2024-01-24 14:55:57

by Chao Yu

Subject: [PATCH] f2fs: fix to avoid potential panic during recovery

During recovery, if FAULT_BLOCK is on, it is possible that
f2fs_reserve_new_block() will return -ENOSPC during recovery,
then it may trigger panic.

Also, if fault injection rate is 1 and only FAULT_BLOCK fault
type is on, it may encounter deadloop in loop of block reservation.

Let's change as below to fix these issues:
- remove bug_on() to avoid panic.
- limit the loop count of block reservation to avoid potential
deadloop.

Fixes: 956fa1ddc132 ("f2fs: fix to check return value of f2fs_reserve_new_block()")
Reported-by: Zhiguo Niu <[email protected]>
Signed-off-by: Chao Yu <[email protected]>
---
fs/f2fs/f2fs.h | 5 +++++
fs/f2fs/recovery.c | 33 ++++++++++++++++-----------------
2 files changed, 21 insertions(+), 17 deletions(-)

diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index a62c7e037456..5993b721e92c 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -76,6 +76,11 @@ struct f2fs_fault_info {

extern const char *f2fs_fault_name[FAULT_MAX];
#define IS_FAULT_SET(fi, type) ((fi)->inject_type & BIT(type))
+
+/* maximum retry count for injected failure */
+#define DEFAULT_FAILURE_RETRY_COUNT 8
+#else
+#define DEFAULT_FAILURE_RETRY_COUNT 1
#endif

/*
diff --git a/fs/f2fs/recovery.c b/fs/f2fs/recovery.c
index d0f24ccbd1ac..aad1d1a9b3d6 100644
--- a/fs/f2fs/recovery.c
+++ b/fs/f2fs/recovery.c
@@ -611,6 +611,19 @@ static int check_index_in_prev_nodes(struct f2fs_sb_info *sbi,
return 0;
}

+static int f2fs_reserve_new_block_retry(struct dnode_of_data *dn)
+{
+ int i, err = 0;
+
+ for (i = DEFAULT_FAILURE_RETRY_COUNT; i > 0; i--) {
+ err = f2fs_reserve_new_block(dn);
+ if (!err)
+ break;
+ }
+
+ return err;
+}
+
static int do_recover_data(struct f2fs_sb_info *sbi, struct inode *inode,
struct page *page)
{
@@ -712,14 +725,8 @@ static int do_recover_data(struct f2fs_sb_info *sbi, struct inode *inode,
*/
if (dest == NEW_ADDR) {
f2fs_truncate_data_blocks_range(&dn, 1);
- do {
- err = f2fs_reserve_new_block(&dn);
- if (err == -ENOSPC) {
- f2fs_bug_on(sbi, 1);
- break;
- }
- } while (err &&
- IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION));
+
+ err = f2fs_reserve_new_block_retry(&dn);
if (err)
goto err;
continue;
@@ -727,16 +734,8 @@ static int do_recover_data(struct f2fs_sb_info *sbi, struct inode *inode,

/* dest is valid block, try to recover from src to dest */
if (f2fs_is_valid_blkaddr(sbi, dest, META_POR)) {
-
if (src == NULL_ADDR) {
- do {
- err = f2fs_reserve_new_block(&dn);
- if (err == -ENOSPC) {
- f2fs_bug_on(sbi, 1);
- break;
- }
- } while (err &&
- IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION));
+ err = f2fs_reserve_new_block_retry(&dn);
if (err)
goto err;
}
--
2.40.1

2024-01-30 20:56:29

by patchwork-bot+f2fs

Subject: Re: [f2fs-dev] [PATCH] f2fs: fix to avoid potential panic during recovery

Hello:

This patch was applied to jaegeuk/f2fs.git (dev)
by Jaegeuk Kim <[email protected]>:

On Wed, 24 Jan 2024 22:49:15 +0800 you wrote:
> During recovery, if FAULT_BLOCK is on, it is possible that
> f2fs_reserve_new_block() will return -ENOSPC during recovery,
> then it may trigger panic.
>
> Also, if fault injection rate is 1 and only FAULT_BLOCK fault
> type is on, it may encounter deadloop in loop of block reservation.
>
> [...]

Here is the summary with links:
- [f2fs-dev] f2fs: fix to avoid potential panic during recovery
https://git.kernel.org/jaegeuk/f2fs/c/f31438c16879

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html