Re: Sony Clie PEG-TJ37 vs. FreeBSD CURRENT (long)

On Thu, 2004-07-01 at 10:25, Brian Fundakowski Feldman wrote:

> Great, glad it works!  Yeah, that's the part where you'd potentially
> want to attach them all; since you're the one with the hardware
> in a position to really test it, I'd appreciate if you'd add the
> appropriate USB_DEBUG printfs for the PALM4 (!VISOR) case to see what
> ports are really available and hopefully get someone who really knows
> the USB/device drivers to finish making multiple ucom attachment work.

OK, now I'm more confused.  Please see the attached patch, which I added
to grab the connection information.

(Yippee...a chance to embarass myself with my lamer un-l33t kernel
hacking skillz.)

My debug output looks like (after MTA wordwrap):

Jul  1 12:43:14 tomcat kernel: ucom0: Palm, Inc. Palm Handheld, rev
1.00/1.00, addr 2
Jul  1 12:43:14 tomcat kernel: ucom0: Palm, Inc. Palm Handheld, rev
1.00/1.00, addr 2
Jul  1 12:43:14 tomcat kernel: ucom0: Connection 0: port 0
end_point_info 33 port_function_id 5f 70 70 70

If this is right, the only reason that I'm getting a successful
attachment is the "if (i == nc) i = nc - 1;" fallthrough code.  Huh?!?

Thanks,

Bruce.


--- uvisor.c-green	Thu Jul  1 10:42:17 2004
+++ uvisor.c	Thu Jul  1 12:41:42 2004
_at__at_ -368,6 +368,18 _at__at_
 		if (nc > UVISOR_MAX_CONN)
 			nc = UVISOR_MAX_CONN;
 
+#ifdef USB_DEBUG
+		for (i = 0; i < nc; ++i) {
+			int j;
+			printf("%s: Connection %d: port %d end_point_info %d port_function_id ", 
+			       devname, i,
+			       coninfo.ci_palm.connections[i].port,
+			       coninfo.ci_palm.connections[i].end_point_info);
+			for (j = 0; j < sizeof(uDWord); ++j)
+				printf("%2x ", coninfo.ci_palm.connections[i].port_function_id[j]);
+			printf("\n");
+		}
+#endif
 		/* XXX Should attach a ucom for each connection. */
 		for (i = 0; i < nc; ++i)
 			if (coninfo.ci_palm.connections[i].port_function_id[0]