]>
Commit | Line | Data |
---|---|---|
2cb7cef9 BS |
1 | Subject: scsi_error: TASK ABORTED status handling improvement |
2 | From: Vladislav Bolkhovitin <vst@vlnb.net> | |
3 | Date: Fri Nov 7 11:21:03 2008 -0600: | |
4 | Git: ab2d67aa39bc63a60fba748b27cf4d962d0dc7f8 | |
5 | ||
6 | This patch improves handling of TASK ABORTED status by Linux SCSI | |
7 | mid-layer. Currently, command returned with this status considered | |
8 | failed and returned to upper layers. It leads to additional error | |
9 | recovery load on file systems and block layer, which sometimes can | |
10 | cause undesired side effects, like I/O errors and file systems | |
11 | corruptions. See http://lkml.org/lkml/2008/11/1/38, for instance. | |
12 | ||
13 | From other side, TASK ABORTED status is returned by SCSI target if the | |
14 | corresponding command was aborted by another initiator and the target | |
15 | has TAS bit set in the control mode page. So, in the majority of cases | |
16 | commands with TASK ABORTED status should be simply retried. In other | |
17 | cases, maybe_retry path will not retry if no retries are allowed. | |
18 | ||
19 | This patch implement suggestion by James Bottomley from | |
20 | http://marc.info/?l=linux-scsi&m=121932916906009&w=2. | |
21 | ||
22 | Signed-off-by: Vladislav Bolkhovitin <vst@vlnb.net> | |
23 | Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com> | |
24 | Signed-off-by: Hannes Reinecke <hare@suse.de> | |
25 | ||
26 | --- | |
27 | drivers/scsi/scsi_error.c | 3 ++- | |
28 | 1 file changed, 2 insertions(+), 1 deletion(-) | |
29 | ||
30 | --- a/drivers/scsi/scsi_error.c | |
31 | +++ b/drivers/scsi/scsi_error.c | |
32 | @@ -1486,8 +1486,9 @@ int scsi_decide_disposition(struct scsi_ | |
33 | return ADD_TO_MLQUEUE; | |
34 | case GOOD: | |
35 | case COMMAND_TERMINATED: | |
36 | - case TASK_ABORTED: | |
37 | return SUCCESS; | |
38 | + case TASK_ABORTED: | |
39 | + goto maybe_retry; | |
40 | case CHECK_CONDITION: | |
41 | rtn = scsi_check_sense(scmd); | |
42 | if (rtn == NEEDS_RETRY) |