Re: USB-to-serial adapter no longer working in 8.0-RC1

On Saturday 19 September 2009 20:13:33 Greg Rivers wrote:
> On Sat, 19 Sep 2009, Hans Petter Selasky wrote:
> > On Saturday 19 September 2009 05:40:28 Greg Rivers wrote:
> >> Last May, Hans Petter Selasky fixed up the umct driver for 8.0-CURRENT
> >> (http://lists.freebsd.org/pipermail/freebsd-current/2009-May/007125.html
> >>). After updating to 8.0-RC1 I find that the driver has a new problem.
> >>
> >> Testing with cu(1) and a modem, I can see by watching the lights on the
> >> modem and on the usb-serial adapter that it still asserts DTR and sends
> >> and receives characters properly, but no characters reach the terminal.
> >> That seems to be the only problem: characters from the attached serial
> >> device seem to disappear before reaching whatever program has the serial
> >> device open.  But strangely enough, power cycling the modem results in a
> >> few "line noise" characters appearing.  Perhaps the driver is no longer
> >> responding appropriately to DSR?
> >>
> >> Any insight or troubleshooting clues would be much appreciated.
> >
> > Hi,
> >
> > You need to provide debug output from:
> >
> > sysctl hw.usb.ucom.debug=15
> >
> > and
> >
> > sysctl hw.usb.umct.debug=15
> >
> > If it exists.
>
> hw.usb.umct does not exist, but hw.usb.ucom does.  Here's debug output with
> hw.usb.ucom.debug set to 15.  I tested with a modem because it's easy to
> interact with.
>
> Attach USB-serial adapter to USB port:
> ugen1.2: <Belkin USB PDA Adapter> at usbus1
> umct0: <Belkin USB PDA Adapter Belkin Components, class 0/0, rev 1.10/1.02,
> addr 2> on usbus1 ucom_attach_tty:317: tp = 0xc5644200, unit = 0
> ucom_attach_tty:346: ttycreate: U0
>
> Open /dev/cuaU0 via 'cu -l cuaU0 -s 115200 dir':
> ucom_open:554: tp = 0xc5644200
> ucom_dtr:827: onoff = 1
> ucom_line_state:799: on=0x01, off=0x00
> ucom_rts:838: onoff = 1
> ucom_line_state:799: on=0x02, off=0x00
> ucom_break:816: onoff = 0
> ucom_line_state:799: on=0x00, off=0x04
> ucom_status_change:901:
> ucom_param:950: sc = 0xc5433230
> ucom_dtr:827: onoff = 1
> ucom_line_state:799: on=0x01, off=0x00
> ucom_rts:838: onoff = 1
> ucom_line_state:799: on=0x02, off=0x00
> ucom_cfg_open:520:
> ucom_ioctl:653: cmd = 0x402c7413
> ucom_ioctl:653: cmd = 0x802c7416
> ucom_ioctl:653: cmd = 0x2000740d
> ucom_ioctl:653: cmd = 0x402c7413
> ucom_ioctl:653: cmd = 0x802c7416
> ucom_param:950: sc = 0xc5433230
> ucom_ioctl:653: cmd = 0x8004667e
> ucom_ioctl:653: cmd = 0x8004667d
> ucom_get_data:1064: cnt=0
>
> Type "atz<enter>" (modem resets, responds with "OK"):
> ucom_outwakeup:1009: sc = 0xc5433230
> ucom_get_data:1064: cnt=1
> ucom_get_data:1064: cnt=0
> ucom_outwakeup:1009: sc = 0xc5433230
> ucom_get_data:1064: cnt=1
> ucom_get_data:1064: cnt=0
> ucom_outwakeup:1009: sc = 0xc5433230
> ucom_get_data:1064: cnt=1
> ucom_get_data:1064: cnt=0
> ucom_outwakeup:1009: sc = 0xc5433230
> ucom_get_data:1064: cnt=1
> ucom_get_data:1064: cnt=0
> ucom_status_change:901:
>
> Type "atdt<enter>" (modem goes off hook,
> finds no dial tone, prints "NO DIALTONE", hangs up):
> ucom_outwakeup:1009: sc = 0xc5433230
> ucom_get_data:1064: cnt=1
> ucom_get_data:1064: cnt=0
> ucom_outwakeup:1009: sc = 0xc5433230
> ucom_get_data:1064: cnt=1
> ucom_get_data:1064: cnt=0
> ucom_outwakeup:1009: sc = 0xc5433230
> ucom_get_data:1064: cnt=1
> ucom_get_data:1064: cnt=0
> ucom_outwakeup:1009: sc = 0xc5433230
> ucom_get_data:1064: cnt=1
> ucom_get_data:1064: cnt=0
> ucom_outwakeup:1009: sc = 0xc5433230
> ucom_get_data:1064: cnt=1
> ucom_get_data:1064: cnt=0
> ucom_status_change:901:
> last message repeated 2 times
>
> Type "atz<enter>" (modem resets, responds with "OK"):
> ucom_outwakeup:1009: sc = 0xc5433230
> ucom_get_data:1064: cnt=1
> ucom_get_data:1064: cnt=0
> ucom_outwakeup:1009: sc = 0xc5433230
> ucom_get_data:1064: cnt=1
> ucom_get_data:1064: cnt=0
> ucom_outwakeup:1009: sc = 0xc5433230
> ucom_get_data:1064: cnt=1
> ucom_get_data:1064: cnt=0
> ucom_outwakeup:1009: sc = 0xc5433230
> ucom_get_data:1064: cnt=1
> ucom_get_data:1064: cnt=0
> ucom_status_change:901:
>
> Type "~." to exit cu:
> ucom_ioctl:653: cmd = 0x2000740e
> ucom_ioctl:653: cmd = 0x2000740e
> ucom_close:623: tp=0xc5644200
> ucom_shutdown:446:
> ucom_dtr:827: onoff = 0
> ucom_line_state:799: on=0x00, off=0x01
> ucom_rts:838: onoff = 1
> ucom_line_state:799: on=0x02, off=0x00
> ucom_cfg_close:605:
>
> Detach USB-serial adapter from USB port:
> ugen1.2: <Belkin USB PDA Adapter> at usbus1 (disconnected)
> umct0: at uhub0, port 1, addr 2 (disconnected)
> ucom_detach_tty:358: sc = 0xc5433230, tp = 0xc5644200
> ucom_close:623: tp=0xc5644200
> ucom_close:626: tp=0xc5644200 already closed
>
> Note that none of the characters output by the modem made it to cu.  Each
> sequence of	ucom_outwakeup:1009: sc = 0xc5433230
>  		ucom_get_data:1064: cnt=1
>  		ucom_get_data:1064: cnt=0
> corresponds to me typing a character.  There are no debug lines
> corresponding to characters output by the modem, so they must be getting
> dropped prior to reaching the ucom driver.
>
> I also tested with a 8.0-BETA2 system from July 18; it has the same
> problem.  But I know that it was working on 8.0-CURRENT as recently as
> the first week of June.  It still works fine on recent 7.2-STABLE.
	/*
	 * The real bulk-in endpoint is also marked as an interrupt.
	 * The only way to differentiate it from the real interrupt
	 * endpoint is to look at the wMaxPacketSize field.
	 */
#ifdef XXX
	maxp = UGETW(sc->sc_xfer[UMCT_BULK_DT_RD]->endpoint->edesc->wMaxPacketSize);
	if (maxp == 0x2) {

		/* guessed wrong - switch around endpoints */

		struct usb_xfer *temp = sc->sc_xfer[UMCT_INTR_DT_RD];

		sc->sc_xfer[UMCT_INTR_DT_RD] = sc->sc_xfer[UMCT_BULK_DT_RD];
		sc->sc_xfer[UMCT_BULK_DT_RD] = temp;

		sc->sc_xfer[UMCT_BULK_DT_RD]->callback = &umct_read_callback;
		sc->sc_xfer[UMCT_INTR_DT_RD]->callback = &umct_intr_callback;
	}
#endif

Could you try to change this into #if 1 ?

--HPS
Hi,

I see some code was commented out in umct.c: