Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1 | dm-io |
| 2 | ===== |
| 3 | |
| 4 | Dm-io provides synchronous and asynchronous I/O services. There are three |
| 5 | types of I/O services available, and each type has a sync and an async |
| 6 | version. |
| 7 | |
| 8 | The user must set up an io_region structure to describe the desired location |
| 9 | of the I/O. Each io_region indicates a block-device along with the starting |
| 10 | sector and size of the region. |
| 11 | |
| 12 | struct io_region { |
| 13 | struct block_device *bdev; |
| 14 | sector_t sector; |
| 15 | sector_t count; |
| 16 | }; |
| 17 | |
| 18 | Dm-io can read from one io_region or write to one or more io_regions. Writes |
| 19 | to multiple regions are specified by an array of io_region structures. |
| 20 | |
| 21 | The first I/O service type takes a list of memory pages as the data buffer for |
| 22 | the I/O, along with an offset into the first page. |
| 23 | |
| 24 | struct page_list { |
| 25 | struct page_list *next; |
| 26 | struct page *page; |
| 27 | }; |
| 28 | |
| 29 | int dm_io_sync(unsigned int num_regions, struct io_region *where, int rw, |
| 30 | struct page_list *pl, unsigned int offset, |
| 31 | unsigned long *error_bits); |
| 32 | int dm_io_async(unsigned int num_regions, struct io_region *where, int rw, |
| 33 | struct page_list *pl, unsigned int offset, |
| 34 | io_notify_fn fn, void *context); |
| 35 | |
| 36 | The second I/O service type takes an array of bio vectors as the data buffer |
| 37 | for the I/O. This service can be handy if the caller has a pre-assembled bio, |
| 38 | but wants to direct different portions of the bio to different devices. |
| 39 | |
| 40 | int dm_io_sync_bvec(unsigned int num_regions, struct io_region *where, |
| 41 | int rw, struct bio_vec *bvec, |
| 42 | unsigned long *error_bits); |
| 43 | int dm_io_async_bvec(unsigned int num_regions, struct io_region *where, |
| 44 | int rw, struct bio_vec *bvec, |
| 45 | io_notify_fn fn, void *context); |
| 46 | |
| 47 | The third I/O service type takes a pointer to a vmalloc'd memory buffer as the |
| 48 | data buffer for the I/O. This service can be handy if the caller needs to do |
| 49 | I/O to a large region but doesn't want to allocate a large number of individual |
| 50 | memory pages. |
| 51 | |
| 52 | int dm_io_sync_vm(unsigned int num_regions, struct io_region *where, int rw, |
| 53 | void *data, unsigned long *error_bits); |
| 54 | int dm_io_async_vm(unsigned int num_regions, struct io_region *where, int rw, |
| 55 | void *data, io_notify_fn fn, void *context); |
| 56 | |
| 57 | Callers of the asynchronous I/O services must include the name of a completion |
| 58 | callback routine and a pointer to some context data for the I/O. |
| 59 | |
| 60 | typedef void (*io_notify_fn)(unsigned long error, void *context); |
| 61 | |
| 62 | The "error" parameter in this callback, as well as the "*error" parameter in |
| 63 | all of the synchronous versions, is a bitset (instead of a simple error value). |
| 64 | In the case of an write-I/O to multiple regions, this bitset allows dm-io to |
| 65 | indicate success or failure on each individual region. |
| 66 | |
| 67 | Before using any of the dm-io services, the user should call dm_io_get() |
| 68 | and specify the number of pages they expect to perform I/O on concurrently. |
| 69 | Dm-io will attempt to resize its mempool to make sure enough pages are |
| 70 | always available in order to avoid unnecessary waiting while performing I/O. |
| 71 | |
| 72 | When the user is finished using the dm-io services, they should call |
| 73 | dm_io_put() and specify the same number of pages that were given on the |
| 74 | dm_io_get() call. |
| 75 | |