[SCSI] megaraid_sas: Fix instance access in megasas_reset_timer

The following patch for megaraid_sas will fix a potential bad pointer access
in megasas_reset_timer(), when a MegaRAID 9265/9285 or 9360/9380 gets a
timeout.  megasas_build_io_fusion() sets SCp.ptr to be a struct
megasas_cmd_fusion *, but then megasas_reset_timer() was casting SCp.ptr to be
a struct megasas_cmd *, then trying to access cmd->instance, which is invalid.

Just loading instance from scmd->device->host->hostdata in
megasas_reset_timer() fixes the issue.

Signed-off-by: Adam Radford <aradford@gmail.com>
Cc: stable@kernel.org
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

diff --git a/drivers/scsi/megaraid/megaraid_sas_base.c b/drivers/scsi/megaraid/megaraid_sas_base.c
index 8d587cb73cb70c3842883d6026e8f96fa165bcc1..29a994f9c4f1e6a9c0973ecd3cebe413f4937c7b 100644 (file)
--- a/drivers/scsi/megaraid/megaraid_sas_base.c
+++ b/drivers/scsi/megaraid/megaraid_sas_base.c
@@ -1914,7 +1914,6 @@ static int megasas_generic_reset(struct scsi_cmnd *scmd)
 static enum
 blk_eh_timer_return megasas_reset_timer(struct scsi_cmnd *scmd)
 {
-       struct megasas_cmd *cmd = (struct megasas_cmd *)scmd->SCp.ptr;
        struct megasas_instance *instance;
        unsigned long flags;
 
@@ -1923,7 +1922,7 @@ blk_eh_timer_return megasas_reset_timer(struct scsi_cmnd *scmd)
                return BLK_EH_NOT_HANDLED;
        }
 
-       instance = cmd->instance;
+       instance = (struct megasas_instance *)scmd->device->host->hostdata;
        if (!(instance->flag & MEGASAS_FW_BUSY)) {
                /* FW is busy, throttle IO */
                spin_lock_irqsave(instance->host->host_lock, flags);