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 uffd-wp write-protected (since 5.13) (see 23 Documentation/admin-guide/mm/userfaultfd.rst) 24 * Bits 58-60 zero 25 * Bit 61 page is file-page or shared-anon (since 3.5) 26 * Bit 62 page swapped 27 * Bit 63 page present 28 29 Since Linux 4.0 only users with the CAP_SYS_ADMIN capability can get PFNs. 30 In 4.0 and 4.1 opens by unprivileged fail with -EPERM. Starting from 31 4.2 the PFN field is zeroed if the user does not have CAP_SYS_ADMIN. 32 Reason: information about PFNs helps in exploiting Rowhammer vulnerability. 33 34 If the page is not present but in swap, then the PFN contains an 35 encoding of the swap file number and the page's offset into the 36 swap. Unmapped pages return a null PFN. This allows determining 37 precisely which pages are mapped (or in swap) and comparing mapped 38 pages between processes. 39 40 Efficient users of this interface will use ``/proc/pid/maps`` to 41 determine which areas of memory are actually mapped and llseek to 42 skip over unmapped regions. 43 44 * ``/proc/kpagecount``. This file contains a 64-bit count of the number of 45 times each page is mapped, indexed by PFN. 46 47The page-types tool in the tools/mm directory can be used to query the 48number of times a page is mapped. 49 50 * ``/proc/kpageflags``. This file contains a 64-bit set of flags for each 51 page, indexed by PFN. 52 53 The flags are (from ``fs/proc/page.c``, above kpageflags_read): 54 55 0. LOCKED 56 1. ERROR 57 2. REFERENCED 58 3. UPTODATE 59 4. DIRTY 60 5. LRU 61 6. ACTIVE 62 7. SLAB 63 8. WRITEBACK 64 9. RECLAIM 65 10. BUDDY 66 11. MMAP 67 12. ANON 68 13. SWAPCACHE 69 14. SWAPBACKED 70 15. COMPOUND_HEAD 71 16. COMPOUND_TAIL 72 17. HUGE 73 18. UNEVICTABLE 74 19. HWPOISON 75 20. NOPAGE 76 21. KSM 77 22. THP 78 23. OFFLINE 79 24. ZERO_PAGE 80 25. IDLE 81 26. PGTABLE 82 83 * ``/proc/kpagecgroup``. This file contains a 64-bit inode number of the 84 memory cgroup each page is charged to, indexed by PFN. Only available when 85 CONFIG_MEMCG is set. 86 87Short descriptions to the page flags 88==================================== 89 900 - LOCKED 91 The page is being locked for exclusive access, e.g. by undergoing read/write 92 IO. 937 - SLAB 94 The page is managed by the SLAB/SLUB kernel memory allocator. 95 When compound page is used, either will only set this flag on the head 96 page. 9710 - BUDDY 98 A free memory block managed by the buddy system allocator. 99 The buddy system organizes free memory in blocks of various orders. 100 An order N block has 2^N physically contiguous pages, with the BUDDY flag 101 set for and _only_ for the first page. 10215 - COMPOUND_HEAD 103 A compound page with order N consists of 2^N physically contiguous pages. 104 A compound page with order 2 takes the form of "HTTT", where H donates its 105 head page and T donates its tail page(s). The major consumers of compound 106 pages are hugeTLB pages (Documentation/admin-guide/mm/hugetlbpage.rst), 107 the SLUB etc. memory allocators and various device drivers. 108 However in this interface, only huge/giga pages are made visible 109 to end users. 11016 - COMPOUND_TAIL 111 A compound page tail (see description above). 11217 - HUGE 113 This is an integral part of a HugeTLB page. 11419 - HWPOISON 115 Hardware detected memory corruption on this page: don't touch the data! 11620 - NOPAGE 117 No page frame exists at the requested address. 11821 - KSM 119 Identical memory pages dynamically shared between one or more processes. 12022 - THP 121 Contiguous pages which construct transparent hugepages. 12223 - OFFLINE 123 The page is logically offline. 12424 - ZERO_PAGE 125 Zero page for pfn_zero or huge_zero page. 12625 - IDLE 127 The page has not been accessed since it was marked idle (see 128 Documentation/admin-guide/mm/idle_page_tracking.rst). 129 Note that this flag may be stale in case the page was accessed via 130 a PTE. To make sure the flag is up-to-date one has to read 131 ``/sys/kernel/mm/page_idle/bitmap`` first. 13226 - PGTABLE 133 The page is in use as a page table. 134 135IO related page flags 136--------------------- 137 1381 - ERROR 139 IO error occurred. 1403 - UPTODATE 141 The page has up-to-date data. 142 ie. for file backed page: (in-memory data revision >= on-disk one) 1434 - DIRTY 144 The page has been written to, hence contains new data. 145 i.e. for file backed page: (in-memory data revision > on-disk one) 1468 - WRITEBACK 147 The page is being synced to disk. 148 149LRU related page flags 150---------------------- 151 1525 - LRU 153 The page is in one of the LRU lists. 1546 - ACTIVE 155 The page is in the active LRU list. 15618 - UNEVICTABLE 157 The page is in the unevictable (non-)LRU list It is somehow pinned and 158 not a candidate for LRU page reclaims, e.g. ramfs pages, 159 shmctl(SHM_LOCK) and mlock() memory segments. 1602 - REFERENCED 161 The page has been referenced since last LRU list enqueue/requeue. 1629 - RECLAIM 163 The page will be reclaimed soon after its pageout IO completed. 16411 - MMAP 165 A memory mapped page. 16612 - ANON 167 A memory mapped page that is not part of a file. 16813 - SWAPCACHE 169 The page is mapped to swap space, i.e. has an associated swap entry. 17014 - SWAPBACKED 171 The page is backed by swap/RAM. 172 173The page-types tool in the tools/mm directory can be used to query the 174above flags. 175 176Using pagemap to do something useful 177==================================== 178 179The general procedure for using pagemap to find out about a process' memory 180usage goes like this: 181 182 1. Read ``/proc/pid/maps`` to determine which parts of the memory space are 183 mapped to what. 184 2. Select the maps you are interested in -- all of them, or a particular 185 library, or the stack or the heap, etc. 186 3. Open ``/proc/pid/pagemap`` and seek to the pages you would like to examine. 187 4. Read a u64 for each page from pagemap. 188 5. Open ``/proc/kpagecount`` and/or ``/proc/kpageflags``. For each PFN you 189 just read, seek to that entry in the file, and read the data you want. 190 191For example, to find the "unique set size" (USS), which is the amount of 192memory that a process is using that is not shared with any other process, 193you can go through every map in the process, find the PFNs, look those up 194in kpagecount, and tally up the number of pages that are only referenced 195once. 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 has been written to from the time it was write protected 251- ``PAGE_IS_FILE`` - Page is file backed 252- ``PAGE_IS_PRESENT`` - Page is present in the memory 253- ``PAGE_IS_SWAPPED`` - Page is in swapped 254- ``PAGE_IS_PFNZERO`` - Page has zero PFN 255- ``PAGE_IS_HUGE`` - Page is THP or Hugetlb backed 256- ``PAGE_IS_SOFT_DIRTY`` - Page is soft-dirty 257 258The ``struct pm_scan_arg`` is used as the argument of the IOCTL. 259 260 1. The size of the ``struct pm_scan_arg`` must be specified in the ``size`` 261 field. This field will be helpful in recognizing the structure if extensions 262 are done later. 263 2. The flags can be specified in the ``flags`` field. The ``PM_SCAN_WP_MATCHING`` 264 and ``PM_SCAN_CHECK_WPASYNC`` are the only added flags at this time. The get 265 operation is optionally performed depending upon if the output buffer is 266 provided or not. 267 3. The range is specified through ``start`` and ``end``. 268 4. The walk can abort before visiting the complete range such as the user buffer 269 can get full etc. The walk ending address is specified in``end_walk``. 270 5. The output buffer of ``struct page_region`` array and size is specified in 271 ``vec`` and ``vec_len``. 272 6. The optional maximum requested pages are specified in the ``max_pages``. 273 7. The masks are specified in ``category_mask``, ``category_anyof_mask``, 274 ``category_inverted`` and ``return_mask``. 275 276Find pages which have been written and WP them as well:: 277 278 struct pm_scan_arg arg = { 279 .size = sizeof(arg), 280 .flags = PM_SCAN_CHECK_WPASYNC | PM_SCAN_CHECK_WPASYNC, 281 .. 282 .category_mask = PAGE_IS_WRITTEN, 283 .return_mask = PAGE_IS_WRITTEN, 284 }; 285 286Find pages which have been written, are file backed, not swapped and either 287present or huge:: 288 289 struct pm_scan_arg arg = { 290 .size = sizeof(arg), 291 .flags = 0, 292 .. 293 .category_mask = PAGE_IS_WRITTEN | PAGE_IS_SWAPPED, 294 .category_inverted = PAGE_IS_SWAPPED, 295 .category_anyof_mask = PAGE_IS_PRESENT | PAGE_IS_HUGE, 296 .return_mask = PAGE_IS_WRITTEN | PAGE_IS_SWAPPED | 297 PAGE_IS_PRESENT | PAGE_IS_HUGE, 298 }; 299 300The ``PAGE_IS_WRITTEN`` flag can be considered as a better-performing alternative 301of soft-dirty flag. It doesn't get affected by VMA merging of the kernel and hence 302the user can find the true soft-dirty pages in case of normal pages. (There may 303still be extra dirty pages reported for THP or Hugetlb pages.) 304 305"PAGE_IS_WRITTEN" category is used with uffd write protect-enabled ranges to 306implement memory dirty tracking in userspace: 307 308 1. The userfaultfd file descriptor is created with ``userfaultfd`` syscall. 309 2. The ``UFFD_FEATURE_WP_UNPOPULATED`` and ``UFFD_FEATURE_WP_ASYNC`` features 310 are set by ``UFFDIO_API`` IOCTL. 311 3. The memory range is registered with ``UFFDIO_REGISTER_MODE_WP`` mode 312 through ``UFFDIO_REGISTER`` IOCTL. 313 4. Then any part of the registered memory or the whole memory region must 314 be write protected using ``PAGEMAP_SCAN`` IOCTL with flag ``PM_SCAN_WP_MATCHING`` 315 or the ``UFFDIO_WRITEPROTECT`` IOCTL can be used. Both of these perform the 316 same operation. The former is better in terms of performance. 317 5. Now the ``PAGEMAP_SCAN`` IOCTL can be used to either just find pages which 318 have been written to since they were last marked and/or optionally write protect 319 the pages as well. 320