On Thu, 06 Mar 2025 09:56:32 +0000,
Shameerali Kolothum Thodi <shameerali.kolothum.thodi@xxxxxxxxxx> wrote:
>
>
>
> > -----Original Message-----
> > From: Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx>
> > Sent: Thursday, March 6, 2025 5:46 AM
> > To: Christoffer Dall <cdall@xxxxxxxxxxxxxxx>; Marc Zyngier
> > <maz@xxxxxxxxxx>
> > Cc: Oliver Upton <oliver.upton@xxxxxxxxx>; Shameerali Kolothum Thodi
> > <shameerali.kolothum.thodi@xxxxxxxxxx>; Linux Kernel Mailing List
> > <linux-kernel@xxxxxxxxxxxxxxx>; Linux Next Mailing List <linux-
> > next@xxxxxxxxxxxxxxx>
> > Subject: linux-next: build failure after merge of the kvm-arm tree
> >
> > Hi all,
> >
> > After merging the kvm-arm tree, today's linux-next build (arm
> > multi_v7_defconfig) failed like this:
> >
> > drivers/firmware/smccc/kvm_guest.c:58:14: warning: no previous prototype
> > for 'kvm_arm_target_impl_cpu_init' [-Wmissing-prototypes]
> > 58 | void __init kvm_arm_target_impl_cpu_init(void)
> > | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
> > drivers/firmware/smccc/kvm_guest.c: In function
> > 'kvm_arm_target_impl_cpu_init':
> > drivers/firmware/smccc/kvm_guest.c:89:39: error: invalid application of
> > 'sizeof' to incomplete type 'struct target_impl_cpu'
> > 89 | target = memblock_alloc(sizeof(*target) * max_cpus,
> > __alignof__(*target));
> > | ^
> > drivers/firmware/smccc/kvm_guest.c:89:62: error: invalid application of
> > '__alignof__' to incomplete type 'struct target_impl_cpu'
> > 89 | target = memblock_alloc(sizeof(*target) * max_cpus,
> > __alignof__(*target));
> > | ^~~~~~~~~~~
> > drivers/firmware/smccc/kvm_guest.c:102:23: error: invalid use of undefined
> > type 'struct target_impl_cpu'
> > 102 | target[i].midr = res.a1;
> > | ^
> > drivers/firmware/smccc/kvm_guest.c:102:26: error: invalid use of undefined
> > type 'struct target_impl_cpu'
> > 102 | target[i].midr = res.a1;
> > | ^
> > drivers/firmware/smccc/kvm_guest.c:103:23: error: invalid use of undefined
> > type 'struct target_impl_cpu'
> > 103 | target[i].revidr = res.a2;
> > | ^
> > drivers/firmware/smccc/kvm_guest.c:103:26: error: invalid use of undefined
> > type 'struct target_impl_cpu'
> > 103 | target[i].revidr = res.a2;
> > | ^
> > drivers/firmware/smccc/kvm_guest.c:104:23: error: invalid use of undefined
> > type 'struct target_impl_cpu'
> > 104 | target[i].aidr = res.a3;
> > | ^
> > drivers/firmware/smccc/kvm_guest.c:104:26: error: invalid use of undefined
> > type 'struct target_impl_cpu'
> > 104 | target[i].aidr = res.a3;
> > | ^
> > drivers/firmware/smccc/kvm_guest.c:107:14: error: implicit declaration of
> > function 'cpu_errata_set_target_impl' [-Wimplicit-function-declaration]
> > 107 | if (!cpu_errata_set_target_impl(max_cpus, target)) {
> > | ^~~~~~~~~~~~~~~~~~~~~~~~~~
> > drivers/firmware/smccc/kvm_guest.c:116:37: error: invalid application of
> > 'sizeof' to incomplete type 'struct target_impl_cpu'
> > 116 | memblock_free(target, sizeof(*target) * max_cpus);
> > | ^
> >
> > Caused by commit
> >
> > 86edf6bdcf05 ("smccc/kvm_guest: Enable errata based on implementation
> > CPUs")
> >
> > I have used the kvm-arm tree from next-20250305 for today.
>
> Thanks for reporting this.
>
> Hmm..kvm_guest.c gets build through HAVE_ARM_SMCCC_DISCOVERY
> which is selected by ARM_GIC_V3.
>
> We could limit the kvm_arm_target_impl_cpu_init() to ARM64 to fix this
> like below as these hypercall is only supported for KVM/ARM64.
>
> Or is there a better way to handle this?
>
> Thanks,
> Shameer
>
> ---8---
> diff --git a/drivers/firmware/smccc/kvm_guest.c
> b/drivers/firmware/smccc/kvm_guest.c
> index 2f03b582c298..5767aed25cdc 100644
> --- a/drivers/firmware/smccc/kvm_guest.c
> +++ b/drivers/firmware/smccc/kvm_guest.c
> @@ -55,6 +55,7 @@ bool kvm_arm_hyp_service_available(u32 func_id)
> }
> EXPORT_SYMBOL_GPL(kvm_arm_hyp_service_available);
>
> +#ifdef CONFIG_ARM64
> void __init kvm_arm_target_impl_cpu_init(void)
> {
> int i;
> @@ -115,3 +116,4 @@ void __init kvm_arm_target_impl_cpu_init(void)
> mem_free:
> memblock_free(target, sizeof(*target) * max_cpus);
> }
> +#endif
Yeah, that's probably best for now.
Oliver, do you mind stashing this on top so that -next can build
again? We can revisit this and have a better solution down the line.
Thanks,
M.
--
Without deviation from the norm, progress is not possible.
