diff options
Diffstat (limited to 'mm/hugetlb.c')
| -rw-r--r-- | mm/hugetlb.c | 533 |
1 files changed, 220 insertions, 313 deletions
diff --git a/mm/hugetlb.c b/mm/hugetlb.c index c445e6fd85..92a2e8dcb7 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -1517,7 +1517,7 @@ static void __destroy_compound_gigantic_folio(struct folio *folio, struct page *p; atomic_set(&folio->_entire_mapcount, 0); - atomic_set(&folio->_nr_pages_mapped, 0); + atomic_set(&folio->_large_mapcount, 0); atomic_set(&folio->_pincount, 0); for (i = 1; i < nr_pages; i++) { @@ -1619,27 +1619,16 @@ static inline void destroy_compound_gigantic_folio(struct folio *folio, unsigned int order) { } #endif -static inline void __clear_hugetlb_destructor(struct hstate *h, - struct folio *folio) -{ - lockdep_assert_held(&hugetlb_lock); - - __folio_clear_hugetlb(folio); -} - /* * Remove hugetlb folio from lists. - * If vmemmap exists for the folio, update dtor so that the folio appears - * as just a compound page. Otherwise, wait until after allocating vmemmap - * to update dtor. - * - * A reference is held on the folio, except in the case of demote. + * If vmemmap exists for the folio, clear the hugetlb flag so that the + * folio appears as just a compound page. Otherwise, wait until after + * allocating vmemmap to clear the flag. * * Must be called with hugetlb lock held. */ -static void __remove_hugetlb_folio(struct hstate *h, struct folio *folio, - bool adjust_surplus, - bool demote) +static void remove_hugetlb_folio(struct hstate *h, struct folio *folio, + bool adjust_surplus) { int nid = folio_nid(folio); @@ -1653,6 +1642,7 @@ static void __remove_hugetlb_folio(struct hstate *h, struct folio *folio, list_del(&folio->lru); if (folio_test_hugetlb_freed(folio)) { + folio_clear_hugetlb_freed(folio); h->free_huge_pages--; h->free_huge_pages_node[nid]--; } @@ -1662,40 +1652,20 @@ static void __remove_hugetlb_folio(struct hstate *h, struct folio *folio, } /* - * We can only clear the hugetlb destructor after allocating vmemmap + * We can only clear the hugetlb flag after allocating vmemmap * pages. Otherwise, someone (memory error handling) may try to write * to tail struct pages. */ if (!folio_test_hugetlb_vmemmap_optimized(folio)) - __clear_hugetlb_destructor(h, folio); - - /* - * In the case of demote we do not ref count the page as it will soon - * be turned into a page of smaller size. - */ - if (!demote) - folio_ref_unfreeze(folio, 1); + __folio_clear_hugetlb(folio); h->nr_huge_pages--; h->nr_huge_pages_node[nid]--; } -static void remove_hugetlb_folio(struct hstate *h, struct folio *folio, - bool adjust_surplus) -{ - __remove_hugetlb_folio(h, folio, adjust_surplus, false); -} - -static void remove_hugetlb_folio_for_demote(struct hstate *h, struct folio *folio, - bool adjust_surplus) -{ - __remove_hugetlb_folio(h, folio, adjust_surplus, true); -} - static void add_hugetlb_folio(struct hstate *h, struct folio *folio, bool adjust_surplus) { - int zeroed; int nid = folio_nid(folio); VM_BUG_ON_FOLIO(!folio_test_hugetlb_vmemmap_optimized(folio), folio); @@ -1719,29 +1689,14 @@ static void add_hugetlb_folio(struct hstate *h, struct folio *folio, */ folio_set_hugetlb_vmemmap_optimized(folio); - /* - * This folio is about to be managed by the hugetlb allocator and - * should have no users. Drop our reference, and check for others - * just in case. - */ - zeroed = folio_put_testzero(folio); - if (unlikely(!zeroed)) - /* - * It is VERY unlikely soneone else has taken a ref - * on the folio. In this case, we simply return as - * free_huge_folio() will be called when this other ref - * is dropped. - */ - return; - - arch_clear_hugepage_flags(&folio->page); + arch_clear_hugetlb_flags(folio); enqueue_hugetlb_folio(h, folio); } static void __update_and_free_hugetlb_folio(struct hstate *h, struct folio *folio) { - bool clear_dtor = folio_test_hugetlb_vmemmap_optimized(folio); + bool clear_flag = folio_test_hugetlb_vmemmap_optimized(folio); if (hstate_is_gigantic(h) && !gigantic_page_runtime_supported()) return; @@ -1754,11 +1709,11 @@ static void __update_and_free_hugetlb_folio(struct hstate *h, return; /* - * If folio is not vmemmap optimized (!clear_dtor), then the folio + * If folio is not vmemmap optimized (!clear_flag), then the folio * is no longer identified as a hugetlb page. hugetlb_vmemmap_restore_folio * can only be passed hugetlb pages and will BUG otherwise. */ - if (clear_dtor && hugetlb_vmemmap_restore_folio(h, folio)) { + if (clear_flag && hugetlb_vmemmap_restore_folio(h, folio)) { spin_lock_irq(&hugetlb_lock); /* * If we cannot allocate vmemmap pages, just refuse to free the @@ -1771,23 +1726,25 @@ static void __update_and_free_hugetlb_folio(struct hstate *h, } /* - * Move PageHWPoison flag from head page to the raw error pages, - * which makes any healthy subpages reusable. - */ - if (unlikely(folio_test_hwpoison(folio))) - folio_clear_hugetlb_hwpoison(folio); - - /* * If vmemmap pages were allocated above, then we need to clear the - * hugetlb destructor under the hugetlb lock. + * hugetlb flag under the hugetlb lock. */ if (folio_test_hugetlb(folio)) { spin_lock_irq(&hugetlb_lock); - __clear_hugetlb_destructor(h, folio); + __folio_clear_hugetlb(folio); spin_unlock_irq(&hugetlb_lock); } /* + * Move PageHWPoison flag from head page to the raw error pages, + * which makes any healthy subpages reusable. + */ + if (unlikely(folio_test_hwpoison(folio))) + folio_clear_hugetlb_hwpoison(folio); + + folio_ref_unfreeze(folio, 1); + + /* * Non-gigantic pages demoted from CMA allocated gigantic pages * need to be given back to CMA in free_gigantic_folio. */ @@ -1796,7 +1753,8 @@ static void __update_and_free_hugetlb_folio(struct hstate *h, destroy_compound_gigantic_folio(folio, huge_page_order(h)); free_gigantic_folio(folio, huge_page_order(h)); } else { - __free_pages(&folio->page, huge_page_order(h)); + INIT_LIST_HEAD(&folio->_deferred_list); + folio_put(folio); } } @@ -1884,7 +1842,7 @@ static void bulk_vmemmap_restore_error(struct hstate *h, list_for_each_entry_safe(folio, t_folio, non_hvo_folios, lru) { list_del(&folio->lru); spin_lock_irq(&hugetlb_lock); - __clear_hugetlb_destructor(h, folio); + __folio_clear_hugetlb(folio); spin_unlock_irq(&hugetlb_lock); update_and_free_hugetlb_folio(h, folio, false); cond_resched(); @@ -1909,7 +1867,7 @@ static void bulk_vmemmap_restore_error(struct hstate *h, } else { list_del(&folio->lru); spin_lock_irq(&hugetlb_lock); - __clear_hugetlb_destructor(h, folio); + __folio_clear_hugetlb(folio); spin_unlock_irq(&hugetlb_lock); update_and_free_hugetlb_folio(h, folio, false); cond_resched(); @@ -1942,14 +1900,14 @@ retry: * should only be pages on the non_hvo_folios list. * Do note that the non_hvo_folios list could be empty. * Without HVO enabled, ret will be 0 and there is no need to call - * __clear_hugetlb_destructor as this was done previously. + * __folio_clear_hugetlb as this was done previously. */ VM_WARN_ON(!list_empty(folio_list)); VM_WARN_ON(ret < 0); if (!list_empty(&non_hvo_folios) && ret) { spin_lock_irq(&hugetlb_lock); list_for_each_entry(folio, &non_hvo_folios, lru) - __clear_hugetlb_destructor(h, folio); + __folio_clear_hugetlb(folio); spin_unlock_irq(&hugetlb_lock); } @@ -1974,7 +1932,7 @@ void free_huge_folio(struct folio *folio) { /* * Can't pass hstate in here because it is called from the - * compound page destructor. + * generic mm code. */ struct hstate *h = folio_hstate(folio); int nid = folio_nid(folio); @@ -2031,7 +1989,7 @@ void free_huge_folio(struct folio *folio) spin_unlock_irqrestore(&hugetlb_lock, flags); update_and_free_hugetlb_folio(h, folio, true); } else { - arch_clear_hugepage_flags(&folio->page); + arch_clear_hugetlb_flags(folio); enqueue_hugetlb_folio(h, folio); spin_unlock_irqrestore(&hugetlb_lock, flags); } @@ -2124,10 +2082,10 @@ static bool __prep_compound_gigantic_folio(struct folio *folio, set_compound_head(p, &folio->page); } __folio_set_head(folio); - /* we rely on prep_new_hugetlb_folio to set the destructor */ + /* we rely on prep_new_hugetlb_folio to set the hugetlb flag */ folio_set_order(folio, order); atomic_set(&folio->_entire_mapcount, -1); - atomic_set(&folio->_nr_pages_mapped, 0); + atomic_set(&folio->_large_mapcount, -1); atomic_set(&folio->_pincount, 0); return true; @@ -2162,13 +2120,13 @@ static bool prep_compound_gigantic_folio_for_demote(struct folio *folio, /* * Find and lock address space (mapping) in write mode. * - * Upon entry, the page is locked which means that page_mapping() is + * Upon entry, the folio is locked which means that folio_mapping() is * stable. Due to locking order, we can only trylock_write. If we can * not get the lock, simply return NULL to caller. */ -struct address_space *hugetlb_page_mapping_lock_write(struct page *hpage) +struct address_space *hugetlb_folio_mapping_lock_write(struct folio *folio) { - struct address_space *mapping = page_mapping(hpage); + struct address_space *mapping = folio_mapping(folio); if (!mapping) return mapping; @@ -2184,13 +2142,13 @@ static struct folio *alloc_buddy_hugetlb_folio(struct hstate *h, nodemask_t *node_alloc_noretry) { int order = huge_page_order(h); - struct page *page; + struct folio *folio; bool alloc_try_hard = true; bool retry = true; /* - * By default we always try hard to allocate the page with - * __GFP_RETRY_MAYFAIL flag. However, if we are allocating pages in + * By default we always try hard to allocate the folio with + * __GFP_RETRY_MAYFAIL flag. However, if we are allocating folios in * a loop (to adjust global huge page counts) and previous allocation * failed, do not continue to try hard on the same node. Use the * node_alloc_noretry bitmap to manage this state information. @@ -2203,43 +2161,45 @@ static struct folio *alloc_buddy_hugetlb_folio(struct hstate *h, if (nid == NUMA_NO_NODE) nid = numa_mem_id(); retry: - page = __alloc_pages(gfp_mask, order, nid, nmask); + folio = __folio_alloc(gfp_mask, order, nid, nmask); + /* Ensure hugetlb folio won't have large_rmappable flag set. */ + if (folio) + folio_clear_large_rmappable(folio); - /* Freeze head page */ - if (page && !page_ref_freeze(page, 1)) { - __free_pages(page, order); + if (folio && !folio_ref_freeze(folio, 1)) { + folio_put(folio); if (retry) { /* retry once */ retry = false; goto retry; } /* WOW! twice in a row. */ - pr_warn("HugeTLB head page unexpected inflated ref count\n"); - page = NULL; + pr_warn("HugeTLB unexpected inflated folio ref count\n"); + folio = NULL; } /* - * If we did not specify __GFP_RETRY_MAYFAIL, but still got a page this - * indicates an overall state change. Clear bit so that we resume - * normal 'try hard' allocations. + * If we did not specify __GFP_RETRY_MAYFAIL, but still got a + * folio this indicates an overall state change. Clear bit so + * that we resume normal 'try hard' allocations. */ - if (node_alloc_noretry && page && !alloc_try_hard) + if (node_alloc_noretry && folio && !alloc_try_hard) node_clear(nid, *node_alloc_noretry); /* - * If we tried hard to get a page but failed, set bit so that + * If we tried hard to get a folio but failed, set bit so that * subsequent attempts will not try as hard until there is an * overall state change. */ - if (node_alloc_noretry && !page && alloc_try_hard) + if (node_alloc_noretry && !folio && alloc_try_hard) node_set(nid, *node_alloc_noretry); - if (!page) { + if (!folio) { __count_vm_event(HTLB_BUDDY_PGALLOC_FAIL); return NULL; } __count_vm_event(HTLB_BUDDY_PGALLOC); - return page_folio(page); + return folio; } static struct folio *__alloc_fresh_hugetlb_folio(struct hstate *h, @@ -2385,8 +2345,8 @@ static struct folio *remove_pool_hugetlb_folio(struct hstate *h, } /* - * Dissolve a given free hugepage into free buddy pages. This function does - * nothing for in-use hugepages and non-hugepages. + * Dissolve a given free hugetlb folio into free buddy pages. This function + * does nothing for in-use hugetlb folios and non-hugetlb folios. * This function returns values like below: * * -ENOMEM: failed to allocate vmemmap pages to free the freed hugepages @@ -2398,10 +2358,9 @@ static struct folio *remove_pool_hugetlb_folio(struct hstate *h, * 0: successfully dissolved free hugepages or the page is not a * hugepage (considered as already dissolved) */ -int dissolve_free_huge_page(struct page *page) +int dissolve_free_hugetlb_folio(struct folio *folio) { int rc = -EBUSY; - struct folio *folio = page_folio(page); retry: /* Not to disrupt normal path by vainly holding hugetlb_lock */ @@ -2478,13 +2437,13 @@ out: * make specified memory blocks removable from the system. * Note that this will dissolve a free gigantic hugepage completely, if any * part of it lies within the given range. - * Also note that if dissolve_free_huge_page() returns with an error, all - * free hugepages that were dissolved before that error are lost. + * Also note that if dissolve_free_hugetlb_folio() returns with an error, all + * free hugetlb folios that were dissolved before that error are lost. */ -int dissolve_free_huge_pages(unsigned long start_pfn, unsigned long end_pfn) +int dissolve_free_hugetlb_folios(unsigned long start_pfn, unsigned long end_pfn) { unsigned long pfn; - struct page *page; + struct folio *folio; int rc = 0; unsigned int order; struct hstate *h; @@ -2497,8 +2456,8 @@ int dissolve_free_huge_pages(unsigned long start_pfn, unsigned long end_pfn) order = min(order, huge_page_order(h)); for (pfn = start_pfn; pfn < end_pfn; pfn += 1 << order) { - page = pfn_to_page(pfn); - rc = dissolve_free_huge_page(page); + folio = pfn_folio(pfn); + rc = dissolve_free_hugetlb_folio(folio); if (rc) break; } @@ -2605,7 +2564,7 @@ struct folio *alloc_buddy_hugetlb_folio_with_mpol(struct hstate *h, /* folio migration callback function */ struct folio *alloc_hugetlb_folio_nodemask(struct hstate *h, int preferred_nid, - nodemask_t *nmask, gfp_t gfp_mask) + nodemask_t *nmask, gfp_t gfp_mask, bool allow_alloc_fallback) { spin_lock_irq(&hugetlb_lock); if (available_huge_pages(h)) { @@ -2620,9 +2579,30 @@ struct folio *alloc_hugetlb_folio_nodemask(struct hstate *h, int preferred_nid, } spin_unlock_irq(&hugetlb_lock); + /* We cannot fallback to other nodes, as we could break the per-node pool. */ + if (!allow_alloc_fallback) + gfp_mask |= __GFP_THISNODE; + return alloc_migrate_hugetlb_folio(h, gfp_mask, preferred_nid, nmask); } +static nodemask_t *policy_mbind_nodemask(gfp_t gfp) +{ +#ifdef CONFIG_NUMA + struct mempolicy *mpol = get_task_policy(current); + + /* + * Only enforce MPOL_BIND policy which overlaps with cpuset policy + * (from policy_nodemask) specifically for hugetlb case + */ + if (mpol->mode == MPOL_BIND && + (apply_policy_zone(mpol, gfp_zone(gfp)) && + cpuset_nodemask_valid_mems_allowed(&mpol->nodes))) + return &mpol->nodes; +#endif + return NULL; +} + /* * Increase the hugetlb pool such that it can accommodate a reservation * of size 'delta'. @@ -2636,6 +2616,8 @@ static int gather_surplus_pages(struct hstate *h, long delta) long i; long needed, allocated; bool alloc_ok = true; + int node; + nodemask_t *mbind_nodemask = policy_mbind_nodemask(htlb_alloc_mask(h)); lockdep_assert_held(&hugetlb_lock); needed = (h->resv_huge_pages + delta) - h->free_huge_pages; @@ -2650,8 +2632,15 @@ static int gather_surplus_pages(struct hstate *h, long delta) retry: spin_unlock_irq(&hugetlb_lock); for (i = 0; i < needed; i++) { - folio = alloc_surplus_hugetlb_folio(h, htlb_alloc_mask(h), - NUMA_NO_NODE, NULL); + folio = NULL; + for_each_node_mask(node, cpuset_current_mems_allowed) { + if (!mbind_nodemask || node_isset(node, *mbind_nodemask)) { + folio = alloc_surplus_hugetlb_folio(h, htlb_alloc_mask(h), + node, NULL); + if (folio) + break; + } + } if (!folio) { alloc_ok = false; break; @@ -3084,11 +3073,8 @@ retry: free_new: spin_unlock_irq(&hugetlb_lock); - if (new_folio) { - /* Folio has a zero ref count, but needs a ref to be freed */ - folio_ref_unfreeze(new_folio, 1); + if (new_folio) update_and_free_hugetlb_folio(h, new_folio, false); - } return ret; } @@ -3943,7 +3929,7 @@ static int demote_free_hugetlb_folio(struct hstate *h, struct folio *folio) target_hstate = size_to_hstate(PAGE_SIZE << h->demote_order); - remove_hugetlb_folio_for_demote(h, folio, false); + remove_hugetlb_folio(h, folio, false); spin_unlock_irq(&hugetlb_lock); /* @@ -3957,7 +3943,6 @@ static int demote_free_hugetlb_folio(struct hstate *h, struct folio *folio) if (rc) { /* Allocation of vmemmmap failed, we can not demote folio */ spin_lock_irq(&hugetlb_lock); - folio_ref_unfreeze(folio, 1); add_hugetlb_folio(h, folio, false); return rc; } @@ -4658,7 +4643,7 @@ void __init hugetlb_add_hstate(unsigned int order) BUG_ON(hugetlb_max_hstate >= HUGE_MAX_HSTATE); BUG_ON(order < order_base_2(__NR_USED_SUBPAGE)); h = &hstates[hugetlb_max_hstate++]; - mutex_init(&h->resize_lock); + __mutex_init(&h->resize_lock, "resize mutex", &h->resize_key); h->order = order; h->mask = ~(huge_page_size(h) - 1); for (i = 0; i < MAX_NUMNODES; ++i) @@ -4881,23 +4866,6 @@ static int __init default_hugepagesz_setup(char *s) } __setup("default_hugepagesz=", default_hugepagesz_setup); -static nodemask_t *policy_mbind_nodemask(gfp_t gfp) -{ -#ifdef CONFIG_NUMA - struct mempolicy *mpol = get_task_policy(current); - - /* - * Only enforce MPOL_BIND policy which overlaps with cpuset policy - * (from policy_nodemask) specifically for hugetlb case - */ - if (mpol->mode == MPOL_BIND && - (apply_policy_zone(mpol, gfp_zone(gfp)) && - cpuset_nodemask_valid_mems_allowed(&mpol->nodes))) - return &mpol->nodes; -#endif - return NULL; -} - static unsigned int allowed_mems_nr(struct hstate *h) { int node; @@ -5032,7 +5000,6 @@ static struct ctl_table hugetlb_table[] = { .mode = 0644, .proc_handler = hugetlb_overcommit_handler, }, - { } }; static void hugetlb_sysctl_init(void) @@ -5935,19 +5902,18 @@ static void unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma, * cannot race with other handlers or page migration. * Keep the pte_same checks anyway to make transition from the mutex easier. */ -static vm_fault_t hugetlb_wp(struct mm_struct *mm, struct vm_area_struct *vma, - unsigned long address, pte_t *ptep, unsigned int flags, - struct folio *pagecache_folio, spinlock_t *ptl, +static vm_fault_t hugetlb_wp(struct folio *pagecache_folio, struct vm_fault *vmf) { - const bool unshare = flags & FAULT_FLAG_UNSHARE; - pte_t pte = huge_ptep_get(ptep); + struct vm_area_struct *vma = vmf->vma; + struct mm_struct *mm = vma->vm_mm; + const bool unshare = vmf->flags & FAULT_FLAG_UNSHARE; + pte_t pte = huge_ptep_get(vmf->pte); struct hstate *h = hstate_vma(vma); struct folio *old_folio; struct folio *new_folio; int outside_reserve = 0; vm_fault_t ret = 0; - unsigned long haddr = address & huge_page_mask(h); struct mmu_notifier_range range; /* @@ -5970,7 +5936,7 @@ static vm_fault_t hugetlb_wp(struct mm_struct *mm, struct vm_area_struct *vma, /* Let's take out MAP_SHARED mappings first. */ if (vma->vm_flags & VM_MAYSHARE) { - set_huge_ptep_writable(vma, haddr, ptep); + set_huge_ptep_writable(vma, vmf->address, vmf->pte); return 0; } @@ -5982,6 +5948,13 @@ retry_avoidcopy: /* * If no-one else is actually using this page, we're the exclusive * owner and can reuse this page. + * + * Note that we don't rely on the (safer) folio refcount here, because + * copying the hugetlb folio when there are unexpected (temporary) + * folio references could harm simple fork()+exit() users when + * we run out of free hugetlb folios: we would have to kill processes + * in scenarios that used to work. As a side effect, there can still + * be leaks between processes, for example, with FOLL_GET users. */ if (folio_mapcount(old_folio) == 1 && folio_test_anon(old_folio)) { if (!PageAnonExclusive(&old_folio->page)) { @@ -5989,7 +5962,7 @@ retry_avoidcopy: SetPageAnonExclusive(&old_folio->page); } if (likely(!unshare)) - set_huge_ptep_writable(vma, haddr, ptep); + set_huge_ptep_writable(vma, vmf->address, vmf->pte); delayacct_wpcopy_end(); return 0; @@ -6016,8 +5989,8 @@ retry_avoidcopy: * Drop page table lock as buddy allocator may be called. It will * be acquired again before returning to the caller, as expected. */ - spin_unlock(ptl); - new_folio = alloc_hugetlb_folio(vma, haddr, outside_reserve); + spin_unlock(vmf->ptl); + new_folio = alloc_hugetlb_folio(vma, vmf->address, outside_reserve); if (IS_ERR(new_folio)) { /* @@ -6042,19 +6015,21 @@ retry_avoidcopy: * * Reacquire both after unmap operation. */ - idx = vma_hugecache_offset(h, vma, haddr); + idx = vma_hugecache_offset(h, vma, vmf->address); hash = hugetlb_fault_mutex_hash(mapping, idx); hugetlb_vma_unlock_read(vma); mutex_unlock(&hugetlb_fault_mutex_table[hash]); - unmap_ref_private(mm, vma, &old_folio->page, haddr); + unmap_ref_private(mm, vma, &old_folio->page, + vmf->address); mutex_lock(&hugetlb_fault_mutex_table[hash]); hugetlb_vma_lock_read(vma); - spin_lock(ptl); - ptep = hugetlb_walk(vma, haddr, huge_page_size(h)); - if (likely(ptep && - pte_same(huge_ptep_get(ptep), pte))) + spin_lock(vmf->ptl); + vmf->pte = hugetlb_walk(vma, vmf->address, + huge_page_size(h)); + if (likely(vmf->pte && + pte_same(huge_ptep_get(vmf->pte), pte))) goto retry_avoidcopy; /* * race occurs while re-acquiring page table @@ -6076,37 +6051,38 @@ retry_avoidcopy: if (unlikely(ret)) goto out_release_all; - if (copy_user_large_folio(new_folio, old_folio, address, vma)) { - ret = VM_FAULT_HWPOISON_LARGE; + if (copy_user_large_folio(new_folio, old_folio, vmf->real_address, vma)) { + ret = VM_FAULT_HWPOISON_LARGE | VM_FAULT_SET_HINDEX(hstate_index(h)); goto out_release_all; } __folio_mark_uptodate(new_folio); - mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, mm, haddr, - haddr + huge_page_size(h)); + mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, mm, vmf->address, + vmf->address + huge_page_size(h)); mmu_notifier_invalidate_range_start(&range); /* * Retake the page table lock to check for racing updates * before the page tables are altered */ - spin_lock(ptl); - ptep = hugetlb_walk(vma, haddr, huge_page_size(h)); - if (likely(ptep && pte_same(huge_ptep_get(ptep), pte))) { + spin_lock(vmf->ptl); + vmf->pte = hugetlb_walk(vma, vmf->address, huge_page_size(h)); + if (likely(vmf->pte && pte_same(huge_ptep_get(vmf->pte), pte))) { pte_t newpte = make_huge_pte(vma, &new_folio->page, !unshare); /* Break COW or unshare */ - huge_ptep_clear_flush(vma, haddr, ptep); + huge_ptep_clear_flush(vma, vmf->address, vmf->pte); hugetlb_remove_rmap(old_folio); - hugetlb_add_new_anon_rmap(new_folio, vma, haddr); + hugetlb_add_new_anon_rmap(new_folio, vma, vmf->address); if (huge_pte_uffd_wp(pte)) newpte = huge_pte_mkuffd_wp(newpte); - set_huge_pte_at(mm, haddr, ptep, newpte, huge_page_size(h)); + set_huge_pte_at(mm, vmf->address, vmf->pte, newpte, + huge_page_size(h)); folio_set_hugetlb_migratable(new_folio); /* Make the old page be freed below */ new_folio = old_folio; } - spin_unlock(ptl); + spin_unlock(vmf->ptl); mmu_notifier_invalidate_range_end(&range); out_release_all: /* @@ -6114,12 +6090,12 @@ out_release_all: * unshare) */ if (new_folio != old_folio) - restore_reserve_on_error(h, vma, haddr, new_folio); + restore_reserve_on_error(h, vma, vmf->address, new_folio); folio_put(new_folio); out_release_old: folio_put(old_folio); - spin_lock(ptl); /* Caller expects lock to be held */ + spin_lock(vmf->ptl); /* Caller expects lock to be held */ delayacct_wpcopy_end(); return ret; @@ -6128,8 +6104,8 @@ out_release_old: /* * Return whether there is a pagecache page to back given address within VMA. */ -static bool hugetlbfs_pagecache_present(struct hstate *h, - struct vm_area_struct *vma, unsigned long address) +bool hugetlbfs_pagecache_present(struct hstate *h, + struct vm_area_struct *vma, unsigned long address) { struct address_space *mapping = vma->vm_file->f_mapping; pgoff_t idx = linear_page_index(vma, address); @@ -6205,23 +6181,19 @@ static bool hugetlb_pte_stable(struct hstate *h, struct mm_struct *mm, return same; } -static vm_fault_t hugetlb_no_page(struct mm_struct *mm, - struct vm_area_struct *vma, - struct address_space *mapping, pgoff_t idx, - unsigned long address, pte_t *ptep, - pte_t old_pte, unsigned int flags, +static vm_fault_t hugetlb_no_page(struct address_space *mapping, struct vm_fault *vmf) { + struct vm_area_struct *vma = vmf->vma; + struct mm_struct *mm = vma->vm_mm; struct hstate *h = hstate_vma(vma); vm_fault_t ret = VM_FAULT_SIGBUS; int anon_rmap = 0; unsigned long size; struct folio *folio; pte_t new_pte; - spinlock_t *ptl; - unsigned long haddr = address & huge_page_mask(h); bool new_folio, new_pagecache_folio = false; - u32 hash = hugetlb_fault_mutex_hash(mapping, idx); + u32 hash = hugetlb_fault_mutex_hash(mapping, vmf->pgoff); /* * Currently, we are forced to kill the process in the event the @@ -6240,10 +6212,10 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm, * before we get page_table_lock. */ new_folio = false; - folio = filemap_lock_hugetlb_folio(h, mapping, idx); + folio = filemap_lock_hugetlb_folio(h, mapping, vmf->pgoff); if (IS_ERR(folio)) { size = i_size_read(mapping->host) >> huge_page_shift(h); - if (idx >= size) + if (vmf->pgoff >= size) goto out; /* Check for page in userfault range */ if (userfaultfd_missing(vma)) { @@ -6264,7 +6236,7 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm, * never happen on the page after UFFDIO_COPY has * correctly installed the page and returned. */ - if (!hugetlb_pte_stable(h, mm, ptep, old_pte)) { + if (!hugetlb_pte_stable(h, mm, vmf->pte, vmf->orig_pte)) { ret = 0; goto out; } @@ -6279,7 +6251,7 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm, goto out; } - folio = alloc_hugetlb_folio(vma, haddr, 0); + folio = alloc_hugetlb_folio(vma, vmf->address, 0); if (IS_ERR(folio)) { /* * Returning error will result in faulting task being @@ -6293,18 +6265,20 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm, * here. Before returning error, get ptl and make * sure there really is no pte entry. */ - if (hugetlb_pte_stable(h, mm, ptep, old_pte)) + if (hugetlb_pte_stable(h, mm, vmf->pte, vmf->orig_pte)) ret = vmf_error(PTR_ERR(folio)); else ret = 0; goto out; } - clear_huge_page(&folio->page, address, pages_per_huge_page(h)); + clear_huge_page(&folio->page, vmf->real_address, + pages_per_huge_page(h)); __folio_mark_uptodate(folio); new_folio = true; if (vma->vm_flags & VM_MAYSHARE) { - int err = hugetlb_add_to_page_cache(folio, mapping, idx); + int err = hugetlb_add_to_page_cache(folio, mapping, + vmf->pgoff); if (err) { /* * err can't be -EEXIST which implies someone @@ -6313,7 +6287,8 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm, * to the page cache. So it's safe to call * restore_reserve_on_error() here. */ - restore_reserve_on_error(h, vma, haddr, folio); + restore_reserve_on_error(h, vma, vmf->address, + folio); folio_put(folio); ret = VM_FAULT_SIGBUS; goto out; @@ -6340,7 +6315,7 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm, folio_unlock(folio); folio_put(folio); /* See comment in userfaultfd_missing() block above */ - if (!hugetlb_pte_stable(h, mm, ptep, old_pte)) { + if (!hugetlb_pte_stable(h, mm, vmf->pte, vmf->orig_pte)) { ret = 0; goto out; } @@ -6355,23 +6330,23 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm, * any allocations necessary to record that reservation occur outside * the spinlock. */ - if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) { - if (vma_needs_reservation(h, vma, haddr) < 0) { + if ((vmf->flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) { + if (vma_needs_reservation(h, vma, vmf->address) < 0) { ret = VM_FAULT_OOM; goto backout_unlocked; } /* Just decrements count, does not deallocate */ - vma_end_reservation(h, vma, haddr); + vma_end_reservation(h, vma, vmf->address); } - ptl = huge_pte_lock(h, mm, ptep); + vmf->ptl = huge_pte_lock(h, mm, vmf->pte); ret = 0; /* If pte changed from under us, retry */ - if (!pte_same(huge_ptep_get(ptep), old_pte)) + if (!pte_same(huge_ptep_get(vmf->pte), vmf->orig_pte)) goto backout; if (anon_rmap) - hugetlb_add_new_anon_rmap(folio, vma, haddr); + hugetlb_add_new_anon_rmap(folio, vma, vmf->address); else hugetlb_add_file_rmap(folio); new_pte = make_huge_pte(vma, &folio->page, ((vma->vm_flags & VM_WRITE) @@ -6380,17 +6355,17 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm, * If this pte was previously wr-protected, keep it wr-protected even * if populated. */ - if (unlikely(pte_marker_uffd_wp(old_pte))) + if (unlikely(pte_marker_uffd_wp(vmf->orig_pte))) new_pte = huge_pte_mkuffd_wp(new_pte); - set_huge_pte_at(mm, haddr, ptep, new_pte, huge_page_size(h)); + set_huge_pte_at(mm, vmf->address, vmf->pte, new_pte, huge_page_size(h)); hugetlb_count_add(pages_per_huge_page(h), mm); - if ((flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) { + if ((vmf->flags & FAULT_FLAG_WRITE) && !(vma->vm_flags & VM_SHARED)) { /* Optimization, do the COW without a second fault */ - ret = hugetlb_wp(mm, vma, address, ptep, flags, folio, ptl, vmf); + ret = hugetlb_wp(folio, vmf); } - spin_unlock(ptl); + spin_unlock(vmf->ptl); /* * Only set hugetlb_migratable in newly allocated pages. Existing pages @@ -6407,10 +6382,10 @@ out: return ret; backout: - spin_unlock(ptl); + spin_unlock(vmf->ptl); backout_unlocked: if (new_folio && !new_pagecache_folio) - restore_reserve_on_error(h, vma, haddr, folio); + restore_reserve_on_error(h, vma, vmf->address, folio); folio_unlock(folio); folio_put(folio); @@ -6444,8 +6419,6 @@ u32 hugetlb_fault_mutex_hash(struct address_space *mapping, pgoff_t idx) vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long address, unsigned int flags) { - pte_t *ptep, entry; - spinlock_t *ptl; vm_fault_t ret; u32 hash; struct folio *folio = NULL; @@ -6453,13 +6426,13 @@ vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, struct hstate *h = hstate_vma(vma); struct address_space *mapping; int need_wait_lock = 0; - unsigned long haddr = address & huge_page_mask(h); struct vm_fault vmf = { .vma = vma, - .address = haddr, + .address = address & huge_page_mask(h), .real_address = address, .flags = flags, - .pgoff = vma_hugecache_offset(h, vma, haddr), + .pgoff = vma_hugecache_offset(h, vma, + address & huge_page_mask(h)), /* TODO: Track hugetlb faults using vm_fault */ /* @@ -6479,25 +6452,26 @@ vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, /* * Acquire vma lock before calling huge_pte_alloc and hold - * until finished with ptep. This prevents huge_pmd_unshare from - * being called elsewhere and making the ptep no longer valid. + * until finished with vmf.pte. This prevents huge_pmd_unshare from + * being called elsewhere and making the vmf.pte no longer valid. */ hugetlb_vma_lock_read(vma); - ptep = huge_pte_alloc(mm, vma, haddr, huge_page_size(h)); - if (!ptep) { + vmf.pte = huge_pte_alloc(mm, vma, vmf.address, huge_page_size(h)); + if (!vmf.pte) { hugetlb_vma_unlock_read(vma); mutex_unlock(&hugetlb_fault_mutex_table[hash]); return VM_FAULT_OOM; } - entry = huge_ptep_get(ptep); - if (huge_pte_none_mostly(entry)) { - if (is_pte_marker(entry)) { + vmf.orig_pte = huge_ptep_get(vmf.pte); + if (huge_pte_none_mostly(vmf.orig_pte)) { + if (is_pte_marker(vmf.orig_pte)) { pte_marker marker = - pte_marker_get(pte_to_swp_entry(entry)); + pte_marker_get(pte_to_swp_entry(vmf.orig_pte)); if (marker & PTE_MARKER_POISONED) { - ret = VM_FAULT_HWPOISON_LARGE; + ret = VM_FAULT_HWPOISON_LARGE | + VM_FAULT_SET_HINDEX(hstate_index(h)); goto out_mutex; } } @@ -6508,21 +6482,20 @@ vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, * hugetlb_no_page will drop vma lock and hugetlb fault * mutex internally, which make us return immediately. */ - return hugetlb_no_page(mm, vma, mapping, vmf.pgoff, address, - ptep, entry, flags, &vmf); + return hugetlb_no_page(mapping, &vmf); } ret = 0; /* - * entry could be a migration/hwpoison entry at this point, so this - * check prevents the kernel from going below assuming that we have - * an active hugepage in pagecache. This goto expects the 2nd page - * fault, and is_hugetlb_entry_(migration|hwpoisoned) check will - * properly handle it. + * vmf.orig_pte could be a migration/hwpoison vmf.orig_pte at this + * point, so this check prevents the kernel from going below assuming + * that we have an active hugepage in pagecache. This goto expects + * the 2nd page fault, and is_hugetlb_entry_(migration|hwpoisoned) + * check will properly handle it. */ - if (!pte_present(entry)) { - if (unlikely(is_hugetlb_entry_migration(entry))) { + if (!pte_present(vmf.orig_pte)) { + if (unlikely(is_hugetlb_entry_migration(vmf.orig_pte))) { /* * Release the hugetlb fault lock now, but retain * the vma lock, because it is needed to guard the @@ -6531,9 +6504,9 @@ vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, * be released there. */ mutex_unlock(&hugetlb_fault_mutex_table[hash]); - migration_entry_wait_huge(vma, ptep); + migration_entry_wait_huge(vma, vmf.pte); return 0; - } else if (unlikely(is_hugetlb_entry_hwpoisoned(entry))) + } else if (unlikely(is_hugetlb_entry_hwpoisoned(vmf.orig_pte))) ret = VM_FAULT_HWPOISON_LARGE | VM_FAULT_SET_HINDEX(hstate_index(h)); goto out_mutex; @@ -6547,13 +6520,13 @@ vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, * determine if a reservation has been consumed. */ if ((flags & (FAULT_FLAG_WRITE|FAULT_FLAG_UNSHARE)) && - !(vma->vm_flags & VM_MAYSHARE) && !huge_pte_write(entry)) { - if (vma_needs_reservation(h, vma, haddr) < 0) { + !(vma->vm_flags & VM_MAYSHARE) && !huge_pte_write(vmf.orig_pte)) { + if (vma_needs_reservation(h, vma, vmf.address) < 0) { ret = VM_FAULT_OOM; goto out_mutex; } /* Just decrements count, does not deallocate */ - vma_end_reservation(h, vma, haddr); + vma_end_reservation(h, vma, vmf.address); pagecache_folio = filemap_lock_hugetlb_folio(h, mapping, vmf.pgoff); @@ -6561,17 +6534,17 @@ vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, pagecache_folio = NULL; } - ptl = huge_pte_lock(h, mm, ptep); + vmf.ptl = huge_pte_lock(h, mm, vmf.pte); /* Check for a racing update before calling hugetlb_wp() */ - if (unlikely(!pte_same(entry, huge_ptep_get(ptep)))) + if (unlikely(!pte_same(vmf.orig_pte, huge_ptep_get(vmf.pte)))) goto out_ptl; /* Handle userfault-wp first, before trying to lock more pages */ - if (userfaultfd_wp(vma) && huge_pte_uffd_wp(huge_ptep_get(ptep)) && - (flags & FAULT_FLAG_WRITE) && !huge_pte_write(entry)) { + if (userfaultfd_wp(vma) && huge_pte_uffd_wp(huge_ptep_get(vmf.pte)) && + (flags & FAULT_FLAG_WRITE) && !huge_pte_write(vmf.orig_pte)) { if (!userfaultfd_wp_async(vma)) { - spin_unlock(ptl); + spin_unlock(vmf.ptl); if (pagecache_folio) { folio_unlock(pagecache_folio); folio_put(pagecache_folio); @@ -6581,18 +6554,18 @@ vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, return handle_userfault(&vmf, VM_UFFD_WP); } - entry = huge_pte_clear_uffd_wp(entry); - set_huge_pte_at(mm, haddr, ptep, entry, + vmf.orig_pte = huge_pte_clear_uffd_wp(vmf.orig_pte); + set_huge_pte_at(mm, vmf.address, vmf.pte, vmf.orig_pte, huge_page_size(hstate_vma(vma))); /* Fallthrough to CoW */ } /* - * hugetlb_wp() requires page locks of pte_page(entry) and + * hugetlb_wp() requires page locks of pte_page(vmf.orig_pte) and * pagecache_folio, so here we need take the former one * when folio != pagecache_folio or !pagecache_folio. */ - folio = page_folio(pte_page(entry)); + folio = page_folio(pte_page(vmf.orig_pte)); if (folio != pagecache_folio) if (!folio_trylock(folio)) { need_wait_lock = 1; @@ -6602,24 +6575,23 @@ vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma, folio_get(folio); if (flags & (FAULT_FLAG_WRITE|FAULT_FLAG_UNSHARE)) { - if (!huge_pte_write(entry)) { - ret = hugetlb_wp(mm, vma, address, ptep, flags, - pagecache_folio, ptl, &vmf); + if (!huge_pte_write(vmf.orig_pte)) { + ret = hugetlb_wp(pagecache_folio, &vmf); goto out_put_page; } else if (likely(flags & FAULT_FLAG_WRITE)) { - entry = huge_pte_mkdirty(entry); + vmf.orig_pte = huge_pte_mkdirty(vmf.orig_pte); } } - entry = pte_mkyoung(entry); - if (huge_ptep_set_access_flags(vma, haddr, ptep, entry, + vmf.orig_pte = pte_mkyoung(vmf.orig_pte); + if (huge_ptep_set_access_flags(vma, vmf.address, vmf.pte, vmf.orig_pte, flags & FAULT_FLAG_WRITE)) - update_mmu_cache(vma, haddr, ptep); + update_mmu_cache(vma, vmf.address, vmf.pte); out_put_page: if (folio != pagecache_folio) folio_unlock(folio); folio_put(folio); out_ptl: - spin_unlock(ptl); + spin_unlock(vmf.ptl); if (pagecache_folio) { folio_unlock(pagecache_folio); @@ -6655,7 +6627,13 @@ static struct folio *alloc_hugetlb_folio_vma(struct hstate *h, gfp_mask = htlb_alloc_mask(h); node = huge_node(vma, address, gfp_mask, &mpol, &nodemask); - folio = alloc_hugetlb_folio_nodemask(h, node, nodemask, gfp_mask); + /* + * This is used to allocate a temporary hugetlb to hold the copied + * content, which will then be copied again to the final hugetlb + * consuming a reservation. Set the alloc_fallback to false to indicate + * that breaking the per-node hugetlb pool is not allowed in this case. + */ + folio = alloc_hugetlb_folio_nodemask(h, node, nodemask, gfp_mask, false); mpol_cond_put(mpol); return folio; @@ -6885,77 +6863,6 @@ out_release_nounlock: } #endif /* CONFIG_USERFAULTFD */ -struct page *hugetlb_follow_page_mask(struct vm_area_struct *vma, - unsigned long address, unsigned int flags, - unsigned int *page_mask) -{ - struct hstate *h = hstate_vma(vma); - struct mm_struct *mm = vma->vm_mm; - unsigned long haddr = address & huge_page_mask(h); - struct page *page = NULL; - spinlock_t *ptl; - pte_t *pte, entry; - int ret; - - hugetlb_vma_lock_read(vma); - pte = hugetlb_walk(vma, haddr, huge_page_size(h)); - if (!pte) - goto out_unlock; - - ptl = huge_pte_lock(h, mm, pte); - entry = huge_ptep_get(pte); - if (pte_present(entry)) { - page = pte_page(entry); - - if (!huge_pte_write(entry)) { - if (flags & FOLL_WRITE) { - page = NULL; - goto out; - } - - if (gup_must_unshare(vma, flags, page)) { - /* Tell the caller to do unsharing */ - page = ERR_PTR(-EMLINK); - goto out; - } - } - - page = nth_page(page, ((address & ~huge_page_mask(h)) >> PAGE_SHIFT)); - - /* - * Note that page may be a sub-page, and with vmemmap - * optimizations the page struct may be read only. - * try_grab_page() will increase the ref count on the - * head page, so this will be OK. - * - * try_grab_page() should always be able to get the page here, - * because we hold the ptl lock and have verified pte_present(). - */ - ret = try_grab_page(page, flags); - - if (WARN_ON_ONCE(ret)) { - page = ERR_PTR(ret); - goto out; - } - - *page_mask = (1U << huge_page_order(h)) - 1; - } -out: - spin_unlock(ptl); -out_unlock: - hugetlb_vma_unlock_read(vma); - - /* - * Fixup retval for dump requests: if pagecache doesn't exist, - * don't try to allocate a new page but just skip it. - */ - if (!page && (flags & FOLL_DUMP) && - !hugetlbfs_pagecache_present(h, vma, address)) - page = ERR_PTR(-EFAULT); - - return page; -} - long hugetlb_change_protection(struct vm_area_struct *vma, unsigned long address, unsigned long end, pgprot_t newprot, unsigned long cp_flags) @@ -7879,7 +7786,7 @@ void __init hugetlb_cma_reserve(int order) * huge page demotion. */ res = cma_declare_contiguous_nid(0, size, 0, - PAGE_SIZE << HUGETLB_PAGE_ORDER, + PAGE_SIZE << order, HUGETLB_PAGE_ORDER, false, name, &hugetlb_cma[nid], nid); if (res) { |