Re: [PATCH 2/2] zonefs: Fix O_APPEND async write handling

[Nathan Chancellor <nathan@kernel.org>]

On Mon, Mar 15, 2021 at 07:22:56AM +0000, Damien Le Moal wrote:
> On 2021/03/15 16:21, Johannes Thumshirn wrote:
> > On 15/03/2021 08:16, kernel test robot wrote:
> >> 818	static ssize_t zonefs_file_dio_write(struct kiocb *iocb, struct iov_iter *from)
> >>    819	{
> >>    820		struct inode *inode = file_inode(iocb->ki_filp);
> >>    821		struct zonefs_inode_info *zi = ZONEFS_I(inode);
> >>    822		struct super_block *sb = inode->i_sb;
> >>    823		bool sync = is_sync_kiocb(iocb);
> >>    824		bool append = false;
> >>    825		ssize_t ret, count;
> > 
> >>    843		count = zonefs_write_checks(iocb, from);
> >>  > 844		if (count <= 0)
> >>    845			goto inode_unlock;
> > 
> > Args that needs to be:
> > 			if (count <= 0) {
> > 				ret = count;
> > 				goto inode_unlock;
> > 			}
> > 
> > Sorry for not spotting it.
> 
> Yep. Sending v2. Weird that gcc does not complain on my local compile...

Unfortunately, GCC's version of this warning was disabled for default
compiles by Linus in commit 78a5255ffb6a ("Stop the ad-hoc games with
-Wno-maybe-initialized"). W=2 is required, which can be quite noisy from
my understanding. KCFLAGS=-Wmaybe-uninitialized is a good option.

Cheers,
Nathan

> > 
> >>    878	inode_unlock:
> >>    879		inode_unlock(inode);
> >>    880	
> >>    881		return ret;
> >>    882	}
> >>    883	
> > 
> 
> -- 
> Damien Le Moal
> Western Digital Research