mm/page_alloc: fix initialization of tags of the huge zero folio with init_on_free
Subsystems (5)
David Hildenbrand (Arm)
maintainerPATCH v2 1/1 mm/page_alloc: fix initialization of tags of the huge zero folio with init_on_free
__GFP_ZEROTAGS semantics are currently a bit weird, but effectively this flag is only ever set alongside __GFP_ZERO and __GFP_SKIP_KASAN.
If we run with init_on_free, we will zero out pages during __free_pages_prepare(), to skip zeroing on the allocation path.
However, when allocating with __GFP_ZEROTAG set, post_alloc_hook() will consequently not only skip clearing page content, but also skip clearing tag memory.
Not clearing tags through __GFP_ZEROTAGS is irrelevant for most pages that will get mapped to user space through set_pte_at() later: set_pte_at() and friends will detect that the tags have not been initialized yet (PG_mte_tagged not set), and initialize them.
However, for the huge zero folio, which will be mapped through a PMD marked as special, this initialization will not be performed, ending up exposing whatever tags were still set for the pages.
The docs (Documentation/arch/arm64/memory-tagging-extension.rst) state that allocation tags are set to 0 when a page is first mapped to user space. That no longer holds with the huge zero folio when init_on_free is enabled.
Fix it by decoupling __GFP_ZEROTAGS from __GFP_ZERO, passing to tag_clear_highpages() whether we want to also clear page content.
Invert the meaning of the tag_clear_highpages() return value to have clearer semantics.
Reproduced with the huge zero folio by modifying the check_buffer_fill arm64/mte selftest to use a 2 MiB area, after making sure that pages have a non-0 tag set when freeing (note that, during boot, we will not actually initialize tags, but only set KASAN_TAG_KERNEL in the page flags).
$ ./check_buffer_fill 1..20 ... not ok 17 Check initial tags with private mapping, sync error mode and mmap memory not ok 18 Check initial tags with private mapping, sync error mode and mmap/mprotect memory ...
This code needs more cleanups; we'll tackle that next, like decoupling __GFP_ZEROTAGS from __GFP_SKIP_KASAN.
---
Changes in v2:
- Drop kasan_hw_tags_enabled() handling, as it missed the case of user-space MTE without KASAN.
- Keep letting_clear_highpages() return a bool and re-instantiate system_supports_mte() handling in the arm64 variant.
- Rephrase __GFP_ZEROTAGS comment, making it clearer that this is not just a performance improvement.
- Retested and more excessively build tested
- Using a new b4 template, hopefully that doesn't mess things up
- Link to v1: https://lore.kernel.org/r/[email protected]
--- arch/arm64/include/asm/page.h | 2 +- arch/arm64/mm/fault.c | 11 +++++++---- include/linux/gfp_types.h | 10 +++++----- include/linux/highmem.h | 7 ++++--- mm/page_alloc.c | 8 ++++---- 5 files changed, 21 insertions(+), 17 deletions(-)
arch/arm64/include/asm/page.h +1 -1
@@ -33,7 +33,7 @@ struct folio *vma_alloc_zeroed_movable_folio(struct vm_area_struct *vma, unsigned long vaddr);#define vma_alloc_zeroed_movable_folio vma_alloc_zeroed_movable_folio -bool tag_clear_highpages (struct page *to,int numpages); +bool tag_clear_highpages (struct page *to,int numpages,bool clear_pages);#define __HAVE_ARCH_TAG_CLEAR_HIGHPAGES#define copy_user_page (to, from, vaddr, pg) copy_page(to, from)
arch/arm64/mm/fault.c +7 -4
@@ -1018,7 +1018,7 @@ struct folio *vma_alloc_zeroed_movable_folio(struct vm_area_struct *vma,return vma_alloc_folio (flags,0 , vma, vaddr); } -bool tag_clear_highpages (struct page *page,int numpages) +bool tag_clear_highpages (struct page *page,int numpages,bool clear_pages) {/* * Check if MTE is supported and fall back to clear_highpage().
@@ -1026,13 +1026,16 @@ bool tag_clear_highpages(struct page *page, int numpages)* post_alloc_hook() will invoke tag_clear_highpages(). */ if (!system_supports_mte ()) -return false; +return clear_pages;/* Newly allocated pages, shouldn't have been tagged yet */ for (int i =0 ; i < numpages; i++, page++) {WARN_ON_ONCE (!try_page_mte_tagging (page)); -mte_zero_clear_page_tags (page_address (page)); +if (clear_pages) +mte_zero_clear_page_tags (page_address (page)); +else +mte_clear_page_tags (page_address (page));set_page_mte_tagged (page); } -return true; +return false; }
include/linux/gfp_types.h +5 -5
@@ -273,11 +273,11 @@ enum {* * %__GFP_ZERO returns a zeroed page on success. * -* %__GFP_ZEROTAGS zeroes memory tags at allocation time if the memory itself -* is being zeroed (either via __GFP_ZERO or via init_on_alloc, provided that -* __GFP_SKIP_ZERO is not set). This flag is intended for optimization: setting -* memory tags at the same time as zeroing memory has minimal additional -* performance impact. +* %__GFP_ZEROTAGS zeroes memory tags at allocation time. Setting memory tags at +* the same time as zeroing memory (e.g., with __GPF_ZERO) has minimal +* additional performance impact. However, __GFP_ZEROTAGS also zeroes the tags +* even if memory is not getting zeroed at allocation time (e.g., +* with init_on_free). * * %__GFP_SKIP_KASAN makes KASAN skip unpoisoning on page allocation. * Used for userspace and vmalloc pages; the latter are unpoisoned by
include/linux/highmem.h +4 -3
@@ -347,10 +347,11 @@ static inline void clear_highpage_kasan_tagged(struct page *page)#ifndef __HAVE_ARCH_TAG_CLEAR_HIGHPAGES -/* Return false to let people know we did not initialize the pages */ -static inline bool tag_clear_highpages (struct page *page,int numpages) +/* Returns true if the caller has to initialize the pages */ +static inline bool tag_clear_highpages (struct page *page,int numpages, +bool clear_pages) { -return false; +return clear_pages; }#endif
mm/page_alloc.c +4 -4
@@ -1808,9 +1808,9 @@ static inline bool should_skip_init(gfp_t flags)inline void post_alloc_hook (struct page *page,unsigned int order,gfp_t gfp_flags) { +const bool zero_tags = gfp_flags & __GFP_ZEROTAGS;bool init = !want_init_on_free () &&want_init_on_alloc (gfp_flags) && !should_skip_init (gfp_flags); -bool zero_tags = init && (gfp_flags & __GFP_ZEROTAGS);int i;set_page_private (page,0 );
@@ -1832,11 +1832,11 @@ inline void post_alloc_hook(struct page *page, unsigned int order,*/ /* -* If memory tags should be zeroed -* (which happens only when memory should be initialized as well). +* Clearing tags can efficiently clear the memory for us as well, if +* required. */ if (zero_tags) - init = !tag_clear_highpages (page,1 << order); + init =tag_clear_highpages (page,1 << order,/* clear_pages= */ init);if (!should_skip_kasan_unpoison (gfp_flags) &&kasan_unpoison_pages (page, order, init)) {
---
base-commit: f1541b40cd422d7e22273be9b7e9edfc9ea4f0d7
change-id: 20260417-zerotags-343a3673e18d
--
Cheers,
David
Quoted 43 lines · David Hildenbrand (Arm)
__GFP_ZEROTAGSsemantics are currently a bit weird, but effectively this flag is only ever set alongside__GFP_ZEROand__GFP_SKIP_KASAN.If we run with
init_on_free, we will zero out pages during__free_pages_prepare(), to skip zeroing on the allocation path.However, when allocating with
__GFP_ZEROTAGset,post_alloc_hook()will consequently not only skip clearing page content, but also skip clearing tag memory.Not clearing tags through
__GFP_ZEROTAGSis irrelevant for most pages that will get mapped to user space throughset_pte_at()later:set_pte_at()and friends will detect that the tags have not been initialized yet (PG_mte_taggednot set), and initialize them.However, for the huge zero folio, which will be mapped through a PMD marked as special, this initialization will not be performed, ending up exposing whatever tags were still set for the pages.
The docs (
Documentation/arch/arm64/memory-tagging-extension.rst) state that allocation tags are set to 0 when a page is first mapped to user space. That no longer holds with the huge zero folio wheninit_on_freeis enabled.Fix it by decoupling
__GFP_ZEROTAGSfrom__GFP_ZERO, passing totag_clear_highpages()whether we want to also clear page content.Invert the meaning of the
tag_clear_highpages()return value to have clearer semantics.Reproduced with the huge zero folio by modifying the
check_buffer_fillarm64/mte selftest to use a 2 MiB area, after making sure that pages have a non-0 tag set when freeing (note that, during boot, we will not actually initialize tags, but only setKASAN_TAG_KERNELin the page flags).$ ./check_buffer_fill 1..20 ... not ok 17 Check initial tags with private mapping, sync error mode and mmap memory not ok 18 Check initial tags with private mapping, sync error mode and mmap/mprotect memory ...
Quoted 7 lines
This code needs more cleanups; we'll tackle that next, like decoupling
__GFP_ZEROTAGSfrom__GFP_SKIP_KASAN.Fixes adfb6609c680 ("mm/huge_memory: initialise the tags of the huge zero folio")Signed-off-by David Hildenbrand (Arm) <[email protected]>---
Tested this fix on MTE with both kasan=on and kasan=off. Works as expected in both cases.
Nothing jumped out at me, LGTM!
David Hildenbrand (Arm)
maintainerQuoted 7 lines
}
include/linux/gfp_types.h +0
-0
include/linux/gfp_types.h enum { · line 277
Quoted 10 lines
include/linux/gfp_types.h +2
-5@@ -273,11 +273,11 @@ enum {* * %__GFP_ZERO returns a zeroed page on success. * -* %__GFP_ZEROTAGS zeroes memory tags at allocation time if the memory itself -* is being zeroed (either via __GFP_ZERO or via init_on_alloc, provided that -* __GFP_SKIP_ZERO is not set). This flag is intended for optimization: setting -* memory tags at the same time as zeroing memory has minimal additional -* performance impact. +* %__GFP_ZEROTAGS zeroes memory tags at allocation time. Setting memory tags at +* the same time as zeroing memory (e.g., with __GPF_ZERO) has minimal
As pointed out by AI review s/__GPF_ZERO/__GFP_ZERO/.
I assume this can be fixed up when applying.
Signature · 3 lines
Cheers, David
Quoted 43 lines · David Hildenbrand
__GFP_ZEROTAGSsemantics are currently a bit weird, but effectively this flag is only ever set alongside__GFP_ZEROand__GFP_SKIP_KASAN.If we run with
init_on_free, we will zero out pages during__free_pages_prepare(), to skip zeroing on the allocation path.However, when allocating with
__GFP_ZEROTAGset,post_alloc_hook()will consequently not only skip clearing page content, but also skip clearing tag memory.Not clearing tags through
__GFP_ZEROTAGSis irrelevant for most pages that will get mapped to user space throughset_pte_at()later:set_pte_at()and friends will detect that the tags have not been initialized yet (PG_mte_taggednot set), and initialize them.However, for the huge zero folio, which will be mapped through a PMD marked as special, this initialization will not be performed, ending up exposing whatever tags were still set for the pages.
The docs (
Documentation/arch/arm64/memory-tagging-extension.rst) state that allocation tags are set to 0 when a page is first mapped to user space. That no longer holds with the huge zero folio wheninit_on_freeis enabled.Fix it by decoupling
__GFP_ZEROTAGSfrom__GFP_ZERO, passing totag_clear_highpages()whether we want to also clear page content.Invert the meaning of the
tag_clear_highpages()return value to have clearer semantics.Reproduced with the huge zero folio by modifying the
check_buffer_fillarm64/mte selftest to use a 2 MiB area, after making sure that pages have a non-0 tag set when freeing (note that, during boot, we will not actually initialize tags, but only setKASAN_TAG_KERNELin the page flags).$ ./check_buffer_fill 1..20 ... not ok 17 Check initial tags with private mapping, sync error mode and mmap memory not ok 18 Check initial tags with private mapping, sync error mode and mmap/mprotect memory ...
Quoted 6 lines
This code needs more cleanups; we'll tackle that next, like decoupling
__GFP_ZEROTAGSfrom__GFP_SKIP_KASAN.Fixes adfb6609c680 ("mm/huge_memory: initialise the tags of the huge zero folio")Signed-off-by David Hildenbrand (Arm) <[email protected]>
The logic looks fine to me. Thanks!