selftests: prctl: introduce tests for disabling THPs completely

The test will set the global system THP setting to never, madvise or
always depending on the fixture variant and the 2M setting to inherit
before it starts (and reset to original at teardown). The fixture setup
will also test if PR_SET_THP_DISABLE prctl call can be made to disable all
THPs and skip if it fails.

This tests if the process can:
- successfully get the policy to disable THPs completely.
- never get a hugepage when the THPs are completely disabled
with the prctl, including with MADV_HUGE and MADV_COLLAPSE.
- successfully reset the policy of the process.
- after reset, only get hugepages with:
- MADV_COLLAPSE when policy is set to never.
- MADV_HUGE and MADV_COLLAPSE when policy is set to madvise.
- always when policy is set to "always".
- never get a THP with MADV_NOHUGEPAGE.
- repeat the above tests in a forked process to make sure
the policy is carried across forks.

[usamaarif642@gmail.com: return after executing test in child process]
Link: https://lkml.kernel.org/r/2d0ea708-ecba-4021-b6ca-e93f1413d60a@gmail.com
[usamaarif642@gmail.com: include linux/mman.h for prctl_thp_disable]
Link: https://lkml.kernel.org/r/20250910204609.1720498-1-usamaarif642@gmail.com
Link: https://lore.kernel.org/all/c8249725-e91d-4c51-b9bb-40305e61e20d@sirena.org.uk/
Link: https://lkml.kernel.org/r/20250815135549.130506-7-usamaarif642@gmail.com
Signed-off-by: Usama Arif <usamaarif642@gmail.com>
Acked-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Barry Song <baohua@kernel.org>
Cc: Dev Jain <dev.jain@arm.com>
Cc: Jann Horn <jannh@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Mariano Pache <npache@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: SeongJae Park <sj@kernel.org>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yafang <laoar.shao@gmail.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

authored by

Usama Arif and committed by

Andrew Morton 9 months ago 681f45de 49850bd0

+189 -1

5 changed files

expand all

tools

testing

selftests

.gitignore

Makefile

prctl_thp_disable.c

thp_settings.c

thp_settings.h

tools/testing/selftests/mm/.gitignore

··· 58 58 pkey_sighandler_tests_64 59 59 guard-regions 60 60 merge 61 + prctl_thp_disable

tools/testing/selftests/mm/Makefile

··· 86 86 TEST_GEN_FILES += pagemap_ioctl 87 87 TEST_GEN_FILES += pfnmap 88 88 TEST_GEN_FILES += process_madv 89 + TEST_GEN_FILES += prctl_thp_disable 89 90 TEST_GEN_FILES += thuge-gen 90 91 TEST_GEN_FILES += transhuge-stress 91 92 TEST_GEN_FILES += uffd-stress

+178

tools/testing/selftests/mm/prctl_thp_disable.c

··· 1 + // SPDX-License-Identifier: GPL-2.0 2 + /* 3 + * Basic tests for PR_GET/SET_THP_DISABLE prctl calls 4 + * 5 + * Author(s): Usama Arif <usamaarif642@gmail.com> 6 + */ 7 + #include <stdio.h> 8 + #include <stdlib.h> 9 + #include <string.h> 10 + #include <unistd.h> 11 + #include <sys/mman.h> 12 + #include <linux/mman.h> 13 + #include <sys/prctl.h> 14 + #include <sys/wait.h> 15 + 16 + #include "../kselftest_harness.h" 17 + #include "thp_settings.h" 18 + #include "vm_util.h" 19 + 20 + enum thp_collapse_type { 21 + THP_COLLAPSE_NONE, 22 + THP_COLLAPSE_MADV_NOHUGEPAGE, 23 + THP_COLLAPSE_MADV_HUGEPAGE, /* MADV_HUGEPAGE before access */ 24 + THP_COLLAPSE_MADV_COLLAPSE, /* MADV_COLLAPSE after access */ 25 + }; 26 + 27 + /* 28 + * Function to mmap a buffer, fault it in, madvise it appropriately (before 29 + * page fault for MADV_HUGE, and after for MADV_COLLAPSE), and check if the 30 + * mmap region is huge. 31 + * Returns: 32 + * 0 if test doesn't give hugepage 33 + * 1 if test gives a hugepage 34 + * -errno if mmap fails 35 + */ 36 + static int test_mmap_thp(enum thp_collapse_type madvise_buf, size_t pmdsize) 37 + { 38 + char *mem, *mmap_mem; 39 + size_t mmap_size; 40 + int ret; 41 + 42 + /* For alignment purposes, we need twice the THP size. */ 43 + mmap_size = 2 * pmdsize; 44 + mmap_mem = (char *)mmap(NULL, mmap_size, PROT_READ | PROT_WRITE, 45 + MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); 46 + if (mmap_mem == MAP_FAILED) 47 + return -errno; 48 + 49 + /* We need a THP-aligned memory area. */ 50 + mem = (char *)(((uintptr_t)mmap_mem + pmdsize) & ~(pmdsize - 1)); 51 + 52 + if (madvise_buf == THP_COLLAPSE_MADV_HUGEPAGE) 53 + madvise(mem, pmdsize, MADV_HUGEPAGE); 54 + else if (madvise_buf == THP_COLLAPSE_MADV_NOHUGEPAGE) 55 + madvise(mem, pmdsize, MADV_NOHUGEPAGE); 56 + 57 + /* Ensure memory is allocated */ 58 + memset(mem, 1, pmdsize); 59 + 60 + if (madvise_buf == THP_COLLAPSE_MADV_COLLAPSE) 61 + madvise(mem, pmdsize, MADV_COLLAPSE); 62 + 63 + /* HACK: make sure we have a separate VMA that we can check reliably. */ 64 + mprotect(mem, pmdsize, PROT_READ); 65 + 66 + ret = check_huge_anon(mem, 1, pmdsize); 67 + munmap(mmap_mem, mmap_size); 68 + return ret; 69 + } 70 + 71 + static void prctl_thp_disable_completely_test(struct __test_metadata *const _metadata, 72 + size_t pmdsize, 73 + enum thp_enabled thp_policy) 74 + { 75 + ASSERT_EQ(prctl(PR_GET_THP_DISABLE, NULL, NULL, NULL, NULL), 1); 76 + 77 + /* tests after prctl overrides global policy */ 78 + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_NONE, pmdsize), 0); 79 + 80 + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_MADV_NOHUGEPAGE, pmdsize), 0); 81 + 82 + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_MADV_HUGEPAGE, pmdsize), 0); 83 + 84 + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_MADV_COLLAPSE, pmdsize), 0); 85 + 86 + /* Reset to global policy */ 87 + ASSERT_EQ(prctl(PR_SET_THP_DISABLE, 0, NULL, NULL, NULL), 0); 88 + 89 + /* tests after prctl is cleared, and only global policy is effective */ 90 + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_NONE, pmdsize), 91 + thp_policy == THP_ALWAYS ? 1 : 0); 92 + 93 + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_MADV_NOHUGEPAGE, pmdsize), 0); 94 + 95 + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_MADV_HUGEPAGE, pmdsize), 96 + thp_policy == THP_NEVER ? 0 : 1); 97 + 98 + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_MADV_COLLAPSE, pmdsize), 1); 99 + } 100 + 101 + FIXTURE(prctl_thp_disable_completely) 102 + { 103 + struct thp_settings settings; 104 + size_t pmdsize; 105 + }; 106 + 107 + FIXTURE_VARIANT(prctl_thp_disable_completely) 108 + { 109 + enum thp_enabled thp_policy; 110 + }; 111 + 112 + FIXTURE_VARIANT_ADD(prctl_thp_disable_completely, never) 113 + { 114 + .thp_policy = THP_NEVER, 115 + }; 116 + 117 + FIXTURE_VARIANT_ADD(prctl_thp_disable_completely, madvise) 118 + { 119 + .thp_policy = THP_MADVISE, 120 + }; 121 + 122 + FIXTURE_VARIANT_ADD(prctl_thp_disable_completely, always) 123 + { 124 + .thp_policy = THP_ALWAYS, 125 + }; 126 + 127 + FIXTURE_SETUP(prctl_thp_disable_completely) 128 + { 129 + if (!thp_available()) 130 + SKIP(return, "Transparent Hugepages not available\n"); 131 + 132 + self->pmdsize = read_pmd_pagesize(); 133 + if (!self->pmdsize) 134 + SKIP(return, "Unable to read PMD size\n"); 135 + 136 + if (prctl(PR_SET_THP_DISABLE, 1, NULL, NULL, NULL)) 137 + SKIP(return, "Unable to disable THPs completely for the process\n"); 138 + 139 + thp_save_settings(); 140 + thp_read_settings(&self->settings); 141 + self->settings.thp_enabled = variant->thp_policy; 142 + self->settings.hugepages[sz2ord(self->pmdsize, getpagesize())].enabled = THP_INHERIT; 143 + thp_write_settings(&self->settings); 144 + } 145 + 146 + FIXTURE_TEARDOWN(prctl_thp_disable_completely) 147 + { 148 + thp_restore_settings(); 149 + } 150 + 151 + TEST_F(prctl_thp_disable_completely, nofork) 152 + { 153 + prctl_thp_disable_completely_test(_metadata, self->pmdsize, variant->thp_policy); 154 + } 155 + 156 + TEST_F(prctl_thp_disable_completely, fork) 157 + { 158 + int ret = 0; 159 + pid_t pid; 160 + 161 + /* Make sure prctl changes are carried across fork */ 162 + pid = fork(); 163 + ASSERT_GE(pid, 0); 164 + 165 + if (!pid) { 166 + prctl_thp_disable_completely_test(_metadata, self->pmdsize, variant->thp_policy); 167 + return; 168 + } 169 + 170 + wait(&ret); 171 + if (WIFEXITED(ret)) 172 + ret = WEXITSTATUS(ret); 173 + else 174 + ret = -EINVAL; 175 + ASSERT_EQ(ret, 0); 176 + } 177 + 178 + TEST_HARNESS_MAIN

+8 -1

tools/testing/selftests/mm/thp_settings.c

··· 382 382 return __thp_supported_orders(true); 383 383 } 384 384 385 - bool thp_is_enabled(void) 385 + bool thp_available(void) 386 386 { 387 387 if (access(THP_SYSFS, F_OK) != 0) 388 + return false; 389 + return true; 390 + } 391 + 392 + bool thp_is_enabled(void) 393 + { 394 + if (!thp_available()) 388 395 return false; 389 396 390 397 int mode = thp_read_string("enabled", thp_enabled_strings);

tools/testing/selftests/mm/thp_settings.h

··· 84 84 unsigned long thp_supported_orders(void); 85 85 unsigned long thp_shmem_supported_orders(void); 86 86 87 + bool thp_available(void); 87 88 bool thp_is_enabled(void); 88 89 89 90 #endif /* __THP_SETTINGS_H__ */

Configure Feed

Configure Feed