Re: Illegal instruction trap in libgcc_s.so

["William D. Jones" <thor0505%comcast.net@localhost>]

I'm not sure what the difference between my configuration and Jarle's is, 
but something is different between our configurations. I brought up my local 
copy of the NetBSD tree and it's not patched. I then took a look at the CVS 
repository itself, and noticed that cpuinfo.c hasn't been changed in 15 
months- this was before I started using NetBSD on a real 486 at all.

So even if that libgcc has code that unconditionally calls CPUID, that 
particular code path never executes on my 486 machine (and I know it will 
panic because back in January, I decided to upgrade and to my chagrin 
someone forgot to put a 486 guard for CPUID in the kernel :P). Perhaps it's 
(supposed to be) dead code based on 486 guards elsewhere in libgcc, that 
either are not firing during runtime or while libgcc was compiled?

-----Original Message----- 
From: Martin Husemann
Sent: Monday, June 01, 2015 3:12 AM
To: Jarle Greipsland
Cc: thor0505%comcast.net@localhost ; port-i386%netbsd.org@localhost
Subject: Re: Illegal instruction trap in libgcc_s.so

On Mon, Jun 01, 2015 at 08:25:50AM +0200, Jarle Greipsland wrote:
> NetBSD bug tracker:
>   http://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=49741

This one is tricky as it is an upstream gcc bug - can you bring it up
in their bugtracker? (Which does not mean we can't commit a local workaround
for now...)

Martin

--
William D. Jones
thor0505%comcast.net@localhost