Skip to contentSkip to conversation

usb EHCI failing with 'EHCI fail timeout STD_ASS reset'

U-Boot New Dan Lykowski 4 participants 8 messages Apr 8, 2010 – Oct 21, 2010

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hi everyone,

I've seen this question posted a few times in the past with no answer. Since I was having the same problem, I thought I would share the solution I came up with.
I'll try to provide a patch if I can get my mess cleaned up.

After some digging, I was able to see the EHCI HC getting a Master Abort over the PCI bus.
It turns out that some.. all??.. PCI EHCI controllers have 64-bits for the buffer registers in the qTD structure.
See Appendix B in the EHCI specification.

This is what I changed the qTD structure to.

/* Queue Element Transfer Descriptor (qTD). */

struct qTD {
    uint32_t qt_next;
    uint32_t qt_altnext;
    uint32_t qt_token;
    uint32_t qt_buffer[5];
    uint32_t qt_buffer_hi[5];
    uint32_t unused[3];
};

Adding the qt_buffer_hi array, realigning the buffers on 32-byte boundries, and filling qt_buffer_hi with zeros when qt_buffer is being filled fixes the problem.

Thanks,
Dan Lykowski

#1 lore raw

PATCH 1/1 USB: sync Queue Element Transfer Descriptor against EHCI spec

Appendix B "EHCI 64-Bit Data Structures" of the "Enhanced Host Controller Interface Specification for Universal Serial Bus" (Rev. 1.0, March 12, 2002) defines additional fields which were missing in U-Boot's struct qTD; as these are also present in recent versions of struct ehci_qtd in the Linux kernel, we add them here, too.

This fixes some nasty memory corruption problems.

See http://permalink.gmane.org/gmane.comp.boot-loaders.u-boot/76942

---

Dear Remy,

this patch goes back to an old posting by Dan Lykowski.

Unfortunately Dan never submitted a patch, so this went unfixed until now, when the bug bit Stefano...

Best regards,

Wolfgang

 drivers/usb/host/ehci-hcd.c |    1 +
 drivers/usb/host/ehci.h     |   14 +++++++++-----
 2 files changed, 10 insertions(+), 5 deletions(-)
drivers/usb/host/ehci-hcd.c +1 -0
index 37d056e..f44fc4e 100644
@@ -288,6 +288,7 @@ static int ehci_td_buffer(struct qTD *td, void *buf, size_t sz)
 	idx = 0;
 	while (idx < 5) {
 		td->qt_buffer[idx] = cpu_to_hc32(addr);
+		td->qt_buffer_hi[idx] = 0;
 		next = (addr + 4096) & ~4095;
 		delta = next - addr;
 		if (delta >= sz)
drivers/usb/host/ehci.h +9 -5
index 6fae8ba..d3aa55b 100644
@@ -166,12 +166,16 @@ struct usb_linux_config_descriptor {
 
 /* Queue Element Transfer Descriptor (qTD). */
 struct qTD {
-	uint32_t qt_next;
+	/* this part defined by EHCI spec */
+	uint32_t qt_next;		/* see EHCI 3.5.1 */
 #define	QT_NEXT_TERMINATE	1
-	uint32_t qt_altnext;
-	uint32_t qt_token;
-	uint32_t qt_buffer[5];
-};
+	uint32_t qt_altnext;		/* see EHCI 3.5.2 */
+	uint32_t qt_token;		/* see EHCI 3.5.3 */
+	uint32_t qt_buffer[5];		/* see EHCI 3.5.4 */
+	uint32_t qt_buffer_hi[5];	/* Appendix B */
+	/* pad struct for 32 byte alignment */
+	uint32_t unused[3];
+} __attribute__ ((aligned (32)));
 
 /* Queue Head (QH). */
 struct QH {
Signature · 1 line
1.7.2.3
#2 lore raw

PATCH 1/1 USB: sync Queue Element Transfer Descriptor against EHCI spec

Quoted 25 lines · In message <[email protected]> I wrote:

Appendix B "EHCI 64-Bit Data Structures" of the "Enhanced Host Controller Interface Specification for Universal Serial Bus" (Rev. 1.0, March 12, 2002) defines additional fields which were missing in U-Boot's struct qTD; as these are also present in recent versions of struct ehci_qtd in the Linux kernel, we add them here, too.

This fixes some nasty memory corruption problems.

See http://permalink.gmane.org/gmane.comp.boot-loaders.u-boot/76942

---

Dear Remy,

this patch goes back to an old posting by Dan Lykowski.

Unfortunately Dan never submitted a patch, so this went unfixed until now, when the bug bit Stefano...

Quoted 8 lines

Best regards,

Wolfgang

 drivers/usb/host/ehci-hcd.c |    1 +
 drivers/usb/host/ehci.h     |   14 +++++++++-----
 2 files changed, 10 insertions(+), 5 deletions(-)

Applied directly.

Remy, I hope this is OK with you.

Best regards,

Wolfgang Denk

Signature · 4 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
I must follow the people.  Am I not their leader? - Benjamin Disraeli
#3 lore raw

[U-Boot] Data bus error

Did you have the data bus error with CN5010-based board at porting u-boot?

How to debug? Any suggestions on this?

Thanks!
Shuyou

#4 lore raw

PATCH 1/1 USB: sync Queue Element Transfer Descriptor against EHCI spec

In reply to Wolfgang Denk

Hi,

2010/10/20 Wolfgang Denk <[email protected]>:

Quoted 19 lines
Quoted 8 lines · In message <[email protected]> I wrote:

Appendix B "EHCI 64-Bit Data Structures" of the "Enhanced Host Controller Interface Specification for Universal Serial Bus" (Rev. 1.0, March 12, 2002) defines additional fields which were missing in U-Boot's struct qTD; as these are also present in recent versions of struct ehci_qtd in the Linux kernel, we add them here, too.

This fixes some nasty memory corruption problems.

Quoted 8 lines

See http://permalink.gmane.org/gmane.comp.boot-loaders.u-boot/76942

Quoted 3 lines

Applied directly.

Remy, I hope this is OK with you.

You seem to be in a hurry...
Patch is OK...

Remy

#5 lore raw

PATCH 1/1 USB: sync Queue Element Transfer Descriptor against EHCI spec

Dear Remy Bohmer,

Quoted 4 lines · In message <[email protected]> you…
Quoted 2 lines

Remy, I hope this is OK with you.

You seem to be in a hurry...

I thought we could be close to a -rc1. Alas, I was wrong once more.

Quoted 1 line

Patch is OK...

Unfortunately it seems it ain't so :-( Patch follows...

Best regards,

Wolfgang Denk

Signature · 4 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
"Free markets select for winning solutions."        - Eric S. Raymond
#6 lore raw

PATCH 1/1 USB: fix Queue Element Transfer Descriptor changes

In reply to Dan Lykowski

Commit 3ed1607 "USB: sync Queue Element Transfer Descriptor against EHCI spec" added an "__attribute__ ((aligned (32)))" to the declaration of struct qTD, as used for example in the Linux kernel as well.

However, it turns out that this attribute causes errors in "usb start" (like "ERROR: NOT USB_CONFIG_DESC 7b" and similar). Drop the attribute again.

Cc Dan Lykowski <[email protected]>
Cc Remy Bohmer <[email protected]>
Cc Stefano Babic <[email protected]>
---
 drivers/usb/host/ehci.h |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
drivers/usb/host/ehci.h +1 -1
index d3aa55b..945ab64 100644
@@ -175,7 +175,7 @@ struct qTD {
 	uint32_t qt_buffer_hi[5];	/* Appendix B */
 	/* pad struct for 32 byte alignment */
 	uint32_t unused[3];
-} __attribute__ ((aligned (32)));
+};
 
 /* Queue Head (QH). */
 struct QH {
Signature · 1 line
1.7.2.3
#7 lore raw

PATCH 1/1 USB: fix Queue Element Transfer Descriptor changes

Hi,

2010/10/20 Wolfgang Denk <[email protected]>:

Quoted 9 lines

Commit 3ed1607 "USB: sync Queue Element Transfer Descriptor against EHCI spec" added an "__attribute__ ((aligned (32)))" to the declaration of struct qTD, as used for example in the Linux kernel as well.

However, it turns out that this attribute causes errors in "usb start" (like "ERROR: NOT USB_CONFIG_DESC 7b" and similar). Drop the attribute again.

Quoted 5 lines
Cc Dan Lykowski <[email protected]>
Cc Remy Bohmer <[email protected]>
Cc Stefano Babic <[email protected]>

---

Applied to u-boot-usb
Thanks.

Remy