1============================= 2Examining Process Page Tables 3============================= 4 5pagemap is a new (as of 2.6.25) set of interfaces in the kernel that allow 6userspace programs to examine the page tables and related information by 7reading files in ``/proc``. 8 9There are four components to pagemap: 10 11 * ``/proc/pid/pagemap``. This file lets a userspace process find out which 12 physical frame each virtual page is mapped to. It contains one 64-bit 13 value for each virtual page, containing the following data (from 14 ``fs/proc/task_mmu.c``, above pagemap_read): 15 16 * Bits 0-54 page frame number (PFN) if present 17 * Bits 0-4 swap type if swapped 18 * Bits 5-54 swap offset if swapped 19 * Bit 55 pte is soft-dirty (see 20 Documentation/admin-guide/mm/soft-dirty.rst) 21 * Bit 56 page exclusively mapped (since 4.2) 22 * Bit 57 pte is tracked by userfaultfd (since 5.13) — in a 23 ``VM_UFFD_WP`` VMA this indicates a write-protected PTE; in a 24 ``VM_UFFD_RWP`` VMA it indicates an RWP-protected PTE. WP and 25 RWP are mutually exclusive per VMA, so the meaning is 26 unambiguous. See Documentation/admin-guide/mm/userfaultfd.rst. 27 * Bit 58 pte is a guard region (since 6.15) (see madvise (2) man page) 28 * Bits 59-60 zero 29 * Bit 61 page is file-page or shared-anon (since 3.5) 30 * Bit 62 page swapped 31 * Bit 63 page present 32 33 Since Linux 4.0 only users with the CAP_SYS_ADMIN capability can get PFNs. 34 In 4.0 and 4.1 opens by unprivileged fail with -EPERM. Starting from 35 4.2 the PFN field is zeroed if the user does not have CAP_SYS_ADMIN. 36 Reason: information about PFNs helps in exploiting Rowhammer vulnerability. 37 38 If the page is not present but in swap, then the PFN contains an 39 encoding of the swap file number and the page's offset into the 40 swap. Unmapped pages return a null PFN. This allows determining 41 precisely which pages are mapped (or in swap) and comparing mapped 42 pages between processes. 43 44 Traditionally, bit 56 indicates that a page is mapped exactly once and bit 45 56 is clear when a page is mapped multiple times, even when mapped in the 46 same process multiple times. In some kernel configurations, the semantics 47 for pages part of a larger allocation (e.g., THP) can differ: bit 56 is set 48 if all pages part of the corresponding large allocation are *certainly* 49 mapped in the same process, even if the page is mapped multiple times in that 50 process. Bit 56 is clear when any page page of the larger allocation 51 is *maybe* mapped in a different process. In some cases, a large allocation 52 might be treated as "maybe mapped by multiple processes" even though this 53 is no longer the case. 54 55 Efficient users of this interface will use ``/proc/pid/maps`` to 56 determine which areas of memory are actually mapped and llseek to 57 skip over unmapped regions. 58 59 * ``/proc/kpagecount``. This file contains a 64-bit count of the number of 60 times each page is mapped, indexed by PFN. Some kernel configurations do 61 not track the precise number of times a page part of a larger allocation 62 (e.g., THP) is mapped. In these configurations, the average number of 63 mappings per page in this larger allocation is returned instead. However, 64 if any page of the large allocation is mapped, the returned value will 65 be at least 1. 66 67The page-types tool in the tools/mm directory can be used to query the 68number of times a page is mapped. 69 70 * ``/proc/kpageflags``. This file contains a 64-bit set of flags for each 71 page, indexed by PFN. 72 73 The flags are (from ``include/uapi/linux/kernel-page-flags.h``): 74 75 0. LOCKED 76 1. ERROR 77 2. REFERENCED 78 3. UPTODATE 79 4. DIRTY 80 5. LRU 81 6. ACTIVE 82 7. SLAB 83 8. WRITEBACK 84 9. RECLAIM 85 10. BUDDY 86 11. MMAP 87 12. ANON 88 13. SWAPCACHE 89 14. SWAPBACKED 90 15. COMPOUND_HEAD 91 16. COMPOUND_TAIL 92 17. HUGE 93 18. UNEVICTABLE 94 19. HWPOISON 95 20. NOPAGE 96 21. KSM 97 22. THP 98 23. OFFLINE 99 24. ZERO_PAGE 100 25. IDLE 101 26. PGTABLE 102 103 * ``/proc/kpagecgroup``. This file contains a 64-bit inode number of the 104 memory cgroup each page is charged to, indexed by PFN. Only available when 105 CONFIG_MEMCG is set. 106 107Short descriptions to the page flags 108==================================== 109 1100 - LOCKED 111 The page is being locked for exclusive access, e.g. by undergoing read/write 112 IO. 1137 - SLAB 114 The page is managed by the SLAB/SLUB kernel memory allocator. 115 When compound page is used, either will only set this flag on the head 116 page. 11710 - BUDDY 118 A free memory block managed by the buddy system allocator. 119 The buddy system organizes free memory in blocks of various orders. 120 An order N block has 2^N physically contiguous pages, with the BUDDY flag 121 set for all pages. 122 Before 4.6 only the first page of the block had the flag set. 12315 - COMPOUND_HEAD 124 A compound page with order N consists of 2^N physically contiguous pages. 125 A compound page with order 2 takes the form of "HTTT", where H donates its 126 head page and T donates its tail page(s). The major consumers of compound 127 pages are hugeTLB pages (Documentation/admin-guide/mm/hugetlbpage.rst), 128 the SLUB etc. memory allocators and various device drivers. 129 However in this interface, only huge/giga pages are made visible 130 to end users. 13116 - COMPOUND_TAIL 132 A compound page tail (see description above). 13317 - HUGE 134 This is an integral part of a HugeTLB page. 13519 - HWPOISON 136 Hardware detected memory corruption on this page: don't touch the data! 13720 - NOPAGE 138 No page frame exists at the requested address. 13921 - KSM 140 Identical memory pages dynamically shared between one or more processes. 14122 - THP 142 Contiguous pages which construct THP of any size and mapped by any granularity. 14323 - OFFLINE 144 The page is logically offline. 14524 - ZERO_PAGE 146 Zero page for pfn_zero or huge_zero page. 14725 - IDLE 148 The page has not been accessed since it was marked idle (see 149 Documentation/admin-guide/mm/idle_page_tracking.rst). 150 Note that this flag may be stale in case the page was accessed via 151 a PTE. To make sure the flag is up-to-date one has to read 152 ``/sys/kernel/mm/page_idle/bitmap`` first. 15326 - PGTABLE 154 The page is in use as a page table. 155 156IO related page flags 157--------------------- 158 1591 - ERROR 160 IO error occurred. 1613 - UPTODATE 162 The page has up-to-date data. 163 ie. for file backed page: (in-memory data revision >= on-disk one) 1644 - DIRTY 165 The page has been written to, hence contains new data. 166 i.e. for file backed page: (in-memory data revision > on-disk one) 1678 - WRITEBACK 168 The page is being synced to disk. 169 170LRU related page flags 171---------------------- 172 1735 - LRU 174 The page is in one of the LRU lists. 1756 - ACTIVE 176 The page is in the active LRU list. 17718 - UNEVICTABLE 178 The page is in the unevictable (non-)LRU list It is somehow pinned and 179 not a candidate for LRU page reclaims, e.g. ramfs pages, 180 shmctl(SHM_LOCK) and mlock() memory segments. 1812 - REFERENCED 182 The page has been referenced since last LRU list enqueue/requeue. 1839 - RECLAIM 184 The page will be reclaimed soon after its pageout IO completed. 18511 - MMAP 186 A memory mapped page. 18712 - ANON 188 A memory mapped page that is not part of a file. 18913 - SWAPCACHE 190 The page is mapped to swap space, i.e. has an associated swap entry. 19114 - SWAPBACKED 192 The page is backed by swap/RAM. 193 194The page-types tool in the tools/mm directory can be used to query the 195above flags. 196 197Exceptions for Shared Memory 198============================ 199 200Page table entries for shared pages are cleared when the pages are zapped or 201swapped out. This makes swapped out pages indistinguishable from never-allocated 202ones. 203 204In kernel space, the swap location can still be retrieved from the page cache. 205However, values stored only on the normal PTE get lost irretrievably when the 206page is swapped out (i.e. SOFT_DIRTY). 207 208In user space, whether the page is present, swapped or none can be deduced with 209the help of lseek and/or mincore system calls. 210 211lseek() can differentiate between accessed pages (present or swapped out) and 212holes (none/non-allocated) by specifying the SEEK_DATA flag on the file where 213the pages are backed. For anonymous shared pages, the file can be found in 214``/proc/pid/map_files/``. 215 216mincore() can differentiate between pages in memory (present, including swap 217cache) and out of memory (swapped out or none/non-allocated). 218 219Other notes 220=========== 221 222Reading from any of the files will return -EINVAL if you are not starting 223the read on an 8-byte boundary (e.g., if you sought an odd number of bytes 224into the file), or if the size of the read is not a multiple of 8 bytes. 225 226Before Linux 3.11 pagemap bits 55-60 were used for "page-shift" (which is 227always 12 at most architectures). Since Linux 3.11 their meaning changes 228after first clear of soft-dirty bits. Since Linux 4.2 they are used for 229flags unconditionally. 230 231Pagemap Scan IOCTL 232================== 233 234The ``PAGEMAP_SCAN`` IOCTL on the pagemap file can be used to get or optionally 235clear the info about page table entries. The following operations are supported 236in this IOCTL: 237 238- Scan the address range and get the memory ranges matching the provided criteria. 239 This is performed when the output buffer is specified. 240- Write-protect the pages. The ``PM_SCAN_WP_MATCHING`` is used to write-protect 241 the pages of interest. The ``PM_SCAN_CHECK_WPASYNC`` aborts the operation if 242 non-Async Write Protected pages are found. The ``PM_SCAN_WP_MATCHING`` can be 243 used with or without ``PM_SCAN_CHECK_WPASYNC``. 244- Both of those operations can be combined into one atomic operation where we can 245 get and write protect the pages as well. 246 247Following flags about pages are currently supported: 248 249- ``PAGE_IS_WPALLOWED`` - Page has async-write-protection enabled 250- ``PAGE_IS_WRITTEN`` - Page in a ``UFFDIO_REGISTER_MODE_WP`` VMA has been 251 written to since it was write-protected. Only reported inside such VMAs. 252- ``PAGE_IS_FILE`` - Page is file backed 253- ``PAGE_IS_PRESENT`` - Page is present in the memory 254- ``PAGE_IS_SWAPPED`` - Page is in swapped 255- ``PAGE_IS_PFNZERO`` - Page has zero PFN 256- ``PAGE_IS_HUGE`` - Page is PMD-mapped THP or Hugetlb backed 257- ``PAGE_IS_SOFT_DIRTY`` - Page is soft-dirty 258- ``PAGE_IS_GUARD`` - Page is a part of a guard region 259- ``PAGE_IS_ACCESSED`` - Page in a ``UFFDIO_REGISTER_MODE_RWP`` VMA has been 260 accessed since RWP was applied. Only reported inside such VMAs. See 261 Documentation/admin-guide/mm/userfaultfd.rst for the RWP workflow. 262 263The ``struct pm_scan_arg`` is used as the argument of the IOCTL. 264 265 1. The size of the ``struct pm_scan_arg`` must be specified in the ``size`` 266 field. This field will be helpful in recognizing the structure if extensions 267 are done later. 268 2. The flags can be specified in the ``flags`` field. The ``PM_SCAN_WP_MATCHING`` 269 and ``PM_SCAN_CHECK_WPASYNC`` are the only added flags at this time. The get 270 operation is optionally performed depending upon if the output buffer is 271 provided or not. 272 3. The range is specified through ``start`` and ``end``. 273 4. The walk can abort before visiting the complete range such as the user buffer 274 can get full etc. The walk ending address is specified in ``walk_end``. 275 5. The output buffer of ``struct page_region`` array and size is specified in 276 ``vec`` and ``vec_len``. 277 6. The optional maximum requested pages are specified in the ``max_pages``. 278 7. The masks are specified in ``category_mask``, ``category_anyof_mask``, 279 ``category_inverted`` and ``return_mask``. 280 281Find pages which have been written and WP them as well:: 282 283 struct pm_scan_arg arg = { 284 .size = sizeof(arg), 285 .flags = PM_SCAN_WP_MATCHING | PM_SCAN_CHECK_WPASYNC, 286 .. 287 .category_mask = PAGE_IS_WRITTEN, 288 .return_mask = PAGE_IS_WRITTEN, 289 }; 290 291Find pages which have been written, are file backed, not swapped and either 292present or huge:: 293 294 struct pm_scan_arg arg = { 295 .size = sizeof(arg), 296 .flags = 0, 297 .. 298 .category_mask = PAGE_IS_WRITTEN | PAGE_IS_FILE, 299 .category_inverted = PAGE_IS_SWAPPED, 300 .category_anyof_mask = PAGE_IS_PRESENT | PAGE_IS_HUGE, 301 .return_mask = PAGE_IS_WRITTEN | PAGE_IS_SWAPPED | 302 PAGE_IS_PRESENT | PAGE_IS_HUGE, 303 }; 304 305The ``PAGE_IS_WRITTEN`` flag can be considered as a better-performing alternative 306of soft-dirty flag. It doesn't get affected by VMA merging of the kernel and hence 307the user can find the true soft-dirty pages in case of normal pages. (There may 308still be extra dirty pages reported for THP or Hugetlb pages.) 309 310"PAGE_IS_WRITTEN" category is used with uffd write protect-enabled ranges to 311implement memory dirty tracking in userspace: 312 313 1. The userfaultfd file descriptor is created with ``userfaultfd`` syscall. 314 2. The ``UFFD_FEATURE_WP_UNPOPULATED`` and ``UFFD_FEATURE_WP_ASYNC`` features 315 are set by ``UFFDIO_API`` IOCTL. 316 3. The memory range is registered with ``UFFDIO_REGISTER_MODE_WP`` mode 317 through ``UFFDIO_REGISTER`` IOCTL. 318 4. Then any part of the registered memory or the whole memory region must 319 be write protected using ``PAGEMAP_SCAN`` IOCTL with flag ``PM_SCAN_WP_MATCHING`` 320 or the ``UFFDIO_WRITEPROTECT`` IOCTL can be used. Both of these perform the 321 same operation. The former is better in terms of performance. 322 5. Now the ``PAGEMAP_SCAN`` IOCTL can be used to either just find pages which 323 have been written to since they were last marked and/or optionally write protect 324 the pages as well. 325