Skip to contentSkip to conversation

USB: isp1760: Soften DW3 X/transaction error bit handling

PATCH In review Anton Vorontsov 5 participants 5 messages May 6, 2010 – May 7, 2010 USB

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 USB: isp1760: Soften DW3 X/transaction error bit handling

There were some reports[1] of isp1760 USB driver malfunctioning with high speed devices, noticed on Blackfin and PowerPC targets. These reports indicated that the original Philips 'pehcd'[2] driver worked fine.

We've noticed the same issue with an ARM RealView platform. This happens under load (with only some mass storage devices, not all, just as in another report[3]):

error bit is set in DW3
error bit is set in DW3
error bit is set in DW3
usb 1-1.2: device descriptor read/64, error -32

It appears that the 'pehcd' driver checks the X bit only if the transaction is halted, otherwise the error is so far insignificant.

I didn't find where exactly ISP1760 spec mandates 'H && X' handling (maybe it's in the EHCI spec?), but the approach that is used in the original driver (and in the EHCI driver, FWIW) fixes the issue.

[1] http://markmail.org/message/lx4qrlbrs2uhcnly
[2] svn co svn://sources.blackfin.uclinux.org/linux-kernel/trunk/drivers/usb/host -r 5494 See pehci.c:pehci_hcd_update_error_status().
[3] http://blackfin.uclinux.org/gf/project/uclinux-dist/tracker/?action=TrackerItemEdit&tracker_item_id=5148

---

Note that this isn't related to the recent cache issues on ARM platforms, as the kernel I'm running has commit 3f54ccf38eb51 'isp1760: Flush the D-cache for the pipe-in transfer buffers'.

 drivers/usb/host/isp1760-hcd.c |    7 +++----
 1 files changed, 3 insertions(+), 4 deletions(-)
drivers/usb/host/isp1760-hcd.c +3 -4
index 9f01293..9989c0b 100644
@@ -713,12 +713,11 @@ static int check_error(struct ptd *ptd)
 	u32 dw3;
 
 	dw3 = le32_to_cpu(ptd->dw3);
-	if (dw3 & DW3_HALT_BIT)
+	if (dw3 & DW3_HALT_BIT) {
 		error = -EPIPE;
 
-	if (dw3 & DW3_ERROR_BIT) {
-		printk(KERN_ERR "error bit is set in DW3\n");
-		error = -EPIPE;
+		if (dw3 & DW3_ERROR_BIT)
+			pr_err("error bit is set in DW3\n");
 	}
 
 	if (dw3 & DW3_QTD_ACTIVE) {
Signature · 1 line
1.7.0.5
#2 lore raw
In reply to Anton Vorontsov
Quoted 6 lines · Anton Vorontsov

It appears that the 'pehcd' driver checks the X bit only if the transaction is halted, otherwise the error is so far insignificant.

I didn't find where exactly ISP1760 spec mandates 'H && X' handling (maybe it's in the EHCI spec?),

Yes, it is described implicitly in the EHCI spec, section 4.10.3. Transaction errors cause the status field to be updated to reflect the type of error, but the transaction continues to be retried until the Active bit is set to 0. When the error counter reaches 0, the Halt bit is set and the Active bit is cleared.

Alan Stern

#3 lore raw

PATCH v2 1/1 USB: isp1760: Soften DW3 X/transaction error bit handling

There were some reports[1] of isp1760 USB driver malfunctioning with high speed devices, noticed on Blackfin and PowerPC targets. These reports indicated that the original Philips 'pehcd'[2] driver worked fine.

We've noticed the same issue with an ARM RealView platform. This happens under load (with only some mass storage devices, not all, just as in another report[3]):

error bit is set in DW3
error bit is set in DW3
error bit is set in DW3
usb 1-1.2: device descriptor read/64, error -32

It appears that the 'pehcd' driver checks the X bit only if the transaction is halted (H bit), otherwise the error is so far insignificant.

The ISP176x chips were modeled after EHCI, and EHCI spec says (thanks to Alan Stern for pointing out):

"Transaction errors cause the status field to be updated to reflect
the type of error, but the transaction continues to be retried until the Active bit is set to 0. When the error counter reaches 0, the Halt bit is set and the Active bit is cleared."

So, just as the original Philips driver, isp1760 must report the error only if the transaction error and the halt bits are set.

[1] http://markmail.org/message/lx4qrlbrs2uhcnly
[2] svn co svn://sources.blackfin.uclinux.org/linux-kernel/trunk/drivers/usb/host -r 5494 See pehci.c:pehci_hcd_update_error_status().
[3] http://blackfin.uclinux.org/gf/tracker/5148

---

On Thu, May 06, 2010 at 03:53:54PM -0400, Alan Stern wrote:
[...]

Quoted 4 lines
Quoted 2 lines

I didn't find where exactly ISP1760 spec mandates 'H && X' handling (maybe it's in the EHCI spec?),

Yes, it is described implicitly in the EHCI spec, section 4.10.3.

Thanks! I've added this to the commit message.

Quoted 5 lines · Mike Frysinger
Quoted 1 line · Anton Vorontsov

[3] http://blackfin.uclinux.org/gf/project/uclinux-dist/tracker/?action=TrackerItemEdit&tracker_item_id=5148

i doubt you feel like resending, but this is a better link:
http://blackfin.uclinux.org/gf/tracker/5148

Yeah. I've changed the link, thanks.

 drivers/usb/host/isp1760-hcd.c |    7 +++----
 1 files changed, 3 insertions(+), 4 deletions(-)
drivers/usb/host/isp1760-hcd.c +3 -4
index 9f01293..9989c0b 100644
@@ -713,12 +713,11 @@ static int check_error(struct ptd *ptd)
 	u32 dw3;
 
 	dw3 = le32_to_cpu(ptd->dw3);
-	if (dw3 & DW3_HALT_BIT)
+	if (dw3 & DW3_HALT_BIT) {
 		error = -EPIPE;
 
-	if (dw3 & DW3_ERROR_BIT) {
-		printk(KERN_ERR "error bit is set in DW3\n");
-		error = -EPIPE;
+		if (dw3 & DW3_ERROR_BIT)
+			pr_err("error bit is set in DW3\n");
 	}
 
 	if (dw3 & DW3_QTD_ACTIVE) {
Signature · 1 line
1.7.0.5