xiaomi-sm8450/android_kernel_xiaomi_sm8450
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[SCSI] ibmvscsi: treat busy and error conditions separately

Browse Source 
This patch fixes a condition where ibmvscsi treats a transport error as a
"busy" condition, so no errors were returned to the scsi mid-layer.
In a RAID environment this means that I/O hung rather than failing
over.

Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
This commit is contained in:
Dave C Boutcher
2006-06-12 21:22:51 -05:00
committed by James Bottomley
parent c65b1445d1
commit cefbda2d6c
2 changed files with 52 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
drivers/scsi/ibmvscsi/rpa_vscsi.c
View File

@@ -208,6 +208,7 @@ int ibmvscsi_init_crq_queue(struct crq_queue *queue,
int max_requests)
{
int rc;
int retrc;
struct vio_dev *vdev = to_vio_dev(hostdata->dev);
queue->msgs = (struct viosrp_crq *)get_zeroed_page(GFP_KERNEL);
@@ -226,7 +227,7 @@ int ibmvscsi_init_crq_queue(struct crq_queue *queue,
gather_partition_info();
set_adapter_info(hostdata);
rc = plpar_hcall_norets(H_REG_CRQ,
retrc = rc = plpar_hcall_norets(H_REG_CRQ,
vdev->unit_address,
queue->msg_token, PAGE_SIZE);
if (rc == H_RESOURCE)
@@ -263,7 +264,7 @@ int ibmvscsi_init_crq_queue(struct crq_queue *queue,
tasklet_init(&hostdata->srp_task, (void *)ibmvscsi_task,
(unsigned long)hostdata);
return 0;
return retrc;
req_irq_failed:
do {