Re: [PATCH RFC v2 14/20] x86: mm: skip pud setup when using generic compile-time folded pagetable

Yeoreum Yun <[email protected]> Wed, 22 Jul 2026 21:40:42 +0100
Newsgroups org.kernel.vger.linux-openrisc,dev.linux.lists.loongarch,org.infradead.lists.kvm-riscv,org.infradead.lists.linux-riscv,org.kernel.vger.kvm,org.kernel.vger.linux-arch,org.kernel.vger.linux-kernel,org.kernel.vger.linux-mips,org.kvack.linux-mm
Message-ID <[email protected]>
On Wed, Jul 22, 2026 at 01:28:40PM -0700, H. Peter Anvin wrote:
> On July 22, 2026 1:22:04 PM PDT, Dave Hansen <[email protected]> wrote:
> >On 7/22/26 13:18, Yeoreum Yun wrote:
> >> Sorry. But do you mean check with cpu_feature_enabled() not with
> >> boot_cpu_has()?
> >> IIRC, the boot_cpu_has() doesn't use the DISABLE_MASK_BIT_SET() but
> >> uses REQUIRED_MASK_BIT_SET() and I don't expect it wouldn't make
> >> an optimisation to constant check which is contrast to DISABLE_MASK_BIT_SET().
> >
> >Ahh, yes. I forgot about that nuance! Moving to cpu_feature_enabled()
> >would be great!
> 
> It is supposed to. That's the point of the required mask.

Sorry, IIUC if the X86_DIRECT_GBPAGES is not enabled, it would fallback
to test_cpu_cap() which wouldn't be optimised at compile time IIUC.
So It would be better to use cpu_feature_enabled().

Am I missing something?

-- 
Sincerely,
Yeoreum Yun