Re: [PATCH 3/3] btrfs: migrate to the newer memparse_safe() helper

[Qu Wenruo]

On 2023/12/27 16:57, David Disseldorp wrote:
> On Sat, 23 Dec 2023 20:28:07 +1030, Qu Wenruo wrote:
>
> > The new helper has better error report and correct overflow detection,
> > furthermore the old @retptr behavior is also kept, thus there should be
> > no behavior change.
> >
> > Signed-off-by: Qu Wenruo <wqu@xxxxxxxx>
> > ---
> >   fs/btrfs/ioctl.c |  8 ++++++--
> >   fs/btrfs/super.c |  8 ++++++++
> >   fs/btrfs/sysfs.c | 14 +++++++++++---
> >   3 files changed, 25 insertions(+), 5 deletions(-)
> >
> > diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
> > index 4e50b62db2a8..8bfd4b4ccf02 100644
> > --- a/fs/btrfs/ioctl.c
> > +++ b/fs/btrfs/ioctl.c
> > @@ -1175,8 +1175,12 @@ static noinline int btrfs_ioctl_resize(struct file *file,
> >   			mod = 1;
> >   			sizestr++;
> >   		}
> > -		new_size = memparse(sizestr, &retptr);
> > -		if (*retptr != '\0' || new_size == 0) {
> > +
> > +		ret = memparse_safe(sizestr, MEMPARSE_SUFFIXES_DEFAULT,
> > +				    &new_size, &retptr);
> > +		if (ret < 0)
> > +			goto out_finish;
> > +		if (*retptr != '\0') {
>
> Was dropping the -EINVAL return for new_size=0 intentional?

Oh, that's unintentional. Although we would reject the invalid string, a
dedicated "0" can still be parsed.
In that case we should still return -EINVAL.

I just got it confused with the old behavior for invalid string (where 0
is returned and @retptr is not advanced).
> >   			ret = -EINVAL;
> >   			goto out_finish;
> >   		}
> > diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
> > index 3a677b808f0f..2bb6ea525e89 100644
> > --- a/fs/btrfs/super.c
> > +++ b/fs/btrfs/super.c
> > @@ -400,6 +400,14 @@ static int btrfs_parse_param(struct fs_context *fc, struct fs_parameter *param)
> >   		ctx->thread_pool_size = result.uint_32;
> >   		break;
> >   	case Opt_max_inline:
> > +		int ret;
> > +
> > +		ret = memparse_safe(param->string, MEMPARSE_SUFFIXES_DEFAULT,
> > +				    &ctx->max_inline, NULL);
> > +		if (ret < 0) {
> > +			btrfs_err(NULL, "invalid string \"%s\"", param->string);
> > +			return ret;
> > +		}
> >   		ctx->max_inline = memparse(param->string, NULL);
>
> Looks like you overlooked removal of the old memparse() call above.

My bad, I forgot to remove the old line.

Furthermore, the declaration of "ret" inside case block is not allowed,
I'll fix it anyway.

Thanks,
Qu
> Cheers, David