[PATCH v1 RESEND 0/6] mm: (pte|pmd)_mkdirty() should not unconditionally allow for write access
This is the follow-up on [1], adding selftests (testing for known issues we added workarounds for and other issues that haven't been fixed yet), fixing sparc64, reverting the workarounds, and perform one cleanup.
The patch from [1] was modified slightly (updated/extended patch description, dropped one unnecessary NOP instruction from the ASM in __pte_mkhwwrite()).
Retested on x86_64 and sparc64 (sun4u in QEMU).
I scanned most architectures to make sure their (pte|pmd)_mkdirty() handling is correct. To be sure, we can run the selftests and find out if other architectures are still affectes (loongarch was fixed recently as well).
Based on master for now. I don't expect surprises regarding mm-tress, but I can rebase if there are any problems.
[1] https://lkml.kernel.org/r/20221212130213.136267-1-david@redhat.com
Cc: Andrew Morton akpm@linux-foundation.org Cc: "David S. Miller" davem@davemloft.net Cc: Peter Xu peterx@redhat.com Cc: Hugh Dickins hughd@google.com Cc: Shuah Khan shuah@kernel.org Cc: Sam Ravnborg sam@ravnborg.org Cc: Yu Zhao yuzhao@google.com Cc: Anshuman Khandual anshuman.khandual@arm.com
David Hildenbrand (6): selftests/mm: reuse read_pmd_pagesize() in COW selftest selftests/mm: mkdirty: test behavior of (pte|pmd)_mkdirty on VMAs without write permissions sparc/mm: don't unconditionally set HW writable bit when setting PTE dirty on 64bit mm/migrate: revert "mm/migrate: fix wrongly apply write bit after mkdirty on sparc64" mm/huge_memory: revert "Partly revert "mm/thp: carry over dirty bit when thp splits on pmd"" mm/huge_memory: conditionally call maybe_mkwrite() and drop pte_wrprotect() in __split_huge_pmd_locked()
arch/sparc/include/asm/pgtable_64.h | 116 +++--- mm/huge_memory.c | 16 +- mm/migrate.c | 2 - tools/testing/selftests/mm/Makefile | 2 + tools/testing/selftests/mm/cow.c | 33 +- tools/testing/selftests/mm/khugepaged.c | 4 + tools/testing/selftests/mm/mkdirty.c | 379 ++++++++++++++++++ tools/testing/selftests/mm/soft-dirty.c | 3 + .../selftests/mm/split_huge_page_test.c | 4 + tools/testing/selftests/mm/vm_util.c | 4 +- 10 files changed, 468 insertions(+), 95 deletions(-) create mode 100644 tools/testing/selftests/mm/mkdirty.c
The COW selftest can deal with THP not being configured. So move error handling of read_pmd_pagesize() into the callers such that we can reuse it in the COW selftest.
Signed-off-by: David Hildenbrand david@redhat.com --- tools/testing/selftests/mm/cow.c | 33 +++---------------- tools/testing/selftests/mm/khugepaged.c | 4 +++ tools/testing/selftests/mm/soft-dirty.c | 3 ++ .../selftests/mm/split_huge_page_test.c | 4 +++ tools/testing/selftests/mm/vm_util.c | 4 +-- 5 files changed, 17 insertions(+), 31 deletions(-)
diff --git a/tools/testing/selftests/mm/cow.c b/tools/testing/selftests/mm/cow.c index 0eb2e8180aa5..dc9d6fe86028 100644 --- a/tools/testing/selftests/mm/cow.c +++ b/tools/testing/selftests/mm/cow.c @@ -45,34 +45,6 @@ static size_t hugetlbsizes[10]; static int gup_fd; static bool has_huge_zeropage;
-static void detect_thpsize(void) -{ - int fd = open("/sys/kernel/mm/transparent_hugepage/hpage_pmd_size", - O_RDONLY); - size_t size = 0; - char buf[15]; - int ret; - - if (fd < 0) - return; - - ret = pread(fd, buf, sizeof(buf), 0); - if (ret > 0 && ret < sizeof(buf)) { - buf[ret] = 0; - - size = strtoul(buf, NULL, 10); - if (size < pagesize) - size = 0; - if (size > 0) { - thpsize = size; - ksft_print_msg("[INFO] detected THP size: %zu KiB\n", - thpsize / 1024); - } - } - - close(fd); -} - static void detect_huge_zeropage(void) { int fd = open("/sys/kernel/mm/transparent_hugepage/use_zero_page", @@ -1741,7 +1713,10 @@ int main(int argc, char **argv) int err;
pagesize = getpagesize(); - detect_thpsize(); + thpsize = read_pmd_pagesize(); + if (thpsize) + ksft_print_msg("[INFO] detected THP size: %zu KiB\n", + thpsize / 1024); detect_hugetlbsizes(); detect_huge_zeropage();
diff --git a/tools/testing/selftests/mm/khugepaged.c b/tools/testing/selftests/mm/khugepaged.c index 64126c8cd561..97adc0f34f9c 100644 --- a/tools/testing/selftests/mm/khugepaged.c +++ b/tools/testing/selftests/mm/khugepaged.c @@ -1476,6 +1476,10 @@ int main(int argc, const char **argv)
page_size = getpagesize(); hpage_pmd_size = read_pmd_pagesize(); + if (!hpage_pmd_size) { + printf("Reading PMD pagesize failed"); + exit(EXIT_FAILURE); + } hpage_pmd_nr = hpage_pmd_size / page_size;
default_settings.khugepaged.max_ptes_none = hpage_pmd_nr - 1; diff --git a/tools/testing/selftests/mm/soft-dirty.c b/tools/testing/selftests/mm/soft-dirty.c index 21d8830c5f24..cc5f144430d4 100644 --- a/tools/testing/selftests/mm/soft-dirty.c +++ b/tools/testing/selftests/mm/soft-dirty.c @@ -80,6 +80,9 @@ static void test_hugepage(int pagemap_fd, int pagesize) int i, ret; size_t hpage_len = read_pmd_pagesize();
+ if (!hpage_len) + ksft_exit_fail_msg("Reading PMD pagesize failed"); + map = memalign(hpage_len, hpage_len); if (!map) ksft_exit_fail_msg("memalign failed\n"); diff --git a/tools/testing/selftests/mm/split_huge_page_test.c b/tools/testing/selftests/mm/split_huge_page_test.c index 76e1c36dd9e5..1dc5804b8b2b 100644 --- a/tools/testing/selftests/mm/split_huge_page_test.c +++ b/tools/testing/selftests/mm/split_huge_page_test.c @@ -300,6 +300,10 @@ int main(int argc, char **argv) pagesize = getpagesize(); pageshift = ffs(pagesize) - 1; pmd_pagesize = read_pmd_pagesize(); + if (!pmd_pagesize) { + printf("Reading PMD pagesize failed\n"); + exit(EXIT_FAILURE); + }
split_pmd_thp(); split_pte_mapped_thp(); diff --git a/tools/testing/selftests/mm/vm_util.c b/tools/testing/selftests/mm/vm_util.c index 40e795624ff3..8dc74dd022c2 100644 --- a/tools/testing/selftests/mm/vm_util.c +++ b/tools/testing/selftests/mm/vm_util.c @@ -84,12 +84,12 @@ uint64_t read_pmd_pagesize(void)
fd = open(PMD_SIZE_FILE_PATH, O_RDONLY); if (fd == -1) - ksft_exit_fail_msg("Open hpage_pmd_size failed\n"); + return 0;
num_read = read(fd, buf, 19); if (num_read < 1) { close(fd); - ksft_exit_fail_msg("Read hpage_pmd_size failed\n"); + return 0; } buf[num_read] = '\0'; close(fd);
Let's add some tests that trigger (pte|pmd)_mkdirty on VMAs without write permissions. If an architecture implementation is wrong, we might accidentally set the PTE/PMD writable and allow for write access in a VMA without write permissions.
The tests include reproducers for the two issues recently discovered and worked-around in core-MM for now:
(1) commit 624a2c94f5b7 ("Partly revert "mm/thp: carry over dirty bit when thp splits on pmd"") (2) commit 96a9c287e25d ("mm/migrate: fix wrongly apply write bit after mkdirty on sparc64")
In addition, some other tests that reveal further issues.
All tests pass under x86_64: ./mkdirty # [INFO] detected THP size: 2048 KiB TAP version 13 1..6 # [INFO] PTRACE write access ok 1 SIGSEGV generated, page not modified # [INFO] PTRACE write access to THP ok 2 SIGSEGV generated, page not modified # [INFO] Page migration ok 3 SIGSEGV generated, page not modified # [INFO] Page migration of THP ok 4 SIGSEGV generated, page not modified # [INFO] PTE-mapping a THP ok 5 SIGSEGV generated, page not modified # [INFO] UFFDIO_COPY ok 6 SIGSEGV generated, page not modified # Totals: pass:6 fail:0 xfail:0 xpass:0 skip:0 error:0
But some fail on sparc64: ./mkdirty # [INFO] detected THP size: 8192 KiB TAP version 13 1..6 # [INFO] PTRACE write access not ok 1 SIGSEGV generated, page not modified # [INFO] PTRACE write access to THP not ok 2 SIGSEGV generated, page not modified # [INFO] Page migration ok 3 SIGSEGV generated, page not modified # [INFO] Page migration of THP ok 4 SIGSEGV generated, page not modified # [INFO] PTE-mapping a THP ok 5 SIGSEGV generated, page not modified # [INFO] UFFDIO_COPY not ok 6 SIGSEGV generated, page not modified Bail out! 3 out of 6 tests failed # Totals: pass:3 fail:3 xfail:0 xpass:0 skip:0 error:0
Reverting both above commits makes all tests fail on sparc64: ./mkdirty # [INFO] detected THP size: 8192 KiB TAP version 13 1..6 # [INFO] PTRACE write access not ok 1 SIGSEGV generated, page not modified # [INFO] PTRACE write access to THP not ok 2 SIGSEGV generated, page not modified # [INFO] Page migration not ok 3 SIGSEGV generated, page not modified # [INFO] Page migration of THP not ok 4 SIGSEGV generated, page not modified # [INFO] PTE-mapping a THP not ok 5 SIGSEGV generated, page not modified # [INFO] UFFDIO_COPY not ok 6 SIGSEGV generated, page not modified Bail out! 6 out of 6 tests failed # Totals: pass:0 fail:6 xfail:0 xpass:0 skip:0 error:0
The tests are useful to detect other problematic archs, to verify new arch fixes, and to stop such issues from reappearing in the future.
For now, we don't add any hugetlb tests.
Signed-off-by: David Hildenbrand david@redhat.com --- tools/testing/selftests/mm/Makefile | 2 + tools/testing/selftests/mm/mkdirty.c | 379 +++++++++++++++++++++++++++ 2 files changed, 381 insertions(+) create mode 100644 tools/testing/selftests/mm/mkdirty.c
diff --git a/tools/testing/selftests/mm/Makefile b/tools/testing/selftests/mm/Makefile index c31d952cff68..8235dddbbbc6 100644 --- a/tools/testing/selftests/mm/Makefile +++ b/tools/testing/selftests/mm/Makefile @@ -47,6 +47,7 @@ TEST_GEN_FILES += map_hugetlb TEST_GEN_FILES += map_populate TEST_GEN_FILES += memfd_secret TEST_GEN_FILES += migration +TEST_GEN_PROGS += mkdirty TEST_GEN_FILES += mlock-random-test TEST_GEN_FILES += mlock2-tests TEST_GEN_FILES += mrelease_test @@ -108,6 +109,7 @@ $(OUTPUT)/cow: vm_util.c $(OUTPUT)/khugepaged: vm_util.c $(OUTPUT)/ksm_functional_tests: vm_util.c $(OUTPUT)/madv_populate: vm_util.c +$(OUTPUT)/mkdirty: vm_util.c $(OUTPUT)/soft-dirty: vm_util.c $(OUTPUT)/split_huge_page_test: vm_util.c $(OUTPUT)/userfaultfd: vm_util.c diff --git a/tools/testing/selftests/mm/mkdirty.c b/tools/testing/selftests/mm/mkdirty.c new file mode 100644 index 000000000000..6d71d972997b --- /dev/null +++ b/tools/testing/selftests/mm/mkdirty.c @@ -0,0 +1,379 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Test handling of code that might set PTE/PMD dirty in read-only VMAs. + * Setting a PTE/PMD dirty must not accidentally set the PTE/PMD writable. + * + * Copyright 2023, Red Hat, Inc. + * + * Author(s): David Hildenbrand david@redhat.com + */ +#include <fcntl.h> +#include <signal.h> +#include <unistd.h> +#include <string.h> +#include <errno.h> +#include <stdlib.h> +#include <stdbool.h> +#include <stdint.h> +#include <sys/mman.h> +#include <setjmp.h> +#include <sys/syscall.h> +#include <sys/ioctl.h> +#include <linux/userfaultfd.h> +#include <linux/mempolicy.h> + +#include "../kselftest.h" +#include "vm_util.h" + +static size_t pagesize; +static size_t thpsize; +static int mem_fd; +static int pagemap_fd; +static sigjmp_buf env; + +static void signal_handler(int sig) +{ + if (sig == SIGSEGV) + siglongjmp(env, 1); + siglongjmp(env, 2); +} + +static void do_test_write_sigsegv(char *mem) +{ + char orig = *mem; + int ret; + + if (signal(SIGSEGV, signal_handler) == SIG_ERR) { + ksft_test_result_fail("signal() failed\n"); + return; + } + + ret = sigsetjmp(env, 1); + if (!ret) + *mem = orig + 1; + + if (signal(SIGSEGV, SIG_DFL) == SIG_ERR) + ksft_test_result_fail("signal() failed\n"); + + ksft_test_result(ret == 1 && *mem == orig, + "SIGSEGV generated, page not modified\n"); +} + +static char *mmap_thp_range(int prot, char **_mmap_mem, size_t *_mmap_size) +{ + const size_t mmap_size = 2 * thpsize; + char *mem, *mmap_mem; + + mmap_mem = mmap(NULL, mmap_size, prot, MAP_PRIVATE|MAP_ANON, + -1, 0); + if (mmap_mem == MAP_FAILED) { + ksft_test_result_fail("mmap() failed\n"); + return MAP_FAILED; + } + mem = (char *)(((uintptr_t)mmap_mem + thpsize) & ~(thpsize - 1)); + + if (madvise(mem, thpsize, MADV_HUGEPAGE)) { + ksft_test_result_skip("MADV_HUGEPAGE failed\n"); + munmap(mmap_mem, mmap_size); + return MAP_FAILED; + } + + *_mmap_mem = mmap_mem; + *_mmap_size = mmap_size; + return mem; +} + +static void test_ptrace_write(void) +{ + char data = 1; + char *mem; + int ret; + + ksft_print_msg("[INFO] PTRACE write access\n"); + + mem = mmap(NULL, pagesize, PROT_READ, MAP_PRIVATE|MAP_ANON, -1, 0); + if (mem == MAP_FAILED) { + ksft_test_result_fail("mmap() failed\n"); + return; + } + + /* Fault in the shared zeropage. */ + if (*mem != 0) { + ksft_test_result_fail("Memory not zero\n"); + goto munmap; + } + + /* + * Unshare the page (populating a fresh anon page that might be set + * dirty in the PTE) in the read-only VMA using ptrace (FOLL_FORCE). + */ + lseek(mem_fd, (uintptr_t) mem, SEEK_SET); + ret = write(mem_fd, &data, 1); + if (ret != 1 || *mem != data) { + ksft_test_result_fail("write() failed\n"); + goto munmap; + } + + do_test_write_sigsegv(mem); +munmap: + munmap(mem, pagesize); +} + +static void test_ptrace_write_thp(void) +{ + char *mem, *mmap_mem; + size_t mmap_size; + char data = 1; + int ret; + + ksft_print_msg("[INFO] PTRACE write access to THP\n"); + + mem = mmap_thp_range(PROT_READ, &mmap_mem, &mmap_size); + if (mem == MAP_FAILED) + return; + + /* + * Write to the first subpage in the read-only VMA using + * ptrace(FOLL_FORCE), eventually placing a fresh THP that is marked + * dirty in the PMD. + */ + lseek(mem_fd, (uintptr_t) mem, SEEK_SET); + ret = write(mem_fd, &data, 1); + if (ret != 1 || *mem != data) { + ksft_test_result_fail("write() failed\n"); + goto munmap; + } + + /* MM populated a THP if we got the last subpage populated as well. */ + if (!pagemap_is_populated(pagemap_fd, mem + thpsize - pagesize)) { + ksft_test_result_skip("Did not get a THP populated\n"); + goto munmap; + } + + do_test_write_sigsegv(mem); +munmap: + munmap(mmap_mem, mmap_size); +} + +static void test_page_migration(void) +{ + char *mem; + + ksft_print_msg("[INFO] Page migration\n"); + + mem = mmap(NULL, pagesize, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANON, + -1, 0); + if (mem == MAP_FAILED) { + ksft_test_result_fail("mmap() failed\n"); + return; + } + + /* Populate a fresh page and dirty it. */ + memset(mem, 1, pagesize); + if (mprotect(mem, pagesize, PROT_READ)) { + ksft_test_result_fail("mprotect() failed\n"); + goto munmap; + } + + /* Trigger page migration. Might not be available or fail. */ + if (syscall(__NR_mbind, mem, pagesize, MPOL_LOCAL, NULL, 0x7fful, + MPOL_MF_MOVE)) { + ksft_test_result_skip("mbind() failed\n"); + goto munmap; + } + + do_test_write_sigsegv(mem); +munmap: + munmap(mem, pagesize); +} + +static void test_page_migration_thp(void) +{ + char *mem, *mmap_mem; + size_t mmap_size; + + ksft_print_msg("[INFO] Page migration of THP\n"); + + mem = mmap_thp_range(PROT_READ|PROT_WRITE, &mmap_mem, &mmap_size); + if (mem == MAP_FAILED) + return; + + /* + * Write to the first page, which might populate a fresh anon THP + * and dirty it. + */ + memset(mem, 1, pagesize); + if (mprotect(mem, thpsize, PROT_READ)) { + ksft_test_result_fail("mprotect() failed\n"); + goto munmap; + } + + /* MM populated a THP if we got the last subpage populated as well. */ + if (!pagemap_is_populated(pagemap_fd, mem + thpsize - pagesize)) { + ksft_test_result_skip("Did not get a THP populated\n"); + goto munmap; + } + + /* Trigger page migration. Might not be available or fail. */ + if (syscall(__NR_mbind, mem, thpsize, MPOL_LOCAL, NULL, 0x7fful, + MPOL_MF_MOVE)) { + ksft_test_result_skip("mbind() failed\n"); + goto munmap; + } + + do_test_write_sigsegv(mem); +munmap: + munmap(mmap_mem, mmap_size); +} + +static void test_pte_mapped_thp(void) +{ + char *mem, *mmap_mem; + size_t mmap_size; + + ksft_print_msg("[INFO] PTE-mapping a THP\n"); + + mem = mmap_thp_range(PROT_READ|PROT_WRITE, &mmap_mem, &mmap_size); + if (mem == MAP_FAILED) + return; + + /* + * Write to the first page, which might populate a fresh anon THP + * and dirty it. + */ + memset(mem, 1, pagesize); + if (mprotect(mem, thpsize, PROT_READ)) { + ksft_test_result_fail("mprotect() failed\n"); + goto munmap; + } + + /* MM populated a THP if we got the last subpage populated as well. */ + if (!pagemap_is_populated(pagemap_fd, mem + thpsize - pagesize)) { + ksft_test_result_skip("Did not get a THP populated\n"); + goto munmap; + } + + /* Trigger PTE-mapping the THP by mprotect'ing the last subpage. */ + if (mprotect(mem + thpsize - pagesize, pagesize, + PROT_READ|PROT_WRITE)) { + ksft_test_result_fail("mprotect() failed\n"); + goto munmap; + } + + do_test_write_sigsegv(mem); +munmap: + munmap(mmap_mem, mmap_size); +} + +#ifdef __NR_userfaultfd +static void test_uffdio_copy(void) +{ + struct uffdio_register uffdio_register; + struct uffdio_copy uffdio_copy; + struct uffdio_api uffdio_api; + char *dst, *src; + int uffd; + + ksft_print_msg("[INFO] UFFDIO_COPY\n"); + + src = malloc(pagesize); + memset(src, 1, pagesize); + dst = mmap(NULL, pagesize, PROT_READ, MAP_PRIVATE|MAP_ANON, -1, 0); + if (dst == MAP_FAILED) { + ksft_test_result_fail("mmap() failed\n"); + return; + } + + uffd = syscall(__NR_userfaultfd, O_CLOEXEC | O_NONBLOCK); + if (uffd < 0) { + ksft_test_result_skip("__NR_userfaultfd failed\n"); + goto munmap; + } + + uffdio_api.api = UFFD_API; + uffdio_api.features = 0; + if (ioctl(uffd, UFFDIO_API, &uffdio_api) < 0) { + ksft_test_result_fail("UFFDIO_API failed\n"); + goto close_uffd; + } + + uffdio_register.range.start = (unsigned long) dst; + uffdio_register.range.len = pagesize; + uffdio_register.mode = UFFDIO_REGISTER_MODE_MISSING; + if (ioctl(uffd, UFFDIO_REGISTER, &uffdio_register)) { + ksft_test_result_fail("UFFDIO_REGISTER failed\n"); + goto close_uffd; + } + + /* Place a page in a read-only VMA, which might set the PTE dirty. */ + uffdio_copy.dst = (unsigned long) dst; + uffdio_copy.src = (unsigned long) src; + uffdio_copy.len = pagesize; + uffdio_copy.mode = 0; + if (ioctl(uffd, UFFDIO_COPY, &uffdio_copy)) { + ksft_test_result_fail("UFFDIO_COPY failed\n"); + goto close_uffd; + } + + do_test_write_sigsegv(dst); +close_uffd: + close(uffd); +munmap: + munmap(dst, pagesize); + free(src); +#endif /* __NR_userfaultfd */ +} + +int main(void) +{ + int err, tests = 2; + + pagesize = getpagesize(); + thpsize = read_pmd_pagesize(); + if (thpsize) { + ksft_print_msg("[INFO] detected THP size: %zu KiB\n", + thpsize / 1024); + tests += 3; + } +#ifdef __NR_userfaultfd + tests += 1; +#endif /* __NR_userfaultfd */ + + ksft_print_header(); + ksft_set_plan(tests); + + mem_fd = open("/proc/self/mem", O_RDWR); + if (mem_fd < 0) + ksft_exit_fail_msg("opening /proc/self/mem failed\n"); + pagemap_fd = open("/proc/self/pagemap", O_RDONLY); + if (pagemap_fd < 0) + ksft_exit_fail_msg("opening /proc/self/pagemap failed\n"); + + /* + * On some ptrace(FOLL_FORCE) write access via /proc/self/mem in + * read-only VMAs, the kernel may set the PTE/PMD dirty. + */ + test_ptrace_write(); + if (thpsize) + test_ptrace_write_thp(); + /* + * On page migration, the kernel may set the PTE/PMD dirty when + * remapping the page. + */ + test_page_migration(); + if (thpsize) + test_page_migration_thp(); + /* PTE-mapping a THP might propagate the dirty PMD bit to the PTEs. */ + if (thpsize) + test_pte_mapped_thp(); + /* Placing a fresh page via userfaultfd may set the PTE dirty. */ +#ifdef __NR_userfaultfd + test_uffdio_copy(); +#endif /* __NR_userfaultfd */ + + err = ksft_get_fail_cnt(); + if (err) + ksft_exit_fail_msg("%d out of %d tests failed\n", + err, ksft_test_num()); + return ksft_exit_pass(); +}
On sparc64, there is no HW modified bit, therefore, SW tracks via a SW bit if the PTE is dirty via pte_mkdirty(). However, pte_mkdirty() currently also unconditionally sets the HW writable bit, which is wrong.
pte_mkdirty() is not supposed to make a PTE actually writable, unless the SW writable bit -- pte_write() -- indicates that the PTE is not write-protected. Fortunately, sparc64 also defines a SW writable bit.
For example, this already turned into a problem in the context of THP splitting as documented in commit 624a2c94f5b7 ("Partly revert "mm/thp: carry over dirty bit when thp splits on pmd""), and for page migration, as documented in commit 96a9c287e25d ("mm/migrate: fix wrongly apply write bit after mkdirty on sparc64").
Also, we might want to use the dirty PTE bit in the context of KSM with shared zeropage [1], whereby setting the page writable would be problematic.
But more general, any code that might end up setting a PTE/PMD dirty inside a VM without write permissions is possibly broken,
Before this commit (sun4u in QEMU): root@debian:~/linux/tools/testing/selftests/mm# ./mkdirty # [INFO] detected THP size: 8192 KiB TAP version 13 1..6 # [INFO] PTRACE write access not ok 1 SIGSEGV generated, page not modified # [INFO] PTRACE write access to THP not ok 2 SIGSEGV generated, page not modified # [INFO] Page migration ok 3 SIGSEGV generated, page not modified # [INFO] Page migration of THP ok 4 SIGSEGV generated, page not modified # [INFO] PTE-mapping a THP ok 5 SIGSEGV generated, page not modified # [INFO] UFFDIO_COPY not ok 6 SIGSEGV generated, page not modified Bail out! 3 out of 6 tests failed # Totals: pass:3 fail:3 xfail:0 xpass:0 skip:0 error:0
Test #3,#4,#5 pass ever since we added some MM workarounds, the underlying issue remains.
Let's fix the remaining issues and prepare for reverting the workarounds by setting the HW writable bit only if both, the SW dirty bit and the SW writable bit are set.
We have to move pte_dirty() and pte_dirty() up. The code patching mechanism and handling constants > 22bit is a bit special on sparc64.
The ASM logic in pte_mkdirty() and pte_mkwrite() match the logic in pte_mkold() to create the mask depending on the machine type. The ASM logic in __pte_mkhwwrite() matches the logic in pte_present(), just using an "or" instead of an "and" instruction.
With this commit (sun4u in QEMU): root@debian:~/linux/tools/testing/selftests/mm# ./mkdirty # [INFO] detected THP size: 8192 KiB TAP version 13 1..6 # [INFO] PTRACE write access ok 1 SIGSEGV generated, page not modified # [INFO] PTRACE write access to THP ok 2 SIGSEGV generated, page not modified # [INFO] Page migration ok 3 SIGSEGV generated, page not modified # [INFO] Page migration of THP ok 4 SIGSEGV generated, page not modified # [INFO] PTE-mapping a THP ok 5 SIGSEGV generated, page not modified # [INFO] UFFDIO_COPY ok 6 SIGSEGV generated, page not modified # Totals: pass:6 fail:0 xfail:0 xpass:0 skip:0 error:0
This handling seems to have been in place forever.
[1] https://lkml.kernel.org/r/533a7c3d-3a48-b16b-b421-6e8386e0b142@redhat.com
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: David Hildenbrand david@redhat.com --- arch/sparc/include/asm/pgtable_64.h | 116 ++++++++++++++++------------ 1 file changed, 66 insertions(+), 50 deletions(-)
diff --git a/arch/sparc/include/asm/pgtable_64.h b/arch/sparc/include/asm/pgtable_64.h index 2dc8d4641734..5563efa1a19f 100644 --- a/arch/sparc/include/asm/pgtable_64.h +++ b/arch/sparc/include/asm/pgtable_64.h @@ -357,6 +357,42 @@ static inline pgprot_t pgprot_noncached(pgprot_t prot) */ #define pgprot_noncached pgprot_noncached
+static inline unsigned long pte_dirty(pte_t pte) +{ + unsigned long mask; + + __asm__ __volatile__( + "\n661: mov %1, %0\n" + " nop\n" + " .section .sun4v_2insn_patch, "ax"\n" + " .word 661b\n" + " sethi %%uhi(%2), %0\n" + " sllx %0, 32, %0\n" + " .previous\n" + : "=r" (mask) + : "i" (_PAGE_MODIFIED_4U), "i" (_PAGE_MODIFIED_4V)); + + return (pte_val(pte) & mask); +} + +static inline unsigned long pte_write(pte_t pte) +{ + unsigned long mask; + + __asm__ __volatile__( + "\n661: mov %1, %0\n" + " nop\n" + " .section .sun4v_2insn_patch, "ax"\n" + " .word 661b\n" + " sethi %%uhi(%2), %0\n" + " sllx %0, 32, %0\n" + " .previous\n" + : "=r" (mask) + : "i" (_PAGE_WRITE_4U), "i" (_PAGE_WRITE_4V)); + + return (pte_val(pte) & mask); +} + #if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE) pte_t arch_make_huge_pte(pte_t entry, unsigned int shift, vm_flags_t flags); #define arch_make_huge_pte arch_make_huge_pte @@ -418,28 +454,43 @@ static inline bool is_hugetlb_pte(pte_t pte) } #endif
+static inline pte_t __pte_mkhwwrite(pte_t pte) +{ + unsigned long val = pte_val(pte); + + /* + * Note: we only want to set the HW writable bit if the SW writable bit + * and the SW dirty bit are set. + */ + __asm__ __volatile__( + "\n661: or %0, %2, %0\n" + " .section .sun4v_1insn_patch, "ax"\n" + " .word 661b\n" + " or %0, %3, %0\n" + " .previous\n" + : "=r" (val) + : "0" (val), "i" (_PAGE_W_4U), "i" (_PAGE_W_4V)); + + return __pte(val); +} + static inline pte_t pte_mkdirty(pte_t pte) { - unsigned long val = pte_val(pte), tmp; + unsigned long val = pte_val(pte), mask;
__asm__ __volatile__( - "\n661: or %0, %3, %0\n" - " nop\n" - "\n662: nop\n" + "\n661: mov %1, %0\n" " nop\n" " .section .sun4v_2insn_patch, "ax"\n" " .word 661b\n" - " sethi %%uhi(%4), %1\n" - " sllx %1, 32, %1\n" - " .word 662b\n" - " or %1, %%lo(%4), %1\n" - " or %0, %1, %0\n" + " sethi %%uhi(%2), %0\n" + " sllx %0, 32, %0\n" " .previous\n" - : "=r" (val), "=r" (tmp) - : "0" (val), "i" (_PAGE_MODIFIED_4U | _PAGE_W_4U), - "i" (_PAGE_MODIFIED_4V | _PAGE_W_4V)); + : "=r" (mask) + : "i" (_PAGE_MODIFIED_4U), "i" (_PAGE_MODIFIED_4V));
- return __pte(val); + pte = __pte(val | mask); + return pte_write(pte) ? __pte_mkhwwrite(pte) : pte; }
static inline pte_t pte_mkclean(pte_t pte) @@ -481,7 +532,8 @@ static inline pte_t pte_mkwrite(pte_t pte) : "=r" (mask) : "i" (_PAGE_WRITE_4U), "i" (_PAGE_WRITE_4V));
- return __pte(val | mask); + pte = __pte(val | mask); + return pte_dirty(pte) ? __pte_mkhwwrite(pte) : pte; }
static inline pte_t pte_wrprotect(pte_t pte) @@ -584,42 +636,6 @@ static inline unsigned long pte_young(pte_t pte) return (pte_val(pte) & mask); }
-static inline unsigned long pte_dirty(pte_t pte) -{ - unsigned long mask; - - __asm__ __volatile__( - "\n661: mov %1, %0\n" - " nop\n" - " .section .sun4v_2insn_patch, "ax"\n" - " .word 661b\n" - " sethi %%uhi(%2), %0\n" - " sllx %0, 32, %0\n" - " .previous\n" - : "=r" (mask) - : "i" (_PAGE_MODIFIED_4U), "i" (_PAGE_MODIFIED_4V)); - - return (pte_val(pte) & mask); -} - -static inline unsigned long pte_write(pte_t pte) -{ - unsigned long mask; - - __asm__ __volatile__( - "\n661: mov %1, %0\n" - " nop\n" - " .section .sun4v_2insn_patch, "ax"\n" - " .word 661b\n" - " sethi %%uhi(%2), %0\n" - " sllx %0, 32, %0\n" - " .previous\n" - : "=r" (mask) - : "i" (_PAGE_WRITE_4U), "i" (_PAGE_WRITE_4V)); - - return (pte_val(pte) & mask); -} - static inline unsigned long pte_exec(pte_t pte) { unsigned long mask;
Hi David.
On Tue, Apr 11, 2023 at 04:25:09PM +0200, David Hildenbrand wrote:
On sparc64, there is no HW modified bit, therefore, SW tracks via a SW bit if the PTE is dirty via pte_mkdirty(). However, pte_mkdirty() currently also unconditionally sets the HW writable bit, which is wrong.
pte_mkdirty() is not supposed to make a PTE actually writable, unless the SW writable bit -- pte_write() -- indicates that the PTE is not write-protected. Fortunately, sparc64 also defines a SW writable bit.
For example, this already turned into a problem in the context of THP splitting as documented in commit 624a2c94f5b7 ("Partly revert "mm/thp: carry over dirty bit when thp splits on pmd""), and for page migration, as documented in commit 96a9c287e25d ("mm/migrate: fix wrongly apply write bit after mkdirty on sparc64").
Also, we might want to use the dirty PTE bit in the context of KSM with shared zeropage [1], whereby setting the page writable would be problematic.
But more general, any code that might end up setting a PTE/PMD dirty inside a VM without write permissions is possibly broken,
Before this commit (sun4u in QEMU): root@debian:~/linux/tools/testing/selftests/mm# ./mkdirty # [INFO] detected THP size: 8192 KiB TAP version 13 1..6 # [INFO] PTRACE write access not ok 1 SIGSEGV generated, page not modified # [INFO] PTRACE write access to THP not ok 2 SIGSEGV generated, page not modified # [INFO] Page migration ok 3 SIGSEGV generated, page not modified # [INFO] Page migration of THP ok 4 SIGSEGV generated, page not modified # [INFO] PTE-mapping a THP ok 5 SIGSEGV generated, page not modified # [INFO] UFFDIO_COPY not ok 6 SIGSEGV generated, page not modified Bail out! 3 out of 6 tests failed # Totals: pass:3 fail:3 xfail:0 xpass:0 skip:0 error:0
Test #3,#4,#5 pass ever since we added some MM workarounds, the underlying issue remains.
Let's fix the remaining issues and prepare for reverting the workarounds by setting the HW writable bit only if both, the SW dirty bit and the SW writable bit are set.
We have to move pte_dirty() and pte_dirty() up. The code patching
One of the pte_dirty() should be replaced with pte_write().
It would have been nice to separate moving and changes in two patches, but keeping it together works too.
mechanism and handling constants > 22bit is a bit special on sparc64.
The ASM logic in pte_mkdirty() and pte_mkwrite() match the logic in pte_mkold() to create the mask depending on the machine type. The ASM logic in __pte_mkhwwrite() matches the logic in pte_present(), just using an "or" instead of an "and" instruction.
With this commit (sun4u in QEMU): root@debian:~/linux/tools/testing/selftests/mm# ./mkdirty # [INFO] detected THP size: 8192 KiB TAP version 13 1..6 # [INFO] PTRACE write access ok 1 SIGSEGV generated, page not modified # [INFO] PTRACE write access to THP ok 2 SIGSEGV generated, page not modified # [INFO] Page migration ok 3 SIGSEGV generated, page not modified # [INFO] Page migration of THP ok 4 SIGSEGV generated, page not modified # [INFO] PTE-mapping a THP ok 5 SIGSEGV generated, page not modified # [INFO] UFFDIO_COPY ok 6 SIGSEGV generated, page not modified # Totals: pass:6 fail:0 xfail:0 xpass:0 skip:0 error:0
Nice!
This handling seems to have been in place forever.
[1] https://lkml.kernel.org/r/533a7c3d-3a48-b16b-b421-6e8386e0b142@redhat.com
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: David Hildenbrand david@redhat.com
I tried to follow your changes, but my knowledge of gcc assembler failed me. But based on the nice and detailed change log and the code I managed to understand: Acked-by: Sam Ravnborg sam@ravnborg.org
arch/sparc/include/asm/pgtable_64.h | 116 ++++++++++++++++------------ 1 file changed, 66 insertions(+), 50 deletions(-)
diff --git a/arch/sparc/include/asm/pgtable_64.h b/arch/sparc/include/asm/pgtable_64.h index 2dc8d4641734..5563efa1a19f 100644 --- a/arch/sparc/include/asm/pgtable_64.h +++ b/arch/sparc/include/asm/pgtable_64.h @@ -357,6 +357,42 @@ static inline pgprot_t pgprot_noncached(pgprot_t prot) */ #define pgprot_noncached pgprot_noncached +static inline unsigned long pte_dirty(pte_t pte) +{
- unsigned long mask;
- __asm__ __volatile__(
- "\n661: mov %1, %0\n"
- " nop\n"
- " .section .sun4v_2insn_patch, "ax"\n"
- " .word 661b\n"
- " sethi %%uhi(%2), %0\n"
- " sllx %0, 32, %0\n"
- " .previous\n"
- : "=r" (mask)
- : "i" (_PAGE_MODIFIED_4U), "i" (_PAGE_MODIFIED_4V));
- return (pte_val(pte) & mask);
+}
+static inline unsigned long pte_write(pte_t pte) +{
- unsigned long mask;
- __asm__ __volatile__(
- "\n661: mov %1, %0\n"
- " nop\n"
- " .section .sun4v_2insn_patch, "ax"\n"
- " .word 661b\n"
- " sethi %%uhi(%2), %0\n"
- " sllx %0, 32, %0\n"
- " .previous\n"
- : "=r" (mask)
- : "i" (_PAGE_WRITE_4U), "i" (_PAGE_WRITE_4V));
- return (pte_val(pte) & mask);
+}
#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE) pte_t arch_make_huge_pte(pte_t entry, unsigned int shift, vm_flags_t flags); #define arch_make_huge_pte arch_make_huge_pte @@ -418,28 +454,43 @@ static inline bool is_hugetlb_pte(pte_t pte) } #endif +static inline pte_t __pte_mkhwwrite(pte_t pte) +{
- unsigned long val = pte_val(pte);
- /*
* Note: we only want to set the HW writable bit if the SW writable bit* and the SW dirty bit are set.*/- __asm__ __volatile__(
- "\n661: or %0, %2, %0\n"
- " .section .sun4v_1insn_patch, "ax"\n"
- " .word 661b\n"
- " or %0, %3, %0\n"
- " .previous\n"
- : "=r" (val)
- : "0" (val), "i" (_PAGE_W_4U), "i" (_PAGE_W_4V));
- return __pte(val);
+}
static inline pte_t pte_mkdirty(pte_t pte) {
- unsigned long val = pte_val(pte), tmp;
- unsigned long val = pte_val(pte), mask;
__asm__ __volatile__(
- "\n661: or %0, %3, %0\n"
- " nop\n"
- "\n662: nop\n"
- "\n661: mov %1, %0\n" " nop\n" " .section .sun4v_2insn_patch, "ax"\n" " .word 661b\n"
- " sethi %%uhi(%4), %1\n"
- " sllx %1, 32, %1\n"
- " .word 662b\n"
- " or %1, %%lo(%4), %1\n"
- " or %0, %1, %0\n"
- " sethi %%uhi(%2), %0\n"
- " sllx %0, 32, %0\n" " .previous\n"
- : "=r" (val), "=r" (tmp)
- : "0" (val), "i" (_PAGE_MODIFIED_4U | _PAGE_W_4U),
"i" (_PAGE_MODIFIED_4V | _PAGE_W_4V));
- : "=r" (mask)
- : "i" (_PAGE_MODIFIED_4U), "i" (_PAGE_MODIFIED_4V));
- return __pte(val);
- pte = __pte(val | mask);
- return pte_write(pte) ? __pte_mkhwwrite(pte) : pte;
} static inline pte_t pte_mkclean(pte_t pte) @@ -481,7 +532,8 @@ static inline pte_t pte_mkwrite(pte_t pte) : "=r" (mask) : "i" (_PAGE_WRITE_4U), "i" (_PAGE_WRITE_4V));
- return __pte(val | mask);
- pte = __pte(val | mask);
- return pte_dirty(pte) ? __pte_mkhwwrite(pte) : pte;
} static inline pte_t pte_wrprotect(pte_t pte) @@ -584,42 +636,6 @@ static inline unsigned long pte_young(pte_t pte) return (pte_val(pte) & mask); } -static inline unsigned long pte_dirty(pte_t pte) -{
- unsigned long mask;
- __asm__ __volatile__(
- "\n661: mov %1, %0\n"
- " nop\n"
- " .section .sun4v_2insn_patch, "ax"\n"
- " .word 661b\n"
- " sethi %%uhi(%2), %0\n"
- " sllx %0, 32, %0\n"
- " .previous\n"
- : "=r" (mask)
- : "i" (_PAGE_MODIFIED_4U), "i" (_PAGE_MODIFIED_4V));
- return (pte_val(pte) & mask);
-}
-static inline unsigned long pte_write(pte_t pte) -{
- unsigned long mask;
- __asm__ __volatile__(
- "\n661: mov %1, %0\n"
- " nop\n"
- " .section .sun4v_2insn_patch, "ax"\n"
- " .word 661b\n"
- " sethi %%uhi(%2), %0\n"
- " sllx %0, 32, %0\n"
- " .previous\n"
- : "=r" (mask)
- : "i" (_PAGE_WRITE_4U), "i" (_PAGE_WRITE_4V));
- return (pte_val(pte) & mask);
-}
static inline unsigned long pte_exec(pte_t pte) { unsigned long mask; -- 2.39.2
[...]
Let's fix the remaining issues and prepare for reverting the workarounds by setting the HW writable bit only if both, the SW dirty bit and the SW writable bit are set.
We have to move pte_dirty() and pte_dirty() up. The code patching
One of the pte_dirty() should be replaced with pte_write().
Indeed, thanks. I assume Andrew can change the latter to pte_write(). [unless I have to resend, of course]
It would have been nice to separate moving and changes in two patches, but keeping it together works too.
I prefer to not have "move code within the same file around" in separate patches as long as it doesn't add too much noise. Here, I think it's acceptable.
mechanism and handling constants > 22bit is a bit special on sparc64.
The ASM logic in pte_mkdirty() and pte_mkwrite() match the logic in pte_mkold() to create the mask depending on the machine type. The ASM logic in __pte_mkhwwrite() matches the logic in pte_present(), just using an "or" instead of an "and" instruction.
With this commit (sun4u in QEMU): root@debian:~/linux/tools/testing/selftests/mm# ./mkdirty # [INFO] detected THP size: 8192 KiB TAP version 13 1..6 # [INFO] PTRACE write access ok 1 SIGSEGV generated, page not modified # [INFO] PTRACE write access to THP ok 2 SIGSEGV generated, page not modified # [INFO] Page migration ok 3 SIGSEGV generated, page not modified # [INFO] Page migration of THP ok 4 SIGSEGV generated, page not modified # [INFO] PTE-mapping a THP ok 5 SIGSEGV generated, page not modified # [INFO] UFFDIO_COPY ok 6 SIGSEGV generated, page not modified # Totals: pass:6 fail:0 xfail:0 xpass:0 skip:0 error:0
Nice!
This handling seems to have been in place forever.
[1] https://lkml.kernel.org/r/533a7c3d-3a48-b16b-b421-6e8386e0b142@redhat.com
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: David Hildenbrand david@redhat.com
I tried to follow your changes, but my knowledge of gcc assembler failed me. But based on the nice and detailed change log and the code I managed to understand: Acked-by: Sam Ravnborg sam@ravnborg.org
Thanks Sam!
On Wed, 12 Apr 2023 11:48:15 +0200 David Hildenbrand david@redhat.com wrote:
We have to move pte_dirty() and pte_dirty() up. The code patching
One of the pte_dirty() should be replaced with pte_write().
Indeed, thanks. I assume Andrew can change the latter to pte_write().
It was a struggle, but I managed to do this ;)
This reverts commit 96a9c287e25d ("mm/migrate: fix wrongly apply write bit after mkdirty on sparc64").
Now that sparc64 mkdirty handling is fixed and no longer sets a PTE/PMD writable that shouldn't be writable, let's revert the temporary fix.
The mkdirty mm selftest still passes with this change on sparc64.
Note that loongarch handling was fixed in commit bf2f34a506e6 ("LoongArch: Set _PAGE_DIRTY only if _PAGE_WRITE is set in {pmd,pte}_mkdirty()").
Signed-off-by: David Hildenbrand david@redhat.com --- mm/huge_memory.c | 6 ++---- mm/migrate.c | 2 -- 2 files changed, 2 insertions(+), 6 deletions(-)
diff --git a/mm/huge_memory.c b/mm/huge_memory.c index 032fb0ef9cd1..ec86bf1d4e81 100644 --- a/mm/huge_memory.c +++ b/mm/huge_memory.c @@ -3276,6 +3276,8 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new) pmde = mk_huge_pmd(new, READ_ONCE(vma->vm_page_prot)); if (pmd_swp_soft_dirty(*pvmw->pmd)) pmde = pmd_mksoft_dirty(pmde); + if (is_writable_migration_entry(entry)) + pmde = maybe_pmd_mkwrite(pmde, vma); if (pmd_swp_uffd_wp(*pvmw->pmd)) pmde = pmd_mkuffd_wp(pmde); if (!is_migration_entry_young(entry)) @@ -3283,10 +3285,6 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new) /* NOTE: this may contain setting soft-dirty on some archs */ if (PageDirty(new) && is_migration_entry_dirty(entry)) pmde = pmd_mkdirty(pmde); - if (is_writable_migration_entry(entry)) - pmde = maybe_pmd_mkwrite(pmde, vma); - else - pmde = pmd_wrprotect(pmde);
if (PageAnon(new)) { rmap_t rmap_flags = RMAP_COMPOUND; diff --git a/mm/migrate.c b/mm/migrate.c index db3f154446af..3b21c44e2176 100644 --- a/mm/migrate.c +++ b/mm/migrate.c @@ -225,8 +225,6 @@ static bool remove_migration_pte(struct folio *folio, pte = maybe_mkwrite(pte, vma); else if (pte_swp_uffd_wp(*pvmw.pte)) pte = pte_mkuffd_wp(pte); - else - pte = pte_wrprotect(pte);
if (folio_test_anon(folio) && !is_readable_migration_entry(entry)) rmap_flags |= RMAP_EXCLUSIVE;
This reverts commit 624a2c94f5b7 ("Partly revert "mm/thp: carry over dirty bit when thp splits on pmd"") and the fixup in commit e833bc503405 ("mm/thp: re-apply mkdirty for small pages after split").
Now that sparc64 mkdirty handling is fixed and no longer sets a PTE/PMD writable that shouldn't be writable, let's revert the temporary fix and remove the stale comment.
The mkdirty mm selftest still passes with this change on sparc64.
Note that loongarch handling was fixed in commit bf2f34a506e6 ("LoongArch: Set _PAGE_DIRTY only if _PAGE_WRITE is set in {pmd,pte}_mkdirty()")
Signed-off-by: David Hildenbrand david@redhat.com --- mm/huge_memory.c | 9 ++------- 1 file changed, 2 insertions(+), 7 deletions(-)
diff --git a/mm/huge_memory.c b/mm/huge_memory.c index ec86bf1d4e81..6f3af65435c8 100644 --- a/mm/huge_memory.c +++ b/mm/huge_memory.c @@ -2238,18 +2238,13 @@ static void __split_huge_pmd_locked(struct vm_area_struct *vma, pmd_t *pmd, entry = maybe_mkwrite(entry, vma); if (anon_exclusive) SetPageAnonExclusive(page + i); + if (!write) + entry = pte_wrprotect(entry); if (!young) entry = pte_mkold(entry); /* NOTE: this may set soft-dirty too on some archs */ if (dirty) entry = pte_mkdirty(entry); - /* - * NOTE: this needs to happen after pte_mkdirty, - * because some archs (sparc64, loongarch) could - * set hw write bit when mkdirty. - */ - if (!write) - entry = pte_wrprotect(entry); if (soft_dirty) entry = pte_mksoft_dirty(entry); if (uffd_wp)
No need to call maybe_mkwrite() to then wrprotect if the source PMD was not writable.
It's worth nothing that this now allows for PTEs to be writable even if the source PMD was not writable: if vma->vm_page_prot includes write permissions.
As documented in commit 931298e103c2 ("mm/userfaultfd: rely on vma->vm_page_prot in uffd_wp_range()"), any mechanism that intends to have pages wrprotected (COW, writenotify, mprotect, uffd-wp, softdirty, ...) has to properly adjust vma->vm_page_prot upfront, to not include write permissions. If vma->vm_page_prot includes write permissions, the PTE/PMD can be writable as default.
This now mimics the handling in mm/migrate.c:remove_migration_pte() and in mm/huge_memory.c:remove_migration_pmd(), which has been in place for a long time (except that 96a9c287e25d ("mm/migrate: fix wrongly apply write bit after mkdirty on sparc64") temporarily changed it).
Signed-off-by: David Hildenbrand david@redhat.com --- mm/huge_memory.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)
diff --git a/mm/huge_memory.c b/mm/huge_memory.c index 6f3af65435c8..8332e16ac97b 100644 --- a/mm/huge_memory.c +++ b/mm/huge_memory.c @@ -2235,11 +2235,10 @@ static void __split_huge_pmd_locked(struct vm_area_struct *vma, pmd_t *pmd, entry = pte_swp_mkuffd_wp(entry); } else { entry = mk_pte(page + i, READ_ONCE(vma->vm_page_prot)); - entry = maybe_mkwrite(entry, vma); + if (write) + entry = maybe_mkwrite(entry, vma); if (anon_exclusive) SetPageAnonExclusive(page + i); - if (!write) - entry = pte_wrprotect(entry); if (!young) entry = pte_mkold(entry); /* NOTE: this may set soft-dirty too on some archs */
linux-kselftest-mirror@lists.linaro.org
-
Andrew Morton -
David Hildenbrand -
Sam Ravnborg