From owner-freebsd-hackers  Sun Apr  7 11:56:47 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id LAA25687
          for hackers-outgoing; Sun, 7 Apr 1996 11:56:47 -0700 (PDT)
Received: from gw2.att.com (gw2.att.com [192.20.239.134])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id LAA25682
          for <hackers@freefall.freebsd.org>; Sun, 7 Apr 1996 11:56:45 -0700 (PDT)
From: dob@nasvr1.cb.att.com
Received: from nasvr1.cb.att.com (naserver1.cb.att.com) by ig1.att.att.com id AA17664; Sun, 7 Apr 96 14:54:10 EDT
Received: by nasvr1.cb.att.com (5.x/EMS-1.1 Sol2)
	id AA29112; Sun, 7 Apr 1996 14:57:08 -0400
Received: from cbsky.cb.att.com by nasvr1.cb.att.com (5.x/EMS-1.1 Sol2)
	id AA29108; Sun, 7 Apr 1996 14:57:05 -0400
Received: by cbsky.cb.att.com (5.x/EMS-1.1 client.cf 1/8/94 (SMI-4.1/SVR4))
	id AA11581; Sun, 7 Apr 1996 14:56:36 -0400
Date: Sun, 7 Apr 1996 14:56:36 -0400
Message-Id: <9604071856.AA11581@cbsky.cb.att.com>
To: hackers@freefall.freebsd.org
Subject: Re: pgcc and kernels..
Sender: owner-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

> From: Tony Kimball <alk@Think.COM>
> Date: Sun, 7 Apr 1996 03:25:42 -0500
> Subject: Re: pgcc and kernels..
> 
>    Date: Sat, 06 Apr 1996 16:23:07 -0800
>    From: "Jordan K. Hubbard" <jkh@time.cdrom.com>
> 
>    You must have done more than that, since our stock libgcc doesn't
>    support some of the inlines that pgcc generates.
> 
> Ah, to mention it, I also added sys/libkern/muldi3.c to the sources.  Thereafter,

Didn't readily find the source for muldi3.c so I just grabbed a copy of the 
_muldi3.o file from the pgcc libgcc library, stuffed it into the kernel 
compile directory, and hacked the Makefile to link in _muldi3.o 

> ; make kern_clock.o clock.o CC=kcc # use 2.6.3

This didn't work for me.  Couldn't find kcc, so I just gave it CC=cc.

> ; make CC=gcc COPTFLAGS=-O2 # use pgcc

pgcc was installed from ports as /usr/local/bin/gcc, 
so I used CC=/usr/local/bin/gcc, instead.

Kernel compiles and links ok but the resulting kernel panics out of 
_scsi_attach during probe phase.  

What other voodoo do I need to chant?  Or was my variation of this voodoo
chanted incorrectly?


Thanks,

Dan O'Brien
Lucent Technologies (Bell Labs)
Columbus, Ohio, USA