* [U-Boot] [PATCH] Lower USB_MAX_XFER_BLK to 32767 by default for trimslice
@ 2016-05-26 20:55 Nicolas Chauvet
2016-05-26 21:17 ` Marek Vasut
2016-05-28 8:06 ` [U-Boot] [PATCH] Lower USB_MAX_XFER_BLK to 32767 " Nicolas Chauvet
0 siblings, 2 replies; 4+ messages in thread
From: Nicolas Chauvet @ 2016-05-26 20:55 UTC (permalink / raw)
To: u-boot
I'm experiencing an issue when loading a generic distro initramfs
on trimslice-pro (with ssd using ehci) with the default value of
65535 for USB_MAX_XFER_BLK
EHCI timed out on TD - token=0x80008d80
As adviced by Marek on IRC, using a lower value for
USB_MAX_XFER_BLK allows to load the file
Tested values on tegra20-trimslice:
4095 ok
32767 ok
32768 timeout
36864 timeout
40960 timeout
49150 timeout
Tested on current u-boot master
Signed-off-by: Nicolas Chauvet <kwizart@gmail.com>
---
common/usb_storage.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/common/usb_storage.c b/common/usb_storage.c
index 7e6e52d..c69a7f7 100644
--- a/common/usb_storage.c
+++ b/common/usb_storage.c
@@ -106,7 +106,7 @@ struct us_data {
* enough free heap space left, but the SCSI READ(10) and WRITE(10) commands are
* limited to 65535 blocks.
*/
-#define USB_MAX_XFER_BLK 65535
+#define USB_MAX_XFER_BLK 32767
#else
#define USB_MAX_XFER_BLK 20
#endif
--
2.4.11
^ permalink raw reply related [flat|nested] 4+ messages in thread* [U-Boot] [PATCH] Lower USB_MAX_XFER_BLK to 32767 by default for trimslice
2016-05-26 20:55 [U-Boot] [PATCH] Lower USB_MAX_XFER_BLK to 32767 by default for trimslice Nicolas Chauvet
@ 2016-05-26 21:17 ` Marek Vasut
2016-05-28 8:06 ` [U-Boot] [PATCH] Lower USB_MAX_XFER_BLK to 32767 " Nicolas Chauvet
1 sibling, 0 replies; 4+ messages in thread
From: Marek Vasut @ 2016-05-26 21:17 UTC (permalink / raw)
To: u-boot
On 05/26/2016 10:55 PM, Nicolas Chauvet wrote:
> I'm experiencing an issue when loading a generic distro initramfs
> on trimslice-pro (with ssd using ehci) with the default value of
> 65535 for USB_MAX_XFER_BLK
> EHCI timed out on TD - token=0x80008d80
>
> As adviced by Marek on IRC, using a lower value for
> USB_MAX_XFER_BLK allows to load the file
>
> Tested values on tegra20-trimslice:
> 4095 ok
> 32767 ok
> 32768 timeout
> 36864 timeout
> 40960 timeout
> 49150 timeout
>
> Tested on current u-boot master
I don't really like adding such patch which affects many devices just to
fix one particular device which is crappy. If you could add some
quirk table into u-boot instead and lower this number only for those
buggy devices, that'd be much better.
> Signed-off-by: Nicolas Chauvet <kwizart@gmail.com>
> ---
> common/usb_storage.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/common/usb_storage.c b/common/usb_storage.c
> index 7e6e52d..c69a7f7 100644
> --- a/common/usb_storage.c
> +++ b/common/usb_storage.c
> @@ -106,7 +106,7 @@ struct us_data {
> * enough free heap space left, but the SCSI READ(10) and WRITE(10) commands are
> * limited to 65535 blocks.
> */
> -#define USB_MAX_XFER_BLK 65535
> +#define USB_MAX_XFER_BLK 32767
> #else
> #define USB_MAX_XFER_BLK 20
> #endif
>
--
Best regards,
Marek Vasut
^ permalink raw reply [flat|nested] 4+ messages in thread* [U-Boot] [PATCH] Lower USB_MAX_XFER_BLK to 32767 for trimslice
2016-05-26 20:55 [U-Boot] [PATCH] Lower USB_MAX_XFER_BLK to 32767 by default for trimslice Nicolas Chauvet
2016-05-26 21:17 ` Marek Vasut
@ 2016-05-28 8:06 ` Nicolas Chauvet
2016-05-28 12:02 ` Marek Vasut
1 sibling, 1 reply; 4+ messages in thread
From: Nicolas Chauvet @ 2016-05-28 8:06 UTC (permalink / raw)
To: u-boot
I'm experiencing an issue when loading a generic distro initramfs
on trimslice-pro (with ssd using usb-ehci) with the default value of
65535 for USB_MAX_XFER_BLK
EHCI timed out on TD - token=0x80008d80
As adviced by Marek on IRC, using a lower value for
USB_MAX_XFER_BLK allows to load the file
Tested values on tegra20-trimslice:
4095 ok
32767 ok
32768 timeout
36864 timeout
40960 timeout
49150 timeout
v2:
- Allow to override if undefined when using CONFIG_USB_EHCI
- Only change the value for trimslice
- Update description
Signed-off-by: Nicolas Chauvet <kwizart@gmail.com>
---
common/usb_storage.c | 2 ++
include/configs/trimslice.h | 2 ++
2 files changed, 4 insertions(+)
diff --git a/common/usb_storage.c b/common/usb_storage.c
index 7e6e52d..aea0477 100644
--- a/common/usb_storage.c
+++ b/common/usb_storage.c
@@ -106,7 +106,9 @@ struct us_data {
* enough free heap space left, but the SCSI READ(10) and WRITE(10) commands are
* limited to 65535 blocks.
*/
+#ifndef USB_MAX_XFER_BLK /* Allows a quirk for broken devices */
#define USB_MAX_XFER_BLK 65535
+#endif
#else
#define USB_MAX_XFER_BLK 20
#endif
diff --git a/include/configs/trimslice.h b/include/configs/trimslice.h
index b761640..3a2fe87 100644
--- a/include/configs/trimslice.h
+++ b/include/configs/trimslice.h
@@ -43,6 +43,8 @@
#define CONFIG_USB_EHCI
#define CONFIG_USB_EHCI_TEGRA
#define CONFIG_USB_STORAGE
+/* Quirk for broken usb-ehci on trimslice */
+#define USB_MAX_XFER_BLK 32767
/* USB networking support */
#define CONFIG_USB_HOST_ETHER
--
2.4.11
^ permalink raw reply related [flat|nested] 4+ messages in thread* [U-Boot] [PATCH] Lower USB_MAX_XFER_BLK to 32767 for trimslice
2016-05-28 8:06 ` [U-Boot] [PATCH] Lower USB_MAX_XFER_BLK to 32767 " Nicolas Chauvet
@ 2016-05-28 12:02 ` Marek Vasut
0 siblings, 0 replies; 4+ messages in thread
From: Marek Vasut @ 2016-05-28 12:02 UTC (permalink / raw)
To: u-boot
On 05/28/2016 10:06 AM, Nicolas Chauvet wrote:
> I'm experiencing an issue when loading a generic distro initramfs
> on trimslice-pro (with ssd using usb-ehci) with the default value of
> 65535 for USB_MAX_XFER_BLK
> EHCI timed out on TD - token=0x80008d80
>
> As adviced by Marek on IRC, using a lower value for
> USB_MAX_XFER_BLK allows to load the file
>
> Tested values on tegra20-trimslice:
> 4095 ok
> 32767 ok
> 32768 timeout
> 36864 timeout
> 40960 timeout
> 49150 timeout
>
> v2:
This should go below diffstat, so it's not part of the commit message
> - Allow to override if undefined when using CONFIG_USB_EHCI
> - Only change the value for trimslice
This will annoy trimslice users then ;-)
> - Update description
Going back to my comment about V1:
If you could add some quirk table into u-boot instead and lower this
number only for those buggy devices, that'd be much better.
> Signed-off-by: Nicolas Chauvet <kwizart@gmail.com>
> ---
> common/usb_storage.c | 2 ++
> include/configs/trimslice.h | 2 ++
> 2 files changed, 4 insertions(+)
>
> diff --git a/common/usb_storage.c b/common/usb_storage.c
> index 7e6e52d..aea0477 100644
> --- a/common/usb_storage.c
> +++ b/common/usb_storage.c
> @@ -106,7 +106,9 @@ struct us_data {
> * enough free heap space left, but the SCSI READ(10) and WRITE(10) commands are
> * limited to 65535 blocks.
> */
> +#ifndef USB_MAX_XFER_BLK /* Allows a quirk for broken devices */
> #define USB_MAX_XFER_BLK 65535
> +#endif
This has multiple problems:
a) This override is added only for EHCI while it is generic and would
work for all USB revisions
b) Any such override should be added to Kconfig instead
c) Tweaking this value will negatively impact multiple devices only to
fix problems on a small amount of crappy devices (see above)
> #else
> #define USB_MAX_XFER_BLK 20
> #endif
> diff --git a/include/configs/trimslice.h b/include/configs/trimslice.h
> index b761640..3a2fe87 100644
> --- a/include/configs/trimslice.h
> +++ b/include/configs/trimslice.h
> @@ -43,6 +43,8 @@
> #define CONFIG_USB_EHCI
> #define CONFIG_USB_EHCI_TEGRA
> #define CONFIG_USB_STORAGE
> +/* Quirk for broken usb-ehci on trimslice */
> +#define USB_MAX_XFER_BLK 32767
>
> /* USB networking support */
> #define CONFIG_USB_HOST_ETHER
>
--
Best regards,
Marek Vasut
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2016-05-28 12:02 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2016-05-26 20:55 [U-Boot] [PATCH] Lower USB_MAX_XFER_BLK to 32767 by default for trimslice Nicolas Chauvet
2016-05-26 21:17 ` Marek Vasut
2016-05-28 8:06 ` [U-Boot] [PATCH] Lower USB_MAX_XFER_BLK to 32767 " Nicolas Chauvet
2016-05-28 12:02 ` Marek Vasut
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox