Date: Wed, 11 May 2005 13:32:52 +0200 From: Hans Petter Selasky <hselasky@c2i.net> To: freebsd-usb@freebsd.org, paradox <algorist2000@yahoo.com> Subject: Re: a problem about ugen.c Message-ID: <200505111332.53170.hselasky@c2i.net> In-Reply-To: <20050511062048.37521.qmail@web53602.mail.yahoo.com> References: <20050511062048.37521.qmail@web53602.mail.yahoo.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wednesday 11 May 2005 08:20, paradox wrote:
> Thanks,all problems are solved.
>
> The usb key device has two hardware bugs, one is the
> CLEAR_FEATURE packet problem just like you said, the
> other is an error in the toggle bit changing.
>
> after I change the usb-subr.c, I can use ugen.ko to
> read and write the device directly.
>
> but when I read data from the device(low speed,
> interrupt pipe),the data length I read must be less
> than the max packet size,or some data will be lost.
> The read command only read data in one interval. but I
> thought the standard read function should read the
> data till all the data has been read or we got an
> incomplete packet. Am I right?
The current ugen implementation for INTERRUPT transfers is going to loose
packet synchronization if too many packets arrive at the same time, because
it uses a ring-buffer to buffer data. Maybe it should use mbufs instead?
You read data from an ugen INTERRUPT endpoint like this:
length = read(fd, buf, sizeof(buf));
/* got one packet (should be like this) */
if(length == -1)
error;
else
got length bytes of data (and not sizeof(buf) bytes).
buf should have a size greater than maxpacket size, and if
(length == sizeof(buf)) packet synchronization will be lost
until the buffer drains. This should be fixed in the ugen driver.
>
> --- Hans Petter Selasky <hselasky@c2i.net> wrote:
> > On Saturday 30 April 2005 09:00, paradox wrote:
> >
> > It might be the CLEAR_FEATURE or clear stall packet
> > that does it. Some devices
> > will only work if clear stall is issued as a part of
> > a reset sequence. This
> > has been discussed before on this list.
> >
> > In the file "/sys/dev/usb/usb_subr.c" in the
> > function "usbd_setup_pipe" could
> > you change:
> >
> > err = usbd_clear_endpoint_stall(p);
> >
> > into
> >
> > err = 0; /* usbd_clear_endpoint_stall(p); */
I see that the memory allocated is not zeroed, so you
might have to add:
(p->methods->cleartoggle)(p);
after
err = 0;
> >
> > Then recompile the USB module:
> > "make -C/sys/modules/usb depend all install clean"
> >
> > Reboot and see if there is any change.
> >
--HPS
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200505111332.53170.hselasky>