Skip to contentSkip to conversation

nvme target 2.1 and independent identify ns

PATCH In review Keith Busch 6 participants 38 messages Nov 7, 2024 – Nov 13, 2024

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

v4 cover nvme target 2.1 and independent identify ns

From: Keith Busch <[email protected]>

Changes from v3:

Defined the log structs for logs supported and features supported

Used defines for those log attribute bits

Fixed some patch ordering to avoid compiler error hazards part way through the last series.

Removed unnecessary type-casts

Fixed setting correct request completion status on one of the logs

Used the correct and reviewed version of the host side rotational support

Added the "no_vwc" patches from Guixin at the end since it is all in the same area of the code

Added more comments explaining the relationship between the namespaces and their "endurance groups".

Added reviews.

Guixin Liu (2):
nvme: check ns's volatile write cache not present
nvmet: report ns's vwc not present

Keith Busch (9):
nvmet: implement id ns for nvm command set
nvmet: implement active command set ns list
nvmet: implement supported log pages
nvmet: implement supported features log
nvmet: implement crto property
nvmet: declare 2.1 version compliance
nvmet: implement endurance groups
nvmet: implement rotational media information log
nvmet: support for csi identify ns

Matias Bjørling (1):
nvme: use command set independent id ns if available

Wang Yugui (1):
nvme: add rotational support

 drivers/nvme/host/core.c          |  19 ++-
 drivers/nvme/host/nvme.h          |   1 +
 drivers/nvme/target/admin-cmd.c   | 257 +++++++++++++++++++++++++++++-
 drivers/nvme/target/fabrics-cmd.c |   3 +
 drivers/nvme/target/nvmet.h       |   2 +-
 include/linux/nvme.h              |  67 +++++++-
 6 files changed, 337 insertions(+), 12 deletions(-)
Signature · 1 line
2.43.5
#1 lore raw

PATCH v4 1/13 nvmet: implement id ns for nvm command set

From: Keith Busch <[email protected]>

We don't report anything here, but it's a mandatory identification for nvme 2.1.

---
 drivers/nvme/target/admin-cmd.c | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)
drivers/nvme/target/admin-cmd.c +16 -2
index 081f0473cd9ea..8d06dba42bcb3 100644
@@ -685,6 +685,20 @@ static void nvmet_execute_identify_ctrl_nvm(struct nvmet_req *req)
 		   nvmet_zero_sgl(req, 0, sizeof(struct nvme_id_ctrl_nvm)));
 }
 
+static void nvme_execute_identify_ns_nvm(struct nvmet_req *req)
+{
+	u16 status;
+
+	status = nvmet_req_find_ns(req);
+	if (status)
+		goto out;
+
+	status = nvmet_copy_to_sgl(req, 0, ZERO_PAGE(0),
+				   NVME_IDENTIFY_DATA_SIZE);
+out:
+	nvmet_req_complete(req, status);
+}
+
 static void nvmet_execute_identify(struct nvmet_req *req)
 {
 	if (!nvmet_check_transfer_len(req, NVME_IDENTIFY_DATA_SIZE))
@@ -706,8 +720,8 @@ static void nvmet_execute_identify(struct nvmet_req *req)
 	case NVME_ID_CNS_CS_NS:
 		switch (req->cmd->identify.csi) {
 		case NVME_CSI_NVM:
-			/* Not supported */
-			break;
+			nvme_execute_identify_ns_nvm(req);
+			return;
 		case NVME_CSI_ZNS:
 			if (IS_ENABLED(CONFIG_BLK_DEV_ZONED)) {
 				nvmet_execute_identify_ns_zns(req);
Signature · 1 line
2.43.5
Reviewed-by on v4 1/13
#2 lore raw

Re: [PATCHv4 01/13] nvmet: implement id ns for nvm command set

Quoted 37 lines · Keith Busch

From: Keith Busch <[email protected]>

We don't report anything here, but it's a mandatory identification for nvme 2.1.

---
  drivers/nvme/target/admin-cmd.c | 18 ++++++++++++++++--
  1 file changed, 16 insertions(+), 2 deletions(-)
drivers/nvme/target/admin-cmd.c +14 -0
index 081f0473cd9ea..8d06dba42bcb3 100644
@@ -685,6 +685,20 @@ static void nvmet_execute_identify_ctrl_nvm(struct nvmet_req *req)
  		   nvmet_zero_sgl(req, 0, sizeof(struct nvme_id_ctrl_nvm)));
  }
 
+static void nvme_execute_identify_ns_nvm(struct nvmet_req *req)
+{
+	u16 status;
+
+	status = nvmet_req_find_ns(req);
+	if (status)
+		goto out;
+
+	status = nvmet_copy_to_sgl(req, 0, ZERO_PAGE(0),
+				   NVME_IDENTIFY_DATA_SIZE);
+out:
+	nvmet_req_complete(req, status);
+}
+
  static void nvmet_execute_identify(struct nvmet_req *req)
  {
  	if (!nvmet_check_transfer_len(req, NVME_IDENTIFY_DATA_SIZE))

drivers/nvme/target/admin-cmd.c static void nvmet_execute_identify(struct nvmet_req *req) · line 727

Quoted 10 lines
drivers/nvme/target/admin-cmd.c +2 -2
@@ -706,8 +720,8 @@ static void nvmet_execute_identify(struct nvmet_req *req)
  	case NVME_ID_CNS_CS_NS:
  		switch (req->cmd->identify.csi) {
  		case NVME_CSI_NVM:
-			/* Not supported */
-			break;
+			nvme_execute_identify_ns_nvm(req);
+			return;
  		case NVME_CSI_ZNS:
  			if (IS_ENABLED(CONFIG_BLK_DEV_ZONED)) {
  				nvmet_execute_identify_ns_zns(req);

Looks good.

#4 lore raw

PATCH v4 2/13 nvmet: implement active command set ns list

In reply to Keith Busch

From: Keith Busch <[email protected]>

This is required for nvme 2.1 for targets that support multiple command sets. We support NVM and ZNS, so are required to support this identification.

---
 drivers/nvme/target/admin-cmd.c | 9 +++++++--
 include/linux/nvme.h            | 1 +
 2 files changed, 8 insertions(+), 2 deletions(-)
drivers/nvme/target/admin-cmd.c +7 -2
index 8d06dba42bcb3..a13242e791c0f 100644
@@ -576,7 +576,7 @@ static void nvmet_execute_identify_ns(struct nvmet_req *req)
 	nvmet_req_complete(req, status);
 }
 
-static void nvmet_execute_identify_nslist(struct nvmet_req *req)
+static void nvmet_execute_identify_nslist(struct nvmet_req *req, bool match_css)
 {
 	static const int buf_size = NVME_IDENTIFY_DATA_SIZE;
 	struct nvmet_ctrl *ctrl = req->sq->ctrl;
@@ -606,6 +606,8 @@ static void nvmet_execute_identify_nslist(struct nvmet_req *req)
 	xa_for_each(&ctrl->subsys->namespaces, idx, ns) {
 		if (ns->nsid <= min_nsid)
 			continue;
+		if (match_css && req->ns->csi != req->cmd->identify.csi)
+			continue;
 		list[i++] = cpu_to_le32(ns->nsid);
 		if (i == buf_size / sizeof(__le32))
 			break;
@@ -712,7 +714,7 @@ static void nvmet_execute_identify(struct nvmet_req *req)
 		nvmet_execute_identify_ctrl(req);
 		return;
 	case NVME_ID_CNS_NS_ACTIVE_LIST:
-		nvmet_execute_identify_nslist(req);
+		nvmet_execute_identify_nslist(req, false);
 		return;
 	case NVME_ID_CNS_NS_DESC_LIST:
 		nvmet_execute_identify_desclist(req);
@@ -743,6 +745,9 @@ static void nvmet_execute_identify(struct nvmet_req *req)
 			break;
 		}
 		break;
+	case NVME_ID_CNS_NS_ACTIVE_LIST_CS:
+		nvmet_execute_identify_nslist(req, true);
+		return;
 	}
 
 	pr_debug("unhandled identify cns %d on qid %d\n",
include/linux/nvme.h +1 -0
index b58d9405d65e0..0f263c7e63192 100644
@@ -522,6 +522,7 @@ enum {
 	NVME_ID_CNS_NS_DESC_LIST	= 0x03,
 	NVME_ID_CNS_CS_NS		= 0x05,
 	NVME_ID_CNS_CS_CTRL		= 0x06,
+	NVME_ID_CNS_NS_ACTIVE_LIST_CS	= 0x07,
 	NVME_ID_CNS_NS_CS_INDEP		= 0x08,
 	NVME_ID_CNS_NS_PRESENT_LIST	= 0x10,
 	NVME_ID_CNS_NS_PRESENT		= 0x11,
Signature · 1 line
2.43.5
Reviewed-by on v4 2/13
#5 lore raw

Re: [PATCHv4 02/13] nvmet: implement active command set ns list

Quoted 60 lines · Keith Busch

From: Keith Busch <[email protected]>

This is required for nvme 2.1 for targets that support multiple command sets. We support NVM and ZNS, so are required to support this identification.

---
  drivers/nvme/target/admin-cmd.c | 9 +++++++--
  include/linux/nvme.h            | 1 +
  2 files changed, 8 insertions(+), 2 deletions(-)
drivers/nvme/target/admin-cmd.c +7 -2
index 8d06dba42bcb3..a13242e791c0f 100644
@@ -576,7 +576,7 @@ static void nvmet_execute_identify_ns(struct nvmet_req *req)
  	nvmet_req_complete(req, status);
  }
 
-static void nvmet_execute_identify_nslist(struct nvmet_req *req)
+static void nvmet_execute_identify_nslist(struct nvmet_req *req, bool match_css)
  {
  	static const int buf_size = NVME_IDENTIFY_DATA_SIZE;
  	struct nvmet_ctrl *ctrl = req->sq->ctrl;
@@ -606,6 +606,8 @@ static void nvmet_execute_identify_nslist(struct nvmet_req *req)
  	xa_for_each(&ctrl->subsys->namespaces, idx, ns) {
  		if (ns->nsid <= min_nsid)
  			continue;
+		if (match_css && req->ns->csi != req->cmd->identify.csi)
+			continue;
  		list[i++] = cpu_to_le32(ns->nsid);
  		if (i == buf_size / sizeof(__le32))
  			break;
@@ -712,7 +714,7 @@ static void nvmet_execute_identify(struct nvmet_req *req)
  		nvmet_execute_identify_ctrl(req);
  		return;
  	case NVME_ID_CNS_NS_ACTIVE_LIST:
-		nvmet_execute_identify_nslist(req);
+		nvmet_execute_identify_nslist(req, false);
  		return;
  	case NVME_ID_CNS_NS_DESC_LIST:
  		nvmet_execute_identify_desclist(req);
@@ -743,6 +745,9 @@ static void nvmet_execute_identify(struct nvmet_req *req)
  			break;
  		}
  		break;
+	case NVME_ID_CNS_NS_ACTIVE_LIST_CS:
+		nvmet_execute_identify_nslist(req, true);
+		return;
  	}
 
  	pr_debug("unhandled identify cns %d on qid %d\n",
include/linux/nvme.h +0 -0
index b58d9405d65e0..0f263c7e63192 100644

include/linux/nvme.h enum { · line 528

Quoted 7 lines
include/linux/nvme.h +1 -0
@@ -522,6 +522,7 @@ enum {
  	NVME_ID_CNS_NS_DESC_LIST	= 0x03,
  	NVME_ID_CNS_CS_NS		= 0x05,
  	NVME_ID_CNS_CS_CTRL		= 0x06,
+	NVME_ID_CNS_NS_ACTIVE_LIST_CS	= 0x07,
  	NVME_ID_CNS_NS_CS_INDEP		= 0x08,
  	NVME_ID_CNS_NS_PRESENT_LIST	= 0x10,
  	NVME_ID_CNS_NS_PRESENT		= 0x11,

Looks good.

#6 lore raw

PATCH v4 3/13 nvmet: implement supported log pages

In reply to Keith Busch

From: Keith Busch <[email protected]>

This log is required for nvme 2.1.

---
 drivers/nvme/target/admin-cmd.c | 27 +++++++++++++++++++++++++++
 include/linux/nvme.h            |  9 +++++++++
 2 files changed, 36 insertions(+)
drivers/nvme/target/admin-cmd.c +27 -0
index a13242e791c0f..712b962fe77ac 100644
@@ -71,6 +71,31 @@ static void nvmet_execute_get_log_page_error(struct nvmet_req *req)
 	nvmet_req_complete(req, 0);
 }
 
+static void nvmet_execute_get_supported_log_pages(struct nvmet_req *req)
+{
+	struct nvme_supported_log *logs;
+	u16 status;
+
+	logs = kzalloc(sizeof(*logs), GFP_KERNEL);
+	if (!logs) {
+		status = NVME_SC_INTERNAL;
+		goto out;
+	}
+
+	logs->lids[NVME_LOG_SUPPORTED] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_ERROR] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_SMART] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_FW_SLOT] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_CHANGED_NS] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_CMD_EFFECTS] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_ANA] = cpu_to_le32(NVME_LIDS_LSUPP);
+
+	status = nvmet_copy_to_sgl(req, 0, logs, sizeof(*logs));
+	kfree(logs);
+out:
+	nvmet_req_complete(req, status);
+}
+
 static u16 nvmet_get_smart_log_nsid(struct nvmet_req *req,
 		struct nvme_smart_log *slog)
 {
@@ -323,6 +348,8 @@ static void nvmet_execute_get_log_page(struct nvmet_req *req)
 		return;
 
 	switch (req->cmd->get_log_page.lid) {
+	case NVME_LOG_SUPPORTED:
+		return nvmet_execute_get_supported_log_pages(req);
 	case NVME_LOG_ERROR:
 		return nvmet_execute_get_log_page_error(req);
 	case NVME_LOG_SMART:
include/linux/nvme.h +9 -0
index 0f263c7e63192..692d606edeed3 100644
@@ -1245,6 +1245,7 @@ enum {
 	NVME_FEAT_WRITE_PROTECT	= 0x84,
 	NVME_FEAT_VENDOR_START	= 0xC0,
 	NVME_FEAT_VENDOR_END	= 0xFF,
+	NVME_LOG_SUPPORTED	= 0x00,
 	NVME_LOG_ERROR		= 0x01,
 	NVME_LOG_SMART		= 0x02,
 	NVME_LOG_FW_SLOT	= 0x03,
@@ -1262,6 +1263,14 @@ enum {
 	NVME_FWACT_ACTV		= (2 << 3),
 };
 
+struct nvme_supported_log {
+	__le32	lids[256];
+};
+
+enum {
+	NVME_LIDS_LSUPP	= 1 << 0,
+};
+
 /* NVMe Namespace Write Protect State */
 enum {
 	NVME_NS_NO_WRITE_PROTECT = 0,
Signature · 1 line
2.43.5
Reviewed-by on v4 3/13
#7 lore raw

Re: [PATCHv4 03/13] nvmet: implement supported log pages

Quoted 72 lines · Keith Busch

From: Keith Busch <[email protected]>

This log is required for nvme 2.1.

---
  drivers/nvme/target/admin-cmd.c | 27 +++++++++++++++++++++++++++
  include/linux/nvme.h            |  9 +++++++++
  2 files changed, 36 insertions(+)
drivers/nvme/target/admin-cmd.c +27 -0
index a13242e791c0f..712b962fe77ac 100644
@@ -71,6 +71,31 @@ static void nvmet_execute_get_log_page_error(struct nvmet_req *req)
  	nvmet_req_complete(req, 0);
  }
 
+static void nvmet_execute_get_supported_log_pages(struct nvmet_req *req)
+{
+	struct nvme_supported_log *logs;
+	u16 status;
+
+	logs = kzalloc(sizeof(*logs), GFP_KERNEL);
+	if (!logs) {
+		status = NVME_SC_INTERNAL;
+		goto out;
+	}
+
+	logs->lids[NVME_LOG_SUPPORTED] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_ERROR] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_SMART] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_FW_SLOT] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_CHANGED_NS] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_CMD_EFFECTS] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_ANA] = cpu_to_le32(NVME_LIDS_LSUPP);
+
+	status = nvmet_copy_to_sgl(req, 0, logs, sizeof(*logs));
+	kfree(logs);
+out:
+	nvmet_req_complete(req, status);
+}
+
  static u16 nvmet_get_smart_log_nsid(struct nvmet_req *req,
  		struct nvme_smart_log *slog)
  {
@@ -323,6 +348,8 @@ static void nvmet_execute_get_log_page(struct nvmet_req *req)
  		return;
 
  	switch (req->cmd->get_log_page.lid) {
+	case NVME_LOG_SUPPORTED:
+		return nvmet_execute_get_supported_log_pages(req);
  	case NVME_LOG_ERROR:
  		return nvmet_execute_get_log_page_error(req);
  	case NVME_LOG_SMART:
include/linux/nvme.h +2 -0
index 0f263c7e63192..692d606edeed3 100644
@@ -1245,6 +1245,7 @@ enum {
  	NVME_FEAT_WRITE_PROTECT	= 0x84,
  	NVME_FEAT_VENDOR_START	= 0xC0,
  	NVME_FEAT_VENDOR_END	= 0xFF,
+	NVME_LOG_SUPPORTED	= 0x00,
  	NVME_LOG_ERROR		= 0x01,
  	NVME_LOG_SMART		= 0x02,
  	NVME_LOG_FW_SLOT	= 0x03,
@@ -1262,6 +1263,14 @@ enum {
  	NVME_FWACT_ACTV		= (2 << 3),
  };
 
+struct nvme_supported_log {

include/linux/nvme.h enum { · line 1276

Quoted 10 lines
include/linux/nvme.h +7 -0
+	__le32	lids[256];
+};
+
+enum {
+	NVME_LIDS_LSUPP	= 1 << 0,
+};
+
  /* NVMe Namespace Write Protect State */
  enum {
  	NVME_NS_NO_WRITE_PROTECT = 0,

Looks good. The patch description could be updated to say it's required for NVMe 2.0 and later devices.

#10 lore raw

PATCH v4 4/13 nvmet: implement supported features log

In reply to Keith Busch

From: Keith Busch <[email protected]>

This log is required for nvme 2.1.

---
 drivers/nvme/target/admin-cmd.c | 31 +++++++++++++++++++++++++++++++
 include/linux/nvme.h            | 11 +++++++++++
 2 files changed, 42 insertions(+)
drivers/nvme/target/admin-cmd.c +31 -0
index 712b962fe77ac..7a879e2cd8555 100644
@@ -89,6 +89,7 @@ static void nvmet_execute_get_supported_log_pages(struct nvmet_req *req)
 	logs->lids[NVME_LOG_CHANGED_NS] = cpu_to_le32(NVME_LIDS_LSUPP);
 	logs->lids[NVME_LOG_CMD_EFFECTS] = cpu_to_le32(NVME_LIDS_LSUPP);
 	logs->lids[NVME_LOG_ANA] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_FEATURES] = cpu_to_le32(NVME_LIDS_LSUPP);
 
 	status = nvmet_copy_to_sgl(req, 0, logs, sizeof(*logs));
 	kfree(logs);
@@ -342,6 +343,34 @@ static void nvmet_execute_get_log_page_ana(struct nvmet_req *req)
 	nvmet_req_complete(req, status);
 }
 
+static void nvmet_execute_get_log_page_features(struct nvmet_req *req)
+{
+	struct nvme_supported_features_log *features;
+	u16 status;
+
+	features = kzalloc(sizeof(*features), GFP_KERNEL);
+	if (!features) {
+		status = NVME_SC_INTERNAL;
+		goto out;
+	}
+
+	features->fis[NVME_FEAT_NUM_QUEUES] = cpu_to_le32(NVME_FIS_FSUPP |
+							  NVME_FIS_CSCPE);
+	features->fis[NVME_FEAT_KATO] = cpu_to_le32(NVME_FIS_FSUPP |
+						    NVME_FIS_CSCPE);
+	features->fis[NVME_FEAT_ASYNC_EVENT] = cpu_to_le32(NVME_FIS_FSUPP |
+							   NVME_FIS_CSCPE);
+	features->fis[NVME_FEAT_HOST_ID] = cpu_to_le32(NVME_FIS_FSUPP |
+						       NVME_FIS_CSCPE);
+	features->fis[NVME_FEAT_WRITE_PROTECT] = cpu_to_le32(NVME_FIS_FSUPP |
+							     NVME_FIS_NSCPE);
+
+	status = nvmet_copy_to_sgl(req, 0, features, sizeof(*features));
+	kfree(features);
+out:
+	nvmet_req_complete(req, status);
+}
+
 static void nvmet_execute_get_log_page(struct nvmet_req *req)
 {
 	if (!nvmet_check_transfer_len(req, nvmet_get_log_page_len(req->cmd)))
@@ -367,6 +396,8 @@ static void nvmet_execute_get_log_page(struct nvmet_req *req)
 		return nvmet_execute_get_log_cmd_effects_ns(req);
 	case NVME_LOG_ANA:
 		return nvmet_execute_get_log_page_ana(req);
+	case NVME_LOG_FEATURES:
+		return nvmet_execute_get_log_page_features(req);
 	}
 	pr_debug("unhandled lid %d on qid %d\n",
 	       req->cmd->get_log_page.lid, req->sq->qid);
include/linux/nvme.h +11 -0
index 692d606edeed3..4b896fa651e02 100644
@@ -1256,6 +1256,7 @@ enum {
 	NVME_LOG_TELEMETRY_CTRL = 0x08,
 	NVME_LOG_ENDURANCE_GROUP = 0x09,
 	NVME_LOG_ANA		= 0x0c,
+	NVME_LOG_FEATURES	= 0x12,
 	NVME_LOG_DISC		= 0x70,
 	NVME_LOG_RESERVATION	= 0x80,
 	NVME_FWACT_REPL		= (0 << 3),
@@ -1271,6 +1272,16 @@ enum {
 	NVME_LIDS_LSUPP	= 1 << 0,
 };
 
+struct nvme_supported_features_log {
+	__le32	fis[256];
+};
+
+enum {
+	NVME_FIS_FSUPP	= 1 << 0,
+	NVME_FIS_NSCPE	= 1 << 20,
+	NVME_FIS_CSCPE	= 1 << 21,
+};
+
 /* NVMe Namespace Write Protect State */
 enum {
 	NVME_NS_NO_WRITE_PROTECT = 0,
Signature · 1 line
2.43.5
Reviewed-by on v4 4/13
#11 lore raw

Re: [PATCHv4 04/13] nvmet: implement supported features log

Quoted 85 lines · Keith Busch

From: Keith Busch <[email protected]>

This log is required for nvme 2.1.

---
  drivers/nvme/target/admin-cmd.c | 31 +++++++++++++++++++++++++++++++
  include/linux/nvme.h            | 11 +++++++++++
  2 files changed, 42 insertions(+)
drivers/nvme/target/admin-cmd.c +31 -0
index 712b962fe77ac..7a879e2cd8555 100644
@@ -89,6 +89,7 @@ static void nvmet_execute_get_supported_log_pages(struct nvmet_req *req)
  	logs->lids[NVME_LOG_CHANGED_NS] = cpu_to_le32(NVME_LIDS_LSUPP);
  	logs->lids[NVME_LOG_CMD_EFFECTS] = cpu_to_le32(NVME_LIDS_LSUPP);
  	logs->lids[NVME_LOG_ANA] = cpu_to_le32(NVME_LIDS_LSUPP);
+	logs->lids[NVME_LOG_FEATURES] = cpu_to_le32(NVME_LIDS_LSUPP);
 
  	status = nvmet_copy_to_sgl(req, 0, logs, sizeof(*logs));
  	kfree(logs);
@@ -342,6 +343,34 @@ static void nvmet_execute_get_log_page_ana(struct nvmet_req *req)
  	nvmet_req_complete(req, status);
  }
 
+static void nvmet_execute_get_log_page_features(struct nvmet_req *req)
+{
+	struct nvme_supported_features_log *features;
+	u16 status;
+
+	features = kzalloc(sizeof(*features), GFP_KERNEL);
+	if (!features) {
+		status = NVME_SC_INTERNAL;
+		goto out;
+	}
+
+	features->fis[NVME_FEAT_NUM_QUEUES] = cpu_to_le32(NVME_FIS_FSUPP |
+							  NVME_FIS_CSCPE);
+	features->fis[NVME_FEAT_KATO] = cpu_to_le32(NVME_FIS_FSUPP |
+						    NVME_FIS_CSCPE);
+	features->fis[NVME_FEAT_ASYNC_EVENT] = cpu_to_le32(NVME_FIS_FSUPP |
+							   NVME_FIS_CSCPE);
+	features->fis[NVME_FEAT_HOST_ID] = cpu_to_le32(NVME_FIS_FSUPP |
+						       NVME_FIS_CSCPE);
+	features->fis[NVME_FEAT_WRITE_PROTECT] = cpu_to_le32(NVME_FIS_FSUPP |
+							     NVME_FIS_NSCPE);
+
+	status = nvmet_copy_to_sgl(req, 0, features, sizeof(*features));
+	kfree(features);
+out:
+	nvmet_req_complete(req, status);
+}
+
  static void nvmet_execute_get_log_page(struct nvmet_req *req)
  {
  	if (!nvmet_check_transfer_len(req, nvmet_get_log_page_len(req->cmd)))
@@ -367,6 +396,8 @@ static void nvmet_execute_get_log_page(struct nvmet_req *req)
  		return nvmet_execute_get_log_cmd_effects_ns(req);
  	case NVME_LOG_ANA:
  		return nvmet_execute_get_log_page_ana(req);
+	case NVME_LOG_FEATURES:
+		return nvmet_execute_get_log_page_features(req);
  	}
  	pr_debug("unhandled lid %d on qid %d\n",
  	       req->cmd->get_log_page.lid, req->sq->qid);
include/linux/nvme.h +4 -0
index 692d606edeed3..4b896fa651e02 100644
@@ -1256,6 +1256,7 @@ enum {
  	NVME_LOG_TELEMETRY_CTRL = 0x08,
  	NVME_LOG_ENDURANCE_GROUP = 0x09,
  	NVME_LOG_ANA		= 0x0c,
+	NVME_LOG_FEATURES	= 0x12,
  	NVME_LOG_DISC		= 0x70,
  	NVME_LOG_RESERVATION	= 0x80,
  	NVME_FWACT_REPL		= (0 << 3),
@@ -1271,6 +1272,16 @@ enum {
  	NVME_LIDS_LSUPP	= 1 << 0,
  };
 
+struct nvme_supported_features_log {
+	__le32	fis[256];
+};

include/linux/nvme.h enum { · line 1287

Quoted 10 lines
include/linux/nvme.h +7 -0
+
+enum {
+	NVME_FIS_FSUPP	= 1 << 0,
+	NVME_FIS_NSCPE	= 1 << 20,
+	NVME_FIS_CSCPE	= 1 << 21,
+};
+
  /* NVMe Namespace Write Protect State */
  enum {
  	NVME_NS_NO_WRITE_PROTECT = 0,

Looks good. Same comment wrt to 2.1 vs 2.0 as the previous patch.

maintainer Reviewed-by on v4 4/13
#12 lore raw

Re: [PATCHv4 04/13] nvmet: implement supported features log

In reply to Keith Busch

drivers/nvme/target/admin-cmd.c static void nvmet_execute_get_log_page_ana(struct nvmet_req *req) · line 360

Quoted 4 lines · Keith Busch
drivers/nvme/target/admin-cmd.c +4 -0
+	features->fis[NVME_FEAT_NUM_QUEUES] = cpu_to_le32(NVME_FIS_FSUPP |
+							  NVME_FIS_CSCPE);
+	features->fis[NVME_FEAT_KATO] = cpu_to_le32(NVME_FIS_FSUPP |
+						    NVME_FIS_CSCPE);

Totally nitpicky, but I find this version much easier to read:

	features->fis[NVME_FEAT_NUM_QUEUES] =
		cpu_to_le32(NVME_FIS_FSUPP | NVME_FIS_CSCPE);
	features->fis[NVME_FEAT_KATO] =
		cpu_to_le32(NVME_FIS_FSUPP | NVME_FIS_CSCPE);

especially if the number of flags keeps growing.

Either way this looks good:

#13 lore raw

PATCH v4 5/13 nvmet: implement crto property

In reply to Keith Busch

From: Keith Busch <[email protected]>

This property is required for nvme 2.1. The target only supports ready with media, so this is just the same value as CAP.TO.

---
 drivers/nvme/target/fabrics-cmd.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/nvme/target/fabrics-cmd.c +3 -0
index c4b2eddd5666a..bb80db6a7950b 100644
@@ -64,6 +64,9 @@ static void nvmet_execute_prop_get(struct nvmet_req *req)
 		case NVME_REG_CSTS:
 			val = ctrl->csts;
 			break;
+		case NVME_REG_CRTO:
+			val = NVME_CAP_TIMEOUT(ctrl->csts);
+			break;
 		default:
 			status = NVME_SC_INVALID_FIELD | NVME_STATUS_DNR;
 			break;
Signature · 1 line
2.43.5
Reviewed-by on v4 5/13
#14 lore raw

Re: [PATCHv4 05/13] nvmet: implement crto property

Quoted 16 lines · Keith Busch

From: Keith Busch <[email protected]>

This property is required for nvme 2.1. The target only supports ready with media, so this is just the same value as CAP.TO.

---
  drivers/nvme/target/fabrics-cmd.c | 3 +++
  1 file changed, 3 insertions(+)
drivers/nvme/target/fabrics-cmd.c +0 -0
index c4b2eddd5666a..bb80db6a7950b 100644

drivers/nvme/target/fabrics-cmd.c static void nvmet_execute_prop_get(struct nvmet_req *req) · line 72

Quoted 9 lines
drivers/nvme/target/fabrics-cmd.c +3 -0
@@ -64,6 +64,9 @@ static void nvmet_execute_prop_get(struct nvmet_req *req)
  		case NVME_REG_CSTS:
  			val = ctrl->csts;
  			break;
+		case NVME_REG_CRTO:
+			val = NVME_CAP_TIMEOUT(ctrl->csts);
+			break;
  		default:
  			status = NVME_SC_INVALID_FIELD | NVME_STATUS_DNR;
  			break;

Looks good.

#15 lore raw

PATCH v4 6/13 nvmet: declare 2.1 version compliance

In reply to Keith Busch

From: Keith Busch <[email protected]>

The target driver implements all the mandatory logs, identifications, features, and properties up to nvme sepcification 2.1.

---
 drivers/nvme/target/nvmet.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/nvme/target/nvmet.h +1 -1
index 190f55e6d7532..f902a630a5cf4 100644
@@ -21,7 +21,7 @@
 #include <linux/radix-tree.h>
 #include <linux/t10-pi.h>
 
-#define NVMET_DEFAULT_VS		NVME_VS(1, 3, 0)
+#define NVMET_DEFAULT_VS		NVME_VS(2, 1, 0)
 
 #define NVMET_ASYNC_EVENTS		4
 #define NVMET_ERROR_LOG_SLOTS		128
Signature · 1 line
2.43.5
Reviewed-by on v4 6/13
#16 lore raw

Re: [PATCHv4 06/13] nvmet: declare 2.1 version compliance

Quoted 17 lines · Keith Busch

From: Keith Busch <[email protected]>

The target driver implements all the mandatory logs, identifications, features, and properties up to nvme sepcification 2.1.

---
  drivers/nvme/target/nvmet.h | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
drivers/nvme/target/nvmet.h +0 -0
index 190f55e6d7532..f902a630a5cf4 100644

drivers/nvme/target/nvmet.h line 27

Quoted 8 lines
drivers/nvme/target/nvmet.h +1 -1
@@ -21,7 +21,7 @@
  #include <linux/radix-tree.h>
  #include <linux/t10-pi.h>
 
-#define NVMET_DEFAULT_VS		NVME_VS(1, 3, 0)
+#define NVMET_DEFAULT_VS		NVME_VS(2, 1, 0)
 
  #define NVMET_ASYNC_EVENTS		4
  #define NVMET_ERROR_LOG_SLOTS		128

Looks good.

#17 lore raw

PATCH v4 7/13 nvmet: implement endurance groups

In reply to Keith Busch

From: Keith Busch <[email protected]>

Most of the returned information is just stubbed data. The target must support these in order to report rotational media. Since this driver doesn't know any better, each namespace is its own endurance group with the engid value matching the nsid.

---
 drivers/nvme/host/core.c        |  1 +
 drivers/nvme/target/admin-cmd.c | 95 +++++++++++++++++++++++++++++++++
 include/linux/nvme.h            | 29 +++++++++-
 3 files changed, 123 insertions(+), 2 deletions(-)
drivers/nvme/host/core.c +1 -0
index e9aac07f4c26d..426d4b90ecd7e 100644
@@ -5001,6 +5001,7 @@ static inline void _nvme_check_size(void)
 	BUILD_BUG_ON(sizeof(struct nvme_id_ctrl_nvm) != NVME_IDENTIFY_DATA_SIZE);
 	BUILD_BUG_ON(sizeof(struct nvme_lba_range_type) != 64);
 	BUILD_BUG_ON(sizeof(struct nvme_smart_log) != 512);
+	BUILD_BUG_ON(sizeof(struct nvme_endurance_group_log) != 512);
 	BUILD_BUG_ON(sizeof(struct nvme_dbbuf) != 64);
 	BUILD_BUG_ON(sizeof(struct nvme_directive_cmd) != 64);
 	BUILD_BUG_ON(sizeof(struct nvme_feat_host_behavior) != 512);
drivers/nvme/target/admin-cmd.c +95 -0

137 lines not on this page · raw message

include/linux/nvme.h +27 -2
index 4b896fa651e02..9b5867f1aa591 100644
@@ -327,7 +327,8 @@ struct nvme_id_ctrl {
 	__le32			sanicap;
 	__le32			hmminds;
 	__le16			hmmaxd;
-	__u8			rsvd338[4];
+	__le16			nvmsetidmax;
+	__le16			endgidmax;
 	__u8			anatt;
 	__u8			anacap;
 	__le32			anagrpmax;
@@ -531,6 +532,7 @@ enum {
 	NVME_ID_CNS_SCNDRY_CTRL_LIST	= 0x15,
 	NVME_ID_CNS_NS_GRANULARITY	= 0x16,
 	NVME_ID_CNS_UUID_LIST		= 0x17,
+	NVME_ID_CNS_ENDGRP_LIST		= 0x19,
 };
 
 enum {
@@ -618,6 +620,28 @@ enum {
 	NVME_NIDT_CSI		= 0x04,
 };
 
+struct nvme_endurance_group_log {
+	__u8	egcw;
+	__u8	egfeat;
+	__u8	rsvd2;
+	__u8	avsp;
+	__u8	avspt;
+	__u8	pused;
+	__le16	did;
+	__u8	rsvd8[24];
+	__u8	ee[16];
+	__u8	dur[16];
+	__u8	duw[16];
+	__u8	muw[16];
+	__u8	hrc[16];
+	__u8	hwc[16];
+	__u8	mdie[16];
+	__u8	neile[16];
+	__u8	tegcap[16];
+	__u8	uegcap[16];
+	__u8	rsvd192[320];
+};
+
 struct nvme_smart_log {
 	__u8			critical_warning;
 	__u8			temperature[2];
@@ -1302,7 +1326,8 @@ struct nvme_identify {
 	__u8			cns;
 	__u8			rsvd3;
 	__le16			ctrlid;
-	__u8			rsvd11[3];
+	__le16			cnssid;
+	__u8			rsvd11;
 	__u8			csi;
 	__u32			rsvd12[4];
 };
Signature · 1 line
2.43.5
#19 lore raw

PATCH v4 8/13 nvmet: implement rotational media information log

In reply to Keith Busch

From: Keith Busch <[email protected]>

Most of the information is stubbed. Supporting this log is a requirement for supporting rotational media.

---
 drivers/nvme/host/core.c        |  1 +
 drivers/nvme/target/admin-cmd.c | 42 +++++++++++++++++++++++++++++++++
 include/linux/nvme.h            | 15 +++++++++++-
 3 files changed, 57 insertions(+), 1 deletion(-)
drivers/nvme/host/core.c +1 -0
index 426d4b90ecd7e..279b0f445904d 100644
@@ -5002,6 +5002,7 @@ static inline void _nvme_check_size(void)
 	BUILD_BUG_ON(sizeof(struct nvme_lba_range_type) != 64);
 	BUILD_BUG_ON(sizeof(struct nvme_smart_log) != 512);
 	BUILD_BUG_ON(sizeof(struct nvme_endurance_group_log) != 512);
+	BUILD_BUG_ON(sizeof(struct nvme_rotational_media_log) != 512);
 	BUILD_BUG_ON(sizeof(struct nvme_dbbuf) != 64);
 	BUILD_BUG_ON(sizeof(struct nvme_directive_cmd) != 64);
 	BUILD_BUG_ON(sizeof(struct nvme_feat_host_behavior) != 512);
drivers/nvme/target/admin-cmd.c +42 -0

60 lines not on this page · raw message

include/linux/nvme.h +14 -1
index 9b5867f1aa591..93a0abfab5b0e 100644
@@ -642,6 +642,18 @@ struct nvme_endurance_group_log {
 	__u8	rsvd192[320];
 };
 
+struct nvme_rotational_media_log {
+	__le16	endgid;
+	__le16	numa;
+	__le16	nrs;
+	__u8	rsvd6[2];
+	__le32	spinc;
+	__le32	fspinc;
+	__le32	ldc;
+	__le32	fldc;
+	__u8	rsvd24[488];
+};
+
 struct nvme_smart_log {
 	__u8			critical_warning;
 	__u8			temperature[2];
@@ -1281,6 +1293,7 @@ enum {
 	NVME_LOG_ENDURANCE_GROUP = 0x09,
 	NVME_LOG_ANA		= 0x0c,
 	NVME_LOG_FEATURES	= 0x12,
+	NVME_LOG_RMI		= 0x16,
 	NVME_LOG_DISC		= 0x70,
 	NVME_LOG_RESERVATION	= 0x80,
 	NVME_FWACT_REPL		= (0 << 3),
@@ -1435,7 +1448,7 @@ struct nvme_get_log_page_command {
 	__u8			lsp; /* upper 4 bits reserved */
 	__le16			numdl;
 	__le16			numdu;
-	__u16			rsvd11;
+	__le16			lsi;
 	union {
 		struct {
 			__le32 lpol;
Signature · 1 line
2.43.5
#20 lore raw

PATCH v4 9/13 nvmet: implement csi identify ns

In reply to Keith Busch

From: Keith Busch <[email protected]>

Implements reporting the I/O Command Set Independent Identify Namespace command.

---
 drivers/nvme/target/admin-cmd.c | 32 ++++++++++++++++++++++++++++++++
 include/linux/nvme.h            |  1 +
 2 files changed, 33 insertions(+)
drivers/nvme/target/admin-cmd.c +32 -0
index 6d3a1d2103bf3..516af7b419189 100644
@@ -893,6 +893,35 @@ static void nvme_execute_identify_ns_nvm(struct nvmet_req *req)
 	nvmet_req_complete(req, status);
 }
 
+static void nvmet_execute_id_cs_indep(struct nvmet_req *req)
+{
+	struct nvme_id_ns_cs_indep *id;
+	u16 status;
+
+	status = nvmet_req_find_ns(req);
+	if (status)
+		goto out;
+
+	id = kzalloc(sizeof(*id), GFP_KERNEL);
+	if (!id) {
+		status = NVME_SC_INTERNAL;
+		goto out;
+	}
+
+	id->nstat = NVME_NSTAT_NRDY;
+	id->anagrpid = req->ns->anagrpid;
+	id->nmic = NVME_NS_NMIC_SHARED;
+	if (req->ns->readonly)
+		id->nsattr |= NVME_NS_ATTR_RO;
+	if (req->ns->bdev && !bdev_nonrot(req->ns->bdev))
+		id->nsfeat |= NVME_NS_ROTATIONAL;
+
+	status = nvmet_copy_to_sgl(req, 0, id, sizeof(*id));
+	kfree(id);
+out:
+	nvmet_req_complete(req, status);
+}
+
 static void nvmet_execute_identify(struct nvmet_req *req)
 {
 	if (!nvmet_check_transfer_len(req, NVME_IDENTIFY_DATA_SIZE))
@@ -940,6 +969,9 @@ static void nvmet_execute_identify(struct nvmet_req *req)
 	case NVME_ID_CNS_NS_ACTIVE_LIST_CS:
 		nvmet_execute_identify_nslist(req, true);
 		return;
+	case NVME_ID_CNS_NS_CS_INDEP:
+		nvmet_execute_id_cs_indep(req);
+		return;
 	case NVME_ID_CNS_ENDGRP_LIST:
 		nvmet_execute_identify_endgrp_list(req);
 		return;
include/linux/nvme.h +1 -0
index 93a0abfab5b0e..22375c87591a2 100644
@@ -563,6 +563,7 @@ enum {
 	NVME_NS_FLBAS_LBA_SHIFT	= 1,
 	NVME_NS_FLBAS_META_EXT	= 0x10,
 	NVME_NS_NMIC_SHARED	= 1 << 0,
+	NVME_NS_ROTATIONAL	= 1 << 4,
 	NVME_LBAF_RP_BEST	= 0,
 	NVME_LBAF_RP_BETTER	= 1,
 	NVME_LBAF_RP_GOOD	= 2,
Signature · 1 line
2.43.5
#21 lore raw

Re: [PATCHv4 09/13] nvmet: implement csi identify ns

Quoted 26 lines · Keith Busch

From: Keith Busch <[email protected]>

Implements reporting the I/O Command Set Independent Identify Namespace command.

---
  drivers/nvme/target/admin-cmd.c | 32 ++++++++++++++++++++++++++++++++
  include/linux/nvme.h            |  1 +
  2 files changed, 33 insertions(+)
drivers/nvme/target/admin-cmd.c +7 -0
index 6d3a1d2103bf3..516af7b419189 100644
@@ -893,6 +893,35 @@ static void nvme_execute_identify_ns_nvm(struct nvmet_req *req)
  	nvmet_req_complete(req, status);
  }
 
+static void nvmet_execute_id_cs_indep(struct nvmet_req *req)
+{
+	struct nvme_id_ns_cs_indep *id;
+	u16 status;
+
+	status = nvmet_req_find_ns(req);
+	if (status)

drivers/nvme/target/admin-cmd.c static void nvme_execute_identify_ns_nvm(struct nvmet_req *req) · line 912

Quoted 10 lines
drivers/nvme/target/admin-cmd.c +10 -0
+		goto out;
+
+	id = kzalloc(sizeof(*id), GFP_KERNEL);
+	if (!id) {
+		status = NVME_SC_INTERNAL;
+		goto out;
+	}
+
+	id->nstat = NVME_NSTAT_NRDY;
+	id->anagrpid = req->ns->anagrpid;

Well, I think you overlooked my comments for v3,

should use cpu_to_le32(req->ns->anagrpid) here.

Best Regards,

Guixin Liu

Quoted 37 lines
drivers/nvme/target/admin-cmd.c +15 -0
+	id->nmic = NVME_NS_NMIC_SHARED;
+	if (req->ns->readonly)
+		id->nsattr |= NVME_NS_ATTR_RO;
+	if (req->ns->bdev && !bdev_nonrot(req->ns->bdev))
+		id->nsfeat |= NVME_NS_ROTATIONAL;
+
+	status = nvmet_copy_to_sgl(req, 0, id, sizeof(*id));
+	kfree(id);
+out:
+	nvmet_req_complete(req, status);
+}
+
  static void nvmet_execute_identify(struct nvmet_req *req)
  {
  	if (!nvmet_check_transfer_len(req, NVME_IDENTIFY_DATA_SIZE))
@@ -940,6 +969,9 @@ static void nvmet_execute_identify(struct nvmet_req *req)
  	case NVME_ID_CNS_NS_ACTIVE_LIST_CS:
  		nvmet_execute_identify_nslist(req, true);
  		return;
+	case NVME_ID_CNS_NS_CS_INDEP:
+		nvmet_execute_id_cs_indep(req);
+		return;
  	case NVME_ID_CNS_ENDGRP_LIST:
  		nvmet_execute_identify_endgrp_list(req);
  		return;
include/linux/nvme.h +1 -0
index 93a0abfab5b0e..22375c87591a2 100644
@@ -563,6 +563,7 @@ enum {
  	NVME_NS_FLBAS_LBA_SHIFT	= 1,
  	NVME_NS_FLBAS_META_EXT	= 0x10,
  	NVME_NS_NMIC_SHARED	= 1 << 0,
+	NVME_NS_ROTATIONAL	= 1 << 4,
  	NVME_LBAF_RP_BEST	= 0,
  	NVME_LBAF_RP_BETTER	= 1,
  	NVME_LBAF_RP_GOOD	= 2,
#23 lore raw

PATCH v4 10/13 nvme: use command set independent id ns if available

In reply to Keith Busch

From: Matias Bjørling <[email protected]>

The NVMe 2.0 specification adds an independent identify namespace data structure that contains generic attributes that apply to all namespace types. Some attributes carry over from the NVM command set identify namespace data structure, and others are new.

Currently, the data structure only considered when CRIMS is enabled or when the namespace type is key-value.

However, the independent namespace data structure is mandatory for devices that implement features from the 2.0+ specification. Therefore, we can check this data structure first. If unavailable, retrieve the generic attributes from the NVM command set identify namespace data structure.

---
 drivers/nvme/host/core.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/nvme/host/core.c +4 -3
index 279b0f445904d..162889b6382aa 100644
@@ -3988,7 +3988,7 @@ static void nvme_scan_ns(struct nvme_ctrl *ctrl, unsigned nsid)
 {
 	struct nvme_ns_info info = { .nsid = nsid };
 	struct nvme_ns *ns;
-	int ret;
+	int ret = 1;
 
 	if (nvme_identify_ns_descs(ctrl, &info))
 		return;
@@ -4005,9 +4005,10 @@ static void nvme_scan_ns(struct nvme_ctrl *ctrl, unsigned nsid)
 	 * set up a namespace.  If not fall back to the legacy version.
 	 */
 	if ((ctrl->cap & NVME_CAP_CRMS_CRIMS) ||
-	    (info.ids.csi != NVME_CSI_NVM && info.ids.csi != NVME_CSI_ZNS))
+	    (info.ids.csi != NVME_CSI_NVM && info.ids.csi != NVME_CSI_ZNS) ||
+	    ctrl->vs >= NVME_VS(2, 0, 0))
 		ret = nvme_ns_info_from_id_cs_indep(ctrl, &info);
-	else
+	if (ret > 0)
 		ret = nvme_ns_info_from_identify(ctrl, &info);
 
 	if (info.is_removed)
Signature · 1 line
2.43.5
#25 lore raw

PATCH v4 11/13 nvme: add rotational support

In reply to Keith Busch

From: Wang Yugui <[email protected]>

Rotational devices, such as hard-drives, can be detected using the rotational bit in the namespace independent identify namespace data structure. Make the bit visible to the block layer through the rotational queue setting.

---
 drivers/nvme/host/core.c | 6 ++++++
 drivers/nvme/host/nvme.h | 1 +
 2 files changed, 7 insertions(+)
drivers/nvme/host/core.c +6 -0
index 162889b6382aa..6f51dde7de6ce 100644
@@ -42,6 +42,7 @@ struct nvme_ns_info {
 	bool is_readonly;
 	bool is_ready;
 	bool is_removed;
+	bool is_rotational;
 };
 
 unsigned int admin_timeout = 60;
@@ -1615,6 +1616,7 @@ static int nvme_ns_info_from_id_cs_indep(struct nvme_ctrl *ctrl,
 		info->is_shared = id->nmic & NVME_NS_NMIC_SHARED;
 		info->is_readonly = id->nsattr & NVME_NS_ATTR_RO;
 		info->is_ready = id->nstat & NVME_NSTAT_NRDY;
+		info->is_rotational = id->nsfeat & NVME_NS_ROTATIONAL;
 	}
 	kfree(id);
 	return ret;
@@ -2162,6 +2164,9 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns,
 	else
 		lim.features &= ~(BLK_FEAT_WRITE_CACHE | BLK_FEAT_FUA);
 
+	if (info->is_rotational)
+		lim.features |= BLK_FEAT_ROTATIONAL;
+
 	/*
 	 * Register a metadata profile for PI, or the plain non-integrity NVMe
 	 * metadata masquerading as Type 0 if supported, otherwise reject block
@@ -3608,6 +3613,7 @@ static struct nvme_ns_head *nvme_alloc_ns_head(struct nvme_ctrl *ctrl,
 	head->ns_id = info->nsid;
 	head->ids = info->ids;
 	head->shared = info->is_shared;
+	head->rotational = info->is_rotational;
 	ratelimit_state_init(&head->rs_nuse, 5 * HZ, 1);
 	ratelimit_set_flags(&head->rs_nuse, RATELIMIT_MSG_ON_RELEASE);
 	kref_init(&head->ref);
drivers/nvme/host/nvme.h +1 -0
index 093cb423f536b..900719c4c70c1 100644
@@ -474,6 +474,7 @@ struct nvme_ns_head {
 	struct list_head	entry;
 	struct kref		ref;
 	bool			shared;
+	bool			rotational;
 	bool			passthru_err_log_enabled;
 	struct nvme_effects_log *effects;
 	u64			nuse;
Signature · 1 line
2.43.5
#27 lore raw

PATCH v4 12/13 nvme: check ns's volatile write cache not present

In reply to Keith Busch

From: Guixin Liu <[email protected]>

When the VWC of a namespace does not exist, the BLK_FEAT_WRITE_CACHE flag should not be set when registering the block device, regardless of whether the controller supports VWC.

---
 drivers/nvme/host/core.c | 4 +++-
 include/linux/nvme.h     | 1 +
 2 files changed, 4 insertions(+), 1 deletion(-)
drivers/nvme/host/core.c +3 -1
index 6f51dde7de6ce..e119ba0f8ab8b 100644
@@ -43,6 +43,7 @@ struct nvme_ns_info {
 	bool is_ready;
 	bool is_removed;
 	bool is_rotational;
+	bool no_vwc;
 };
 
 unsigned int admin_timeout = 60;
@@ -1617,6 +1618,7 @@ static int nvme_ns_info_from_id_cs_indep(struct nvme_ctrl *ctrl,
 		info->is_readonly = id->nsattr & NVME_NS_ATTR_RO;
 		info->is_ready = id->nstat & NVME_NSTAT_NRDY;
 		info->is_rotational = id->nsfeat & NVME_NS_ROTATIONAL;
+		info->no_vwc = id->nsfeat & NVME_NS_VWC_NOT_PRESENT;
 	}
 	kfree(id);
 	return ret;
@@ -2159,7 +2161,7 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns,
 	    ns->head->ids.csi == NVME_CSI_ZNS)
 		nvme_update_zone_info(ns, &lim, &zi);
 
-	if (ns->ctrl->vwc & NVME_CTRL_VWC_PRESENT)
+	if ((ns->ctrl->vwc & NVME_CTRL_VWC_PRESENT) && !info->no_vwc)
 		lim.features |= BLK_FEAT_WRITE_CACHE | BLK_FEAT_FUA;
 	else
 		lim.features &= ~(BLK_FEAT_WRITE_CACHE | BLK_FEAT_FUA);
include/linux/nvme.h +1 -0
index 22375c87591a2..5c6d065c2f178 100644
@@ -564,6 +564,7 @@ enum {
 	NVME_NS_FLBAS_META_EXT	= 0x10,
 	NVME_NS_NMIC_SHARED	= 1 << 0,
 	NVME_NS_ROTATIONAL	= 1 << 4,
+	NVME_NS_VWC_NOT_PRESENT = 1 << 5,
 	NVME_LBAF_RP_BEST	= 0,
 	NVME_LBAF_RP_BETTER	= 1,
 	NVME_LBAF_RP_GOOD	= 2,
Signature · 1 line
2.43.5
#28 lore raw

PATCH v4 13/13 nvmet: report ns's vwc not present

In reply to Keith Busch

From: Guixin Liu <[email protected]>

Currently, we report that controller has vwc even though the ns may not have vwc. Report ns's vwc not present when not buffered_io or backdev doesn't have vwc.

---
 drivers/nvme/target/admin-cmd.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/nvme/target/admin-cmd.c +3 -0
index 516af7b419189..f688d7fdf7d67 100644
@@ -915,6 +915,9 @@ static void nvmet_execute_id_cs_indep(struct nvmet_req *req)
 		id->nsattr |= NVME_NS_ATTR_RO;
 	if (req->ns->bdev && !bdev_nonrot(req->ns->bdev))
 		id->nsfeat |= NVME_NS_ROTATIONAL;
+	if ((req->ns->file && !req->ns->buffered_io) ||
+	    (req->ns->bdev && !bdev_write_cache(req->ns->bdev)))
+		id->nsfeat |= NVME_NS_VWC_NOT_PRESENT;
 
 	status = nvmet_copy_to_sgl(req, 0, id, sizeof(*id));
 	kfree(id);
Signature · 1 line
2.43.5
#29 lore raw

Re: [PATCHv4 13/13] nvmet: report ns's vwc not present

drivers/nvme/target/admin-cmd.c static void nvmet_execute_id_cs_indep(struct nvmet_req *req) · line 918

Quoted 9 lines · Keith Busch
drivers/nvme/target/admin-cmd.c +1 -0
index 516af7b419189..f688d7fdf7d67 100644
@@ -915,6 +915,9 @@ static void nvmet_execute_id_cs_indep(struct nvmet_req *req)
 		id->nsattr |= NVME_NS_ATTR_RO;
 	if (req->ns->bdev && !bdev_nonrot(req->ns->bdev))
 		id->nsfeat |= NVME_NS_ROTATIONAL;
+	if ((req->ns->file && !req->ns->buffered_io) ||

I don't think this check is correct. Direct I/O still requires cache flushes for durability.

maintainer
#30 lore raw

Re: [PATCHv4 13/13] nvmet: report ns's vwc not present

Quoted 13 lines · Christoph Hellwig

drivers/nvme/target/admin-cmd.c static void nvmet_execute_id_cs_indep(struct nvmet_req *req) · line 918

Quoted 9 lines · Keith Busch
drivers/nvme/target/admin-cmd.c +1 -0
index 516af7b419189..f688d7fdf7d67 100644
@@ -915,6 +915,9 @@ static void nvmet_execute_id_cs_indep(struct nvmet_req *req)
 		id->nsattr |= NVME_NS_ATTR_RO;
 	if (req->ns->bdev && !bdev_nonrot(req->ns->bdev))
 		id->nsfeat |= NVME_NS_ROTATIONAL;
+	if ((req->ns->file && !req->ns->buffered_io) ||

I don't think this check is correct. Direct I/O still requires cache flushes for durability.

Oh, right. Would it be sufficient to check the file's block device for write caching?

	if ((req->ns->file && !req->ns->buffered_io &&
		bdev_write_cache(file_inode(req->ns->file)->i_sb->s_bdev))
#31 lore raw

Re: [PATCHv4 13/13] nvmet: report ns's vwc not present

Quoted 13 lines · Keith Busch
Quoted 12 lines · Christoph Hellwig

drivers/nvme/target/admin-cmd.c static void nvmet_execute_id_cs_indep(struct nvmet_req *req) · line 918

Quoted 9 lines · Keith Busch
drivers/nvme/target/admin-cmd.c +1 -0
index 516af7b419189..f688d7fdf7d67 100644
@@ -915,6 +915,9 @@ static void nvmet_execute_id_cs_indep(struct nvmet_req *req)
  		id->nsattr |= NVME_NS_ATTR_RO;
  	if (req->ns->bdev && !bdev_nonrot(req->ns->bdev))
  		id->nsfeat |= NVME_NS_ROTATIONAL;
+	if ((req->ns->file && !req->ns->buffered_io) ||

I don't think this check is correct. Direct I/O still requires cache flushes for durability.

Quoted 5 lines

Oh, right. Would it be sufficient to check the file's block device for write caching?

	if ((req->ns->file && !req->ns->buffered_io &&
		bdev_write_cache(file_inode(req->ns->file)->i_sb->s_bdev))

Or just remove !req->ns->buffered_io check? All file backend return

support vwc.

Best Regards,

Guixin Liu

#32 lore raw

Re: [PATCHv4 13/13] nvmet: report ns's vwc not present

Quoted 19 lines · Guixin Liu
Quoted 17 lines · Keith Busch
Quoted 13 lines · Christoph Hellwig

drivers/nvme/target/admin-cmd.c static void nvmet_execute_id_cs_indep(struct · line 919

Quoted 10 lines · Keith Busch
drivers/nvme/target/admin-cmd.c +1 -0
index 516af7b419189..f688d7fdf7d67 100644
@@ -915,6 +915,9 @@ static void nvmet_execute_id_cs_indep(struct 
 nvmet_req *req)
           id->nsattr |= NVME_NS_ATTR_RO;
       if (req->ns->bdev && !bdev_nonrot(req->ns->bdev))
           id->nsfeat |= NVME_NS_ROTATIONAL;
+    if ((req->ns->file && !req->ns->buffered_io) ||

I don't think this check is correct.  Direct I/O still requires cache flushes for durability.

Oh, right. Would it be sufficient to check the file's block device for write caching?

Quoted 6 lines
Quoted 2 lines

    if ((req->ns->file && !req->ns->buffered_io && bdev_write_cache(file_inode(req->ns->file)->i_sb->s_bdev))

Or just remove !req->ns->buffered_io check? All file backend return

support vwc.

Sorry, remove "req->ns->file && !req->ns->buffered_io".

Quoted 5 lines

Best Regards,

Guixin Liu

maintainer
#33 lore raw

Re: [PATCHv4 13/13] nvmet: report ns's vwc not present

Quoted 6 lines · Guixin Liu
Quoted 4 lines · Guixin Liu

Or just remove !req->ns->buffered_io check? All file backend return

support vwc.

Sorry, remove "req->ns->file && !req->ns->buffered_io".

Do you want to send an update the with the updated check?

#34 lore raw

Re: [PATCHv4 13/13] nvmet: report ns's vwc not present

Quoted 7 lines · Keith Busch
Quoted 5 lines · Guixin Liu
Quoted 3 lines · Guixin Liu

Or just remove !req->ns->buffered_io check? All file backend return

support vwc.

Sorry, remove "req->ns->file && !req->ns->buffered_io".

Do you want to send an update the with the updated check?

Sure, I will send an update one.

Best Regards,

Guixin Liu

#35 lore raw

Re: [PATCHv4 13/13] nvmet: report ns's vwc not present

In reply to Guixin Liu
Quoted 6 lines · Guixin Liu
Quoted 2 lines
	if ((req->ns->file && !req->ns->buffered_io &&
		bdev_write_cache(file_inode(req->ns->file)->i_sb->s_bdev))

Or just remove !req->ns->buffered_io check? All file backend return

support vwc.

Yes, that is the right thing to do.

#36 lore raw

Re: [PATCHv4 13/13] nvmet: report ns's vwc not present

In reply to Keith Busch
Quoted 10 lines · Keith Busch
Quoted 4 lines

drivers/nvme/target/admin-cmd.c static void nvmet_execute_id_cs_indep(struct nvmet_req *req) · line 918

Quoted 1 line
drivers/nvme/target/admin-cmd.c +1 -0
+	if ((req->ns->file && !req->ns->buffered_io) ||

I don't think this check is correct. Direct I/O still requires cache flushes for durability.

Oh, right. Would it be sufficient to check the file's block device for write caching?

	if ((req->ns->file && !req->ns->buffered_io &&
		bdev_write_cache(file_inode(req->ns->file)->i_sb->s_bdev))

No, that's wrong for two reasons:

1) sb->s_bdev is only really relevant for mounting and st_dev reported in stat. It might not relate to where the file data is placed at all
2) the presence of cache in the underlying block device only has a minimal relation to the need to fdatasync a file you wrote to. Except for the corner case of overwritting a fully allocated (and not just preallocated) extent on an in-place write file system there always is metadata for a write that needs to be synchronized. You could work around that with an O_DYNC write that does that sync per I/O, but performance would be awful (and nvmet donsn't support that anyway)

maintainer
#37 lore raw
In reply to Keith Busch

Patches 1-12 applied after merging to nvme-6.13, folding in the reservation logs and features. Also included feedback from patch 4 (code style alignment suggestion) and patch 9 (cpu_to_le32(anagrpid)).

Patch 13 is on hold.