Re: [PATCH v2 02/22] ieee1275: claim more memory

[Daniel Kiper]

On Fri, Jul 16, 2021 at 05:59:20AM +0200, Patrick Steinhardt wrote:
> On Thu, Jul 15, 2021 at 11:51:04PM +0200, Daniel Kiper wrote:
> > CC-in a few people who can be interested in this...
> >
> > On Wed, Jun 30, 2021 at 06:40:11PM +1000, Daniel Axtens wrote:
> > > On powerpc-ieee1275, we are running out of memory trying to verify
> > > anything. This is because:
> > >
> > >  - we have to load an entire file into memory to verify it. This is
> > >    extremely difficult to change with appended signatures.
> > >  - We only have 32MB of heap.
> > >  - Distro kernels are now often around 30MB.
> > >
> > > So we want to claim more memory from OpenFirmware for our heap.
> >
> > AFAICT it is common problem in the GRUB right now. Please take a look at
> > [1], [2]. It would be nice to find general solution for all. Of course
> > if possible. So, if somebody could take a look at memory management in
> > the GRUB and propose a solution for the problem that would be perfect.
> > Any volunteers?
> >
> > Daniel
> >
> > [1] https://lists.gnu.org/archive/html/grub-devel/2020-06/msg00009.html
> > [2] https://lists.gnu.org/archive/html/grub-devel/2021-01/msg00031.html
>
> I think that my [1] should solve the issue generically. Instead of
> bumping any of the static limits we have in place, we just drop them
> altogether in favor of dynamically requesting additional EFI regions
> whenever we realize that the currently mapped regions cannot satisfy our
> needs. Like this, we can lower the initially requested regions, but
> scale them to the specific needs if need be.
>
> I had planned to revisit this patch series much earlier, but somehow I
> didn't yet find the time. Any comments on my approach would be welcome
> though, and if we agree that this may be a viable route to go down then
> I'd be happy to further pursue it.

Patrick, I went quickly through the patches and I like the idea. I found
some minor issues there but they are not important at this point. If you
could rebase your patch set on current master that would be nice. Though
please remember that the interface should be sufficiently generic to
make use of it on the other architectures and platforms.

When you send new patch set please CC all people who are CC-ed in this email.

Daniel