On Thu, Feb 07, 2019 at 04:40:45PM +0100, David Sterba wrote:
> On Thu, Feb 07, 2019 at 01:36:47PM +0300, Dan Carpenter wrote:
> > Hello Dennis Zhou,
> >
> > The patch a67dc67a2cb8: "btrfs: change set_level() to bound the level
> > passed in" from Jan 28, 2019, leads to the following static checker
> > warning:
> >
> > fs/btrfs/compression.c:1576 btrfs_compress_str2level()
> > error: uninitialized symbol 'level'.
> >
> > fs/btrfs/compression.c
> > 1566 unsigned int btrfs_compress_str2level(unsigned int type, const char *str)
> > 1567 {
> > 1568 unsigned int level;
> > 1569 int ret;
> > 1570
> > 1571 if (!type)
> > 1572 return 0;
> > 1573
> > 1574 if (str[0] == ':') {
> > 1575 ret = kstrtouint(str + 1, 10, &level);
> > --> 1576 if (ret)
> > 1577 level = 0;
> >
> > I feel like if the user gives bad input then we should just return an
> > error code instead of picking a level.
>
> We've debated a bit and a warning with fallback to default is the
> preferred option.
> >
> > 1578 }
> >
> > level is not initialized if the first character is not ':'.
>
> Right that's a bug, can be fixed by initializing level to 0, the
> set_level() implementations will fall back to default then.
>
> Thanks for the report.
>
Yeah that was a blunder on my part. I fixed it in v2 of 0009 in that
series:
https://lore.kernel.org/lkml/20190206164854.GB73401@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/
> >
> > 1579
> > 1580 level = btrfs_compress_op[type]->set_level(level);
> > 1581
> > 1582 return level;
> > 1583 }
Thanks,
Dennis
