* [dm-devel] [PATCH]multipath-tools: fix "multipath -ll" bug for Native NVME Multipath devices
@ 2021-09-27 1:42 chengjike
0 siblings, 0 replies; 4+ messages in thread
From: chengjike @ 2021-09-27 1:42 UTC (permalink / raw)
To: dm-devel, christophe.varoqui; +Cc: sunao.sun, jiangtao62, chengjike.cheng
After "Native NVME Multipath" is configured,
the content displayed is incorrect when you run "multipath -ll" command.
Each NVME devices have the same path name. For example:
[root@localhost home]# multipath -ll
eui.710032e8fb22a86c24a52c1000000db8 [nvme]:nvme1n1 NVMe,Huawei-XSG1,1000001
size=10485760 features='n/a' hwhandler='ANA' wp=rw
|-+- policy='n/a' prio=50 status=optimized
| `- 1:4:1 nvme1c4n1 0:0 n/a optimized live
`-+- policy='n/a' prio=50 status=optimized
`- 1:9:1 nvme1c9n1 0:0 n/a optimized live
eui.710032e8fb22a86b24a52c7c00000db7 [nvme]:nvme1n2 NVMe,Huawei-XSG1,1000001
size=10485760 features='n/a' hwhandler='ANA' wp=rw
|-+- policy='n/a' prio=50 status=optimized
| `- 1:4:1 nvme1c4n1 0:0 n/a optimized live
`-+- policy='n/a' prio=50 status=optimized
`- 1:9:1 nvme1c9n1 0:0 n/a optimized live
[root@localhost home]#
The logical paths of "nvme1n1" and "nvme1n2" are both "nvme1c4n1" and "nvme1c9n1".
So when filtering logical paths, use "nvme_ns_head->instance" for matching.
Signed-off-by: chengjike <chengjike.cheng@huawei.com>
---
libmultipath/foreign/nvme.c | 26 ++++++++++++++++++++------
1 file changed, 20 insertions(+), 6 deletions(-)
diff --git a/libmultipath/foreign/nvme.c b/libmultipath/foreign/nvme.c
index d40c0869..38580880 100644
--- a/libmultipath/foreign/nvme.c
+++ b/libmultipath/foreign/nvme.c
@@ -530,14 +530,18 @@ static int _dirent_controller(const struct dirent *di)
/* Find the block device for a given nvme controller */
struct udev_device *get_ctrl_blkdev(const struct context *ctx,
- struct udev_device *ctrl)
+ struct udev_device *ctrl, const char *ctrl_name)
{
+ int ctrl_num, ns_num;
struct udev_list_entry *item;
struct udev_device *blkdev = NULL;
struct udev_enumerate *enm = udev_enumerate_new(ctx->udev);
const char *devtype;
- if (enm == NULL)
+ if (enm == NULL || ctrl_name == NULL)
+ return NULL;
+
+ if (sscanf(ctrl_name, "nvme%dn%d", &ctrl_num, &ns_num) != 2)
return NULL;
pthread_cleanup_push(_udev_enumerate_unref, enm);
@@ -555,6 +559,8 @@ struct udev_device *get_ctrl_blkdev(const struct context *ctx,
item != NULL;
item = udev_list_entry_get_next(item)) {
struct udev_device *tmp;
+ const char *name = NULL ;
+ int m, n, l;
tmp = udev_device_new_from_syspath(ctx->udev,
udev_list_entry_get_name(item));
@@ -562,11 +568,19 @@ struct udev_device *get_ctrl_blkdev(const struct context *ctx,
continue;
devtype = udev_device_get_devtype(tmp);
- if (devtype && !strcmp(devtype, "disk")) {
+ if (devtype == NULL || strcmp(devtype, "disk")) {
+ udev_device_unref(tmp);
+ continue;
+ }
+
+ name = udev_device_get_sysname(tmp);
+ if (name != NULL &&
+ sscanf(name, "nvme%dc%dn%d", &m, &n, &l) == 3 &&
+ l == ns_num) {
blkdev = tmp;
break;
- } else
- udev_device_unref(tmp);
+ }
+ udev_device_unref(tmp);
}
if (blkdev == NULL)
@@ -679,7 +693,7 @@ static void _find_controllers(struct context *ctx, struct nvme_map *map)
}
pthread_cleanup_push(_udev_device_unref, ctrl);
- udev = get_ctrl_blkdev(ctx, ctrl);
+ udev = get_ctrl_blkdev(ctx, ctrl, udev_device_get_sysname(map->udev));
/*
* We give up the reference to the nvme device here and get
* it back from the child below.
--
2.21.0.windows.1
--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [dm-devel] [PATCH]multipath-tools: fix "multipath -ll" bug for Native NVME Multipath devices
2021-10-08 12:24 chengjike
2021-10-20 16:03 ` Martin Wilck
@ 2022-06-17 14:33 ` Martin Wilck
1 sibling, 0 replies; 4+ messages in thread
From: Martin Wilck @ 2022-06-17 14:33 UTC (permalink / raw)
To: chengjike, dm-devel, christophe.varoqui, xose.vazquez
Cc: sunao.sun, jiangtao62
On Fri, 2021-10-08 at 20:24 +0800, chengjike wrote:
> After "Native NVME Multipath" is configured,
> the content displayed is incorrect when you run "multipath -ll"
> command.
> Each NVME devices have the same path name. For example:
>
> [root@localhost home]# multipath -ll
> eui.710032e8fb22a86c24a52c1000000db8 [nvme]:nvme1n1 NVMe,Huawei-
> XSG1,1000001
> size=10485760 features='n/a' hwhandler='ANA' wp=rw
> > -+- policy='n/a' prio=50 status=optimized
> > `- 1:4:1 nvme1c4n1 0:0 n/a optimized live
> `-+- policy='n/a' prio=50 status=optimized
> `- 1:9:1 nvme1c9n1 0:0 n/a optimized live
> eui.710032e8fb22a86b24a52c7c00000db7 [nvme]:nvme1n2 NVMe,Huawei-
> XSG1,1000001
> size=10485760 features='n/a' hwhandler='ANA' wp=rw
> > -+- policy='n/a' prio=50 status=optimized
> > `- 1:4:1 nvme1c4n1 0:0 n/a optimized live
> `-+- policy='n/a' prio=50 status=optimized
> `- 1:9:1 nvme1c9n1 0:0 n/a optimized live
> [root@localhost home]#
>
> The logical paths of "nvme1n1" and "nvme1n2" are both "nvme1c4n1" and
> "nvme1c9n1".
> So when multipath-tools aggregates disks, use "nvme_ns_head-
> >instance" for matching.
> such as ,Use "b" in "nvmeanb" string to match "z" in
> "nvmexcynz"(a,b,x,y,z can be any number),
> and if "b" and "z" are the same, they are related.
>
> Signed-off-by: chengjike <chengjike.cheng@huawei.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
Thanks a lot, and sorry for the late reply. This somehow got lost in my
inbox.
Regards,
Martin
--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [dm-devel] [PATCH]multipath-tools: fix "multipath -ll" bug for Native NVME Multipath devices
2021-10-08 12:24 chengjike
@ 2021-10-20 16:03 ` Martin Wilck
2022-06-17 14:33 ` Martin Wilck
1 sibling, 0 replies; 4+ messages in thread
From: Martin Wilck @ 2021-10-20 16:03 UTC (permalink / raw)
To: chengjike, dm-devel, christophe.varoqui, xose.vazquez
Cc: sunao.sun, jiangtao62
On Fri, 2021-10-08 at 20:24 +0800, chengjike wrote:
> After "Native NVME Multipath" is configured,
> the content displayed is incorrect when you run "multipath -ll"
> command.
> Each NVME devices have the same path name. For example:
>
> [root@localhost home]# multipath -ll
> eui.710032e8fb22a86c24a52c1000000db8 [nvme]:nvme1n1 NVMe,Huawei-
> XSG1,1000001
> size=10485760 features='n/a' hwhandler='ANA' wp=rw
> > -+- policy='n/a' prio=50 status=optimized
> > `- 1:4:1 nvme1c4n1 0:0 n/a optimized live
> `-+- policy='n/a' prio=50 status=optimized
> `- 1:9:1 nvme1c9n1 0:0 n/a optimized live
> eui.710032e8fb22a86b24a52c7c00000db7 [nvme]:nvme1n2 NVMe,Huawei-
> XSG1,1000001
> size=10485760 features='n/a' hwhandler='ANA' wp=rw
> > -+- policy='n/a' prio=50 status=optimized
> > `- 1:4:1 nvme1c4n1 0:0 n/a optimized live
> `-+- policy='n/a' prio=50 status=optimized
> `- 1:9:1 nvme1c9n1 0:0 n/a optimized live
> [root@localhost home]#
>
> The logical paths of "nvme1n1" and "nvme1n2" are both "nvme1c4n1" and
> "nvme1c9n1".
> So when multipath-tools aggregates disks, use "nvme_ns_head-
> >instance" for matching.
> such as ,Use "b" in "nvmeanb" string to match "z" in
> "nvmexcynz"(a,b,x,y,z can be any number),
> and if "b" and "z" are the same, they are related.
>
> Signed-off-by: chengjike <chengjike.cheng@huawei.com>
Reviewed-by: Martin Wilck <mwilck@suse.com>
Thanks!!
--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel
^ permalink raw reply [flat|nested] 4+ messages in thread
* [dm-devel] [PATCH]multipath-tools: fix "multipath -ll" bug for Native NVME Multipath devices
@ 2021-10-08 12:24 chengjike
2021-10-20 16:03 ` Martin Wilck
2022-06-17 14:33 ` Martin Wilck
0 siblings, 2 replies; 4+ messages in thread
From: chengjike @ 2021-10-08 12:24 UTC (permalink / raw)
To: dm-devel, christophe.varoqui, mwilck, xose.vazquez
Cc: sunao.sun, jiangtao62, chengjike.cheng
After "Native NVME Multipath" is configured,
the content displayed is incorrect when you run "multipath -ll" command.
Each NVME devices have the same path name. For example:
[root@localhost home]# multipath -ll
eui.710032e8fb22a86c24a52c1000000db8 [nvme]:nvme1n1 NVMe,Huawei-XSG1,1000001
size=10485760 features='n/a' hwhandler='ANA' wp=rw
|-+- policy='n/a' prio=50 status=optimized
| `- 1:4:1 nvme1c4n1 0:0 n/a optimized live
`-+- policy='n/a' prio=50 status=optimized
`- 1:9:1 nvme1c9n1 0:0 n/a optimized live
eui.710032e8fb22a86b24a52c7c00000db7 [nvme]:nvme1n2 NVMe,Huawei-XSG1,1000001
size=10485760 features='n/a' hwhandler='ANA' wp=rw
|-+- policy='n/a' prio=50 status=optimized
| `- 1:4:1 nvme1c4n1 0:0 n/a optimized live
`-+- policy='n/a' prio=50 status=optimized
`- 1:9:1 nvme1c9n1 0:0 n/a optimized live
[root@localhost home]#
The logical paths of "nvme1n1" and "nvme1n2" are both "nvme1c4n1" and "nvme1c9n1".
So when multipath-tools aggregates disks, use "nvme_ns_head->instance" for matching.
such as ,Use "b" in "nvmeanb" string to match "z" in "nvmexcynz"(a,b,x,y,z can be any number),
and if "b" and "z" are the same, they are related.
Signed-off-by: chengjike <chengjike.cheng@huawei.com>
---
libmultipath/foreign/nvme.c | 26 ++++++++++++++++++++------
1 file changed, 20 insertions(+), 6 deletions(-)
diff --git a/libmultipath/foreign/nvme.c b/libmultipath/foreign/nvme.c
index d40c0869..38580880 100644
--- a/libmultipath/foreign/nvme.c
+++ b/libmultipath/foreign/nvme.c
@@ -530,14 +530,18 @@ static int _dirent_controller(const struct dirent *di)
/* Find the block device for a given nvme controller */
struct udev_device *get_ctrl_blkdev(const struct context *ctx,
- struct udev_device *ctrl)
+ struct udev_device *ctrl, const char *ctrl_name)
{
+ int ctrl_num, ns_num;
struct udev_list_entry *item;
struct udev_device *blkdev = NULL;
struct udev_enumerate *enm = udev_enumerate_new(ctx->udev);
const char *devtype;
- if (enm == NULL)
+ if (enm == NULL || ctrl_name == NULL)
+ return NULL;
+
+ if (sscanf(ctrl_name, "nvme%dn%d", &ctrl_num, &ns_num) != 2)
return NULL;
pthread_cleanup_push(_udev_enumerate_unref, enm);
@@ -555,6 +559,8 @@ struct udev_device *get_ctrl_blkdev(const struct context *ctx,
item != NULL;
item = udev_list_entry_get_next(item)) {
struct udev_device *tmp;
+ const char *name = NULL ;
+ int m, n, l;
tmp = udev_device_new_from_syspath(ctx->udev,
udev_list_entry_get_name(item));
@@ -562,11 +568,19 @@ struct udev_device *get_ctrl_blkdev(const struct context *ctx,
continue;
devtype = udev_device_get_devtype(tmp);
- if (devtype && !strcmp(devtype, "disk")) {
+ if (devtype == NULL || strcmp(devtype, "disk")) {
+ udev_device_unref(tmp);
+ continue;
+ }
+
+ name = udev_device_get_sysname(tmp);
+ if (name != NULL &&
+ sscanf(name, "nvme%dc%dn%d", &m, &n, &l) == 3 &&
+ l == ns_num) {
blkdev = tmp;
break;
- } else
- udev_device_unref(tmp);
+ }
+ udev_device_unref(tmp);
}
if (blkdev == NULL)
@@ -679,7 +693,7 @@ static void _find_controllers(struct context *ctx, struct nvme_map *map)
}
pthread_cleanup_push(_udev_device_unref, ctrl);
- udev = get_ctrl_blkdev(ctx, ctrl);
+ udev = get_ctrl_blkdev(ctx, ctrl, udev_device_get_sysname(map->udev));
/*
* We give up the reference to the nvme device here and get
* it back from the child below.
--
2.21.0.windows.1
--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel
^ permalink raw reply related [flat|nested] 4+ messages in thread
end of thread, other threads:[~2022-06-17 14:34 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-09-27 1:42 [dm-devel] [PATCH]multipath-tools: fix "multipath -ll" bug for Native NVME Multipath devices chengjike
2021-10-08 12:24 chengjike
2021-10-20 16:03 ` Martin Wilck
2022-06-17 14:33 ` Martin Wilck
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).