| 1 |
On Fri, 07 Jun 2013 23:47:33 -0400 |
| 2 |
Richard Yao <ryao@g.o> wrote: |
| 3 |
|
| 4 |
> When you use dm-crypt, block IO requests to a dm-* device will invoke |
| 5 |
> dm_request_fn() -> map_request() -> crypt_map(). If a BIO is a write |
| 6 |
> barrier, crypt_map() will return DM_MAPIO_REMAPPED to map_request(), |
| 7 |
> which will immediately queue it to the device. |
| 8 |
> |
| 9 |
> If a few dozen IOs are queued in rapid succession with multiple write |
| 10 |
> barriers, all write barriers will be executed before any actual write |
| 11 |
> BIOs occur because the write IOs will be processed asynchronously in a |
| 12 |
> work queue. Since the barriers will be long gone by the time the write |
| 13 |
> IOs are queued, they can be queued in any order. |
| 14 |
> |
| 15 |
> Am I misunderstanding this or is dm-crypt ignoring proper write barrier |
| 16 |
> semantics? |
| 17 |
> |
| 18 |
http://www.saout.de/pipermail/dm-crypt/2012-April/002441.html |
| 19 |
http://lwn.net/Articles/400541/ |
| 20 |
|
| 21 |
-- |
| 22 |
|
| 23 |
Sergei |
Archives