From: Artem Blagodarenko <artem.blagodarenko@gmail.com>
Date: Wed, 13 Apr 2022 20:01:58 +0000 (-0400)
Subject: LU-15754 lfsck: skip an inode if iget() returns -ENOMEM
X-Git-Tag: 2.15.51~211
X-Git-Url: https://git.whamcloud.com/?a=commitdiff_plain;h=9523e99046b9d03ab1e5b75622e1f80e02fcc8e9;p=fs%2Flustre-release.git

LU-15754 lfsck: skip an inode if iget() returns -ENOMEM

After the change
commit c2b6d621c4ffe9936adf7a55c8b1c769672c306f
Author: Al Viro <viro@zeniv.linux.org.uk>
Date: Thu Jun 28 15:53:17 2018 -0400 new primitive: discard_new_inode()

find_inode_fast() returns -ESTALE, but iget_locked() replaces
it to the NULL and finally ldiskfs_inode_attach_jinode()
returns -ENOMEM.

So this check in osd_iit_iget() doesn't work.

if (rc == -ENOENT || rc == -ESTALE)
    RETURN(SCRUB_NEXT_CONTINUE);

As a solution we can skip an inode if -ENOMEM returned

Hpe-bug-id: LUS-10833
Change-Id: Icb30610e46e2ab899a512761b63aea248c4f2ada
Signed-off-by: Artem Blagodarenko <artem.blagodarenko@hpe.com>
Reviewed-on: https://review.whamcloud.com/47079
Tested-by: jenkins <devops@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Andrew Perepechko <andrew.perepechko@hpe.com>
Reviewed-by: Alexander Zarochentsev <alexander.zarochentsev@hpe.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
---

diff --git a/lustre/osd-ldiskfs/osd_scrub.c b/lustre/osd-ldiskfs/osd_scrub.c
index 0605c31..abc3295 100644
--- a/lustre/osd-ldiskfs/osd_scrub.c
+++ b/lustre/osd-ldiskfs/osd_scrub.c
@@ -608,8 +608,14 @@ static int osd_iit_iget(struct osd_thread_info *info, struct osd_device *dev,
 	if (IS_ERR(inode)) {
 		rc = PTR_ERR(inode);
 		/* The inode may be removed after bitmap searching, or the
-		 * file is new created without inode initialized yet. */
-		if (rc == -ENOENT || rc == -ESTALE)
+		 * file is new created without inode initialized yet.
+		 * LU-15754: After "new primitive: discard_new_inode()" change
+		 * in the kernel find_inode_fast() returns -ESTALE, but
+		 * iget_locked replaces it to the NULL and finally
+		 * ldiskfs_inode_attach_jinode() returns -ENOMEM
+		 * Let's skip an inode if -ENOMEM returned.
+		 */
+		if (rc == -ENOENT || rc == -ESTALE || rc == -ENOMEM)
 			RETURN(SCRUB_NEXT_CONTINUE);
 
 		CDEBUG(D_LFSCK, "%s: fail to read inode, ino# = %u: "