Re: [PATCH v13 11/40] arm64/gcs: Provide basic EL2 setup to allow GCS usage at EL0 and EL1

[Catalin Marinas]

On Fri, Oct 11, 2024 at 01:55:05PM +0100, Marc Zyngier wrote:
> On Thu, 10 Oct 2024 18:16:52 +0100,
> Catalin Marinas <catalin.marinas@xxxxxxx> wrote:
> > 
> > On Thu, Oct 10, 2024 at 04:18:13PM +0100, Marc Zyngier wrote:
> > > From 20c98d2647c11db1e40768f92c5998ff5d764a3a Mon Sep 17 00:00:00 2001
> > > From: Marc Zyngier <maz@xxxxxxxxxx>
> > > Date: Thu, 10 Oct 2024 16:13:26 +0100
> > > Subject: [PATCH] KVM: arm64: Shave a few bytes from the EL2 idmap code
> > > 
> > > Our idmap is becoming too big, to the point where it doesn't fit in
> > > a 4kB page anymore.
> > > 
> > > There are some low-hanging fruits though, such as the el2_init_state
> > > horror that is expanded 3 times in the kernel. Let's at least limit
> > > ourselves to two copies, which makes the kernel link again.
> > > 
> > > At some point, we'll have to have a better way of doing this.
> > > 
> > > Reported-by: Nathan Chancellor <nathan@xxxxxxxxxx>
> > > Signed-off-by: Marc Zyngier <maz@xxxxxxxxxx>
> > > Link: https://lore.kernel.org/r/20241009204903.GA3353168@thelio-3990X
> > 
> > Thanks Marc for the quick fix. It looks fine to me, it will keep the
> > linker quiet for a while. I pushed it to arm64 for-kernelci for the time
> > being, see if anything falls apart. I'll apply it properly once it gets
> > a bit more testing.
> 
> Works for me. But if that helps, I can also queue it as a KVM fix for
> 6.12, leaving the GCS branch unencumbered. Just let me know.

That would help, thanks. The for-kernelci branch merges the latest mainline
rc anyway, so it will no longer blow up allmodconfig tests.

-- 
Catalin