Skip to contentSkip to conversation

nvme-cli: fix disconnect-all for pcie tp

PATCH New Minwoo Im 5 participants 9 messages May 16, 2019 – May 27, 2019 NVMe

lore
Share & tools
lore

Minwoo Im

author
#0 lore raw

cover nvme-cli: fix disconnect-all for pcie tp

The first patch fixes that it tries to delete_controller even it's a controller with PCIe transport type. If so, the rest of the controller with fabrics type might not be disconnected properly.

The second one is just a trivial clean-up patch.

Minwoo Im (2):
nvme-cli: don't try to disconnect for ctrl with pcie transport
nvme-cli: remove unnecessary initialize of local var

 fabrics.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
Signature · 1 line
2.21.0
author
#1 lore raw

PATCH 1/2 nvme-cli: don't try to disconnect for ctrl with pcie transport

If a host system has both pcie and fabrics controllers, the following nvme-cli disconnect-all command will might not be working because contrlller with pcie transport does not have delete_controller in sysfs.

root at target:~# nvme list
Node SN Model Namespace Usage Format FW Rev
---------------- -------------------- ---------------------------------------- --------- -------------------------- ---------------- --------

/dev/nvme0n1     foo                  QEMU NVMe Ctrl                           1         536.87  MB / 536.87  MB    512   B +  0 B   1.0
/dev/nvme1n1     b92326b9b2323bf0     Linux                                    1         134.22  MB / 134.22  MB    512   B +  0 B   5.1.0-rc
/dev/nvme2n1     65fa04eddd9bbac0     Linux                                    1         134.22  MB / 134.22  MB    512   B +  0 B   5.1.0-rc

root at target:~/nvme-cli.git# nvme disconnect-all
Failed to open /sys/class/nvme/nvme0/delete_controller: No such file or directory

This patch checks the transport type of the controller in iteration.

Signed-off-by Minwoo Im <minwoo.im.dev at gmail.com>
---
 fabrics.c | 2 ++
 1 file changed, 2 insertions(+)
fabrics.c +2 -0
index 511de06..733a24b 100644
@@ -1196,6 +1196,8 @@ int disconnect_all(const char *desc, int argc, char **argv)
 
 		for (j = 0; j < subsys->nctrls; j++) {
 			struct ctrl_list_item *ctrl = &subsys->ctrls[j];
+			if (!strcmp(ctrl->transport, "pcie"))
+				continue;
 
 			ret = disconnect_by_device(ctrl->name);
 			if (ret)
Signature · 1 line
2.21.0
#2 lore raw

PATCH 1/2 nvme-cli: don't try to disconnect for ctrl with pcie transport

Looks good.

Reviewed-by Chaitanya Kulkarni <chaitanya.kulkarni at wdc.com>
Quoted 3 lines · Minwoo Im

If a host system has both pcie and fabrics controllers, the following nvme-cli disconnect-all command will might not be working because contrlller with pcie transport does not have delete_controller in sysfs.

Can we please move following block of the output into the comment(not sure if that is the right name)

of the patch section ? i.e. after signoff... line. This can be done at the merge time also.

---

output

---

fabrics.c | 2 ++

Quoted 23 lines

root at target:~# nvme list
Node SN Model Namespace Usage Format FW Rev
---------------- -------------------- ---------------------------------------- --------- -------------------------- ---------------- --------

/dev/nvme0n1     foo                  QEMU NVMe Ctrl                           1         536.87  MB / 536.87  MB    512   B +  0 B   1.0
/dev/nvme1n1     b92326b9b2323bf0     Linux                                    1         134.22  MB / 134.22  MB    512   B +  0 B   5.1.0-rc
/dev/nvme2n1     65fa04eddd9bbac0     Linux                                    1         134.22  MB / 134.22  MB    512   B +  0 B   5.1.0-rc

root at target:~/nvme-cli.git# nvme disconnect-all
Failed to open /sys/class/nvme/nvme0/delete_controller: No such file or directory

This patch checks the transport type of the controller in iteration.

Signed-off-by Minwoo Im <minwoo.im.dev at gmail.com>
---
 fabrics.c | 2 ++
 1 file changed, 2 insertions(+)
fabrics.c +0 -0
index 511de06..733a24b 100644

fabrics.c int disconnect_all(const char *desc, int argc, char **argv) · line 1199

Quoted 4 lines
fabrics.c +1 -0
@@ -1196,6 +1196,8 @@ int disconnect_all(const char *desc, int argc, char **argv)
 
 		for (j = 0; j < subsys->nctrls; j++) {
 			struct ctrl_list_item *ctrl = &subsys->ctrls[j];
+			if (!strcmp(ctrl->transport, "pcie"))

I didn't looked into the case but it is safe to use strcasecmp()? if you are sure strcmp()

will do the job, please disregard this comment.

Quoted 4 lines
fabrics.c +1 -0
+				continue;
 
 			ret = disconnect_by_device(ctrl->name);
 			if (ret)
author
#3 lore raw

PATCH 1/2 nvme-cli: don't try to disconnect for ctrl with pcie transport

Quoted 9 lines · Chaitanya Kulkarni

Looks good.

Reviewed-by Chaitanya Kulkarni <chaitanya.kulkarni at wdc.com>
Quoted 3 lines · Minwoo Im

If a host system has both pcie and fabrics controllers, the following nvme-cli disconnect-all command will might not be working because contrlller with pcie transport does not have delete_controller in sysfs.

Quoted 5 lines

Can we please move following block of the output into the comment(not sure if that is the right name)

of the patch section ? i.e. after signoff... line. This can be done at the merge time also.

Chaitanya,

Sorry for lately response here. I don't get what you really mean here. Could you please explain why should we move the following output lines to after --- blocks ?

Quoted 33 lines

---

output

---

fabrics.c | 2 ++

Quoted 24 lines

root at target:~# nvme list
Node SN Model Namespace Usage Format FW Rev
---------------- -------------------- ---------------------------------------- --------- -------------------------- ---------------- --------

/dev/nvme0n1     foo                  QEMU NVMe Ctrl                           1         536.87  MB / 536.87  MB    512   B +  0 B   1.0
/dev/nvme1n1     b92326b9b2323bf0     Linux                                    1         134.22  MB / 134.22  MB    512   B +  0 B   5.1.0-rc
/dev/nvme2n1     65fa04eddd9bbac0     Linux                                    1         134.22  MB / 134.22  MB    512   B +  0 B   5.1.0-rc

root at target:~/nvme-cli.git# nvme disconnect-all
Failed to open /sys/class/nvme/nvme0/delete_controller: No such file or directory

This patch checks the transport type of the controller in iteration.

Signed-off-by Minwoo Im <minwoo.im.dev at gmail.com>
---
 fabrics.c | 2 ++
 1 file changed, 2 insertions(+)
fabrics.c +0 -0
index 511de06..733a24b 100644
@@ -1196,6 +1196,8 @@ int disconnect_all(const char *desc, int argc, char **argv)
 
Quoted 8 lines

fabrics.c int disconnect_all(const char *desc, int argc, char **argv) · line 1199

Quoted 3 lines
fabrics.c +1 -0
 		for (j = 0; j < subsys->nctrls; j++) {
 			struct ctrl_list_item *ctrl = &subsys->ctrls[j];
+			if (!strcmp(ctrl->transport, "pcie"))

I didn't looked into the case but it is safe to use strcasecmp()? if you are sure strcmp()

will do the job, please disregard this comment.

strcmp() is enough to go here :)

author
#4 lore raw

PATCH 2/2 nvme-cli: remove unnecessary initialize of local var

In reply to Minwoo Im

This patch does not have any functional changes here. Just clean-up the unnecessary initialize.

Signed-off-by Minwoo Im <minwoo.im.dev at gmail.com>
---
 fabrics.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fabrics.c +1 -1
index 733a24b..573a6ef 100644
@@ -1180,7 +1180,7 @@ int disconnect(const char *desc, int argc, char **argv)
 int disconnect_all(const char *desc, int argc, char **argv)
 {
 	struct subsys_list_item *slist;
-	int i, j, ret = 0, subcnt = 0;
+	int i, j, ret, subcnt = 0;
 	const struct argconfig_commandline_options command_line_options[] = {
 		{NULL},
 	};
Signature · 1 line
2.21.0
#5 lore raw

PATCH 2/2 nvme-cli: remove unnecessary initialize of local var

Looks good.

Reviewed-by: Chaitanya Kulkarni <chaitanya.kulkarni at wdc.com>

Quoted 21 lines · Minwoo Im

This patch does not have any functional changes here. Just clean-up the unnecessary initialize.

Signed-off-by Minwoo Im <minwoo.im.dev at gmail.com>
---
 fabrics.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fabrics.c +1 -1
index 733a24b..573a6ef 100644
@@ -1180,7 +1180,7 @@ int disconnect(const char *desc, int argc, char **argv)
 int disconnect_all(const char *desc, int argc, char **argv)
 {
 	struct subsys_list_item *slist;
-	int i, j, ret = 0, subcnt = 0;
+	int i, j, ret, subcnt = 0;
 	const struct argconfig_commandline_options command_line_options[] = {
 		{NULL},
 	};
maintainer
#7 lore raw

cover nvme-cli: fix disconnect-all for pcie tp

For both patches:

Reviewed-by Sagi Grimberg <sagi at grimberg.me>
maintainer
#8 lore raw

cover nvme-cli: fix disconnect-all for pcie tp

In reply to Minwoo Im
Quoted 5 lines · Minwoo Im

The first patch fixes that it tries to delete_controller even it's a controller with PCIe transport type. If so, the rest of the controller with fabrics type might not be disconnected properly.

The second one is just a trivial clean-up patch.

Thanks, added with Sagi and Chaitanya's reviews.