[PATCH 0/2] mm/huge_memory: take a reference on a device DAX PUD copied by fork

[PATCH 0/2] mm/huge_memory: take a reference on a device DAX PUD copied by fork

[Konstantin Khorenko]

copy_huge_pud() copies a device DAX PUD into the child without the
reference, rmap and counter that vmf_insert_folio_pud() took, so the
child's unmap takes them from the parent. fork copies a device DAX VMA
once it has a guard region, so it takes no more than:

1. map a devdax instance with 1 GiB alignment and fault in a PUD;
2. MADV_GUARD_INSTALL a page elsewhere in the VMA;
3. fork, and let the child exit.

Patch 1 is the fix, patch 2 a selftest for it. I found it while
reworking copy_huge_pmd() for fork after David's review [1]; it is
independent of that series and applies to v7.3-rc7 on its own.

Testing
=======

 * v7.3-rc7 x86_64 KVM guest with CONFIG_DEBUG_VM
 * pmem from memmap=3G!4G
 * devdax made with "ndctl create-namespace -m devdax -a 1G":

  kernel              fork_dax_pud   log
  7.3-rc7             not ok         BUG: Bad rss-counter state mm:...
                                     type:MM_FILEPAGES val:-262144
  7.3-rc7 + patch 1   ok             clean

A variant that writes to the device and forks three times also hit,
on 7.3-rc7:

  WARNING: ./include/linux/rmap.h:159 at
  __folio_large_mapcount_sanity_checks+0x90/0x150

The selftest only runs on a device given in FORK_DAX_PUD_DEVICE or
argv[1], as on an unfixed kernel it leaves that device's folio with a
mapcount below zero. On pmem emulated in RAM, with
CONFIG_X86_PMEM_LEGACY and CONFIG_DEV_DAX_PMEM:

  boot with memmap=3G!4G
  ndctl create-namespace -f -e namespace0.0 -m devdax -a 1G
  FORK_DAX_PUD_DEVICE=/dev/dax0.0 ./fork_dax_pud

[1] https://lore.kernel.org/linux-mm/20261014-b4-prep-vstor-149847-copy_huge_pmd-recheck-special-pmd-v3-0-85eaec8417f6@virtuozzo.com/

Signed-off-by: Konstantin Khorenko <khorenko@virtuozzo.com>
---
Konstantin Khorenko (2):
      mm/huge_memory: take a reference on a folio PUD copied by fork
      selftests/mm: test fork of a device DAX PUD

 MAINTAINERS                               |   1 +
 include/linux/huge_mm.h                   |   2 +-
 include/linux/rmap.h                      |  22 +++++
 mm/huge_memory.c                          |  13 ++-
 mm/memory.c                               |   4 +-
 tools/testing/selftests/mm/Makefile       |   1 +
 tools/testing/selftests/mm/fork_dax_pud.c | 136 ++++++++++++++++++++++++++++++
 tools/testing/selftests/mm/run_vmtests.sh |   1 +
 8 files changed, 175 insertions(+), 5 deletions(-)
---
base-commit: a804c5742bff224d4cb4d25dd383554d598be32b
change-id: 20261021-vstor-150143-copy_huge_pud-folio-refcount-ea269e668270

Best regards,
--  
Konstantin Khorenko,
Virtuozzo Linux Kernel Team

[PATCH 1/2] mm/huge_memory: take a reference on a folio PUD copied by fork

[Konstantin Khorenko]

Device DAX maps its memory with vmf_insert_folio_pud(), which takes a
folio reference, adds a PUD rmap and accounts the pages to the mm.
zap_huge_pud() drops all three, but copy_huge_pud() copies the PUD into
the child without taking any of them. Each child's unmap then drops
what the parent took:

- the folio is a reference short, so with two children its refcount
  drops to zero and free_zone_device_folio() runs while the parent
  still maps it;
- the child's file counter goes negative, "Bad rss-counter state" when
  its mm is freed;
- the parent's unmap drives the entire mapcount below zero.

fork copies the page tables of a device DAX mapping that has a guard
region: commit f807123d578d ("mm: allow guard regions in file-backed and
read-only mappings") allowed MADV_GUARD_INSTALL there, and it makes fork
copy the VMA.

Take the reference, the rmap and the counter for a PUD that
vm_normal_page_pud() resolves to a page, as copy_huge_pmd() does for a
PMD, with a new folio_dup_file_rmap_pud() helper.

Fixes: aed877c2b425 ("device/dax: properly refcount device dax pages when mapping")
Cc: stable@vger.kernel.org
Assisted-by: LLM
Signed-off-by: Konstantin Khorenko <khorenko@virtuozzo.com>
---
 include/linux/huge_mm.h |  2 +-
 include/linux/rmap.h    | 22 ++++++++++++++++++++++
 mm/huge_memory.c        | 13 +++++++++++--
 mm/memory.c             |  4 ++--
 4 files changed, 36 insertions(+), 5 deletions(-)

diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
index c745f7ad22987..c5094f9d95ba9 100644
--- a/include/linux/huge_mm.h
+++ b/include/linux/huge_mm.h
@@ -14,7 +14,7 @@ int copy_huge_pmd(struct mm_struct *dst_mm, struct mm_struct *src_mm,
 bool huge_pmd_set_accessed(struct vm_fault *vmf);
 int copy_huge_pud(struct mm_struct *dst_mm, struct mm_struct *src_mm,
 		  pud_t *dst_pud, pud_t *src_pud, unsigned long addr,
-		  struct vm_area_struct *vma);
+		  struct vm_area_struct *dst_vma, struct vm_area_struct *src_vma);
 
 #ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
 void huge_pud_set_accessed(struct vm_fault *vmf, pud_t orig_pud);
diff --git a/include/linux/rmap.h b/include/linux/rmap.h
index 0b332770abeed..3c9ed8e27f0e4 100644
--- a/include/linux/rmap.h
+++ b/include/linux/rmap.h
@@ -564,6 +564,28 @@ static inline void folio_dup_file_rmap_pmd(struct folio *folio,
 #endif
 }
 
+/**
+ * folio_dup_file_rmap_pud - duplicate a PUD mapping of a page range of a folio
+ * @folio:	The folio to duplicate the mapping of
+ * @page:	The first page to duplicate the mapping of
+ * @dst_vma:	The destination vm area
+ *
+ * The page range of the folio is defined by [page, page + HPAGE_PUD_NR)
+ *
+ * The caller needs to hold the page table lock.
+ */
+static inline void folio_dup_file_rmap_pud(struct folio *folio,
+		struct page *page, struct vm_area_struct *dst_vma)
+{
+#if defined(CONFIG_TRANSPARENT_HUGEPAGE) && \
+	defined(CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD)
+	__folio_dup_file_rmap(folio, page, HPAGE_PUD_NR, dst_vma,
+			      PGTABLE_LEVEL_PUD);
+#else
+	WARN_ON_ONCE(true);
+#endif
+}
+
 static __always_inline int __folio_try_dup_anon_rmap(struct folio *folio,
 		struct page *page, int nr_pages, struct vm_area_struct *dst_vma,
 		struct vm_area_struct *src_vma, enum pgtable_level level)
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index 1e5d68acf62a5..7e8b41f293a49 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -2075,9 +2075,10 @@ void touch_pud(struct vm_area_struct *vma, unsigned long addr,
 
 int copy_huge_pud(struct mm_struct *dst_mm, struct mm_struct *src_mm,
 		  pud_t *dst_pud, pud_t *src_pud, unsigned long addr,
-		  struct vm_area_struct *vma)
+		  struct vm_area_struct *dst_vma, struct vm_area_struct *src_vma)
 {
 	spinlock_t *dst_ptl, *src_ptl;
+	struct page *page;
 	pud_t pud;
 	int ret;
 
@@ -2094,7 +2095,15 @@ int copy_huge_pud(struct mm_struct *dst_mm, struct mm_struct *src_mm,
 	 * TODO: once we support anonymous pages, use
 	 * folio_try_dup_anon_rmap_*() and split if duplicating fails.
 	 */
-	if (vma_is_cow_mapping(vma) && pud_write(pud)) {
+	page = vm_normal_page_pud(src_vma, addr, pud);
+	if (page) {
+		struct folio *folio = page_folio(page);
+
+		folio_get(folio);
+		folio_dup_file_rmap_pud(folio, page, dst_vma);
+		add_mm_counter(dst_mm, mm_counter_file(folio), HPAGE_PUD_NR);
+	}
+	if (vma_is_cow_mapping(src_vma) && pud_write(pud)) {
 		pudp_set_wrprotect(src_mm, addr, src_pud);
 		pud = pud_wrprotect(pud);
 	}
diff --git a/mm/memory.c b/mm/memory.c
index 8b0c2c735d3de..8acd4d5bb2310 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -1514,8 +1514,8 @@ copy_pud_range(struct vm_area_struct *dst_vma, struct vm_area_struct *src_vma,
 			int err;
 
 			VM_BUG_ON_VMA(next-addr != HPAGE_PUD_SIZE, src_vma);
-			err = copy_huge_pud(dst_mm, src_mm,
-					    dst_pud, src_pud, addr, src_vma);
+			err = copy_huge_pud(dst_mm, src_mm, dst_pud, src_pud,
+					    addr, dst_vma, src_vma);
 			if (err == -ENOMEM)
 				return -ENOMEM;
 			if (!err)

-- 
2.43.0

[PATCH 2/2] selftests/mm: test fork of a device DAX PUD

[Konstantin Khorenko]

Fork a process that maps a device DAX instance with a PUD, in a VMA
whose page tables fork copies because it has a guard region. The child
must account the copied PUD in its RssFile.

Without a reference, an rmap and a counter for the copied PUD, the
child's RssFile misses the PUD, and its exit leaves the child's file
counter negative: "Bad rss-counter state".

It runs only on a device named in argv[1] or $FORK_DAX_PUD_DEVICE: on
a kernel without the fix it leaves the device folio's mapcount below
zero.

Assisted-by: LLM
Signed-off-by: Konstantin Khorenko <khorenko@virtuozzo.com>
---
 MAINTAINERS                               |   1 +
 tools/testing/selftests/mm/Makefile       |   1 +
 tools/testing/selftests/mm/fork_dax_pud.c | 136 ++++++++++++++++++++++++++++++
 tools/testing/selftests/mm/run_vmtests.sh |   1 +
 4 files changed, 139 insertions(+)

diff --git a/MAINTAINERS b/MAINTAINERS
index 7b5cab72d851b..48653704007f3 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -17481,6 +17481,7 @@ F:	include/trace/events/huge_memory.h
 F:	mm/huge_memory.c
 F:	mm/khugepaged.c
 F:	mm/mm_slot.h
+F:	tools/testing/selftests/mm/fork_dax_pud.c
 F:	tools/testing/selftests/mm/khugepaged.c
 F:	tools/testing/selftests/mm/split_huge_page_test.c
 F:	tools/testing/selftests/mm/transhuge-stress.c
diff --git a/tools/testing/selftests/mm/Makefile b/tools/testing/selftests/mm/Makefile
index 2d5366196e309..86914aaf8e6ad 100644
--- a/tools/testing/selftests/mm/Makefile
+++ b/tools/testing/selftests/mm/Makefile
@@ -104,6 +104,7 @@ TEST_GEN_FILES += guard-regions
 TEST_GEN_FILES += merge
 TEST_GEN_FILES += rmap
 TEST_GEN_FILES += folio_split_race_test
+TEST_GEN_FILES += fork_dax_pud
 
 ifneq ($(ARCH),arm64)
 TEST_GEN_FILES += soft-dirty
diff --git a/tools/testing/selftests/mm/fork_dax_pud.c b/tools/testing/selftests/mm/fork_dax_pud.c
new file mode 100644
index 0000000000000..1fcf46341ace2
--- /dev/null
+++ b/tools/testing/selftests/mm/fork_dax_pud.c
@@ -0,0 +1,136 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * fork() of a process that maps a device DAX instance with a PUD, in a VMA
+ * whose page tables fork copies because it has a guard region. The child
+ * must account the copied PUD in its RssFile.
+ *
+ * Runs only on a device named in argv[1] or $FORK_DAX_PUD_DEVICE that
+ * nothing else maps, with 1 GiB alignment and at least 2 GiB. To make one
+ * on pmem emulated in RAM, with CONFIG_X86_PMEM_LEGACY and
+ * CONFIG_DEV_DAX_PMEM:
+ *
+ *	boot with memmap=3G!4G
+ *	ndctl create-namespace -f -e namespace0.0 -m devdax -a 1G
+ *	FORK_DAX_PUD_DEVICE=/dev/dax0.0 ./fork_dax_pud
+ */
+#define _GNU_SOURCE
+#include <asm-generic/mman.h> /* Force the import of the tools version. */
+#include <errno.h>
+#include <fcntl.h>
+#include <limits.h>
+#include <setjmp.h>
+#include <signal.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <sys/mman.h>
+#include <sys/wait.h>
+
+#include "kselftest.h"
+#include "vm_util.h"
+
+#define SZ_1G		(1UL << 30)
+#define DAX_SYSFS	"/sys/bus/dax/devices"
+
+static sigjmp_buf sigbus_jmp;
+
+static void sigbus_handler(int sig)
+{
+	siglongjmp(sigbus_jmp, 1);
+}
+
+static unsigned long sysfs_ul(const char *dev, const char *attr)
+{
+	char path[PATH_MAX], buf[64];
+	unsigned long val = 0;
+	FILE *f;
+
+	snprintf(path, sizeof(path), DAX_SYSFS "/%s/%s", dev, attr);
+	f = fopen(path, "r");
+	if (!f)
+		return 0;
+	if (fgets(buf, sizeof(buf), f))
+		val = strtoul(buf, NULL, 0);
+	fclose(f);
+	return val;
+}
+
+static bool dax_dev_ok(const char *path)
+{
+	const char *name = strrchr(path, '/');
+
+	name = name ? name + 1 : path;
+	return sysfs_ul(name, "align") == SZ_1G &&
+	       sysfs_ul(name, "size") >= 2 * SZ_1G;
+}
+
+static long rss_file_kb(void)
+{
+	char line[256];
+	long kb = -1;
+	FILE *f = fopen("/proc/self/status", "r");
+
+	if (!f)
+		return -1;
+	while (fgets(line, sizeof(line), f))
+		if (sscanf(line, "RssFile: %ld", &kb) == 1)
+			break;
+	fclose(f);
+	return kb;
+}
+
+int main(int argc, char **argv)
+{
+	const char *path = argc > 1 ? argv[1] : getenv("FORK_DAX_PUD_DEVICE");
+	char *area, *mem;
+	int fd, status;
+	pid_t pid;
+
+	ksft_print_header();
+	if (!path)
+		ksft_exit_skip("no devdax device given\n");
+	if (!dax_dev_ok(path))
+		ksft_exit_skip("%s: needs 1 GiB alignment and 2 GiB\n", path);
+
+	fd = open(path, O_RDONLY);
+	if (fd < 0)
+		ksft_exit_skip("%s: %s\n", path, strerror(errno));
+	area = mmap(NULL, 3 * SZ_1G, PROT_NONE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+	if (area == MAP_FAILED)
+		ksft_exit_fail_perror("mmap");
+	mem = (char *)(((unsigned long)area + SZ_1G - 1) & ~(SZ_1G - 1));
+	mem = mmap(mem, 2 * SZ_1G, PROT_READ, MAP_SHARED | MAP_FIXED, fd, 0);
+	if (mem == MAP_FAILED)
+		ksft_exit_skip("mmap %s: %s\n", path, strerror(errno));
+
+	/* A PUD in the first GiB, a guard page at the end of the second. */
+	signal(SIGBUS, sigbus_handler);
+	if (sigsetjmp(sigbus_jmp, 1))
+		ksft_exit_skip("%s cannot be mapped with a PUD\n", path);
+	FORCE_READ(mem[0]);
+	signal(SIGBUS, SIG_DFL);
+	if (madvise(mem + 2 * SZ_1G - getpagesize(), getpagesize(),
+		    MADV_GUARD_INSTALL))
+		ksft_exit_skip("MADV_GUARD_INSTALL: %s\n", strerror(errno));
+
+	ksft_set_plan(1);
+	fflush(stdout);
+	pid = fork();
+	if (pid < 0)
+		ksft_exit_fail_perror("fork");
+	if (!pid)
+		_exit(rss_file_kb() < (long)(SZ_1G >> 10) ? 10 : 0);
+	if (waitpid(pid, &status, 0) != pid)
+		ksft_exit_fail_perror("waitpid");
+
+	if (!WIFEXITED(status) || WEXITSTATUS(status))
+		ksft_test_result_fail("child does not account the copied PUD\n");
+	else
+		ksft_test_result_pass("devdax PUD copied by fork\n");
+
+	munmap(area, 3 * SZ_1G);
+	close(fd);
+	ksft_finished();
+}
diff --git a/tools/testing/selftests/mm/run_vmtests.sh b/tools/testing/selftests/mm/run_vmtests.sh
index d09f9f6a384ee..7294e1334d1da 100755
--- a/tools/testing/selftests/mm/run_vmtests.sh
+++ b/tools/testing/selftests/mm/run_vmtests.sh
@@ -445,6 +445,7 @@ fi
 CATEGORY="thp" run_test ./transhuge-stress -d 20
 
 CATEGORY="thp" run_test ./folio_split_race_test
+CATEGORY="thp" run_test ./fork_dax_pud
 
 CATEGORY="migration" run_test ./migration
 

-- 
2.43.0

Re: [PATCH 1/2] mm/huge_memory: take a reference on a folio PUD copied by fork

[David Hildenbrand (Arm)]

On 10/21/26 20:10, Konstantin Khorenko wrote:
> Device DAX maps its memory with vmf_insert_folio_pud(), which takes a
> folio reference, adds a PUD rmap and accounts the pages to the mm.
> zap_huge_pud() drops all three, but copy_huge_pud() copies the PUD into
> the child without taking any of them. Each child's unmap then drops
> what the parent took:
> 
> - the folio is a reference short, so with two children its refcount
>   drops to zero and free_zone_device_folio() runs while the parent
>   still maps it;
> - the child's file counter goes negative, "Bad rss-counter state" when
>   its mm is freed;
> - the parent's unmap drives the entire mapcount below zero.
> 
> fork copies the page tables of a device DAX mapping that has a guard
> region: commit f807123d578d ("mm: allow guard regions in file-backed and
> read-only mappings") allowed MADV_GUARD_INSTALL there, and it makes fork
> copy the VMA.
> 
> Take the reference, the rmap and the counter for a PUD that
> vm_normal_page_pud() resolves to a page, as copy_huge_pmd() does for a
> PMD, with a new folio_dup_file_rmap_pud() helper.
> 
> Fixes: aed877c2b425 ("device/dax: properly refcount device dax pages when mapping")
> Cc: stable@vger.kernel.org
> Assisted-by: LLM
> Signed-off-by: Konstantin Khorenko <khorenko@virtuozzo.com>
> ---
>  include/linux/huge_mm.h |  2 +-
>  include/linux/rmap.h    | 22 ++++++++++++++++++++++
>  mm/huge_memory.c        | 13 +++++++++++--
>  mm/memory.c             |  4 ++--
>  4 files changed, 36 insertions(+), 5 deletions(-)
> 
> diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
> index c745f7ad22987..c5094f9d95ba9 100644
> --- a/include/linux/huge_mm.h
> +++ b/include/linux/huge_mm.h
> @@ -14,7 +14,7 @@ int copy_huge_pmd(struct mm_struct *dst_mm, struct mm_struct *src_mm,
>  bool huge_pmd_set_accessed(struct vm_fault *vmf);
>  int copy_huge_pud(struct mm_struct *dst_mm, struct mm_struct *src_mm,
>  		  pud_t *dst_pud, pud_t *src_pud, unsigned long addr,
> -		  struct vm_area_struct *vma);
> +		  struct vm_area_struct *dst_vma, struct vm_area_struct *src_vma);
>  
>  #ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
>  void huge_pud_set_accessed(struct vm_fault *vmf, pud_t orig_pud);
> diff --git a/include/linux/rmap.h b/include/linux/rmap.h
> index 0b332770abeed..3c9ed8e27f0e4 100644
> --- a/include/linux/rmap.h
> +++ b/include/linux/rmap.h
> @@ -564,6 +564,28 @@ static inline void folio_dup_file_rmap_pmd(struct folio *folio,
>  #endif
>  }
>  
> +/**
> + * folio_dup_file_rmap_pud - duplicate a PUD mapping of a page range of a folio
> + * @folio:	The folio to duplicate the mapping of
> + * @page:	The first page to duplicate the mapping of
> + * @dst_vma:	The destination vm area
> + *
> + * The page range of the folio is defined by [page, page + HPAGE_PUD_NR)
> + *
> + * The caller needs to hold the page table lock.
> + */
> +static inline void folio_dup_file_rmap_pud(struct folio *folio,
> +		struct page *page, struct vm_area_struct *dst_vma)
> +{
> +#if defined(CONFIG_TRANSPARENT_HUGEPAGE) && \
> +	defined(CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD)
> +	__folio_dup_file_rmap(folio, page, HPAGE_PUD_NR, dst_vma,
> +			      PGTABLE_LEVEL_PUD);
> +#else
> +	WARN_ON_ONCE(true);
> +#endif
> +}
> +
>  static __always_inline int __folio_try_dup_anon_rmap(struct folio *folio,
>  		struct page *page, int nr_pages, struct vm_area_struct *dst_vma,
>  		struct vm_area_struct *src_vma, enum pgtable_level level)
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index 1e5d68acf62a5..7e8b41f293a49 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -2075,9 +2075,10 @@ void touch_pud(struct vm_area_struct *vma, unsigned long addr,
>  
>  int copy_huge_pud(struct mm_struct *dst_mm, struct mm_struct *src_mm,
>  		  pud_t *dst_pud, pud_t *src_pud, unsigned long addr,
> -		  struct vm_area_struct *vma)
> +		  struct vm_area_struct *dst_vma, struct vm_area_struct *src_vma)
>  {
>  	spinlock_t *dst_ptl, *src_ptl;
> +	struct page *page;
>  	pud_t pud;
>  	int ret;
>  
> @@ -2094,7 +2095,15 @@ int copy_huge_pud(struct mm_struct *dst_mm, struct mm_struct *src_mm,
>  	 * TODO: once we support anonymous pages, use
>  	 * folio_try_dup_anon_rmap_*() and split if duplicating fails.
>  	 */
> -	if (vma_is_cow_mapping(vma) && pud_write(pud)) {
> +	page = vm_normal_page_pud(src_vma, addr, pud);
> +	if (page) {
> +		struct folio *folio = page_folio(page);
> +
> +		folio_get(folio);
> +		folio_dup_file_rmap_pud(folio, page, dst_vma);
> +		add_mm_counter(dst_mm, mm_counter_file(folio), HPAGE_PUD_NR);
> +	}
> +	if (vma_is_cow_mapping(src_vma) && pud_write(pud)) {
>  		pudp_set_wrprotect(src_mm, addr, src_pud);
>  		pud = pud_wrprotect(pud);
>  	}
> diff --git a/mm/memory.c b/mm/memory.c
> index 8b0c2c735d3de..8acd4d5bb2310 100644
> --- a/mm/memory.c
> +++ b/mm/memory.c
> @@ -1514,8 +1514,8 @@ copy_pud_range(struct vm_area_struct *dst_vma, struct vm_area_struct *src_vma,
>  			int err;
>  
>  			VM_BUG_ON_VMA(next-addr != HPAGE_PUD_SIZE, src_vma);
> -			err = copy_huge_pud(dst_mm, src_mm,
> -					    dst_pud, src_pud, addr, src_vma);
> +			err = copy_huge_pud(dst_mm, src_mm, dst_pud, src_pud,
> +					    addr, dst_vma, src_vma);
>  			if (err == -ENOMEM)
>  				return -ENOMEM;
>  			if (!err)
> 

I have the following fix lying around here.

Introducing folio_dup_file_rmap_pud() is better. But please see how I move the comment and
how passing mm+vma to the helper is not really reasonable.


Subject: [PATCH 01/36] mm/huge_memory: properly handle refcounted PUD mappings
 on fork

copy_huge_pud() copies device-DAX entries without taking a folio reference
or updating mapcounts and the child's RSS.

For example, MADV_GUARD_INSTALL sets VM_MAYBE_GUARD on a device-DAX VMA,
forcing copy_page_range() to traverse the whole VMA on fork. Any remaining
huge PUDs are then copied without accounting for the new mappings.
Unmapping them in the child decrements accounting that was never added.

Use vm_normal_page_pud() to identify refcounted mappings. Take a folio
reference and add the child's PUD rmap and RSS accounting before installing
the copied entry. Pass src and dst VMAs consistently instead of src and
dst MM. Move the TODO.

Fixes: aed877c2b4257 ("device/dax: properly refcount device dax pages when mapping")
Cc: stable@vger.kernel.org
Signed-off-by: David Hildenbrand (Arm) <david@kernel.org>
---
 include/linux/huge_mm.h |  5 ++---
 mm/huge_memory.c        | 27 +++++++++++++++++++--------
 mm/memory.c             |  4 +---
 3 files changed, 22 insertions(+), 14 deletions(-)

diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
index c745f7ad22987..fac05120ab774 100644
--- a/include/linux/huge_mm.h
+++ b/include/linux/huge_mm.h
@@ -12,9 +12,8 @@ int copy_huge_pmd(struct mm_struct *dst_mm, struct mm_struct *src_mm,
 		  pmd_t *dst_pmd, pmd_t *src_pmd, unsigned long addr,
 		  struct vm_area_struct *dst_vma, struct vm_area_struct *src_vma);
 bool huge_pmd_set_accessed(struct vm_fault *vmf);
-int copy_huge_pud(struct mm_struct *dst_mm, struct mm_struct *src_mm,
-		  pud_t *dst_pud, pud_t *src_pud, unsigned long addr,
-		  struct vm_area_struct *vma);
+int copy_huge_pud(struct vm_area_struct *dst_vma, struct vm_area_struct *src_vma,
+		pud_t *dst_pud, pud_t *src_pud, unsigned long addr);
 
 #ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
 void huge_pud_set_accessed(struct vm_fault *vmf, pud_t orig_pud);
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index 766ea4d458bc2..8cfaad8168948 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -2078,11 +2078,13 @@ void touch_pud(struct vm_area_struct *vma, unsigned long addr,
 		update_mmu_cache_pud(vma, addr, pud);
 }
 
-int copy_huge_pud(struct mm_struct *dst_mm, struct mm_struct *src_mm,
-		  pud_t *dst_pud, pud_t *src_pud, unsigned long addr,
-		  struct vm_area_struct *vma)
+int copy_huge_pud(struct vm_area_struct *dst_vma, struct vm_area_struct *src_vma,
+		pud_t *dst_pud, pud_t *src_pud, unsigned long addr)
 {
+	struct mm_struct *dst_mm = dst_vma->vm_mm;
+	struct mm_struct *src_mm = src_vma->vm_mm;
 	spinlock_t *dst_ptl, *src_ptl;
+	struct page *page;
 	pud_t pud;
 	int ret;
 
@@ -2095,11 +2097,20 @@ int copy_huge_pud(struct mm_struct *dst_mm, struct mm_struct *src_mm,
 	if (unlikely(!pud_trans_huge(pud)))
 		goto out_unlock;
 
-	/*
-	 * TODO: once we support anonymous pages, use
-	 * folio_try_dup_anon_rmap_*() and split if duplicating fails.
-	 */
-	if (vma_is_cow_mapping(vma) && pud_write(pud)) {
+	page = vm_normal_page_pud(src_vma, addr, pud);
+	if (page) {
+		struct folio *folio = page_folio(page);
+
+		folio_get(folio);
+		/*
+		 * TODO: once we support anonymous pages, use
+		 * folio_try_dup_anon_rmap_*() and split if duplicating fails.
+		 */
+		folio_add_file_rmap_pud(folio, page, dst_vma);
+		add_mm_counter(dst_mm, mm_counter_file(folio), HPAGE_PUD_NR);
+	}
+
+	if (vma_is_cow_mapping(src_vma) && pud_write(pud)) {
 		pudp_set_wrprotect(src_mm, addr, src_pud);
 		pud = pud_wrprotect(pud);
 	}
diff --git a/mm/memory.c b/mm/memory.c
index 797d39572a09d..8a673b411c197 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -1463,7 +1463,6 @@ copy_pud_range(struct vm_area_struct *dst_vma, struct vm_area_struct *src_vma,
 	       unsigned long end)
 {
 	struct mm_struct *dst_mm = dst_vma->vm_mm;
-	struct mm_struct *src_mm = src_vma->vm_mm;
 	pud_t *src_pud, *dst_pud;
 	unsigned long next;
 
@@ -1477,8 +1476,7 @@ copy_pud_range(struct vm_area_struct *dst_vma, struct vm_area_struct *src_vma,
 			int err;
 
 			VM_BUG_ON_VMA(next-addr != HPAGE_PUD_SIZE, src_vma);
-			err = copy_huge_pud(dst_mm, src_mm,
-					    dst_pud, src_pud, addr, src_vma);
+			err = copy_huge_pud(dst_vma, src_vma, dst_pud, src_pud, addr);
 			if (err == -ENOMEM)
 				return -ENOMEM;
 			if (!err)
-- 
2.43.0


-- 
Cheers,

David