[lvc-project] [PATCH v2] gfs2: fix unreachable error handling after freeze/thaw rework

Alexey Velichayshiy a.velichayshiy at ispras.ru
Tue Nov 11 23:40:50 MSK 2025


After commit b77b4a4815a9 ("gfs2: Rework freeze / thaw logic"), the second
error handling block in the freeze process became unreachable due to
misplaced gfs2_do_thaw() call. The thaw operation was incorrectly placed
within the error handling block of gfs2_lock_fs_check_clean(), causing
EBUSY/EIO error conditions to be processed as if they originated from
thaw rather than from the initial freeze check.

Move gfs2_do_thaw() call outside of gfs2_lock_fs_check_clean() error
handling block to restore proper error code processing and ensure
error messages accurately reflect their source.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Signed-off-by: Alexey Velichayshiy <a.velichayshiy at ispras.ru>
---
 fs/gfs2/super.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c
index 03b407a925dc..320bea5215f5 100644
--- a/fs/gfs2/super.c
+++ b/fs/gfs2/super.c
@@ -747,11 +747,7 @@ static int gfs2_freeze_super(struct super_block *sb, enum freeze_holder who,
 			set_bit(SDF_FREEZE_INITIATOR, &sdp->sd_flags);
 			set_bit(SDF_FROZEN, &sdp->sd_flags);
 			break;
-		}
-
-		error = gfs2_do_thaw(sdp, who, freeze_owner);
-
-		if (error == -EBUSY)
+		} else if (error == -EBUSY)
 			fs_err(sdp, "waiting for recovery before freeze\n");
 		else if (error == -EIO) {
 			fs_err(sdp, "Fatal IO error: cannot freeze gfs2 due "
@@ -760,6 +756,11 @@ static int gfs2_freeze_super(struct super_block *sb, enum freeze_holder who,
 		} else {
 			fs_err(sdp, "error freezing FS: %d\n", error);
 		}
+
+		error = gfs2_do_thaw(sdp, who, freeze_owner);
+		if (error)
+			goto out;
+
 		fs_err(sdp, "retrying...\n");
 		msleep(1000);
 	}
-- 
2.43.0




More information about the lvc-project mailing list