--- zzzz-none-000/linux-3.10.107/arch/powerpc/mm/hugetlbpage.c 2017-06-27 09:49:32.000000000 +0000 +++ scorpion-7490-727/linux-3.10.107/arch/powerpc/mm/hugetlbpage.c 2021-02-04 17:41:59.000000000 +0000 @@ -21,6 +21,9 @@ #include #include #include +#include + +#ifdef CONFIG_HUGETLB_PAGE #define PAGE_SHIFT_64K 16 #define PAGE_SHIFT_16M 24 @@ -59,6 +62,9 @@ /* * 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; + * + * Defined in such a way that we can optimize away code block at build time + * if CONFIG_HUGETLB_PAGE=n. */ int pmd_huge(pmd_t pmd) { @@ -83,6 +89,25 @@ */ return ((pgd_val(pgd) & 0x3) != 0x0); } + +#if defined(CONFIG_PPC_64K_PAGES) && defined(CONFIG_DEBUG_VM) +/* + * This enables us to catch the wrong page directory format + * Moved here so that we can use WARN() in the call. + */ +int hugepd_ok(hugepd_t hpd) +{ + bool is_hugepd; + + /* + * We should not find this format in page directory, warn otherwise. + */ + is_hugepd = (((hpd.pd & 0x3) == 0x0) && ((hpd.pd & HUGEPD_SHIFT_MASK) != 0)); + WARN(is_hugepd, "Found wrong page directory format\n"); + return 0; +} +#endif + #else int pmd_huge(pmd_t pmd) { @@ -100,69 +125,10 @@ } #endif -/* - * 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 - */ -pte_t *find_linux_pte_or_hugepte(pgd_t *pgdir, unsigned long ea, unsigned *shift) -{ - pgd_t *pg; - pud_t *pu; - pmd_t *pm; - pte_t *ret_pte; - hugepd_t *hpdp = NULL; - unsigned pdshift = PGDIR_SHIFT; - - if (shift) - *shift = 0; - - pg = pgdir + pgd_index(ea); - - if (pgd_huge(*pg)) { - ret_pte = (pte_t *) pg; - goto out; - } else if (is_hugepd(pg)) - hpdp = (hugepd_t *)pg; - else if (!pgd_none(*pg)) { - pdshift = PUD_SHIFT; - pu = pud_offset(pg, ea); - - if (pud_huge(*pu)) { - ret_pte = (pte_t *) pu; - goto out; - } else if (is_hugepd(pu)) - hpdp = (hugepd_t *)pu; - else if (!pud_none(*pu)) { - pdshift = PMD_SHIFT; - pm = pmd_offset(pu, ea); - - if (pmd_huge(*pm)) { - ret_pte = (pte_t *) pm; - goto out; - } else if (is_hugepd(pm)) - hpdp = (hugepd_t *)pm; - else if (!pmd_none(*pm)) - return pte_offset_kernel(pm, 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); - pte_t *huge_pte_offset(struct mm_struct *mm, unsigned long addr) { - return find_linux_pte_or_hugepte(mm->pgd, addr, NULL); + /* Only called for hugetlbfs pages, hence can ignore THP */ + return __find_linux_pte_or_hugepte(mm->pgd, addr, NULL, NULL); } static int __hugepte_alloc(struct mm_struct *mm, hugepd_t *hpdp, @@ -286,7 +252,7 @@ if (hugepd_none(*hpdp) && __hugepte_alloc(mm, hpdp, addr, pdshift, pshift)) return NULL; - return hugepte_offset(hpdp, addr, pdshift); + return hugepte_offset(*hpdp, addr, pdshift); } #else @@ -326,13 +292,13 @@ if (hugepd_none(*hpdp) && __hugepte_alloc(mm, hpdp, addr, pdshift, pshift)) return NULL; - return hugepte_offset(hpdp, addr, pdshift); + 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 - * boot before the buddy or bootmem allocator is setup. + * boot before the buddy allocator is setup. */ void add_gpage(u64 addr, u64 page_size, unsigned long number_of_pages) { @@ -357,7 +323,7 @@ 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) @@ -368,7 +334,7 @@ * If gpages can be in highmem we can't use the trick of storing the * data structure in the page; allocate space for this */ - m = alloc_bootmem(sizeof(struct huge_bootmem_page)); + m = memblock_virt_alloc(sizeof(struct huge_bootmem_page), 0); m->phys = gpage_freearray[idx].gpage_list[--nr_gpages]; #else m = phys_to_virt(gpage_freearray[idx].gpage_list[--nr_gpages]); @@ -389,7 +355,7 @@ unsigned long gpage_npages[MMU_PAGE_COUNT]; static int __init do_gpage_early_setup(char *param, char *val, - const char *unused) + const char *unused, void *arg) { static phys_addr_t size; unsigned long npages; @@ -408,6 +374,13 @@ if (size != 0) { if (sscanf(val, "%lu", &npages) <= 0) npages = 0; + if (npages > MAX_NUMBER_GPAGES) { + pr_warn("MMU: %lu pages requested for page " + "size %llu KB, limiting to " + __stringify(MAX_NUMBER_GPAGES) "\n", + npages, size / 1024); + npages = MAX_NUMBER_GPAGES; + } gpage_npages[shift_to_mmu_psize(__ffs(size))] = npages; size = 0; } @@ -431,7 +404,7 @@ strlcpy(cmdline, boot_command_line, COMMAND_LINE_SIZE); parse_args("hugetlb gpages", cmdline, NULL, 0, 0, 0, - &do_gpage_early_setup); + NULL, &do_gpage_early_setup); /* * Walk gpage list in reverse, allocating larger page sizes first. @@ -455,7 +428,7 @@ #else /* !PPC_FSL_BOOK3E */ /* Build list of addresses of gigantic pages. This function is used in early - * boot before the buddy or bootmem allocator is setup. + * boot before the buddy allocator is setup. */ void add_gpage(u64 addr, u64 page_size, unsigned long number_of_pages) { @@ -485,11 +458,6 @@ } #endif -int huge_pmd_unshare(struct mm_struct *mm, unsigned long *addr, pte_t *ptep) -{ - return 0; -} - #ifdef CONFIG_PPC_FSL_BOOK3E #define HUGEPD_FREELIST_SIZE \ ((PAGE_SIZE - sizeof(struct hugepd_freelist)) / sizeof(pte_t)) @@ -518,12 +486,13 @@ { 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; } @@ -537,6 +506,7 @@ call_rcu_sched(&(*batchp)->rcu, hugepd_free_rcu_callback); *batchp = NULL; } + put_cpu_var(hugepd_freelist_cur); } #endif @@ -571,8 +541,6 @@ for (i = 0; i < num_hugepd; i++, hpdp++) hpdp->pd = 0; - tlb->need_flush = 1; - #ifdef CONFIG_PPC_FSL_BOOK3E hugepd_free(tlb, hugepte); #else @@ -592,7 +560,7 @@ do { pmd = pmd_offset(pud, addr); next = pmd_addr_end(addr, end); - if (!is_hugepd(pmd)) { + if (!is_hugepd(__hugepd(pmd_val(*pmd)))) { /* * if it is not hugepd pointer, we should already find * it cleared. @@ -627,6 +595,7 @@ pmd = pmd_offset(pud, start); pud_clear(pud); pmd_free_tlb(tlb, pmd, start); + mm_dec_nr_pmds(tlb->mm); } static void hugetlb_free_pud_range(struct mmu_gather *tlb, pgd_t *pgd, @@ -641,7 +610,7 @@ do { pud = pud_offset(pgd, addr); next = pud_addr_end(addr, end); - if (!is_hugepd(pud)) { + if (!is_hugepd(__hugepd(pud_val(*pud)))) { if (pud_none_or_clear_bad(pud)) continue; hugetlb_free_pmd_range(tlb, pud, addr, next, floor, @@ -679,8 +648,6 @@ /* * 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, @@ -709,7 +676,7 @@ do { next = pgd_addr_end(addr, end); pgd = pgd_offset(tlb->mm, addr); - if (!is_hugepd(pgd)) { + if (!is_hugepd(__hugepd(pgd_val(*pgd)))) { if (pgd_none_or_clear_bad(pgd)) continue; hugetlb_free_pud_range(tlb, pgd, addr, next, floor, ceiling); @@ -729,25 +696,43 @@ } while (addr = next, addr != end); } +/* + * We are holding mmap_sem, so a parallel huge page collapse cannot run. + * To prevent hugepage split, disable irq. + */ struct page * follow_huge_addr(struct mm_struct *mm, unsigned long address, int write) { - pte_t *ptep; - struct page *page; + bool is_thp; + pte_t *ptep, pte; unsigned shift; - unsigned long mask; + unsigned long mask, flags; + struct page *page = ERR_PTR(-EINVAL); - ptep = find_linux_pte_or_hugepte(mm->pgd, address, &shift); - - /* Verify it is a huge page else bail. */ - if (!ptep || !shift) - return ERR_PTR(-EINVAL); + local_irq_save(flags); + ptep = find_linux_pte_or_hugepte(mm->pgd, address, &is_thp, &shift); + if (!ptep) + goto no_page; + pte = READ_ONCE(*ptep); + /* + * Verify it is a huge page else bail. + * Transparent hugepages are handled by generic code. We can skip them + * here. + */ + if (!shift || is_thp) + goto no_page; + if (!pte_present(pte)) { + page = NULL; + goto no_page; + } mask = (1UL << shift) - 1; - page = pte_page(*ptep); + page = pte_page(pte); if (page) page += (address & mask) / PAGE_SIZE; +no_page: + local_irq_restore(flags); return page; } @@ -759,67 +744,12 @@ return NULL; } -int gup_hugepte(pte_t *ptep, unsigned long sz, unsigned long addr, - unsigned long end, int write, struct page **pages, int *nr) +struct page * +follow_huge_pud(struct mm_struct *mm, unsigned long address, + pud_t *pud, int write) { - 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; + BUG(); + return NULL; } static unsigned long hugepte_addr_end(unsigned long addr, unsigned long end, @@ -829,12 +759,11 @@ return (__boundary - 1 < end - 1) ? __boundary : end; } -int gup_hugepd(hugepd_t *hugepd, unsigned pdshift, - unsigned long addr, unsigned long end, - int write, struct page **pages, int *nr) +int gup_huge_pd(hugepd_t hugepd, unsigned long addr, unsigned pdshift, + unsigned long end, int write, struct page **pages, int *nr) { pte_t *ptep; - unsigned long sz = 1UL << hugepd_shift(*hugepd); + unsigned long sz = 1UL << hugepd_shift(hugepd); unsigned long next; ptep = hugepte_offset(hugepd, addr, pdshift); @@ -899,14 +828,6 @@ if ((mmu_psize = shift_to_mmu_psize(shift)) < 0) return -EINVAL; -#ifdef CONFIG_SPU_FS_64K_LS - /* Disable support for 64K huge pages when 64K SPU local store - * support is enabled as the current implementation conflicts. - */ - if (shift == PAGE_SHIFT_64K) - return -EINVAL; -#endif /* CONFIG_SPU_FS_64K_LS */ - BUG_ON(mmu_psize_defs[mmu_psize].shift != shift); /* Return if huge page size has already been setup */ @@ -1019,7 +940,7 @@ return 0; } #endif -module_init(hugetlbpage_init); +arch_initcall(hugetlbpage_init); void flush_dcache_icache_hugepage(struct page *page) { @@ -1038,3 +959,172 @@ } } } + +#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. + * This function need to be called with interrupts disabled. We use this variant + * when we have MSR[EE] = 0 but the paca->soft_enabled = 1 + */ + +pte_t *__find_linux_pte_or_hugepte(pgd_t *pgdir, unsigned long ea, + bool *is_thp, 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; + + if (is_thp) + *is_thp = false; + + pgdp = pgdir + pgd_index(ea); + pgd = READ_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(__hugepd(pgd_val(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 = READ_ONCE(*pudp); + + if (pud_none(pud)) + return NULL; + else if (pud_huge(pud)) { + ret_pte = (pte_t *) pudp; + goto out; + } else if (is_hugepd(__hugepd(pud_val(pud)))) + hpdp = (hugepd_t *)&pud; + else { + pdshift = PMD_SHIFT; + pmdp = pmd_offset(&pud, ea); + pmd = READ_ONCE(*pmdp); + /* + * A hugepage collapse is captured by pmd_none, because + * it mark the pmd none and do a hpte invalidate. + * + * We don't worry about pmd_trans_splitting here, The + * caller if it needs to handle the splitting case + * should check for that. + */ + if (pmd_none(pmd)) + return NULL; + + if (pmd_trans_huge(pmd)) { + if (is_thp) + *is_thp = true; + ret_pte = (pte_t *) pmdp; + goto out; + } + + if (pmd_huge(pmd)) { + ret_pte = (pte_t *) pmdp; + goto out; + } else if (is_hugepd(__hugepd(pmd_val(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 = READ_ONCE(*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; +}