Skip to contentSkip to conversation

ide: fix drive side 80c cable check

PATCH NAK Tejun Heo 4 participants 7 messages Feb 5, 2007 – Jul 12, 2007 1 message not in the archive libata

lore
Share & tools
lore

Tejun Heo

author
#0 lore raw

PATCH 1/1 ide: fix drive side 80c cable check

The 80c wire bit is bit 13, not 14. This increases the chance of incorrect wire detection especially because host side cable detection is often unreliable and we sometimes soley depend on drive side cable detection. Fix it.

---
Please consider for -stable.

 drivers/ide/ide-iops.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/ide/ide-iops.c +1 -1
Index: work/drivers/ide/ide-iops.c
===================================================================
@@ -604,7 +604,7 @@ u8 eighty_ninty_three (ide_drive_t *driv
 	if (!(drive->id->hw_config & 0x6000))
 		return 0;
 #ifndef CONFIG_IDEDMA_IVB
-	if(!(drive->id->hw_config & 0x4000))
+	if (!(drive->id->hw_config & 0x2000))
 		return 0;
 #endif /* CONFIG_IDEDMA_IVB */
 	return 1;
NAK
#1 lore raw
Quoted 17 lines · On Mon, 5 Feb 2007 16:47:13 +0900 Tejun Heo <[email protected]> wrote:

The 80c wire bit is bit 13, not 14. This increases the chance of incorrect wire detection especially because host side cable detection is often unreliable and we sometimes soley depend on drive side cable detection. Fix it.

---
Please consider for -stable.

 drivers/ide/ide-iops.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: work/drivers/ide/ide-iops.c
===================================================================

drivers/ide/ide-iops.c u8 eighty_ninty_three (ide_drive_t *driv · line 609

Quoted 7 lines
drivers/ide/ide-iops.c +1 -1
@@ -604,7 +604,7 @@ u8 eighty_ninty_three (ide_drive_t *driv
 	if (!(drive->id->hw_config & 0x6000))
 		return 0;
 #ifndef CONFIG_IDEDMA_IVB
-	if(!(drive->id->hw_config & 0x4000))
+	if (!(drive->id->hw_config & 0x2000))
 		return 0;
 #endif /* CONFIG_IDEDMA_IVB */

NAK

While the old code is a mess, your changes don't fix it. The code above is correct before you touch it as far as I can tell. Incomplete but correct as far as it went.

The logic in the function as far as it goes is correct

If neither valid bit nor 80pin bit set -> 40pin [00]
If checking valid bit && valid bit clear -> 40pin [0x]

The 0x2000 test is needed as an additional test (as per the
ide_ata66_check function directly below) [10] v [11]

Without this an id value of 0x2000 will trigger 80pin but is not valid.

(the IDEDMA_IVB check is more relaxted to handle some confused ATA4 drives, and is probably something we don't want in libata anyway, or should blacklist the afflicted for this)

Alan

author
#2 lore raw

PATCH 1/1 ide: fix drive side 80c cable check, take 2

eighty_ninty_three() had word 93 validitity check but not the 80c bit test itself (bit 12). This increases the chance of incorrect wire detection especially because host side cable detection is often unreliable and we sometimes soley depend on drive side cable detection. Fix it.

---
Ah... thanks. That explains the code much better. Fixed accordingly.

drivers/ide/ide-iops.c +2 -0
index badde63..6558055 100644
@@ -607,6 +607,8 @@ u8 eighty_ninty_three (ide_drive_t *drive)
 	if(!(drive->id->hw_config & 0x4000))
 		return 0;
 #endif /* CONFIG_IDEDMA_IVB */
+	if (!(drive->id->hw_config & 0x2000))
+		return 0;
 	return 1;
 }
 
Acked-by Alan Cox on v1 1/1
#3 lore raw

Re: [PATCH] ide: fix drive side 80c cable check, take 2

Quoted 15 lines · On Mon, 5 Feb 2007 21:47:13 +0900 Tejun Heo <[email protected]> wrote:

eighty_ninty_three() had word 93 validitity check but not the 80c bit test itself (bit 12). This increases the chance of incorrect wire detection especially because host side cable detection is often unreliable and we sometimes soley depend on drive side cable detection. Fix it.

---
Ah... thanks. That explains the code much better. Fixed accordingly.

drivers/ide/ide-iops.c +0 -0
index badde63..6558055 100644

drivers/ide/ide-iops.c u8 eighty_ninty_three (ide_drive_t *drive) · line 612

Quoted 6 lines
drivers/ide/ide-iops.c +2 -0
@@ -607,6 +607,8 @@ u8 eighty_ninty_three (ide_drive_t *drive)
 	if(!(drive->id->hw_config & 0x4000))
 		return 0;
 #endif /* CONFIG_IDEDMA_IVB */
+	if (!(drive->id->hw_config & 0x2000))
+		return 0;
 	return 1;

A message that is not in the archive (sent off-list or not yet received): [email protected]

#5 lore raw

Re: [PATCH] ide: fix drive side 80c cable check, take 2

Quoted 13 lines · Alan
Quoted 10 lines · On Mon, 5 Feb 2007 21:47:13 +0900 Tejun Heo <[email protected]> wrote:

eighty_ninty_three() had word 93 validitity check but not the 80c bit test itself (bit 12). This increases the chance of incorrect wire detection especially because host side cable detection is often unreliable and we sometimes soley depend on drive side cable detection. Fix it.

---
Ah... thanks. That explains the code much better. Fixed accordingly.

Quoted 13 lines

drivers/ide/ide-iops.c u8 eighty_ninty_three (ide_drive_t *drive) · line 612

Quoted 11 lines
drivers/ide/ide-iops.c +2 -0
index badde63..6558055 100644
@@ -607,6 +607,8 @@ u8 eighty_ninty_three (ide_drive_t *drive)
      if(!(drive->id->hw_config & 0x4000))
              return 0;
 #endif /* CONFIG_IDEDMA_IVB */
+     if (!(drive->id->hw_config & 0x2000))
+             return 0;
      return 1;

applied

#6 lore raw

Re: [PATCH] ide: fix drive side 80c cable check, take 2

In reply to Tejun Heo

Hello.

Quoted 13 lines · Tejun Heo

eighty_ninty_three() had word 93 validitity check but not the 80c bit test itself (bit 12). This increases the chance of incorrect wire detection especially because host side cable detection is often unreliable and we sometimes soley depend on drive side cable detection. Fix it.

drivers/ide/ide-iops.c +0 -0
index badde63..6558055 100644

drivers/ide/ide-iops.c u8 eighty_ninty_three (ide_drive_t *drive) · line 611

Quoted 5 lines
drivers/ide/ide-iops.c +2 -0
@@ -607,6 +607,8 @@ u8 eighty_ninty_three (ide_drive_t *drive)
 	if(!(drive->id->hw_config & 0x4000))
 		return 0;
 #endif /* CONFIG_IDEDMA_IVB */
+	if (!(drive->id->hw_config & 0x2000))
+		return 0;

Haha, you know just *why* this was wrong? Bit 13 of the word 93 when *set*
means 40c cable, not 80c! Look at the table 9 in ATA/PI-6, for example, and then into the bit description in the table 27.

drivers/ide/ide-iops.c u8 eighty_ninty_three (ide_drive_t *drive) · line 613

Quoted 2 lines
drivers/ide/ide-iops.c +0 -0
 	return 1;
 }

MBR, Sergei

#7 lore raw

Re: [PATCH] ide: fix drive side 80c cable check, take 2

Quoted 8 lines · Hello, I wrote:
Quoted 7 lines

eighty_ninty_three() had word 93 validitity check but not the 80c bit test itself (bit 12). This increases the chance of incorrect wire detection especially because host side cable detection is often unreliable and we sometimes soley depend on drive side cable detection. Fix it.

Quoted 13 lines

drivers/ide/ide-iops.c u8 eighty_ninty_three (ide_drive_t *drive) · line 611

Quoted 10 lines
drivers/ide/ide-iops.c +2 -0
index badde63..6558055 100644
@@ -607,6 +607,8 @@ u8 eighty_ninty_three (ide_drive_t *drive)
     if(!(drive->id->hw_config & 0x4000))
         return 0;
 #endif /* CONFIG_IDEDMA_IVB */
+    if (!(drive->id->hw_config & 0x2000))
+        return 0;

Haha, you know just *why* this was wrong? Bit 13 of the word 93 when
*set* means 40c cable, not 80c! Look at the table 9 in ATA/PI-6, for example, and then into the bit description in the table 27.

Oops, that's a host side detection, the device side CBLID- seems to have
the opposite levels. Should have really be looking at the table 8. :-<

Quoted 2 lines

drivers/ide/ide-iops.c u8 eighty_ninty_three (ide_drive_t *drive) · line 613

Quoted 2 lines
drivers/ide/ide-iops.c +0 -0
     return 1;
 }

Actually, trying to find out why 'hdparm -i' clips the modes to udma2
sometimes...

MBR, Sergei