If the HW reports the endpoint as Halted, there's no need to start any URB for that endpoint, we can simply return -EPIPE and this will tell upper layers that the endpoint is, indeed, halted. This patch fixes usbtest's set/clear halt test #13 on xHCI-based hosts. Cc: <stable@xxxxxxxxxxxxxxx> Signed-off-by: Felipe Balbi <balbi@xxxxxx> --- Hi Sarah, seems like this has been broken for quite a long time. I tested the patch on an AM437x which has XHCI Host and the same IP configured as device as well (two dwc3 instances, basically). It has been running for the past hour or so without any failures when running testusb -t 13 -c 5000 -s 2048 -a in a loop, but if you think there's a better way to fix this, let me know. I just didn't understand why xhci-hcd still queues the URB even though HW already told us the endpoint is halted. Any comments ? cheers drivers/usb/host/xhci-ring.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c index 53c2e29..e9df61a 100644 --- a/drivers/usb/host/xhci-ring.c +++ b/drivers/usb/host/xhci-ring.c @@ -2959,7 +2959,8 @@ static int prepare_ring(struct xhci_hcd *xhci, struct xhci_ring *ep_ring, /* XXX not sure if this should be -ENOENT or not */ return -EINVAL; case EP_STATE_HALTED: - xhci_dbg(xhci, "WARN halted endpoint, queueing URB anyway.\n"); + xhci_dbg(xhci, "WARN halted endpoint.\n"); + return -EPIPE; case EP_STATE_STOPPED: case EP_STATE_RUNNING: break; -- 1.8.4.GIT -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
