[PATCH 21/21] libmultipath: update_pathvec_from_dm: don't force DI_WWID

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



From: Martin Wilck <mwilck@xxxxxxxx>

For existing paths, we don't need to force setting DI_WWID. It's
pointless because it would matter only in multipathd, and all
callers of update_multipath_table() / update_pathvec_from_dm()
in multipathd call it with flags=0, so this code won't be executed.
And this makes sense, because the path would either have been
cleanly initialized via path_discover() or an uevent, or pathinfo()
would have been called in update_pathvec_from_dm() in an earlier
invocation.

Also remove the misleading comment, and make coverity happy by
checking the return value of pathinfo().

---

Note: I haven't fully made up my mind whether we should act
on a pathinfo() failure here. For now, let the log message suffice.
Because the pathinfo flags are usually clear and pp->udev exists,
the only possible failure would be filtering by foreign libraries
or by devnode, which seems highly unlikely if it hasn't happened
beforehand.

Signed-off-by: Martin Wilck <mwilck@xxxxxxxx>
---
 libmultipath/structs_vec.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/libmultipath/structs_vec.c b/libmultipath/structs_vec.c
index df5709a..bfec840 100644
--- a/libmultipath/structs_vec.c
+++ b/libmultipath/structs_vec.c
@@ -149,16 +149,15 @@ bool update_pathvec_from_dm(vector pathvec, struct multipath *mpp,
 			 * uninitialized struct path to pgp->paths, with only
 			 * pp->dev_t filled in. Thus if pp->udev is set here,
 			 * we know that the path is in pathvec already.
-			 * However, it's possible that the path in pathvec is
-			 * different from the one the kernel still had in its
-			 * map.
 			 */
 			if (pp->udev) {
 				if (pathinfo_flags & ~DI_NOIO) {
 					conf = get_multipath_config();
 					pthread_cleanup_push(put_multipath_config,
 							     conf);
-					pathinfo(pp, conf, pathinfo_flags|DI_WWID);
+					if (pathinfo(pp, conf, pathinfo_flags) != PATHINFO_OK)
+						condlog(2, "%s: pathinfo failed for existing path %s (flags=0x%x)",
+							__func__, pp->dev, pathinfo_flags);
 					pthread_cleanup_pop(1);
 				}
 			} else {
-- 
2.33.1


--
dm-devel mailing list
dm-devel@xxxxxxxxxx
https://listman.redhat.com/mailman/listinfo/dm-devel




[Index of Archives]     [DM Crypt]     [Fedora Desktop]     [ATA RAID]     [Fedora Marketing]     [Fedora Packaging]     [Fedora SELinux]     [Yosemite Discussion]     [KDE Users]     [Fedora Docs]

  Powered by Linux