Re: [stable:linux-5.15.y 55/9999] arch/x86/kvm/hyperv.c:2185:5: error: stack frame size (1036) exceeds limit (1024) in 'kvm_hv_hypercall'

[Sean Christopherson <seanjc@google.com>]

On Mon, Jul 10, 2023, Vitaly Kuznetsov wrote:
> kernel test robot <lkp@intel.com> writes:
> 
> > Hi Vitaly,
> >
> > First bad commit (maybe != root cause):
> >
> > tree:   https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git linux-5.15.y
> > head:   d54cfc420586425d418a53871290cc4a59d33501
> > commit: cb188e07105f2216f5efbefac95df4b6ce266906 [55/9999] KVM: x86: hyper-v: HVCALL_SEND_IPI_EX is an XMM fast hypercall
> > config: i386-buildonly-randconfig-r006-20230708 (https://download.01.org/0day-ci/archive/20230708/202307080326.zDp7E3o0-lkp@intel.com/config)
> > compiler: clang version 15.0.7 (https://github.com/llvm/llvm-project.git 8dfdcc7b7bf66834a761bd8de445840ef68e4d1a)
> > reproduce: (https://download.01.org/0day-ci/archive/20230708/202307080326.zDp7E3o0-lkp@intel.com/reproduce)
> >
> > If you fix the issue in a separate patch/commit (i.e. not just a new version of
> > the same patch/commit), kindly add following tags
> > | Reported-by: kernel test robot <lkp@intel.com>
> > | Closes: https://lore.kernel.org/oe-kbuild-all/202307080326.zDp7E3o0-lkp@intel.com/
> >
> > All errors (new ones prefixed by >>):
> >
> >>> arch/x86/kvm/hyperv.c:2185:5: error: stack frame size (1036) exceeds limit (1024) in 'kvm_hv_hypercall' [-Werror,-Wframe-larger-than]
> >    int kvm_hv_hypercall(struct kvm_vcpu *vcpu)
> >        ^
> >    1 error generated.
> 
> (sorry for delayed reply)
> 
> This used to be a warning (without CONFIG_KVM_WERROR I guess?) :-) E.g.
> 
> https://lore.kernel.org/kvm/87zgg6sza8.fsf@redhat.com/#t
> 
> where Nathan explained LLVM's behavior:
> 
> https://lore.kernel.org/kvm/Yvp87jlVWg0e376v@dev-arch.thelio-3990X/
> 
> This was 'fixed' upstream with
> 
> commit 7d5e88d301f84a7b64602dbe3640f288223095ea
> Author: Vitaly Kuznetsov <vkuznets@redhat.com>
> Date:   Tue Nov 1 15:53:56 2022 +0100
> 
>     KVM: x86: hyper-v: Use preallocated buffer in 'struct kvm_vcpu_hv' instead of on-stack 'sparse_banks'
>   
> and personally, I'm not against backporting it to 5.15.y but I seriously
> doubt it is worth the hassle (i386 KVM + llvm + CONFIG_KVM_WERROR is
> likely an impossible combo).
> 
> Also, there seems to be another build problem with CONFIG_KVM_WERROR I
> met with clan-16 and the same config:
> 
> ../arch/x86/kvm/x86.c:2315:19: error: unused function 'gtod_is_based_on_tsc' [-Werror,-Wunused-function]
> static inline int gtod_is_based_on_tsc(int mode)
> 
> TL;DR: Let's ignore this for 5.15, not worth fixing IMO. Cc: kvm@ to
> check if anyone thinks differently.

Ya, ignore it.  KVM_WERROR is off-by-default for 32-bit builds, and all evidence
suggests that no one uses KVM with 32-bit kernels these days, so I can't imagine
this negatively affects anyone.

  config KVM_WERROR
        bool "Compile KVM with -Werror"
        # KASAN may cause the build to fail due to larger frames
        default y if X86_64 && !KASAN