my usb prog

[Hartmut Brandt]

Chuck Robey wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Man, I'm just not sure if I am making a mistake posting here, but I tried the
> usb list, and no answer to either subject, so I am risking bothering you.  If
> I'm wrong, and you jump all over me, I will learn that way, at least, not to do
> it again.  Ok, 2 parts below, here's part 1:
> 
> I've been working trying to get my UC-Logic graphic tablet to work under
> - -current.  First thing I did, I began with an equivalent of a "Hello, world" and
>  added different features so I could get more experience playing with uhid0
> (where dmesg puts my tablet).  One part of it is a reader/dumper of the
> descriptor using libusbhid, and one part it printed out, the usage page, kept on
>  printing as a 0x0D (the number of a Digitizer) instead of what I thought it
> should have been, the word "Digitizer".  I looked over libusbhid, and found a
> small change I could put into usage.c to get the right part to print ...
> 
>  cvs diff: Diffing .
> Index: usage.c
> ===================================================================
> RCS file: /home/ncvs/src/lib/libusbhid/usage.c,v
> retrieving revision 1.8
> diff -u -r1.8 usage.c
> - --- usage.c     9 Apr 2003 01:52:48 -0000       1.8
> +++ usage.c     24 May 2008 16:34:17 -0000
> @@ -164,7 +164,7 @@
>                 errx(1, "no hid table");
> 
>         for (k = 0; k < npages; k++)
> - -               if (pages[k].usage == i)
> +               if (pages[k].usage == HID_PAGE(i))
>                         return pages[k].name;
>         sprintf(b, "0x%04x", i);
>         return b;
> 
> I wanted to apply this myself, but I wanted someone else to verify it looks ok
> to apply.  I used to be pretty active wtih ports, but didn't work in src before
> (publicly), so if you'd just NOT apply this, just tell me if it's ok or not, I
> will see about applying it.  I  have a freebsd.org ssh login from long while
> back, it still works, so if this is a good diff, please leave it to me to do, ok?
> 
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 2nd part is that my other prog, a uhid0 dumper, which is the first part of a
> Xinput module for my tablet, is not able to correctly dump the uhid0.  At first
> it looked like:
> 
>         while(nread = read(udev, buf, 1024) > 0){
>             for(i=0; i<nread; i++)
>                 fprintf(stderr,"i;%d:%x",nread,buf[i]);
>         }
> 
> This always returns ";1;9" in endless repetitions, no changes, and nread is
> always 1.  nread is a ssize_t.  Well, finally, I got to wondering if it might be
> possible that nread was wrong, so I made these changes to show me the first few
> bytes following, no matter what the nread was:
> 
>         while(nread = read(udev, buf, 1024) > 0){
>             fprintf(stderr,";%d",nread);
>             for(i=0; i<5; i++)
>                 fprintf(stderr,":%x",buf[i]);
>         }
> 


Cannot comment on the other stuff, but you probably want

	while ((nread = read(udev, buf, 1024)) > 0) {
		fprintf(stderr, ";%zd", nread);

As a matter of style, I even would write

	while ((nread = read(udev, buf, 1024)) != -1) {

(read is supposed to return -1 on errors, not 'something negative', I 
think).

harti


> Well, the funny thing is, the data coming out changes!  I get the strong notion
> that I really outght to simply read the number of bytes that the report
> descriptor tells me to, but the first code sample above, shouldn't it work just
> as I wrote it?  What am I doing wrong?  Or, is it possible that the uhid code is
> wrong, I should maybe read the uhid driver and fix something?  I can easily dump
> enough of the desciptor to find out how much to read, no problem there, I just
> can't figure out why the "correct" approach fails.
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v2.0.4 (FreeBSD)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
> 
> iD8DBQFIOZRez62J6PPcoOkRAi0TAJ9aV1dY4SMBSmyqv9hHski3J0IlkQCeLHex
> zz42d1+2Tf6SERctKLE/+gM=
> =mWZN
> -----END PGP SIGNATURE-----
> _______________________________________________
> freebsd-current at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to "freebsd-current-unsubscribe at freebsd.org"
>