Re: [f2fs-dev] [PATCH v2] f2fs: relax sanity check if checkpoint is corrupted

[Chao Yu <chao@kernel.org>]

On 2023/4/13 23:52, Jaegeuk Kim wrote:
> On 04/13, Chao Yu wrote:
>> On 2023/4/13 0:17, Jaegeuk Kim wrote:
>>> 1. extent_cache
>>>    - let's drop the largest extent_cache
>>> 2. invalidate_block
>>>    - don't show the warnings
>>>
>>> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
>>> ---
>>>
>>>    Change log from v1:
>>>     - add one more case to skip the error message
>>>
>>>    fs/f2fs/checkpoint.c   | 10 ++++++++++
>>>    fs/f2fs/extent_cache.c | 22 +++++++++++++++-------
>>>    2 files changed, 25 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c
>>> index 448ecf5000b8..64b3860f50ee 100644
>>> --- a/fs/f2fs/checkpoint.c
>>> +++ b/fs/f2fs/checkpoint.c
>>> @@ -152,6 +152,11 @@ static bool __is_bitmap_valid(struct f2fs_sb_info *sbi, block_t blkaddr,
>>>    	se = get_seg_entry(sbi, segno);
>>>    	exist = f2fs_test_bit(offset, se->cur_valid_map);
>>> +
>>> +	/* skip data, if we already have an error in checkpoint. */
>>> +	if (unlikely(f2fs_cp_error(sbi)))
>>> +		return exist;
>>> +
>>>    	if (exist && type == DATA_GENERIC_ENHANCE_UPDATE) {
>>>    		f2fs_err(sbi, "Inconsistent error blkaddr:%u, sit bitmap:%d",
>>>    			 blkaddr, exist);
>>> @@ -202,6 +207,11 @@ bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi,
>>>    	case DATA_GENERIC_ENHANCE_UPDATE:
>>>    		if (unlikely(blkaddr >= MAX_BLKADDR(sbi) ||
>>>    				blkaddr < MAIN_BLKADDR(sbi))) {
>>> +
>>> +			/* Skip to emit an error message. */
>>> +			if (unlikely(f2fs_cp_error(sbi)))
>>> +				return false;
>>> +
>>>    			f2fs_warn(sbi, "access invalid blkaddr:%u",
>>>    				  blkaddr);
>>>    			set_sbi_flag(sbi, SBI_NEED_FSCK);
>>> diff --git a/fs/f2fs/extent_cache.c b/fs/f2fs/extent_cache.c
>>> index 9a8153895d20..bea6ab9d846a 100644
>>> --- a/fs/f2fs/extent_cache.c
>>> +++ b/fs/f2fs/extent_cache.c
>>> @@ -23,18 +23,26 @@ bool sanity_check_extent_cache(struct inode *inode)
>>>    {
>>>    	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
>>>    	struct f2fs_inode_info *fi = F2FS_I(inode);
>>> +	struct extent_tree *et = fi->extent_tree[EX_READ];
>>>    	struct extent_info *ei;
>>> -	if (!fi->extent_tree[EX_READ])
>>> +	if (!et)
>>> +		return true;
>>> +
>>> +	ei = &et->largest;
>>> +	if (!ei->len)
>>>    		return true;
>>> -	ei = &fi->extent_tree[EX_READ]->largest;
>>> +	/* Let's drop, if checkpoint got corrupted. */
>>> +	if (is_set_ckpt_flags(sbi, CP_ERROR_FLAG)) {
>>> +		ei->len = 0;
>>> +		et->largest_updated = true;
>>
>> Thsi field indicates whether it is needed to update largest extent into
>> inode page, since cp_error is unrecoverable, it seems we don't needed to
>> update it.
> 
> Intention is to avoid any assumption like that, but sync the same logic when
> updating this.

Sure, fine to me.

Reviewed-by: Chao Yu <chao@kernel.org>

Thanks,

> 
>>
>> Thanks,
>>
>>> +		return true;
>>> +	}
>>> -	if (ei->len &&
>>> -		(!f2fs_is_valid_blkaddr(sbi, ei->blk,
>>> -					DATA_GENERIC_ENHANCE) ||
>>> -		!f2fs_is_valid_blkaddr(sbi, ei->blk + ei->len - 1,
>>> -					DATA_GENERIC_ENHANCE))) {
>>> +	if (!f2fs_is_valid_blkaddr(sbi, ei->blk, DATA_GENERIC_ENHANCE) ||
>>> +	    !f2fs_is_valid_blkaddr(sbi, ei->blk + ei->len - 1,
>>> +					DATA_GENERIC_ENHANCE)) {
>>>    		set_sbi_flag(sbi, SBI_NEED_FSCK);
>>>    		f2fs_warn(sbi, "%s: inode (ino=%lx) extent info [%u, %u, %u] is incorrect, run fsck to fix",
>>>    			  __func__, inode->i_ino,