A. Wik | 3 May 22:57 2009

kernel compilers


Does anyone know which compilers are capable - or incapable -
or building different versions of the NetBSD/vax kernel.
I'm particularly interested in 1.5.2, because that's the
one I've been working on.

A while back I reverted to gcc-2.7.2.3 because of the
horrible resource consumption[1] AND the internal errors[2]
of egcs (1.1.1 according to --version, but 2.91.60
according to the opinion of -dumpspecs).

My newly rebuilt kernel crashes after reporting that the
filesystem is of type FFS.  See note [3] below for details.

-aw

[1]  Indeed, I had to use ulimit for the first time ever (I've
     used it before, but that was for curiosity.).

[2]  It repeatedly died attempting to build a part of Samba.
     I now have Samba running on the VAX, so at least in this
     instance, gcc-2.7.2.3 turned out to be a more capable
     compiler (not that I'm particularly surprised - newer
     versions do little but increase wastefulness, reduce
     compatibility and introduce new bugs in the name of the
     holy GCC religion "optimisation" - what kind of
     optimisation is it if it doesn't run at all or with
     significantly lower reliability, or if it runs on a
     greatly reduced subset of hardware, or otherwise only
     under certain special conditions?
(Continue reading)

Michael L. Hitch | 5 May 19:05 2009

Re: (nearly) everything works on a Vaxstation 4000/90

On Mon, 30 Mar 2009, Bella Riosa wrote:

> a big thank you to everyone!
> NetBSD5-RC3 is running very well on my VaxStation. I've built many packages already and nearly
everything works! I've found only two bugs:
>
> 1. the groff bug, while building world (is anyone working oh this?)

   I finally took some time to dig into this, and believe I have fixed it. 
It should be fixed in -current, and pullups are pending for the netbsd-5 
branches.  [I had kind of  hoped to have gotten it fixed for the 5.0 
release, but missed that.]

--
Michael L. Hitch			mhitch <at> montana.edu
Computer Consultant
Information Technology Center
Montana State University	Bozeman, MT	USA

Dave McGuire | 5 May 20:02 2009

Re: (nearly) everything works on a Vaxstation 4000/90

On May 5, 2009, at 1:05 PM, Michael L. Hitch wrote:
>> a big thank you to everyone!
>> NetBSD5-RC3 is running very well on my VaxStation. I've built many  
>> packages already and nearly everything works! I've found only two  
>> bugs:
>>
>> 1. the groff bug, while building world (is anyone working oh this?)
>
>   I finally took some time to dig into this, and believe I have  
> fixed it. It should be fixed in -current, and pullups are pending  
> for the netbsd-5 branches.  [I had kind of  hoped to have gotten it  
> fixed for the 5.0 release, but missed that.]

   Was this different from the floating point problem that Brad  
nailed a few months ago??

             -Dave

--

-- 
Dave McGuire
Port Charlotte, FL

Michael L. Hitch | 5 May 19:34 2009

Re: (nearly) everything works on a Vaxstation 4000/90

On Tue, 5 May 2009, Dave McGuire wrote:

> Was this different from the floating point problem that Brad nailed a few 
> months ago??

   The native build groff bad relocations had nothing to do with floating
point, so I'd say it was completely unrelated.

--
Michael L. Hitch			mhitch <at> montana.edu
Computer Consultant
Information Technology Center
Montana State University	Bozeman, MT	USA

Dave McGuire | 5 May 20:25 2009

recent fixes for VAX


   I know that a great many VAX port problems were fixed recently,  
bringing NetBSD/vax back into the realm of usefulness.  Did those  
fixes make it into the recent 5.0 release?  Meaning, if I'm about to  
put together a VAX on which to run NetBSD, would I be better off  
starting with the 5.0 release, or -current?

             -Dave

--

-- 
Dave McGuire
Port Charlotte, FL

Michael L. Hitch | 5 May 19:54 2009

Re: recent fixes for VAX

On Tue, 5 May 2009, Dave McGuire wrote:

> I know that a great many VAX port problems were fixed recently, bringing 
> NetBSD/vax back into the realm of usefulness.  Did those fixes make it into 
> the recent 5.0 release?  Meaning, if I'm about to put together a VAX on which 
> to run NetBSD, would I be better off starting with the 5.0 release, or 
> -current?

   As far as I know, all the recent fixes with the exception of the one I 
just committed yesterday should be in the 5.0 release.  The one exception 
is the one that causes problems doing a native build (a couple of the 
toolchain groff programs get build with a bad relocation item, causing 
them to abort).  Restarting the build (multiple times, unfortunately) 
would be sufficient to eventually finish a full build.

   That last fix has now been pulled up into the netbsd-5 branch, so the 
next autobuild should have the bad relocation fixed.

   [Note:  I think that /usr/bin/troff in the NetBSD 5.0 release probably 
has the bad relocation problem as well, which will cause problems trying 
to build something that uses troff.]

--
Michael L. Hitch			mhitch <at> montana.edu
Computer Consultant
Information Technology Center
Montana State University	Bozeman, MT	USA

Dave McGuire | 5 May 20:22 2009

Re: (nearly) everything works on a Vaxstation 4000/90

On May 5, 2009, at 1:34 PM, Michael L. Hitch wrote:
>> Was this different from the floating point problem that Brad  
>> nailed a few months ago??
>
>   The native build groff bad relocations had nothing to do with  
> floating
> point, so I'd say it was completely unrelated.

   Ahhhh ok.

         -Dave

--

-- 
Dave McGuire
Port Charlotte, FL

Dave McGuire | 5 May 20:38 2009

Re: recent fixes for VAX

On May 5, 2009, at 1:54 PM, Michael L. Hitch wrote:
>> I know that a great many VAX port problems were fixed recently,  
>> bringing NetBSD/vax back into the realm of usefulness.  Did those  
>> fixes make it into the recent 5.0 release?  Meaning, if I'm about  
>> to put together a VAX on which to run NetBSD, would I be better  
>> off starting with the 5.0 release, or -current?
>
>   As far as I know, all the recent fixes with the exception of the  
> one I just committed yesterday should be in the 5.0 release.  The  
> one exception is the one that causes problems doing a native build  
> (a couple of the toolchain groff programs get build with a bad  
> relocation item, causing them to abort).  Restarting the build  
> (multiple times, unfortunately) would be sufficient to eventually  
> finish a full build.
>
>   That last fix has now been pulled up into the netbsd-5 branch, so  
> the next autobuild should have the bad relocation fixed.
>
>   [Note:  I think that /usr/bin/troff in the NetBSD 5.0 release  
> probably has the bad relocation problem as well, which will cause  
> problems trying to build something that uses troff.]

   Excellent, thanks for the info!  Now I just have to choose a  
machine.  I'm sorta torn between a 4000-90 (fast!) and a 4000VLC  
(tiny!).  I'll probably go with the -90.

          -Dave

--

-- 
Dave McGuire
(Continue reading)

Dave McGuire | 6 May 00:08 2009

Re: recent fixes for VAX

On May 5, 2009, at 6:03 PM, Patrick Finnegan wrote:
>>    Excellent, thanks for the info!  Now I just have to choose a
>> machine.  I'm sorta torn between a 4000-90 (fast!) and a 4000VLC
>> (tiny!).  I'll probably go with the -90.
>
> Dave, you should realize that both of those are tiny VAXes...

   How very true, but I don't think NetBSD/vax runs on the '7000 yet. ;)

            -Dave

--

-- 
Dave McGuire
Port Charlotte, FL

Patrick Finnegan | 6 May 00:03 2009
Picon

Re: recent fixes for VAX

On Tuesday 05 May 2009, Dave McGuire wrote:
>    Excellent, thanks for the info!  Now I just have to choose a
> machine.  I'm sorta torn between a 4000-90 (fast!) and a 4000VLC
> (tiny!).  I'll probably go with the -90.

Dave, you should realize that both of those are tiny VAXes...

Pat
--

-- 
Purdue University Research Computing --  http://www.itap.purdue.edu/rcac


Gmane