Here's another demo that indicates that sparc64 doesn't work properly. In this case it's a CONS_SETKBD ioctl which is used by kbdcontrol(8). I "kldload vkbd" and make sure two vkbd instances exist, /dev/vkbdctl0 and /dev/vkbdctl1.
Now, I open the first like this:
sleep 1200 < /dev/vkbdctl0
and attempt to program /dev/vkbdctl1 as a keyboard for
virtual terminals. On amd64, I get:
# kbdcontrol -K < /dev/ttyv0 > /dev/null ; kbdcontrol -k /dev/vkbdctl1 -i <
/dev/ttyv0
kbd2 at vkbd1
kbd2
vkbd1, type:generic (0)
kbd2:
vkbd1, type:generic (0)
It's fine; the "-i" returns the same console as as was programmed.
On sparc64, I get:
# kbdcontrol -K < /dev/ttyv0 > /dev/null ; kbdcontrol -k /dev/vkbdctl1 -i <
/dev/ttyv0
kbd1
vkbd1, type:generic (0)
kbd0:
vkbd0, type:generic (0)
Surprise! It assigned already open (by sleep(1)) vkbd0 while I
asked for vkbd1.
Oh yes, if you ever want to repeat this, make sure to temporary
apply this patch to kbdcontrol.c, because vkbd is only valid
when it's open:
%%%
Index: kbdcontrol.c
===================================================================
RCS file: /home/ncvs/src/usr.sbin/kbdcontrol/kbdcontrol.c,v
retrieving revision 1.50
diff -u -p -r1.50 kbdcontrol.c
--- kbdcontrol.c 15 Sep 2006 18:41:12 -0000 1.50
+++ kbdcontrol.c 20 Sep 2006 16:18:48 -0000
@@ -1046,7 +1046,9 @@ set_keyboard(char *device)
* is closed. So, we don't check error here.
*/
ioctl(fd, CONS_RELKBD, 0);
+#if 0
close(fd);
+#endif
#if 1
printf("kbd%d\n", info.kb_index);
printf(" %.*s%d, type:%s (%d)\n",
%%%
What this means practically is that currently, on sparc64, you
can only have a /dev/kbd0 keyboard attached to a syscons console.
A fix is in the works...
Cheers,
--
Ruslan Ermilov
[EMAIL PROTECTED]
FreeBSD committer
pgpeCRSKBF7W0.pgp
Description: PGP signature
