ata: pata_platform: Disable prereset logic.

PATCH In review Paul Mundt 2 participants 6 messages May 23, 2007 libata

lore

Share & tools

lore

Paul Mundt

author

2007-05-23 06:45 UTC #0 lore raw

PATCH 1/1 ata: pata_platform: Disable prereset logic.

On a number of boards the current prereset logic seems to misbehave:

scsi0 : pata_platform
ata1: PATA max PIO0 cmd 0xb06001f0 ctl 0xb06003f6 bmdma 0x00000000 irq 0
ata1: device not ready (errno=-19), forcing hardreset
ata1: BUG: prereset() requested invalid reset type

This triggers when there is no card inserted in the slot.

Simply disabling the prereset gets rid of this, and doesn't seem to cause any problems for either PCMCIA or CF cards when they're actually present.

Signed-off-by Paul Mundt <[email protected]>

 drivers/ata/pata_platform.c |   15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)

drivers/ata/pata_platform.c +11 -4

index cbb7866..8a569c7 100644
@@ -1,7 +1,7 @@
 /*
  * Generic platform device PATA driver
  *
- * Copyright (C) 2006  Paul Mundt
+ * Copyright (C) 2006 - 2007  Paul Mundt
  *
  * Based on pata_pcmcia:
  *

@@ -22,7 +22,7 @@
 #include <linux/pata_platform.h>
 
 #define DRV_NAME "pata_platform"
-#define DRV_VERSION "1.0"
+#define DRV_VERSION "1.1"
 
 static int pio_mask = 1;

@@ -30,7 +30,8 @@ static int pio_mask = 1;
  * Provide our own set_mode() as we don't want to change anything that has
  * already been configured..
  */
-static int pata_platform_set_mode(struct ata_port *ap, struct ata_device **unused)
+static int pata_platform_set_mode(struct ata_port *ap,
+				  struct ata_device **unused)
 {
 	int i;

@@ -48,6 +49,12 @@ static int pata_platform_set_mode(struct ata_port *ap, struct ata_device **unuse
 	return 0;
 }
 
+static void pata_platform_error_handler(struct ata_port *ap)
+{
+	ata_bmdma_drive_eh(ap, NULL, ata_std_softreset, NULL,
+			ata_std_postreset);
+}
+
 static int ata_dummy_ret0(struct ata_port *ap)	{ return 0; }
 
 static struct scsi_host_template pata_platform_sht = {

@@ -80,7 +87,7 @@ static struct ata_port_operations pata_platform_port_ops = {
 
 	.freeze			= ata_bmdma_freeze,
 	.thaw			= ata_bmdma_thaw,
-	.error_handler		= ata_bmdma_error_handler,
+	.error_handler		= pata_platform_error_handler,
 	.post_internal_cmd	= ata_bmdma_post_internal_cmd,
 	.cable_detect		= ata_cable_unknown,

Tejun Heo

2007-05-23 08:07 UTC #1 lore raw

Quoted 11 lines · Paul Mundt

On a number of boards the current prereset logic seems to misbehave:
scsi0 : pata_platform
ata1: PATA max PIO0 cmd 0xb06001f0 ctl 0xb06003f6 bmdma 0x00000000 irq 0
ata1: device not ready (errno=-19), forcing hardreset
ata1: BUG: prereset() requested invalid reset type
This triggers when there is no card inserted in the slot.
Simply disabling the prereset gets rid of this, and doesn't seem to cause any problems for either PCMCIA or CF cards when they're actually present.

NACK. The BUG printking needs fixing but you can't just kill prereset(). Did it work properly on 2.6.21.1? Can you modify ata_wait_ready() such that it prints out the status value while waiting?

Signature · 1 line

tejun

Paul Mundt

author

2007-05-23 08:31 UTC #2 lore raw

Quoted 8 lines · Tejun Heo

Quoted 7 lines · Paul Mundt
On a number of boards the current prereset logic seems to misbehave:
scsi0 : pata_platform
ata1: PATA max PIO0 cmd 0xb06001f0 ctl 0xb06003f6 bmdma 0x00000000 irq 0
ata1: device not ready (errno=-19), forcing hardreset
ata1: BUG: prereset() requested invalid reset type

Quoted 9 lines

Quoted 4 lines
This triggers when there is no card inserted in the slot.
Simply disabling the prereset gets rid of this, and doesn't seem to cause any problems for either PCMCIA or CF cards when they're actually present.
NACK. The BUG printking needs fixing but you can't just kill prereset(). Did it work properly on 2.6.21.1? Can you modify ata_wait_ready() such that it prints out the status value while waiting?

ata_wait_ready() works fine, it reports 0xff immediately (and this is what I would expect when there is no card inserted).

The board that exhibits this behaviour wasn't supported in the older kernels, I can backport and test if it will be useful, though.

Tejun Heo

2007-05-23 09:30 UTC #3 lore raw

Quoted 11 lines · Paul Mundt

Quoted 10 lines · Tejun Heo
Quoted 9 lines · Paul Mundt
On a number of boards the current prereset logic seems to misbehave:
scsi0 : pata_platform
ata1: PATA max PIO0 cmd 0xb06001f0 ctl 0xb06003f6 bmdma 0x00000000 irq 0
ata1: device not ready (errno=-19), forcing hardreset
ata1: BUG: prereset() requested invalid reset type
This triggers when there is no card inserted in the slot.

Quoted 8 lines

Quoted 6 lines
Quoted 2 lines
Simply disabling the prereset gets rid of this, and doesn't seem to cause any problems for either PCMCIA or CF cards when they're actually present.
NACK. The BUG printking needs fixing but you can't just kill prereset(). Did it work properly on 2.6.21.1? Can you modify ata_wait_ready() such that it prints out the status value while waiting?
ata_wait_ready() works fine, it reports 0xff immediately (and this is what I would expect when there is no card inserted).

Does the attached patch fix your problem?

Signature · 1 line

tejun

drivers/ata/libata-core.c +1 -1

index a6de57e..79f2175 100644
@@ -3368,7 +3368,7 @@ int ata_std_prereset(struct ata_port *ap
 	 */
 	if (!(ap->flags & ATA_FLAG_SKIP_D2H_BSY) && !ata_port_offline(ap)) {
 		rc = ata_wait_ready(ap, deadline);
-		if (rc) {
+		if (rc && rc != -ENODEV) {
 			ata_port_printk(ap, KERN_WARNING, "device not ready "
 					"(errno=%d), forcing hardreset\n", rc);
 			ehc->i.action |= ATA_EH_HARDRESET;

Paul Mundt

author

2007-05-23 09:49 UTC #4 lore raw

Quoted 10 lines · Tejun Heo

Quoted 9 lines · Paul Mundt
Quoted 8 lines · Tejun Heo
Quoted 7 lines · Paul Mundt
On a number of boards the current prereset logic seems to misbehave:
scsi0 : pata_platform
ata1: PATA max PIO0 cmd 0xb06001f0 ctl 0xb06003f6 bmdma 0x00000000 irq 0
ata1: device not ready (errno=-19), forcing hardreset
ata1: BUG: prereset() requested invalid reset type

Quoted 7 lines

Quoted 4 lines
Quoted 2 lines
Quoted 2 lines
This triggers when there is no card inserted in the slot.
ata_wait_ready() works fine, it reports 0xff immediately (and this is what I would expect when there is no card inserted).
Does the attached patch fix your problem?

Yes, that works fine, thanks.

Tejun Heo

2007-05-23 09:59 UTC #5 lore raw

PATCH 1/1 libata: -ENODEV during prereset isn't an error

During prereset, -ENODEV return from ata_wait_ready() is not an error. This causes unnecessary bug message on controllers which uses 0xff to indicate empty port. Fix it.

Signed-off-by Tejun Heo <[email protected]>

Cc Paul Mundt <[email protected]>

---

This one is for 2.6.22 too. Thanks.

 libata-core.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

drivers/ata/libata-core.c +1 -1

index a6de57e..79f2175 100644
@@ -3368,7 +3368,7 @@ int ata_std_prereset(struct ata_port *ap
 	 */
 	if (!(ap->flags & ATA_FLAG_SKIP_D2H_BSY) && !ata_port_offline(ap)) {
 		rc = ata_wait_ready(ap, deadline);
-		if (rc) {
+		if (rc && rc != -ENODEV) {
 			ata_port_printk(ap, KERN_WARNING, "device not ready "
 					"(errno=%d), forcing hardreset\n", rc);
 			ehc->i.action |= ATA_EH_HARDRESET;