1MPAGE_READPAGES(9)               The Linux VFS              MPAGE_READPAGES(9)
2
3
4

NAME

6       mpage_readpages - populate an address space with some pages & start
7       reads against them
8

SYNOPSIS

10       int mpage_readpages(struct address_space * mapping,
11                           struct list_head * pages, unsigned nr_pages,
12                           get_block_t get_block);
13

ARGUMENTS

15       mapping
16           the address_space
17
18       pages
19           The address of a list_head which contains the target pages. These
20           pages have their ->index populated and are otherwise uninitialised.
21           The page at pages->prev has the lowest file offset, and reads
22           should be issued in pages->prev to pages->next order.
23
24       nr_pages
25           The number of pages at *pages
26
27       get_block
28           The filesystem´s block mapper function.
29

DESCRIPTION

31       This function walks the pages and the blocks within each page, building
32       and emitting large BIOs.
33
34       If anything unusual happens, such as:
35
36       - encountering a page which has buffers - encountering a page which has
37       a non-hole after a hole - encountering a page with non-contiguous
38       blocks
39
40       then this code just gives up and calls the buffer_head-based read
41       function. It does handle a page which has holes at the end - that is a
42       common case: the end-of-file on blocksize < PAGE_CACHE_SIZE setups.
43

BH_BOUNDARY EXPLANATION

45       There is a problem. The mpage read code assembles several pages, gets
46       all their disk mappings, and then submits them all. That´s fine, but
47       obtaining the disk mappings may require I/O. Reads of indirect blocks,
48       for example.
49
50       So an mpage read of the first 16 blocks of an ext2 file will cause I/O
51       to be
52

SUBMITTED IN THE FOLLOWING ORDER

54       12 0 1 2 3 4 5 6 7 8 9 10 11 13 14 15 16
55
56       because the indirect block has to be read to get the mappings of blocks
57       13,14,15,16. Obviously, this impacts performance.
58
59       So what we do it to allow the filesystem´s get_block function to set
60       BH_Boundary when it maps block 11. BH_Boundary says: mapping of the
61       block after this one will require I/O against a block which is probably
62       close to this one. So you should push what I/O you have currently
63       accumulated.
64
65       This all causes the disk requests to be issued in the correct order.
66

68Kernel Hackers Manual 2.6.         June 2019                MPAGE_READPAGES(9)
Impressum