Skip to contentSkip to conversation

ext4: Add missing brelse() in add_new_gdb_meta_bg()

PATCH Applied Lukas Czerner 2 participants 5 messages Mar 5, 2019 – Mar 15, 2019 ext4 file system

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/2 ext4: Add missing brelse() in add_new_gdb_meta_bg()

Currently in add_new_gdb_meta_bg() there is a missing brelse of gdb_bh in case ext4_journal_get_write_access() fails.
Additionally kvfree() is missing in the same error path. Fix it by moving the ext4_journal_get_write_access() before the ext4 sb update as Ted suggested and release n_group_desc and gdb_bh in case it fails.

Fixes 61a9c11e5e7a ("ext4: add missing brelse() add_new_gdb_meta_bg()'s error path")

---
v2: Mofe ext4_journal_get_write_access() before we update ext4 sb

 fs/ext4/resize.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)
fs/ext4/resize.c +9 -2
index 48421de803b7..5f89c19043a2 100644
@@ -932,11 +932,18 @@ static int add_new_gdb_meta_bg(struct super_block *sb,
 	memcpy(n_group_desc, o_group_desc,
 	       EXT4_SB(sb)->s_gdb_count * sizeof(struct buffer_head *));
 	n_group_desc[gdb_num] = gdb_bh;
+
+	BUFFER_TRACE(gdb_bh, "get_write_access");
+	err = ext4_journal_get_write_access(handle, gdb_bh);
+	if (err) {
+		kvfree(n_group_desc);
+		brelse(gdb_bh);
+		return err;
+	}
+
 	EXT4_SB(sb)->s_group_desc = n_group_desc;
 	EXT4_SB(sb)->s_gdb_count++;
 	kvfree(o_group_desc);
-	BUFFER_TRACE(gdb_bh, "get_write_access");
-	err = ext4_journal_get_write_access(handle, gdb_bh);
 	return err;
 }
 
Signature · 1 line
2.20.1
#1 lore raw

PATCH v2 2/2 ext4: Report real fs size after failed resize

Currently when the file system resize using ext4_resize_fs() fails it will report into log that "resized filesystem to <requested block count>". However this may not be true in the case of failure. Use the current block count as returned by ext4_blocks_count() to report the block count.
Additionally report to log that "error occurred during file system resize"

---
v2: this is actually the first version

 fs/ext4/resize.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
fs/ext4/resize.c +4 -1
index 5f89c19043a2..939e91a8c88b 100644
@@ -2079,6 +2079,9 @@ int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count)
 		free_flex_gd(flex_gd);
 	if (resize_inode != NULL)
 		iput(resize_inode);
-	ext4_msg(sb, KERN_INFO, "resized filesystem to %llu", n_blocks_count);
+	if (err)
+		ext4_warning(sb, "error occurred during file system resize");
+	ext4_msg(sb, KERN_INFO, "resized filesystem to %llu",
+		 ext4_blocks_count(es));
 	return err;
 }
Signature · 1 line
2.20.1
maintainer Applied
#2 lore raw

Re: [PATCH v2 2/2] ext4: Report real fs size after failed resize

Quoted 9 lines · Lukas Czerner

Currently when the file system resize using ext4_resize_fs() fails it will report into log that "resized filesystem to <requested block count>". However this may not be true in the case of failure. Use the current block count as returned by ext4_blocks_count() to report the block count.
Additionally report to log that "error occurred during file system resize"

Thanks, applied with one minor change. In call to ext4_warning(), I changed it to include error code so the warning message will be more useful:

fs/ext4/resize.c int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count) · line 2083

Quoted 2 lines
fs/ext4/resize.c +2 -0
+	if (err)
+		ext4_warning(sb, "error occurred during file system resize");

is now

+ if (err)
+ ext4_warning(sb, "error (%d) occurred during "
+ "file system resize", err);

- Ted

#3 lore raw

Re: [PATCH v2 2/2] ext4: Report real fs size after failed resize

Quoted 20 lines · Theodore Ts'o
Quoted 9 lines · Lukas Czerner

Currently when the file system resize using ext4_resize_fs() fails it will report into log that "resized filesystem to <requested block count>". However this may not be true in the case of failure. Use the current block count as returned by ext4_blocks_count() to report the block count.
Additionally report to log that "error occurred during file system resize"

Thanks, applied with one minor change. In call to ext4_warning(), I changed it to include error code so the warning message will be more useful:

fs/ext4/resize.c int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count) · line 2083

Quoted 2 lines
fs/ext4/resize.c +2 -0
+	if (err)
+		ext4_warning(sb, "error occurred during file system resize");

is now

Review

Quoted 3 lines
(quoted diff) +3 -0
+       if (err)
+               ext4_warning(sb, "error (%d) occurred during "
+                            "file system resize", err);

Yeah, that's better.

Thanks!
-Lukas

Quoted 2 lines

- Ted

maintainer Applied
#4 lore raw
In reply to Lukas Czerner
Quoted 8 lines · Lukas Czerner

Currently in add_new_gdb_meta_bg() there is a missing brelse of gdb_bh in case ext4_journal_get_write_access() fails.
Additionally kvfree() is missing in the same error path. Fix it by moving the ext4_journal_get_write_access() before the ext4 sb update as Ted suggested and release n_group_desc and gdb_bh in case it fails.

Fixes 61a9c11e5e7a ("ext4: add missing brelse() add_new_gdb_meta_bg()'s error path")

Thanks, applied.

- Ted