Patch 1 of 2
This is a resubmission of yesterdays patch to detect changes on the MSA2012.
I hope I've addressed all concerns. This patch rearranges some of the code
so we also have coverage in the sg and the ioctl paths as well as the main
data path.
The MSA2012 cannot inform the driver of configuration changes since all
management is out of band. This is a departure from any storage we have
supported in the past. We need some way to detect changes on the topology so
we implement this kernel thread. In some instances there's nothing we can do
from the driver (like LUN failure) so just print out a message. In the case
where logical volumes are added or deleted we call rebuild_lun_table to
refreash the driver's view of the world.
Please consider this for inclusion.
Signed-off-by: Mike Miller <[email protected]>
diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c
index d2cb67b..f51545e 100644
--- a/drivers/block/cciss.c
+++ b/drivers/block/cciss.c
@@ -51,6 +51,7 @@
#include <scsi/scsi_ioctl.h>
#include <linux/cdrom.h>
#include <linux/scatterlist.h>
+#include <linux/kthread.h>
#define CCISS_DRIVER_VERSION(maj,min,submin) ((maj<<16)|(min<<8)|(submin))
#define DRIVER_NAME "HP CISS Driver (v 3.6.20)"
@@ -186,6 +187,8 @@ static int sendcmd_withirq(__u8 cmd, int ctlr, void *buff, size_t size,
__u8 page_code, int cmd_type);
static void fail_all_cmds(unsigned long ctlr);
+static int scan_thread(ctlr_info_t *h);
+static int check_for_unit_attention(ctlr_info_t *h, CommandList_struct *c);
#ifdef CONFIG_PROC_FS
static void cciss_procinit(int i);
@@ -735,6 +738,12 @@ static int cciss_getgeo(struct block_device *bdev, struct hd_geometry *geo)
return 0;
}
+static void check_ioctl_unit_attention(ctlr_info_t *host, CommandList_struct *c)
+{
+ if (c->err_info->CommandStatus == CMD_TARGET_STATUS &&
+ c->err_info->ScsiStatus != SAM_STAT_CHECK_CONDITION)
+ (void)check_for_unit_attention(host, c);
+}
/*
* ioctl
*/
@@ -1029,6 +1038,8 @@ static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
iocommand.buf_size,
PCI_DMA_BIDIRECTIONAL);
+ check_ioctl_unit_attention(host, c);
+
/* Copy the error information out */
iocommand.error_info = *(c->err_info);
if (copy_to_user
@@ -1180,6 +1191,7 @@ static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
(dma_addr_t) temp64.val, buff_size[i],
PCI_DMA_BIDIRECTIONAL);
}
+ check_ioctl_unit_attention(host, c);
/* Copy the error information out */
ioc->error_info = *(c->err_info);
if (copy_to_user(argp, ioc, sizeof(*ioc))) {
@@ -2585,12 +2597,14 @@ static inline unsigned int make_status_bytes(unsigned int scsi_status_byte,
((driver_byte & 0xff) << 24);
}
-static inline int evaluate_target_status(CommandList_struct *cmd)
+static inline int evaluate_target_status(ctlr_info_t *h,
+ CommandList_struct *cmd, int *retry_cmd)
{
unsigned char sense_key;
unsigned char status_byte, msg_byte, host_byte, driver_byte;
int error_value;
+ *retry_cmd = 0;
/* If we get in here, it means we got "target status", that is, scsi status */
status_byte = cmd->err_info->ScsiStatus;
driver_byte = DRIVER_OK;
@@ -2618,6 +2632,11 @@ static inline int evaluate_target_status(CommandList_struct *cmd)
if (((sense_key == 0x0) || (sense_key == 0x1)) && !blk_pc_request(cmd->rq))
error_value = 0;
+ if (check_for_unit_attention(h, cmd)) {
+ *retry_cmd = !blk_pc_request(cmd->rq);
+ return 0;
+ }
+
if (!blk_pc_request(cmd->rq)) { /* Not SG_IO or similar? */
if (error_value != 0)
printk(KERN_WARNING "cciss: cmd %p has CHECK CONDITION"
@@ -2647,6 +2666,7 @@ static inline void complete_command(ctlr_info_t *h, CommandList_struct *cmd,
int retry_cmd = 0;
struct request *rq = cmd->rq;
+
rq->errors = 0;
if (timeout)
@@ -2657,7 +2677,7 @@ static inline void complete_command(ctlr_info_t *h, CommandList_struct *cmd,
switch (cmd->err_info->CommandStatus) {
case CMD_TARGET_STATUS:
- rq->errors = evaluate_target_status(cmd);
+ rq->errors = evaluate_target_status(h, cmd, &retry_cmd);
break;
case CMD_DATA_UNDERRUN:
if (blk_fs_request(cmd->rq)) {
@@ -3008,6 +3028,62 @@ static irqreturn_t do_cciss_intr(int irq, void *dev_id)
return IRQ_HANDLED;
}
+static int scan_thread(ctlr_info_t *h)
+{
+ int rc;
+ DECLARE_COMPLETION_ONSTACK(wait);
+ h->rescan_wait = &wait;
+
+ while (!kthread_should_stop()) {
+ rc = wait_for_completion_timeout(&wait, 300 * HZ);
+ if (!rc)
+ continue;
+ else
+ rebuild_lun_table(h, 0);
+ }
+ return 0;
+}
+
+static int check_for_unit_attention(ctlr_info_t *h, CommandList_struct *c)
+{
+ if (c->err_info->SenseInfo[2] != UNIT_ATTENTION)
+ return 0;
+
+ switch (c->err_info->SenseInfo[12]) {
+ case STATE_CHANGED:
+ printk(KERN_WARNING "cciss%d: a state change "
+ "detected, command retried\n", h->ctlr);
+ return 1;
+ break;
+ case LUN_FAILED:
+ printk(KERN_WARNING "cciss%d: LUN failure "
+ "detected, action required\n", h->ctlr);
+ return 1;
+ break;
+ case REPORT_LUNS_CHANGED:
+ printk(KERN_WARNING "cciss%d: report LUN data "
+ "changed\n", h->ctlr);
+ if (h->rescan_wait)
+ complete(h->rescan_wait);
+ return 1;
+ break;
+ case POWER_OR_RESET:
+ printk(KERN_WARNING "cciss%d: a power on "
+ "or device reset detected\n", h->ctlr);
+ return 1;
+ break;
+ case UNIT_ATTENTION_CLEARED:
+ printk(KERN_WARNING "cciss%d: unit attention "
+ "cleared by another initiator\n", h->ctlr);
+ return 1;
+ break;
+ default:
+ printk(KERN_WARNING "cciss%d: unknown "
+ "unit attention detected\n", h->ctlr);
+ return 1;
+ }
+}
+
/*
* We cannot read the structure directly, for portability we must use
* the io functions.
@@ -3600,6 +3685,7 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
int rc;
int dac, return_code;
InquiryData_struct *inq_buff = NULL;
+ struct task_struct *cciss_scan_thread;
if (reset_devices) {
/* Reset the controller with a PCI power-cycle */
@@ -3751,6 +3837,11 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
hba[i]->busy_initializing = 0;
rebuild_lun_table(hba[i], 1);
+ cciss_scan_thread = kthread_run((void *)scan_thread, hba[i],
+ "scan_thread");
+ if (IS_ERR(cciss_scan_thread))
+ return PTR_ERR(cciss_scan_thread);
+
return 1;
clean4:
diff --git a/drivers/block/cciss.h b/drivers/block/cciss.h
index 15e2b84..c85783f 100644
--- a/drivers/block/cciss.h
+++ b/drivers/block/cciss.h
@@ -121,6 +121,7 @@ struct ctlr_info
struct sendcmd_reject_list scsi_rejects;
#endif
unsigned char alive;
+ struct completion *rescan_wait;
};
/* Defining the diffent access_menthods */
diff --git a/drivers/block/cciss_cmd.h b/drivers/block/cciss_cmd.h
index 24e22de..40b1b92 100644
--- a/drivers/block/cciss_cmd.h
+++ b/drivers/block/cciss_cmd.h
@@ -25,6 +25,29 @@
#define CMD_TIMEOUT 0x000B
#define CMD_UNABORTABLE 0x000C
+/* Unit Attentions ASC's as defined for the MSA2012sa */
+#define POWER_OR_RESET 0x29
+#define STATE_CHANGED 0x2a
+#define UNIT_ATTENTION_CLEARED 0x2f
+#define LUN_FAILED 0x3e
+#define REPORT_LUNS_CHANGED 0x3f
+
+/* Unit Attentions ASCQ's as defined for the MSA2012sa */
+
+ /* These ASCQ's defined for ASC = POWER_OR_RESET */
+#define POWER_ON_RESET 0x00
+#define POWER_ON_REBOOT 0x01
+#define SCSI_BUS_RESET 0x02
+#define MSA_TARGET_RESET 0x03
+#define CONTROLLER_FAILOVER 0x04
+#define TRANSCEIVER_SE 0x05
+#define TRANSCEIVER_LVD 0x06
+
+ /* These ASCQ's defined for ASC = STATE_CHANGED */
+#define RESERVATION_PREEMPTED 0x03
+#define ASYM_ACCESS_CHANGED 0x06
+#define LUN_CAPACITY_CHANGED 0x09
+
//transfer direction
#define XFER_NONE 0x00
#define XFER_WRITE 0x01
On Fri, 2009-03-06 at 12:16 -0600, Mike Miller wrote:
> Patch 1 of 2
>
> This is a resubmission of yesterdays patch to detect changes on the MSA2012.
> I hope I've addressed all concerns. This patch rearranges some of the code
> so we also have coverage in the sg and the ioctl paths as well as the main
> data path.
>
> The MSA2012 cannot inform the driver of configuration changes since all
> management is out of band. This is a departure from any storage we have
> supported in the past. We need some way to detect changes on the topology so
> we implement this kernel thread. In some instances there's nothing we can do
> from the driver (like LUN failure) so just print out a message. In the case
> where logical volumes are added or deleted we call rebuild_lun_table to
> refreash the driver's view of the world.
>
> Please consider this for inclusion.
I still don't quite see how the thread stops on module removal ... there
needs to be an explicit kthread_stop() somewhere in the clean up path.
James
> -----Original Message-----
> From: James Bottomley [mailto:[email protected]]
> Sent: Friday, March 06, 2009 12:24 PM
> To: Miller, Mike (OS Dev)
> Cc: Andrew Morton; Jens Axboe; LKML; LKML-scsi;
> [email protected]; [email protected]
> Subject: Re: [PATCH 1/2] resubmit cciss: kernel thread to
> detect changes on MSA2012
>
> On Fri, 2009-03-06 at 12:16 -0600, Mike Miller wrote:
> > Patch 1 of 2
> >
> > This is a resubmission of yesterdays patch to detect
> changes on the MSA2012.
> > I hope I've addressed all concerns. This patch rearranges
> some of the
> > code so we also have coverage in the sg and the ioctl paths
> as well as
> > the main data path.
> >
> > The MSA2012 cannot inform the driver of configuration changes since
> > all management is out of band. This is a departure from any
> storage we
> > have supported in the past. We need some way to detect
> changes on the
> > topology so we implement this kernel thread. In some
> instances there's
> > nothing we can do from the driver (like LUN failure) so
> just print out
> > a message. In the case where logical volumes are added or
> deleted we
> > call rebuild_lun_table to refreash the driver's view of the world.
> >
> > Please consider this for inclusion.
>
> I still don't quite see how the thread stops on module
> removal ... there needs to be an explicit kthread_stop()
> somewhere in the clean up path.
>
I thought that was probably needed, duh. Is there anything else I should address at the same time?
-- mikem-
On Fri, Mar 06, 2009 at 12:24:27PM -0600, James Bottomley wrote:
> On Fri, 2009-03-06 at 12:16 -0600, Mike Miller wrote:
> > Patch 1 of 2
> >
> > This is a resubmission of yesterdays patch to detect changes on the MSA2012.
> > I hope I've addressed all concerns. This patch rearranges some of the code
> > so we also have coverage in the sg and the ioctl paths as well as the main
> > data path.
> >
> > The MSA2012 cannot inform the driver of configuration changes since all
> > management is out of band. This is a departure from any storage we have
> > supported in the past. We need some way to detect changes on the topology so
> > we implement this kernel thread. In some instances there's nothing we can do
> > from the driver (like LUN failure) so just print out a message. In the case
> > where logical volumes are added or deleted we call rebuild_lun_table to
> > refreash the driver's view of the world.
> >
> > Please consider this for inclusion.
>
> I still don't quite see how the thread stops on module removal ... there
> needs to be an explicit kthread_stop() somewhere in the clean up path.
>
> James
>
>
This time I make a call to kthread_stop in cciss_remove_one. The driver can
be unloaded and the thread gets cleaned up.
KNOWN BUG: it seems the timeout must expire before kthread_stop actually
stops the thread. This causes the driver to hang and wait during rmmod. I've
played around with several things but haven't found the correct way to
address the problem. Looking at other drivers hasn't been much help. Any
advice is greatly appreciated.
Thanks,
mikem
Signed-off-by: Mike Miller <[email protected]>
diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c
index d2cb67b..5b5c61d 100644
--- a/drivers/block/cciss.c
+++ b/drivers/block/cciss.c
@@ -51,6 +51,7 @@
#include <scsi/scsi_ioctl.h>
#include <linux/cdrom.h>
#include <linux/scatterlist.h>
+#include <linux/kthread.h>
#define CCISS_DRIVER_VERSION(maj,min,submin) ((maj<<16)|(min<<8)|(submin))
#define DRIVER_NAME "HP CISS Driver (v 3.6.20)"
@@ -186,6 +187,8 @@ static int sendcmd_withirq(__u8 cmd, int ctlr, void *buff, size_t size,
__u8 page_code, int cmd_type);
static void fail_all_cmds(unsigned long ctlr);
+static int scan_thread(ctlr_info_t *h);
+static int check_for_unit_attention(ctlr_info_t *h, CommandList_struct *c);
#ifdef CONFIG_PROC_FS
static void cciss_procinit(int i);
@@ -735,6 +738,12 @@ static int cciss_getgeo(struct block_device *bdev, struct hd_geometry *geo)
return 0;
}
+static void check_ioctl_unit_attention(ctlr_info_t *host, CommandList_struct *c)
+{
+ if (c->err_info->CommandStatus == CMD_TARGET_STATUS &&
+ c->err_info->ScsiStatus != SAM_STAT_CHECK_CONDITION)
+ (void)check_for_unit_attention(host, c);
+}
/*
* ioctl
*/
@@ -1029,6 +1038,8 @@ static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
iocommand.buf_size,
PCI_DMA_BIDIRECTIONAL);
+ check_ioctl_unit_attention(host, c);
+
/* Copy the error information out */
iocommand.error_info = *(c->err_info);
if (copy_to_user
@@ -1180,6 +1191,7 @@ static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
(dma_addr_t) temp64.val, buff_size[i],
PCI_DMA_BIDIRECTIONAL);
}
+ check_ioctl_unit_attention(host, c);
/* Copy the error information out */
ioc->error_info = *(c->err_info);
if (copy_to_user(argp, ioc, sizeof(*ioc))) {
@@ -2585,12 +2597,14 @@ static inline unsigned int make_status_bytes(unsigned int scsi_status_byte,
((driver_byte & 0xff) << 24);
}
-static inline int evaluate_target_status(CommandList_struct *cmd)
+static inline int evaluate_target_status(ctlr_info_t *h,
+ CommandList_struct *cmd, int *retry_cmd)
{
unsigned char sense_key;
unsigned char status_byte, msg_byte, host_byte, driver_byte;
int error_value;
+ *retry_cmd = 0;
/* If we get in here, it means we got "target status", that is, scsi status */
status_byte = cmd->err_info->ScsiStatus;
driver_byte = DRIVER_OK;
@@ -2618,6 +2632,11 @@ static inline int evaluate_target_status(CommandList_struct *cmd)
if (((sense_key == 0x0) || (sense_key == 0x1)) && !blk_pc_request(cmd->rq))
error_value = 0;
+ if (check_for_unit_attention(h, cmd)) {
+ *retry_cmd = !blk_pc_request(cmd->rq);
+ return 0;
+ }
+
if (!blk_pc_request(cmd->rq)) { /* Not SG_IO or similar? */
if (error_value != 0)
printk(KERN_WARNING "cciss: cmd %p has CHECK CONDITION"
@@ -2657,7 +2676,7 @@ static inline void complete_command(ctlr_info_t *h, CommandList_struct *cmd,
switch (cmd->err_info->CommandStatus) {
case CMD_TARGET_STATUS:
- rq->errors = evaluate_target_status(cmd);
+ rq->errors = evaluate_target_status(h, cmd, &retry_cmd);
break;
case CMD_DATA_UNDERRUN:
if (blk_fs_request(cmd->rq)) {
@@ -3008,6 +3027,62 @@ static irqreturn_t do_cciss_intr(int irq, void *dev_id)
return IRQ_HANDLED;
}
+static int scan_thread(ctlr_info_t *h)
+{
+ int rc;
+ DECLARE_COMPLETION_ONSTACK(wait);
+ h->rescan_wait = &wait;
+
+ while (!kthread_should_stop()) {
+ rc = wait_for_completion_timeout(&wait, 300 * HZ);
+ if (!rc)
+ continue;
+ else
+ rebuild_lun_table(h, 0);
+ }
+ return 0;
+}
+
+static int check_for_unit_attention(ctlr_info_t *h, CommandList_struct *c)
+{
+ if (c->err_info->SenseInfo[2] != UNIT_ATTENTION)
+ return 0;
+
+ switch (c->err_info->SenseInfo[12]) {
+ case STATE_CHANGED:
+ printk(KERN_WARNING "cciss%d: a state change "
+ "detected, command retried\n", h->ctlr);
+ return 1;
+ break;
+ case LUN_FAILED:
+ printk(KERN_WARNING "cciss%d: LUN failure "
+ "detected, action required\n", h->ctlr);
+ return 1;
+ break;
+ case REPORT_LUNS_CHANGED:
+ printk(KERN_WARNING "cciss%d: report LUN data "
+ "changed\n", h->ctlr);
+ if (h->rescan_wait)
+ complete(h->rescan_wait);
+ return 1;
+ break;
+ case POWER_OR_RESET:
+ printk(KERN_WARNING "cciss%d: a power on "
+ "or device reset detected\n", h->ctlr);
+ return 1;
+ break;
+ case UNIT_ATTENTION_CLEARED:
+ printk(KERN_WARNING "cciss%d: unit attention "
+ "cleared by another initiator\n", h->ctlr);
+ return 1;
+ break;
+ default:
+ printk(KERN_WARNING "cciss%d: unknown "
+ "unit attention detected\n", h->ctlr);
+ return 1;
+ }
+}
+
/*
* We cannot read the structure directly, for portability we must use
* the io functions.
@@ -3600,6 +3675,7 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
int rc;
int dac, return_code;
InquiryData_struct *inq_buff = NULL;
+ char cciss_scan[14];
if (reset_devices) {
/* Reset the controller with a PCI power-cycle */
@@ -3751,6 +3827,13 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
hba[i]->busy_initializing = 0;
rebuild_lun_table(hba[i], 1);
+ hba[i]->time_for_thread_to_die = 0;
+ snprintf(cciss_scan, 14, "cciss_scan%02d", i);
+ hba[i]->cciss_scan_thread = kthread_run((void *)scan_thread, hba[i],
+ cciss_scan);
+ if (IS_ERR(hba[i]->cciss_scan_thread))
+ return PTR_ERR(hba[i]->cciss_scan_thread);
+
return 1;
clean4:
@@ -3826,6 +3909,7 @@ static void __devexit cciss_remove_one(struct pci_dev *pdev)
printk(KERN_ERR "cciss: Unable to remove device \n");
return;
}
+
tmp_ptr = pci_get_drvdata(pdev);
i = tmp_ptr->ctlr;
if (hba[i] == NULL) {
@@ -3834,6 +3918,9 @@ static void __devexit cciss_remove_one(struct pci_dev *pdev)
return;
}
+ kthread_stop(hba[i]->cciss_scan_thread);
+ complete(hba[i]->rescan_wait);
+
remove_proc_entry(hba[i]->devname, proc_cciss);
unregister_blkdev(hba[i]->major, hba[i]->devname);
diff --git a/drivers/block/cciss.h b/drivers/block/cciss.h
index 15e2b84..703e080 100644
--- a/drivers/block/cciss.h
+++ b/drivers/block/cciss.h
@@ -121,6 +121,8 @@ struct ctlr_info
struct sendcmd_reject_list scsi_rejects;
#endif
unsigned char alive;
+ struct completion *rescan_wait;
+ struct task_struct *cciss_scan_thread;
};
/* Defining the diffent access_menthods */
diff --git a/drivers/block/cciss_cmd.h b/drivers/block/cciss_cmd.h
index 24e22de..40b1b92 100644
--- a/drivers/block/cciss_cmd.h
+++ b/drivers/block/cciss_cmd.h
@@ -25,6 +25,29 @@
#define CMD_TIMEOUT 0x000B
#define CMD_UNABORTABLE 0x000C
+/* Unit Attentions ASC's as defined for the MSA2012sa */
+#define POWER_OR_RESET 0x29
+#define STATE_CHANGED 0x2a
+#define UNIT_ATTENTION_CLEARED 0x2f
+#define LUN_FAILED 0x3e
+#define REPORT_LUNS_CHANGED 0x3f
+
+/* Unit Attentions ASCQ's as defined for the MSA2012sa */
+
+ /* These ASCQ's defined for ASC = POWER_OR_RESET */
+#define POWER_ON_RESET 0x00
+#define POWER_ON_REBOOT 0x01
+#define SCSI_BUS_RESET 0x02
+#define MSA_TARGET_RESET 0x03
+#define CONTROLLER_FAILOVER 0x04
+#define TRANSCEIVER_SE 0x05
+#define TRANSCEIVER_LVD 0x06
+
+ /* These ASCQ's defined for ASC = STATE_CHANGED */
+#define RESERVATION_PREEMPTED 0x03
+#define ASYM_ACCESS_CHANGED 0x06
+#define LUN_CAPACITY_CHANGED 0x09
+
//transfer direction
#define XFER_NONE 0x00
#define XFER_WRITE 0x01
On Fri, 6 Mar 2009 17:29:18 -0600
"Mike Miller (OS Dev)" <[email protected]> wrote:
> On Fri, Mar 06, 2009 at 12:24:27PM -0600, James Bottomley wrote:
> > On Fri, 2009-03-06 at 12:16 -0600, Mike Miller wrote:
> > > Patch 1 of 2
> > >
> > > This is a resubmission of yesterdays patch to detect changes on the MSA2012.
> > > I hope I've addressed all concerns. This patch rearranges some of the code
> > > so we also have coverage in the sg and the ioctl paths as well as the main
> > > data path.
> > >
> > > The MSA2012 cannot inform the driver of configuration changes since all
> > > management is out of band. This is a departure from any storage we have
> > > supported in the past. We need some way to detect changes on the topology so
> > > we implement this kernel thread. In some instances there's nothing we can do
> > > from the driver (like LUN failure) so just print out a message. In the case
> > > where logical volumes are added or deleted we call rebuild_lun_table to
> > > refreash the driver's view of the world.
> > >
> > > Please consider this for inclusion.
> >
> > I still don't quite see how the thread stops on module removal ... there
> > needs to be an explicit kthread_stop() somewhere in the clean up path.
> >
> > James
> >
> >
> This time I make a call to kthread_stop in cciss_remove_one. The driver can
> be unloaded and the thread gets cleaned up.
Please include a complete (and suitably updated) copy of the changelog
with each iteration of a patch.
> KNOWN BUG: it seems the timeout must expire before kthread_stop actually
> stops the thread. This causes the driver to hang and wait during rmmod. I've
> played around with several things but haven't found the correct way to
> address the problem. Looking at other drivers hasn't been much help. Any
> advice is greatly appreciated.
Well, wait_for_completion_timeout() is only going to return when the
timeout timed out, or someone ran complete().
> +static int scan_thread(ctlr_info_t *h)
> +{
> + int rc;
> + DECLARE_COMPLETION_ONSTACK(wait);
> + h->rescan_wait = &wait;
> +
> + while (!kthread_should_stop()) {
> + rc = wait_for_completion_timeout(&wait, 300 * HZ);
> + if (!rc)
> + continue;
> + else
> + rebuild_lun_table(h, 0);
> + }
> + return 0;
> +}
So.. we shouldn't need the timeout here at all - just use
wait_for_completion().
static int scan_thread(ctlr_info_t *h)
{
DECLARE_COMPLETION_ONSTACK(wait);
h->rescan_wait = &wait;
for ( ; ; ) {
wait_for_completion(&wait);
if (kthread_should_stop())
break;
rebuild_lun_table(h, 0);
}
return 0;
}
And on the teardown path, do
complete(...);
kthread_stop(...);
?
On Fri, Mar 06, 2009 at 12:16:03PM -0600, Miller, Mike (OS Dev)?? wrote:
Patch 1 of 2
This is yet another go at the patch to detect changes on the MSA2012.
I hope I've addressed all concerns. This patch rearranges some of the code
so we also have coverage in the sg and the ioctl paths as well as the main
data path.
The MSA2012 cannot inform the driver of configuration changes since all
management is out of band. This is a departure from any storage we have
supported in the past. We need some way to detect changes on the topology so
we implement this kernel thread. In some instances there's nothing we can do
from the driver (like LUN failure) so just print out a message. In the case
where logical volumes are added or deleted we call rebuild_lun_table to
refresh the driver's view of the world.
Changelog:
1. Do the completion(hba[i]->rescan_wait) before calling kthread_stop,
this resolves the issue of waiting for the timeout on rmmod
2. Added a new function called check_ioctl_unit_attention to cover UA's
from the ioctl patch
3. I preserved the wait_for_completion_timeout to avoid call traces
caused by /proc/sys/kernel/hung_task_timeout_secs expiring
4. Moved the call to check_for_unit_attention to evaluate_target_status
since it's already called from complete_command
5. Add retry_cmd as an argument to evaluate_target_status
6. Added *rescan_wait to the controller info struct
Please consider this for inclusion.
Signed-off-by: Mike Miller <[email protected]>
diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c
index d2cb67b..96e20e1 100644
--- a/drivers/block/cciss.c
+++ b/drivers/block/cciss.c
@@ -51,6 +51,7 @@
#include <scsi/scsi_ioctl.h>
#include <linux/cdrom.h>
#include <linux/scatterlist.h>
+#include <linux/kthread.h>
#define CCISS_DRIVER_VERSION(maj,min,submin) ((maj<<16)|(min<<8)|(submin))
#define DRIVER_NAME "HP CISS Driver (v 3.6.20)"
@@ -186,6 +187,8 @@ static int sendcmd_withirq(__u8 cmd, int ctlr, void *buff, size_t size,
__u8 page_code, int cmd_type);
static void fail_all_cmds(unsigned long ctlr);
+static int scan_thread(ctlr_info_t *h);
+static int check_for_unit_attention(ctlr_info_t *h, CommandList_struct *c);
#ifdef CONFIG_PROC_FS
static void cciss_procinit(int i);
@@ -735,6 +738,12 @@ static int cciss_getgeo(struct block_device *bdev, struct hd_geometry *geo)
return 0;
}
+static void check_ioctl_unit_attention(ctlr_info_t *host, CommandList_struct *c)
+{
+ if (c->err_info->CommandStatus == CMD_TARGET_STATUS &&
+ c->err_info->ScsiStatus != SAM_STAT_CHECK_CONDITION)
+ (void)check_for_unit_attention(host, c);
+}
/*
* ioctl
*/
@@ -1029,6 +1038,8 @@ static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
iocommand.buf_size,
PCI_DMA_BIDIRECTIONAL);
+ check_ioctl_unit_attention(host, c);
+
/* Copy the error information out */
iocommand.error_info = *(c->err_info);
if (copy_to_user
@@ -1180,6 +1191,7 @@ static int cciss_ioctl(struct block_device *bdev, fmode_t mode,
(dma_addr_t) temp64.val, buff_size[i],
PCI_DMA_BIDIRECTIONAL);
}
+ check_ioctl_unit_attention(host, c);
/* Copy the error information out */
ioc->error_info = *(c->err_info);
if (copy_to_user(argp, ioc, sizeof(*ioc))) {
@@ -2585,12 +2597,14 @@ static inline unsigned int make_status_bytes(unsigned int scsi_status_byte,
((driver_byte & 0xff) << 24);
}
-static inline int evaluate_target_status(CommandList_struct *cmd)
+static inline int evaluate_target_status(ctlr_info_t *h,
+ CommandList_struct *cmd, int *retry_cmd)
{
unsigned char sense_key;
unsigned char status_byte, msg_byte, host_byte, driver_byte;
int error_value;
+ *retry_cmd = 0;
/* If we get in here, it means we got "target status", that is, scsi status */
status_byte = cmd->err_info->ScsiStatus;
driver_byte = DRIVER_OK;
@@ -2618,6 +2632,11 @@ static inline int evaluate_target_status(CommandList_struct *cmd)
if (((sense_key == 0x0) || (sense_key == 0x1)) && !blk_pc_request(cmd->rq))
error_value = 0;
+ if (check_for_unit_attention(h, cmd)) {
+ *retry_cmd = !blk_pc_request(cmd->rq);
+ return 0;
+ }
+
if (!blk_pc_request(cmd->rq)) { /* Not SG_IO or similar? */
if (error_value != 0)
printk(KERN_WARNING "cciss: cmd %p has CHECK CONDITION"
@@ -2657,7 +2676,7 @@ static inline void complete_command(ctlr_info_t *h, CommandList_struct *cmd,
switch (cmd->err_info->CommandStatus) {
case CMD_TARGET_STATUS:
- rq->errors = evaluate_target_status(cmd);
+ rq->errors = evaluate_target_status(h, cmd, &retry_cmd);
break;
case CMD_DATA_UNDERRUN:
if (blk_fs_request(cmd->rq)) {
@@ -3008,6 +3027,62 @@ static irqreturn_t do_cciss_intr(int irq, void *dev_id)
return IRQ_HANDLED;
}
+static int scan_thread(ctlr_info_t *h)
+{
+ int rc;
+ DECLARE_COMPLETION_ONSTACK(wait);
+ h->rescan_wait = &wait;
+
+ while (!kthread_should_stop()) {
+ rc = wait_for_completion_timeout(&wait, 300 * HZ);
+ if (!rc)
+ continue;
+ else
+ rebuild_lun_table(h, 0);
+ }
+ return 0;
+}
+
+static int check_for_unit_attention(ctlr_info_t *h, CommandList_struct *c)
+{
+ if (c->err_info->SenseInfo[2] != UNIT_ATTENTION)
+ return 0;
+
+ switch (c->err_info->SenseInfo[12]) {
+ case STATE_CHANGED:
+ printk(KERN_WARNING "cciss%d: a state change "
+ "detected, command retried\n", h->ctlr);
+ return 1;
+ break;
+ case LUN_FAILED:
+ printk(KERN_WARNING "cciss%d: LUN failure "
+ "detected, action required\n", h->ctlr);
+ return 1;
+ break;
+ case REPORT_LUNS_CHANGED:
+ printk(KERN_WARNING "cciss%d: report LUN data "
+ "changed\n", h->ctlr);
+ if (h->rescan_wait)
+ complete(h->rescan_wait);
+ return 1;
+ break;
+ case POWER_OR_RESET:
+ printk(KERN_WARNING "cciss%d: a power on "
+ "or device reset detected\n", h->ctlr);
+ return 1;
+ break;
+ case UNIT_ATTENTION_CLEARED:
+ printk(KERN_WARNING "cciss%d: unit attention "
+ "cleared by another initiator\n", h->ctlr);
+ return 1;
+ break;
+ default:
+ printk(KERN_WARNING "cciss%d: unknown "
+ "unit attention detected\n", h->ctlr);
+ return 1;
+ }
+}
+
/*
* We cannot read the structure directly, for portability we must use
* the io functions.
@@ -3600,6 +3675,7 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
int rc;
int dac, return_code;
InquiryData_struct *inq_buff = NULL;
+ char cciss_scan[14];
if (reset_devices) {
/* Reset the controller with a PCI power-cycle */
@@ -3751,6 +3827,12 @@ static int __devinit cciss_init_one(struct pci_dev *pdev,
hba[i]->busy_initializing = 0;
rebuild_lun_table(hba[i], 1);
+ snprintf(cciss_scan, 14, "cciss_scan%02d", i);
+ hba[i]->cciss_scan_thread = kthread_run((void *)scan_thread, hba[i],
+ cciss_scan);
+ if (IS_ERR(hba[i]->cciss_scan_thread))
+ return PTR_ERR(hba[i]->cciss_scan_thread);
+
return 1;
clean4:
@@ -3826,6 +3908,7 @@ static void __devexit cciss_remove_one(struct pci_dev *pdev)
printk(KERN_ERR "cciss: Unable to remove device \n");
return;
}
+
tmp_ptr = pci_get_drvdata(pdev);
i = tmp_ptr->ctlr;
if (hba[i] == NULL) {
@@ -3834,6 +3917,9 @@ static void __devexit cciss_remove_one(struct pci_dev *pdev)
return;
}
+ complete(hba[i]->rescan_wait);
+ kthread_stop(hba[i]->cciss_scan_thread);
+
remove_proc_entry(hba[i]->devname, proc_cciss);
unregister_blkdev(hba[i]->major, hba[i]->devname);
diff --git a/drivers/block/cciss.h b/drivers/block/cciss.h
index 15e2b84..703e080 100644
--- a/drivers/block/cciss.h
+++ b/drivers/block/cciss.h
@@ -121,6 +121,8 @@ struct ctlr_info
struct sendcmd_reject_list scsi_rejects;
#endif
unsigned char alive;
+ struct completion *rescan_wait;
+ struct task_struct *cciss_scan_thread;
};
/* Defining the diffent access_menthods */
diff --git a/drivers/block/cciss_cmd.h b/drivers/block/cciss_cmd.h
index 24e22de..40b1b92 100644
--- a/drivers/block/cciss_cmd.h
+++ b/drivers/block/cciss_cmd.h
@@ -25,6 +25,29 @@
#define CMD_TIMEOUT 0x000B
#define CMD_UNABORTABLE 0x000C
+/* Unit Attentions ASC's as defined for the MSA2012sa */
+#define POWER_OR_RESET 0x29
+#define STATE_CHANGED 0x2a
+#define UNIT_ATTENTION_CLEARED 0x2f
+#define LUN_FAILED 0x3e
+#define REPORT_LUNS_CHANGED 0x3f
+
+/* Unit Attentions ASCQ's as defined for the MSA2012sa */
+
+ /* These ASCQ's defined for ASC = POWER_OR_RESET */
+#define POWER_ON_RESET 0x00
+#define POWER_ON_REBOOT 0x01
+#define SCSI_BUS_RESET 0x02
+#define MSA_TARGET_RESET 0x03
+#define CONTROLLER_FAILOVER 0x04
+#define TRANSCEIVER_SE 0x05
+#define TRANSCEIVER_LVD 0x06
+
+ /* These ASCQ's defined for ASC = STATE_CHANGED */
+#define RESERVATION_PREEMPTED 0x03
+#define ASYM_ACCESS_CHANGED 0x06
+#define LUN_CAPACITY_CHANGED 0x09
+
//transfer direction
#define XFER_NONE 0x00
#define XFER_WRITE 0x01
On Fri, 2009-03-06 at 15:56 -0800, Andrew Morton wrote:
> On Fri, 6 Mar 2009 17:29:18 -0600
> "Mike Miller (OS Dev)" <[email protected]> wrote:
>
> > On Fri, Mar 06, 2009 at 12:24:27PM -0600, James Bottomley wrote:
> > > On Fri, 2009-03-06 at 12:16 -0600, Mike Miller wrote:
> > > > Patch 1 of 2
> > > >
> > > > This is a resubmission of yesterdays patch to detect changes on the MSA2012.
> > > > I hope I've addressed all concerns. This patch rearranges some of the code
> > > > so we also have coverage in the sg and the ioctl paths as well as the main
> > > > data path.
> > > >
> > > > The MSA2012 cannot inform the driver of configuration changes since all
> > > > management is out of band. This is a departure from any storage we have
> > > > supported in the past. We need some way to detect changes on the topology so
> > > > we implement this kernel thread. In some instances there's nothing we can do
> > > > from the driver (like LUN failure) so just print out a message. In the case
> > > > where logical volumes are added or deleted we call rebuild_lun_table to
> > > > refreash the driver's view of the world.
> > > >
> > > > Please consider this for inclusion.
> > >
> > > I still don't quite see how the thread stops on module removal ... there
> > > needs to be an explicit kthread_stop() somewhere in the clean up path.
> > >
> > > James
> > >
> > >
> > This time I make a call to kthread_stop in cciss_remove_one. The driver can
> > be unloaded and the thread gets cleaned up.
>
> Please include a complete (and suitably updated) copy of the changelog
> with each iteration of a patch.
>
>
> > KNOWN BUG: it seems the timeout must expire before kthread_stop actually
> > stops the thread. This causes the driver to hang and wait during rmmod. I've
> > played around with several things but haven't found the correct way to
> > address the problem. Looking at other drivers hasn't been much help. Any
> > advice is greatly appreciated.
>
> Well, wait_for_completion_timeout() is only going to return when the
> timeout timed out, or someone ran complete().
>
> > +static int scan_thread(ctlr_info_t *h)
> > +{
> > + int rc;
> > + DECLARE_COMPLETION_ONSTACK(wait);
> > + h->rescan_wait = &wait;
> > +
> > + while (!kthread_should_stop()) {
> > + rc = wait_for_completion_timeout(&wait, 300 * HZ);
> > + if (!rc)
> > + continue;
> > + else
> > + rebuild_lun_table(h, 0);
> > + }
> > + return 0;
> > +}
>
> So.. we shouldn't need the timeout here at all - just use
> wait_for_completion().
>
> static int scan_thread(ctlr_info_t *h)
> {
> DECLARE_COMPLETION_ONSTACK(wait);
>
> h->rescan_wait = &wait;
> for ( ; ; ) {
> wait_for_completion(&wait);
> if (kthread_should_stop())
> break;
> rebuild_lun_table(h, 0);
> }
> return 0;
> }
>
> And on the teardown path, do
>
> complete(...);
> kthread_stop(...);
This is racy ... although I think the race would only show in a pre-empt
kernel: complete causes the thread to run immediately pre-empting us.
Now it runs around the loop, through kthread_should_stop() and back to
wait_for_completion() before we get a chance to run kthread_stop().
The only way to avoid this seems to be to use wait queues and wake up
(kthread_stop does an automatic wake_up of the process, which is ignored
by completions).
James