Re: [PATCH RFC 09/13] mm/powerpc: Redefine pXd_huge() with pXd_leaf()

[Peter Xu <peterx@redhat.com>]

On Wed, Mar 06, 2024 at 11:56:56PM +1100, Michael Ellerman wrote:
> peterx@redhat.com writes:
> > From: Peter Xu <peterx@redhat.com>
> >
> > PowerPC book3s 4K mostly has the same definition on both, except pXd_huge()
> > constantly returns 0 for hash MMUs.  AFAICT that is fine to be removed,
> > because pXd_huge() reflects a hugetlb entry, while it's own hugetlb pgtable
> > lookup function (__find_linux_pte() shared by all powerpc code) already use
> > pXd_leaf() irrelevant of the MMU type.  It means pXd_leaf() should work all
> > fine with hash MMU pgtables or something could already went wrong.
> 
> Yes I think that's correct.
> 
> 4K Hash MMU doesn't support any hugepage size at PMD or PUD level (the
> geometry is wrong), so pmd/pud_huge() were written with that in mind,
> ie. they are hard coded to return false.
> 
> But it should be OK to use pmd/pud_leaf(), they will actually look for
> _PAGE_PTE, but it should never be set for 4K Hash.
> 
> See eg. arch/powerpc/include/asm/book3s/64/hash-4k.h:
> 
> static inline pmd_t hash__pmd_mkhuge(pmd_t pmd)
> {
> 	BUG();
> 	return pmd;
> }

Good to get confirmation on this, thanks, Michael.  These explanations also
look better than what I wrote, I'll amend the commit message.

> 
> > The goal should be that we will have one API pXd_leaf() to detect all kinds
> > of huge mappings.  AFAICT we need to use the pXd_leaf() impl (rather than
> > pXd_huge() ones) to make sure ie. THPs on hash MMU will also return true.
> >
> > This helps to simplify a follow up patch to drop pXd_huge() treewide.
> >
> > Cc: Michael Ellerman <mpe@ellerman.id.au>
> > Cc: Nicholas Piggin <npiggin@gmail.com>
> > Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
> > Cc: "Aneesh Kumar K.V" <aneesh.kumar@kernel.org>
> > Cc: "Naveen N. Rao" <naveen.n.rao@linux.ibm.com>
> > Cc: linuxppc-dev@lists.ozlabs.org
> > Signed-off-by: Peter Xu <peterx@redhat.com>
> > ---
> >  arch/powerpc/include/asm/book3s/64/pgtable-4k.h | 14 ++------------
> >  1 file changed, 2 insertions(+), 12 deletions(-)
> >
> > diff --git a/arch/powerpc/include/asm/book3s/64/pgtable-4k.h b/arch/powerpc/include/asm/book3s/64/pgtable-4k.h
> > index 48f21820afe2..92545981bb49 100644
> > --- a/arch/powerpc/include/asm/book3s/64/pgtable-4k.h
> > +++ b/arch/powerpc/include/asm/book3s/64/pgtable-4k.h
> > @@ -8,22 +8,12 @@
> >  #ifdef CONFIG_HUGETLB_PAGE
> >  static inline int pmd_huge(pmd_t pmd)
> >  {
> > -	/*
> > -	 * leaf pte for huge page
> > -	 */
> > -	if (radix_enabled())
> > -		return !!(pmd_raw(pmd) & cpu_to_be64(_PAGE_PTE));
> > -	return 0;
> > +	return pmd_leaf(pmd);
> >  }
> >  
> >  static inline int pud_huge(pud_t pud)
> >  {
> > -	/*
> > -	 * leaf pte for huge page
> > -	 */
> > -	if (radix_enabled())
> > -		return !!(pud_raw(pud) & cpu_to_be64(_PAGE_PTE));
> > -	return 0;
> > +	return pud_leaf(pud);
> >  }
> 
> This doesn't actually compile though.
> 
>   arch/powerpc/include/asm/book3s/64/pgtable-4k.h:11:16: error: implicit declaration of function ‘pmd_leaf’; did you mean ‘pgd_clear’? [-Werror=implicit-function-declaration]
> 
> etc.
> 
> To make it compile we'd need to relocate the pmd/pud_leaf() definitions:
> 
> diff --git a/arch/powerpc/include/asm/book3s/64/pgtable.h b/arch/powerpc/include/asm/book3s/64/pgtable.h
> index df66dce8306f..fd7180fded75 100644
> --- a/arch/powerpc/include/asm/book3s/64/pgtable.h
> +++ b/arch/powerpc/include/asm/book3s/64/pgtable.h
> @@ -262,6 +262,18 @@ extern unsigned long __kernel_io_end;
> 
>  extern struct page *vmemmap;
>  extern unsigned long pci_io_base;
> +
> +#define pmd_leaf pmd_leaf
> +static inline bool pmd_leaf(pmd_t pmd)
> +{
> +       return !!(pmd_raw(pmd) & cpu_to_be64(_PAGE_PTE));
> +}
> +
> +#define pud_leaf pud_leaf
> +static inline bool pud_leaf(pud_t pud)
> +{
> +       return !!(pud_raw(pud) & cpu_to_be64(_PAGE_PTE));
> +}
>  #endif /* __ASSEMBLY__ */
> 
>  #include <asm/book3s/64/hash.h>
> @@ -1436,20 +1448,5 @@ static inline bool is_pte_rw_upgrade(unsigned long old_val, unsigned long new_va
>         return false;
>  }
> 
> -/*
> - * Like pmd_huge(), but works regardless of config options
> - */
> -#define pmd_leaf pmd_leaf
> -static inline bool pmd_leaf(pmd_t pmd)
> -{
> -       return !!(pmd_raw(pmd) & cpu_to_be64(_PAGE_PTE));
> -}
> -
> -#define pud_leaf pud_leaf
> -static inline bool pud_leaf(pud_t pud)
> -{
> -       return !!(pud_raw(pud) & cpu_to_be64(_PAGE_PTE));
> -}
> -
>  #endif /* __ASSEMBLY__ */
>  #endif /* _ASM_POWERPC_BOOK3S_64_PGTABLE_H_ */

Thanks for the help, I'll fix that.  I'm wondering when syzbot will start
to feed my series into the testers; I do still rely on those feedbacks on
compilation issues with such treewide changes, but so far I didn't yet
receive any reports.

I've also attached the new patch directly here in case of any further
comment.

Thanks,

==========8<===========
From 9e75aef2141170f241577e7786aaa4bbbfd93360 Mon Sep 17 00:00:00 2001
From: Peter Xu <peterx@redhat.com>
Date: Wed, 6 Mar 2024 14:49:48 +0800
Subject: [PATCH] mm/powerpc: Redefine pXd_huge() with pXd_leaf()

PowerPC book3s 4K mostly has the same definition on both, except pXd_huge()
constantly returns 0 for hash MMUs.  As Michael Ellerman pointed out [1],
it is safe to check _PAGE_PTE on hash MMUs, as the bit will never be set so
it will keep returning false.

As a reference, __p[mu]d_mkhuge() will trigger a BUG_ON trying to create
such huge mappings for 4K hash MMUs.  Meanwhile, the major powerpc hugetlb
pgtable walker __find_linux_pte(), already used pXd_leaf() to check hugetlb
mappings.

The goal should be that we will have one API pXd_leaf() to detect all kinds
of huge mappings.  AFAICT we need to use the pXd_leaf() impl (rather than
pXd_huge() ones) to make sure ie. THPs on hash MMU will also return true.

This helps to simplify a follow up patch to drop pXd_huge() treewide.

NOTE: *_leaf() definition need to be moved before the inclusion of
asm/book3s/64/pgtable-4k.h, which defines pXd_huge() with it.

[1] https://lore.kernel.org/r/87v85zo6w7.fsf@mail.lhotse

Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@kernel.org>
Cc: "Naveen N. Rao" <naveen.n.rao@linux.ibm.com>
Cc: linuxppc-dev@lists.ozlabs.org
Signed-off-by: Peter Xu <peterx@redhat.com>
---
 .../include/asm/book3s/64/pgtable-4k.h        | 14 ++--------
 arch/powerpc/include/asm/book3s/64/pgtable.h  | 27 +++++++++----------
 2 files changed, 14 insertions(+), 27 deletions(-)

diff --git a/arch/powerpc/include/asm/book3s/64/pgtable-4k.h b/arch/powerpc/include/asm/book3s/64/pgtable-4k.h
index 48f21820afe2..92545981bb49 100644
--- a/arch/powerpc/include/asm/book3s/64/pgtable-4k.h
+++ b/arch/powerpc/include/asm/book3s/64/pgtable-4k.h
@@ -8,22 +8,12 @@
 #ifdef CONFIG_HUGETLB_PAGE
 static inline int pmd_huge(pmd_t pmd)
 {
-	/*
-	 * leaf pte for huge page
-	 */
-	if (radix_enabled())
-		return !!(pmd_raw(pmd) & cpu_to_be64(_PAGE_PTE));
-	return 0;
+	return pmd_leaf(pmd);
 }
 
 static inline int pud_huge(pud_t pud)
 {
-	/*
-	 * leaf pte for huge page
-	 */
-	if (radix_enabled())
-		return !!(pud_raw(pud) & cpu_to_be64(_PAGE_PTE));
-	return 0;
+	return pud_leaf(pud);
 }
 
 /*
diff --git a/arch/powerpc/include/asm/book3s/64/pgtable.h b/arch/powerpc/include/asm/book3s/64/pgtable.h
index df66dce8306f..fd7180fded75 100644
--- a/arch/powerpc/include/asm/book3s/64/pgtable.h
+++ b/arch/powerpc/include/asm/book3s/64/pgtable.h
@@ -262,6 +262,18 @@ extern unsigned long __kernel_io_end;
 
 extern struct page *vmemmap;
 extern unsigned long pci_io_base;
+
+#define pmd_leaf pmd_leaf
+static inline bool pmd_leaf(pmd_t pmd)
+{
+	return !!(pmd_raw(pmd) & cpu_to_be64(_PAGE_PTE));
+}
+
+#define pud_leaf pud_leaf
+static inline bool pud_leaf(pud_t pud)
+{
+	return !!(pud_raw(pud) & cpu_to_be64(_PAGE_PTE));
+}
 #endif /* __ASSEMBLY__ */
 
 #include <asm/book3s/64/hash.h>
@@ -1436,20 +1448,5 @@ static inline bool is_pte_rw_upgrade(unsigned long old_val, unsigned long new_va
 	return false;
 }
 
-/*
- * Like pmd_huge(), but works regardless of config options
- */
-#define pmd_leaf pmd_leaf
-static inline bool pmd_leaf(pmd_t pmd)
-{
-	return !!(pmd_raw(pmd) & cpu_to_be64(_PAGE_PTE));
-}
-
-#define pud_leaf pud_leaf
-static inline bool pud_leaf(pud_t pud)
-{
-	return !!(pud_raw(pud) & cpu_to_be64(_PAGE_PTE));
-}
-
 #endif /* __ASSEMBLY__ */
 #endif /* _ASM_POWERPC_BOOK3S_64_PGTABLE_H_ */
-- 
2.44.0

-- 
Peter Xu