANY, 0, 0);
+			NVME_QID_ANY, 0, 0, false);
 
 	if (ret >= 0)
 		*val = le64_to_cpu(res.u64);
@@ -271,7 +271,7 @@ int nvmf_reg_write32(struct nvme_ctrl *ctrl, u32 off, u32 val)
 	cmd.prop_set.value = cpu_to_le64(val);
 
 	ret = __nvme_submit_sync_cmd(ctrl->fabrics_q, &cmd, NULL, NULL, 0,
-			NVME_QID_ANY, 0, 0);
+			NVME_QID_ANY, 0, 0, false);
 	if (unlikely(ret))
 		dev_err(ctrl->device,
 			"Property Set error: %d, offset %#x\n",
@@ -451,7 +451,7 @@ int nvmf_connect_admin_queue(struct nvme_ctrl *ctrl)
 
 	ret = __nvme_submit_sync_cmd(ctrl->fabrics_q, &cmd, &res,
 			data, sizeof(*data), NVME_QID_ANY, 1,
-			BLK_MQ_REQ_RESERVED | BLK_MQ_REQ_NOWAIT);
+			BLK_MQ_REQ_RESERVED | BLK_MQ_REQ_NOWAIT, false);
 	if (ret) {
 		nvmf_log_connect_error(ctrl, ret, le32_to_cpu(res.u32),
 				       &cmd, data);
@@ -527,7 +527,7 @@ int nvmf_connect_io_queue(struct nvme_ctrl *ctrl, u16 qid)
 
 	ret = __nvme_submit_sync_cmd(ctrl->connect_q, &cmd, &res,
 			data, sizeof(*data), qid, 1,
-			BLK_MQ_REQ_RESERVED | BLK_MQ_REQ_NOWAIT);
+			BLK_MQ_REQ_RESERVED | BLK_MQ_REQ_NOWAIT, false);
 	if (ret) {
 		nvmf_log_connect_error(ctrl, ret, le32_to_cpu(res.u32),
 				       &cmd, data);
drivers/nvme/host/nvme.h +1 -1
index 396e6e72b6a3..743ef474d282 100644
@@ -829,7 +829,7 @@ int nvme_submit_sync_cmd(struct request_queue *q, struct nvme_command *cmd,
 int __nvme_submit_sync_cmd(struct request_queue *q, struct nvme_command *cmd,
 		union nvme_result *result, void *buffer, unsigned bufflen,
 		int qid, int at_head,
-		blk_mq_req_flags_t flags);
+		blk_mq_req_flags_t flags, bool retry);
 int nvme_set_features(struct nvme_ctrl *dev, unsigned int fid,
 		      unsigned int dword11, void *buffer, size_t buflen,
 		      u32 *result);
Signature · 1 line
2.35.3
maintainer
#1 lore raw
Quoted 9 lines · On 1/25/24 6:09 AM, [email protected] wrote:

From: Hannes Reinecke <[email protected]>

Authentication commands are normal NVMe commands, and as such can return a status where NVME_SC_DNR is not set, indicating that the command should be retried.
Rather than checking NVME_SC_DNR manually for each command completion this patch adds a flag 'retry' to __nvme_submit_sync_cmd(), causing the REQ_FAILFAST_DRIVER option to be cleared and the command to be retried via the normal mechanism.

Can we please do this without adding Yet Another parameter to that function?

Signature · 1 line
Jens Axboe
#2 lore raw
Quoted 11 lines · Jens Axboe
Quoted 9 lines · On 1/25/24 6:09 AM, [email protected] wrote:

From: Hannes Reinecke <[email protected]>

Authentication commands are normal NVMe commands, and as such can return a status where NVME_SC_DNR is not set, indicating that the command should be retried.
Rather than checking NVME_SC_DNR manually for each command completion this patch adds a flag 'retry' to __nvme_submit_sync_cmd(), causing the REQ_FAILFAST_DRIVER option to be cleared and the command to be retried via the normal mechanism.

Quoted 2 lines

Can we please do this without adding Yet Another parameter to that function?

Heh, I wrote about the same reply a few minutes ago. I'll try to think what we can do instead.

maintainer
#3 lore raw
Quoted 12 lines · Christoph Hellwig
Quoted 11 lines · Jens Axboe
Quoted 9 lines · On 1/25/24 6:09 AM, [email protected] wrote:

From: Hannes Reinecke <[email protected]>

Authentication commands are normal NVMe commands, and as such can return a status where NVME_SC_DNR is not set, indicating that the command should be retried.
Rather than checking NVME_SC_DNR manually for each command completion this patch adds a flag 'retry' to __nvme_submit_sync_cmd(), causing the REQ_FAILFAST_DRIVER option to be cleared and the command to be retried via the normal mechanism.

Quoted 5 lines
Quoted 2 lines

Can we please do this without adding Yet Another parameter to that function?

Heh, I wrote about the same reply a few minutes ago. I'll try to think what we can do instead.

We could do something like this:

drivers/nvme/host/core.c +2 -0
index 8ebdfd623e0f..200326dc33c1 100644
@@ -1057,6 +1057,8 @@ int __nvme_submit_sync_cmd(struct request_queue *q, struct nvme_command *cmd,
        if (IS_ERR(req))
                return PTR_ERR(req);
        nvme_init_request(req, cmd);
+       if (flags & BLK_MQ_REQ_RETRY)
+               req->cmd_flags &= ~REQ_FAILFAST_DRIVER;
 
        if (buffer && bufflen) {
                ret = blk_rq_map_kern(q, req, buffer, bufflen, GFP_KERNEL);
include/linux/blk-mq.h +2 -0
index 1ab3081c82ed..7983e1abb622 100644
@@ -719,6 +719,8 @@ enum {
        BLK_MQ_REQ_RESERVED     = (__force blk_mq_req_flags_t)(1 << 1),
        /* set RQF_PM */
        BLK_MQ_REQ_PM           = (__force blk_mq_req_flags_t)(1 << 2),
+       /* enable command retries */
+       BLK_MQ_REQ_RETRY        = (__force blk_mq_req_flags_t)(1 << 3),
 };
 
 struct request *blk_mq_alloc_request(struct request_queue *q, 

blk_opf_t opf,

but that's arguably a misuse of block layer flags ...

Cheers,

Hannes

Signature · 5 lines
Dr. Hannes Reinecke                Kernel Storage Architect
[email protected]                              +49 911 74053 688
SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), GF: Ivo Totev, Andrew McDonald,
Werner Knoblich
maintainer
#4 lore raw
Quoted 13 lines · Hannes Reinecke
Quoted 12 lines · Christoph Hellwig
Quoted 11 lines · Jens Axboe
Quoted 9 lines · On 1/25/24 6:09 AM, [email protected] wrote:

From: Hannes Reinecke <[email protected]>

Authentication commands are normal NVMe commands, and as such can return a status where NVME_SC_DNR is not set, indicating that the command should be retried.
Rather than checking NVME_SC_DNR manually for each command completion this patch adds a flag 'retry' to __nvme_submit_sync_cmd(), causing the REQ_FAILFAST_DRIVER option to be cleared and the command to be retried via the normal mechanism.

Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

Can we please do this without adding Yet Another parameter to that function?

Heh, I wrote about the same reply a few minutes ago. I'll try to think what we can do instead.

We could do something like this:

That won't really do much, and yeah wouldn't be the way even if complete. How about just not setting the flag in the first place? nvme_init_request() defaults to doing it, so either pass in the init flags to that, or just clear it in there and have the callers decide if they want to set it.

Signature · 1 line
Jens Axboe
maintainer
#5 lore raw
Quoted 20 lines · Jens Axboe
Quoted 19 lines · Hannes Reinecke
Quoted 17 lines · Christoph Hellwig
Quoted 11 lines · Jens Axboe
Quoted 9 lines · On 1/25/24 6:09 AM, [email protected] wrote:

From: Hannes Reinecke <[email protected]>

Authentication commands are normal NVMe commands, and as such can return a status where NVME_SC_DNR is not set, indicating that the command should be retried.
Rather than checking NVME_SC_DNR manually for each command completion this patch adds a flag 'retry' to __nvme_submit_sync_cmd(), causing the REQ_FAILFAST_DRIVER option to be cleared and the command to be retried via the normal mechanism.

Quoted 2 lines

Can we please do this without adding Yet Another parameter to that function?

Heh, I wrote about the same reply a few minutes ago. I'll try to think what we can do instead.

Quoted 8 lines
Quoted 1 line

We could do something like this:

That won't really do much, and yeah wouldn't be the way even if complete. How about just not setting the flag in the first place? nvme_init_request() defaults to doing it, so either pass in the init flags to that, or just clear it in there and have the callers decide if they want to set it.

But that's the thing; the request is never visible to the caller. So moving things out of the nvme_init_request() doesn't really help.

And open-coding __nvme_submit_sync_cmd() also doesn't look very appealing.

Cheers,

Hannes

Signature · 5 lines
Dr. Hannes Reinecke                Kernel Storage Architect
[email protected]                              +49 911 74053 688
SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), GF: Ivo Totev, Andrew McDonald,
Werner Knoblich
maintainer
#6 lore raw
Quoted 29 lines · Hannes Reinecke
Quoted 20 lines · Jens Axboe
Quoted 19 lines · Hannes Reinecke
Quoted 17 lines · Christoph Hellwig
Quoted 11 lines · Jens Axboe
Quoted 9 lines · On 1/25/24 6:09 AM, [email protected] wrote:

From: Hannes Reinecke <[email protected]>

Authentication commands are normal NVMe commands, and as such can return a status where NVME_SC_DNR is not set, indicating that the command should be retried.
Rather than checking NVME_SC_DNR manually for each command completion this patch adds a flag 'retry' to __nvme_submit_sync_cmd(), causing the REQ_FAILFAST_DRIVER option to be cleared and the command to be retried via the normal mechanism.

Quoted 2 lines

Can we please do this without adding Yet Another parameter to that function?

Heh, I wrote about the same reply a few minutes ago. I'll try to think what we can do instead.

Quoted 8 lines
Quoted 1 line

We could do something like this:

That won't really do much, and yeah wouldn't be the way even if complete. How about just not setting the flag in the first place? nvme_init_request() defaults to doing it, so either pass in the init flags to that, or just clear it in there and have the callers decide if they want to set it.

Quoted 5 lines

But that's the thing; the request is never visible to the caller. So moving things out of the nvme_init_request() doesn't really help.

And open-coding __nvme_submit_sync_cmd() also doesn't look very appealing.

__nvme_submit_sync_cmd() already takes another bool (at_head) and blk-mq flags. You could have it take some nvme specific flags instead, and fold those two parameters into a single flags. Once that is done, adding the retry flag on top would be trivial.

And as a side benefit, it gets RID of a bool flag rather than add another random one.

Signature · 1 line
Jens Axboe