summaryrefslogtreecommitdiff
path: root/mm/hugetlb.c
diff options
context:
space:
mode:
Diffstat (limited to 'mm/hugetlb.c')
-rw-r--r--mm/hugetlb.c195
1 files changed, 120 insertions, 75 deletions
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index e3e6ac991b9c..32ab14aa4074 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -58,6 +58,7 @@ int hugetlb_max_hstate __read_mostly;
unsigned int default_hstate_idx;
struct hstate hstates[HUGE_MAX_HSTATE];
+__initdata nodemask_t hugetlb_bootmem_nodes;
__initdata struct list_head huge_boot_pages[MAX_NUMNODES];
static unsigned long hstate_boot_nrinvalid[HUGE_MAX_HSTATE] __initdata;
@@ -1250,7 +1251,7 @@ void hugetlb_dup_vma_private(struct vm_area_struct *vma)
/*
* Reset and decrement one ref on hugepage private reservation.
* Called with mm->mmap_lock writer semaphore held.
- * This function should be only used by move_vma() and operate on
+ * This function should be only used by mremap and operate on
* same sized vma. It should never come here with last ref on the
* reservation.
*/
@@ -1950,7 +1951,6 @@ static struct folio *alloc_buddy_hugetlb_folio(struct hstate *h,
int order = huge_page_order(h);
struct folio *folio;
bool alloc_try_hard = true;
- bool retry = true;
/*
* By default we always try hard to allocate the folio with
@@ -1965,22 +1965,8 @@ static struct folio *alloc_buddy_hugetlb_folio(struct hstate *h,
gfp_mask |= __GFP_RETRY_MAYFAIL;
if (nid == NUMA_NO_NODE)
nid = numa_mem_id();
-retry:
- 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);
- 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 unexpected inflated folio ref count\n");
- folio = NULL;
- }
+ folio = (struct folio *)__alloc_frozen_pages(gfp_mask, order, nid, nmask);
/*
* If we did not specify __GFP_RETRY_MAYFAIL, but still got a
@@ -2419,7 +2405,6 @@ 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, alloc_nodemask;
mbind_nodemask = policy_mbind_nodemask(htlb_alloc_mask(h));
@@ -2443,21 +2428,12 @@ retry:
for (i = 0; i < needed; i++) {
folio = NULL;
- /* Prioritize current node */
- if (node_isset(numa_mem_id(), alloc_nodemask))
- folio = alloc_surplus_hugetlb_folio(h, htlb_alloc_mask(h),
- numa_mem_id(), NULL);
-
- if (!folio) {
- for_each_node_mask(node, alloc_nodemask) {
- if (node == numa_mem_id())
- continue;
- folio = alloc_surplus_hugetlb_folio(h, htlb_alloc_mask(h),
- node, NULL);
- if (folio)
- break;
- }
- }
+ /*
+ * It is okay to use NUMA_NO_NODE because we use numa_mem_id()
+ * down the road to pick the current node if that is the case.
+ */
+ folio = alloc_surplus_hugetlb_folio(h, htlb_alloc_mask(h),
+ NUMA_NO_NODE, &alloc_nodemask);
if (!folio) {
alloc_ok = false;
break;
@@ -2896,10 +2872,9 @@ free_new:
return ret;
}
-int isolate_or_dissolve_huge_page(struct page *page, struct list_head *list)
+int isolate_or_dissolve_huge_folio(struct folio *folio, struct list_head *list)
{
struct hstate *h;
- struct folio *folio = page_folio(page);
int ret = -EBUSY;
/*
@@ -2949,12 +2924,20 @@ int replace_free_hugepage_folios(unsigned long start_pfn, unsigned long end_pfn)
while (start_pfn < end_pfn) {
folio = pfn_folio(start_pfn);
+
+ /*
+ * The folio might have been dissolved from under our feet, so make sure
+ * to carefully check the state under the lock.
+ */
+ spin_lock_irq(&hugetlb_lock);
if (folio_test_hugetlb(folio)) {
h = folio_hstate(folio);
} else {
+ spin_unlock_irq(&hugetlb_lock);
start_pfn++;
continue;
}
+ spin_unlock_irq(&hugetlb_lock);
if (!folio_ref_count(folio)) {
ret = alloc_and_dissolve_hugetlb_folio(h, folio,
@@ -3010,7 +2993,7 @@ struct folio *alloc_hugetlb_folio(struct vm_area_struct *vma,
struct hugepage_subpool *spool = subpool_vma(vma);
struct hstate *h = hstate_vma(vma);
struct folio *folio;
- long retval, gbl_chg;
+ long retval, gbl_chg, gbl_reserve;
map_chg_state map_chg;
int ret, idx;
struct hugetlb_cgroup *h_cg = NULL;
@@ -3163,8 +3146,16 @@ out_uncharge_cgroup_reservation:
hugetlb_cgroup_uncharge_cgroup_rsvd(idx, pages_per_huge_page(h),
h_cg);
out_subpool_put:
- if (map_chg)
- hugepage_subpool_put_pages(spool, 1);
+ /*
+ * put page to subpool iff the quota of subpool's rsv_hpages is used
+ * during hugepage_subpool_get_pages.
+ */
+ if (map_chg && !gbl_chg) {
+ gbl_reserve = hugepage_subpool_put_pages(spool, 1);
+ hugetlb_acct_memory(h, -gbl_reserve);
+ }
+
+
out_end_reservation:
if (map_chg != MAP_CHG_ENFORCED)
vma_end_reservation(h, vma, addr);
@@ -3237,7 +3228,8 @@ int __alloc_bootmem_huge_page(struct hstate *h, int nid)
}
/* allocate from next node when distributing huge pages */
- for_each_node_mask_to_alloc(&h->next_nid_to_alloc, nr_nodes, node, &node_states[N_ONLINE]) {
+ for_each_node_mask_to_alloc(&h->next_nid_to_alloc, nr_nodes, node,
+ &hugetlb_bootmem_nodes) {
m = alloc_bootmem(h, node, false);
if (!m)
return 0;
@@ -3701,6 +3693,15 @@ static void __init hugetlb_init_hstates(void)
struct hstate *h, *h2;
for_each_hstate(h) {
+ /*
+ * Always reset to first_memory_node here, even if
+ * next_nid_to_alloc was set before - we can't
+ * reference hugetlb_bootmem_nodes after init, and
+ * first_memory_node is right for all further allocations.
+ */
+ h->next_nid_to_alloc = first_memory_node;
+ h->next_nid_to_free = first_memory_node;
+
/* oversize hugepages were init'ed in early boot */
if (!hstate_is_gigantic(h))
hugetlb_hstate_alloc_pages(h);
@@ -4034,10 +4035,13 @@ static long demote_free_hugetlb_folios(struct hstate *src, struct hstate *dst,
list_for_each_entry_safe(folio, next, src_list, lru) {
int i;
+ bool cma;
if (folio_test_hugetlb_vmemmap_optimized(folio))
continue;
+ cma = folio_test_hugetlb_cma(folio);
+
list_del(&folio->lru);
split_page_owner(&folio->page, huge_page_order(src), huge_page_order(dst));
@@ -4053,6 +4057,9 @@ static long demote_free_hugetlb_folios(struct hstate *src, struct hstate *dst,
new_folio->mapping = NULL;
init_new_hugetlb_folio(dst, new_folio);
+ /* Copy the CMA flag so that it is freed correctly */
+ if (cma)
+ folio_set_hugetlb_cma(new_folio);
list_add(&new_folio->lru, &dst_list);
}
}
@@ -5007,6 +5014,20 @@ static int __init default_hugepagesz_setup(char *s)
}
hugetlb_early_param("default_hugepagesz", default_hugepagesz_setup);
+void __init hugetlb_bootmem_set_nodes(void)
+{
+ int i, nid;
+ unsigned long start_pfn, end_pfn;
+
+ if (!nodes_empty(hugetlb_bootmem_nodes))
+ return;
+
+ for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid) {
+ if (end_pfn > start_pfn)
+ node_set(nid, hugetlb_bootmem_nodes);
+ }
+}
+
static bool __hugetlb_bootmem_allocated __initdata;
bool __init hugetlb_bootmem_allocated(void)
@@ -5022,6 +5043,8 @@ void __init hugetlb_bootmem_alloc(void)
if (__hugetlb_bootmem_allocated)
return;
+ hugetlb_bootmem_set_nodes();
+
for (i = 0; i < MAX_NUMNODES; i++)
INIT_LIST_HEAD(&huge_boot_pages[i]);
@@ -5029,7 +5052,6 @@ void __init hugetlb_bootmem_alloc(void)
for_each_hstate(h) {
h->next_nid_to_alloc = first_online_node;
- h->next_nid_to_free = first_online_node;
if (hstate_is_gigantic(h))
hugetlb_hstate_alloc_pages(h);
@@ -5458,18 +5480,16 @@ const struct vm_operations_struct hugetlb_vm_ops = {
.pagesize = hugetlb_vm_op_pagesize,
};
-static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page,
+static pte_t make_huge_pte(struct vm_area_struct *vma, struct folio *folio,
bool try_mkwrite)
{
- pte_t entry;
+ pte_t entry = folio_mk_pte(folio, vma->vm_page_prot);
unsigned int shift = huge_page_shift(hstate_vma(vma));
if (try_mkwrite && (vma->vm_flags & VM_WRITE)) {
- entry = huge_pte_mkwrite(huge_pte_mkdirty(mk_huge_pte(page,
- vma->vm_page_prot)));
+ entry = pte_mkwrite_novma(pte_mkdirty(entry));
} else {
- entry = huge_pte_wrprotect(mk_huge_pte(page,
- vma->vm_page_prot));
+ entry = pte_wrprotect(entry);
}
entry = pte_mkyoung(entry);
entry = arch_make_huge_pte(entry, shift, vma->vm_flags);
@@ -5524,7 +5544,7 @@ static void
hugetlb_install_folio(struct vm_area_struct *vma, pte_t *ptep, unsigned long addr,
struct folio *new_folio, pte_t old, unsigned long sz)
{
- pte_t newpte = make_huge_pte(vma, &new_folio->page, true);
+ pte_t newpte = make_huge_pte(vma, new_folio, true);
__folio_mark_uptodate(new_folio);
hugetlb_add_new_anon_rmap(new_folio, vma, addr);
@@ -5828,14 +5848,14 @@ int move_hugetlb_page_tables(struct vm_area_struct *vma,
void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
unsigned long start, unsigned long end,
- struct page *ref_page, zap_flags_t zap_flags)
+ struct folio *folio, zap_flags_t zap_flags)
{
struct mm_struct *mm = vma->vm_mm;
+ const bool folio_provided = !!folio;
unsigned long address;
pte_t *ptep;
pte_t pte;
spinlock_t *ptl;
- struct page *page;
struct hstate *h = hstate_vma(vma);
unsigned long sz = huge_page_size(h);
bool adjust_reservation = false;
@@ -5899,14 +5919,13 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
continue;
}
- page = pte_page(pte);
/*
- * If a reference page is supplied, it is because a specific
- * page is being unmapped, not a range. Ensure the page we
- * are about to unmap is the actual page of interest.
+ * If a folio is supplied, it is because a specific
+ * folio is being unmapped, not a range. Ensure the folio we
+ * are about to unmap is the actual folio of interest.
*/
- if (ref_page) {
- if (page != ref_page) {
+ if (folio_provided) {
+ if (folio != page_folio(pte_page(pte))) {
spin_unlock(ptl);
continue;
}
@@ -5916,12 +5935,14 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
* looking like data was lost
*/
set_vma_resv_flags(vma, HPAGE_RESV_UNMAPPED);
+ } else {
+ folio = page_folio(pte_page(pte));
}
pte = huge_ptep_get_and_clear(mm, address, ptep, sz);
tlb_remove_huge_tlb_entry(h, tlb, ptep, address);
if (huge_pte_dirty(pte))
- set_page_dirty(page);
+ folio_mark_dirty(folio);
/* Leave a uffd-wp pte marker if needed */
if (huge_pte_uffd_wp(pte) &&
!(zap_flags & ZAP_FLAG_DROP_MARKER))
@@ -5929,7 +5950,7 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
make_pte_marker(PTE_MARKER_UFFD_WP),
sz);
hugetlb_count_sub(pages_per_huge_page(h), mm);
- hugetlb_remove_rmap(page_folio(page));
+ hugetlb_remove_rmap(folio);
/*
* Restore the reservation for anonymous page, otherwise the
@@ -5938,8 +5959,8 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
* reservation bit.
*/
if (!h->surplus_huge_pages && __vma_private_lock(vma) &&
- folio_test_anon(page_folio(page))) {
- folio_set_hugetlb_restore_reserve(page_folio(page));
+ folio_test_anon(folio)) {
+ folio_set_hugetlb_restore_reserve(folio);
/* Reservation to be adjusted after the spin lock */
adjust_reservation = true;
}
@@ -5963,16 +5984,17 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma,
* count will not be incremented by free_huge_folio.
* Act as if we consumed the reservation.
*/
- folio_clear_hugetlb_restore_reserve(page_folio(page));
+ folio_clear_hugetlb_restore_reserve(folio);
else if (rc)
vma_add_reservation(h, vma, address);
}
- tlb_remove_page_size(tlb, page, huge_page_size(h));
+ tlb_remove_page_size(tlb, folio_page(folio, 0),
+ folio_size(folio));
/*
- * Bail out after unmapping reference page if supplied
+ * If we were instructed to unmap a specific folio, we're done.
*/
- if (ref_page)
+ if (folio_provided)
break;
}
tlb_end_vma(tlb, vma);
@@ -6034,7 +6056,7 @@ void __hugetlb_zap_end(struct vm_area_struct *vma,
}
void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
- unsigned long end, struct page *ref_page,
+ unsigned long end, struct folio *folio,
zap_flags_t zap_flags)
{
struct mmu_notifier_range range;
@@ -6046,7 +6068,8 @@ void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
mmu_notifier_invalidate_range_start(&range);
tlb_gather_mmu(&tlb, vma->vm_mm);
- __unmap_hugepage_range(&tlb, vma, start, end, ref_page, zap_flags);
+ __unmap_hugepage_range(&tlb, vma, start, end,
+ folio, zap_flags);
mmu_notifier_invalidate_range_end(&range);
tlb_finish_mmu(&tlb);
@@ -6059,7 +6082,7 @@ void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
* same region.
*/
static void unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma,
- struct page *page, unsigned long address)
+ struct folio *folio, unsigned long address)
{
struct hstate *h = hstate_vma(vma);
struct vm_area_struct *iter_vma;
@@ -6103,7 +6126,8 @@ static void unmap_ref_private(struct mm_struct *mm, struct vm_area_struct *vma,
*/
if (!is_vma_resv_set(iter_vma, HPAGE_RESV_OWNER))
unmap_hugepage_range(iter_vma, address,
- address + huge_page_size(h), page, 0);
+ address + huge_page_size(h),
+ folio, 0);
}
i_mmap_unlock_write(mapping);
}
@@ -6226,8 +6250,7 @@ retry_avoidcopy:
hugetlb_vma_unlock_read(vma);
mutex_unlock(&hugetlb_fault_mutex_table[hash]);
- unmap_ref_private(mm, vma, &old_folio->page,
- vmf->address);
+ unmap_ref_private(mm, vma, old_folio, vmf->address);
mutex_lock(&hugetlb_fault_mutex_table[hash]);
hugetlb_vma_lock_read(vma);
@@ -6274,7 +6297,7 @@ retry_avoidcopy:
spin_lock(vmf->ptl);
vmf->pte = hugetlb_walk(vma, vmf->address, huge_page_size(h));
if (likely(vmf->pte && pte_same(huge_ptep_get(mm, vmf->address, vmf->pte), pte))) {
- pte_t newpte = make_huge_pte(vma, &new_folio->page, !unshare);
+ pte_t newpte = make_huge_pte(vma, new_folio, !unshare);
/* Break COW or unshare */
huge_ptep_clear_flush(vma, vmf->address, vmf->pte);
@@ -6554,7 +6577,7 @@ static vm_fault_t hugetlb_no_page(struct address_space *mapping,
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_SHARED);
+ new_pte = make_huge_pte(vma, folio, vma->vm_flags & VM_SHARED);
/*
* If this pte was previously wr-protected, keep it wr-protected even
* if populated.
@@ -7039,7 +7062,7 @@ int hugetlb_mfill_atomic_pte(pte_t *dst_pte,
* For either: (1) CONTINUE on a non-shared VMA, or (2) UFFDIO_COPY
* with wp flag set, don't set pte write bit.
*/
- _dst_pte = make_huge_pte(dst_vma, &folio->page,
+ _dst_pte = make_huge_pte(dst_vma, folio,
!wp_enabled && !(is_continue && !vm_shared));
/*
* Always mark UFFDIO_COPY page dirty; note that this may not be
@@ -7233,7 +7256,7 @@ bool hugetlb_reserve_pages(struct inode *inode,
struct vm_area_struct *vma,
vm_flags_t vm_flags)
{
- long chg = -1, add = -1;
+ long chg = -1, add = -1, spool_resv, gbl_resv;
struct hstate *h = hstate_inode(inode);
struct hugepage_subpool *spool = subpool_inode(inode);
struct resv_map *resv_map;
@@ -7368,8 +7391,16 @@ bool hugetlb_reserve_pages(struct inode *inode,
return true;
out_put_pages:
- /* put back original number of pages, chg */
- (void)hugepage_subpool_put_pages(spool, chg);
+ spool_resv = chg - gbl_reserve;
+ if (spool_resv) {
+ /* put sub pool's reservation back, chg - gbl_reserve */
+ gbl_resv = hugepage_subpool_put_pages(spool, spool_resv);
+ /*
+ * subpool's reserved pages can not be put back due to race,
+ * return to hstate.
+ */
+ hugetlb_acct_memory(h, -gbl_resv);
+ }
out_uncharge_cgroup:
hugetlb_cgroup_uncharge_cgroup_rsvd(hstate_index(h),
chg * pages_per_huge_page(h), h_cg);
@@ -7909,3 +7940,17 @@ void hugetlb_unshare_all_pmds(struct vm_area_struct *vma)
hugetlb_unshare_pmds(vma, ALIGN(vma->vm_start, PUD_SIZE),
ALIGN_DOWN(vma->vm_end, PUD_SIZE));
}
+
+/*
+ * For hugetlb, mremap() is an odd edge case - while the VMA copying is
+ * performed, we permit both the old and new VMAs to reference the same
+ * reservation.
+ *
+ * We fix this up after the operation succeeds, or if a newly allocated VMA
+ * is closed as a result of a failure to allocate memory.
+ */
+void fixup_hugetlb_reservations(struct vm_area_struct *vma)
+{
+ if (is_vm_hugetlb_page(vma))
+ clear_vma_resv_huge_pages(vma);
+}
generated by cgit (git 2.34.1) at 2025-06-03 23:29:43 +0000