Based on kernel version 4.16.1. Page generated on 2018-04-09 11:53 EST.
1 MOTIVATION 2 3 The idle page tracking feature allows to track which memory pages are being 4 accessed by a workload and which are idle. This information can be useful for 5 estimating the workload's working set size, which, in turn, can be taken into 6 account when configuring the workload parameters, setting memory cgroup limits, 7 or deciding where to place the workload within a compute cluster. 8 9 It is enabled by CONFIG_IDLE_PAGE_TRACKING=y. 10 11 USER API 12 13 The idle page tracking API is located at /sys/kernel/mm/page_idle. Currently, 14 it consists of the only read-write file, /sys/kernel/mm/page_idle/bitmap. 15 16 The file implements a bitmap where each bit corresponds to a memory page. The 17 bitmap is represented by an array of 8-byte integers, and the page at PFN #i is 18 mapped to bit #i%64 of array element #i/64, byte order is native. When a bit is 19 set, the corresponding page is idle. 20 21 A page is considered idle if it has not been accessed since it was marked idle 22 (for more details on what "accessed" actually means see the IMPLEMENTATION 23 DETAILS section). To mark a page idle one has to set the bit corresponding to 24 the page by writing to the file. A value written to the file is OR-ed with the 25 current bitmap value. 26 27 Only accesses to user memory pages are tracked. These are pages mapped to a 28 process address space, page cache and buffer pages, swap cache pages. For other 29 page types (e.g. SLAB pages) an attempt to mark a page idle is silently ignored, 30 and hence such pages are never reported idle. 31 32 For huge pages the idle flag is set only on the head page, so one has to read 33 /proc/kpageflags in order to correctly count idle huge pages. 34 35 Reading from or writing to /sys/kernel/mm/page_idle/bitmap will return 36 -EINVAL if you are not starting the read/write on an 8-byte boundary, or 37 if the size of the read/write is not a multiple of 8 bytes. Writing to 38 this file beyond max PFN will return -ENXIO. 39 40 That said, in order to estimate the amount of pages that are not used by a 41 workload one should: 42 43 1. Mark all the workload's pages as idle by setting corresponding bits in 44 /sys/kernel/mm/page_idle/bitmap. The pages can be found by reading 45 /proc/pid/pagemap if the workload is represented by a process, or by 46 filtering out alien pages using /proc/kpagecgroup in case the workload is 47 placed in a memory cgroup. 48 49 2. Wait until the workload accesses its working set. 50 51 3. Read /sys/kernel/mm/page_idle/bitmap and count the number of bits set. If 52 one wants to ignore certain types of pages, e.g. mlocked pages since they 53 are not reclaimable, he or she can filter them out using /proc/kpageflags. 54 55 See Documentation/vm/pagemap.txt for more information about /proc/pid/pagemap, 56 /proc/kpageflags, and /proc/kpagecgroup. 57 58 IMPLEMENTATION DETAILS 59 60 The kernel internally keeps track of accesses to user memory pages in order to 61 reclaim unreferenced pages first on memory shortage conditions. A page is 62 considered referenced if it has been recently accessed via a process address 63 space, in which case one or more PTEs it is mapped to will have the Accessed bit 64 set, or marked accessed explicitly by the kernel (see mark_page_accessed()). The 65 latter happens when: 66 67 - a userspace process reads or writes a page using a system call (e.g. read(2) 68 or write(2)) 69 70 - a page that is used for storing filesystem buffers is read or written, 71 because a process needs filesystem metadata stored in it (e.g. lists a 72 directory tree) 73 74 - a page is accessed by a device driver using get_user_pages() 75 76 When a dirty page is written to swap or disk as a result of memory reclaim or 77 exceeding the dirty memory limit, it is not marked referenced. 78 79 The idle memory tracking feature adds a new page flag, the Idle flag. This flag 80 is set manually, by writing to /sys/kernel/mm/page_idle/bitmap (see the USER API 81 section), and cleared automatically whenever a page is referenced as defined 82 above. 83 84 When a page is marked idle, the Accessed bit must be cleared in all PTEs it is 85 mapped to, otherwise we will not be able to detect accesses to the page coming 86 from a process address space. To avoid interference with the reclaimer, which, 87 as noted above, uses the Accessed bit to promote actively referenced pages, one 88 more page flag is introduced, the Young flag. When the PTE Accessed bit is 89 cleared as a result of setting or updating a page's Idle flag, the Young flag 90 is set on the page. The reclaimer treats the Young flag as an extra PTE 91 Accessed bit and therefore will consider such a page as referenced. 92 93 Since the idle memory tracking feature is based on the memory reclaimer logic, 94 it only works with pages that are on an LRU list, other pages are silently 95 ignored. That means it will ignore a user memory page if it is isolated, but 96 since there are usually not many of them, it should not affect the overall 97 result noticeably. In order not to stall scanning of the idle page bitmap, 98 locked pages may be skipped too.