Re: [f2fs-dev] [PATCH] f2fs: allow to set compression for inlined file

[Chao Yu]

On 2022/10/26 1:46, Jaegeuk Kim wrote:
> On 10/25, Chao Yu wrote:
> > On 2022/10/25 7:36, Jaegeuk Kim wrote:
> > > The below commit disallows to set compression on empty created file which
> > > has a inline_data. Let's fix it.
> > >
> > > Fixes: 7165841d578e ("f2fs: fix to check inline_data during compressed inode conversion")
> > > Signed-off-by: Jaegeuk Kim <jaegeuk@xxxxxxxxxx>
> > > ---
> > >    fs/f2fs/file.c | 3 +++
> > >    1 file changed, 3 insertions(+)
> > >
> > > diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c
> > > index 1c4004c9245f..304fe08edc61 100644
> > > --- a/fs/f2fs/file.c
> > > +++ b/fs/f2fs/file.c
> > > @@ -1915,6 +1915,9 @@ static int f2fs_setflags_common(struct inode *inode, u32 iflags, u32 mask)
> > >    			if (!f2fs_disable_compressed_file(inode))
> > >    				return -EINVAL;
> > >    		} else {
> > > +			/* try to convert inline_data to support compression */
> > > +			f2fs_convert_inline_inode(inode);
> >
> > It needs to check return value of f2fs_convert_inline_inode()?
>
> I intended to catch that in the below checks?

But it may hide the real error number, e.g. -ENOMEM returned from
f2fs_convert_inline_inode()?

Thanks,

> > Thanks,
> >
> > > +
> > >    			if (!f2fs_may_compress(inode))
> > >    				return -EINVAL;
> > >    			if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))