New patch to review for coreboot: e7b585e libpayload/usb: wait a millisecond to work around device bugs

[Patrick Georgi]

Patrick Georgi ([e-mail]) just uploaded a new patch set to gerrit, which you can find at http://review.coreboot.org/8735

-gerrit

commit e7b585e051ee5f05d0c54c8940475aeb566a367e
Author: Patrick Georgi 
Date:   Mon Nov 10 16:50:21 2014 +0100

libpayload/usb: wait a millisecond to work around device bugs

Some USB sticks seem to send a NAK at a place where they mustn't
    by spec, leading to a controller side error condition.

To avoid it, wait a millisecond which is enough to get past the
    NAK condition. That delay only happens on device discovery so it
    won't affect boot time by more than 1ms per device.

BUG=chromium:414959
    BRANCH=none
    TEST=depthcharge recognizes a Lexar 16GB USB stick after applying
    this change.

Change-Id: I0e385702a5259b16fda0a253fc121d8f66e6705c
    Signed-off-by: Patrick Georgi 
    Original-Commit-Id: 10bbfda8395af009e7f910cc503f50c2ad969ae8
    Original-Change-Id: I6dd5ca34e9f3767003ccb0ca9daaf16116f4a2df
    Original-Signed-off-by: Patrick Georgi 
    Original-Reviewed-on: https://chromium-review.googlesource.com/228791
    Original-Reviewed-by: Julius Werner 
    Original-Reviewed-by: Sheng-liang Song 
---
 payloads/libpayload/drivers/usb/usb.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/payloads/libpayload/drivers/usb/usb.c b/payloads/libpayload/drivers/usb/usb.c
index 53006ff..54b113c 100644
--- a/payloads/libpayload/drivers/usb/usb.c
+++ b/payloads/libpayload/drivers/usb/usb.c
@@ -393,6 +393,9 @@ set_address (hci_t *controller, usb_speed speed, int hubport, int hubaddr)
    	usb_detach_device (controller, dev->address);
    	return -1;
    }
+	/* workaround for some USB devices: wait until they're ready, or
+	 * they send a NAK when they're not allowed to do. 1ms is enough */
+	mdelay(1);
    dev->configuration = malloc(buf[1]);
    if (!dev->configuration) {
    	usb_debug ("could not allocate %d bytes for DT_CFG\n", buf[1]);