linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] mm: bail out when the PMD has been set in bloom filter
@ 2026-02-27  7:52 zhaoyang.huang
  2026-02-27 11:42 ` kernel test robot
  0 siblings, 1 reply; 3+ messages in thread
From: zhaoyang.huang @ 2026-02-27  7:52 UTC (permalink / raw)
  To: Andrew Morton, Yu Zhao, linux-mm, linux-kernel, Zhaoyang Huang,
	steve.kang

From: Zhaoyang Huang <zhaoyang.huang@unisoc.com>

There are two reasons to have the recorded PMD bail out from doing
the following iteration
1. It is worth of doing such a trade off thing in terms of reclaiming
efficiency as test_bloom_filter only consume 20~30 instructions in modern
processors(25 instructions in ARM64).
2. The PMD needs to accumulate young pages until aging happens while the
new arrived folio reference checking under current max_seq refuse to do so
which will affect carrying hot PMDs to new generation.

Signed-off-by: Zhaoyang Huang <zhaoyang.huang@unisoc.com>
---
 mm/vmscan.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 9d900be478ea..e50e98291d0d 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -4227,6 +4227,10 @@ bool lru_gen_look_around(struct page_vma_mapped_walk *pvmw)
 	/* avoid taking the LRU lock under the PTL when possible */
 	walk = current->reclaim_state ? current->reclaim_state->mm_walk : NULL;
 
+	/* may the pmd has been set in bloom filter */
+	if (test_bloom_filter(mm_state, max_seq, pvmw->pmd))
+		return true;
+
 	start = max(addr & PMD_MASK, vma->vm_start);
 	end = min(addr | ~PMD_MASK, vma->vm_end - 1) + 1;
 
-- 
2.25.1



^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH] mm: bail out when the PMD has been set in bloom filter
  2026-02-27  7:52 [PATCH] mm: bail out when the PMD has been set in bloom filter zhaoyang.huang
@ 2026-02-27 11:42 ` kernel test robot
  2026-02-28  0:50   ` Zhaoyang Huang
  0 siblings, 1 reply; 3+ messages in thread
From: kernel test robot @ 2026-02-27 11:42 UTC (permalink / raw)
  To: zhaoyang.huang, Andrew Morton, Yu Zhao, linux-kernel,
	Zhaoyang Huang, steve.kang
  Cc: llvm, oe-kbuild-all, Linux Memory Management List

Hi zhaoyang.huang,

kernel test robot noticed the following build warnings:

[auto build test WARNING on akpm-mm/mm-everything]

url:    https://github.com/intel-lab-lkp/linux/commits/zhaoyang-huang/mm-bail-out-when-the-PMD-has-been-set-in-bloom-filter/20260227-155729
base:   https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-everything
patch link:    https://lore.kernel.org/r/20260227075250.1128175-1-zhaoyang.huang%40unisoc.com
patch subject: [PATCH] mm: bail out when the PMD has been set in bloom filter
config: sparc64-randconfig-002-20260227 (https://download.01.org/0day-ci/archive/20260227/202602271916.OBNa34QU-lkp@intel.com/config)
compiler: clang version 23.0.0git (https://github.com/llvm/llvm-project 9a109fbb6e184ec9bcce10615949f598f4c974a9)
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260227/202602271916.OBNa34QU-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202602271916.OBNa34QU-lkp@intel.com/

All warnings (new ones prefixed by >>):

>> mm/vmscan.c:4206:24: warning: variable 'mm_state' is uninitialized when used here [-Wuninitialized]
    4206 |         if (test_bloom_filter(mm_state, max_seq, pvmw->pmd))
         |                               ^~~~~~~~
   mm/vmscan.c:4185:35: note: initialize the variable 'mm_state' to silence this warning
    4185 |         struct lru_gen_mm_state *mm_state;
         |                                          ^
         |                                           = NULL
>> mm/vmscan.c:4206:34: warning: variable 'max_seq' is uninitialized when used here [-Wuninitialized]
    4206 |         if (test_bloom_filter(mm_state, max_seq, pvmw->pmd))
         |                                         ^~~~~~~
   mm/vmscan.c:4186:23: note: initialize the variable 'max_seq' to silence this warning
    4186 |         unsigned long max_seq;
         |                              ^
         |                               = 0
   2 warnings generated.


vim +/mm_state +4206 mm/vmscan.c

  4157	
  4158	/******************************************************************************
  4159	 *                          rmap/PT walk feedback
  4160	 ******************************************************************************/
  4161	
  4162	/*
  4163	 * This function exploits spatial locality when shrink_folio_list() walks the
  4164	 * rmap. It scans the adjacent PTEs of a young PTE and promotes hot pages. If
  4165	 * the scan was done cacheline efficiently, it adds the PMD entry pointing to
  4166	 * the PTE table to the Bloom filter. This forms a feedback loop between the
  4167	 * eviction and the aging.
  4168	 */
  4169	bool lru_gen_look_around(struct page_vma_mapped_walk *pvmw)
  4170	{
  4171		int i;
  4172		bool dirty;
  4173		unsigned long start;
  4174		unsigned long end;
  4175		struct lru_gen_mm_walk *walk;
  4176		struct folio *last = NULL;
  4177		int young = 1;
  4178		pte_t *pte = pvmw->pte;
  4179		unsigned long addr = pvmw->address;
  4180		struct vm_area_struct *vma = pvmw->vma;
  4181		struct folio *folio = pfn_folio(pvmw->pfn);
  4182		struct mem_cgroup *memcg;
  4183		struct pglist_data *pgdat = folio_pgdat(folio);
  4184		struct lruvec *lruvec;
  4185		struct lru_gen_mm_state *mm_state;
  4186		unsigned long max_seq;
  4187		int gen;
  4188	
  4189		lockdep_assert_held(pvmw->ptl);
  4190		VM_WARN_ON_ONCE_FOLIO(folio_test_lru(folio), folio);
  4191	
  4192		if (!ptep_clear_young_notify(vma, addr, pte))
  4193			return false;
  4194	
  4195		if (spin_is_contended(pvmw->ptl))
  4196			return true;
  4197	
  4198		/* exclude special VMAs containing anon pages from COW */
  4199		if (vma->vm_flags & VM_SPECIAL)
  4200			return true;
  4201	
  4202		/* avoid taking the LRU lock under the PTL when possible */
  4203		walk = current->reclaim_state ? current->reclaim_state->mm_walk : NULL;
  4204	
  4205		/* may the pmd has been set in bloom filter */
> 4206		if (test_bloom_filter(mm_state, max_seq, pvmw->pmd))
  4207			return true;
  4208	
  4209		start = max(addr & PMD_MASK, vma->vm_start);
  4210		end = min(addr | ~PMD_MASK, vma->vm_end - 1) + 1;
  4211	
  4212		if (end - start == PAGE_SIZE)
  4213			return true;
  4214	
  4215		if (end - start > MIN_LRU_BATCH * PAGE_SIZE) {
  4216			if (addr - start < MIN_LRU_BATCH * PAGE_SIZE / 2)
  4217				end = start + MIN_LRU_BATCH * PAGE_SIZE;
  4218			else if (end - addr < MIN_LRU_BATCH * PAGE_SIZE / 2)
  4219				start = end - MIN_LRU_BATCH * PAGE_SIZE;
  4220			else {
  4221				start = addr - MIN_LRU_BATCH * PAGE_SIZE / 2;
  4222				end = addr + MIN_LRU_BATCH * PAGE_SIZE / 2;
  4223			}
  4224		}
  4225	
  4226		memcg = get_mem_cgroup_from_folio(folio);
  4227		lruvec = mem_cgroup_lruvec(memcg, pgdat);
  4228		max_seq = READ_ONCE((lruvec)->lrugen.max_seq);
  4229		gen = lru_gen_from_seq(max_seq);
  4230		mm_state = get_mm_state(lruvec);
  4231	
  4232		lazy_mmu_mode_enable();
  4233	
  4234		pte -= (addr - start) / PAGE_SIZE;
  4235	
  4236		for (i = 0, addr = start; addr != end; i++, addr += PAGE_SIZE) {
  4237			unsigned long pfn;
  4238			pte_t ptent = ptep_get(pte + i);
  4239	
  4240			pfn = get_pte_pfn(ptent, vma, addr, pgdat);
  4241			if (pfn == -1)
  4242				continue;
  4243	
  4244			folio = get_pfn_folio(pfn, memcg, pgdat);
  4245			if (!folio)
  4246				continue;
  4247	
  4248			if (!ptep_clear_young_notify(vma, addr, pte + i))
  4249				continue;
  4250	
  4251			if (last != folio) {
  4252				walk_update_folio(walk, last, gen, dirty);
  4253	
  4254				last = folio;
  4255				dirty = false;
  4256			}
  4257	
  4258			if (pte_dirty(ptent))
  4259				dirty = true;
  4260	
  4261			young++;
  4262		}
  4263	
  4264		walk_update_folio(walk, last, gen, dirty);
  4265	
  4266		lazy_mmu_mode_disable();
  4267	
  4268		/* feedback from rmap walkers to page table walkers */
  4269		if (mm_state && suitable_to_scan(i, young))
  4270			update_bloom_filter(mm_state, max_seq, pvmw->pmd);
  4271	
  4272		mem_cgroup_put(memcg);
  4273	
  4274		return true;
  4275	}
  4276	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH] mm: bail out when the PMD has been set in bloom filter
  2026-02-27 11:42 ` kernel test robot
@ 2026-02-28  0:50   ` Zhaoyang Huang
  0 siblings, 0 replies; 3+ messages in thread
From: Zhaoyang Huang @ 2026-02-28  0:50 UTC (permalink / raw)
  To: kernel test robot
  Cc: zhaoyang.huang, Andrew Morton, Yu Zhao, linux-kernel, steve.kang,
	llvm, oe-kbuild-all, Linux Memory Management List

On Fri, Feb 27, 2026 at 7:43 PM kernel test robot <lkp@intel.com> wrote:
>
> Hi zhaoyang.huang,
>
> kernel test robot noticed the following build warnings:
>
> [auto build test WARNING on akpm-mm/mm-everything]
>
> url:    https://github.com/intel-lab-lkp/linux/commits/zhaoyang-huang/mm-bail-out-when-the-PMD-has-been-set-in-bloom-filter/20260227-155729
> base:   https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-everything
> patch link:    https://lore.kernel.org/r/20260227075250.1128175-1-zhaoyang.huang%40unisoc.com
> patch subject: [PATCH] mm: bail out when the PMD has been set in bloom filter
> config: sparc64-randconfig-002-20260227 (https://download.01.org/0day-ci/archive/20260227/202602271916.OBNa34QU-lkp@intel.com/config)
> compiler: clang version 23.0.0git (https://github.com/llvm/llvm-project 9a109fbb6e184ec9bcce10615949f598f4c974a9)
> reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260227/202602271916.OBNa34QU-lkp@intel.com/reproduce)
>
> If you fix the issue in a separate patch/commit (i.e. not just a new version of
> the same patch/commit), kindly add following tags
> | Reported-by: kernel test robot <lkp@intel.com>
> | Closes: https://lore.kernel.org/oe-kbuild-all/202602271916.OBNa34QU-lkp@intel.com/
>
> All warnings (new ones prefixed by >>):
>
> >> mm/vmscan.c:4206:24: warning: variable 'mm_state' is uninitialized when used here [-Wuninitialized]
>     4206 |         if (test_bloom_filter(mm_state, max_seq, pvmw->pmd))
>          |                               ^~~~~~~~
>    mm/vmscan.c:4185:35: note: initialize the variable 'mm_state' to silence this warning
>     4185 |         struct lru_gen_mm_state *mm_state;
>          |                                          ^
>          |                                           = NULL
> >> mm/vmscan.c:4206:34: warning: variable 'max_seq' is uninitialized when used here [-Wuninitialized]
>     4206 |         if (test_bloom_filter(mm_state, max_seq, pvmw->pmd))
>          |                                         ^~~~~~~
>    mm/vmscan.c:4186:23: note: initialize the variable 'max_seq' to silence this warning
>     4186 |         unsigned long max_seq;
>          |                              ^
>          |                               = 0
>    2 warnings generated.
>
>
> vim +/mm_state +4206 mm/vmscan.c
>
>   4157
>   4158  /******************************************************************************
>   4159   *                          rmap/PT walk feedback
>   4160   ******************************************************************************/
>   4161
>   4162  /*
>   4163   * This function exploits spatial locality when shrink_folio_list() walks the
>   4164   * rmap. It scans the adjacent PTEs of a young PTE and promotes hot pages. If
>   4165   * the scan was done cacheline efficiently, it adds the PMD entry pointing to
>   4166   * the PTE table to the Bloom filter. This forms a feedback loop between the
>   4167   * eviction and the aging.
>   4168   */
>   4169  bool lru_gen_look_around(struct page_vma_mapped_walk *pvmw)
>   4170  {
>   4171          int i;
>   4172          bool dirty;
>   4173          unsigned long start;
>   4174          unsigned long end;
>   4175          struct lru_gen_mm_walk *walk;
>   4176          struct folio *last = NULL;
>   4177          int young = 1;
>   4178          pte_t *pte = pvmw->pte;
>   4179          unsigned long addr = pvmw->address;
>   4180          struct vm_area_struct *vma = pvmw->vma;
>   4181          struct folio *folio = pfn_folio(pvmw->pfn);
>   4182          struct mem_cgroup *memcg;
>   4183          struct pglist_data *pgdat = folio_pgdat(folio);
>   4184          struct lruvec *lruvec;
>   4185          struct lru_gen_mm_state *mm_state;
>   4186          unsigned long max_seq;
>   4187          int gen;
>   4188
>   4189          lockdep_assert_held(pvmw->ptl);
>   4190          VM_WARN_ON_ONCE_FOLIO(folio_test_lru(folio), folio);
>   4191
>   4192          if (!ptep_clear_young_notify(vma, addr, pte))
>   4193                  return false;
>   4194
>   4195          if (spin_is_contended(pvmw->ptl))
>   4196                  return true;
>   4197
>   4198          /* exclude special VMAs containing anon pages from COW */
>   4199          if (vma->vm_flags & VM_SPECIAL)
>   4200                  return true;
>   4201
>   4202          /* avoid taking the LRU lock under the PTL when possible */
>   4203          walk = current->reclaim_state ? current->reclaim_state->mm_walk : NULL;
>   4204
>   4205          /* may the pmd has been set in bloom filter */
> > 4206          if (test_bloom_filter(mm_state, max_seq, pvmw->pmd))
>   4207                  return true;
>   4208
>   4209          start = max(addr & PMD_MASK, vma->vm_start);
>   4210          end = min(addr | ~PMD_MASK, vma->vm_end - 1) + 1;
>   4211
>   4212          if (end - start == PAGE_SIZE)
>   4213                  return true;
>   4214
>   4215          if (end - start > MIN_LRU_BATCH * PAGE_SIZE) {
>   4216                  if (addr - start < MIN_LRU_BATCH * PAGE_SIZE / 2)
>   4217                          end = start + MIN_LRU_BATCH * PAGE_SIZE;
>   4218                  else if (end - addr < MIN_LRU_BATCH * PAGE_SIZE / 2)
>   4219                          start = end - MIN_LRU_BATCH * PAGE_SIZE;
>   4220                  else {
>   4221                          start = addr - MIN_LRU_BATCH * PAGE_SIZE / 2;
>   4222                          end = addr + MIN_LRU_BATCH * PAGE_SIZE / 2;
>   4223                  }
>   4224          }
>   4225
>   4226          memcg = get_mem_cgroup_from_folio(folio);
>   4227          lruvec = mem_cgroup_lruvec(memcg, pgdat);
>   4228          max_seq = READ_ONCE((lruvec)->lrugen.max_seq);
>   4229          gen = lru_gen_from_seq(max_seq);
>   4230          mm_state = get_mm_state(lruvec);
I am confused about the code base which this test is based on. By
checking the base that listed above and linux-next and 7.1-rc1, I
can't find above merged result or I miss anything?
>   4231
>   4232          lazy_mmu_mode_enable();
>   4233
>   4234          pte -= (addr - start) / PAGE_SIZE;
>   4235
>   4236          for (i = 0, addr = start; addr != end; i++, addr += PAGE_SIZE) {
>   4237                  unsigned long pfn;
>   4238                  pte_t ptent = ptep_get(pte + i);
>   4239
>   4240                  pfn = get_pte_pfn(ptent, vma, addr, pgdat);
>   4241                  if (pfn == -1)
>   4242                          continue;
>   4243
>   4244                  folio = get_pfn_folio(pfn, memcg, pgdat);
>   4245                  if (!folio)
>   4246                          continue;
>   4247
>   4248                  if (!ptep_clear_young_notify(vma, addr, pte + i))
>   4249                          continue;
>   4250
>   4251                  if (last != folio) {
>   4252                          walk_update_folio(walk, last, gen, dirty);
>   4253
>   4254                          last = folio;
>   4255                          dirty = false;
>   4256                  }
>   4257
>   4258                  if (pte_dirty(ptent))
>   4259                          dirty = true;
>   4260
>   4261                  young++;
>   4262          }
>   4263
>   4264          walk_update_folio(walk, last, gen, dirty);
>   4265
>   4266          lazy_mmu_mode_disable();
>   4267
>   4268          /* feedback from rmap walkers to page table walkers */
>   4269          if (mm_state && suitable_to_scan(i, young))
>   4270                  update_bloom_filter(mm_state, max_seq, pvmw->pmd);
>   4271
>   4272          mem_cgroup_put(memcg);
>   4273
>   4274          return true;
>   4275  }
>   4276
>
> --
> 0-DAY CI Kernel Test Service
> https://github.com/intel/lkp-tests/wiki


^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2026-02-28  0:50 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-02-27  7:52 [PATCH] mm: bail out when the PMD has been set in bloom filter zhaoyang.huang
2026-02-27 11:42 ` kernel test robot
2026-02-28  0:50   ` Zhaoyang Huang

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox