Re: [f2fs-dev] [PATCH v3 3/6] f2fs: compress: fix to check unreleased compressed cluster

[Chao Yu]

On 2024/1/12 1:15, Daeho Jeong wrote:
> On Wed, Jan 10, 2024 at 5:33 PM Chao Yu <chao@xxxxxxxxxx> wrote:
> > On 2024/1/11 9:18, Daeho Jeong wrote:
> > > On Thu, Dec 28, 2023 at 6:33 AM Chao Yu <chao@xxxxxxxxxx> wrote:
> > > > From: Sheng Yong <shengyong@xxxxxxxx>
> > > >
> > > > Compressed cluster may not be released due to we can fail in
> > > > release_compress_blocks(), fix to handle reserved compressed
> > > > cluster correctly in reserve_compress_blocks().
> > > >
> > > > Fixes: 4c8ff7095bef ("f2fs: support data compression")
> > > > Signed-off-by: Sheng Yong <shengyong@xxxxxxxx>
> > > > Signed-off-by: Chao Yu <chao@xxxxxxxxxx>
> > > > ---
> > > >    fs/f2fs/file.c | 12 ++++++++++++
> > > >    1 file changed, 12 insertions(+)
> > > >
> > > > diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c
> > > > index 026d05a7edd8..782ae3be48f6 100644
> > > > --- a/fs/f2fs/file.c
> > > > +++ b/fs/f2fs/file.c
> > > > @@ -3624,6 +3624,15 @@ static int reserve_compress_blocks(struct dnode_of_data *dn, pgoff_t count)
> > > >                                   goto next;
> > > >                           }
> > > >
> > > > +                       /*
> > > > +                        * compressed cluster was not released due to
> > > > +                        * it fails in release_compress_blocks().
> > > > +                        */
> > > > +                       if (blkaddr == NEW_ADDR) {
> > > > +                               compr_blocks++;
> > > > +                               continue;
> > > > +                       }
> > > > +
> > > >                           if (__is_valid_data_blkaddr(blkaddr)) {
> > > >                                   compr_blocks++;
> > > >                                   continue;
> > >
> > > How about merging two conditions like "blkaddr == NEW_ADDR ||
> > > __is_valid_data_blkaddr(blkaddr)"?
> >
> > Oh, sure.
> >
> > > > @@ -3633,6 +3642,9 @@ static int reserve_compress_blocks(struct dnode_of_data *dn, pgoff_t count)
> > > >                   }
> > > >
> > > >                   reserved = cluster_size - compr_blocks;
> > > > +               if (!reserved)
> > > > +                       goto next;
> > > > +
> > >
> > > How can the reserved variable be zero?
> >
> > I guess it can happen if a cluster was not released during
> > release_compress_blocks(), then all blocks in the cluster should
> > has been reserved, so, in this round of reserving, it needs to skip
> > reserve blocks, right?
>
> Let's assume cluster_size is 4. How can compr_blocks be 4?
>
>                          if (i == 0) {
>                                  if (blkaddr == COMPRESS_ADDR)
>                                          continue;
>                                  dn->ofs_in_node += cluster_size;
>                                  goto next;
>                          }
>
> We skip the block having COMPRESS_ADDR when counting compr_blocks.
> So, the maximum value of compr_blocks should be 3, right?

Ah, got it, and I think you're right.

Should fix the condition as below?

		/* for the case all blocks in cluster were reserved */
		if (reserved == 1)
			goto next;

Thanks,

> > Thanks,
> >
> > > >                   ret = inc_valid_block_count(sbi, dn->inode, &reserved);
> > > >                   if (ret)
> > > >                           return ret;
> > > > --
> > > > 2.40.1
> > > >
> > > >
> > > >
> > > > _______________________________________________
> > > > Linux-f2fs-devel mailing list
> > > > Linux-f2fs-devel@xxxxxxxxxxxxxxxxxxxxx
> > > > https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel