Re: [PATCH] btrfs: disable zstd support for i386-pc

[Michael Chang]

On Fri, Nov 15, 2019 at 12:42:52PM +0100, Daniel Kiper wrote:
> On Thu, Nov 14, 2019 at 09:53:54AM +0000, Michael Chang wrote:
> > On Wed, Nov 13, 2019 at 12:00:50PM +0100, Daniel Kiper wrote:
> > > On Thu, Nov 07, 2019 at 12:52:35PM +0100, Daniel Kiper wrote:
> > > > On Wed, Nov 06, 2019 at 11:15:04AM -0800, Vladimir 'phcoder' Serbinenko 
> > > > wrote:
> > > > > Please don't do it this way. The right solution is to move it to 
> > > > > separate
> > > > > module and include zstd module when needed. Not everybody uses btrfs
> > > > > embedded area. I recommend not to use it. Using mbr gap or BBP is the
> > > > > recommended way.
> > > >
> > > > I will put the cat among the pigeons. Maybe we should finally stop
> > > > pretending that the GRUB supports small MBR gaps. Otherwise we will
> > > > be fighting with such issues endlessly.
> > >
> > > Bumping the thread... Nobody objects? Hmmm...
> >
> > I don't think we are able to give up MBR gap support, simply because no
> > other way out in creating the area for bootloader embedding. However we
> 
> Nope, I do not propose that...
> 
> > should consider to correct this claim in manual.
> >
> >  "You must ensure that the first partition starts at least 31 KiB (63
> >  sectors) from the start of the disk"
> >
> > to reflect the fact that 31 KiB is no longer applicable and the ideal
> > size should be above 1MB (or such). You should go check with your disk
> > tools to find the suitable parameter to fulfill the requirement, for eg,
> > the partition alignment would mostly affect this.
> 
> Exactly! However, there are some legacy systems which do not boot if MBR
> gap does not end at 63 sectors boundary. Hence, maybe we should suggest
> chainloading, using e.g. SYSLINUX, in a such cases. Anyway, may I ask
> you to prepare a patch for GRUB manual which describes the problem?

Alright, sounds good to me. :) 

Regards,
Michael

> 
> Daniel