Re: [PATCH 1/2] fs/erofs: Add support for EROFS

[Gao Xiang]

Hi,

On 2023/9/8 03:57, Daniel Kiper wrote:
> Hi Gao,
>
> On Wed, Sep 06, 2023 at 01:46:25PM +0800, Gao Xiang wrote:
> > Hi Daniel,
> >
> > On 2023/7/21 01:37, Glenn Washburn wrote:
> > > On Thu, 20 Jul 2023 21:14:22 +0800
> > > <zhaoyifan@sjtu.edu.cn> wrote:
> > >
> > > > Thank you for the reply!
> > > >
> > > > As Xiang pointed out, we believe that uncompressed support for
> > > > EROFS is still valuable, so I would be very grateful if this
> > > > support could be merged into the GRUB mainline.
> > >
> > > Yes, I believe this will get merged into master. The question is
> > > "when", after or before release. Now that I know that the compression is
> > > on a per-file basis and so this version can still be widely used (with
> > > the condition that the kernel and initrd are uncompressed by EROFS), I
> > > am in support of having this included in the upcoming release.
> > > Ultimately the decision is Daniel's though.
> >
> > ping.. May I ask what's your opinion of EROFS uncompressed support
> > for GRUB.  I assume compressed EROFS support needs more work to do,
> > so could we support uncompressed support as the first step since
> > EROFS supports per-file compression configuration?
> >
> > The v4 version is already finished with Glenn's RVB:
> > https://lists.gnu.org/archive/html/grub-devel/2023-07/msg00102.html
>
> We are in the GRUB 2.12 release process and busy with it. I will take
> a look at the patch set after the release. Sorry for delay.

Okay, thanks.

> > > > By the way, we expect to bring support for the EROFS compression
> > > > part within a few weeks. However, AFAIK, GRUB lacks generic
> > > > support for lz4 compression algorithms. Would it be more
> > > > appropriate to introduce a dedicated lz4 library for EROFS, like
> > > > zfs, or generic lz4 support for grub?
> > >
> > > I have not looked into ZFS's usage of its internal library, nor how
> > > hard it would be to make it a separate module. But I prefer the
> > > separate module approach. Perhaps Daniel can speak to his preferred> method.
>
> I agree with Glenn, please put/move lz4 compression code to separate module.

Thanks for the confirmation on this stuff!

Thanks,
Gao Xiang

> Daniel