On 03/10/2019 06:49 AM, Yu Zhao wrote:
> Switch from per mm_struct to per pmd page table lock by enabling
> ARCH_ENABLE_SPLIT_PMD_PTLOCK. This provides better granularity for
> large system.
>
> I'm not sure if there is contention on mm->page_table_lock. Given
> the option comes at no cost (apart from initializing more spin
> locks), why not enable it now.
>
> We only do so when pmd is not folded, so we don't mistakenly call
> pgtable_pmd_page_ctor() on pud or p4d in pgd_pgtable_alloc(). (We
> check shift against PMD_SHIFT, which is same as PUD_SHIFT when pmd
> is folded).
>
> Signed-off-by: Yu Zhao <yuzhao@xxxxxxxxxx>
> ---
> arch/arm64/Kconfig | 3 +++
> arch/arm64/include/asm/pgalloc.h | 12 +++++++++++-
> arch/arm64/include/asm/tlb.h | 5 ++++-
> 3 files changed, 18 insertions(+), 2 deletions(-)
>
> diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
> index cfbf307d6dc4..a3b1b789f766 100644
> --- a/arch/arm64/Kconfig
> +++ b/arch/arm64/Kconfig
> @@ -872,6 +872,9 @@ config ARCH_WANT_HUGE_PMD_SHARE
> config ARCH_HAS_CACHE_LINE_SIZE
> def_bool y
>
> +config ARCH_ENABLE_SPLIT_PMD_PTLOCK
> + def_bool y if PGTABLE_LEVELS > 2
> +
> config SECCOMP
> bool "Enable seccomp to safely compute untrusted bytecode"
> ---help---
> diff --git a/arch/arm64/include/asm/pgalloc.h b/arch/arm64/include/asm/pgalloc.h
> index 52fa47c73bf0..dabba4b2c61f 100644
> --- a/arch/arm64/include/asm/pgalloc.h
> +++ b/arch/arm64/include/asm/pgalloc.h
> @@ -33,12 +33,22 @@
>
> static inline pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long addr)
> {
> - return (pmd_t *)__get_free_page(PGALLOC_GFP);
> + struct page *page;
> +
> + page = alloc_page(PGALLOC_GFP);
> + if (!page)
> + return NULL;
> + if (!pgtable_pmd_page_ctor(page)) {
> + __free_page(page);
> + return NULL;
> + }
> + return page_address(page);
> }
>
> static inline void pmd_free(struct mm_struct *mm, pmd_t *pmdp)
> {
> BUG_ON((unsigned long)pmdp & (PAGE_SIZE-1));
> + pgtable_pmd_page_dtor(virt_to_page(pmdp));
> free_page((unsigned long)pmdp);
> }
There is just one problem here. ARM KVM's stage2_pmd_free() calls into pmd_free() on a page
originally allocated with __get_free_page() and never went through pgtable_pmd_page_ctor().
So when ARCH_ENABLE_SPLIT_PMD_PTLOCK is enabled
stage2_pmd_free()
pgtable_pmd_page_dtor()
ptlock_free()
kmem_cache_free(page_ptl_cachep, page->ptl)
Though SLUB implementation for kmem_cache_free() seems to be handling NULL page->ptl (as the
page never got it's lock allocated or initialized) correctly I am not sure if it is a right
thing to do.
