[SCSI] ipr: Fix sparse error in ipr driver
authorwenxiong@linux.vnet.ibm.com <wenxiong@linux.vnet.ibm.com>
Fri, 11 Jan 2013 23:43:54 +0000 (17:43 -0600)
committerJames Bottomley <JBottomley@Parallels.com>
Tue, 29 Jan 2013 23:48:53 +0000 (10:48 +1100)
This patch fixes the following sparse error:

 CHECK   drivers/scsi/ipr.c
spinlock.h:147:9: warning: context imbalance in 'ipr_reset_reload' - unexpected unlock

Signed-off-by: Wen Xiong <wenxiong@linux.vnet.ibm.com>
Acked-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
drivers/scsi/ipr.c

index 17aea2d1ec7a6b7aadc8b740f57a1be4f7f11e6a..8fa79b83f2d38de823569ed5e9533b23d0238157 100644 (file)
@@ -2552,36 +2552,6 @@ static void ipr_oper_timeout(struct ipr_cmnd *ipr_cmd)
        LEAVE;
 }
 
-/**
- * ipr_reset_reload - Reset/Reload the IOA
- * @ioa_cfg:           ioa config struct
- * @shutdown_type:     shutdown type
- *
- * This function resets the adapter and re-initializes it.
- * This function assumes that all new host commands have been stopped.
- * Return value:
- *     SUCCESS / FAILED
- **/
-static int ipr_reset_reload(struct ipr_ioa_cfg *ioa_cfg,
-                           enum ipr_shutdown_type shutdown_type)
-{
-       if (!ioa_cfg->in_reset_reload)
-               ipr_initiate_ioa_reset(ioa_cfg, shutdown_type);
-
-       spin_unlock_irq(ioa_cfg->host->host_lock);
-       wait_event(ioa_cfg->reset_wait_q, !ioa_cfg->in_reset_reload);
-       spin_lock_irq(ioa_cfg->host->host_lock);
-
-       /* If we got hit with a host reset while we were already resetting
-        the adapter for some reason, and the reset failed. */
-       if (ioa_cfg->hrrq[IPR_INIT_HRRQ].ioa_is_dead) {
-               ipr_trace;
-               return FAILED;
-       }
-
-       return SUCCESS;
-}
-
 /**
  * ipr_find_ses_entry - Find matching SES in SES table
  * @res:       resource entry struct of SES
@@ -4797,22 +4767,18 @@ static int ipr_slave_alloc(struct scsi_device *sdev)
        return rc;
 }
 
-/**
- * ipr_eh_host_reset - Reset the host adapter
- * @scsi_cmd:  scsi command struct
- *
- * Return value:
- *     SUCCESS / FAILED
- **/
-static int __ipr_eh_host_reset(struct scsi_cmnd *scsi_cmd)
+static int ipr_eh_host_reset(struct scsi_cmnd *cmd)
 {
        struct ipr_ioa_cfg *ioa_cfg;
-       int rc;
+       unsigned long lock_flags = 0;
+       int rc = SUCCESS;
 
        ENTER;
-       ioa_cfg = (struct ipr_ioa_cfg *) scsi_cmd->device->host->hostdata;
+       ioa_cfg = (struct ipr_ioa_cfg *) cmd->device->host->hostdata;
+       spin_lock_irqsave(ioa_cfg->host->host_lock, lock_flags);
 
        if (!ioa_cfg->in_reset_reload) {
+               ipr_initiate_ioa_reset(ioa_cfg, IPR_SHUTDOWN_ABBREV);
                dev_err(&ioa_cfg->pdev->dev,
                        "Adapter being reset as a result of error recovery.\n");
 
@@ -4820,20 +4786,19 @@ static int __ipr_eh_host_reset(struct scsi_cmnd *scsi_cmd)
                        ioa_cfg->sdt_state = GET_DUMP;
        }
 
-       rc = ipr_reset_reload(ioa_cfg, IPR_SHUTDOWN_ABBREV);
-
-       LEAVE;
-       return rc;
-}
-
-static int ipr_eh_host_reset(struct scsi_cmnd *cmd)
-{
-       int rc;
+       spin_unlock_irqrestore(ioa_cfg->host->host_lock, lock_flags);
+       wait_event(ioa_cfg->reset_wait_q, !ioa_cfg->in_reset_reload);
+       spin_lock_irqsave(ioa_cfg->host->host_lock, lock_flags);
 
-       spin_lock_irq(cmd->device->host->host_lock);
-       rc = __ipr_eh_host_reset(cmd);
-       spin_unlock_irq(cmd->device->host->host_lock);
+       /* If we got hit with a host reset while we were already resetting
+        the adapter for some reason, and the reset failed. */
+       if (ioa_cfg->hrrq[IPR_INIT_HRRQ].ioa_is_dead) {
+               ipr_trace;
+               rc = FAILED;
+       }
 
+       spin_unlock_irqrestore(ioa_cfg->host->host_lock, lock_flags);
+       LEAVE;
        return rc;
 }
 
This page took 0.029806 seconds and 5 git commands to generate.