linux-f2fs-devel.lists.sourceforge.net archive mirror
 help / color / mirror / Atom feed
From: Chao Yu <yuchao0@huawei.com>
To: Sahitya Tummala <stummala@codeaurora.org>
Cc: Jaegeuk Kim <jaegeuk@kernel.org>,
	linux-kernel@vger.kernel.org,
	linux-f2fs-devel@lists.sourceforge.net
Subject: Re: [f2fs-dev] [PATCH v4] f2fs: Fix indefinite loop in f2fs_gc()
Date: Tue, 17 Sep 2019 14:35:52 +0800	[thread overview]
Message-ID: <447ade2b-0199-6b02-c1cf-8835851bbd00@huawei.com> (raw)
In-Reply-To: <20190917053316.GB12730@codeaurora.org>

Hi Sahitya,

On 2019/9/17 13:33, Sahitya Tummala wrote:
> Hi Chao,
> 
> On Fri, Sep 06, 2019 at 07:00:32PM +0800, Chao Yu wrote:
>> Hi Sahitya,
>>
>> Luckily, I can reproduce this issue with generic/269, and have sent another
>> patch for the issue, could you please check that one?
>>
> 
> Thanks for the fix. The issue could not get reproduced yet, so could not make
> much progress on the customer case.

Alright, let me know if you get any clue or make progress on this issue.

Thanks,

> 
> thanks,
> 
>> Thanks,
>>
>> On 2019/8/7 22:06, Chao Yu wrote:
>>> On 2019-8-7 21:40, Sahitya Tummala wrote:
>>>> Policy - Foreground GC, LFS and greedy GC mode.
>>>>
>>>> Under this policy, f2fs_gc() loops forever to GC as it doesn't have
>>>> enough free segements to proceed and thus it keeps calling gc_more
>>>> for the same victim segment.  This can happen if the selected victim
>>>> segment could not be GC'd due to failed blkaddr validity check i.e.
>>>> is_alive() returns false for the blocks set in current validity map.
>>>>
>>>> Fix this by keeping track of such invalid segments and skip those
>>>> segments for selection in get_victim_by_default() to avoid endless
>>>> GC loop under such error scenarios. Currently, add this logic under
>>>> CONFIG_F2FS_CHECK_FS to be able to root cause the issue in debug
>>>> version.
>>>>
>>>> Signed-off-by: Sahitya Tummala <stummala@codeaurora.org>
>>>> ---
>>>> v4: Cover all logic with CONFIG_F2FS_CHECK_FS
>>>>
>>>>  fs/f2fs/gc.c      | 31 +++++++++++++++++++++++++++++--
>>>>  fs/f2fs/segment.c | 14 +++++++++++++-
>>>>  fs/f2fs/segment.h |  3 +++
>>>>  3 files changed, 45 insertions(+), 3 deletions(-)
>>>>
>>>> diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
>>>> index 8974672..cbcacbd 100644
>>>> --- a/fs/f2fs/gc.c
>>>> +++ b/fs/f2fs/gc.c
>>>> @@ -382,6 +382,16 @@ static int get_victim_by_default(struct f2fs_sb_info *sbi,
>>>>  			nsearched++;
>>>>  		}
>>>>  
>>>> +#ifdef CONFIG_F2FS_CHECK_FS
>>>> +		/*
>>>> +		 * skip selecting the invalid segno (that is failed due to block
>>>> +		 * validity check failure during GC) to avoid endless GC loop in
>>>> +		 * such cases.
>>>> +		 */
>>>> +		if (test_bit(segno, sm->invalid_segmap))
>>>> +			goto next;
>>>> +#endif
>>>> +
>>>>  		secno = GET_SEC_FROM_SEG(sbi, segno);
>>>>  
>>>>  		if (sec_usage_check(sbi, secno))
>>>> @@ -602,8 +612,15 @@ static bool is_alive(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
>>>>  {
>>>>  	struct page *node_page;
>>>>  	nid_t nid;
>>>> -	unsigned int ofs_in_node;
>>>> +	unsigned int ofs_in_node, segno;
>>>>  	block_t source_blkaddr;
>>>> +	unsigned long offset;
>>>> +#ifdef CONFIG_F2FS_CHECK_FS
>>>> +	struct sit_info *sit_i = SIT_I(sbi);
>>>> +#endif
>>>> +
>>>> +	segno = GET_SEGNO(sbi, blkaddr);
>>>> +	offset = GET_BLKOFF_FROM_SEG0(sbi, blkaddr);
>>>>  
>>>>  	nid = le32_to_cpu(sum->nid);
>>>>  	ofs_in_node = le16_to_cpu(sum->ofs_in_node);
>>>> @@ -627,8 +644,18 @@ static bool is_alive(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
>>>>  	source_blkaddr = datablock_addr(NULL, node_page, ofs_in_node);
>>>>  	f2fs_put_page(node_page, 1);
>>>>  
>>>> -	if (source_blkaddr != blkaddr)
>>>> +	if (source_blkaddr != blkaddr) {
>>>> +#ifdef CONFIG_F2FS_CHECK_FS
>>>
>>> 		unsigned int segno = GET_SEGNO(sbi, blkaddr);
>>> 		unsigned int offset = GET_BLKOFF_FROM_SEG0(sbi, blkaddr);
>>>
>>> Should be local, otherwise it looks good to me, I think Jaegeuk can help to fix
>>> this while merging.
>>>
>>> Reviewed-by: Chao Yu <yuchao0@huawei.com>
>>>
>>> Thanks,
>>>
>>>> +		if (unlikely(check_valid_map(sbi, segno, offset))) {
>>>> +			if (!test_and_set_bit(segno, sit_i->invalid_segmap)) {
>>>> +				f2fs_err(sbi, "mismatched blkaddr %u (source_blkaddr %u) in seg %u\n",
>>>> +						blkaddr, source_blkaddr, segno);
>>>> +				f2fs_bug_on(sbi, 1);
>>>> +			}
>>>> +		}
>>>> +#endif
>>>>  		return false;
>>>> +	}
>>>>  	return true;
>>>>  }
>>>>  
>>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
>>>> index a661ac3..ee795b1 100644
>>>> --- a/fs/f2fs/segment.c
>>>> +++ b/fs/f2fs/segment.c
>>>> @@ -806,6 +806,9 @@ static void __remove_dirty_segment(struct f2fs_sb_info *sbi, unsigned int segno,
>>>>  		enum dirty_type dirty_type)
>>>>  {
>>>>  	struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
>>>> +#ifdef CONFIG_F2FS_CHECK_FS
>>>> +	struct sit_info *sit_i = SIT_I(sbi);
>>>> +#endif
>>>>  
>>>>  	if (test_and_clear_bit(segno, dirty_i->dirty_segmap[dirty_type]))
>>>>  		dirty_i->nr_dirty[dirty_type]--;
>>>> @@ -817,9 +820,13 @@ static void __remove_dirty_segment(struct f2fs_sb_info *sbi, unsigned int segno,
>>>>  		if (test_and_clear_bit(segno, dirty_i->dirty_segmap[t]))
>>>>  			dirty_i->nr_dirty[t]--;
>>>>  
>>>> -		if (get_valid_blocks(sbi, segno, true) == 0)
>>>> +		if (get_valid_blocks(sbi, segno, true) == 0) {
>>>>  			clear_bit(GET_SEC_FROM_SEG(sbi, segno),
>>>>  						dirty_i->victim_secmap);
>>>> +#ifdef CONFIG_F2FS_CHECK_FS
>>>> +			clear_bit(segno, sit_i->invalid_segmap);
>>>> +#endif
>>>> +		}
>>>>  	}
>>>>  }
>>>>  
>>>> @@ -4015,6 +4022,10 @@ static int build_sit_info(struct f2fs_sb_info *sbi)
>>>>  	sit_i->sit_bitmap_mir = kmemdup(src_bitmap, bitmap_size, GFP_KERNEL);
>>>>  	if (!sit_i->sit_bitmap_mir)
>>>>  		return -ENOMEM;
>>>> +
>>>> +	sit_i->invalid_segmap = f2fs_kvzalloc(sbi, bitmap_size, GFP_KERNEL);
>>>> +	if (!sit_i->invalid_segmap)
>>>> +		return -ENOMEM;
>>>>  #endif
>>>>  
>>>>  	/* init SIT information */
>>>> @@ -4517,6 +4528,7 @@ static void destroy_sit_info(struct f2fs_sb_info *sbi)
>>>>  	kvfree(sit_i->sit_bitmap);
>>>>  #ifdef CONFIG_F2FS_CHECK_FS
>>>>  	kvfree(sit_i->sit_bitmap_mir);
>>>> +	kvfree(sit_i->invalid_segmap);
>>>>  #endif
>>>>  	kvfree(sit_i);
>>>>  }
>>>> diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h
>>>> index b746028..9370d53 100644
>>>> --- a/fs/f2fs/segment.h
>>>> +++ b/fs/f2fs/segment.h
>>>> @@ -229,6 +229,9 @@ struct sit_info {
>>>>  	char *sit_bitmap;		/* SIT bitmap pointer */
>>>>  #ifdef CONFIG_F2FS_CHECK_FS
>>>>  	char *sit_bitmap_mir;		/* SIT bitmap mirror */
>>>> +
>>>> +	/* bitmap of segments to be ignored by GC in case of errors */
>>>> +	unsigned long *invalid_segmap;
>>>>  #endif
>>>>  	unsigned int bitmap_size;	/* SIT bitmap size */
>>>>  
>>>>
>>> .
>>>
> 


_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

      reply	other threads:[~2019-09-17  6:36 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-07 13:40 [f2fs-dev] [PATCH v4] f2fs: Fix indefinite loop in f2fs_gc() Sahitya Tummala
2019-08-07 14:06 ` Chao Yu
2019-08-09 15:36   ` Jaegeuk Kim
2019-09-06 11:00   ` Chao Yu
2019-09-17  5:33     ` Sahitya Tummala
2019-09-17  6:35       ` Chao Yu [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=447ade2b-0199-6b02-c1cf-8835851bbd00@huawei.com \
    --to=yuchao0@huawei.com \
    --cc=jaegeuk@kernel.org \
    --cc=linux-f2fs-devel@lists.sourceforge.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=stummala@codeaurora.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).