diff options
Diffstat (limited to 'arch/powerpc/mm/hugetlbpage.c')
| -rw-r--r-- | arch/powerpc/mm/hugetlbpage.c | 435 |
1 files changed, 301 insertions, 134 deletions
diff --git a/arch/powerpc/mm/hugetlbpage.c b/arch/powerpc/mm/hugetlbpage.c index 1a6de0a7d8e..7e70ae968e5 100644 --- a/arch/powerpc/mm/hugetlbpage.c +++ b/arch/powerpc/mm/hugetlbpage.c @@ -21,6 +21,9 @@ #include <asm/pgalloc.h> #include <asm/tlb.h> #include <asm/setup.h> +#include <asm/hugetlb.h> + +#ifdef CONFIG_HUGETLB_PAGE #define PAGE_SHIFT_64K 16 #define PAGE_SHIFT_16M 24 @@ -48,66 +51,61 @@ static u64 gpage_freearray[MAX_NUMBER_GPAGES]; static unsigned nr_gpages; #endif -static inline int shift_to_mmu_psize(unsigned int shift) -{ - int psize; +#define hugepd_none(hpd) ((hpd).pd == 0) - for (psize = 0; psize < MMU_PAGE_COUNT; ++psize) - if (mmu_psize_defs[psize].shift == shift) - return psize; - return -1; -} +#ifdef CONFIG_PPC_BOOK3S_64 +/* + * At this point we do the placement change only for BOOK3S 64. This would + * possibly work on other subarchs. + */ -static inline unsigned int mmu_psize_to_shift(unsigned int mmu_psize) +/* + * We have PGD_INDEX_SIZ = 12 and PTE_INDEX_SIZE = 8, so that we can have + * 16GB hugepage pte in PGD and 16MB hugepage pte at PMD; + */ +int pmd_huge(pmd_t pmd) { - if (mmu_psize_defs[mmu_psize].shift) - return mmu_psize_defs[mmu_psize].shift; - BUG(); + /* + * leaf pte for huge page, bottom two bits != 00 + */ + return ((pmd_val(pmd) & 0x3) != 0x0); } -#define hugepd_none(hpd) ((hpd).pd == 0) - -pte_t *find_linux_pte_or_hugepte(pgd_t *pgdir, unsigned long ea, unsigned *shift) +int pud_huge(pud_t pud) { - pgd_t *pg; - pud_t *pu; - pmd_t *pm; - hugepd_t *hpdp = NULL; - unsigned pdshift = PGDIR_SHIFT; - - if (shift) - *shift = 0; + /* + * leaf pte for huge page, bottom two bits != 00 + */ + return ((pud_val(pud) & 0x3) != 0x0); +} - pg = pgdir + pgd_index(ea); - if (is_hugepd(pg)) { - hpdp = (hugepd_t *)pg; - } else if (!pgd_none(*pg)) { - pdshift = PUD_SHIFT; - pu = pud_offset(pg, ea); - if (is_hugepd(pu)) - hpdp = (hugepd_t *)pu; - else if (!pud_none(*pu)) { - pdshift = PMD_SHIFT; - pm = pmd_offset(pu, ea); - if (is_hugepd(pm)) - hpdp = (hugepd_t *)pm; - else if (!pmd_none(*pm)) { - return pte_offset_kernel(pm, ea); - } - } - } +int pgd_huge(pgd_t pgd) +{ + /* + * leaf pte for huge page, bottom two bits != 00 + */ + return ((pgd_val(pgd) & 0x3) != 0x0); +} +#else +int pmd_huge(pmd_t pmd) +{ + return 0; +} - if (!hpdp) - return NULL; +int pud_huge(pud_t pud) +{ + return 0; +} - if (shift) - *shift = hugepd_shift(*hpdp); - return hugepte_offset(hpdp, ea, pdshift); +int pgd_huge(pgd_t pgd) +{ + return 0; } -EXPORT_SYMBOL_GPL(find_linux_pte_or_hugepte); +#endif pte_t *huge_pte_offset(struct mm_struct *mm, unsigned long addr) { + /* Only called for hugetlbfs pages, hence can ignore THP */ return find_linux_pte_or_hugepte(mm->pgd, addr, NULL); } @@ -145,6 +143,7 @@ static int __hugepte_alloc(struct mm_struct *mm, hugepd_t *hpdp, if (unlikely(!hugepd_none(*hpdp))) break; else + /* We use the old format for PPC_FSL_BOOK3E */ hpdp->pd = ((unsigned long)new & ~PD_HUGE) | pshift; } /* If we bailed from the for loop early, an error occurred, clean up */ @@ -156,9 +155,15 @@ static int __hugepte_alloc(struct mm_struct *mm, hugepd_t *hpdp, #else if (!hugepd_none(*hpdp)) kmem_cache_free(cachep, new); - else + else { +#ifdef CONFIG_PPC_BOOK3S_64 + hpdp->pd = (unsigned long)new | + (shift_to_mmu_psize(pshift) << 2); +#else hpdp->pd = ((unsigned long)new & ~PD_HUGE) | pshift; #endif + } +#endif spin_unlock(&mm->page_table_lock); return 0; } @@ -175,6 +180,61 @@ static int __hugepte_alloc(struct mm_struct *mm, hugepd_t *hpdp, #define HUGEPD_PUD_SHIFT PMD_SHIFT #endif +#ifdef CONFIG_PPC_BOOK3S_64 +/* + * At this point we do the placement change only for BOOK3S 64. This would + * possibly work on other subarchs. + */ +pte_t *huge_pte_alloc(struct mm_struct *mm, unsigned long addr, unsigned long sz) +{ + pgd_t *pg; + pud_t *pu; + pmd_t *pm; + hugepd_t *hpdp = NULL; + unsigned pshift = __ffs(sz); + unsigned pdshift = PGDIR_SHIFT; + + addr &= ~(sz-1); + pg = pgd_offset(mm, addr); + + if (pshift == PGDIR_SHIFT) + /* 16GB huge page */ + return (pte_t *) pg; + else if (pshift > PUD_SHIFT) + /* + * We need to use hugepd table + */ + hpdp = (hugepd_t *)pg; + else { + pdshift = PUD_SHIFT; + pu = pud_alloc(mm, pg, addr); + if (pshift == PUD_SHIFT) + return (pte_t *)pu; + else if (pshift > PMD_SHIFT) + hpdp = (hugepd_t *)pu; + else { + pdshift = PMD_SHIFT; + pm = pmd_alloc(mm, pu, addr); + if (pshift == PMD_SHIFT) + /* 16MB hugepage */ + return (pte_t *)pm; + else + hpdp = (hugepd_t *)pm; + } + } + if (!hpdp) + return NULL; + + BUG_ON(!hugepd_none(*hpdp) && !hugepd_ok(*hpdp)); + + if (hugepd_none(*hpdp) && __hugepte_alloc(mm, hpdp, addr, pdshift, pshift)) + return NULL; + + return hugepte_offset(hpdp, addr, pdshift); +} + +#else + pte_t *huge_pte_alloc(struct mm_struct *mm, unsigned long addr, unsigned long sz) { pgd_t *pg; @@ -212,6 +272,7 @@ pte_t *huge_pte_alloc(struct mm_struct *mm, unsigned long addr, unsigned long sz return hugepte_offset(hpdp, addr, pdshift); } +#endif #ifdef CONFIG_PPC_FSL_BOOK3E /* Build list of addresses of gigantic pages. This function is used in early @@ -240,7 +301,7 @@ void add_gpage(u64 addr, u64 page_size, unsigned long number_of_pages) int alloc_bootmem_huge_page(struct hstate *hstate) { struct huge_bootmem_page *m; - int idx = shift_to_mmu_psize(hstate->order + PAGE_SHIFT); + int idx = shift_to_mmu_psize(huge_page_shift(hstate)); int nr_gpages = gpage_freearray[idx].nr_gpages; if (nr_gpages == 0) @@ -401,12 +462,13 @@ static void hugepd_free(struct mmu_gather *tlb, void *hugepte) { struct hugepd_freelist **batchp; - batchp = &__get_cpu_var(hugepd_freelist_cur); + batchp = &get_cpu_var(hugepd_freelist_cur); if (atomic_read(&tlb->mm->mm_users) < 2 || cpumask_equal(mm_cpumask(tlb->mm), cpumask_of(smp_processor_id()))) { kmem_cache_free(hugepte_cache, hugepte); + put_cpu_var(hugepd_freelist_cur); return; } @@ -420,6 +482,7 @@ static void hugepd_free(struct mmu_gather *tlb, void *hugepte) call_rcu_sched(&(*batchp)->rcu, hugepd_free_rcu_callback); *batchp = NULL; } + put_cpu_var(hugepd_freelist_cur); } #endif @@ -475,8 +538,14 @@ static void hugetlb_free_pmd_range(struct mmu_gather *tlb, pud_t *pud, do { pmd = pmd_offset(pud, addr); next = pmd_addr_end(addr, end); - if (pmd_none(*pmd)) + if (!is_hugepd(pmd)) { + /* + * if it is not hugepd pointer, we should already find + * it cleared. + */ + WARN_ON(!pmd_none_or_clear_bad(pmd)); continue; + } #ifdef CONFIG_PPC_FSL_BOOK3E /* * Increment next by the size of the huge mapping since @@ -556,8 +625,6 @@ static void hugetlb_free_pud_range(struct mmu_gather *tlb, pgd_t *pgd, /* * This function frees user-level page tables of a process. - * - * Must be called with pagetable lock held. */ void hugetlb_free_pgd_range(struct mmu_gather *tlb, unsigned long addr, unsigned long end, @@ -613,11 +680,14 @@ follow_huge_addr(struct mm_struct *mm, unsigned long address, int write) struct page *page; unsigned shift; unsigned long mask; - + /* + * Transparent hugepages are handled by generic code. We can skip them + * here. + */ ptep = find_linux_pte_or_hugepte(mm->pgd, address, &shift); /* Verify it is a huge page else bail. */ - if (!ptep || !shift) + if (!ptep || !shift || pmd_trans_huge(*(pmd_t *)ptep)) return ERR_PTR(-EINVAL); mask = (1UL << shift) - 1; @@ -628,16 +698,6 @@ follow_huge_addr(struct mm_struct *mm, unsigned long address, int write) return page; } -int pmd_huge(pmd_t pmd) -{ - return 0; -} - -int pud_huge(pud_t pud) -{ - return 0; -} - struct page * follow_huge_pmd(struct mm_struct *mm, unsigned long address, pmd_t *pmd, int write) @@ -646,69 +706,6 @@ follow_huge_pmd(struct mm_struct *mm, unsigned long address, return NULL; } -static noinline int gup_hugepte(pte_t *ptep, unsigned long sz, unsigned long addr, - unsigned long end, int write, struct page **pages, int *nr) -{ - unsigned long mask; - unsigned long pte_end; - struct page *head, *page, *tail; - pte_t pte; - int refs; - - pte_end = (addr + sz) & ~(sz-1); - if (pte_end < end) - end = pte_end; - - pte = *ptep; - mask = _PAGE_PRESENT | _PAGE_USER; - if (write) - mask |= _PAGE_RW; - - if ((pte_val(pte) & mask) != mask) - return 0; - - /* hugepages are never "special" */ - VM_BUG_ON(!pfn_valid(pte_pfn(pte))); - - refs = 0; - head = pte_page(pte); - - page = head + ((addr & (sz-1)) >> PAGE_SHIFT); - tail = page; - do { - VM_BUG_ON(compound_head(page) != head); - pages[*nr] = page; - (*nr)++; - page++; - refs++; - } while (addr += PAGE_SIZE, addr != end); - - if (!page_cache_add_speculative(head, refs)) { - *nr -= refs; - return 0; - } - - if (unlikely(pte_val(pte) != pte_val(*ptep))) { - /* Could be optimized better */ - *nr -= refs; - while (refs--) - put_page(head); - return 0; - } - - /* - * Any tail page need their mapcount reference taken before we - * return. - */ - while (refs--) { - if (PageTail(tail)) - get_huge_page_tail(tail); - tail++; - } - - return 1; -} - static unsigned long hugepte_addr_end(unsigned long addr, unsigned long end, unsigned long sz) { @@ -742,7 +739,7 @@ unsigned long hugetlb_get_unmapped_area(struct file *file, unsigned long addr, struct hstate *hstate = hstate_file(file); int mmu_psize = shift_to_mmu_psize(huge_page_shift(hstate)); - return slice_get_unmapped_area(addr, len, flags, mmu_psize, 1, 0); + return slice_get_unmapped_area(addr, len, flags, mmu_psize, 1); } #endif @@ -883,11 +880,16 @@ static int __init hugetlbpage_init(void) pdshift = PUD_SHIFT; else pdshift = PGDIR_SHIFT; - - pgtable_cache_add(pdshift - shift, NULL); - if (!PGT_CACHE(pdshift - shift)) - panic("hugetlbpage_init(): could not create " - "pgtable cache for %d bit pagesize\n", shift); + /* + * if we have pdshift and shift value same, we don't + * use pgt cache for hugepd. + */ + if (pdshift != shift) { + pgtable_cache_add(pdshift - shift, NULL); + if (!PGT_CACHE(pdshift - shift)) + panic("hugetlbpage_init(): could not create " + "pgtable cache for %d bit pagesize\n", shift); + } } /* Set default large page size. Currently, we pick 16M or 1M @@ -920,3 +922,168 @@ void flush_dcache_icache_hugepage(struct page *page) } } } + +#endif /* CONFIG_HUGETLB_PAGE */ + +/* + * We have 4 cases for pgds and pmds: + * (1) invalid (all zeroes) + * (2) pointer to next table, as normal; bottom 6 bits == 0 + * (3) leaf pte for huge page, bottom two bits != 00 + * (4) hugepd pointer, bottom two bits == 00, next 4 bits indicate size of table + * + * So long as we atomically load page table pointers we are safe against teardown, + * we can follow the address down to the the page and take a ref on it. + */ + +pte_t *find_linux_pte_or_hugepte(pgd_t *pgdir, unsigned long ea, unsigned *shift) +{ + pgd_t pgd, *pgdp; + pud_t pud, *pudp; + pmd_t pmd, *pmdp; + pte_t *ret_pte; + hugepd_t *hpdp = NULL; + unsigned pdshift = PGDIR_SHIFT; + + if (shift) + *shift = 0; + + pgdp = pgdir + pgd_index(ea); + pgd = ACCESS_ONCE(*pgdp); + /* + * Always operate on the local stack value. This make sure the + * value don't get updated by a parallel THP split/collapse, + * page fault or a page unmap. The return pte_t * is still not + * stable. So should be checked there for above conditions. + */ + if (pgd_none(pgd)) + return NULL; + else if (pgd_huge(pgd)) { + ret_pte = (pte_t *) pgdp; + goto out; + } else if (is_hugepd(&pgd)) + hpdp = (hugepd_t *)&pgd; + else { + /* + * Even if we end up with an unmap, the pgtable will not + * be freed, because we do an rcu free and here we are + * irq disabled + */ + pdshift = PUD_SHIFT; + pudp = pud_offset(&pgd, ea); + pud = ACCESS_ONCE(*pudp); + + if (pud_none(pud)) + return NULL; + else if (pud_huge(pud)) { + ret_pte = (pte_t *) pudp; + goto out; + } else if (is_hugepd(&pud)) + hpdp = (hugepd_t *)&pud; + else { + pdshift = PMD_SHIFT; + pmdp = pmd_offset(&pud, ea); + pmd = ACCESS_ONCE(*pmdp); + /* + * A hugepage collapse is captured by pmd_none, because + * it mark the pmd none and do a hpte invalidate. + * + * A hugepage split is captured by pmd_trans_splitting + * because we mark the pmd trans splitting and do a + * hpte invalidate + * + */ + if (pmd_none(pmd) || pmd_trans_splitting(pmd)) + return NULL; + + if (pmd_huge(pmd) || pmd_large(pmd)) { + ret_pte = (pte_t *) pmdp; + goto out; + } else if (is_hugepd(&pmd)) + hpdp = (hugepd_t *)&pmd; + else + return pte_offset_kernel(&pmd, ea); + } + } + if (!hpdp) + return NULL; + + ret_pte = hugepte_offset(hpdp, ea, pdshift); + pdshift = hugepd_shift(*hpdp); +out: + if (shift) + *shift = pdshift; + return ret_pte; +} +EXPORT_SYMBOL_GPL(find_linux_pte_or_hugepte); + +int gup_hugepte(pte_t *ptep, unsigned long sz, unsigned long addr, + unsigned long end, int write, struct page **pages, int *nr) +{ + unsigned long mask; + unsigned long pte_end; + struct page *head, *page, *tail; + pte_t pte; + int refs; + + pte_end = (addr + sz) & ~(sz-1); + if (pte_end < end) + end = pte_end; + + pte = ACCESS_ONCE(*ptep); + mask = _PAGE_PRESENT | _PAGE_USER; + if (write) + mask |= _PAGE_RW; + + if ((pte_val(pte) & mask) != mask) + return 0; + +#ifdef CONFIG_TRANSPARENT_HUGEPAGE + /* + * check for splitting here + */ + if (pmd_trans_splitting(pte_pmd(pte))) + return 0; +#endif + + /* hugepages are never "special" */ + VM_BUG_ON(!pfn_valid(pte_pfn(pte))); + + refs = 0; + head = pte_page(pte); + + page = head + ((addr & (sz-1)) >> PAGE_SHIFT); + tail = page; + do { + VM_BUG_ON(compound_head(page) != head); + pages[*nr] = page; + (*nr)++; + page++; + refs++; + } while (addr += PAGE_SIZE, addr != end); + + if (!page_cache_add_speculative(head, refs)) { + *nr -= refs; + return 0; + } + + if (unlikely(pte_val(pte) != pte_val(*ptep))) { + /* Could be optimized better */ + *nr -= refs; + while (refs--) + put_page(head); + return 0; + } + + /* + * Any tail page need their mapcount reference taken before we + * return. + */ + while (refs--) { + if (PageTail(tail)) + get_huge_page_tail(tail); + tail++; + } + + return 1; +} |