When hugetlb vmemmap optimization was introduced, the overhead of enabling
the option was measured as described in commit 426e5c429d16 [1]. The summary
states that allocating a hugetlb page should be ~2x slower with optimization
and freeing a hugetlb page should be ~2-3x slower. Such overhead was deemed
an acceptable trade off for the memory savings obtained by freeing vmemmap
pages.
It was recently reported that the overhead associated with enabling vmemmap
optimization could be as high as 190x for hugetlb page allocations.
Yes, 190x! Some actual numbers from other environments are:
Bare Metal 8 socket Intel(R) Xeon(R) CPU E7-8895
------------------------------------------------
Unmodified next-20230824, vm.hugetlb_optimize_vmemmap = 0
time echo 500000 > .../hugepages-2048kB/nr_hugepages
real 0m4.119s
time echo 0 > .../hugepages-2048kB/nr_hugepages
real 0m4.477s
Unmodified next-20230824, vm.hugetlb_optimize_vmemmap = 1
time echo 500000 > .../hugepages-2048kB/nr_hugepages
real 0m28.973s
time echo 0 > .../hugepages-2048kB/nr_hugepages
real 0m36.748s
VM with 252 vcpus on host with 2 socket AMD EPYC 7J13 Milan
-----------------------------------------------------------
Unmodified next-20230824, vm.hugetlb_optimize_vmemmap = 0
time echo 524288 > .../hugepages-2048kB/nr_hugepages
real 0m2.463s
time echo 0 > .../hugepages-2048kB/nr_hugepages
real 0m2.931s
Unmodified next-20230824, vm.hugetlb_optimize_vmemmap = 1
time echo 524288 > .../hugepages-2048kB/nr_hugepages
real 2m27.609s
time echo 0 > .../hugepages-2048kB/nr_hugepages
real 2m29.924s
In the VM environment, the slowdown of enabling hugetlb vmemmap optimization
resulted in allocation times being 61x slower.
A quick profile showed that the vast majority of this overhead was due to
TLB flushing. Each time we modify the kernel pagetable we need to flush
the TLB. For each hugetlb that is optimized, there could be potentially
two TLB flushes performed. One for the vmemmap pages associated with the
hugetlb page, and potentially another one if the vmemmap pages are mapped
at the PMD level and must be split. The TLB flushes required for the kernel
pagetable, result in a broadcast IPI with each CPU having to flush a range
of pages, or do a global flush if a threshold is exceeded. So, the flush
time increases with the number of CPUs. In addition, in virtual environments
the broadcast IPI can’t be accelerated by hypervisor hardware and leads to
traps that need to wakeup/IPI all vCPUs which is very expensive. Because of
this the slowdown in virtual environments is even worse than bare metal as
the number of vCPUS/CPUs is increased.
The following series attempts to reduce amount of time spent in TLB flushing.
The idea is to batch the vmemmap modification operations for multiple hugetlb
pages. Instead of doing one or two TLB flushes for each page, we do two TLB
flushes for each batch of pages. One flush after splitting pages mapped at
the PMD level, and another after remapping vmemmap associated with all
hugetlb pages. Results of such batching are as follows:
Bare Metal 8 socket Intel(R) Xeon(R) CPU E7-8895
------------------------------------------------
next-20230824 + Batching patches, vm.hugetlb_optimize_vmemmap = 0
time echo 500000 > .../hugepages-2048kB/nr_hugepages
real 0m4.719s
time echo 0 > .../hugepages-2048kB/nr_hugepages
real 0m4.245s
next-20230824 + Batching patches, vm.hugetlb_optimize_vmemmap = 1
time echo 500000 > .../hugepages-2048kB/nr_hugepages
real 0m7.267s
time echo 0 > .../hugepages-2048kB/nr_hugepages
real 0m13.199s
VM with 252 vcpus on host with 2 socket AMD EPYC 7J13 Milan
-----------------------------------------------------------
next-20230824 + Batching patches, vm.hugetlb_optimize_vmemmap = 0
time echo 524288 > .../hugepages-2048kB/nr_hugepages
real 0m2.715s
time echo 0 > .../hugepages-2048kB/nr_hugepages
real 0m3.186s
next-20230824 + Batching patches, vm.hugetlb_optimize_vmemmap = 1
time echo 524288 > .../hugepages-2048kB/nr_hugepages
real 0m4.799s
time echo 0 > .../hugepages-2048kB/nr_hugepages
real 0m5.273s
With batching, results are back in the 2-3x slowdown range.
This series is based on mm-stable (October 18)
Changed v7 -> v8:
- patch 2 initialized h in gather_bootmem_prealloc so clang would not
cause loop to be executed when entry condition false.
- patch 2 changed locking in prep_and_add_allocated_folios to take into
account that IRQs may be disabled.
- patch 3 same locking changes for patch 2 above for
prep_and_add_bootmem_folios
- Removed Muchun RBs from patches although they are much the same
Changes v6 -> v7:
- Fixed hugetlb_vmemmap_restore_folios stub for the
!CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP case
- Added Muchun RB for patches 4 and 8
Changes v5 -> v6:
- patch 4 in bulk_vmemmap_restore_error remove folio from list before
calling add_hugetlb_folio.
- Added Muchun RB for patches 2 and 3
Changes v4 -> v5:
- patch 3 comment style updated, unnecessary INIT_LIST_HEAD
- patch 4 updated hugetlb_vmemmap_restore_folios to pass back number of
restored folios in non-error case. In addition, routine passes back
list of folios with vmemmmap. Naming more consistent.
- patch 5 remover over optimization and added Muchun RB
- patch 6 break and early return in ENOMEM case. Updated comments.
Added Muchun RB.
- patch 7 Updated comments about splitting failure. Added Muchun RB.
- patch 8 Made comments consistent.
Changes v3 -> v4:
- Rebased on mm-unstable and dropped requisite patches.
- patch 2 updated to take bootmem vmemmap initialization into account
- patch 3 more changes for bootmem hugetlb pages. added routine
prep_and_add_bootmem_folios.
- patch 5 in hugetlb_vmemmap_optimize_folios on ENOMEM check for
list_empty before freeing and retry. This is more important in
subsequent patch where we flush_tlb_all after ENOMEM.
Changes v2 -> v3:
- patch 5 was part of an earlier series that was not picked up. It is
included here as it helps with batching optimizations.
- patch 6 hugetlb_vmemmap_restore_folios is changed from type void to
returning an error code as well as an additional output parameter providing
the number folios for which vmemmap was actually restored. The caller can
then be more intelligent about processing the list.
- patch 9 eliminate local list in vmemmap_restore_pte. The routine
hugetlb_vmemmap_optimize_folios checks for ENOMEM and frees accumulated
vmemmap pages while processing the list.
- patch 10 introduce flags field to struct vmemmap_remap_walk and
VMEMMAP_SPLIT_NO_TLB_FLUSH for not flushing during pass to split PMDs.
- patch 11 rename flag VMEMMAP_REMAP_NO_TLB_FLUSH and pass in from callers.
Changes v1 -> v2:
- patch 5 now takes into account the requirement that only compound
pages with hugetlb flag set can be passed to vmemmmap routines. This
involved separating the 'prep' of hugetlb pages even further. The code
dealing with bootmem allocations was also modified so that batching is
possible. Adding a 'batch' of hugetlb pages to their respective free
lists is now done in one lock cycle.
- patch 7 added description of routine hugetlb_vmemmap_restore_folios
(Muchun).
- patch 8 rename bulk_pages to vmemmap_pages and let caller be responsible
for freeing (Muchun)
- patch 9 use 'walk->remap_pte' to determine if a split only operation
is being performed (Muchun). Removed unused variable and
hugetlb_optimize_vmemmap_key (Muchun).
- Patch 10 pass 'flags variable' instead of bool to indicate behavior and
allow for future expansion (Muchun). Single flag VMEMMAP_NO_TLB_FLUSH.
Provide detailed comment about the need to keep old and new vmemmap pages
in sync (Muchun).
- Patch 11 pass flag variable as in patch 10 (Muchun).
Joao Martins (2):
hugetlb: batch PMD split for bulk vmemmap dedup
hugetlb: batch TLB flushes when freeing vmemmap
Mike Kravetz (6):
hugetlb: optimize update_and_free_pages_bulk to avoid lock cycles
hugetlb: restructure pool allocations
hugetlb: perform vmemmap optimization on a list of pages
hugetlb: perform vmemmap restoration on a list of pages
hugetlb: batch freeing of vmemmap pages
hugetlb: batch TLB flushes when restoring vmemmap
mm/hugetlb.c | 303 ++++++++++++++++++++++++++++++++++++-------
mm/hugetlb_vmemmap.c | 273 ++++++++++++++++++++++++++++++++------
mm/hugetlb_vmemmap.h | 16 +++
3 files changed, 509 insertions(+), 83 deletions(-)
--
2.41.0
Now that batching of hugetlb vmemmap optimization processing is possible,
batch the freeing of vmemmap pages. When freeing vmemmap pages for a
hugetlb page, we add them to a list that is freed after the entire batch
has been processed.
This enhances the ability to return contiguous ranges of memory to the
low level allocators.
Signed-off-by: Mike Kravetz <[email protected]>
Reviewed-by: Muchun Song <[email protected]>
---
mm/hugetlb_vmemmap.c | 82 ++++++++++++++++++++++++++++++--------------
1 file changed, 56 insertions(+), 26 deletions(-)
diff --git a/mm/hugetlb_vmemmap.c b/mm/hugetlb_vmemmap.c
index 77f44b81ff01..4ac521e596db 100644
--- a/mm/hugetlb_vmemmap.c
+++ b/mm/hugetlb_vmemmap.c
@@ -251,7 +251,7 @@ static void vmemmap_remap_pte(pte_t *pte, unsigned long addr,
}
entry = mk_pte(walk->reuse_page, pgprot);
- list_add_tail(&page->lru, walk->vmemmap_pages);
+ list_add(&page->lru, walk->vmemmap_pages);
set_pte_at(&init_mm, addr, pte, entry);
}
@@ -306,18 +306,20 @@ static void vmemmap_restore_pte(pte_t *pte, unsigned long addr,
* @end: end address of the vmemmap virtual address range that we want to
* remap.
* @reuse: reuse address.
+ * @vmemmap_pages: list to deposit vmemmap pages to be freed. It is callers
+ * responsibility to free pages.
*
* Return: %0 on success, negative error code otherwise.
*/
static int vmemmap_remap_free(unsigned long start, unsigned long end,
- unsigned long reuse)
+ unsigned long reuse,
+ struct list_head *vmemmap_pages)
{
int ret;
- LIST_HEAD(vmemmap_pages);
struct vmemmap_remap_walk walk = {
.remap_pte = vmemmap_remap_pte,
.reuse_addr = reuse,
- .vmemmap_pages = &vmemmap_pages,
+ .vmemmap_pages = vmemmap_pages,
};
int nid = page_to_nid((struct page *)reuse);
gfp_t gfp_mask = GFP_KERNEL | __GFP_NORETRY | __GFP_NOWARN;
@@ -334,7 +336,7 @@ static int vmemmap_remap_free(unsigned long start, unsigned long end,
if (walk.reuse_page) {
copy_page(page_to_virt(walk.reuse_page),
(void *)walk.reuse_addr);
- list_add(&walk.reuse_page->lru, &vmemmap_pages);
+ list_add(&walk.reuse_page->lru, vmemmap_pages);
}
/*
@@ -365,15 +367,13 @@ static int vmemmap_remap_free(unsigned long start, unsigned long end,
walk = (struct vmemmap_remap_walk) {
.remap_pte = vmemmap_restore_pte,
.reuse_addr = reuse,
- .vmemmap_pages = &vmemmap_pages,
+ .vmemmap_pages = vmemmap_pages,
};
vmemmap_remap_range(reuse, end, &walk);
}
mmap_read_unlock(&init_mm);
- free_vmemmap_page_list(&vmemmap_pages);
-
return ret;
}
@@ -389,7 +389,7 @@ static int alloc_vmemmap_page_list(unsigned long start, unsigned long end,
page = alloc_pages_node(nid, gfp_mask, 0);
if (!page)
goto out;
- list_add_tail(&page->lru, list);
+ list_add(&page->lru, list);
}
return 0;
@@ -577,24 +577,17 @@ static bool vmemmap_should_optimize(const struct hstate *h, const struct page *h
return true;
}
-/**
- * hugetlb_vmemmap_optimize - optimize @head page's vmemmap pages.
- * @h: struct hstate.
- * @head: the head page whose vmemmap pages will be optimized.
- *
- * This function only tries to optimize @head's vmemmap pages and does not
- * guarantee that the optimization will succeed after it returns. The caller
- * can use HPageVmemmapOptimized(@head) to detect if @head's vmemmap pages
- * have been optimized.
- */
-void hugetlb_vmemmap_optimize(const struct hstate *h, struct page *head)
+static int __hugetlb_vmemmap_optimize(const struct hstate *h,
+ struct page *head,
+ struct list_head *vmemmap_pages)
{
+ int ret = 0;
unsigned long vmemmap_start = (unsigned long)head, vmemmap_end;
unsigned long vmemmap_reuse;
VM_WARN_ON_ONCE(!PageHuge(head));
if (!vmemmap_should_optimize(h, head))
- return;
+ return ret;
static_branch_inc(&hugetlb_optimize_vmemmap_key);
@@ -604,21 +597,58 @@ void hugetlb_vmemmap_optimize(const struct hstate *h, struct page *head)
/*
* Remap the vmemmap virtual address range [@vmemmap_start, @vmemmap_end)
- * to the page which @vmemmap_reuse is mapped to, then free the pages
- * which the range [@vmemmap_start, @vmemmap_end] is mapped to.
+ * to the page which @vmemmap_reuse is mapped to. Add pages previously
+ * mapping the range to vmemmap_pages list so that they can be freed by
+ * the caller.
*/
- if (vmemmap_remap_free(vmemmap_start, vmemmap_end, vmemmap_reuse))
+ ret = vmemmap_remap_free(vmemmap_start, vmemmap_end, vmemmap_reuse, vmemmap_pages);
+ if (ret)
static_branch_dec(&hugetlb_optimize_vmemmap_key);
else
SetHPageVmemmapOptimized(head);
+
+ return ret;
+}
+
+/**
+ * hugetlb_vmemmap_optimize - optimize @head page's vmemmap pages.
+ * @h: struct hstate.
+ * @head: the head page whose vmemmap pages will be optimized.
+ *
+ * This function only tries to optimize @head's vmemmap pages and does not
+ * guarantee that the optimization will succeed after it returns. The caller
+ * can use HPageVmemmapOptimized(@head) to detect if @head's vmemmap pages
+ * have been optimized.
+ */
+void hugetlb_vmemmap_optimize(const struct hstate *h, struct page *head)
+{
+ LIST_HEAD(vmemmap_pages);
+
+ __hugetlb_vmemmap_optimize(h, head, &vmemmap_pages);
+ free_vmemmap_page_list(&vmemmap_pages);
}
void hugetlb_vmemmap_optimize_folios(struct hstate *h, struct list_head *folio_list)
{
struct folio *folio;
+ LIST_HEAD(vmemmap_pages);
+
+ list_for_each_entry(folio, folio_list, lru) {
+ int ret = __hugetlb_vmemmap_optimize(h, &folio->page,
+ &vmemmap_pages);
+
+ /*
+ * Pages to be freed may have been accumulated. If we
+ * encounter an ENOMEM, free what we have and try again.
+ */
+ if (ret == -ENOMEM && !list_empty(&vmemmap_pages)) {
+ free_vmemmap_page_list(&vmemmap_pages);
+ INIT_LIST_HEAD(&vmemmap_pages);
+ __hugetlb_vmemmap_optimize(h, &folio->page, &vmemmap_pages);
+ }
+ }
- list_for_each_entry(folio, folio_list, lru)
- hugetlb_vmemmap_optimize(h, &folio->page);
+ free_vmemmap_page_list(&vmemmap_pages);
}
static struct ctl_table hugetlb_vmemmap_sysctls[] = {
--
2.41.0
Allocation of a hugetlb page for the hugetlb pool is done by the routine
alloc_pool_huge_page. This routine will allocate contiguous pages from
a low level allocator, prep the pages for usage as a hugetlb page and
then add the resulting hugetlb page to the pool.
In the 'prep' stage, optional vmemmap optimization is done. For
performance reasons we want to perform vmemmap optimization on multiple
hugetlb pages at once. To do this, restructure the hugetlb pool
allocation code such that vmemmap optimization can be isolated and later
batched.
The code to allocate hugetlb pages from bootmem was also modified to
allow batching.
No functional changes, only code restructure.
Signed-off-by: Mike Kravetz <[email protected]>
---
mm/hugetlb.c | 180 ++++++++++++++++++++++++++++++++++++++++-----------
1 file changed, 141 insertions(+), 39 deletions(-)
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index b839080a2a6b..559f7c71c596 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -1996,16 +1996,21 @@ static void __prep_account_new_huge_page(struct hstate *h, int nid)
h->nr_huge_pages_node[nid]++;
}
-static void __prep_new_hugetlb_folio(struct hstate *h, struct folio *folio)
+static void init_new_hugetlb_folio(struct hstate *h, struct folio *folio)
{
folio_set_hugetlb(folio);
- hugetlb_vmemmap_optimize(h, &folio->page);
INIT_LIST_HEAD(&folio->lru);
hugetlb_set_folio_subpool(folio, NULL);
set_hugetlb_cgroup(folio, NULL);
set_hugetlb_cgroup_rsvd(folio, NULL);
}
+static void __prep_new_hugetlb_folio(struct hstate *h, struct folio *folio)
+{
+ init_new_hugetlb_folio(h, folio);
+ hugetlb_vmemmap_optimize(h, &folio->page);
+}
+
static void prep_new_hugetlb_folio(struct hstate *h, struct folio *folio, int nid)
{
__prep_new_hugetlb_folio(h, folio);
@@ -2202,16 +2207,9 @@ static struct folio *alloc_buddy_hugetlb_folio(struct hstate *h,
return page_folio(page);
}
-/*
- * Common helper to allocate a fresh hugetlb page. All specific allocators
- * should use this function to get new hugetlb pages
- *
- * Note that returned page is 'frozen': ref count of head page and all tail
- * pages is zero.
- */
-static struct folio *alloc_fresh_hugetlb_folio(struct hstate *h,
- gfp_t gfp_mask, int nid, nodemask_t *nmask,
- nodemask_t *node_alloc_noretry)
+static struct folio *__alloc_fresh_hugetlb_folio(struct hstate *h,
+ gfp_t gfp_mask, int nid, nodemask_t *nmask,
+ nodemask_t *node_alloc_noretry)
{
struct folio *folio;
bool retry = false;
@@ -2224,6 +2222,7 @@ static struct folio *alloc_fresh_hugetlb_folio(struct hstate *h,
nid, nmask, node_alloc_noretry);
if (!folio)
return NULL;
+
if (hstate_is_gigantic(h)) {
if (!prep_compound_gigantic_folio(folio, huge_page_order(h))) {
/*
@@ -2238,32 +2237,81 @@ static struct folio *alloc_fresh_hugetlb_folio(struct hstate *h,
return NULL;
}
}
- prep_new_hugetlb_folio(h, folio, folio_nid(folio));
return folio;
}
+static struct folio *only_alloc_fresh_hugetlb_folio(struct hstate *h,
+ gfp_t gfp_mask, int nid, nodemask_t *nmask,
+ nodemask_t *node_alloc_noretry)
+{
+ struct folio *folio;
+
+ folio = __alloc_fresh_hugetlb_folio(h, gfp_mask, nid, nmask,
+ node_alloc_noretry);
+ if (folio)
+ init_new_hugetlb_folio(h, folio);
+ return folio;
+}
+
/*
- * Allocates a fresh page to the hugetlb allocator pool in the node interleaved
- * manner.
+ * Common helper to allocate a fresh hugetlb page. All specific allocators
+ * should use this function to get new hugetlb pages
+ *
+ * Note that returned page is 'frozen': ref count of head page and all tail
+ * pages is zero.
*/
-static int alloc_pool_huge_page(struct hstate *h, nodemask_t *nodes_allowed,
- nodemask_t *node_alloc_noretry)
+static struct folio *alloc_fresh_hugetlb_folio(struct hstate *h,
+ gfp_t gfp_mask, int nid, nodemask_t *nmask,
+ nodemask_t *node_alloc_noretry)
{
struct folio *folio;
- int nr_nodes, node;
+
+ folio = __alloc_fresh_hugetlb_folio(h, gfp_mask, nid, nmask,
+ node_alloc_noretry);
+ if (!folio)
+ return NULL;
+
+ prep_new_hugetlb_folio(h, folio, folio_nid(folio));
+ return folio;
+}
+
+static void prep_and_add_allocated_folios(struct hstate *h,
+ struct list_head *folio_list)
+{
+ unsigned long flags;
+ struct folio *folio, *tmp_f;
+
+ /* Add all new pool pages to free lists in one lock cycle */
+ spin_lock_irqsave(&hugetlb_lock, flags);
+ list_for_each_entry_safe(folio, tmp_f, folio_list, lru) {
+ __prep_account_new_huge_page(h, folio_nid(folio));
+ enqueue_hugetlb_folio(h, folio);
+ }
+ spin_unlock_irqrestore(&hugetlb_lock, flags);
+}
+
+/*
+ * Allocates a fresh hugetlb page in a node interleaved manner. The page
+ * will later be added to the appropriate hugetlb pool.
+ */
+static struct folio *alloc_pool_huge_folio(struct hstate *h,
+ nodemask_t *nodes_allowed,
+ nodemask_t *node_alloc_noretry)
+{
gfp_t gfp_mask = htlb_alloc_mask(h) | __GFP_THISNODE;
+ int nr_nodes, node;
for_each_node_mask_to_alloc(h, nr_nodes, node, nodes_allowed) {
- folio = alloc_fresh_hugetlb_folio(h, gfp_mask, node,
+ struct folio *folio;
+
+ folio = only_alloc_fresh_hugetlb_folio(h, gfp_mask, node,
nodes_allowed, node_alloc_noretry);
- if (folio) {
- free_huge_folio(folio); /* free it into the hugepage allocator */
- return 1;
- }
+ if (folio)
+ return folio;
}
- return 0;
+ return NULL;
}
/*
@@ -3302,25 +3350,35 @@ static void __init hugetlb_folio_init_vmemmap(struct folio *folio,
*/
static void __init gather_bootmem_prealloc(void)
{
+ LIST_HEAD(folio_list);
struct huge_bootmem_page *m;
+ struct hstate *h = NULL, *prev_h = NULL;
list_for_each_entry(m, &huge_boot_pages, list) {
struct page *page = virt_to_page(m);
struct folio *folio = (void *)page;
- struct hstate *h = m->hstate;
+
+ h = m->hstate;
+ /*
+ * It is possible to have multiple huge page sizes (hstates)
+ * in this list. If so, process each size separately.
+ */
+ if (h != prev_h && prev_h != NULL)
+ prep_and_add_allocated_folios(prev_h, &folio_list);
+ prev_h = h;
VM_BUG_ON(!hstate_is_gigantic(h));
WARN_ON(folio_ref_count(folio) != 1);
hugetlb_folio_init_vmemmap(folio, h,
HUGETLB_VMEMMAP_RESERVE_PAGES);
- prep_new_hugetlb_folio(h, folio, folio_nid(folio));
+ __prep_new_hugetlb_folio(h, folio);
/* If HVO fails, initialize all tail struct pages */
if (!HPageVmemmapOptimized(&folio->page))
hugetlb_folio_init_tail_vmemmap(folio,
HUGETLB_VMEMMAP_RESERVE_PAGES,
pages_per_huge_page(h));
- free_huge_folio(folio); /* add to the hugepage allocator */
+ list_add(&folio->lru, &folio_list);
/*
* We need to restore the 'stolen' pages to totalram_pages
@@ -3330,6 +3388,8 @@ static void __init gather_bootmem_prealloc(void)
adjust_managed_page_count(page, pages_per_huge_page(h));
cond_resched();
}
+
+ prep_and_add_allocated_folios(h, &folio_list);
}
static void __init hugetlb_hstate_alloc_pages_onenode(struct hstate *h, int nid)
@@ -3363,9 +3423,22 @@ static void __init hugetlb_hstate_alloc_pages_onenode(struct hstate *h, int nid)
h->max_huge_pages_node[nid] = i;
}
+/*
+ * NOTE: this routine is called in different contexts for gigantic and
+ * non-gigantic pages.
+ * - For gigantic pages, this is called early in the boot process and
+ * pages are allocated from memblock allocated or something similar.
+ * Gigantic pages are actually added to pools later with the routine
+ * gather_bootmem_prealloc.
+ * - For non-gigantic pages, this is called later in the boot process after
+ * all of mm is up and functional. Pages are allocated from buddy and
+ * then added to hugetlb pools.
+ */
static void __init hugetlb_hstate_alloc_pages(struct hstate *h)
{
unsigned long i;
+ struct folio *folio;
+ LIST_HEAD(folio_list);
nodemask_t *node_alloc_noretry;
bool node_specific_alloc = false;
@@ -3407,14 +3480,25 @@ static void __init hugetlb_hstate_alloc_pages(struct hstate *h)
for (i = 0; i < h->max_huge_pages; ++i) {
if (hstate_is_gigantic(h)) {
+ /*
+ * gigantic pages not added to list as they are not
+ * added to pools now.
+ */
if (!alloc_bootmem_huge_page(h, NUMA_NO_NODE))
break;
- } else if (!alloc_pool_huge_page(h,
- &node_states[N_MEMORY],
- node_alloc_noretry))
- break;
+ } else {
+ folio = alloc_pool_huge_folio(h, &node_states[N_MEMORY],
+ node_alloc_noretry);
+ if (!folio)
+ break;
+ list_add(&folio->lru, &folio_list);
+ }
cond_resched();
}
+
+ /* list will be empty if hstate_is_gigantic */
+ prep_and_add_allocated_folios(h, &folio_list);
+
if (i < h->max_huge_pages) {
char buf[32];
@@ -3548,7 +3632,9 @@ static int adjust_pool_surplus(struct hstate *h, nodemask_t *nodes_allowed,
static int set_max_huge_pages(struct hstate *h, unsigned long count, int nid,
nodemask_t *nodes_allowed)
{
- unsigned long min_count, ret;
+ unsigned long min_count;
+ unsigned long allocated;
+ struct folio *folio;
LIST_HEAD(page_list);
NODEMASK_ALLOC(nodemask_t, node_alloc_noretry, GFP_KERNEL);
@@ -3625,7 +3711,8 @@ static int set_max_huge_pages(struct hstate *h, unsigned long count, int nid,
break;
}
- while (count > persistent_huge_pages(h)) {
+ allocated = 0;
+ while (count > (persistent_huge_pages(h) + allocated)) {
/*
* If this allocation races such that we no longer need the
* page, free_huge_folio will handle it by freeing the page
@@ -3636,15 +3723,32 @@ static int set_max_huge_pages(struct hstate *h, unsigned long count, int nid,
/* yield cpu to avoid soft lockup */
cond_resched();
- ret = alloc_pool_huge_page(h, nodes_allowed,
+ folio = alloc_pool_huge_folio(h, nodes_allowed,
node_alloc_noretry);
- spin_lock_irq(&hugetlb_lock);
- if (!ret)
+ if (!folio) {
+ prep_and_add_allocated_folios(h, &page_list);
+ spin_lock_irq(&hugetlb_lock);
goto out;
+ }
+
+ list_add(&folio->lru, &page_list);
+ allocated++;
/* Bail for signals. Probably ctrl-c from user */
- if (signal_pending(current))
+ if (signal_pending(current)) {
+ prep_and_add_allocated_folios(h, &page_list);
+ spin_lock_irq(&hugetlb_lock);
goto out;
+ }
+
+ spin_lock_irq(&hugetlb_lock);
+ }
+
+ /* Add allocated pages to the pool */
+ if (!list_empty(&page_list)) {
+ spin_unlock_irq(&hugetlb_lock);
+ prep_and_add_allocated_folios(h, &page_list);
+ spin_lock_irq(&hugetlb_lock);
}
/*
@@ -3670,8 +3774,6 @@ static int set_max_huge_pages(struct hstate *h, unsigned long count, int nid,
* Collect pages to be removed on list without dropping lock
*/
while (min_count < persistent_huge_pages(h)) {
- struct folio *folio;
-
folio = remove_pool_hugetlb_folio(h, nodes_allowed, 0);
if (!folio)
break;
--
2.41.0
From: Joao Martins <[email protected]>
In an effort to minimize amount of TLB flushes, batch all PMD splits
belonging to a range of pages in order to perform only 1 (global) TLB
flush.
Add a flags field to the walker and pass whether it's a bulk allocation
or just a single page to decide to remap. First value
(VMEMMAP_SPLIT_NO_TLB_FLUSH) designates the request to not do the TLB
flush when we split the PMD.
Rebased and updated by Mike Kravetz
Signed-off-by: Joao Martins <[email protected]>
Signed-off-by: Mike Kravetz <[email protected]>
Reviewed-by: Muchun Song <[email protected]>
---
mm/hugetlb_vmemmap.c | 92 ++++++++++++++++++++++++++++++++++++++++++--
1 file changed, 88 insertions(+), 4 deletions(-)
diff --git a/mm/hugetlb_vmemmap.c b/mm/hugetlb_vmemmap.c
index 4ac521e596db..10739e4285d5 100644
--- a/mm/hugetlb_vmemmap.c
+++ b/mm/hugetlb_vmemmap.c
@@ -27,6 +27,8 @@
* @reuse_addr: the virtual address of the @reuse_page page.
* @vmemmap_pages: the list head of the vmemmap pages that can be freed
* or is mapped from.
+ * @flags: used to modify behavior in vmemmap page table walking
+ * operations.
*/
struct vmemmap_remap_walk {
void (*remap_pte)(pte_t *pte, unsigned long addr,
@@ -35,9 +37,13 @@ struct vmemmap_remap_walk {
struct page *reuse_page;
unsigned long reuse_addr;
struct list_head *vmemmap_pages;
+
+/* Skip the TLB flush when we split the PMD */
+#define VMEMMAP_SPLIT_NO_TLB_FLUSH BIT(0)
+ unsigned long flags;
};
-static int split_vmemmap_huge_pmd(pmd_t *pmd, unsigned long start)
+static int split_vmemmap_huge_pmd(pmd_t *pmd, unsigned long start, bool flush)
{
pmd_t __pmd;
int i;
@@ -80,7 +86,8 @@ static int split_vmemmap_huge_pmd(pmd_t *pmd, unsigned long start)
/* Make pte visible before pmd. See comment in pmd_install(). */
smp_wmb();
pmd_populate_kernel(&init_mm, pmd, pgtable);
- flush_tlb_kernel_range(start, start + PMD_SIZE);
+ if (flush)
+ flush_tlb_kernel_range(start, start + PMD_SIZE);
} else {
pte_free_kernel(&init_mm, pgtable);
}
@@ -127,11 +134,20 @@ static int vmemmap_pmd_range(pud_t *pud, unsigned long addr,
do {
int ret;
- ret = split_vmemmap_huge_pmd(pmd, addr & PMD_MASK);
+ ret = split_vmemmap_huge_pmd(pmd, addr & PMD_MASK,
+ !(walk->flags & VMEMMAP_SPLIT_NO_TLB_FLUSH));
if (ret)
return ret;
next = pmd_addr_end(addr, end);
+
+ /*
+ * We are only splitting, not remapping the hugetlb vmemmap
+ * pages.
+ */
+ if (!walk->remap_pte)
+ continue;
+
vmemmap_pte_range(pmd, addr, next, walk);
} while (pmd++, addr = next, addr != end);
@@ -198,7 +214,8 @@ static int vmemmap_remap_range(unsigned long start, unsigned long end,
return ret;
} while (pgd++, addr = next, addr != end);
- flush_tlb_kernel_range(start, end);
+ if (walk->remap_pte)
+ flush_tlb_kernel_range(start, end);
return 0;
}
@@ -297,6 +314,36 @@ static void vmemmap_restore_pte(pte_t *pte, unsigned long addr,
set_pte_at(&init_mm, addr, pte, mk_pte(page, pgprot));
}
+/**
+ * vmemmap_remap_split - split the vmemmap virtual address range [@start, @end)
+ * backing PMDs of the directmap into PTEs
+ * @start: start address of the vmemmap virtual address range that we want
+ * to remap.
+ * @end: end address of the vmemmap virtual address range that we want to
+ * remap.
+ * @reuse: reuse address.
+ *
+ * Return: %0 on success, negative error code otherwise.
+ */
+static int vmemmap_remap_split(unsigned long start, unsigned long end,
+ unsigned long reuse)
+{
+ int ret;
+ struct vmemmap_remap_walk walk = {
+ .remap_pte = NULL,
+ .flags = VMEMMAP_SPLIT_NO_TLB_FLUSH,
+ };
+
+ /* See the comment in the vmemmap_remap_free(). */
+ BUG_ON(start - reuse != PAGE_SIZE);
+
+ mmap_read_lock(&init_mm);
+ ret = vmemmap_remap_range(reuse, end, &walk);
+ mmap_read_unlock(&init_mm);
+
+ return ret;
+}
+
/**
* vmemmap_remap_free - remap the vmemmap virtual address range [@start, @end)
* to the page which @reuse is mapped to, then free vmemmap
@@ -320,6 +367,7 @@ static int vmemmap_remap_free(unsigned long start, unsigned long end,
.remap_pte = vmemmap_remap_pte,
.reuse_addr = reuse,
.vmemmap_pages = vmemmap_pages,
+ .flags = 0,
};
int nid = page_to_nid((struct page *)reuse);
gfp_t gfp_mask = GFP_KERNEL | __GFP_NORETRY | __GFP_NOWARN;
@@ -368,6 +416,7 @@ static int vmemmap_remap_free(unsigned long start, unsigned long end,
.remap_pte = vmemmap_restore_pte,
.reuse_addr = reuse,
.vmemmap_pages = vmemmap_pages,
+ .flags = 0,
};
vmemmap_remap_range(reuse, end, &walk);
@@ -419,6 +468,7 @@ static int vmemmap_remap_alloc(unsigned long start, unsigned long end,
.remap_pte = vmemmap_restore_pte,
.reuse_addr = reuse,
.vmemmap_pages = &vmemmap_pages,
+ .flags = 0,
};
/* See the comment in the vmemmap_remap_free(). */
@@ -628,11 +678,45 @@ void hugetlb_vmemmap_optimize(const struct hstate *h, struct page *head)
free_vmemmap_page_list(&vmemmap_pages);
}
+static int hugetlb_vmemmap_split(const struct hstate *h, struct page *head)
+{
+ unsigned long vmemmap_start = (unsigned long)head, vmemmap_end;
+ unsigned long vmemmap_reuse;
+
+ if (!vmemmap_should_optimize(h, head))
+ return 0;
+
+ vmemmap_end = vmemmap_start + hugetlb_vmemmap_size(h);
+ vmemmap_reuse = vmemmap_start;
+ vmemmap_start += HUGETLB_VMEMMAP_RESERVE_SIZE;
+
+ /*
+ * Split PMDs on the vmemmap virtual address range [@vmemmap_start,
+ * @vmemmap_end]
+ */
+ return vmemmap_remap_split(vmemmap_start, vmemmap_end, vmemmap_reuse);
+}
+
void hugetlb_vmemmap_optimize_folios(struct hstate *h, struct list_head *folio_list)
{
struct folio *folio;
LIST_HEAD(vmemmap_pages);
+ list_for_each_entry(folio, folio_list, lru) {
+ int ret = hugetlb_vmemmap_split(h, &folio->page);
+
+ /*
+ * Spliting the PMD requires allocating a page, thus lets fail
+ * early once we encounter the first OOM. No point in retrying
+ * as it can be dynamically done on remap with the memory
+ * we get back from the vmemmap deduplication.
+ */
+ if (ret == -ENOMEM)
+ break;
+ }
+
+ flush_tlb_all();
+
list_for_each_entry(folio, folio_list, lru) {
int ret = __hugetlb_vmemmap_optimize(h, &folio->page,
&vmemmap_pages);
--
2.41.0
The routine update_and_free_pages_bulk already performs vmemmap
restoration on the list of hugetlb pages in a separate step. In
preparation for more functionality to be added in this step, create a
new routine hugetlb_vmemmap_restore_folios() that will restore
vmemmap for a list of folios.
This new routine must provide sufficient feedback about errors and
actual restoration performed so that update_and_free_pages_bulk can
perform optimally.
Special care must be taken when encountering an error from
hugetlb_vmemmap_restore_folios. We want to continue making as much
forward progress as possible. A new routine bulk_vmemmap_restore_error
handles this specific situation.
Signed-off-by: Mike Kravetz <[email protected]>
Reviewed-by: Muchun Song <[email protected]>
---
mm/hugetlb.c | 99 +++++++++++++++++++++++++++++++-------------
mm/hugetlb_vmemmap.c | 38 +++++++++++++++++
mm/hugetlb_vmemmap.h | 11 +++++
3 files changed, 120 insertions(+), 28 deletions(-)
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 8b171f866d0a..cf834bb7f820 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -1859,50 +1859,93 @@ static void update_and_free_hugetlb_folio(struct hstate *h, struct folio *folio,
schedule_work(&free_hpage_work);
}
-static void update_and_free_pages_bulk(struct hstate *h, struct list_head *list)
+static void bulk_vmemmap_restore_error(struct hstate *h,
+ struct list_head *folio_list,
+ struct list_head *non_hvo_folios)
{
struct folio *folio, *t_folio;
- bool clear_dtor = false;
- /*
- * First allocate required vmemmmap (if necessary) for all folios on
- * list. If vmemmap can not be allocated, we can not free folio to
- * lower level allocator, so add back as hugetlb surplus page.
- * add_hugetlb_folio() removes the page from THIS list.
- * Use clear_dtor to note if vmemmap was successfully allocated for
- * ANY page on the list.
- */
- list_for_each_entry_safe(folio, t_folio, list, lru) {
- if (folio_test_hugetlb_vmemmap_optimized(folio)) {
+ if (!list_empty(non_hvo_folios)) {
+ /*
+ * Free any restored hugetlb pages so that restore of the
+ * entire list can be retried.
+ * The idea is that in the common case of ENOMEM errors freeing
+ * hugetlb pages with vmemmap we will free up memory so that we
+ * can allocate vmemmap for more hugetlb pages.
+ */
+ 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);
+ spin_unlock_irq(&hugetlb_lock);
+ update_and_free_hugetlb_folio(h, folio, false);
+ cond_resched();
+ }
+ } else {
+ /*
+ * In the case where there are no folios which can be
+ * immediately freed, we loop through the list trying to restore
+ * vmemmap individually in the hope that someone elsewhere may
+ * have done something to cause success (such as freeing some
+ * memory). If unable to restore a hugetlb page, the hugetlb
+ * page is made a surplus page and removed from the list.
+ * If are able to restore vmemmap and free one hugetlb page, we
+ * quit processing the list to retry the bulk operation.
+ */
+ list_for_each_entry_safe(folio, t_folio, folio_list, lru)
if (hugetlb_vmemmap_restore(h, &folio->page)) {
+ list_del(&folio->lru);
spin_lock_irq(&hugetlb_lock);
add_hugetlb_folio(h, folio, true);
spin_unlock_irq(&hugetlb_lock);
- } else
- clear_dtor = true;
- }
+ } else {
+ list_del(&folio->lru);
+ spin_lock_irq(&hugetlb_lock);
+ __clear_hugetlb_destructor(h, folio);
+ spin_unlock_irq(&hugetlb_lock);
+ update_and_free_hugetlb_folio(h, folio, false);
+ cond_resched();
+ break;
+ }
}
+}
+
+static void update_and_free_pages_bulk(struct hstate *h,
+ struct list_head *folio_list)
+{
+ long ret;
+ struct folio *folio, *t_folio;
+ LIST_HEAD(non_hvo_folios);
/*
- * If vmemmmap allocation was performed on any folio above, take lock
- * to clear destructor of all folios on list. This avoids the need to
- * lock/unlock for each individual folio.
- * The assumption is vmemmap allocation was performed on all or none
- * of the folios on the list. This is true expect in VERY rare cases.
+ * First allocate required vmemmmap (if necessary) for all folios.
+ * Carefully handle errors and free up any available hugetlb pages
+ * in an effort to make forward progress.
*/
- if (clear_dtor) {
+retry:
+ ret = hugetlb_vmemmap_restore_folios(h, folio_list, &non_hvo_folios);
+ if (ret < 0) {
+ bulk_vmemmap_restore_error(h, folio_list, &non_hvo_folios);
+ goto retry;
+ }
+
+ /*
+ * At this point, list should be empty, ret should be >= 0 and there
+ * 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.
+ */
+ 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, list, lru)
+ list_for_each_entry(folio, &non_hvo_folios, lru)
__clear_hugetlb_destructor(h, folio);
spin_unlock_irq(&hugetlb_lock);
}
- /*
- * Free folios back to low level allocators. vmemmap and destructors
- * were taken care of above, so update_and_free_hugetlb_folio will
- * not need to take hugetlb lock.
- */
- list_for_each_entry_safe(folio, t_folio, list, lru) {
+ list_for_each_entry_safe(folio, t_folio, &non_hvo_folios, lru) {
update_and_free_hugetlb_folio(h, folio, false);
cond_resched();
}
diff --git a/mm/hugetlb_vmemmap.c b/mm/hugetlb_vmemmap.c
index 4558b814ffab..77f44b81ff01 100644
--- a/mm/hugetlb_vmemmap.c
+++ b/mm/hugetlb_vmemmap.c
@@ -480,6 +480,44 @@ int hugetlb_vmemmap_restore(const struct hstate *h, struct page *head)
return ret;
}
+/**
+ * hugetlb_vmemmap_restore_folios - restore vmemmap for every folio on the list.
+ * @h: hstate.
+ * @folio_list: list of folios.
+ * @non_hvo_folios: Output list of folios for which vmemmap exists.
+ *
+ * Return: number of folios for which vmemmap was restored, or an error code
+ * if an error was encountered restoring vmemmap for a folio.
+ * Folios that have vmemmap are moved to the non_hvo_folios
+ * list. Processing of entries stops when the first error is
+ * encountered. The folio that experienced the error and all
+ * non-processed folios will remain on folio_list.
+ */
+long hugetlb_vmemmap_restore_folios(const struct hstate *h,
+ struct list_head *folio_list,
+ struct list_head *non_hvo_folios)
+{
+ struct folio *folio, *t_folio;
+ long restored = 0;
+ long ret = 0;
+
+ list_for_each_entry_safe(folio, t_folio, folio_list, lru) {
+ if (folio_test_hugetlb_vmemmap_optimized(folio)) {
+ ret = hugetlb_vmemmap_restore(h, &folio->page);
+ if (ret)
+ break;
+ restored++;
+ }
+
+ /* Add non-optimized folios to output list */
+ list_move(&folio->lru, non_hvo_folios);
+ }
+
+ if (!ret)
+ ret = restored;
+ return ret;
+}
+
/* Return true iff a HugeTLB whose vmemmap should and can be optimized. */
static bool vmemmap_should_optimize(const struct hstate *h, const struct page *head)
{
diff --git a/mm/hugetlb_vmemmap.h b/mm/hugetlb_vmemmap.h
index c512e388dbb4..a0dcf49f46ba 100644
--- a/mm/hugetlb_vmemmap.h
+++ b/mm/hugetlb_vmemmap.h
@@ -19,6 +19,9 @@
#ifdef CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP
int hugetlb_vmemmap_restore(const struct hstate *h, struct page *head);
+long hugetlb_vmemmap_restore_folios(const struct hstate *h,
+ struct list_head *folio_list,
+ struct list_head *non_hvo_folios);
void hugetlb_vmemmap_optimize(const struct hstate *h, struct page *head);
void hugetlb_vmemmap_optimize_folios(struct hstate *h, struct list_head *folio_list);
@@ -45,6 +48,14 @@ static inline int hugetlb_vmemmap_restore(const struct hstate *h, struct page *h
return 0;
}
+static long hugetlb_vmemmap_restore_folios(const struct hstate *h,
+ struct list_head *folio_list,
+ struct list_head *non_hvo_folios)
+{
+ list_splice_init(folio_list, non_hvo_folios);
+ return 0;
+}
+
static inline void hugetlb_vmemmap_optimize(const struct hstate *h, struct page *head)
{
}
--
2.41.0
When adding hugetlb pages to the pool, we first create a list of the
allocated pages before adding to the pool. Pass this list of pages to a
new routine hugetlb_vmemmap_optimize_folios() for vmemmap optimization.
Due to significant differences in vmemmmap initialization for bootmem
allocated hugetlb pages, a new routine prep_and_add_bootmem_folios
is created.
We also modify the routine vmemmap_should_optimize() to check for pages
that are already optimized. There are code paths that might request
vmemmap optimization twice and we want to make sure this is not
attempted.
Signed-off-by: Mike Kravetz <[email protected]>
---
mm/hugetlb.c | 43 +++++++++++++++++++++++++++++++++++--------
mm/hugetlb_vmemmap.c | 11 +++++++++++
mm/hugetlb_vmemmap.h | 5 +++++
3 files changed, 51 insertions(+), 8 deletions(-)
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 559f7c71c596..8b171f866d0a 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -2282,6 +2282,9 @@ static void prep_and_add_allocated_folios(struct hstate *h,
unsigned long flags;
struct folio *folio, *tmp_f;
+ /* Send list for bulk vmemmap optimization processing */
+ hugetlb_vmemmap_optimize_folios(h, folio_list);
+
/* Add all new pool pages to free lists in one lock cycle */
spin_lock_irqsave(&hugetlb_lock, flags);
list_for_each_entry_safe(folio, tmp_f, folio_list, lru) {
@@ -3344,6 +3347,35 @@ static void __init hugetlb_folio_init_vmemmap(struct folio *folio,
prep_compound_head((struct page *)folio, huge_page_order(h));
}
+static void __init prep_and_add_bootmem_folios(struct hstate *h,
+ struct list_head *folio_list)
+{
+ unsigned long flags;
+ struct folio *folio, *tmp_f;
+
+ /* Send list for bulk vmemmap optimization processing */
+ hugetlb_vmemmap_optimize_folios(h, folio_list);
+
+ /* Add all new pool pages to free lists in one lock cycle */
+ spin_lock_irqsave(&hugetlb_lock, flags);
+ list_for_each_entry_safe(folio, tmp_f, folio_list, lru) {
+ if (!folio_test_hugetlb_vmemmap_optimized(folio)) {
+ /*
+ * If HVO fails, initialize all tail struct pages
+ * We do not worry about potential long lock hold
+ * time as this is early in boot and there should
+ * be no contention.
+ */
+ hugetlb_folio_init_tail_vmemmap(folio,
+ HUGETLB_VMEMMAP_RESERVE_PAGES,
+ pages_per_huge_page(h));
+ }
+ __prep_account_new_huge_page(h, folio_nid(folio));
+ enqueue_hugetlb_folio(h, folio);
+ }
+ spin_unlock_irqrestore(&hugetlb_lock, flags);
+}
+
/*
* Put bootmem huge pages into the standard lists after mem_map is up.
* Note: This only applies to gigantic (order > MAX_ORDER) pages.
@@ -3364,7 +3396,7 @@ static void __init gather_bootmem_prealloc(void)
* in this list. If so, process each size separately.
*/
if (h != prev_h && prev_h != NULL)
- prep_and_add_allocated_folios(prev_h, &folio_list);
+ prep_and_add_bootmem_folios(prev_h, &folio_list);
prev_h = h;
VM_BUG_ON(!hstate_is_gigantic(h));
@@ -3372,12 +3404,7 @@ static void __init gather_bootmem_prealloc(void)
hugetlb_folio_init_vmemmap(folio, h,
HUGETLB_VMEMMAP_RESERVE_PAGES);
- __prep_new_hugetlb_folio(h, folio);
- /* If HVO fails, initialize all tail struct pages */
- if (!HPageVmemmapOptimized(&folio->page))
- hugetlb_folio_init_tail_vmemmap(folio,
- HUGETLB_VMEMMAP_RESERVE_PAGES,
- pages_per_huge_page(h));
+ init_new_hugetlb_folio(h, folio);
list_add(&folio->lru, &folio_list);
/*
@@ -3389,7 +3416,7 @@ static void __init gather_bootmem_prealloc(void)
cond_resched();
}
- prep_and_add_allocated_folios(h, &folio_list);
+ prep_and_add_bootmem_folios(h, &folio_list);
}
static void __init hugetlb_hstate_alloc_pages_onenode(struct hstate *h, int nid)
diff --git a/mm/hugetlb_vmemmap.c b/mm/hugetlb_vmemmap.c
index 76682d1d79a7..4558b814ffab 100644
--- a/mm/hugetlb_vmemmap.c
+++ b/mm/hugetlb_vmemmap.c
@@ -483,6 +483,9 @@ int hugetlb_vmemmap_restore(const struct hstate *h, struct page *head)
/* Return true iff a HugeTLB whose vmemmap should and can be optimized. */
static bool vmemmap_should_optimize(const struct hstate *h, const struct page *head)
{
+ if (HPageVmemmapOptimized((struct page *)head))
+ return false;
+
if (!READ_ONCE(vmemmap_optimize_enabled))
return false;
@@ -572,6 +575,14 @@ void hugetlb_vmemmap_optimize(const struct hstate *h, struct page *head)
SetHPageVmemmapOptimized(head);
}
+void hugetlb_vmemmap_optimize_folios(struct hstate *h, struct list_head *folio_list)
+{
+ struct folio *folio;
+
+ list_for_each_entry(folio, folio_list, lru)
+ hugetlb_vmemmap_optimize(h, &folio->page);
+}
+
static struct ctl_table hugetlb_vmemmap_sysctls[] = {
{
.procname = "hugetlb_optimize_vmemmap",
diff --git a/mm/hugetlb_vmemmap.h b/mm/hugetlb_vmemmap.h
index 4573899855d7..c512e388dbb4 100644
--- a/mm/hugetlb_vmemmap.h
+++ b/mm/hugetlb_vmemmap.h
@@ -20,6 +20,7 @@
#ifdef CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP
int hugetlb_vmemmap_restore(const struct hstate *h, struct page *head);
void hugetlb_vmemmap_optimize(const struct hstate *h, struct page *head);
+void hugetlb_vmemmap_optimize_folios(struct hstate *h, struct list_head *folio_list);
static inline unsigned int hugetlb_vmemmap_size(const struct hstate *h)
{
@@ -48,6 +49,10 @@ static inline void hugetlb_vmemmap_optimize(const struct hstate *h, struct page
{
}
+static inline void hugetlb_vmemmap_optimize_folios(struct hstate *h, struct list_head *folio_list)
+{
+}
+
static inline unsigned int hugetlb_vmemmap_optimizable_size(const struct hstate *h)
{
return 0;
--
2.41.0
From: Joao Martins <[email protected]>
Now that a list of pages is deduplicated at once, the TLB
flush can be batched for all vmemmap pages that got remapped.
Expand the flags field value to pass whether to skip the TLB flush
on remap of the PTE.
The TLB flush is global as we don't have guarantees from caller
that the set of folios is contiguous, or to add complexity in
composing a list of kVAs to flush.
Modified by Mike Kravetz to perform TLB flush on single folio if an
error is encountered.
Signed-off-by: Joao Martins <[email protected]>
Signed-off-by: Mike Kravetz <[email protected]>
Reviewed-by: Muchun Song <[email protected]>
---
mm/hugetlb_vmemmap.c | 49 ++++++++++++++++++++++++++++++++++----------
1 file changed, 38 insertions(+), 11 deletions(-)
diff --git a/mm/hugetlb_vmemmap.c b/mm/hugetlb_vmemmap.c
index 10739e4285d5..9df350372046 100644
--- a/mm/hugetlb_vmemmap.c
+++ b/mm/hugetlb_vmemmap.c
@@ -40,6 +40,8 @@ struct vmemmap_remap_walk {
/* Skip the TLB flush when we split the PMD */
#define VMEMMAP_SPLIT_NO_TLB_FLUSH BIT(0)
+/* Skip the TLB flush when we remap the PTE */
+#define VMEMMAP_REMAP_NO_TLB_FLUSH BIT(1)
unsigned long flags;
};
@@ -214,7 +216,7 @@ static int vmemmap_remap_range(unsigned long start, unsigned long end,
return ret;
} while (pgd++, addr = next, addr != end);
- if (walk->remap_pte)
+ if (walk->remap_pte && !(walk->flags & VMEMMAP_REMAP_NO_TLB_FLUSH))
flush_tlb_kernel_range(start, end);
return 0;
@@ -355,19 +357,21 @@ static int vmemmap_remap_split(unsigned long start, unsigned long end,
* @reuse: reuse address.
* @vmemmap_pages: list to deposit vmemmap pages to be freed. It is callers
* responsibility to free pages.
+ * @flags: modifications to vmemmap_remap_walk flags
*
* Return: %0 on success, negative error code otherwise.
*/
static int vmemmap_remap_free(unsigned long start, unsigned long end,
unsigned long reuse,
- struct list_head *vmemmap_pages)
+ struct list_head *vmemmap_pages,
+ unsigned long flags)
{
int ret;
struct vmemmap_remap_walk walk = {
.remap_pte = vmemmap_remap_pte,
.reuse_addr = reuse,
.vmemmap_pages = vmemmap_pages,
- .flags = 0,
+ .flags = flags,
};
int nid = page_to_nid((struct page *)reuse);
gfp_t gfp_mask = GFP_KERNEL | __GFP_NORETRY | __GFP_NOWARN;
@@ -629,7 +633,8 @@ static bool vmemmap_should_optimize(const struct hstate *h, const struct page *h
static int __hugetlb_vmemmap_optimize(const struct hstate *h,
struct page *head,
- struct list_head *vmemmap_pages)
+ struct list_head *vmemmap_pages,
+ unsigned long flags)
{
int ret = 0;
unsigned long vmemmap_start = (unsigned long)head, vmemmap_end;
@@ -640,6 +645,18 @@ static int __hugetlb_vmemmap_optimize(const struct hstate *h,
return ret;
static_branch_inc(&hugetlb_optimize_vmemmap_key);
+ /*
+ * Very Subtle
+ * If VMEMMAP_REMAP_NO_TLB_FLUSH is set, TLB flushing is not performed
+ * immediately after remapping. As a result, subsequent accesses
+ * and modifications to struct pages associated with the hugetlb
+ * page could be to the OLD struct pages. Set the vmemmap optimized
+ * flag here so that it is copied to the new head page. This keeps
+ * the old and new struct pages in sync.
+ * If there is an error during optimization, we will immediately FLUSH
+ * the TLB and clear the flag below.
+ */
+ SetHPageVmemmapOptimized(head);
vmemmap_end = vmemmap_start + hugetlb_vmemmap_size(h);
vmemmap_reuse = vmemmap_start;
@@ -651,11 +668,12 @@ static int __hugetlb_vmemmap_optimize(const struct hstate *h,
* mapping the range to vmemmap_pages list so that they can be freed by
* the caller.
*/
- ret = vmemmap_remap_free(vmemmap_start, vmemmap_end, vmemmap_reuse, vmemmap_pages);
- if (ret)
+ ret = vmemmap_remap_free(vmemmap_start, vmemmap_end, vmemmap_reuse,
+ vmemmap_pages, flags);
+ if (ret) {
static_branch_dec(&hugetlb_optimize_vmemmap_key);
- else
- SetHPageVmemmapOptimized(head);
+ ClearHPageVmemmapOptimized(head);
+ }
return ret;
}
@@ -674,7 +692,7 @@ void hugetlb_vmemmap_optimize(const struct hstate *h, struct page *head)
{
LIST_HEAD(vmemmap_pages);
- __hugetlb_vmemmap_optimize(h, head, &vmemmap_pages);
+ __hugetlb_vmemmap_optimize(h, head, &vmemmap_pages, 0);
free_vmemmap_page_list(&vmemmap_pages);
}
@@ -719,19 +737,28 @@ void hugetlb_vmemmap_optimize_folios(struct hstate *h, struct list_head *folio_l
list_for_each_entry(folio, folio_list, lru) {
int ret = __hugetlb_vmemmap_optimize(h, &folio->page,
- &vmemmap_pages);
+ &vmemmap_pages,
+ VMEMMAP_REMAP_NO_TLB_FLUSH);
/*
* Pages to be freed may have been accumulated. If we
* encounter an ENOMEM, free what we have and try again.
+ * This can occur in the case that both spliting fails
+ * halfway and head page allocation also failed. In this
+ * case __hugetlb_vmemmap_optimize() would free memory
+ * allowing more vmemmap remaps to occur.
*/
if (ret == -ENOMEM && !list_empty(&vmemmap_pages)) {
+ flush_tlb_all();
free_vmemmap_page_list(&vmemmap_pages);
INIT_LIST_HEAD(&vmemmap_pages);
- __hugetlb_vmemmap_optimize(h, &folio->page, &vmemmap_pages);
+ __hugetlb_vmemmap_optimize(h, &folio->page,
+ &vmemmap_pages,
+ VMEMMAP_REMAP_NO_TLB_FLUSH);
}
}
+ flush_tlb_all();
free_vmemmap_page_list(&vmemmap_pages);
}
--
2.41.0
Update the internal hugetlb restore vmemmap code path such that TLB
flushing can be batched. Use the existing mechanism of passing the
VMEMMAP_REMAP_NO_TLB_FLUSH flag to indicate flushing should not be
performed for individual pages. The routine hugetlb_vmemmap_restore_folios
is the only user of this new mechanism, and it will perform a global
flush after all vmemmap is restored.
Signed-off-by: Joao Martins <[email protected]>
Signed-off-by: Mike Kravetz <[email protected]>
Reviewed-by: Muchun Song <[email protected]>
---
mm/hugetlb_vmemmap.c | 39 ++++++++++++++++++++++++---------------
1 file changed, 24 insertions(+), 15 deletions(-)
diff --git a/mm/hugetlb_vmemmap.c b/mm/hugetlb_vmemmap.c
index 9df350372046..d2999c303031 100644
--- a/mm/hugetlb_vmemmap.c
+++ b/mm/hugetlb_vmemmap.c
@@ -461,18 +461,19 @@ static int alloc_vmemmap_page_list(unsigned long start, unsigned long end,
* @end: end address of the vmemmap virtual address range that we want to
* remap.
* @reuse: reuse address.
+ * @flags: modifications to vmemmap_remap_walk flags
*
* Return: %0 on success, negative error code otherwise.
*/
static int vmemmap_remap_alloc(unsigned long start, unsigned long end,
- unsigned long reuse)
+ unsigned long reuse, unsigned long flags)
{
LIST_HEAD(vmemmap_pages);
struct vmemmap_remap_walk walk = {
.remap_pte = vmemmap_restore_pte,
.reuse_addr = reuse,
.vmemmap_pages = &vmemmap_pages,
- .flags = 0,
+ .flags = flags,
};
/* See the comment in the vmemmap_remap_free(). */
@@ -494,17 +495,7 @@ EXPORT_SYMBOL(hugetlb_optimize_vmemmap_key);
static bool vmemmap_optimize_enabled = IS_ENABLED(CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP_DEFAULT_ON);
core_param(hugetlb_free_vmemmap, vmemmap_optimize_enabled, bool, 0);
-/**
- * hugetlb_vmemmap_restore - restore previously optimized (by
- * hugetlb_vmemmap_optimize()) vmemmap pages which
- * will be reallocated and remapped.
- * @h: struct hstate.
- * @head: the head page whose vmemmap pages will be restored.
- *
- * Return: %0 if @head's vmemmap pages have been reallocated and remapped,
- * negative error code otherwise.
- */
-int hugetlb_vmemmap_restore(const struct hstate *h, struct page *head)
+static int __hugetlb_vmemmap_restore(const struct hstate *h, struct page *head, unsigned long flags)
{
int ret;
unsigned long vmemmap_start = (unsigned long)head, vmemmap_end;
@@ -525,7 +516,7 @@ int hugetlb_vmemmap_restore(const struct hstate *h, struct page *head)
* When a HugeTLB page is freed to the buddy allocator, previously
* discarded vmemmap pages must be allocated and remapping.
*/
- ret = vmemmap_remap_alloc(vmemmap_start, vmemmap_end, vmemmap_reuse);
+ ret = vmemmap_remap_alloc(vmemmap_start, vmemmap_end, vmemmap_reuse, flags);
if (!ret) {
ClearHPageVmemmapOptimized(head);
static_branch_dec(&hugetlb_optimize_vmemmap_key);
@@ -534,6 +525,21 @@ int hugetlb_vmemmap_restore(const struct hstate *h, struct page *head)
return ret;
}
+/**
+ * hugetlb_vmemmap_restore - restore previously optimized (by
+ * hugetlb_vmemmap_optimize()) vmemmap pages which
+ * will be reallocated and remapped.
+ * @h: struct hstate.
+ * @head: the head page whose vmemmap pages will be restored.
+ *
+ * Return: %0 if @head's vmemmap pages have been reallocated and remapped,
+ * negative error code otherwise.
+ */
+int hugetlb_vmemmap_restore(const struct hstate *h, struct page *head)
+{
+ return __hugetlb_vmemmap_restore(h, head, 0);
+}
+
/**
* hugetlb_vmemmap_restore_folios - restore vmemmap for every folio on the list.
* @h: hstate.
@@ -557,7 +563,8 @@ long hugetlb_vmemmap_restore_folios(const struct hstate *h,
list_for_each_entry_safe(folio, t_folio, folio_list, lru) {
if (folio_test_hugetlb_vmemmap_optimized(folio)) {
- ret = hugetlb_vmemmap_restore(h, &folio->page);
+ ret = __hugetlb_vmemmap_restore(h, &folio->page,
+ VMEMMAP_REMAP_NO_TLB_FLUSH);
if (ret)
break;
restored++;
@@ -567,6 +574,8 @@ long hugetlb_vmemmap_restore_folios(const struct hstate *h,
list_move(&folio->lru, non_hvo_folios);
}
+ if (restored)
+ flush_tlb_all();
if (!ret)
ret = restored;
return ret;
--
2.41.0
> On Oct 19, 2023, at 10:31, Mike Kravetz <[email protected]> wrote:
>
> Allocation of a hugetlb page for the hugetlb pool is done by the routine
> alloc_pool_huge_page. This routine will allocate contiguous pages from
> a low level allocator, prep the pages for usage as a hugetlb page and
> then add the resulting hugetlb page to the pool.
>
> In the 'prep' stage, optional vmemmap optimization is done. For
> performance reasons we want to perform vmemmap optimization on multiple
> hugetlb pages at once. To do this, restructure the hugetlb pool
> allocation code such that vmemmap optimization can be isolated and later
> batched.
>
> The code to allocate hugetlb pages from bootmem was also modified to
> allow batching.
>
> No functional changes, only code restructure.
>
> Signed-off-by: Mike Kravetz <[email protected]>
The difference is small compared with v7. So
Reviewed-by: Muchun Song <[email protected]>
Thanks.
> On Oct 19, 2023, at 10:31, Mike Kravetz <[email protected]> wrote:
>
> When adding hugetlb pages to the pool, we first create a list of the
> allocated pages before adding to the pool. Pass this list of pages to a
> new routine hugetlb_vmemmap_optimize_folios() for vmemmap optimization.
>
> Due to significant differences in vmemmmap initialization for bootmem
> allocated hugetlb pages, a new routine prep_and_add_bootmem_folios
> is created.
>
> We also modify the routine vmemmap_should_optimize() to check for pages
> that are already optimized. There are code paths that might request
> vmemmap optimization twice and we want to make sure this is not
> attempted.
>
> Signed-off-by: Mike Kravetz <[email protected]>
Reviewed-by: Muchun Song <[email protected]>
Thanks.
On (23/10/18 19:31), Mike Kravetz wrote:
> Allocation of a hugetlb page for the hugetlb pool is done by the routine
> alloc_pool_huge_page. This routine will allocate contiguous pages from
> a low level allocator, prep the pages for usage as a hugetlb page and
> then add the resulting hugetlb page to the pool.
>
> In the 'prep' stage, optional vmemmap optimization is done. For
> performance reasons we want to perform vmemmap optimization on multiple
> hugetlb pages at once. To do this, restructure the hugetlb pool
> allocation code such that vmemmap optimization can be isolated and later
> batched.
>
> The code to allocate hugetlb pages from bootmem was also modified to
> allow batching.
>
> No functional changes, only code restructure.
>
> Signed-off-by: Mike Kravetz <[email protected]>
Tested-by: Sergey Senozhatsky <[email protected]>
Hi, Mike,
On 10/18/2023 7:31 PM, Mike Kravetz wrote:
> From: Joao Martins <[email protected]>
>
> Now that a list of pages is deduplicated at once, the TLB
> flush can be batched for all vmemmap pages that got remapped.
>
[..]
>
> @@ -719,19 +737,28 @@ void hugetlb_vmemmap_optimize_folios(struct hstate *h, struct list_head *folio_l
>
> list_for_each_entry(folio, folio_list, lru) {
> int ret = __hugetlb_vmemmap_optimize(h, &folio->page,
> - &vmemmap_pages);
> + &vmemmap_pages,
> + VMEMMAP_REMAP_NO_TLB_FLUSH);
>
> /*
> * Pages to be freed may have been accumulated. If we
> * encounter an ENOMEM, free what we have and try again.
> + * This can occur in the case that both spliting fails
> + * halfway and head page allocation also failed. In this
> + * case __hugetlb_vmemmap_optimize() would free memory
> + * allowing more vmemmap remaps to occur.
> */
> if (ret == -ENOMEM && !list_empty(&vmemmap_pages)) {
> + flush_tlb_all();
> free_vmemmap_page_list(&vmemmap_pages);
> INIT_LIST_HEAD(&vmemmap_pages);
> - __hugetlb_vmemmap_optimize(h, &folio->page, &vmemmap_pages);
> + __hugetlb_vmemmap_optimize(h, &folio->page,
> + &vmemmap_pages,
> + VMEMMAP_REMAP_NO_TLB_FLUSH);
> }
> }
>
> + flush_tlb_all();
It seems that if folio_list is empty, we could spend a tlb flush here.
perhaps it's worth to check against empty list up front and return ?
thanks,
-jane
> free_vmemmap_page_list(&vmemmap_pages);
> }
>
On 10/21/23 11:20, Jane Chu wrote:
> Hi, Mike,
>
> On 10/18/2023 7:31 PM, Mike Kravetz wrote:
> > From: Joao Martins <[email protected]>
> >
> > Now that a list of pages is deduplicated at once, the TLB
> > flush can be batched for all vmemmap pages that got remapped.
> >
> [..]
>
> > @@ -719,19 +737,28 @@ void hugetlb_vmemmap_optimize_folios(struct hstate *h, struct list_head *folio_l
> > list_for_each_entry(folio, folio_list, lru) {
> > int ret = __hugetlb_vmemmap_optimize(h, &folio->page,
> > - &vmemmap_pages);
> > + &vmemmap_pages,
> > + VMEMMAP_REMAP_NO_TLB_FLUSH);
> > /*
> > * Pages to be freed may have been accumulated. If we
> > * encounter an ENOMEM, free what we have and try again.
> > + * This can occur in the case that both spliting fails
> > + * halfway and head page allocation also failed. In this
> > + * case __hugetlb_vmemmap_optimize() would free memory
> > + * allowing more vmemmap remaps to occur.
> > */
> > if (ret == -ENOMEM && !list_empty(&vmemmap_pages)) {
> > + flush_tlb_all();
> > free_vmemmap_page_list(&vmemmap_pages);
> > INIT_LIST_HEAD(&vmemmap_pages);
> > - __hugetlb_vmemmap_optimize(h, &folio->page, &vmemmap_pages);
> > + __hugetlb_vmemmap_optimize(h, &folio->page,
> > + &vmemmap_pages,
> > + VMEMMAP_REMAP_NO_TLB_FLUSH);
> > }
> > }
> > + flush_tlb_all();
>
> It seems that if folio_list is empty, we could spend a tlb flush here.
> perhaps it's worth to check against empty list up front and return ?
Good point.
hugetlb_vmemmap_optimize_folios is only called from
prep_and_add_allocated_folios and prep_and_add_bootmem_folios. I
previously thought about adding a check like the following at the
beginning of those routines.
if (list_empty(folio_list))
return;
However that seemed like over optimizing. But, such a check would avoid
the tlb flush as you point out above as well as an unnecessary
hugetlb_lock lock/unlock cycle.
We can add something like this as an optimization. I am not too concerned
about this right now because these these routines are generally called very
infrequently as the result of a user request to change the size of hugetlb
pools.
--
Mike Kravetz