refactor and merge PTE scanning logic
Subsystems (2)
Lance Yang
reviewerv3 cover refactor and merge PTE scanning logic
Hi all,
This series cleans up the almost-duplicated PTE scanning logic in the collapse path.
The first one is a preparatory step that refactors both loops to use a single if-else-if-else-if chain for checking disjoint PTEs.
The second one replaces VM_BUG_ON_FOLIO() with a more graceful VM_WARN_ON_FOLIO() for handling non-anonymous folios.
The last one then extracts the common logic into a shared helper.
Thanks,
Lance
---
This series applies on top of patch[1], which is based on mm-new.
[1] https://lore.kernel.org/linux-mm/[email protected]
v2 -> v3:
- #02 Collect Reviewed-by from Wei and Dev - thanks!
- #03 Use vm_normal_folio() and drop struct page altogether (per Dev)
- #03 Move the cc parameter to the end (per Dev)
- https://lore.kernel.org/linux-mm/[email protected]
v1 -> v2:
- #01 Update the changelog (per Dev)
- #01 Collect Reviewed-by from Wei, Dev and Zi - thanks!
- #03 Make more of the scanning logic common between scan_pmd() and _isolate() (per Dev)
- https://lore.kernel.org/linux-mm/[email protected]
Lance Yang (3):
mm/khugepaged: optimize PTE scanning with if-else-if-else-if chain mm/khugepaged: use VM_WARN_ON_FOLIO instead of VM_BUG_ON_FOLIO for
non-anon folios
mm/khugepaged: merge PTE scanning logic into a new helper
mm/khugepaged.c | 242 ++++++++++++++++++++++++++---------------------- 1 file changed, 130 insertions(+), 112 deletions(-)
Signature · 1 line
2.49.0
Lance Yang
reviewerPATCH v3 1/3 mm/khugepaged: optimize PTE scanning with if-else-if-else-if chain
From: Lance Yang <[email protected]>
As pointed out by Dev, the PTE checks for disjoint conditions in the scanning loops can be optimized. is_swap_pte, (pte_none && is_zero_pfn), and pte_uffd_wp are mutually exclusive.
This patch refactors the loops in both __collapse_huge_page_isolate() and hpage_collapse_scan_pmd() to use a continuous if-else-if-else-if chain instead of separate if blocks. While at it, the redundant pte_present() check before is_zero_pfn() is also removed.
Also, this is a preparatory step to make it easier to merge the almost-duplicated scanning logic in these two functions, as suggested by David.
--- mm/khugepaged.c | 12 ++++-------- 1 file changed, 4 insertions(+), 8 deletions(-)
mm/khugepaged.c +4 -8
@@ -548,8 +548,7 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma,for (_pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) || (pte_present (pteval) && -is_zero_pfn (pte_pfn (pteval)))) { +if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { ++none_or_zero; if (!userfaultfd_armed(vma) && (!cc->is_khugepaged ||
@@ -560,12 +559,10 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma,count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); goto out; } - } -if (!pte_present (pteval)) { + }else if (!pte_present (pteval)) { result =SCAN_PTE_NON_PRESENT ; goto out; - } -if (pte_uffd_wp (pteval)) { + }else if (pte_uffd_wp (pteval)) { result =SCAN_PTE_UFFD_WP ; goto out; }
@@ -1321,8 +1318,7 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); goto out_unmap; } - } -if (pte_uffd_wp (pteval)) { + }else if (pte_uffd_wp (pteval)) {/* * Don't collapse the page if any of the small * PTEs are armed with uffd write protection.
Signature · 1 line
2.49.0
Lance Yang
reviewerPATCH v3 2/3 mm/khugepaged: use VM_WARN_ON_FOLIO instead of VM_BUG_ON_FOLIO for non-anon folios
In reply to Lance YangFrom: Lance Yang <[email protected]>
As Zi pointed out, we should avoid crashing the kernel for conditions that can be handled gracefully. Encountering a non-anonymous folio in an anonymous VMA is a bug, but a warning is sufficient.
This patch changes the VM_BUG_ON_FOLIO(!folio_test_anon(folio)) to a VM_WARN_ON_FOLIO() in both __collapse_huge_page_isolate() and hpage_collapse_scan_pmd(), and then aborts the scan with SCAN_PAGE_ANON.
Making more of the scanning logic common between hpage_collapse_scan_pmd() and __collapse_huge_page_isolate(), as suggested by Dev.
--- mm/khugepaged.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
mm/khugepaged.c +6 -1
@@ -573,7 +573,11 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma, } folio =page_folio (page); -VM_BUG_ON_FOLIO (!folio_test_anon (folio), folio); +if (!folio_test_anon (folio)) { +VM_WARN_ON_FOLIO (true, folio); + result =SCAN_PAGE_ANON ; + goto out; + }/* See hpage_collapse_scan_pmd(). */ if (folio_maybe_mapped_shared (folio)) {
@@ -1340,6 +1344,7 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm, folio =page_folio (page);if (!folio_test_anon (folio)) { +VM_WARN_ON_FOLIO (true, folio); result =SCAN_PAGE_ANON ; goto out_unmap; }
Signature · 1 line
2.49.0
Lance Yang
reviewerPATCH v3 3/3 mm/khugepaged: merge PTE scanning logic into a new helper
In reply to Lance YangFrom: Lance Yang <[email protected]>
As David suggested, the PTE scanning logic in hpage_collapse_scan_pmd() and __collapse_huge_page_isolate() was almost duplicated.
This patch cleans things up by moving all the common PTE checking logic into a new shared helper, thp_collapse_check_pte(). While at it, we use vm_normal_folio() instead of vm_normal_page().
--- mm/khugepaged.c | 243 ++++++++++++++++++++++++++---------------------- 1 file changed, 130 insertions(+), 113 deletions(-)
mm/khugepaged.c +130 -113
@@ -61,6 +61,12 @@ enum scan_result { SCAN_PAGE_FILLED, }; +enum pte_check_result { +PTE_CHECK_SUCCEED , +PTE_CHECK_CONTINUE , +PTE_CHECK_FAIL , +}; +#define CREATE_TRACE_POINTS #include <trace/events/huge_memory.h>
@@ -533,62 +539,139 @@ static void release_pte_pages(pte_t *pte, pte_t *_pte, } } +/* +* thp_collapse_check_pte - Check if a PTE is suitable for THP collapse +* @pte: The PTE to check +* @vma: The VMA the PTE belongs to +* @addr: The virtual address corresponding to this PTE +* @foliop: On success, used to return a pointer to the folio +* Must be non-NULL +* @none_or_zero: Counter for none/zero PTEs. Must be non-NULL +* @unmapped: Counter for swap PTEs. Can be NULL if not scanning swaps +* @shared: Counter for shared pages. Must be non-NULL +* @scan_result: Used to return the failure reason (SCAN_*) on a +* PTE_CHECK_FAIL return. Must be non-NULL +* @cc: Collapse control settings +* +* Returns: +* PTE_CHECK_SUCCEED - PTE is suitable, proceed with further checks +* PTE_CHECK_CONTINUE - Skip this PTE and continue scanning +* PTE_CHECK_FAIL - Abort collapse scan +*/ +static inline int thp_collapse_check_pte (pte_t pte,struct vm_area_struct *vma, +unsigned long addr,struct folio **foliop,int *none_or_zero, +int *unmapped,int *shared,int *scan_result, +struct collapse_control *cc) +{ +struct folio *folio =NULL ; + +if (pte_none (pte) ||is_zero_pfn (pte_pfn (pte))) { + (*none_or_zero)++; +if (!userfaultfd_armed (vma) && + (!cc->is_khugepaged || + *none_or_zero <= khugepaged_max_ptes_none)) { +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_NONE_PTE ; +count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (!pte_present (pte)) { +if (!unmapped) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + +if (non_swap_entry (pte_to_swp_entry (pte))) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + + (*unmapped)++; +if (!cc->is_khugepaged || + *unmapped <= khugepaged_max_ptes_swap) { +/* +* Always be strict with uffd-wp enabled swap +* entries. Please see comment below for +* pte_uffd_wp(). +*/ +if (pte_swp_uffd_wp (pte)) { + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_SWAP_PTE ; +count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (pte_uffd_wp (pte)) { +/* +* Don't collapse the page if any of the small PTEs are +* armed with uffd write protection. Here we can also mark +* the new huge pmd as write protected if any of the small +* ones is marked but that could bring unknown userfault +* messages that falls outside of the registered range. +* So, just be simple. +*/ + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } + + folio =vm_normal_folio (vma, addr, pte); +if (unlikely (!folio) ||unlikely (folio_is_zone_device (folio))) { + *scan_result =SCAN_PAGE_NULL ; +return PTE_CHECK_FAIL ; + } + +if (!folio_test_anon (folio)) { +VM_WARN_ON_FOLIO (true, folio); + *scan_result =SCAN_PAGE_ANON ; +return PTE_CHECK_FAIL ; + } + +/* +* We treat a single page as shared if any part of the THP +* is shared. +*/ +if (folio_maybe_mapped_shared (folio)) { + (*shared)++; +if (cc->is_khugepaged && *shared > khugepaged_max_ptes_shared) { + *scan_result =SCAN_EXCEED_SHARED_PTE ; +count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); +return PTE_CHECK_FAIL ; + } + } + + *foliop = folio; + +return PTE_CHECK_SUCCEED ; +} +static int __collapse_huge_page_isolate (struct vm_area_struct *vma,unsigned long start_addr,pte_t *pte,struct collapse_control *cc,struct list_head *compound_pagelist) { -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr = start_addr;pte_t *_pte;int none_or_zero =0 , shared =0 , result =SCAN_FAIL , referenced =0 ; +int pte_check_res;for (_pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out; - } - }else if (!pte_present (pteval)) { - result =SCAN_PTE_NON_PRESENT ; - goto out; - }else if (pte_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out; - } - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out; - } - folio =page_folio (page); -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; - goto out; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero,NULL , &shared, + &result, cc); -/* See hpage_collapse_scan_pmd(). */ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out; - } - } +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) + goto out;if (folio_test_large (folio)) {struct folio *f;
@@ -1264,11 +1347,11 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,pte_t *pte, *_pte;int result =SCAN_FAIL , referenced =0 ;int none_or_zero =0 , shared =0 ; -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr;spinlock_t *ptl;int node =NUMA_NO_NODE , unmapped =0 ; +int pte_check_res;VM_BUG_ON (start_addr & ~HPAGE_PMD_MASK );
@@ -1287,81 +1370,15 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,for (addr = start_addr, _pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out_unmap; - } - }else if (!pte_present (pteval)) { -if (non_swap_entry (pte_to_swp_entry (pteval))) { - result =SCAN_PTE_NON_PRESENT ; - goto out_unmap; - } - ++unmapped; -if (!cc->is_khugepaged || - unmapped <= khugepaged_max_ptes_swap) { -/* -* Always be strict with uffd-wp -* enabled swap entries. Please see -* comment below for pte_uffd_wp(). -*/ -if (pte_swp_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } -continue ; - }else { - result =SCAN_EXCEED_SWAP_PTE ; -count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); - goto out_unmap; - } - }else if (pte_uffd_wp (pteval)) { -/* -* Don't collapse the page if any of the small -* PTEs are armed with uffd write protection. -* Here we can also mark the new huge pmd as -* write protected if any of the small ones is -* marked but that could bring unknown -* userfault messages that falls outside of -* the registered range. So, just be simple. -*/ - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero, &unmapped, + &shared, &result, cc); - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out_unmap; - } - folio =page_folio (page); - -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) goto out_unmap; - } - -/* -* We treat a single page as shared if any part of the THP -* is shared. -*/ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out_unmap; - } - }/* * Record which node the original page is from and save this
Signature · 1 line
2.49.0
Andrew Morton
maintainerRe: [PATCH mm-new v3 3/3] mm/khugepaged: merge PTE scanning logic into a new helper
Quoted 7 lines · On Wed, 8 Oct 2025 12:37:48 +0800 Lance Yang <[email protected]> wrote:
(quoted diff) +7
-0+ if (!cc->is_khugepaged || + *unmapped <= khugepaged_max_ptes_swap) { + /* + * Always be strict with uffd-wp enabled swap + * entries. Please see comment below for + * pte_uffd_wp(). + */
Review
Quoted 10 lines
(quoted diff) +10
-0+ if (pte_swp_uffd_wp(pte)) { + *scan_result = SCAN_PTE_UFFD_WP; + return PTE_CHECK_FAIL; + } + return PTE_CHECK_CONTINUE; + } else { + *scan_result = SCAN_EXCEED_SWAP_PTE; + count_vm_event(THP_SCAN_EXCEED_SWAP_PTE); + return PTE_CHECK_FAIL; + }
I'm inclined to agree with checkpatch here.
WARNING: else is not generally useful after a break or return #81: FILE: mm/khugepaged.c:574:
+ return PTE_CHECK_CONTINUE;
+ } else {
did you see this and disagree or did you forget to run checkpatch?
mm/khugepaged.c renamed from mm/khugepaged.c~mm-khugepaged-merge-pte-scanning-logic-into-a-new-helper-checkpatch-fixes +6 -8
@@ -571,11 +571,10 @@ static inline int thp_collapse_check_pte (!cc->is_khugepaged || *none_or_zero <= khugepaged_max_ptes_none)) {return PTE_CHECK_CONTINUE ; - }else { - *scan_result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); -return PTE_CHECK_FAIL ; } + *scan_result =SCAN_EXCEED_NONE_PTE ; +count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); +return PTE_CHECK_FAIL ; }else if (!pte_present (pte)) {if (!unmapped) { *scan_result =SCAN_PTE_NON_PRESENT ;
@@ -600,11 +599,10 @@ static inline int thp_collapse_check_ptereturn PTE_CHECK_FAIL ; }return PTE_CHECK_CONTINUE ; - }else { - *scan_result =SCAN_EXCEED_SWAP_PTE ; -count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); -return PTE_CHECK_FAIL ; } + *scan_result =SCAN_EXCEED_SWAP_PTE ; +count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); +return PTE_CHECK_FAIL ; }else if (pte_uffd_wp (pte)) {/* * Don't collapse the page if any of the small PTEs are
_
Lance Yang
reviewerRe: [PATCH mm-new v3 3/3] mm/khugepaged: merge PTE scanning logic into a new helper
Quoted 13 lines · Andrew Morton
Quoted 11 lines · On Wed, 8 Oct 2025 12:37:48 +0800 Lance Yang <[email protected]> wrote:
(quoted diff) +11
-0+ if (!cc->is_khugepaged || + *unmapped <= khugepaged_max_ptes_swap) { + /* + * Always be strict with uffd-wp enabled swap + * entries. Please see comment below for + * pte_uffd_wp(). + */ + if (pte_swp_uffd_wp(pte)) { + *scan_result = SCAN_PTE_UFFD_WP; + return PTE_CHECK_FAIL; + }
Quoted 8 lines
mm/khugepaged.c static inline int thp_collapse_check_pte
Quoted 6 lines
mm/khugepaged.c renamed from mm/khugepaged.c~mm-khugepaged-merge-pte-scanning-logic-into-a-new-helper-checkpatch-fixes +6
-0+return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_SWAP_PTE ; +count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); +return PTE_CHECK_FAIL ; + }I'm inclined to agree with checkpatch here.
Thanks!
Quoted 7 lines
WARNING: else is not generally useful after a break or return #81: FILE:
mm/khugepaged.c:574:(quoted diff) +2
-0+ return PTE_CHECK_CONTINUE; + } else {did you see this and disagree or did you forget to run checkpatch?
Yes, I saw the warning. I kept the original style because this is just a code move ...
Quoted 35 lines
mm/khugepaged.c renamed from mm/khugepaged.c~mm-khugepaged-merge-pte-scanning-logic-into-a-new-helper-checkpatch-fixes +6
-8@@ -571,11 +571,10 @@ static inline int thp_collapse_check_pte (!cc->is_khugepaged || *none_or_zero <= khugepaged_max_ptes_none)) {return PTE_CHECK_CONTINUE ; - }else { - *scan_result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); -return PTE_CHECK_FAIL ; } + *scan_result =SCAN_EXCEED_NONE_PTE ; +count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); +return PTE_CHECK_FAIL ; }else if (!pte_present (pte)) {if (!unmapped) { *scan_result =SCAN_PTE_NON_PRESENT ;@@ -600,11 +599,10 @@ static inline int thp_collapse_check_ptereturn PTE_CHECK_FAIL ; }return PTE_CHECK_CONTINUE ; - }else { - *scan_result =SCAN_EXCEED_SWAP_PTE ; -count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); -return PTE_CHECK_FAIL ; } + *scan_result =SCAN_EXCEED_SWAP_PTE ; +count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); +return PTE_CHECK_FAIL ; }else if (pte_uffd_wp (pte)) {/* * Don't collapse the page if any of the small PTEs are _
Dev Jain
reviewerRe: [PATCH mm-new v3 3/3] mm/khugepaged: merge PTE scanning logic into a new helper
In reply to Lance YangQuoted 18 lines · Lance Yang
} }mm/khugepaged.c +15
-0+/* +* thp_collapse_check_pte - Check if a PTE is suitable for THP collapse +* @pte: The PTE to check +* @vma: The VMA the PTE belongs to +* @addr: The virtual address corresponding to this PTE +* @foliop: On success, used to return a pointer to the folio +* Must be non-NULL +* @none_or_zero: Counter for none/zero PTEs. Must be non-NULL +* @unmapped: Counter for swap PTEs. Can be NULL if not scanning swaps +* @shared: Counter for shared pages. Must be non-NULL +* @scan_result: Used to return the failure reason (SCAN_*) on a +* PTE_CHECK_FAIL return. Must be non-NULL +* @cc: Collapse control settings +* +* Returns:
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 566
Quoted 10 lines
mm/khugepaged.c +10
-0+* PTE_CHECK_SUCCEED - PTE is suitable, proceed with further checks +* PTE_CHECK_CONTINUE - Skip this PTE and continue scanning +* PTE_CHECK_FAIL - Abort collapse scan +*/ +static inline int thp_collapse_check_pte (pte_t pte,struct vm_area_struct *vma, +unsigned long addr,struct folio **foliop,int *none_or_zero, +int *unmapped,int *shared,int *scan_result, +struct collapse_control *cc) +{ +struct folio *folio =NULL ;
I think initialization is not needed here?Otherwise, LGTM. Reviewed-by: Dev Jain <[email protected]>
Lance Yang
reviewerRe: [PATCH mm-new v3 3/3] mm/khugepaged: merge PTE scanning logic into a new helper
Quoted 24 lines · Dev Jain
Quoted 22 lines · Lance Yang
} }+/*
+ *thp_collapse_check_pte- Check if a PTE is suitable for THP collapse+ * @pte: The PTE to check + * @vma: The VMA the PTE belongs to + * @addr: The virtual address corresponding to this PTE + * @foliop: On success, used to return a pointer to the folio + * Must be non-NULL + * @none_or_zero: Counter for none/zero PTEs. Must be non-NULL + * @unmapped: Counter for swap PTEs. Can be NULL if not scanning swaps + * @shared: Counter for shared pages. Must be non-NULL + * @scan_result: Used to return the failure reason (SCAN_*) on a + * PTE_CHECK_FAIL return. Must be non-NULL + * @cc: Collapse control settings + * + * Returns: + * PTE_CHECK_SUCCEED - PTE is suitable, proceed with further checks + * PTE_CHECK_CONTINUE - Skip this PTE and continue scanning + * PTE_CHECK_FAIL - Abort collapse scan+ */
Quoted 8 lines
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 566
Quoted 6 lines
mm/khugepaged.c +6
-0+static inline int thp_collapse_check_pte (pte_t pte,struct vm_area_struct *vma, +unsigned long addr,struct folio **foliop,int *none_or_zero, +int *unmapped,int *shared,int *scan_result, +struct collapse_control *cc) +{ +struct folio *folio =NULL ;I think initialization is not needed here?Otherwise, LGTM. Reviewed-by: Dev Jain <[email protected]>
Yep. It's a minor thing, so I'll fold that in if a new version is needed. Thanks!
Re: [PATCH mm-new v3 3/3] mm/khugepaged: merge PTE scanning logic into a new helper
In reply to Lance YangQuoted 136 lines · Lance Yang
From: Lance Yang <[email protected]>
As David suggested, the PTE scanning logic in
hpage_collapse_scan_pmd()and__collapse_huge_page_isolate()was almost duplicated.This patch cleans things up by moving all the common PTE checking logic into a new shared helper,
thp_collapse_check_pte(). While at it, we usevm_normal_folio()instead ofvm_normal_page().Suggested-by David Hildenbrand <[email protected]>Suggested-by Dev Jain <[email protected]>Signed-off-by Lance Yang <[email protected]>---
mm/khugepaged.c| 243 ++++++++++++++++++++++++++----------------------1 file changed, 130 insertions(+), 113 deletions(-)mm/khugepaged.c +105
-0@@ -61,6 +61,12 @@ enum scan_result { SCAN_PAGE_FILLED, }; +enum pte_check_result { +PTE_CHECK_SUCCEED , +PTE_CHECK_CONTINUE , +PTE_CHECK_FAIL , +}; +#define CREATE_TRACE_POINTS #include <trace/events/huge_memory.h> @@ -533,62 +539,139 @@ static void release_pte_pages(pte_t *pte, pte_t *_pte, } } +/* +* thp_collapse_check_pte - Check if a PTE is suitable for THP collapse +* @pte: The PTE to check +* @vma: The VMA the PTE belongs to +* @addr: The virtual address corresponding to this PTE +* @foliop: On success, used to return a pointer to the folio +* Must be non-NULL +* @none_or_zero: Counter for none/zero PTEs. Must be non-NULL +* @unmapped: Counter for swap PTEs. Can be NULL if not scanning swaps +* @shared: Counter for shared pages. Must be non-NULL +* @scan_result: Used to return the failure reason (SCAN_*) on a +* PTE_CHECK_FAIL return. Must be non-NULL +* @cc: Collapse control settings +* +* Returns: +* PTE_CHECK_SUCCEED - PTE is suitable, proceed with further checks +* PTE_CHECK_CONTINUE - Skip this PTE and continue scanning +* PTE_CHECK_FAIL - Abort collapse scan +*/ +static inline int thp_collapse_check_pte (pte_t pte,struct vm_area_struct *vma, +unsigned long addr,struct folio **foliop,int *none_or_zero, +int *unmapped,int *shared,int *scan_result, +struct collapse_control *cc) +{ +struct folio *folio =NULL ; + +if (pte_none (pte) ||is_zero_pfn (pte_pfn (pte))) { + (*none_or_zero)++; +if (!userfaultfd_armed (vma) && + (!cc->is_khugepaged || + *none_or_zero <= khugepaged_max_ptes_none)) { +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_NONE_PTE ; +count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (!pte_present (pte)) { +if (!unmapped) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + +if (non_swap_entry (pte_to_swp_entry (pte))) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + + (*unmapped)++; +if (!cc->is_khugepaged || + *unmapped <= khugepaged_max_ptes_swap) { +/* +* Always be strict with uffd-wp enabled swap +* entries. Please see comment below for +* pte_uffd_wp(). +*/ +if (pte_swp_uffd_wp (pte)) { + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_SWAP_PTE ; +count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (pte_uffd_wp (pte)) { +/* +* Don't collapse the page if any of the small PTEs are +* armed with uffd write protection. Here we can also mark +* the new huge pmd as write protected if any of the small +* ones is marked but that could bring unknown userfault +* messages that falls outside of the registered range. +* So, just be simple. +*/ + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } + + folio =vm_normal_folio (vma, addr, pte); +if (unlikely (!folio) ||unlikely (folio_is_zone_device (folio))) { + *scan_result =SCAN_PAGE_NULL ; +return PTE_CHECK_FAIL ; + } + +if (!folio_test_anon (folio)) { +VM_WARN_ON_FOLIO (true, folio); + *scan_result =SCAN_PAGE_ANON ; +return PTE_CHECK_FAIL ; + } + +/* +* We treat a single page as shared if any part of the THP +* is shared. +*/ +if (folio_maybe_mapped_shared (folio)) { + (*shared)++; +if (cc->is_khugepaged && *shared > khugepaged_max_ptes_shared) { + *scan_result =SCAN_EXCEED_SHARED_PTE ;
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 650
Quoted 10 lines
mm/khugepaged.c +10
-0+count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); +return PTE_CHECK_FAIL ; + } + } + + *foliop = folio; + +return PTE_CHECK_SUCCEED ; +} +
This one looks much better.
While my personal feeling is this is not a complete work to merge the scanning logic. We still have folio_expected_ref_count() and pte_young() check present both in __collapse_huge_page_isolate() and huge_collapse_scan_pmd().
Signature · 2 lines
Wei Yang Help you, Help me
Lance Yang
reviewerRe: [PATCH mm-new v3 3/3] mm/khugepaged: merge PTE scanning logic into a new helper
Quoted 150 lines · Wei Yang
Quoted 136 lines · Lance Yang
From: Lance Yang <[email protected]>
As David suggested, the PTE scanning logic in
hpage_collapse_scan_pmd()and__collapse_huge_page_isolate()was almost duplicated.This patch cleans things up by moving all the common PTE checking logic into a new shared helper,
thp_collapse_check_pte(). While at it, we usevm_normal_folio()instead ofvm_normal_page().Suggested-by David Hildenbrand <[email protected]>Suggested-by Dev Jain <[email protected]>Signed-off-by Lance Yang <[email protected]>---
mm/khugepaged.c| 243 ++++++++++++++++++++++++++----------------------1 file changed, 130 insertions(+), 113 deletions(-)mm/khugepaged.c +105
-0@@ -61,6 +61,12 @@ enum scan_result { SCAN_PAGE_FILLED, }; +enum pte_check_result { +PTE_CHECK_SUCCEED , +PTE_CHECK_CONTINUE , +PTE_CHECK_FAIL , +}; +#define CREATE_TRACE_POINTS #include <trace/events/huge_memory.h> @@ -533,62 +539,139 @@ static void release_pte_pages(pte_t *pte, pte_t *_pte, } } +/* +* thp_collapse_check_pte - Check if a PTE is suitable for THP collapse +* @pte: The PTE to check +* @vma: The VMA the PTE belongs to +* @addr: The virtual address corresponding to this PTE +* @foliop: On success, used to return a pointer to the folio +* Must be non-NULL +* @none_or_zero: Counter for none/zero PTEs. Must be non-NULL +* @unmapped: Counter for swap PTEs. Can be NULL if not scanning swaps +* @shared: Counter for shared pages. Must be non-NULL +* @scan_result: Used to return the failure reason (SCAN_*) on a +* PTE_CHECK_FAIL return. Must be non-NULL +* @cc: Collapse control settings +* +* Returns: +* PTE_CHECK_SUCCEED - PTE is suitable, proceed with further checks +* PTE_CHECK_CONTINUE - Skip this PTE and continue scanning +* PTE_CHECK_FAIL - Abort collapse scan +*/ +static inline int thp_collapse_check_pte (pte_t pte,struct vm_area_struct *vma, +unsigned long addr,struct folio **foliop,int *none_or_zero, +int *unmapped,int *shared,int *scan_result, +struct collapse_control *cc) +{ +struct folio *folio =NULL ; + +if (pte_none (pte) ||is_zero_pfn (pte_pfn (pte))) { + (*none_or_zero)++; +if (!userfaultfd_armed (vma) && + (!cc->is_khugepaged || + *none_or_zero <= khugepaged_max_ptes_none)) { +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_NONE_PTE ; +count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (!pte_present (pte)) { +if (!unmapped) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + +if (non_swap_entry (pte_to_swp_entry (pte))) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + + (*unmapped)++; +if (!cc->is_khugepaged || + *unmapped <= khugepaged_max_ptes_swap) { +/* +* Always be strict with uffd-wp enabled swap +* entries. Please see comment below for +* pte_uffd_wp(). +*/ +if (pte_swp_uffd_wp (pte)) { + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_SWAP_PTE ; +count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (pte_uffd_wp (pte)) { +/* +* Don't collapse the page if any of the small PTEs are +* armed with uffd write protection. Here we can also mark +* the new huge pmd as write protected if any of the small +* ones is marked but that could bring unknown userfault +* messages that falls outside of the registered range. +* So, just be simple. +*/ + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } + + folio =vm_normal_folio (vma, addr, pte); +if (unlikely (!folio) ||unlikely (folio_is_zone_device (folio))) { + *scan_result =SCAN_PAGE_NULL ; +return PTE_CHECK_FAIL ; + } + +if (!folio_test_anon (folio)) { +VM_WARN_ON_FOLIO (true, folio); + *scan_result =SCAN_PAGE_ANON ; +return PTE_CHECK_FAIL ; + } + +/* +* We treat a single page as shared if any part of the THP +* is shared. +*/ +if (folio_maybe_mapped_shared (folio)) { + (*shared)++; +if (cc->is_khugepaged && *shared > khugepaged_max_ptes_shared) { + *scan_result =SCAN_EXCEED_SHARED_PTE ;mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 650
Quoted 10 lines
mm/khugepaged.c +10
-0+count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); +return PTE_CHECK_FAIL ; + } + } + + *foliop = folio; + +return PTE_CHECK_SUCCEED ; +} +This one looks much better.
Quoted 3 lines
While my personal feeling is this is not a complete work to merge the scanning logic. We still have
folio_expected_ref_count()andpte_young()check present both in__collapse_huge_page_isolate()andhuge_collapse_scan_pmd().
Yep, good catch!
There's definitely more that can be done. For now, let's keep this patch as-is to avoid making it too complex. Let's get this one in first, and then we can work on the next step :)
Re: [PATCH mm-new v3 1/3] mm/khugepaged: optimize PTE scanning with if-else-if-else-if chain
In reply to Lance YangQuoted 5 lines · Lance Yang
From: Lance Yang <[email protected]>
As pointed out by Dev, the PTE checks for disjoint conditions in the scanning loops can be optimized.
is_swap_pte, (pte_none&&is_zero_pfn), andpte_uffd_wpare mutually exclusive.
But you're not using is_swap_pte anywhere :) This comes back to my review quesiotn on the series this is dependent upon.
Quoted 5 lines
This patch refactors the loops in both
__collapse_huge_page_isolate()andhpage_collapse_scan_pmd()to use a continuous if-else-if-else-if chain instead of separate if blocks. While at it, the redundantpte_present()check beforeis_zero_pfn()is also removed.
I mean see review below, I don't see why you're doing this and I am unconvinced by how redundant that check is.
Also this just feels like it should be part of the series where you change these? I'm not sure why this is separate?
Quoted 20 lines
Also, this is a preparatory step to make it easier to merge the almost-duplicated scanning logic in these two functions, as suggested by David.
Reviewed-by Wei Yang <[email protected]>Reviewed-by Dev Jain <[email protected]>Reviewed-by Zi Yan <[email protected]>Suggested-by Dev Jain <[email protected]>Suggested-by David Hildenbrand <[email protected]>Signed-off-by Lance Yang <[email protected]>--- mm/khugepaged.c | 12 ++++-------- 1 file changed, 4 insertions(+), 8 deletions(-)mm/khugepaged.c +0
-0
mm/khugepaged.c static int __collapse_huge_page_isolate(struct vm_area_struct *vma, · line 551
Quoted 6 lines
mm/khugepaged.c +1
-2@@ -548,8 +548,7 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma,for (_pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) || (pte_present (pteval) && -is_zero_pfn (pte_pfn (pteval)))) { +if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) {
You can have non-pte_none() non-present entries no? Isn't pte_present() a prerequisite for pfe_pfn() to be valid? If it's a swap entry couldn't you end up accidentally (unlikely but still) hitting this?
Seems like this is required isn't it? I may be missing something here...
mm/khugepaged.c static int __collapse_huge_page_isolate(struct vm_area_struct *vma, · line 562
Quoted 10 lines
mm/khugepaged.c +1
-2++none_or_zero; if (!userfaultfd_armed(vma) && (!cc->is_khugepaged ||@@ -560,12 +559,10 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma,count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); goto out; } - } -if (!pte_present (pteval)) { + }else if (!pte_present (pteval)) {
This seems pointless, since either the above logic will continue or goto out right?
mm/khugepaged.c static int __collapse_huge_page_isolate(struct vm_area_struct *vma, · line 565
Quoted 5 lines
mm/khugepaged.c +1
-2result =SCAN_PTE_NON_PRESENT ; goto out; - } -if (pte_uffd_wp (pteval)) { + }else if (pte_uffd_wp (pteval)) {
Again, what is the point of an else when the if() branch unconditionally ->out?
mm/khugepaged.c static int hpage_collapse_scan_pmd(struct mm_struct *mm, · line 1321
Quoted 10 lines
mm/khugepaged.c +1
-2result =SCAN_PTE_UFFD_WP ; goto out; }@@ -1321,8 +1318,7 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); goto out_unmap; } - } -if (pte_uffd_wp (pteval)) { + }else if (pte_uffd_wp (pteval)) {
Same comment as above, I'm really confused about the purpose of this logic?
Quoted 6 lines
mm/khugepaged.c +0
-1/* * Don't collapse the page if any of the small * PTEs are armed with uffd write protection. -- 2.49.0
Re: [PATCH mm-new v3 2/3] mm/khugepaged: use VM_WARN_ON_FOLIO instead of VM_BUG_ON_FOLIO for non-anon folios
In reply to Lance YangQuoted 9 lines · Lance Yang
From: Lance Yang <[email protected]>
As Zi pointed out, we should avoid crashing the kernel for conditions that can be handled gracefully. Encountering a non-anonymous folio in an anonymous VMA is a bug, but a warning is sufficient.
This patch changes the VM_BUG_ON_FOLIO(!folio_test_anon(folio)) to a
VM_WARN_ON_FOLIO()in both__collapse_huge_page_isolate()andhpage_collapse_scan_pmd(), and then aborts the scan withSCAN_PAGE_ANON.
Well no, in hpage_collapse_scan_pmd() there is no warning at all.
Quoted 3 lines
Making more of the scanning logic common between
hpage_collapse_scan_pmd()and__collapse_huge_page_isolate(), as suggested by Dev.
I mean I guess it's fine but I'm not sure it's really necessary to give a blow-by-blow of who suggested what in the actual commit message :) This isn't really useful information for somebody looking at this code in the future.
Quoted 15 lines
Suggested-by Dev Jain <[email protected]>Suggested-by Zi Yan <[email protected]>Reviewed-by Wei Yang <[email protected]>Reviewed-by Dev Jain <[email protected]>Signed-off-by Lance Yang <[email protected]>--- mm/khugepaged.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)mm/khugepaged.c +0
-0
mm/khugepaged.c static int __collapse_huge_page_isolate(struct vm_area_struct *vma, · line 579
Quoted 8 lines
mm/khugepaged.c +4
-1@@ -573,7 +573,11 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma, } folio =page_folio (page); -VM_BUG_ON_FOLIO (!folio_test_anon (folio), folio); +if (!folio_test_anon (folio)) { +VM_WARN_ON_FOLIO (true, folio); + result =SCAN_PAGE_ANON ; + goto out;
Hmm this is iffy I'm not sure I agree with Zi here - the purpose of VM_WARN_ON() etc. is for things that programmatically _should not happen_.
Now every single time we run this code we're doing this check.
AND it implies that it is an actual possiblity, at run time, for this to be the case.
I really don't like this.
Also if it's a runtime check this should be a WARN_ON_ONCE() not a VM_WARN_ON_ONCE(). Of course you lose the folio output then. So this code is very confused.
In general I don't think we should be doing this at all, rather we should just convert the VM_BUG_ON_FOLIO() to a VM_WARN_ON_FOLIO().
mm/khugepaged.c static int hpage_collapse_scan_pmd(struct mm_struct *mm, · line 1347
Quoted 9 lines
mm/khugepaged.c +2
-0+ }/* See hpage_collapse_scan_pmd(). */ if (folio_maybe_mapped_shared (folio)) {@@ -1340,6 +1344,7 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm, folio =page_folio (page);if (!folio_test_anon (folio)) { +VM_WARN_ON_FOLIO (true, folio);
Err, what? This is a condition that should never, ever happen to the point that we warn on it?
This surely is a condition that we expect to happen sometimes otherwise we wouldn't do this no?
Either way the comments above still apply. Also VM_WARN_ON_FOLIO(true, ...) is kinda gross... if this is an actual pattern that exists, VM_WARN_FOLIO() would be preferable.
Quoted 6 lines
mm/khugepaged.c +0
-1result =SCAN_PAGE_ANON ; goto out_unmap; } --2.49.0
Re: [PATCH mm-new v3 1/3] mm/khugepaged: optimize PTE scanning with if-else-if-else-if chain
In reply to Lorenzo StoakesQuoted 16 lines · Lorenzo Stoakes
Quoted 5 lines · Lance Yang
From: Lance Yang <[email protected]>
As pointed out by Dev, the PTE checks for disjoint conditions in the scanning loops can be optimized.
is_swap_pte, (pte_none&&is_zero_pfn), andpte_uffd_wpare mutually exclusive.But you're not using
is_swap_pteanywhere :) This comes back to my review quesiotn on the series this is dependent upon.Quoted 5 lines
This patch refactors the loops in both
__collapse_huge_page_isolate()andhpage_collapse_scan_pmd()to use a continuous if-else-if-else-if chain instead of separate if blocks. While at it, the redundantpte_present()check beforeis_zero_pfn()is also removed.
Quoted 5 lines
I mean see review below, I don't see why you're doing this and I am unconvinced by how redundant that check is.
Also this just feels like it should be part of the series where you change these? I'm not sure why this is separate?
I think Lance is trying to unify both scanning functions to look alike, such that when he refactors them out in patch #3 it looks more straight forward.
The missing pte_present() check in hpage_collapse_scan_pmd() is interesting
Likely there is one such check missing there?
Signature · 3 lines
Cheers David / dhildenb
Re: [PATCH mm-new v3 3/3] mm/khugepaged: merge PTE scanning logic into a new helper
In reply to Lance YangQuoted 5 lines · Lance Yang
From: Lance Yang <[email protected]>
As David suggested, the PTE scanning logic in
hpage_collapse_scan_pmd()and__collapse_huge_page_isolate()was almost duplicated.
Quoted 7 lines
This patch cleans things up by moving all the common PTE checking logic into a new shared helper,
thp_collapse_check_pte(). While at it, we usevm_normal_folio()instead ofvm_normal_page().Suggested-by David Hildenbrand <[email protected]>Suggested-by Dev Jain <[email protected]>Signed-off-by Lance Yang <[email protected]>
In general I like the idea, the implementation needs work though.
Will check in more detail on respin
Quoted 9 lines
--- mm/khugepaged.c | 243 ++++++++++++++++++++++++++---------------------- 1 file changed, 130 insertions(+), 113 deletions(-)mm/khugepaged.c +0
-0
mm/khugepaged.c enum scan_result { · line 66
Quoted 6 lines
mm/khugepaged.c +3
-0@@ -61,6 +61,12 @@ enum scan_result { SCAN_PAGE_FILLED, }; +enum pte_check_result { +PTE_CHECK_SUCCEED , +PTE_CHECK_CONTINUE ,
Don't love this logic - this feels like we're essentially abstracting control flow, I mean what does 'continue' mean here? Other than you're in a loop and please continue, which is relying a little too much on what the caller does.
if we retain this logic something like PTE_CHECK_SKIP would make more sense.
Quoted 13 lines
mm/khugepaged.c +6
-0+ PTE_CHECK_FAIL, +}; +#define CREATE_TRACE_POINTS #include <trace/events/huge_memory.h> @@ -533,62 +539,139 @@ static void release_pte_pages(pte_t *pte, pte_t *_pte, } } +/* +* thp_collapse_check_pte - Check if a PTE is suitable for THP collapse +* @pte: The PTE to check
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 554
Quoted 10 lines
mm/khugepaged.c +10
-0+* @vma: The VMA the PTE belongs to +* @addr: The virtual address corresponding to this PTE +* @foliop: On success, used to return a pointer to the folio +* Must be non-NULL +* @none_or_zero: Counter for none/zero PTEs. Must be non-NULL +* @unmapped: Counter for swap PTEs. Can be NULL if not scanning swaps +* @shared: Counter for shared pages. Must be non-NULL +* @scan_result: Used to return the failure reason (SCAN_*) on a +* PTE_CHECK_FAIL return. Must be non-NULL +* @cc: Collapse control settings
Do we really need this many parameters? THis is hard to follow.
Of course it being me, I'd immediately prefer a helper struct :)
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 561
Quoted 7 lines
mm/khugepaged.c +7
-0+* +* Returns: +* PTE_CHECK_SUCCEED - PTE is suitable, proceed with further checks +* PTE_CHECK_CONTINUE - Skip this PTE and continue scanning +* PTE_CHECK_FAIL - Abort collapse scan +*/ +static inline int thp_collapse_check_pte(pte_t pte, struct vm_area_struct *vma,
There's no need for inline in an internal static function in a file.
Quoted 8 lines
mm/khugepaged.c +8
-0+ unsigned long addr, struct folio **foliop, int *none_or_zero, + int *unmapped, int *shared, int *scan_result, + struct collapse_control *cc) +{ + struct folio *folio = NULL; + +if (pte_none (pte) ||is_zero_pfn (pte_pfn (pte))) { + (*none_or_zero)++;
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 579
Quoted 10 lines
mm/khugepaged.c +10
-0+if (!userfaultfd_armed (vma) && + (!cc->is_khugepaged || + *none_or_zero <= khugepaged_max_ptes_none)) { +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_NONE_PTE ; +count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (!pte_present (pte)) {
You're now making the if-else issues with previous patches worse by returning which gets even checkpatch to warn you.
if (pte_none(pte) || is_zero_pfn(pte_pfn(pte))) {(of course note that I am not convinced you can drop the pte_present() check here)
(*none_or_zero)++;
if (!userfaultfd_armed(vma) &&
(!cc->is_khugepaged ||
*none_or_zero <= khugepaged_max_ptes_none))
return PTE_CHECK_CONTINUE;
*scan_result = SCAN_EXCEED_NONE_PTE;
count_vm_event(THP_SCAN_EXCEED_NONE_PTE);
return PTE_CHECK_FAIL;
}
if (!pte_present(pte)) {
...
}But even that really needs seriously more refactoring - that condition above is horrible for instance so, e.g.:
if (pte_none(pte) || is_zero_pfn(pte_pfn(pte))) {
(*none_or_zero)++;
/* Cases where this is acceptable. */
if (!userfaultfd_armed(vma))
return PTE_CHECK_SKIP;
if (!cc->is_khugepaged)
return PTE_CHECK_SKIP;
if (*none_or_zero <= khugepaged_max_ptes_none)
return PTE_CHECK_SKIP;
/* Otherwise, we must abort the scan. */
*scan_result = SCAN_EXCEED_NONE_PTE;
count_vm_event(THP_SCAN_EXCEED_NONE_PTE);
return PTE_CHECK_FAIL;
}
if (!pte_present(pte)) {
...
}Improves things a lot.
I do however _hate_ this (*blah)++; thing. A helper struct would help :)
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 582
Quoted 3 lines
mm/khugepaged.c +3
-0+if (!unmapped) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ;
Can't we have a helper that sets the result, e.g.:
return pte_check_fail(scan_result, SCAN_PTE_NON_PRESENT);
static enum pte_check_result pte_check_fail(int *scan_result,
enum pte_check_result result)
{
*scan_result = result;
return PTE_CHECK_FAIL;
}And same for success/skip
Then all of these horrible if (blah) { *foo = bar; return baz; } can be made into
if (blah) return pte_check_xxx(..., SCAN_PTE_...);
Quoted 28 lines
mm/khugepaged.c +28
-0+ } + +if (non_swap_entry (pte_to_swp_entry (pte))) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + + (*unmapped)++; +if (!cc->is_khugepaged || + *unmapped <= khugepaged_max_ptes_swap) { +/* +* Always be strict with uffd-wp enabled swap +* entries. Please see comment below for +* pte_uffd_wp(). +*/ +if (pte_swp_uffd_wp (pte)) { + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_SWAP_PTE ; +count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (pte_uffd_wp (pte)) { +/* +* Don't collapse the page if any of the small PTEs are
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 620
Quoted 10 lines
mm/khugepaged.c +10
-0+* armed with uffd write protection. Here we can also mark +* the new huge pmd as write protected if any of the small +* ones is marked but that could bring unknown userfault +* messages that falls outside of the registered range. +* So, just be simple. +*/ + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } +
Again as all of the comments for previous series still apply here so obviously I don't like this as-is :)
And as Andrew has pointed out, now checkpatch is finding the 'pointless else' issue (as mentioned above also).
Quoted 201 lines
mm/khugepaged.c +45
-114+ folio =vm_normal_folio (vma, addr, pte); +if (unlikely (!folio) ||unlikely (folio_is_zone_device (folio))) { + *scan_result =SCAN_PAGE_NULL ; +return PTE_CHECK_FAIL ; + } + +if (!folio_test_anon (folio)) { +VM_WARN_ON_FOLIO (true, folio); + *scan_result =SCAN_PAGE_ANON ; +return PTE_CHECK_FAIL ; + } + +/* +* We treat a single page as shared if any part of the THP +* is shared. +*/ +if (folio_maybe_mapped_shared (folio)) { + (*shared)++; +if (cc->is_khugepaged && *shared > khugepaged_max_ptes_shared) { + *scan_result =SCAN_EXCEED_SHARED_PTE ; +count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); +return PTE_CHECK_FAIL ; + } + } + + *foliop = folio; + +return PTE_CHECK_SUCCEED ; +} +static int __collapse_huge_page_isolate (struct vm_area_struct *vma,unsigned long start_addr,pte_t *pte,struct collapse_control *cc,struct list_head *compound_pagelist) { -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr = start_addr;pte_t *_pte;int none_or_zero =0 , shared =0 , result =SCAN_FAIL , referenced =0 ; +int pte_check_res;for (_pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out; - } - }else if (!pte_present (pteval)) { - result =SCAN_PTE_NON_PRESENT ; - goto out; - }else if (pte_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out; - } - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out; - } - folio =page_folio (page); -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; - goto out; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero,NULL , &shared, + &result, cc); -/* See hpage_collapse_scan_pmd(). */ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out; - } - } +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) + goto out;if (folio_test_large (folio)) {struct folio *f;@@ -1264,11 +1347,11 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,pte_t *pte, *_pte;int result =SCAN_FAIL , referenced =0 ;int none_or_zero =0 , shared =0 ; -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr;spinlock_t *ptl;int node =NUMA_NO_NODE , unmapped =0 ; +int pte_check_res;VM_BUG_ON (start_addr & ~HPAGE_PMD_MASK );@@ -1287,81 +1370,15 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,for (addr = start_addr, _pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out_unmap; - } - }else if (!pte_present (pteval)) { -if (non_swap_entry (pte_to_swp_entry (pteval))) { - result =SCAN_PTE_NON_PRESENT ; - goto out_unmap; - } - ++unmapped; -if (!cc->is_khugepaged || - unmapped <= khugepaged_max_ptes_swap) { -/* -* Always be strict with uffd-wp -* enabled swap entries. Please see -* comment below for pte_uffd_wp(). -*/ -if (pte_swp_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } -continue ; - }else { - result =SCAN_EXCEED_SWAP_PTE ; -count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); - goto out_unmap; - } - }else if (pte_uffd_wp (pteval)) { -/* -* Don't collapse the page if any of the small -* PTEs are armed with uffd write protection. -* Here we can also mark the new huge pmd as -* write protected if any of the small ones is -* marked but that could bring unknown -* userfault messages that falls outside of -* the registered range. So, just be simple. -*/ - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero, &unmapped, + &shared, &result, cc); - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out_unmap; - } - folio =page_folio (page); - -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) goto out_unmap; - } - -/* -* We treat a single page as shared if any part of the THP -* is shared. -*/ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out_unmap; - } - }/* * Record which node the original page is from and save this -- 2.49.0
Re: [PATCH mm-new v3 3/3] mm/khugepaged: merge PTE scanning logic into a new helper
In reply to Lance YangOK so this patch _is_ based on [0] already being applied then?
I don't see any is_swap_pte() here so presumably so right?
Can we just respin these series and put them together? Because now I've reviewed a bunch of stuff in [0], which this series depends upon, and you're saying we should now review on this instead of that and it's a bit of a mess.
Once review here is dealt with can you combine [0] into this series please?
For [0] I would like you to reinstate the is_swap_pte() conditional as (copiously :) discussed over there.
[0}: https://lore.kernel.org/all/[email protected]/
Thanks, Lorenzo
Quoted 317 lines · Lance Yang
From: Lance Yang <[email protected]>
As David suggested, the PTE scanning logic in
hpage_collapse_scan_pmd()and__collapse_huge_page_isolate()was almost duplicated.This patch cleans things up by moving all the common PTE checking logic into a new shared helper,
thp_collapse_check_pte(). While at it, we usevm_normal_folio()instead ofvm_normal_page().Suggested-by David Hildenbrand <[email protected]>Suggested-by Dev Jain <[email protected]>Signed-off-by Lance Yang <[email protected]>--- mm/khugepaged.c | 243 ++++++++++++++++++++++++++---------------------- 1 file changed, 130 insertions(+), 113 deletions(-)mm/khugepaged.c +130
-114@@ -61,6 +61,12 @@ enum scan_result { SCAN_PAGE_FILLED, }; +enum pte_check_result { +PTE_CHECK_SUCCEED , +PTE_CHECK_CONTINUE , +PTE_CHECK_FAIL , +}; +#define CREATE_TRACE_POINTS #include <trace/events/huge_memory.h> @@ -533,62 +539,139 @@ static void release_pte_pages(pte_t *pte, pte_t *_pte, } } +/* +* thp_collapse_check_pte - Check if a PTE is suitable for THP collapse +* @pte: The PTE to check +* @vma: The VMA the PTE belongs to +* @addr: The virtual address corresponding to this PTE +* @foliop: On success, used to return a pointer to the folio +* Must be non-NULL +* @none_or_zero: Counter for none/zero PTEs. Must be non-NULL +* @unmapped: Counter for swap PTEs. Can be NULL if not scanning swaps +* @shared: Counter for shared pages. Must be non-NULL +* @scan_result: Used to return the failure reason (SCAN_*) on a +* PTE_CHECK_FAIL return. Must be non-NULL +* @cc: Collapse control settings +* +* Returns: +* PTE_CHECK_SUCCEED - PTE is suitable, proceed with further checks +* PTE_CHECK_CONTINUE - Skip this PTE and continue scanning +* PTE_CHECK_FAIL - Abort collapse scan +*/ +static inline int thp_collapse_check_pte (pte_t pte,struct vm_area_struct *vma, +unsigned long addr,struct folio **foliop,int *none_or_zero, +int *unmapped,int *shared,int *scan_result, +struct collapse_control *cc) +{ +struct folio *folio =NULL ; + +if (pte_none (pte) ||is_zero_pfn (pte_pfn (pte))) { + (*none_or_zero)++; +if (!userfaultfd_armed (vma) && + (!cc->is_khugepaged || + *none_or_zero <= khugepaged_max_ptes_none)) { +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_NONE_PTE ; +count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (!pte_present (pte)) { +if (!unmapped) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + +if (non_swap_entry (pte_to_swp_entry (pte))) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + + (*unmapped)++; +if (!cc->is_khugepaged || + *unmapped <= khugepaged_max_ptes_swap) { +/* +* Always be strict with uffd-wp enabled swap +* entries. Please see comment below for +* pte_uffd_wp(). +*/ +if (pte_swp_uffd_wp (pte)) { + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_SWAP_PTE ; +count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (pte_uffd_wp (pte)) { +/* +* Don't collapse the page if any of the small PTEs are +* armed with uffd write protection. Here we can also mark +* the new huge pmd as write protected if any of the small +* ones is marked but that could bring unknown userfault +* messages that falls outside of the registered range. +* So, just be simple. +*/ + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } + + folio =vm_normal_folio (vma, addr, pte); +if (unlikely (!folio) ||unlikely (folio_is_zone_device (folio))) { + *scan_result =SCAN_PAGE_NULL ; +return PTE_CHECK_FAIL ; + } + +if (!folio_test_anon (folio)) { +VM_WARN_ON_FOLIO (true, folio); + *scan_result =SCAN_PAGE_ANON ; +return PTE_CHECK_FAIL ; + } + +/* +* We treat a single page as shared if any part of the THP +* is shared. +*/ +if (folio_maybe_mapped_shared (folio)) { + (*shared)++; +if (cc->is_khugepaged && *shared > khugepaged_max_ptes_shared) { + *scan_result =SCAN_EXCEED_SHARED_PTE ; +count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); +return PTE_CHECK_FAIL ; + } + } + + *foliop = folio; + +return PTE_CHECK_SUCCEED ; +} +static int __collapse_huge_page_isolate (struct vm_area_struct *vma,unsigned long start_addr,pte_t *pte,struct collapse_control *cc,struct list_head *compound_pagelist) { -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr = start_addr;pte_t *_pte;int none_or_zero =0 , shared =0 , result =SCAN_FAIL , referenced =0 ; +int pte_check_res;for (_pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out; - } - }else if (!pte_present (pteval)) { - result =SCAN_PTE_NON_PRESENT ; - goto out; - }else if (pte_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out; - } - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out; - } - folio =page_folio (page); -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; - goto out; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero,NULL , &shared, + &result, cc); -/* See hpage_collapse_scan_pmd(). */ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out; - } - } +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) + goto out;if (folio_test_large (folio)) {struct folio *f;@@ -1264,11 +1347,11 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,pte_t *pte, *_pte;int result =SCAN_FAIL , referenced =0 ;int none_or_zero =0 , shared =0 ; -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr;spinlock_t *ptl;int node =NUMA_NO_NODE , unmapped =0 ; +int pte_check_res;VM_BUG_ON (start_addr & ~HPAGE_PMD_MASK );@@ -1287,81 +1370,15 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,for (addr = start_addr, _pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out_unmap; - } - }else if (!pte_present (pteval)) { -if (non_swap_entry (pte_to_swp_entry (pteval))) { - result =SCAN_PTE_NON_PRESENT ; - goto out_unmap; - } - ++unmapped; -if (!cc->is_khugepaged || - unmapped <= khugepaged_max_ptes_swap) { -/* -* Always be strict with uffd-wp -* enabled swap entries. Please see -* comment below for pte_uffd_wp(). -*/ -if (pte_swp_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } -continue ; - }else { - result =SCAN_EXCEED_SWAP_PTE ; -count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); - goto out_unmap; - } - }else if (pte_uffd_wp (pteval)) { -/* -* Don't collapse the page if any of the small -* PTEs are armed with uffd write protection. -* Here we can also mark the new huge pmd as -* write protected if any of the small ones is -* marked but that could bring unknown -* userfault messages that falls outside of -* the registered range. So, just be simple. -*/ - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero, &unmapped, + &shared, &result, cc); - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out_unmap; - } - folio =page_folio (page); - -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) goto out_unmap; - } - -/* -* We treat a single page as shared if any part of the THP -* is shared. -*/ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out_unmap; - } - }/* * Record which node the original page is from and save this -- 2.49.0
Lance Yang
reviewerRe: [PATCH mm-new v3 3/3] mm/khugepaged: merge PTE scanning logic into a new helper
Quoted 10 lines · Lorenzo Stoakes
OK so this patch _is_ based on [0] already being applied then?
I don't see any
is_swap_pte()here so presumably so right?Can we just respin these series and put them together? Because now I've reviewed a bunch of stuff in [0], which this series depends upon, and you're saying we should now review on this instead of that and it's a bit of a mess.
Once review here is dealt with can you combine [0] into this series please?
Got it, thanks!
Quoted 326 lines
For [0] I would like you to reinstate the
is_swap_pte()conditional as (copiously :) discussed over there.[0}: https://lore.kernel.org/all/[email protected]/
Thanks, Lorenzo
Quoted 317 lines · Lance Yang
From: Lance Yang <[email protected]>
As David suggested, the PTE scanning logic in
hpage_collapse_scan_pmd()and__collapse_huge_page_isolate()was almost duplicated.This patch cleans things up by moving all the common PTE checking logic into a new shared helper,
thp_collapse_check_pte(). While at it, we usevm_normal_folio()instead ofvm_normal_page().Suggested-by David Hildenbrand <[email protected]>Suggested-by Dev Jain <[email protected]>Signed-off-by Lance Yang <[email protected]>--- mm/khugepaged.c | 243 ++++++++++++++++++++++++++---------------------- 1 file changed, 130 insertions(+), 113 deletions(-)mm/khugepaged.c +130
-114@@ -61,6 +61,12 @@ enum scan_result { SCAN_PAGE_FILLED, }; +enum pte_check_result { +PTE_CHECK_SUCCEED , +PTE_CHECK_CONTINUE , +PTE_CHECK_FAIL , +}; +#define CREATE_TRACE_POINTS #include <trace/events/huge_memory.h> @@ -533,62 +539,139 @@ static void release_pte_pages(pte_t *pte, pte_t *_pte, } } +/* +* thp_collapse_check_pte - Check if a PTE is suitable for THP collapse +* @pte: The PTE to check +* @vma: The VMA the PTE belongs to +* @addr: The virtual address corresponding to this PTE +* @foliop: On success, used to return a pointer to the folio +* Must be non-NULL +* @none_or_zero: Counter for none/zero PTEs. Must be non-NULL +* @unmapped: Counter for swap PTEs. Can be NULL if not scanning swaps +* @shared: Counter for shared pages. Must be non-NULL +* @scan_result: Used to return the failure reason (SCAN_*) on a +* PTE_CHECK_FAIL return. Must be non-NULL +* @cc: Collapse control settings +* +* Returns: +* PTE_CHECK_SUCCEED - PTE is suitable, proceed with further checks +* PTE_CHECK_CONTINUE - Skip this PTE and continue scanning +* PTE_CHECK_FAIL - Abort collapse scan +*/ +static inline int thp_collapse_check_pte (pte_t pte,struct vm_area_struct *vma, +unsigned long addr,struct folio **foliop,int *none_or_zero, +int *unmapped,int *shared,int *scan_result, +struct collapse_control *cc) +{ +struct folio *folio =NULL ; + +if (pte_none (pte) ||is_zero_pfn (pte_pfn (pte))) { + (*none_or_zero)++; +if (!userfaultfd_armed (vma) && + (!cc->is_khugepaged || + *none_or_zero <= khugepaged_max_ptes_none)) { +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_NONE_PTE ; +count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (!pte_present (pte)) { +if (!unmapped) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + +if (non_swap_entry (pte_to_swp_entry (pte))) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + + (*unmapped)++; +if (!cc->is_khugepaged || + *unmapped <= khugepaged_max_ptes_swap) { +/* +* Always be strict with uffd-wp enabled swap +* entries. Please see comment below for +* pte_uffd_wp(). +*/ +if (pte_swp_uffd_wp (pte)) { + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_SWAP_PTE ; +count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (pte_uffd_wp (pte)) { +/* +* Don't collapse the page if any of the small PTEs are +* armed with uffd write protection. Here we can also mark +* the new huge pmd as write protected if any of the small +* ones is marked but that could bring unknown userfault +* messages that falls outside of the registered range. +* So, just be simple. +*/ + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } + + folio =vm_normal_folio (vma, addr, pte); +if (unlikely (!folio) ||unlikely (folio_is_zone_device (folio))) { + *scan_result =SCAN_PAGE_NULL ; +return PTE_CHECK_FAIL ; + } + +if (!folio_test_anon (folio)) { +VM_WARN_ON_FOLIO (true, folio); + *scan_result =SCAN_PAGE_ANON ; +return PTE_CHECK_FAIL ; + } + +/* +* We treat a single page as shared if any part of the THP +* is shared. +*/ +if (folio_maybe_mapped_shared (folio)) { + (*shared)++; +if (cc->is_khugepaged && *shared > khugepaged_max_ptes_shared) { + *scan_result =SCAN_EXCEED_SHARED_PTE ; +count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); +return PTE_CHECK_FAIL ; + } + } + + *foliop = folio; + +return PTE_CHECK_SUCCEED ; +} +static int __collapse_huge_page_isolate (struct vm_area_struct *vma,unsigned long start_addr,pte_t *pte,struct collapse_control *cc,struct list_head *compound_pagelist) { -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr = start_addr;pte_t *_pte;int none_or_zero =0 , shared =0 , result =SCAN_FAIL , referenced =0 ; +int pte_check_res;for (_pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out; - } - }else if (!pte_present (pteval)) { - result =SCAN_PTE_NON_PRESENT ; - goto out; - }else if (pte_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out; - } - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out; - } - folio =page_folio (page); -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; - goto out; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero,NULL , &shared, + &result, cc); -/* See hpage_collapse_scan_pmd(). */ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out; - } - } +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) + goto out;if (folio_test_large (folio)) {struct folio *f;@@ -1264,11 +1347,11 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,pte_t *pte, *_pte;int result =SCAN_FAIL , referenced =0 ;int none_or_zero =0 , shared =0 ; -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr;spinlock_t *ptl;int node =NUMA_NO_NODE , unmapped =0 ; +int pte_check_res;VM_BUG_ON (start_addr & ~HPAGE_PMD_MASK );@@ -1287,81 +1370,15 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,for (addr = start_addr, _pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out_unmap; - } - }else if (!pte_present (pteval)) { -if (non_swap_entry (pte_to_swp_entry (pteval))) { - result =SCAN_PTE_NON_PRESENT ; - goto out_unmap; - } - ++unmapped; -if (!cc->is_khugepaged || - unmapped <= khugepaged_max_ptes_swap) { -/* -* Always be strict with uffd-wp -* enabled swap entries. Please see -* comment below for pte_uffd_wp(). -*/ -if (pte_swp_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } -continue ; - }else { - result =SCAN_EXCEED_SWAP_PTE ; -count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); - goto out_unmap; - } - }else if (pte_uffd_wp (pteval)) { -/* -* Don't collapse the page if any of the small -* PTEs are armed with uffd write protection. -* Here we can also mark the new huge pmd as -* write protected if any of the small ones is -* marked but that could bring unknown -* userfault messages that falls outside of -* the registered range. So, just be simple. -*/ - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero, &unmapped, + &shared, &result, cc); - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out_unmap; - } - folio =page_folio (page); - -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) goto out_unmap; - } - -/* -* We treat a single page as shared if any part of the THP -* is shared. -*/ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out_unmap; - } - }/* * Record which node the original page is from and save this -- 2.49.0
Lance Yang
reviewerRe: [PATCH mm-new v3 1/3] mm/khugepaged: optimize PTE scanning with if-else-if-else-if chain
In reply to David HildenbrandQuoted 12 lines · David Hildenbrand
Quoted 11 lines · Lorenzo Stoakes
Quoted 5 lines · Lance Yang
From: Lance Yang <[email protected]>
As pointed out by Dev, the PTE checks for disjoint conditions in the scanning loops can be optimized.
is_swap_pte, (pte_none&&is_zero_pfn), andpte_uffd_wpare mutually exclusive.But you're not using
is_swap_pteanywhere :) This comes back to my review quesiotn on the series this is dependent upon.Quoted 1 line
Quoted 7 lines
Quoted 7 lines
Quoted 4 lines
This patch refactors the loops in both
__collapse_huge_page_isolate()andhpage_collapse_scan_pmd()to use a continuous if-else-if-else-if chain instead of separate if blocks. While at it, the redundantpte_present()check beforeis_zero_pfn()is also removed.I mean see review below, I don't see why you're doing this and I am unconvinced by how redundant that check is.
Ah, good catch! Lorenzo, thanks!!!
Quoted 7 lines
Quoted 3 lines
Also this just feels like it should be part of the series where you change these? I'm not sure why this is separate?
I think Lance is trying to unify both scanning functions to look alike, such that when he refactors them out in patch #3 it looks more straight forward.
The missing
pte_present()check inhpage_collapse_scan_pmd()is interesting
Yep, indeed ;)
Quoted 2 lines
Likely there is one such check missing there?
I think the risk is exactly how pte_pfn() would handle a swap PTE ...
A swap PTE contains completely different data(swap type and offset). pte_pfn() doesn't know this, so if we feed a swap entry to it, it will spit out a junk PFN :)
What if that junk PFN happens to match the zeropage's PFN by sheer chance? IHMO, it's really unlikely, but it would be really bad if it did.
Clearly, pte_present() prevents this ;)
By the way, I noticed there are other places in khugepaged.c that call pte_pfn() without being under a pte_present() check.
Perhaps those should all be fixed as well in a separate patch?
Re: [PATCH mm-new v3 1/3] mm/khugepaged: optimize PTE scanning with if-else-if-else-if chain
Quoted 44 lines · Lance Yang
Quoted 12 lines · David Hildenbrand
Quoted 11 lines · Lorenzo Stoakes
Quoted 5 lines · Lance Yang
From: Lance Yang <[email protected]>
As pointed out by Dev, the PTE checks for disjoint conditions in the scanning loops can be optimized.
is_swap_pte, (pte_none&&is_zero_pfn), andpte_uffd_wpare mutually exclusive.But you're not using
is_swap_pteanywhere :) This comes back to my review quesiotn on the series this is dependent upon.Quoted 1 line
Quoted 8 lines
Quoted 8 lines
Quoted 5 lines
This patch refactors the loops in both
__collapse_huge_page_isolate()andhpage_collapse_scan_pmd()to use a continuous if-else-if-else-if chain instead of separate if blocks. While at it, the redundantpte_present()check beforeis_zero_pfn()is also removed.I mean see review below, I don't see why you're doing this and I am unconvinced by how redundant that check is.
Ah, good catch! Lorenzo, thanks!!!
Quoted 10 lines
Quoted 4 lines
Also this just feels like it should be part of the series where you change
these? I'm not sure why this is separate?I think Lance is trying to unify both scanning functions to look alike, such that when he refactors them out in patch #3 it looks more straight forward.
The missing
pte_present()check inhpage_collapse_scan_pmd()is interestingYep, indeed ;)
Quoted 2 lines
Likely there is one such check missing there?
I think the risk is exactly how
pte_pfn()would handle a swap PTE ...
Quoted 8 lines
A swap PTE contains completely different data(swap type and offset).
pte_pfn()doesn't know this, so if we feed a swap entry to it, it will spit out a junk PFN :)What if that junk PFN happens to match the zeropage's PFN by sheer chance? IHMO, it's really unlikely, but it would be really bad if it did.
Clearly,
pte_present()prevents this ;)
Yeah, not so clearly kinda the whole point I'm making here. I mean all this code sucks because we have deeply nested conditions and you have to keep in your mind that 'oh we already checked for X so we can do Y'.
But the THP code is horrible in general.
Anyway let's stay focused (so I can stand a chance of catching up with my post-vacation backlog :), I will check the respin once you send!
Quoted 5 lines
By the way, I noticed there are other places in
khugepaged.cthat callpte_pfn()without being under apte_present()check.Perhaps those should all be fixed as well in a separate patch?
Yeah I'm not surprised and sure indeed.
Thanks, Lorenzo
Lance Yang
reviewerRe: [PATCH mm-new v3 1/3] mm/khugepaged: optimize PTE scanning with if-else-if-else-if chain
Quoted 49 lines · Lorenzo Stoakes
Quoted 48 lines · Lance Yang
Quoted 12 lines · David Hildenbrand
Quoted 11 lines · Lorenzo Stoakes
Quoted 5 lines · Lance Yang
From: Lance Yang <[email protected]>
As pointed out by Dev, the PTE checks for disjoint conditions in the scanning loops can be optimized.
is_swap_pte, (pte_none&&is_zero_pfn), andpte_uffd_wpare mutually exclusive.But you're not using
is_swap_pteanywhere :) This comes back to my review quesiotn on the series this is dependent upon.Quoted 1 line
Quoted 8 lines
Quoted 8 lines
Quoted 5 lines
This patch refactors the loops in both
__collapse_huge_page_isolate()andhpage_collapse_scan_pmd()to use a continuous if-else-if-else-if chain instead of separate if blocks. While at it, the redundantpte_present()check beforeis_zero_pfn()is also removed.I mean see review below, I don't see why you're doing this and I am unconvinced by how redundant that check is.
Ah, good catch! Lorenzo, thanks!!!
Quoted 10 lines
Quoted 4 lines
Also this just feels like it should be part of the series where you change
these? I'm not sure why this is separate?I think Lance is trying to unify both scanning functions to look alike, such that when he refactors them out in patch #3 it looks more straight forward.
The missing
pte_present()check inhpage_collapse_scan_pmd()is interestingYep, indeed ;)
Quoted 2 lines
Likely there is one such check missing there?
I think the risk is exactly how
pte_pfn()would handle a swap PTE ...A swap PTE contains completely different data(swap type and offset).
pte_pfn()doesn't know this, so if we feed a swap entry to it, it will spit out a junk PFN :)
Quoted 8 lines
Quoted 4 lines
What if that junk PFN happens to match the zeropage's PFN by sheer chance? IHMO, it's really unlikely, but it would be really bad if it did.
Clearly,
pte_present()prevents this ;)Yeah, not so clearly kinda the whole point I'm making here. I mean all this code sucks because we have deeply nested conditions and you have to keep in your mind that 'oh we already checked for X so we can do Y'.
I see :)
Quoted 2 lines
But the THP code is horrible in general.
Yep, you'll get no argument from me on that one ;p
The code is indeed tricky ...
Quoted 3 lines
Anyway let's stay focused (so I can stand a chance of catching up with my post-vacation backlog :), I will check the respin once you send!
Cheers!
This series has been dropped from mm-new. I'm going to take a bit to rethink the approach based on the feedback.
Quoted 10 lines
Quoted 6 lines
By the way, I noticed there are other places in
khugepaged.cthat callpte_pfn()without being under apte_present()check.Perhaps those should all be fixed as well in a separate patch?
Yeah I'm not surprised and sure indeed.
Thanks, Lorenzo
In the meantime, I'll send up a separate patch for the missing pte_present() checks first ;)
Thanks,
Lance
Re: [PATCH mm-new v3 3/3] mm/khugepaged: merge PTE scanning logic into a new helper
In reply to Lorenzo StoakesQuoted 69 lines · Lorenzo Stoakes
Quoted 5 lines · Lance Yang
From: Lance Yang <[email protected]>
As David suggested, the PTE scanning logic in
hpage_collapse_scan_pmd()and__collapse_huge_page_isolate()was almost duplicated.Quoted 7 lines
This patch cleans things up by moving all the common PTE checking logic into a new shared helper,
thp_collapse_check_pte(). While at it, we usevm_normal_folio()instead ofvm_normal_page().Suggested-by David Hildenbrand <[email protected]>Suggested-by Dev Jain <[email protected]>Signed-off-by Lance Yang <[email protected]>In general I like the idea, the implementation needs work though.
Will check in more detail on respin
Quoted 9 lines
--- mm/khugepaged.c | 243 ++++++++++++++++++++++++++---------------------- 1 file changed, 130 insertions(+), 113 deletions(-)mm/khugepaged.c +0
-0mm/khugepaged.c enum scan_result { · line 66
Quoted 6 lines
mm/khugepaged.c +3
-0@@ -61,6 +61,12 @@ enum scan_result { SCAN_PAGE_FILLED, }; +enum pte_check_result { +PTE_CHECK_SUCCEED , +PTE_CHECK_CONTINUE ,Don't love this logic - this feels like we're essentially abstracting control flow, I mean what does 'continue' mean here? Other than you're in a loop and please continue, which is relying a little too much on what the caller does.
if we retain this logic something like
PTE_CHECK_SKIPwould make more sense.Quoted 13 lines
mm/khugepaged.c +6
-0+ PTE_CHECK_FAIL, +}; +#define CREATE_TRACE_POINTS #include <trace/events/huge_memory.h> @@ -533,62 +539,139 @@ static void release_pte_pages(pte_t *pte, pte_t *_pte, } } +/* +* thp_collapse_check_pte - Check if a PTE is suitable for THP collapse +* @pte: The PTE to check mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 554
Quoted 10 lines
mm/khugepaged.c +10
-0+* @vma: The VMA the PTE belongs to +* @addr: The virtual address corresponding to this PTE +* @foliop: On success, used to return a pointer to the folio +* Must be non-NULL +* @none_or_zero: Counter for none/zero PTEs. Must be non-NULL +* @unmapped: Counter for swap PTEs. Can be NULL if not scanning swaps +* @shared: Counter for shared pages. Must be non-NULL +* @scan_result: Used to return the failure reason (SCAN_*) on a +* PTE_CHECK_FAIL return. Must be non-NULL +* @cc: Collapse control settings Do we really need this many parameters? THis is hard to follow.
Quoted 1 line
Of course it being me, I'd immediately prefer a helper struct :)
Hey!
Ive been trying to reimplement this patch, and started converting this to a helper struct... At first glance it seems like the right approach, but because the scan/isolate share two slightly different implementations, we need to leave some logic in the parent. This results in polluting the rest of the code significantly just to save one ugly function call with many parameters.
not sure what to do about that tbh.
e.g., (just one of many examples)
- NR_ISOLATED_ANON + folio_is_file_lru(folio),
- folio_nr_pages(folio));
- VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio);
- VM_BUG_ON_FOLIO(folio_test_lru(folio), folio);
+ NR_ISOLATED_ANON + folio_is_file_lru(ctx.folio),
+ folio_nr_pages(ctx.folio));
+ VM_BUG_ON_FOLIO(!folio_test_locked(ctx.folio), ctx.folio);
+ VM_BUG_ON_FOLIO(folio_test_lru(ctx.folio), ctx.folio);
- if (folio_test_large(folio))
- list_add_tail(&folio->lru, compound_pagelist);
+ if (folio_test_large(ctx.folio))
+ list_add_tail(ctx.folio->lru, compound_pagelist);
Are we sure this helper structure is the right approach?
-- Nico
Quoted 353 lines
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 561
Quoted 8 lines
mm/khugepaged.c +7
-0+* +* Returns: +* PTE_CHECK_SUCCEED - PTE is suitable, proceed with further checks +* PTE_CHECK_CONTINUE - Skip this PTE and continue scanning +* PTE_CHECK_FAIL - Abort collapse scan +*/ +static inline int thp_collapse_check_pte(pte_t pte, struct vm_area_struct *vma,There's no need for inline in an internal static function in a file.
Quoted 8 lines
mm/khugepaged.c +8
-0+ unsigned long addr, struct folio **foliop, int *none_or_zero, + int *unmapped, int *shared, int *scan_result, + struct collapse_control *cc) +{ + struct folio *folio = NULL; + +if (pte_none (pte) ||is_zero_pfn (pte_pfn (pte))) { + (*none_or_zero)++;mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 579
Quoted 10 lines
mm/khugepaged.c +10
-0+if (!userfaultfd_armed (vma) && + (!cc->is_khugepaged || + *none_or_zero <= khugepaged_max_ptes_none)) { +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_NONE_PTE ; +count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (!pte_present (pte)) {You're now making the if-else issues with previous patches worse by returning which gets even checkpatch to warn you.
if (pte_none(pte) || is_zero_pfn(pte_pfn(pte))) {(of course note that I am not convinced you can drop the
pte_present()check here)(*none_or_zero)++; if (!userfaultfd_armed(vma) && (!cc->is_khugepaged || *none_or_zero <= khugepaged_max_ptes_none)) return PTE_CHECK_CONTINUE; *scan_result = SCAN_EXCEED_NONE_PTE; count_vm_event(THP_SCAN_EXCEED_NONE_PTE); return PTE_CHECK_FAIL; } if (!pte_present(pte)) { ... }But even that really needs seriously more refactoring - that condition above is horrible for instance so, e.g.:
if (pte_none(pte) || is_zero_pfn(pte_pfn(pte))) { (*none_or_zero)++; /* Cases where this is acceptable. */ if (!userfaultfd_armed(vma)) return PTE_CHECK_SKIP; if (!cc->is_khugepaged) return PTE_CHECK_SKIP; if (*none_or_zero <= khugepaged_max_ptes_none) return PTE_CHECK_SKIP; /* Otherwise, we must abort the scan. */ *scan_result = SCAN_EXCEED_NONE_PTE; count_vm_event(THP_SCAN_EXCEED_NONE_PTE); return PTE_CHECK_FAIL; } if (!pte_present(pte)) { ... }Improves things a lot.
I do however _hate_ this (*blah)++; thing. A helper struct would help :)
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 582
Quoted 3 lines
mm/khugepaged.c +3
-0+if (!unmapped) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ;Can't we have a helper that sets the result, e.g.:
return pte_check_fail(scan_result, SCAN_PTE_NON_PRESENT);static enum
pte_check_resultpte_check_fail(int*scan_result,
enumpte_check_resultresult){ *scan_result = result; return PTE_CHECK_FAIL; }And same for success/skip
Then all of these horrible if (blah) { *foo = bar; return baz; } can be made into
if (blah) return pte_check_xxx(..., SCAN_PTE_...);Quoted 28 lines
mm/khugepaged.c +28
-0+ } + +if (non_swap_entry (pte_to_swp_entry (pte))) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + + (*unmapped)++; +if (!cc->is_khugepaged || + *unmapped <= khugepaged_max_ptes_swap) { +/* +* Always be strict with uffd-wp enabled swap +* entries. Please see comment below for +* pte_uffd_wp(). +*/ +if (pte_swp_uffd_wp (pte)) { + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_SWAP_PTE ; +count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (pte_uffd_wp (pte)) { +/* +* Don't collapse the page if any of the small PTEs are mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 620
Quoted 10 lines
mm/khugepaged.c +10
-0+* armed with uffd write protection. Here we can also mark +* the new huge pmd as write protected if any of the small +* ones is marked but that could bring unknown userfault +* messages that falls outside of the registered range. +* So, just be simple. +*/ + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } +Again as all of the comments for previous series still apply here so obviously I don't like this as-is :)
And as Andrew has pointed out, now checkpatch is finding the 'pointless else' issue (as mentioned above also).
Quoted 201 lines
mm/khugepaged.c +45
-114+ folio =vm_normal_folio (vma, addr, pte); +if (unlikely (!folio) ||unlikely (folio_is_zone_device (folio))) { + *scan_result =SCAN_PAGE_NULL ; +return PTE_CHECK_FAIL ; + } + +if (!folio_test_anon (folio)) { +VM_WARN_ON_FOLIO (true, folio); + *scan_result =SCAN_PAGE_ANON ; +return PTE_CHECK_FAIL ; + } + +/* +* We treat a single page as shared if any part of the THP +* is shared. +*/ +if (folio_maybe_mapped_shared (folio)) { + (*shared)++; +if (cc->is_khugepaged && *shared > khugepaged_max_ptes_shared) { + *scan_result =SCAN_EXCEED_SHARED_PTE ; +count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); +return PTE_CHECK_FAIL ; + } + } + + *foliop = folio; + +return PTE_CHECK_SUCCEED ; +} +static int __collapse_huge_page_isolate (struct vm_area_struct *vma,unsigned long start_addr,pte_t *pte,struct collapse_control *cc,struct list_head *compound_pagelist) { -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr = start_addr;pte_t *_pte;int none_or_zero =0 , shared =0 , result =SCAN_FAIL , referenced =0 ; +int pte_check_res;for (_pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out; - } - }else if (!pte_present (pteval)) { - result =SCAN_PTE_NON_PRESENT ; - goto out; - }else if (pte_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out; - } - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out; - } - folio =page_folio (page); -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; - goto out; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero,NULL , &shared, + &result, cc); -/* See hpage_collapse_scan_pmd(). */ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out; - } - } +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) + goto out;if (folio_test_large (folio)) {struct folio *f;@@ -1264,11 +1347,11 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,pte_t *pte, *_pte;int result =SCAN_FAIL , referenced =0 ;int none_or_zero =0 , shared =0 ; -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr;spinlock_t *ptl;int node =NUMA_NO_NODE , unmapped =0 ; +int pte_check_res;VM_BUG_ON (start_addr & ~HPAGE_PMD_MASK );@@ -1287,81 +1370,15 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,for (addr = start_addr, _pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out_unmap; - } - }else if (!pte_present (pteval)) { -if (non_swap_entry (pte_to_swp_entry (pteval))) { - result =SCAN_PTE_NON_PRESENT ; - goto out_unmap; - } - ++unmapped; -if (!cc->is_khugepaged || - unmapped <= khugepaged_max_ptes_swap) { -/* -* Always be strict with uffd-wp -* enabled swap entries. Please see -* comment below for pte_uffd_wp(). -*/ -if (pte_swp_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } -continue ; - }else { - result =SCAN_EXCEED_SWAP_PTE ; -count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); - goto out_unmap; - } - }else if (pte_uffd_wp (pteval)) { -/* -* Don't collapse the page if any of the small -* PTEs are armed with uffd write protection. -* Here we can also mark the new huge pmd as -* write protected if any of the small ones is -* marked but that could bring unknown -* userfault messages that falls outside of -* the registered range. So, just be simple. -*/ - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero, &unmapped, + &shared, &result, cc); - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out_unmap; - } - folio =page_folio (page); - -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) goto out_unmap; - } - -/* -* We treat a single page as shared if any part of the THP -* is shared. -*/ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out_unmap; - } - }/* * Record which node the original page is from and save this -- 2.49.0
Lance Yang
reviewerRe: [PATCH mm-new v3 3/3] mm/khugepaged: merge PTE scanning logic into a new helper
+Cc David and Lorenzo, address oops fixed :)
Quoted 73 lines · Nico Pache
Quoted 69 lines · Lorenzo Stoakes
Quoted 5 lines · Lance Yang
From: Lance Yang <[email protected]>
As David suggested, the PTE scanning logic in
hpage_collapse_scan_pmd()and__collapse_huge_page_isolate()was almost duplicated.Quoted 7 lines
This patch cleans things up by moving all the common PTE checking logic into a new shared helper,
thp_collapse_check_pte(). While at it, we usevm_normal_folio()instead ofvm_normal_page().Suggested-by David Hildenbrand <[email protected]>Suggested-by Dev Jain <[email protected]>Signed-off-by Lance Yang <[email protected]>In general I like the idea, the implementation needs work though.
Will check in more detail on respin
Quoted 9 lines
--- mm/khugepaged.c | 243 ++++++++++++++++++++++++++---------------------- 1 file changed, 130 insertions(+), 113 deletions(-)mm/khugepaged.c +0
-0mm/khugepaged.c enum scan_result { · line 66
Quoted 6 lines
mm/khugepaged.c +3
-0@@ -61,6 +61,12 @@ enum scan_result { SCAN_PAGE_FILLED, }; +enum pte_check_result { +PTE_CHECK_SUCCEED , +PTE_CHECK_CONTINUE ,Don't love this logic - this feels like we're essentially abstracting control flow, I mean what does 'continue' mean here? Other than you're in a loop and please continue, which is relying a little too much on what the caller does.
if we retain this logic something like
PTE_CHECK_SKIPwould make more sense.Quoted 13 lines
mm/khugepaged.c +6
-0+ PTE_CHECK_FAIL, +}; +#define CREATE_TRACE_POINTS #include <trace/events/huge_memory.h> @@ -533,62 +539,139 @@ static void release_pte_pages(pte_t *pte, pte_t *_pte, } } +/* +* thp_collapse_check_pte - Check if a PTE is suitable for THP collapse +* @pte: The PTE to check mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 554
Quoted 10 lines
mm/khugepaged.c +10
-0+* @vma: The VMA the PTE belongs to +* @addr: The virtual address corresponding to this PTE +* @foliop: On success, used to return a pointer to the folio +* Must be non-NULL +* @none_or_zero: Counter for none/zero PTEs. Must be non-NULL +* @unmapped: Counter for swap PTEs. Can be NULL if not scanning swaps +* @shared: Counter for shared pages. Must be non-NULL +* @scan_result: Used to return the failure reason (SCAN_*) on a +* PTE_CHECK_FAIL return. Must be non-NULL +* @cc: Collapse control settings Do we really need this many parameters? THis is hard to follow.
Quoted 1 line
Of course it being me, I'd immediately prefer a helper struct :)
Quoted 8 lines
Hey!
Ive been trying to reimplement this patch, and started converting this to a helper struct... At first glance it seems like the right approach, but because the scan/isolate share two slightly different implementations, we need to leave some logic in the parent. This results in polluting the rest of the code significantly just to save one ugly function call with many parameters.
Been a while ... so I don't fully recall anymore :)
If you already have a patch or series, maybe just post it and we can look at the actual diff.
Probably easier than talking about it without seeing the code :D
Cheers, Lance
Quoted 379 lines
not sure what to do about that tbh.
e.g., (just one of many examples)
(quoted diff) +6
-6- NR_ISOLATED_ANON + folio_is_file_lru(folio), - folio_nr_pages(folio)); - VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio); - VM_BUG_ON_FOLIO(folio_test_lru(folio), folio); + NR_ISOLATED_ANON + folio_is_file_lru(ctx.folio), + folio_nr_pages(ctx.folio)); + VM_BUG_ON_FOLIO(!folio_test_locked(ctx.folio), ctx.folio); + VM_BUG_ON_FOLIO(folio_test_lru(ctx.folio), ctx.folio); - if (folio_test_large(folio)) - list_add_tail(&folio->lru, compound_pagelist); + if (folio_test_large(ctx.folio)) + list_add_tail(ctx.folio->lru, compound_pagelist);Are we sure this helper structure is the right approach?
-- Nico
Quoted 353 lines
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 561
Quoted 8 lines
mm/khugepaged.c +7
-0+* +* Returns: +* PTE_CHECK_SUCCEED - PTE is suitable, proceed with further checks +* PTE_CHECK_CONTINUE - Skip this PTE and continue scanning +* PTE_CHECK_FAIL - Abort collapse scan +*/ +static inline int thp_collapse_check_pte(pte_t pte, struct vm_area_struct *vma,There's no need for inline in an internal static function in a file.
Quoted 8 lines
mm/khugepaged.c +8
-0+ unsigned long addr, struct folio **foliop, int *none_or_zero, + int *unmapped, int *shared, int *scan_result, + struct collapse_control *cc) +{ + struct folio *folio = NULL; + +if (pte_none (pte) ||is_zero_pfn (pte_pfn (pte))) { + (*none_or_zero)++;mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 579
Quoted 10 lines
mm/khugepaged.c +10
-0+if (!userfaultfd_armed (vma) && + (!cc->is_khugepaged || + *none_or_zero <= khugepaged_max_ptes_none)) { +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_NONE_PTE ; +count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (!pte_present (pte)) {You're now making the if-else issues with previous patches worse by returning which gets even checkpatch to warn you.
if (pte_none(pte) || is_zero_pfn(pte_pfn(pte))) {(of course note that I am not convinced you can drop the
pte_present()check here)(*none_or_zero)++; if (!userfaultfd_armed(vma) && (!cc->is_khugepaged || *none_or_zero <= khugepaged_max_ptes_none)) return PTE_CHECK_CONTINUE; *scan_result = SCAN_EXCEED_NONE_PTE; count_vm_event(THP_SCAN_EXCEED_NONE_PTE); return PTE_CHECK_FAIL; } if (!pte_present(pte)) { ... }But even that really needs seriously more refactoring - that condition above is horrible for instance so, e.g.:
if (pte_none(pte) || is_zero_pfn(pte_pfn(pte))) { (*none_or_zero)++; /* Cases where this is acceptable. */ if (!userfaultfd_armed(vma)) return PTE_CHECK_SKIP; if (!cc->is_khugepaged) return PTE_CHECK_SKIP; if (*none_or_zero <= khugepaged_max_ptes_none) return PTE_CHECK_SKIP; /* Otherwise, we must abort the scan. */ *scan_result = SCAN_EXCEED_NONE_PTE; count_vm_event(THP_SCAN_EXCEED_NONE_PTE); return PTE_CHECK_FAIL; } if (!pte_present(pte)) { ... }Improves things a lot.
I do however _hate_ this (*blah)++; thing. A helper struct would help :)
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 582
Quoted 3 lines
mm/khugepaged.c +3
-0+if (!unmapped) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ;Can't we have a helper that sets the result, e.g.:
return pte_check_fail(scan_result, SCAN_PTE_NON_PRESENT);static enum
pte_check_resultpte_check_fail(int*scan_result,
enumpte_check_resultresult){ *scan_result = result; return PTE_CHECK_FAIL; }And same for success/skip
Then all of these horrible if (blah) { *foo = bar; return baz; } can be made into
if (blah) return pte_check_xxx(..., SCAN_PTE_...);Quoted 28 lines
mm/khugepaged.c +28
-0+ } + +if (non_swap_entry (pte_to_swp_entry (pte))) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + + (*unmapped)++; +if (!cc->is_khugepaged || + *unmapped <= khugepaged_max_ptes_swap) { +/* +* Always be strict with uffd-wp enabled swap +* entries. Please see comment below for +* pte_uffd_wp(). +*/ +if (pte_swp_uffd_wp (pte)) { + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_SWAP_PTE ; +count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (pte_uffd_wp (pte)) { +/* +* Don't collapse the page if any of the small PTEs are mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 620
Quoted 10 lines
mm/khugepaged.c +10
-0+* armed with uffd write protection. Here we can also mark +* the new huge pmd as write protected if any of the small +* ones is marked but that could bring unknown userfault +* messages that falls outside of the registered range. +* So, just be simple. +*/ + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } +Again as all of the comments for previous series still apply here so obviously I don't like this as-is :)
And as Andrew has pointed out, now checkpatch is finding the 'pointless else' issue (as mentioned above also).
Quoted 201 lines
mm/khugepaged.c +45
-114+ folio =vm_normal_folio (vma, addr, pte); +if (unlikely (!folio) ||unlikely (folio_is_zone_device (folio))) { + *scan_result =SCAN_PAGE_NULL ; +return PTE_CHECK_FAIL ; + } + +if (!folio_test_anon (folio)) { +VM_WARN_ON_FOLIO (true, folio); + *scan_result =SCAN_PAGE_ANON ; +return PTE_CHECK_FAIL ; + } + +/* +* We treat a single page as shared if any part of the THP +* is shared. +*/ +if (folio_maybe_mapped_shared (folio)) { + (*shared)++; +if (cc->is_khugepaged && *shared > khugepaged_max_ptes_shared) { + *scan_result =SCAN_EXCEED_SHARED_PTE ; +count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); +return PTE_CHECK_FAIL ; + } + } + + *foliop = folio; + +return PTE_CHECK_SUCCEED ; +} +static int __collapse_huge_page_isolate (struct vm_area_struct *vma,unsigned long start_addr,pte_t *pte,struct collapse_control *cc,struct list_head *compound_pagelist) { -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr = start_addr;pte_t *_pte;int none_or_zero =0 , shared =0 , result =SCAN_FAIL , referenced =0 ; +int pte_check_res;for (_pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out; - } - }else if (!pte_present (pteval)) { - result =SCAN_PTE_NON_PRESENT ; - goto out; - }else if (pte_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out; - } - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out; - } - folio =page_folio (page); -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; - goto out; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero,NULL , &shared, + &result, cc); -/* See hpage_collapse_scan_pmd(). */ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out; - } - } +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) + goto out;if (folio_test_large (folio)) {struct folio *f;@@ -1264,11 +1347,11 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,pte_t *pte, *_pte;int result =SCAN_FAIL , referenced =0 ;int none_or_zero =0 , shared =0 ; -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr;spinlock_t *ptl;int node =NUMA_NO_NODE , unmapped =0 ; +int pte_check_res;VM_BUG_ON (start_addr & ~HPAGE_PMD_MASK );@@ -1287,81 +1370,15 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,for (addr = start_addr, _pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out_unmap; - } - }else if (!pte_present (pteval)) { -if (non_swap_entry (pte_to_swp_entry (pteval))) { - result =SCAN_PTE_NON_PRESENT ; - goto out_unmap; - } - ++unmapped; -if (!cc->is_khugepaged || - unmapped <= khugepaged_max_ptes_swap) { -/* -* Always be strict with uffd-wp -* enabled swap entries. Please see -* comment below for pte_uffd_wp(). -*/ -if (pte_swp_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } -continue ; - }else { - result =SCAN_EXCEED_SWAP_PTE ; -count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); - goto out_unmap; - } - }else if (pte_uffd_wp (pteval)) { -/* -* Don't collapse the page if any of the small -* PTEs are armed with uffd write protection. -* Here we can also mark the new huge pmd as -* write protected if any of the small ones is -* marked but that could bring unknown -* userfault messages that falls outside of -* the registered range. So, just be simple. -*/ - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero, &unmapped, + &shared, &result, cc); - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out_unmap; - } - folio =page_folio (page); - -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) goto out_unmap; - } - -/* -* We treat a single page as shared if any part of the THP -* is shared. -*/ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out_unmap; - } - }/* * Record which node the original page is from and save this -- 2.49.0
Re: [PATCH mm-new v3 3/3] mm/khugepaged: merge PTE scanning logic into a new helper
Quoted 86 lines · Lance Yang
+Cc David and Lorenzo, address oops fixed :)
Quoted 73 lines · Nico Pache
Quoted 69 lines · Lorenzo Stoakes
Quoted 5 lines · Lance Yang
From: Lance Yang <[email protected]>
As David suggested, the PTE scanning logic in
hpage_collapse_scan_pmd()and__collapse_huge_page_isolate()was almost duplicated.Quoted 7 lines
This patch cleans things up by moving all the common PTE checking logic into a new shared helper,
thp_collapse_check_pte(). While at it, we usevm_normal_folio()instead ofvm_normal_page().Suggested-by David Hildenbrand <[email protected]>Suggested-by Dev Jain <[email protected]>Signed-off-by Lance Yang <[email protected]>In general I like the idea, the implementation needs work though.
Will check in more detail on respin
Quoted 9 lines
--- mm/khugepaged.c | 243 ++++++++++++++++++++++++++---------------------- 1 file changed, 130 insertions(+), 113 deletions(-)mm/khugepaged.c +0
-0mm/khugepaged.c enum scan_result { · line 66
Quoted 6 lines
mm/khugepaged.c +3
-0@@ -61,6 +61,12 @@ enum scan_result { SCAN_PAGE_FILLED, }; +enum pte_check_result { +PTE_CHECK_SUCCEED , +PTE_CHECK_CONTINUE ,Don't love this logic - this feels like we're essentially abstracting control flow, I mean what does 'continue' mean here? Other than you're in a loop and please continue, which is relying a little too much on what the caller does.
if we retain this logic something like
PTE_CHECK_SKIPwould make more sense.Quoted 13 lines
mm/khugepaged.c +6
-0+ PTE_CHECK_FAIL, +}; +#define CREATE_TRACE_POINTS #include <trace/events/huge_memory.h> @@ -533,62 +539,139 @@ static void release_pte_pages(pte_t *pte, pte_t *_pte, } } +/* +* thp_collapse_check_pte - Check if a PTE is suitable for THP collapse +* @pte: The PTE to check mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 554
Quoted 10 lines
mm/khugepaged.c +10
-0+* @vma: The VMA the PTE belongs to +* @addr: The virtual address corresponding to this PTE +* @foliop: On success, used to return a pointer to the folio +* Must be non-NULL +* @none_or_zero: Counter for none/zero PTEs. Must be non-NULL +* @unmapped: Counter for swap PTEs. Can be NULL if not scanning swaps +* @shared: Counter for shared pages. Must be non-NULL +* @scan_result: Used to return the failure reason (SCAN_*) on a +* PTE_CHECK_FAIL return. Must be non-NULL +* @cc: Collapse control settings Do we really need this many parameters? THis is hard to follow.
Quoted 1 line
Of course it being me, I'd immediately prefer a helper struct :)
Quoted 8 lines
Hey!
Ive been trying to reimplement this patch, and started converting this to a helper struct... At first glance it seems like the right approach, but because the scan/isolate share two slightly different implementations, we need to leave some logic in the parent. This results in polluting the rest of the code significantly just to save one ugly function call with many parameters.
Quoted 6 lines
Been a while ... so I don't fully recall anymore :)
If you already have a patch or series, maybe just post it and we can look at the actual diff.
Probably easier than talking about it without seeing the code :D
Yeah fair point!
I think I found a good intermediate solution after all. Which was to keep the folio local (not part of the helper), as that was the major source of the noise.
I have a work-in-progress series to address a number of cleanups, the code should be ready at the start of next week. I just want to test and verify things first, and it's best not to post at the end of the week :)
-- Nico
Quoted 383 lines
Cheers, Lance
Quoted 379 lines
not sure what to do about that tbh.
e.g., (just one of many examples)
(quoted diff) +6
-6- NR_ISOLATED_ANON + folio_is_file_lru(folio), - folio_nr_pages(folio)); - VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio); - VM_BUG_ON_FOLIO(folio_test_lru(folio), folio); + NR_ISOLATED_ANON + folio_is_file_lru(ctx.folio), + folio_nr_pages(ctx.folio)); + VM_BUG_ON_FOLIO(!folio_test_locked(ctx.folio), ctx.folio); + VM_BUG_ON_FOLIO(folio_test_lru(ctx.folio), ctx.folio); - if (folio_test_large(folio)) - list_add_tail(&folio->lru, compound_pagelist); + if (folio_test_large(ctx.folio)) + list_add_tail(ctx.folio->lru, compound_pagelist);Are we sure this helper structure is the right approach?
-- Nico
Quoted 353 lines
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 561
Quoted 8 lines
mm/khugepaged.c +7
-0+* +* Returns: +* PTE_CHECK_SUCCEED - PTE is suitable, proceed with further checks +* PTE_CHECK_CONTINUE - Skip this PTE and continue scanning +* PTE_CHECK_FAIL - Abort collapse scan +*/ +static inline int thp_collapse_check_pte(pte_t pte, struct vm_area_struct *vma,There's no need for inline in an internal static function in a file.
Quoted 8 lines
mm/khugepaged.c +8
-0+ unsigned long addr, struct folio **foliop, int *none_or_zero, + int *unmapped, int *shared, int *scan_result, + struct collapse_control *cc) +{ + struct folio *folio = NULL; + +if (pte_none (pte) ||is_zero_pfn (pte_pfn (pte))) { + (*none_or_zero)++;mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 579
Quoted 10 lines
mm/khugepaged.c +10
-0+if (!userfaultfd_armed (vma) && + (!cc->is_khugepaged || + *none_or_zero <= khugepaged_max_ptes_none)) { +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_NONE_PTE ; +count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (!pte_present (pte)) {You're now making the if-else issues with previous patches worse by returning which gets even checkpatch to warn you.
if (pte_none(pte) || is_zero_pfn(pte_pfn(pte))) {(of course note that I am not convinced you can drop the
pte_present()check here)(*none_or_zero)++; if (!userfaultfd_armed(vma) && (!cc->is_khugepaged || *none_or_zero <= khugepaged_max_ptes_none)) return PTE_CHECK_CONTINUE; *scan_result = SCAN_EXCEED_NONE_PTE; count_vm_event(THP_SCAN_EXCEED_NONE_PTE); return PTE_CHECK_FAIL; } if (!pte_present(pte)) { ... }But even that really needs seriously more refactoring - that condition above is horrible for instance so, e.g.:
if (pte_none(pte) || is_zero_pfn(pte_pfn(pte))) { (*none_or_zero)++; /* Cases where this is acceptable. */ if (!userfaultfd_armed(vma)) return PTE_CHECK_SKIP; if (!cc->is_khugepaged) return PTE_CHECK_SKIP; if (*none_or_zero <= khugepaged_max_ptes_none) return PTE_CHECK_SKIP; /* Otherwise, we must abort the scan. */ *scan_result = SCAN_EXCEED_NONE_PTE; count_vm_event(THP_SCAN_EXCEED_NONE_PTE); return PTE_CHECK_FAIL; } if (!pte_present(pte)) { ... }Improves things a lot.
I do however _hate_ this (*blah)++; thing. A helper struct would help :)
mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 582
Quoted 3 lines
mm/khugepaged.c +3
-0+if (!unmapped) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ;Can't we have a helper that sets the result, e.g.:
return pte_check_fail(scan_result, SCAN_PTE_NON_PRESENT);static enum
pte_check_resultpte_check_fail(int*scan_result,
enumpte_check_resultresult){ *scan_result = result; return PTE_CHECK_FAIL; }And same for success/skip
Then all of these horrible if (blah) { *foo = bar; return baz; } can be made into
if (blah) return pte_check_xxx(..., SCAN_PTE_...);Quoted 28 lines
mm/khugepaged.c +28
-0+ } + +if (non_swap_entry (pte_to_swp_entry (pte))) { + *scan_result =SCAN_PTE_NON_PRESENT ; +return PTE_CHECK_FAIL ; + } + + (*unmapped)++; +if (!cc->is_khugepaged || + *unmapped <= khugepaged_max_ptes_swap) { +/* +* Always be strict with uffd-wp enabled swap +* entries. Please see comment below for +* pte_uffd_wp(). +*/ +if (pte_swp_uffd_wp (pte)) { + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } +return PTE_CHECK_CONTINUE ; + }else { + *scan_result =SCAN_EXCEED_SWAP_PTE ; +count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); +return PTE_CHECK_FAIL ; + } + }else if (pte_uffd_wp (pte)) { +/* +* Don't collapse the page if any of the small PTEs are mm/khugepaged.c static void release_pte_pages(pte_t *pte, pte_t *_pte, · line 620
Quoted 10 lines
mm/khugepaged.c +10
-0+* armed with uffd write protection. Here we can also mark +* the new huge pmd as write protected if any of the small +* ones is marked but that could bring unknown userfault +* messages that falls outside of the registered range. +* So, just be simple. +*/ + *scan_result =SCAN_PTE_UFFD_WP ; +return PTE_CHECK_FAIL ; + } +Again as all of the comments for previous series still apply here so obviously I don't like this as-is :)
And as Andrew has pointed out, now checkpatch is finding the 'pointless else' issue (as mentioned above also).
Quoted 201 lines
mm/khugepaged.c +45
-114+ folio =vm_normal_folio (vma, addr, pte); +if (unlikely (!folio) ||unlikely (folio_is_zone_device (folio))) { + *scan_result =SCAN_PAGE_NULL ; +return PTE_CHECK_FAIL ; + } + +if (!folio_test_anon (folio)) { +VM_WARN_ON_FOLIO (true, folio); + *scan_result =SCAN_PAGE_ANON ; +return PTE_CHECK_FAIL ; + } + +/* +* We treat a single page as shared if any part of the THP +* is shared. +*/ +if (folio_maybe_mapped_shared (folio)) { + (*shared)++; +if (cc->is_khugepaged && *shared > khugepaged_max_ptes_shared) { + *scan_result =SCAN_EXCEED_SHARED_PTE ; +count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); +return PTE_CHECK_FAIL ; + } + } + + *foliop = folio; + +return PTE_CHECK_SUCCEED ; +} +static int __collapse_huge_page_isolate (struct vm_area_struct *vma,unsigned long start_addr,pte_t *pte,struct collapse_control *cc,struct list_head *compound_pagelist) { -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr = start_addr;pte_t *_pte;int none_or_zero =0 , shared =0 , result =SCAN_FAIL , referenced =0 ; +int pte_check_res;for (_pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out; - } - }else if (!pte_present (pteval)) { - result =SCAN_PTE_NON_PRESENT ; - goto out; - }else if (pte_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out; - } - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out; - } - folio =page_folio (page); -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; - goto out; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero,NULL , &shared, + &result, cc); -/* See hpage_collapse_scan_pmd(). */ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out; - } - } +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) + goto out;if (folio_test_large (folio)) {struct folio *f;@@ -1264,11 +1347,11 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,pte_t *pte, *_pte;int result =SCAN_FAIL , referenced =0 ;int none_or_zero =0 , shared =0 ; -struct page *page =NULL ;struct folio *folio =NULL ;unsigned long addr;spinlock_t *ptl;int node =NUMA_NO_NODE , unmapped =0 ; +int pte_check_res;VM_BUG_ON (start_addr & ~HPAGE_PMD_MASK );@@ -1287,81 +1370,15 @@ static int hpage_collapse_scan_pmd(struct mm_struct *mm,for (addr = start_addr, _pte = pte; _pte < pte +HPAGE_PMD_NR ; _pte++, addr +=PAGE_SIZE ) {pte_t pteval =ptep_get (_pte); -if (pte_none (pteval) ||is_zero_pfn (pte_pfn (pteval))) { - ++none_or_zero; -if (!userfaultfd_armed (vma) && - (!cc->is_khugepaged || - none_or_zero <= khugepaged_max_ptes_none)) { -continue ; - }else { - result =SCAN_EXCEED_NONE_PTE ; -count_vm_event (THP_SCAN_EXCEED_NONE_PTE ); - goto out_unmap; - } - }else if (!pte_present (pteval)) { -if (non_swap_entry (pte_to_swp_entry (pteval))) { - result =SCAN_PTE_NON_PRESENT ; - goto out_unmap; - } - ++unmapped; -if (!cc->is_khugepaged || - unmapped <= khugepaged_max_ptes_swap) { -/* -* Always be strict with uffd-wp -* enabled swap entries. Please see -* comment below for pte_uffd_wp(). -*/ -if (pte_swp_uffd_wp (pteval)) { - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } -continue ; - }else { - result =SCAN_EXCEED_SWAP_PTE ; -count_vm_event (THP_SCAN_EXCEED_SWAP_PTE ); - goto out_unmap; - } - }else if (pte_uffd_wp (pteval)) { -/* -* Don't collapse the page if any of the small -* PTEs are armed with uffd write protection. -* Here we can also mark the new huge pmd as -* write protected if any of the small ones is -* marked but that could bring unknown -* userfault messages that falls outside of -* the registered range. So, just be simple. -*/ - result =SCAN_PTE_UFFD_WP ; - goto out_unmap; - } + pte_check_res =thp_collapse_check_pte (pteval, vma, addr, + &folio, &none_or_zero, &unmapped, + &shared, &result, cc); - page =vm_normal_page (vma, addr, pteval); -if (unlikely (!page) ||unlikely (is_zone_device_page (page))) { - result =SCAN_PAGE_NULL ; - goto out_unmap; - } - folio =page_folio (page); - -if (!folio_test_anon (folio)) { -VM_WARN_ON_FOLIO (true, folio); - result =SCAN_PAGE_ANON ; +if (pte_check_res ==PTE_CHECK_CONTINUE ) +continue ; +else if (pte_check_res ==PTE_CHECK_FAIL ) goto out_unmap; - } - -/* -* We treat a single page as shared if any part of the THP -* is shared. -*/ -if (folio_maybe_mapped_shared (folio)) { - ++shared; -if (cc->is_khugepaged && - shared > khugepaged_max_ptes_shared) { - result =SCAN_EXCEED_SHARED_PTE ; -count_vm_event (THP_SCAN_EXCEED_SHARED_PTE ); - goto out_unmap; - } - }/* * Record which node the original page is from and save this -- 2.49.0
Re: [PATCH mm-new v3 3/3] mm/khugepaged: merge PTE scanning logic into a new helper
Quoted 5 lines · Nico Pache
I have a work-in-progress series to address a number of cleanups, the code should be ready at the start of next week. I just want to test and verify things first, and it's best not to post at the end of the week :)
Woo! Thanks :) Cleanups very welcome.
But yeah best to send on week of -rc1 please. I'm holding off on a pretty large series myself and we need to get into the habit of treating the merge window as quiet time.
I think Andrew is not taking anything atm anyway and asking people to resend in -rc1 anyway.
(There's a debate as to - send early, get early feedback, resend in -rc1 vs. tidal wave in -rc1, but I'm kinda in favour of the tidal wave, as it at least gives some breathing space. I am very very far behind on review rn :)
Thanks, Lorenzo