openwrt/target/linux/brcm2708/patches-3.10/0060-dwc_otg-fix-NAK-holdoff-and-allow-on-split-transacti.patch
Florian Fainelli bb39b8d99a brcm2708: update against latest rpi-3.10.y branch
Update our copies of the brcm2708 patches to the latest rpi-3.10-y
rebased against linux-3.10.y stable (3.10.32). This should hopefully
make it easier for us in the future to leverage the raspberry/rpi-*
branches.

Signed-off-by: Florian Fainelli <florian@openwrt.org>

SVN-Revision: 39770
2014-02-28 20:30:08 +00:00

68 lines
2.7 KiB
Diff

From 370cbdb4a5ec521d9312b02bbaf269ed520b1451 Mon Sep 17 00:00:00 2001
From: P33M <P33M@github.com>
Date: Mon, 22 Apr 2013 00:08:36 +0100
Subject: [PATCH 060/174] dwc_otg: fix NAK holdoff and allow on split
transactions only
This corrects a bug where if a single active non-periodic endpoint
had at least one transaction in its qh, on frnum == MAX_FRNUM the qh
would get skipped and never get queued again. This would result in
a silent device until error detection (automatic or otherwise) would
either reset the device or flush and requeue the URBs.
Additionally the NAK holdoff was enabled for all transactions - this
would potentially stall a HS endpoint for 1ms if a previous error state
enabled this interrupt and the next response was a NAK. Fix so that
only split transactions get held off.
---
drivers/usb/host/dwc_otg/dwc_otg_hcd.c | 28 ++++++++++++++++++----------
1 file changed, 18 insertions(+), 10 deletions(-)
--- a/drivers/usb/host/dwc_otg/dwc_otg_hcd.c
+++ b/drivers/usb/host/dwc_otg/dwc_otg_hcd.c
@@ -46,7 +46,7 @@
#include "dwc_otg_hcd.h"
#include "dwc_otg_regs.h"
-extern bool microframe_schedule;
+extern bool microframe_schedule, nak_holdoff_enable;
//#define DEBUG_HOST_CHANNELS
#ifdef DEBUG_HOST_CHANNELS
@@ -1349,18 +1349,26 @@ dwc_otg_transaction_type_e dwc_otg_hcd_s
/*
* Check to see if this is a NAK'd retransmit, in which case ignore for retransmission
- * we hold off on bulk retransmissions to reduce NAK interrupt overhead for
+ * we hold off on bulk retransmissions to reduce NAK interrupt overhead for full-speed
* cheeky devices that just hold off using NAKs
*/
- if (dwc_full_frame_num(qh->nak_frame) == dwc_full_frame_num(dwc_otg_hcd_get_frame_number(hcd))) {
- // Make fiq interrupt run on next frame (i.e. 8 uframes)
- g_next_sched_frame = ((qh->nak_frame + 8) & ~7) & DWC_HFNUM_MAX_FRNUM;
- qh_ptr = DWC_LIST_NEXT(qh_ptr);
- continue;
+ if (nak_holdoff_enable && qh->do_split) {
+ if (qh->nak_frame != 0xffff &&
+ dwc_full_frame_num(qh->nak_frame) ==
+ dwc_full_frame_num(dwc_otg_hcd_get_frame_number(hcd))) {
+ /*
+ * Revisit: Need to avoid trampling on periodic scheduling.
+ * Currently we are safe because g_np_count != g_np_sent whenever we hit this,
+ * but if this behaviour is changed then periodic endpoints will get a slower
+ * polling rate.
+ */
+ g_next_sched_frame = ((qh->nak_frame + 8) & ~7) & DWC_HFNUM_MAX_FRNUM;
+ qh_ptr = DWC_LIST_NEXT(qh_ptr);
+ continue;
+ } else {
+ qh->nak_frame = 0xffff;
+ }
}
- else
- qh->nak_frame = 0xffff;
-
if (microframe_schedule) {
DWC_SPINLOCK_IRQSAVE(channel_lock, &flags);
if (hcd->available_host_channels < 1) {