Subject: Re: Shared libraries re-enabled for VAX
To: Michael L. Hitch <mhitch@lightning.msu.montana.edu>
From: Johnny Billquist <bqt@softjar.se>
List: port-vax
Date: 04/08/2007 14:47:40
This is a multi-part message in MIME format.
--------------030203060403040301000400
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Michael L. Hitch wrote:
> On Sat, 7 Apr 2007, Johnny Billquist wrote:
> 
>> Matt Thomas wrote:
> 
> ...
> 
>>> I believe the gcc codegen problems have been fixed.
>>
>> Whatever the problem is, it still exists.
>> I noted that you checked in some new stuff for GCC, as well as the 
>> mutex fixes Michael posted about, Matt. Things look good when booting 
>> up, but as soon as I try to do something heavy on the disk, the cpu 
>> time runs up to 100% system time, and the system hangs.
>> Something is still bad, which was introduced after the 15 of March, 
>> and Michael seems to have narrowed it down to something in GCC... 
>> Hmmm... Any ideas on how to diagnose this?
> 
>   Matt and I have been working on this over the last couple of days, and 
> his fix to src/gnu/dist/gcc4/gcc/config/vax/vax.c (revision 1.9) fixes 
> the excessive kernel time when I was doing file system writes.

Sweet. Yes, now it works fine!

>   I've been running the new kernel for 15 hours now, and haven't seen any
> excessive kernel time when writing files [I did several copies of the 
> binary/sets from several builds, and it also survived the daily and 
> weekly cron jobs this morning].

No. All seems well in that department now.

So let's go back to the other problems we have.

1) Running gcc natively don't work. The result is:
Gnat:bqt/src> gcc -o hello hello.c
/var/tmp//ccT5yaE4.s: Assembler messages:
/var/tmp//ccT5yaE4.s:12: Warning: Symbol used as immediate operand in 
PIC mode.
/var/tmp//ccT5yaE4.s:13: Fatal error: Junk at end of expression "LC0"

This from a simple hello world program. Anyone should be able to 
reproduce this.

2) Matt, have you looked at why the built-in finite() in gcc always 
expands to false for the VAX? (This is the reason top and so on gives 
illegal instruction).


Anyways. Good work guys! We're moving forward. Looking forward to when I 
can try a native build of the whole system again... :-)

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt@softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol

--------------030203060403040301000400
Content-Type: text/x-vcard; charset=utf-8;
 name="bqt.vcf"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
 filename="bqt.vcf"

begin:vcard
fn:Johnny Billquist
n:Billquist;Johnny
email;internet:bqt@softjar.se
tel;cell:+46 70-8641486
x-mozilla-html:FALSE
version:2.1
end:vcard


--------------030203060403040301000400--