Re: [PATCH] KVM: arm64: Advertise ID_AA64PFR2_EL1.GCIE

[Marc Zyngier <maz@kernel.org>]

On Sat, 04 Apr 2026 19:13:30 +0100,
Nathan Chancellor <nathan@kernel.org> wrote:
> 
> Hi Marc,
> 
> On Wed, Apr 01, 2026 at 06:00:17PM +0100, Marc Zyngier wrote:
> > As we are missing ID_AA64PFR2_EL1.GCIE from the kernel feature set,
> > userspace cannot write ID_AA64PFR2_EL1 with GCIE set, even if we are
> > on a GICv5 host.
> > 
> > Add the required field description.
> > 
> > Signed-off-by: Marc Zyngier <maz@kernel.org>
> > ---
> >  arch/arm64/kernel/cpufeature.c | 1 +
> >  1 file changed, 1 insertion(+)
> > 
> > diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
> > index 32c2dbcc0c641..5bca6e064ca72 100644
> > --- a/arch/arm64/kernel/cpufeature.c
> > +++ b/arch/arm64/kernel/cpufeature.c
> > @@ -327,6 +327,7 @@ static const struct arm64_ftr_bits ftr_id_aa64pfr2[] = {
> >  	ARM64_FTR_BITS(FTR_VISIBLE, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64PFR2_EL1_FPMR_SHIFT, 4, 0),
> >  	ARM64_FTR_BITS(FTR_VISIBLE, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64PFR2_EL1_MTEFAR_SHIFT, 4, ID_AA64PFR2_EL1_MTEFAR_NI),
> >  	ARM64_FTR_BITS(FTR_VISIBLE, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64PFR2_EL1_MTESTOREONLY_SHIFT, 4, ID_AA64PFR2_EL1_MTESTOREONLY_NI),
> > +	ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64PFR2_EL1_GCIE_SHIFT, 4, ID_AA64PFR2_EL1_GCIE_NI),
> >  	ARM64_FTR_END,
> >  };
> >  
> > -- 
> > 2.47.3
> > 
> 
> After this change in -next as commit 899ff451fcee ("KVM: arm64:
> Advertise ID_AA64PFR2_EL1.GCIE"), I am seeing a warning on boot in my
> simple QEMU boot tests.
> 
>   $ make -skj"$(nproc)" ARCH=arm64 CROSS_COMPILE=aarch64-linux- mrproper virtconfig Image.gz
> 
>   $ curl -LSs https://github.com/ClangBuiltLinux/boot-utils/releases/download/20241120-044434/arm64-rootfs.cpio.zst | zstd -d >rootfs.cpio
> 
>   $ qemu-system-aarch64 \
>       -display none \
>       -nodefaults \
>       -machine virt,gic-version=max \
>       -append 'console=ttyAMA0 earlycon' \
>       -kernel arch/arm64/boot/Image.gz \
>       -initrd rootfs.cpio \
>       -cpu host \
>       -enable-kvm \
>       -m 1G \
>       -smp 8 \
>       -serial mon:stdio
>   [    0.000000] Booting Linux on physical CPU 0x0000000000 [0x413fd0c1]
>   [    0.000000] Linux version 7.0.0-rc4-00058-g899ff451fcee (nathan@aadp) (aarch64-linux-gcc (GCC) 15.2.0, GNU ld (GNU Binutils) 2.45) #1 SMP PREEMPT Sat Apr  4 06:55:05 MST 2026
>   ...
>   [    0.000000] ------------[ cut here ]------------
>   [    0.000000] SYS_ID_AA64PFR2_EL1 has feature overlap at shift 12
>   [    0.000000] WARNING: arch/arm64/kernel/cpufeature.c:986 at init_cpu_features+0xbc/0x344, CPU#0: swapper/0
>   [    0.000000] Modules linked in:
>   [    0.000000] CPU: 0 UID: 0 PID: 0 Comm: swapper Not tainted 7.0.0-rc4-00058-g899ff451fcee #1 PREEMPT
>   [    0.000000] Hardware name: linux,dummy-virt (DT)
>   [    0.000000] pstate: 600000c5 (nZCv daIF -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
>   [    0.000000] pc : init_cpu_features+0xbc/0x344
>   [    0.000000] lr : init_cpu_features+0xbc/0x344
>   [    0.000000] sp : ffffcd0982373db0
>   [    0.000000] x29: ffffcd0982373db0 x28: 0000000000000010 x27: ffffcd0981c63878
>   [    0.000000] x26: 0000000000000018 x25: ffffcd0982013f38 x24: ffffcd0981c69068
>   [    0.000000] x23: ffffcd0981c635f0 x22: ffffcd0982388640 x21: 0000000000000003
>   [    0.000000] x20: 0000000000000017 x19: ffffcd09824c9308 x18: 000000000000000a
>   [    0.000000] x17: 5d305b203837205d x16: 305b203737205d30 x15: 0000000000000000
>   [    0.000000] x14: 0000000000000000 x13: 3231207466696873 x12: 2074612070616c72
>   [    0.000000] x11: 0000000000000058 x10: 0000000000000018 x9 : ffffcd0982396598
>   [    0.000000] x8 : 0000000000057fa8 x7 : 000000000000002a x6 : ffffcd09823ee598
>   [    0.000000] x5 : ffffcd09823ee598 x4 : 0000000000000000 x3 : 0000000000000000
>   [    0.000000] x2 : 0000000000000000 x1 : 0000000000000000 x0 : ffffcd09823852c0
>   [    0.000000] Call trace:
>   [    0.000000]  init_cpu_features+0xbc/0x344 (P)
>   [    0.000000]  cpuinfo_store_boot_cpu+0x48/0x54
>   [    0.000000]  smp_prepare_boot_cpu+0x28/0x38
>   [    0.000000]  start_kernel+0x248/0x780
>   [    0.000000]  __primary_switched+0x88/0x90
>   [    0.000000] ---[ end trace 0000000000000000 ]---
>   ...
>   ```
> 
> Is this expected? I assume not, hence the report. If there is any
> information I can provide or patches I can test, I am more than happy to
> do so.

Gah. No idea how I managed to miss that: the register fields must be
strictly ordered, and I placed the field in the wrong spot. The
following hack fixes it for me:

diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
index 5bca6e064ca72..1bfaa96881dab 100644
--- a/arch/arm64/kernel/cpufeature.c
+++ b/arch/arm64/kernel/cpufeature.c
@@ -325,9 +325,9 @@ static const struct arm64_ftr_bits ftr_id_aa64pfr1[] = {
 
 static const struct arm64_ftr_bits ftr_id_aa64pfr2[] = {
 	ARM64_FTR_BITS(FTR_VISIBLE, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64PFR2_EL1_FPMR_SHIFT, 4, 0),
+	ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64PFR2_EL1_GCIE_SHIFT, 4, ID_AA64PFR2_EL1_GCIE_NI),
 	ARM64_FTR_BITS(FTR_VISIBLE, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64PFR2_EL1_MTEFAR_SHIFT, 4, ID_AA64PFR2_EL1_MTEFAR_NI),
 	ARM64_FTR_BITS(FTR_VISIBLE, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64PFR2_EL1_MTESTOREONLY_SHIFT, 4, ID_AA64PFR2_EL1_MTESTOREONLY_NI),
-	ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64PFR2_EL1_GCIE_SHIFT, 4, ID_AA64PFR2_EL1_GCIE_NI),
 	ARM64_FTR_END,
 };
 
If that works for you, I'll fold that into the original patch...

Thanks for pointing this out!

	M.

-- 
Jazz isn't dead. It just smells funny.