When suspend, some urb is put into cancelled_td_list. process_bulk_intr_td may process the last trb as below: xhci_transfer_event.transfer_len = 0 xhci_transfer_event.trb_comp_code = 27 urb.transfer_buffer_length = 1024 trb_comp_code is COMP_STOP_INVAL, it should be taken as invalid urb and discarded, but urb actual_length still is calculated as transfer_buffer_length. When handling Stop Endpoint Command completion, the urb in cancelled_td_list is transferred to upper layer(cdc-acm for example). The content of urb transfer_buffer is the overlay of previous two read transfer_buffer. It makes upper layer process wrong buffer. This patch is to set actual_length as transfer_buffer_length only if trb_comp_code is not COMP_STOP_INVAL when process_bulk_intr_td. Signed-off-by: xiao jin <jin.xiao@xxxxxxxxx> --- drivers/usb/host/xhci-ring.c | 7 +++++-- 1 files changed, 5 insertions(+), 2 deletions(-) diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c index 6bfbd80..c9a8863 100644 --- a/drivers/usb/host/xhci-ring.c +++ b/drivers/usb/host/xhci-ring.c @@ -2319,8 +2319,11 @@ static int process_bulk_intr_td(struct xhci_hcd *xhci, struct xhci_td *td, *status = 0; } } else { - td->urb->actual_length = - td->urb->transfer_buffer_length; + if (trb_comp_code != COMP_STOP_INVAL) + td->urb->actual_length = + td->urb->transfer_buffer_length; + else + td->urb->actual_length = 0; /* Ignore a short packet completion if the * untransferred length was zero. */ -- 1.7.0.4 -- 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