Re: [PATCH 2/3] f2fs: keep meta pages in cp_error state

[Jaegeuk Kim]

On 07/16, Chao Yu wrote:
> On 2018/7/15 9:11, Jaegeuk Kim wrote:
> > It turns out losing meta pages in shutdown period makes f2fs very unstable
> > so that I could see many unexpected error conditions.
> > 
> > Let's keep meta pages for fault injection and sudden power-off tests.
> > 
> > Signed-off-by: Jaegeuk Kim <jaegeuk@xxxxxxxxxx>
> > ---
> >  fs/f2fs/checkpoint.c | 28 +++++++++++++---------------
> >  1 file changed, 13 insertions(+), 15 deletions(-)
> > 
> > diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c
> > index 9f1c96caebda..37ec0f16448e 100644
> > --- a/fs/f2fs/checkpoint.c
> > +++ b/fs/f2fs/checkpoint.c
> > @@ -242,11 +242,8 @@ static int __f2fs_write_meta_page(struct page *page,
> >  
> >  	trace_f2fs_writepage(page, META);
> >  
> > -	if (unlikely(f2fs_cp_error(sbi))) {
> > -		dec_page_count(sbi, F2FS_DIRTY_META);
> > -		unlock_page(page);
> > -		return 0;
> > -	}
> > +	if (unlikely(f2fs_cp_error(sbi)))
> > +		goto redirty_out;
> >  	if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
> >  		goto redirty_out;
> >  	if (wbc->for_reclaim && page->index < GET_SUM_BLOCK(sbi, 0))
> > @@ -1129,6 +1126,9 @@ static void wait_on_all_pages_writeback(struct f2fs_sb_info *sbi)
> >  		if (!get_pages(sbi, F2FS_WB_CP_DATA))
> >  			break;
> >  
> > +		if (unlikely(f2fs_cp_error(sbi)))
> > +			break;
> > +
> >  		io_schedule_timeout(5*HZ);
> >  	}
> >  	finish_wait(&sbi->cp_wait, &wait);
> > @@ -1202,8 +1202,12 @@ static void commit_checkpoint(struct f2fs_sb_info *sbi,
> >  
> >  	/* writeout cp pack 2 page */
> >  	err = __f2fs_write_meta_page(page, &wbc, FS_CP_META_IO);
> > -	f2fs_bug_on(sbi, err);
> > +	if (unlikely(err && f2fs_cp_error(sbi))) {
> > +		f2fs_put_page(page, 1);
> > +		return;
> > +	}
> >  
> > +	f2fs_bug_on(sbi, err);
> >  	f2fs_put_page(page, 0);
> >  
> >  	/* submit checkpoint (with barrier if NOBARRIER is not set) */
> > @@ -1229,7 +1233,7 @@ static int do_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc)
> >  	while (get_pages(sbi, F2FS_DIRTY_META)) {
> >  		f2fs_sync_meta_pages(sbi, META, LONG_MAX, FS_CP_META_IO);
> >  		if (unlikely(f2fs_cp_error(sbi)))
> > -			return -EIO;
> > +			break;
> >  	}
> >  
> >  	/*
> > @@ -1309,7 +1313,7 @@ static int do_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc)
> >  			f2fs_sync_meta_pages(sbi, META, LONG_MAX,
> >  							FS_CP_META_IO);
> >  			if (unlikely(f2fs_cp_error(sbi)))
> > -				return -EIO;
> > +				break;
> >  		}
> >  	}
> >  
> > @@ -1350,9 +1354,6 @@ static int do_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc)
> >  	/* wait for previous submitted meta pages writeback */
> >  	wait_on_all_pages_writeback(sbi);
> >  
> > -	if (unlikely(f2fs_cp_error(sbi)))
> > -		return -EIO;
> > -
> >  	/* flush all device cache */
> >  	err = f2fs_flush_device_cache(sbi);
> >  	if (err)
> > @@ -1364,9 +1365,6 @@ static int do_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc)
> >  
> >  	f2fs_release_ino_entry(sbi, false);
> >  
> > -	if (unlikely(f2fs_cp_error(sbi)))
> > -		return -EIO;
> > -
> >  	clear_sbi_flag(sbi, SBI_IS_DIRTY);
> >  	clear_sbi_flag(sbi, SBI_NEED_CP);
> >  	__set_cp_next_pack(sbi);
> > @@ -1381,7 +1379,7 @@ static int do_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc)
> >  
> >  	f2fs_bug_on(sbi, get_pages(sbi, F2FS_DIRTY_DENTS));
> >  
> > -	return 0;
> > +	return unlikely(f2fs_cp_error(sbi));
> 
> return unlikely(f2fs_cp_error(sbi)) ? -EIO : 0;
> 

Done.

> >  }
> >  
> >  /*
> >