On Thu, May 23, 2024 at 07:07:06PM -0400, Peter Xu wrote:
> On Thu, May 23, 2024 at 03:48:22PM -0700, Dave Hansen wrote:
> > On 5/23/24 15:37, Peter Xu wrote:
> > > diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> > > index 317de2afd371..c4a2356b1a54 100644
> > > --- a/mm/huge_memory.c
> > > +++ b/mm/huge_memory.c
> > > @@ -1135,7 +1135,7 @@ static void insert_pfn_pmd(struct vm_area_struct *vma, unsigned long addr,
> > > goto out_unlock;
> > > }
> > >
> > > - entry = pmd_mkhuge(pfn_t_pmd(pfn, prot));
> > > + entry = pmd_mkhuge(pfn_t_pmd(pfn, pgprot_4k_2_large(prot)));
> > > if (pfn_t_devmap(pfn))
> > > entry = pmd_mkdevmap(entry);
> > > if (write) {
> >
> > Does this even compile on non-x86 architectures?
>
> Probably not.. I think I can define a pgprot_to_large() globally, pointing
> that to pgprot_4k_2_large() on x86 and make the fallback to be noop. And
> if there's a new version I'll guarantee to run over my cross compilers.
>
> Any comments on the idea itself? Do we have a problem, or maybe I
> overlooked something?
I also attached one new version of patch 2 that should pass the cross
builds. Please reviewers feel free to look at this one instead. From x86
perspective they should be the same thing.
Thanks,
===8<===
>From 1cce12c872cb01aaa8686d8f5c7cd6b266ca4e38 Mon Sep 17 00:00:00 2001
From: Peter Xu <peterx@xxxxxxxxxx>
Date: Thu, 23 May 2024 18:19:35 -0400
Subject: [PATCH rfcv1.1] mm/x86/pat: Do proper PAT bit shift for large mappings
For large mappings, the pgtable PAT is set on bit 12 (_PAGE_PAT_LARGE)
rather than bit 9 (_PAGE_PAT), while bit 9 is used as PAE hint. Do proper
shifting when inject large pfn pgtable mappings to make cache mode alright.
Cc: Alex Williamson <alex.williamson@xxxxxxxxxx>
Cc: Jason Gunthorpe <jgg@xxxxxxxxxx>
Cc: Al Viro <viro@xxxxxxxxxxxxxxxxxx>
Cc: Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx>
Cc: Andy Lutomirski <luto@xxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: Ingo Molnar <mingo@xxxxxxxxxx>
Cc: Borislav Petkov <bp@xxxxxxxxx>
Cc: Kirill A. Shutemov <kirill@xxxxxxxxxxxxx>
Cc: x86@xxxxxxxxxx
Signed-off-by: Peter Xu <peterx@xxxxxxxxxx>
---
arch/x86/include/asm/pgtable_types.h | 1 +
include/linux/pgtable.h | 4 ++++
mm/huge_memory.c | 4 ++--
3 files changed, 7 insertions(+), 2 deletions(-)
diff --git a/arch/x86/include/asm/pgtable_types.h b/arch/x86/include/asm/pgtable_types.h
index b78644962626..f9edb2bb1512 100644
--- a/arch/x86/include/asm/pgtable_types.h
+++ b/arch/x86/include/asm/pgtable_types.h
@@ -512,6 +512,7 @@ static inline pgprot_t pgprot_large_2_4k(pgprot_t pgprot)
return __pgprot(protval_large_2_4k(pgprot_val(pgprot)));
}
+#define pgprot_to_large(pgprot) pgprot_4k_2_large(pgprot)
typedef struct page *pgtable_t;
diff --git a/include/linux/pgtable.h b/include/linux/pgtable.h
index 18019f037bae..54487d2b3e40 100644
--- a/include/linux/pgtable.h
+++ b/include/linux/pgtable.h
@@ -1956,4 +1956,8 @@ pgprot_t vm_get_page_prot(unsigned long vm_flags) \
} \
EXPORT_SYMBOL(vm_get_page_prot);
+#ifndef pgprot_to_large
+#define pgprot_to_large(pgprot) pgprot
+#endif
+
#endif /* _LINUX_PGTABLE_H */
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index 317de2afd371..4c134a60fb64 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -1135,7 +1135,7 @@ static void insert_pfn_pmd(struct vm_area_struct *vma, unsigned long addr,
goto out_unlock;
}
- entry = pmd_mkhuge(pfn_t_pmd(pfn, prot));
+ entry = pmd_mkhuge(pfn_t_pmd(pfn, pgprot_to_large(prot)));
if (pfn_t_devmap(pfn))
entry = pmd_mkdevmap(entry);
if (write) {
@@ -1233,7 +1233,7 @@ static void insert_pfn_pud(struct vm_area_struct *vma, unsigned long addr,
goto out_unlock;
}
- entry = pud_mkhuge(pfn_t_pud(pfn, prot));
+ entry = pud_mkhuge(pfn_t_pud(pfn, pgprot_to_large(prot)));
if (pfn_t_devmap(pfn))
entry = pud_mkdevmap(entry);
if (write) {
--
2.45.0
--
Peter Xu
