Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20220126205214.2149936-1-jannh@google.com> <CA+fCnZe_p+JwUWwumgGm185vWSdAK_z-UFDp7-HWKANB4YjA=g@mail.gmail.com>
In-Reply-To: <CA+fCnZe_p+JwUWwumgGm185vWSdAK_z-UFDp7-HWKANB4YjA=g@mail.gmail.com>
From:   Jann Horn <jannh@google.com>
Date:   Wed, 26 Jan 2022 23:30:54 +0100
Message-ID: <CAG48ez0pke5fqEuWGecMAKLpsdVoW3JM3M-SkajHcq_dsrQ_4A@mail.gmail.com>
Subject: Re: [PATCH] usb: raw-gadget: fix handling of dual-direction-capable endpoints
To:     Andrey Konovalov <andreyknvl@gmail.com>
Cc:     Felipe Balbi <balbi@kernel.org>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        USB list <linux-usb@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Wed, Jan 26, 2022 at 11:12 PM Andrey Konovalov <andreyknvl@gmail.com> wrote:
> On Wed, Jan 26, 2022 at 9:52 PM Jann Horn <jannh@google.com> wrote:
> >
> > Under dummy_hcd, every available endpoint is *either* IN or OUT capable.
> > But with some real hardware, there are endpoints that support both IN and
> > OUT. In particular, the PLX 2380 has four available endpoints that each
> > support both IN and OUT.
> >
> > raw-gadget currently gets confused and thinks that any endpoint that is
> > usable as an IN endpoint can never be used as an OUT endpoint.
> >
> > Fix it by looking at the direction in the configured endpoint descriptor
> > instead of looking at the hardware capabilities.
> >
> > With this change, I can use the PLX 2380 with raw-gadget.
> >
> > Fixes: f2c2e717642c ("usb: gadget: add raw-gadget interface")
> > Signed-off-by: Jann Horn <jannh@google.com>
> > ---
> >  drivers/usb/gadget/legacy/raw_gadget.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/usb/gadget/legacy/raw_gadget.c b/drivers/usb/gadget/legacy/raw_gadget.c
> > index c5a2c734234a..d86c3a36441e 100644
> > --- a/drivers/usb/gadget/legacy/raw_gadget.c
> > +++ b/drivers/usb/gadget/legacy/raw_gadget.c
> > @@ -1004,7 +1004,7 @@ static int raw_process_ep_io(struct raw_dev *dev, struct usb_raw_ep_io *io,
> >                 ret = -EBUSY;
> >                 goto out_unlock;
> >         }
> > -       if ((in && !ep->ep->caps.dir_in) || (!in && ep->ep->caps.dir_in)) {
> > +       if (in != usb_endpoint_dir_in(ep->ep->desc)) {
> >                 dev_dbg(&dev->gadget->dev, "fail, wrong direction\n");
> >                 ret = -EINVAL;
> >                 goto out_unlock;
> >
> > base-commit: 0280e3c58f92b2fe0e8fbbdf8d386449168de4a8
> > --
> > 2.35.0.rc0.227.g00780c9af4-goog
> >
>
> Awesome! Thanks for finding this!
>
> What do you think about using
>
> if ((in && !ep->ep->caps.dir_in) || (!in && !ep->ep->caps.dir_out))
>
> instead?
>
> It looks less cryptic: if (in and no in caps) or (out and no out caps) => fail.

That's also semantically different, right?
As I understand it, what we should be checking here is whether the
direction of the request matches the direction previously specified in
USB_RAW_IOCTL_EP_ENABLE, not whether the hardware would be capable of
using the endpoint in the requested direction if it had been
configured for that direction?
But I might also be misunderstanding what's going on - it's not like
I've looked at a spec for this or anything like that, I'm just kinda
guessing...