From owner-freebsd-current  Fri May 17 08:54:02 1996
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id IAA26518
          for current-outgoing; Fri, 17 May 1996 08:54:02 -0700 (PDT)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id IAA26512
          for <current@FreeBSD.org>; Fri, 17 May 1996 08:53:54 -0700 (PDT)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.12/8.6.9) id BAA28332; Sat, 18 May 1996 01:47:15 +1000
Date: Sat, 18 May 1996 01:47:15 +1000
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199605171547.BAA28332@godzilla.zeta.org.au>
To: charnier@lirmm.fr, current@FreeBSD.org
Subject: Re: Keyboard reset failed
Sender: owner-current@FreeBSD.org
X-Loop: FreeBSD.org
Precedence: bulk

>I got this message at boot since the last commit to syscons. I tried to
>increase the number of tries to 20 (default is 10) but no luck. The returned
>val is 0xfa (aka KB_ACK)...

I'd like to know why it used to work.  It used to wait forever for valid
input.  Now it reads garbage input if valid input arrives later than
expected, but it partially compensates for this bug by silently retrying
up to 10 times if the garbage happens to equal KB_ACK (which it usually
does).  Try increasing the number of tries a lot more and printing the
number of retries.

Garbage is also read in the `while (retries--)' loop.  inb(KB_DATA) isn't
valid until a few usec after (inb(KB_STAT) & KB_BUF_FULL) != 0, but KB_STAT
isn't read at all in this loop.

I see times of 5ms between sending the KB_RESET and receiving the KB_ACK,
and 3 ms between receiving the KB_ACK and receiving the KB_RESET_DONE.

Bruce