Re: [PATCH v10 0/4] x86/pvh: fix unbootable VMs again (PVH + KASAN)

[Mauricio Faria de Oliveira <mfo@igalia.com>]

On 2026-09-22 00:58, Borislav Petkov wrote:
> On Mon, Sep 21, 2026 at 10:36:31PM -0300, Mauricio Faria de Oliveira wrote:
>> The issue of unbootable VMs with CONFIG_PVH due to CONFIG_KASAN is back.
>> 
>> Booting directly from vmlinux (instead of bzImage) now fails with gcc-14/15
>> (but works with gcc-12/13) if CONFIG_KASAN_GENERIC is set, on Ubuntu 25.10.
>> 
>> The PVH code is required/supposed not to use the KASAN memory access check
>> in the kernel entry point as KASAN has not yet been setup, or an exception
>> is hit and the boot fails.
>> 
>> This was previously described and addressed with __builtin_mem{cmp,set}():
>> - commit 661362e3dcab ("xen, pvh: fix unbootable VMs (PVH + KASAN - AMD_MEM_ENCRYPT)")
>> - commit 416a33c9afce ("x86/cpu: fix unbootable VMs by inlining memcmp() in hypervisor_cpuid_base()")
>> - commit fbe5a6dfe492 ("xen, pvh: fix unbootable VMs by inlining memset() in xen_prepare_pvh()")
>> 
>> However, even with __builtin the compiler may decide to use the out of line
>> function instead of the inline implementation. So, that does not really fix
>> the issue unconditionally; see details below.
> 
> So, this whole deal doesn't sound to me like we need to backport it to stable
> - it rather looks more like fixing some configs which want to enable KASAN on
> PVH guests.
> 
> In that case, I'll queue this for 7.4.
> 
> If this needs to go to stable, then there better be a pretty good reason for
> it.
> 
> Right?

I think that is fine, yes. Even though it's a boot failure, actually
hitting it depends on all of: CONFIG_KASAN, CONFIG_PVH, booting from the
PVH entry point, _plus_ a compiler version that triggers it.

Nonetheless, this may be picked up for stable due to the Fixes: tags,
but I can provide backports as needed.

BTW, I just realized that the title of patch 4/4 is missing memset().
Would you mind adding it, please? Or I can send v11.
-x86/cpuid: fix unbootable VMs by really inlining memcmp() in
cpuid_base_hypervisor() and xen_prepare_pvh()
+x86/cpuid: fix unbootable VMs by really inlining memcmp() and memset()
in cpuid_base_hypervisor() and xen_prepare_pvh()

Thanks,

-- 
Mauricio