From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752253AbcLFWGT (ORCPT ); Tue, 6 Dec 2016 17:06:19 -0500 Received: from mail.free-electrons.com ([62.4.15.54]:54896 "EHLO mail.free-electrons.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750953AbcLFWGQ (ORCPT ); Tue, 6 Dec 2016 17:06:16 -0500 Date: Tue, 6 Dec 2016 23:06:04 +0100 From: Boris Brezillon To: Nicolas Ferre , Alexandre Belloni , Felipe Balbi , Greg Kroah-Hartman , linux-usb@vger.kernel.org Cc: linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH] usb: gadget: udc: atmel: Fix check in usba_ep_enable() Message-ID: <20161206230604.4bcdb3e7@bbrezillon> In-Reply-To: <1481061583-4727-1-git-send-email-boris.brezillon@free-electrons.com> References: <1481061583-4727-1-git-send-email-boris.brezillon@free-electrons.com> X-Mailer: Claws Mail 3.13.2 (GTK+ 2.24.30; x86_64-pc-linux-gnu) MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi Felipe, I realize I sent this patch to your old @ti.com email address. Do you want me to resend it? Regards, Boris On Tue, 6 Dec 2016 22:59:43 +0100 Boris Brezillon wrote: > desc->bEndpointAddress & USB_ENDPOINT_NUMBER_MASK is not necessarily > equal to ep->index and that's perfectly fine. The usba endpoint index is > just an internal identifier used by the driver to know which registers > to use for a USB endpoint. > > Enforcing this constraint is not only useless, but can also lead to > errors since nothing guarantees that the endpoint number and index are > matching when an endpoint is selected for a specific descriptor, thus > leading to errors at ->enable() time when it's already too late to choose > another endpoint. > > Signed-off-by: Boris Brezillon > --- > Hi, > > I intentionally didn't add the Cc stable and Fixes tags because this > bug dates back to the drivers creation, and I fear the index <-> > epnum constraint was actually required at that time. > > Note that I discovered this bug thanks to the WARN_ON_ONCE() in > usb_ep_queue() [1] which was introduced in 4.5. > It might appear that this problem was silently ignored before that > (with part of the usba_ep_enable() code being skipped without any > notice). > > Regards, > > Boris > > [1]http://lxr.free-electrons.com/source/drivers/usb/gadget/udc/core.c#L264 > --- > drivers/usb/gadget/udc/atmel_usba_udc.c | 7 ++----- > 1 file changed, 2 insertions(+), 5 deletions(-) > > diff --git a/drivers/usb/gadget/udc/atmel_usba_udc.c b/drivers/usb/gadget/udc/atmel_usba_udc.c > index bb1f6c8f0f01..981d2639d413 100644 > --- a/drivers/usb/gadget/udc/atmel_usba_udc.c > +++ b/drivers/usb/gadget/udc/atmel_usba_udc.c > @@ -531,11 +531,8 @@ usba_ep_enable(struct usb_ep *_ep, const struct usb_endpoint_descriptor *desc) > > maxpacket = usb_endpoint_maxp(desc) & 0x7ff; > > - if (((desc->bEndpointAddress & USB_ENDPOINT_NUMBER_MASK) != ep->index) > - || ep->index == 0 > - || desc->bDescriptorType != USB_DT_ENDPOINT > - || maxpacket == 0 > - || maxpacket > ep->fifo_size) { > + if (ep->index == 0 || desc->bDescriptorType != USB_DT_ENDPOINT || > + maxpacket == 0 || maxpacket > ep->fifo_size) { > DBG(DBG_ERR, "ep_enable: Invalid argument"); > return -EINVAL; > }