The patch titled
Subject: mm/filemap: remove hugetlb special casing in filemap.c
has been added to the -mm mm-unstable branch. Its filename is
mm-filemap-remove-hugetlb-special-casing-in-filemapc.patch
This patch will shortly appear at
https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patches/mm-filemap-remove-hugetlb-special-casing-in-filemapc.patch
This patch will later appear in the mm-unstable branch at
git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/process/submit-checklist.rst when testing your code ***
The -mm tree is included into linux-next via the mm-everything
branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
and is updated there every 2-3 working days
------------------------------------------------------
From: Sidhartha Kumar <sidhartha.kumar@xxxxxxxxxx>
Subject: mm/filemap: remove hugetlb special casing in filemap.c
Date: Mon, 11 Sep 2023 14:53:19 -0700
Remove special cased hugetlb handling code within the page cache by
changing the granularity of ->index to the base page size rather than the
huge page size. The motivation of this patch is to reduce complexity
within the filemap code while also increasing performance by removing
branches that are evaluated on every page cache lookup.
To support the change in index, new wrappers for hugetlb page cache
interactions are added. These wrappers perform the conversion to a linear
index which is now expected by the page cache for huge pages.
========================= PERFORMANCE ======================================
Perf was used to check the performance differences after the patch.
Overall the performance is similar to mainline with a very small larger
overhead that occurs in __filemap_add_folio() and
hugetlb_add_to_page_cache(). This is because of the larger overhead that
occurs in xa_load() and xa_store() as the xarray is now using more entries
to store hugetlb folios in the page cache.
Timing
aarch64
2MB Page Size
6.5-rc3 + this patch:
[root@sidhakum-ol9-1 hugepages]# time fallocate -l 700GB test.txt
real 1m49.568s
user 0m0.000s
sys 1m49.461s
6.5-rc3:
[root]# time fallocate -l 700GB test.txt
real 1m47.495s
user 0m0.000s
sys 1m47.370s
1GB Page Size
6.5-rc3 + this patch:
[root@sidhakum-ol9-1 hugepages1G]# time fallocate -l 700GB test.txt
real 1m47.024s
user 0m0.000s
sys 1m46.921s
6.5-rc3:
[root@sidhakum-ol9-1 hugepages1G]# time fallocate -l 700GB test.txt
real 1m44.551s
user 0m0.000s
sys 1m44.438s
x86
2MB Page Size
6.5-rc3 + this patch:
[root@sidhakum-ol9-2 hugepages]# time fallocate -l 100GB test.txt
real 0m22.383s
user 0m0.000s
sys 0m22.255s
6.5-rc3:
[opc@sidhakum-ol9-2 hugepages]$ time sudo fallocate -l 100GB /dev/hugepages/test.txt
real 0m22.735s
user 0m0.038s
sys 0m22.567s
1GB Page Size
6.5-rc3 + this patch:
[root@sidhakum-ol9-2 hugepages1GB]# time fallocate -l 100GB test.txt
real 0m25.786s
user 0m0.001s
sys 0m25.589s
6.5-rc3:
[root@sidhakum-ol9-2 hugepages1G]# time fallocate -l 100GB test.txt
real 0m33.454s
user 0m0.001s
sys 0m33.193s
aarch64:
workload - fallocate a 700GB file backed by huge pages
6.5-rc3 + this patch:
2MB Page Size:
--100.00%--__arm64_sys_fallocate
ksys_fallocate
vfs_fallocate
hugetlbfs_fallocate
|
|--95.04%--__pi_clear_page
|
|--3.57%--clear_huge_page
| |
| |--2.63%--rcu_all_qs
| |
| --0.91%--__cond_resched
|
--0.67%--__cond_resched
0.17% 0.00% 0 fallocate [kernel.vmlinux] [k] hugetlb_add_to_page_cache
0.14% 0.10% 11 fallocate [kernel.vmlinux] [k] __filemap_add_folio
6.5-rc3
2MB Page Size:
--100.00%--__arm64_sys_fallocate
ksys_fallocate
vfs_fallocate
hugetlbfs_fallocate
|
|--94.91%--__pi_clear_page
|
|--4.11%--clear_huge_page
| |
| |--3.00%--rcu_all_qs
| |
| --1.10%--__cond_resched
|
--0.59%--__cond_resched
0.08% 0.01% 1 fallocate [kernel.kallsyms] [k] hugetlb_add_to_page_cache
0.05% 0.03% 3 fallocate [kernel.kallsyms] [k] __filemap_add_folio
x86
workload - fallocate a 100GB file backed by huge pages
6.5-rc3 + this patch:
2MB Page Size:
hugetlbfs_fallocate
|
--99.57%--clear_huge_page
|
--98.47%--clear_page_erms
|
--0.53%--asm_sysvec_apic_timer_interrupt
0.04% 0.04% 1 fallocate [kernel.kallsyms] [k] xa_load
0.04% 0.00% 0 fallocate [kernel.kallsyms] [k] hugetlb_add_to_page_cache
0.04% 0.00% 0 fallocate [kernel.kallsyms] [k] __filemap_add_folio
0.04% 0.00% 0 fallocate [kernel.kallsyms] [k] xas_store
6.5-rc3
2MB Page Size:
--99.93%--__x64_sys_fallocate
vfs_fallocate
hugetlbfs_fallocate
|
--99.38%--clear_huge_page
|
|--98.40%--clear_page_erms
|
--0.59%--__cond_resched
0.03% 0.03% 1 fallocate [kernel.kallsyms] [k] __filemap_add_folio
========================= TESTING ======================================
This patch passes libhugetlbfs tests and LTP hugetlb tests
********** TEST SUMMARY
* 2M
* 32-bit 64-bit
* Total testcases: 110 113
* Skipped: 0 0
* PASS: 107 113
* FAIL: 0 0
* Killed by signal: 3 0
* Bad configuration: 0 0
* Expected FAIL: 0 0
* Unexpected PASS: 0 0
* Test not present: 0 0
* Strange test result: 0 0
**********
###############################################################
Done executing testcases.
LTP Version: 20220527-178-g2761a81c4
###############################################################
page migration was also tested using Mike Kravetz's test program.[8]
[1]: https://lore.kernel.org/linux-mm/20230519220142.212051-1-sidhartha.kumar@xxxxxxxxxx/T/
[2]: https://lore.kernel.org/lkml/20230609194947.37196-1-sidhartha.kumar@xxxxxxxxxx/
[3]: https://lore.kernel.org/lkml/ZLtVlJA+V2+2yjxc@xxxxxxxxxxxxxxxxxxxx/T/
[4]: https://lore.kernel.org/lkml/20230811233939.GA105247@monkey/T/
[5]: https://lore.kernel.org/linux-mm/202308151231.L2pfzOmu-lkp@xxxxxxxxx/T/#t
[6]: https://lore.kernel.org/linux-mm/20230815191932.273054-1-sidhartha.kumar@xxxxxxxxxx/T/
[7]: https://lore.kernel.org/linux-mm/ZOTtK1Kfa5gCS3Gb@xxxxxxxxxxxxxxxxxxxx/T/
[8]: https://lore.kernel.org/linux-mm/20230907001832.GA63356@monkey/T/#m35ed80fc2bb21d613bc4cc8771159d6ae0b5a264
Link: https://lkml.kernel.org/r/20230911215319.644224-1-sidhartha.kumar@xxxxxxxxxx
Signed-off-by: Sidhartha Kumar <sidhartha.kumar@xxxxxxxxxx>
Cc: Matthew Wilcox (Oracle) <willy@xxxxxxxxxxxxx>
Cc: Mike Kravetz <mike.kravetz@xxxxxxxxxx>
Cc: Muchun Song <songmuchun@xxxxxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
fs/hugetlbfs/inode.c | 15 ++++++++-------
include/linux/hugetlb.h | 12 ++++++++++++
include/linux/pagemap.h | 32 ++------------------------------
mm/filemap.c | 34 ++++++++++------------------------
mm/hugetlb.c | 32 ++++++--------------------------
mm/migrate.c | 6 +++---
6 files changed, 41 insertions(+), 90 deletions(-)
--- a/fs/hugetlbfs/inode.c~mm-filemap-remove-hugetlb-special-casing-in-filemapc
+++ a/fs/hugetlbfs/inode.c
@@ -661,21 +661,20 @@ static void remove_inode_hugepages(struc
{
struct hstate *h = hstate_inode(inode);
struct address_space *mapping = &inode->i_data;
- const pgoff_t start = lstart >> huge_page_shift(h);
- const pgoff_t end = lend >> huge_page_shift(h);
+ const pgoff_t end = lend >> PAGE_SHIFT;
struct folio_batch fbatch;
pgoff_t next, index;
int i, freed = 0;
bool truncate_op = (lend == LLONG_MAX);
folio_batch_init(&fbatch);
- next = start;
+ next = lstart >> PAGE_SHIFT;
while (filemap_get_folios(mapping, &next, end - 1, &fbatch)) {
for (i = 0; i < folio_batch_count(&fbatch); ++i) {
struct folio *folio = fbatch.folios[i];
u32 hash = 0;
- index = folio->index;
+ index = folio->index >> huge_page_order(h);
hash = hugetlb_fault_mutex_hash(mapping, index);
mutex_lock(&hugetlb_fault_mutex_table[hash]);
@@ -693,7 +692,9 @@ static void remove_inode_hugepages(struc
}
if (truncate_op)
- (void)hugetlb_unreserve_pages(inode, start, LONG_MAX, freed);
+ (void)hugetlb_unreserve_pages(inode,
+ lstart >> huge_page_shift(h),
+ LONG_MAX, freed);
}
static void hugetlbfs_evict_inode(struct inode *inode)
@@ -741,7 +742,7 @@ static void hugetlbfs_zero_partial_page(
pgoff_t idx = start >> huge_page_shift(h);
struct folio *folio;
- folio = filemap_lock_folio(mapping, idx);
+ folio = filemap_lock_hugetlb_folio(h, mapping, idx);
if (IS_ERR(folio))
return;
@@ -886,7 +887,7 @@ static long hugetlbfs_fallocate(struct f
mutex_lock(&hugetlb_fault_mutex_table[hash]);
/* See if already present in mapping to avoid alloc/free */
- folio = filemap_get_folio(mapping, index);
+ folio = filemap_get_folio(mapping, index << huge_page_order(h));
if (!IS_ERR(folio)) {
folio_put(folio);
mutex_unlock(&hugetlb_fault_mutex_table[hash]);
--- a/include/linux/hugetlb.h~mm-filemap-remove-hugetlb-special-casing-in-filemapc
+++ a/include/linux/hugetlb.h
@@ -812,6 +812,12 @@ static inline unsigned int blocks_per_hu
return huge_page_size(h) / 512;
}
+static inline struct folio *filemap_lock_hugetlb_folio(struct hstate *h,
+ struct address_space *mapping, pgoff_t idx)
+{
+ return filemap_lock_folio(mapping, idx << huge_page_order(h));
+}
+
#include <asm/hugetlb.h>
#ifndef is_hugepage_only_range
@@ -1005,6 +1011,12 @@ static inline struct hugepage_subpool *h
{
return NULL;
}
+
+static inline struct folio *filemap_lock_hugetlb_folio(struct hstate *h,
+ struct address_space *mapping, pgoff_t idx)
+{
+ return NULL;
+}
static inline int isolate_or_dissolve_huge_page(struct page *page,
struct list_head *list)
--- a/include/linux/pagemap.h~mm-filemap-remove-hugetlb-special-casing-in-filemapc
+++ a/include/linux/pagemap.h
@@ -789,9 +789,6 @@ static inline pgoff_t folio_next_index(s
*/
static inline struct page *folio_file_page(struct folio *folio, pgoff_t index)
{
- /* HugeTLBfs indexes the page cache in units of hpage_size */
- if (folio_test_hugetlb(folio))
- return &folio->page;
return folio_page(folio, index & (folio_nr_pages(folio) - 1));
}
@@ -807,9 +804,6 @@ static inline struct page *folio_file_pa
*/
static inline bool folio_contains(struct folio *folio, pgoff_t index)
{
- /* HugeTLBfs indexes the page cache in units of hpage_size */
- if (folio_test_hugetlb(folio))
- return folio->index == index;
return index - folio_index(folio) < folio_nr_pages(folio);
}
@@ -867,10 +861,9 @@ static inline struct folio *read_mapping
}
/*
- * Get index of the page within radix-tree (but not for hugetlb pages).
- * (TODO: remove once hugetlb pages will have ->index in PAGE_SIZE)
+ * Get the offset in PAGE_SIZE (even for hugetlb pages).
*/
-static inline pgoff_t page_to_index(struct page *page)
+static inline pgoff_t page_to_pgoff(struct page *page)
{
struct page *head;
@@ -885,19 +878,6 @@ static inline pgoff_t page_to_index(stru
return head->index + page - head;
}
-extern pgoff_t hugetlb_basepage_index(struct page *page);
-
-/*
- * Get the offset in PAGE_SIZE (even for hugetlb pages).
- * (TODO: hugetlb pages should have ->index in PAGE_SIZE)
- */
-static inline pgoff_t page_to_pgoff(struct page *page)
-{
- if (unlikely(PageHuge(page)))
- return hugetlb_basepage_index(page);
- return page_to_index(page);
-}
-
/*
* Return byte-offset into filesystem object for page.
*/
@@ -934,24 +914,16 @@ static inline loff_t folio_file_pos(stru
/*
* Get the offset in PAGE_SIZE (even for hugetlb folios).
- * (TODO: hugetlb folios should have ->index in PAGE_SIZE)
*/
static inline pgoff_t folio_pgoff(struct folio *folio)
{
- if (unlikely(folio_test_hugetlb(folio)))
- return hugetlb_basepage_index(&folio->page);
return folio->index;
}
-extern pgoff_t linear_hugepage_index(struct vm_area_struct *vma,
- unsigned long address);
-
static inline pgoff_t linear_page_index(struct vm_area_struct *vma,
unsigned long address)
{
pgoff_t pgoff;
- if (unlikely(is_vm_hugetlb_page(vma)))
- return linear_hugepage_index(vma, address);
pgoff = (address - vma->vm_start) >> PAGE_SHIFT;
pgoff += vma->vm_pgoff;
return pgoff;
--- a/mm/filemap.c~mm-filemap-remove-hugetlb-special-casing-in-filemapc
+++ a/mm/filemap.c
@@ -131,11 +131,8 @@ static void page_cache_delete(struct add
mapping_set_update(&xas, mapping);
- /* hugetlb pages are represented by a single entry in the xarray */
- if (!folio_test_hugetlb(folio)) {
- xas_set_order(&xas, folio->index, folio_order(folio));
- nr = folio_nr_pages(folio);
- }
+ xas_set_order(&xas, folio->index, folio_order(folio));
+ nr = folio_nr_pages(folio);
VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio);
@@ -234,7 +231,7 @@ void filemap_free_folio(struct address_s
if (free_folio)
free_folio(folio);
- if (folio_test_large(folio) && !folio_test_hugetlb(folio))
+ if (folio_test_large(folio))
refs = folio_nr_pages(folio);
folio_put_refs(folio, refs);
}
@@ -855,14 +852,15 @@ noinline int __filemap_add_folio(struct
if (!huge) {
int error = mem_cgroup_charge(folio, NULL, gfp);
- VM_BUG_ON_FOLIO(index & (folio_nr_pages(folio) - 1), folio);
if (error)
return error;
charged = true;
- xas_set_order(&xas, index, folio_order(folio));
- nr = folio_nr_pages(folio);
}
+ VM_BUG_ON_FOLIO(index & (folio_nr_pages(folio) - 1), folio);
+ xas_set_order(&xas, index, folio_order(folio));
+ nr = folio_nr_pages(folio);
+
gfp &= GFP_RECLAIM_MASK;
folio_ref_add(folio, nr);
folio->mapping = mapping;
@@ -2040,7 +2038,7 @@ unsigned find_get_entries(struct address
int idx = folio_batch_count(fbatch) - 1;
folio = fbatch->folios[idx];
- if (!xa_is_value(folio) && !folio_test_hugetlb(folio))
+ if (!xa_is_value(folio))
nr = folio_nr_pages(folio);
*start = indices[idx] + nr;
}
@@ -2104,7 +2102,7 @@ put:
int idx = folio_batch_count(fbatch) - 1;
folio = fbatch->folios[idx];
- if (!xa_is_value(folio) && !folio_test_hugetlb(folio))
+ if (!xa_is_value(folio))
nr = folio_nr_pages(folio);
*start = indices[idx] + nr;
}
@@ -2145,9 +2143,6 @@ unsigned filemap_get_folios(struct addre
continue;
if (!folio_batch_add(fbatch, folio)) {
unsigned long nr = folio_nr_pages(folio);
-
- if (folio_test_hugetlb(folio))
- nr = 1;
*start = folio->index + nr;
goto out;
}
@@ -2213,9 +2208,6 @@ unsigned filemap_get_folios_contig(struc
if (!folio_batch_add(fbatch, folio)) {
nr = folio_nr_pages(folio);
-
- if (folio_test_hugetlb(folio))
- nr = 1;
*start = folio->index + nr;
goto out;
}
@@ -2232,10 +2224,7 @@ update_start:
if (nr) {
folio = fbatch->folios[nr - 1];
- if (folio_test_hugetlb(folio))
- *start = folio->index + 1;
- else
- *start = folio_next_index(folio);
+ *start = folio->index + folio_nr_pages(folio);
}
out:
rcu_read_unlock();
@@ -2273,9 +2262,6 @@ unsigned filemap_get_folios_tag(struct a
continue;
if (!folio_batch_add(fbatch, folio)) {
unsigned long nr = folio_nr_pages(folio);
-
- if (folio_test_hugetlb(folio))
- nr = 1;
*start = folio->index + nr;
goto out;
}
--- a/mm/hugetlb.c~mm-filemap-remove-hugetlb-special-casing-in-filemapc
+++ a/mm/hugetlb.c
@@ -952,7 +952,7 @@ static long region_count(struct resv_map
/*
* Convert the address within this vma to the page offset within
- * the mapping, in pagecache page units; huge pages here.
+ * the mapping, huge page units here.
*/
static pgoff_t vma_hugecache_offset(struct hstate *h,
struct vm_area_struct *vma, unsigned long address)
@@ -961,13 +961,6 @@ static pgoff_t vma_hugecache_offset(stru
(vma->vm_pgoff >> huge_page_order(h));
}
-pgoff_t linear_hugepage_index(struct vm_area_struct *vma,
- unsigned long address)
-{
- return vma_hugecache_offset(hstate_vma(vma), vma, address);
-}
-EXPORT_SYMBOL_GPL(linear_hugepage_index);
-
/**
* vma_kernel_pagesize - Page size granularity for this VMA.
* @vma: The user mapping.
@@ -2076,20 +2069,6 @@ struct address_space *hugetlb_page_mappi
return NULL;
}
-pgoff_t hugetlb_basepage_index(struct page *page)
-{
- struct page *page_head = compound_head(page);
- pgoff_t index = page_index(page_head);
- unsigned long compound_idx;
-
- if (compound_order(page_head) > MAX_ORDER)
- compound_idx = page_to_pfn(page) - page_to_pfn(page_head);
- else
- compound_idx = page - page_head;
-
- return (index << compound_order(page_head)) + compound_idx;
-}
-
static struct folio *alloc_buddy_hugetlb_folio(struct hstate *h,
gfp_t gfp_mask, int nid, nodemask_t *nmask,
nodemask_t *node_alloc_noretry)
@@ -5722,7 +5701,7 @@ static bool hugetlbfs_pagecache_present(
struct vm_area_struct *vma, unsigned long address)
{
struct address_space *mapping = vma->vm_file->f_mapping;
- pgoff_t idx = vma_hugecache_offset(h, vma, address);
+ pgoff_t idx = linear_page_index(vma, address);
struct folio *folio;
folio = filemap_get_folio(mapping, idx);
@@ -5739,6 +5718,7 @@ int hugetlb_add_to_page_cache(struct fol
struct hstate *h = hstate_inode(inode);
int err;
+ idx <<= huge_page_order(h);
__folio_set_locked(folio);
err = __filemap_add_folio(mapping, folio, idx, GFP_KERNEL, NULL);
@@ -5846,7 +5826,7 @@ static vm_fault_t hugetlb_no_page(struct
* before we get page_table_lock.
*/
new_folio = false;
- folio = filemap_lock_folio(mapping, idx);
+ folio = filemap_lock_hugetlb_folio(h, mapping, idx);
if (IS_ERR(folio)) {
size = i_size_read(mapping->host) >> huge_page_shift(h);
if (idx >= size)
@@ -6155,7 +6135,7 @@ vm_fault_t hugetlb_fault(struct mm_struc
/* Just decrements count, does not deallocate */
vma_end_reservation(h, vma, haddr);
- pagecache_folio = filemap_lock_folio(mapping, idx);
+ pagecache_folio = filemap_lock_hugetlb_folio(h, mapping, idx);
if (IS_ERR(pagecache_folio))
pagecache_folio = NULL;
}
@@ -6287,7 +6267,7 @@ int hugetlb_mfill_atomic_pte(pte_t *dst_
if (is_continue) {
ret = -EFAULT;
- folio = filemap_lock_folio(mapping, idx);
+ folio = filemap_lock_hugetlb_folio(h, mapping, idx);
if (IS_ERR(folio))
goto out;
folio_in_pagecache = true;
--- a/mm/migrate.c~mm-filemap-remove-hugetlb-special-casing-in-filemapc
+++ a/mm/migrate.c
@@ -521,7 +521,7 @@ int migrate_huge_page_move_mapping(struc
int expected_count;
xas_lock_irq(&xas);
- expected_count = 2 + folio_has_private(src);
+ expected_count = folio_expected_refs(mapping, src);
if (!folio_ref_freeze(src, expected_count)) {
xas_unlock_irq(&xas);
return -EAGAIN;
@@ -530,11 +530,11 @@ int migrate_huge_page_move_mapping(struc
dst->index = src->index;
dst->mapping = src->mapping;
- folio_get(dst);
+ folio_ref_add(dst, folio_nr_pages(dst));
xas_store(&xas, dst);
- folio_ref_unfreeze(src, expected_count - 1);
+ folio_ref_unfreeze(src, expected_count - folio_nr_pages(src));
xas_unlock_irq(&xas);
_
Patches currently in -mm which might be from sidhartha.kumar@xxxxxxxxxx are
mm-filemap-remove-hugetlb-special-casing-in-filemapc.patch
