Loading...
1Direct Access for files
2-----------------------
3
4Motivation
5----------
6
7The page cache is usually used to buffer reads and writes to files.
8It is also used to provide the pages which are mapped into userspace
9by a call to mmap.
10
11For block devices that are memory-like, the page cache pages would be
12unnecessary copies of the original storage. The DAX code removes the
13extra copy by performing reads and writes directly to the storage device.
14For file mappings, the storage device is mapped directly into userspace.
15
16
17Usage
18-----
19
20If you have a block device which supports DAX, you can make a filesystem
21on it as usual. The DAX code currently only supports files with a block
22size equal to your kernel's PAGE_SIZE, so you may need to specify a block
23size when creating the filesystem. When mounting it, use the "-o dax"
24option on the command line or add 'dax' to the options in /etc/fstab.
25
26
27Implementation Tips for Block Driver Writers
28--------------------------------------------
29
30To support DAX in your block driver, implement the 'direct_access'
31block device operation. It is used to translate the sector number
32(expressed in units of 512-byte sectors) to a page frame number (pfn)
33that identifies the physical page for the memory. It also returns a
34kernel virtual address that can be used to access the memory.
35
36The direct_access method takes a 'size' parameter that indicates the
37number of bytes being requested. The function should return the number
38of bytes that can be contiguously accessed at that offset. It may also
39return a negative errno if an error occurs.
40
41In order to support this method, the storage must be byte-accessible by
42the CPU at all times. If your device uses paging techniques to expose
43a large amount of memory through a smaller window, then you cannot
44implement direct_access. Equally, if your device can occasionally
45stall the CPU for an extended period, you should also not attempt to
46implement direct_access.
47
48These block devices may be used for inspiration:
49- axonram: Axon DDR2 device driver
50- brd: RAM backed block device driver
51- dcssblk: s390 dcss block device driver
52
53
54Implementation Tips for Filesystem Writers
55------------------------------------------
56
57Filesystem support consists of
58- adding support to mark inodes as being DAX by setting the S_DAX flag in
59 i_flags
60- implementing the direct_IO address space operation, and calling
61 dax_do_io() instead of blockdev_direct_IO() if S_DAX is set
62- implementing an mmap file operation for DAX files which sets the
63 VM_MIXEDMAP and VM_HUGEPAGE flags on the VMA, and setting the vm_ops to
64 include handlers for fault, pmd_fault and page_mkwrite (which should
65 probably call dax_fault(), dax_pmd_fault() and dax_mkwrite(), passing the
66 appropriate get_block() callback)
67- calling dax_truncate_page() instead of block_truncate_page() for DAX files
68- calling dax_zero_page_range() instead of zero_user() for DAX files
69- ensuring that there is sufficient locking between reads, writes,
70 truncates and page faults
71
72The get_block() callback passed to the DAX functions may return
73uninitialised extents. If it does, it must ensure that simultaneous
74calls to get_block() (for example by a page-fault racing with a read()
75or a write()) work correctly.
76
77These filesystems may be used for inspiration:
78- ext2: the second extended filesystem, see Documentation/filesystems/ext2.txt
79- ext4: the fourth extended filesystem, see Documentation/filesystems/ext4.txt
80
81
82Shortcomings
83------------
84
85Even if the kernel or its modules are stored on a filesystem that supports
86DAX on a block device that supports DAX, they will still be copied into RAM.
87
88The DAX code does not work correctly on architectures which have virtually
89mapped caches such as ARM, MIPS and SPARC.
90
91Calling get_user_pages() on a range of user memory that has been mmaped
92from a DAX file will fail as there are no 'struct page' to describe
93those pages. This problem is being worked on. That means that O_DIRECT
94reads/writes to those memory ranges from a non-DAX file will fail (note
95that O_DIRECT reads/writes _of a DAX file_ do work, it is the memory
96that is being accessed that is key here). Other things that will not
97work include RDMA, sendfile() and splice().
1Direct Access for files
2-----------------------
3
4Motivation
5----------
6
7The page cache is usually used to buffer reads and writes to files.
8It is also used to provide the pages which are mapped into userspace
9by a call to mmap.
10
11For block devices that are memory-like, the page cache pages would be
12unnecessary copies of the original storage. The DAX code removes the
13extra copy by performing reads and writes directly to the storage device.
14For file mappings, the storage device is mapped directly into userspace.
15
16
17Usage
18-----
19
20If you have a block device which supports DAX, you can make a filesystem
21on it as usual. The DAX code currently only supports files with a block
22size equal to your kernel's PAGE_SIZE, so you may need to specify a block
23size when creating the filesystem. When mounting it, use the "-o dax"
24option on the command line or add 'dax' to the options in /etc/fstab.
25
26
27Implementation Tips for Block Driver Writers
28--------------------------------------------
29
30To support DAX in your block driver, implement the 'direct_access'
31block device operation. It is used to translate the sector number
32(expressed in units of 512-byte sectors) to a page frame number (pfn)
33that identifies the physical page for the memory. It also returns a
34kernel virtual address that can be used to access the memory.
35
36The direct_access method takes a 'size' parameter that indicates the
37number of bytes being requested. The function should return the number
38of bytes that can be contiguously accessed at that offset. It may also
39return a negative errno if an error occurs.
40
41In order to support this method, the storage must be byte-accessible by
42the CPU at all times. If your device uses paging techniques to expose
43a large amount of memory through a smaller window, then you cannot
44implement direct_access. Equally, if your device can occasionally
45stall the CPU for an extended period, you should also not attempt to
46implement direct_access.
47
48These block devices may be used for inspiration:
49- axonram: Axon DDR2 device driver
50- brd: RAM backed block device driver
51- dcssblk: s390 dcss block device driver
52- pmem: NVDIMM persistent memory driver
53
54
55Implementation Tips for Filesystem Writers
56------------------------------------------
57
58Filesystem support consists of
59- adding support to mark inodes as being DAX by setting the S_DAX flag in
60 i_flags
61- implementing ->read_iter and ->write_iter operations which use dax_iomap_rw()
62 when inode has S_DAX flag set
63- implementing an mmap file operation for DAX files which sets the
64 VM_MIXEDMAP and VM_HUGEPAGE flags on the VMA, and setting the vm_ops to
65 include handlers for fault, pmd_fault, page_mkwrite, pfn_mkwrite. These
66 handlers should probably call dax_iomap_fault() (for fault and page_mkwrite
67 handlers), dax_iomap_pmd_fault(), dax_pfn_mkwrite() passing the appropriate
68 iomap operations.
69- calling iomap_zero_range() passing appropriate iomap operations instead of
70 block_truncate_page() for DAX files
71- ensuring that there is sufficient locking between reads, writes,
72 truncates and page faults
73
74The iomap handlers for allocating blocks must make sure that allocated blocks
75are zeroed out and converted to written extents before being returned to avoid
76exposure of uninitialized data through mmap.
77
78These filesystems may be used for inspiration:
79- ext2: see Documentation/filesystems/ext2.txt
80- ext4: see Documentation/filesystems/ext4.txt
81- xfs: see Documentation/filesystems/xfs.txt
82
83
84Handling Media Errors
85---------------------
86
87The libnvdimm subsystem stores a record of known media error locations for
88each pmem block device (in gendisk->badblocks). If we fault at such location,
89or one with a latent error not yet discovered, the application can expect
90to receive a SIGBUS. Libnvdimm also allows clearing of these errors by simply
91writing the affected sectors (through the pmem driver, and if the underlying
92NVDIMM supports the clear_poison DSM defined by ACPI).
93
94Since DAX IO normally doesn't go through the driver/bio path, applications or
95sysadmins have an option to restore the lost data from a prior backup/inbuilt
96redundancy in the following ways:
97
981. Delete the affected file, and restore from a backup (sysadmin route):
99 This will free the file system blocks that were being used by the file,
100 and the next time they're allocated, they will be zeroed first, which
101 happens through the driver, and will clear bad sectors.
102
1032. Truncate or hole-punch the part of the file that has a bad-block (at least
104 an entire aligned sector has to be hole-punched, but not necessarily an
105 entire filesystem block).
106
107These are the two basic paths that allow DAX filesystems to continue operating
108in the presence of media errors. More robust error recovery mechanisms can be
109built on top of this in the future, for example, involving redundancy/mirroring
110provided at the block layer through DM, or additionally, at the filesystem
111level. These would have to rely on the above two tenets, that error clearing
112can happen either by sending an IO through the driver, or zeroing (also through
113the driver).
114
115
116Shortcomings
117------------
118
119Even if the kernel or its modules are stored on a filesystem that supports
120DAX on a block device that supports DAX, they will still be copied into RAM.
121
122The DAX code does not work correctly on architectures which have virtually
123mapped caches such as ARM, MIPS and SPARC.
124
125Calling get_user_pages() on a range of user memory that has been mmaped
126from a DAX file will fail when there are no 'struct page' to describe
127those pages. This problem has been addressed in some device drivers
128by adding optional struct page support for pages under the control of
129the driver (see CONFIG_NVDIMM_PFN in drivers/nvdimm for an example of
130how to do this). In the non struct page cases O_DIRECT reads/writes to
131those memory ranges from a non-DAX file will fail (note that O_DIRECT
132reads/writes _of a DAX file_ do work, it is the memory that is being
133accessed that is key here). Other things that will not work in the
134non struct page case include RDMA, sendfile() and splice().