Nathan J. Williams | 1 Aug 16:34 2006

Re: crunch binary for miniroot

Izumi Tsutsui <tsutsui <at> ceres.dti.ne.jp> writes:

> Is there any reason why sun3 miniroot just use binaries
> (including shlibs) from DESTDIR rather than to use crunchgen(1)
> to create smaller binaries without shlibs?

I think the history of the sun3 port didn't make it necessary to
shrink the miniroot very much; both the tape one was presumed to be
loading from and the swap partition one was loading into were
relatively large. Admittedly, that predated our dynamic-everything
userland.

> If there is no particular reason, I'll commit the attachd changes
> (which create crunched binaries), so that we'll get smaller image
> and no longer have to sync shlib majors in the list file.

Looks okay.

        - Nathan

schaecsn | 1 Aug 19:32 2006
Picon
Picon

Re: Old Sun 3 hardware

Hi folks,

I have a sun 3/110 including monitor (part number 365-10000-01) to give away for free in Santa Clara, CA.

It worked the last time I booted it - maybe 4 years ago. The monitor is a little bit fuzzy, but one can
definately read from it.

  LG Stefan

> From: Michael Bergman <eclectic <at> mit.edu>
> Precedence: list
> 
> Hi, folks.  I have a couple of empty Sun3 chassis, and also some VME 
> drive control boards and matching drives (in their matching 
> floor-standing cabinets).  Also an old, somewhat dim, but working 
> (when last used, several years ago) color "high res" 21" monitor.
> 
> Worth a separate line, the chassis for a SunArray.  Stands about 6' 
> tall, 3' wide (though it's a standard 19" rack on the inside).  On 
> wheels.
> 
> I also have most or all of a Cognex 4400 VME-based vision system, and 
> a lot of floppies with the development system, as well as 
> documentation.  No camera, but I think it will work with any standard 
> NTSC camera.  Some of the s/w and docs seem to be for a DOS based 
> system; I didn't read through it all to make sure that I have stuff 
> that matches the hardware, but I think I do and that the DOS stuff 
> was for the successor system (the 5x00 series).
> 
> All this stuff is in Waltham.  The only stuff I am *urgently* trying 
(Continue reading)

Martin Etteldorf | 1 Aug 20:28 2006

Re: Old Sun 3 hardware

On 8/1/06, schaecsn <schaecsn <at> gmx.net> wrote:
> Hi folks,
>
> I have a sun 3/110 including monitor (part number 365-10000-01) to give
> away for free in Santa Clara, CA.

I'd be interested, but I assume shipping to good ole yurop will cost a fortune.

Martin

schaecsn | 1 Aug 20:54 2006
Picon
Picon

Re: Old Sun 3 hardware

Hi folks,

> > I have a sun 3/110 including monitor (part number 365-10000-01) to give
> > away for free in Santa Clara, CA.
> 
> I'd be interested, but I assume shipping to good ole yurop will cost a fortune.

I never expected someone asking for shipping. This thing is heaaaavy. I would prefer pick-up. Iff  no one is
interested in pick up then I consider to consider shiping it.

  LG Stefan

Christos Zoulas | 1 Aug 21:06 2006

Re: crunch binary for miniroot

In article <mtuu04wwmn7.fsf <at> contents-vnder-pressvre.mit.edu>,
Nathan J. Williams <nathanw <at> wasabisystems.com> wrote:
>Izumi Tsutsui <tsutsui <at> ceres.dti.ne.jp> writes:
>
>> Is there any reason why sun3 miniroot just use binaries
>> (including shlibs) from DESTDIR rather than to use crunchgen(1)
>> to create smaller binaries without shlibs?
>
>I think the history of the sun3 port didn't make it necessary to
>shrink the miniroot very much; both the tape one was presumed to be
>loading from and the swap partition one was loading into were
>relatively large. Admittedly, that predated our dynamic-everything
>userland.
>
>> If there is no particular reason, I'll commit the attachd changes
>> (which create crunched binaries), so that we'll get smaller image
>> and no longer have to sync shlib majors in the list file.
>
>Looks okay.
>
>        - Nathan
>

Let's do it for consistency's sake.

christos

Izumi Tsutsui | 2 Aug 12:01 2006
Picon

Re: crunch binary for miniroot

christos <at> astron.com wrote:

> >> If there is no particular reason, I'll commit the attachd changes
> >> (which create crunched binaries), so that we'll get smaller image
> >> and no longer have to sync shlib majors in the list file.
> >
> >Looks okay.
> >
> >        - Nathan
> >
> 
> Let's do it for consistency's sake.
> 
> christos

Committed. Thanks for your comments.
---
Izumi Tsutsui

Dream | 26 Aug 04:42 2006

www.thuonghieuviet.com good website of Vietnamese

look for web us:

www.thuonghieuviet.com
http://tendonvi.thuonghieuviet.com. 
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Connection in connection, you will create a series of global commerce.

OPEN LETTER

This is a e-commerce website of high-quality electronic products of Vietnamese, please fill out and send
us, we will connect to and answer
by email for you, all for free ! 

-	Your login name:
-	Company name:
-	Your business (or the owner)::
-	Address:
-	Phone number:
-	Fax number:
-	Contact:
-	Your web:
-	Email:
Please send us a logo with size <50kB.(100 X100 pixel )
For your benefit, for the development of globalizing commerce. We wish you would be interested in this mail.

Thankyou.
-------------------------------------------------------------------------------------------------------------------------------------------------------------
*** please contact  us : hoahongtrang_moden2005 <at> yahoo.com; btvvietnam <at> hcm.vnn.vn 

(Continue reading)

Izumi Tsutsui | 31 Aug 08:32 2006
Picon

wired entry count problem on sun3x pmap

Today I've manage to track the real problem on sun3x pmap
which causes "processes can't get memory" problem and
"panic: get_a_table: out of A tables." (and also B, C tables)
panics during pkgsrc/lang/perl5 build etc.

It isn't a simple resource shortage but bad wired entry accounting:

- pmap_enter(9) doesn't adjust wired counts on wired -> unwired changes
- pmap_remove(9) doesn't adjust wired counts even if removed pages
  have wired mappings

So any tables whose pages are unwired by these functions are never
put back into the active queues, and it causes resource shortage.

These possible problems are already noted in comment. It says
"This may be ok if pmap_unwire() is the only interface
 used to UNWIRE a page" but I guess it's false in these days.

The attached patch seems to fix the panics,
but I'd appreciate comments or suggestions from VM gurus.
---
Izumi Tsutsui

Index: sun3x/pmap.c
===================================================================
RCS file: /cvsroot/src/sys/arch/sun3/sun3x/pmap.c,v
retrieving revision 1.90
diff -u -r1.90 pmap.c
--- sun3x/pmap.c	10 May 2006 06:24:03 -0000	1.90
+++ sun3x/pmap.c	31 Aug 2006 06:04:51 -0000
(Continue reading)

Jeremy Cooper | 31 Aug 11:59 2006
Picon

Re: wired entry count problem on sun3x pmap

Izumi, thank you very much for your hard work finding this bug, and 
especially for working to keep the comments in sync!

-J

On Thu, 31 Aug 2006, Izumi Tsutsui wrote:

> Today I've manage to track the real problem on sun3x pmap
> which causes "processes can't get memory" problem and
> "panic: get_a_table: out of A tables." (and also B, C tables)
> panics during pkgsrc/lang/perl5 build etc.


Gmane