Re: [PATCH] ArmPkg/ArmGicDxe ARM: fix encoding for GICv3 interrupt acknowledge

[Ard Biesheuvel <ard.biesheuvel@linaro.org>]

On Wed, 14 Nov 2018 at 14:11, Laszlo Ersek <lersek@redhat.com> wrote:
>
> On 11/14/18 21:00, Leif Lindholm wrote:
> > On Wed, Nov 14, 2018 at 11:27:24AM -0800, Ard Biesheuvel wrote:
> >> Fix a typo in the 32-bit ARM version of the GICv3 driver, which uses
> >> the wrong system register encoding to access ICC_IAR1, and attempted
> >> to access ICC_IAR0 instead. This results in boot time hangs both
> >> under QEMU emulation and on real hardware.
> >>
> >> Contributed-under: TianoCore Contribution Agreement 1.1
> >> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
> >
> > Reviewed-by: Leif Lindholm <leif.lindholm@linaro.org>
> >
> > I would say given how long we've gone without finding this,
>
> Right, that makes me curious -- what has changed now? What exposed this bug?
>

I was regression testing a EFI workaround I put in the kernel for
GICv3, which was apparently the first time anyone tried running
EDK2/ARM on a GICv3 system (which is one of the reasons I wanted to
get you one of the Socionext SynQuacer boards: it has a GICv3 with
GICv2 compatibility and support for 32-bit guests, but sadly, we still
don't have any with fixed silicon)