Subject: Re: Two problems with 12/24 sources
To: None <>
From: John F. Woods <>
List: current-users
Date: 12/28/1993 10:22:32
> I've encounted two problems with the sources supped from 12/24.  Any
> clues are much appreciated...
> 1) someone broke ps:
> [root@stingray]~# ps aux
> ps: proc size mismatch (got 17484 total, kinfo_proc: 560)
> [root@stingray]~# 
> Top and w still work, and an older copy of ps (week or so) still works
> fine, so I conclude ps itself broke.

I see the same thing out of ps, plus top is broken for me ("Out of memory").

> 2) I can finally get a kernel to compile and link again without
> anything missing, but when I try to boot it, it just gets most of the
> way through the autoconfig and then spontaneously reboots.  I notice
> some of the characters are garbled on the screen too... *shrug*
> Config file appended (kernel from Dec 19 still works fine):

I haven't seen this, but I haven't actually tried the Dec 24 kernel yet.
I do have my own "problem number two" to add, though:  gawk appears to be
acting strangely, which I think is resulting in /bin/sh getting built wrong
(oh, joy).  The obvious symptom I see is that builtins.h and builtins.c,
which get created in /usr/src/bin/sh/obj/, look like this:

	#define BGCMD 0
	#define  1
	#define NEXTCOMMAND 2
	#define  3
	int bgcmd();
	int  ();
	int nextcommand();
	int  ();
respectively (sorry about not remembering the literal contents).  Once I
edit those two files to make sense, I then find that sh misinterprets shell
scripts (usually whining about `` "esac(NEWLINE)" unexpected'', yes there
is a newline in the command name it prints, and it's usually complaining
about an "if" statement).

Since I pick up changes by ftp, it's not inconceivable that I've just missed
some important library change; I'm going to pick up fresh and complete sources
today and try again.