OpenBSD Journal

Call for Testers

Contributed by jose on from the quality-assurance dept.

We're heading towards the final 3.4 release cut, but we need to test stuff. One of the ways you can help is to download the new snapshots and test them out. From Wim Vandeputte , expressing requests from Theo, Pval and others:

At the request of the OpenBSD developers, you are invited to test the latest snapshots on the ftp server to verify a last minute fix in the boot blocks (version 2.02). Please mail your report to: henning@openbsd.org when you test: something like

i386:Serverworks SLC,P42.4:CD34.iso+FTP install(VERSION 2.02
boot block):WORKS:X GOOD (ATI RADEON 7500)
Be sure to say arch, something about what it is, what install method you used, and be sure to indicate whether or not you had the VERSION 2.02 boot block. Test X if you can and say if you did.

Also start testing packages (as opposed to ports), since those also go on the CD. Thanks!

(Comments are closed)


Comments
  1. By Jedi/Sector One () j@pureftpd.org on http://www.skymobile.com/

    The "Vid" package doesn't work at all for me with a Creative Webcam Pro. The same package works perfectly on NetBSD and FreeBSD. I really don't know what could be wrong. I even copied the latest uhci code from NetBSD and it didn't change anything.

    Comments
    1. By Anonymous Coward () on

      Yes, we remember from when you filed a pr.

    2. By Wim () on

      Hmmm I have it working here no problem:

      vid -d /dev/ugen0.00 | xv -
      crw-rw---- 1 root wheel 63, 0 Dec 17 2002 /dev/ugen0.00
      lrwxr-xr-x 1 root wheel 8 Jan 9 2003 /dev/ugen0.00.1@ -> ugen0.01
      crw-rw---- 1 root wheel 63, 1 Dec 17 2002 /dev/ugen0.01


      ugen0 at uhub2 port 1
      ugen0: OmniVision OV511+ Camera, rev 1.00/1.00, addr 4

  2. By chill () on

    It figures they're changing boot blocks now!

    I have a Soekris 4501 and it has problems booting off some CF cards because those cards can't handle multisector transfers during boot.

    There are patches for the FreeBSD and NetBSD bootloader to have pbr.S and biosboot.S fixed to handle only single sector transfers. There was no such patch for OpenBSD that I could find, and my assembly skills were developed on a CBM 6502 and IBM 4381 -- not up to par.

    I just cajoled the individual who did the NetBSD patch to do one for OpenBSD and he e-mailed it over. I've been working on that for the last two days...

    and NOW they change the code! Argh!

    Comments
    1. By Federico () fgsch@lodoss.net on mailto:fgsch@lodoss.net

      Fill a PR and will take a look. BTW, have you checked if 3.4 solves the problem for you?

      Comments
      1. By chill () on

        I was using the 3.4 iso and snapshots from about a week ago, so 3.4 itself doesn't help.

        I'm d/ling the new snapshots and will test later today and fill out reports.

        I have another issue (machine not booting OpenBSD if anything USB is plugged in, but USB stuff works if plugged in after successful boot). I'll file the reports, etc. once I test with the latest snapshots.

    2. By tedu () on

      the change (in both cases) was fairly minor. it's not like anybody rewrote the bootblocks just for kicks the day before release.

    3. By Brian () on

      Between this issue and issues with many CF readers, getting soekris boxes to boot can be a real pain. My company went thru two cf readers and three cf cards before we got a working combination and a bootable unit.

      That said, I haven't had any problems with getting the cards we use to boot 3.4 with the new bootloader.

      Comments
      1. By Robert Granger () robertgranger AT yahoo DOT com on mailto:robertgranger AT yahoo DOT com

        Brian,

        Any chance you would be willing to share that working combo of soekris and CF cards. I am looking at the soekris 4801 and a 1gb SanDisk card.

        Thanks for sharing

        Rob

        Comments
        1. By Nate Tobik () ntobik@hotmail.com on mailto:ntobik@hotmail.com

          Rob

          I worked on some Soekris boxes for a while earlier this summer. They were a pain, we couldn't get the CF to boot right, what we did was do a PXE boot with linux and the Soekris booted over the network then loaded it's file system on NFS. It worked great, then we created the filesystem on the CF/Microdrive and rebooted w/o the PXE boot. The Soekris box booted off the CF without a problem. We did this in linux, we had to compile a custom kernel for it, just remember to compile in support for the processor the Soekris has the (Elan).

          As for the CF cards, we used SanDisk 256, worked great, and a IBM 1gig Microdrive, also worked great. Don't know about readers because we netbooted. Hope this helps.

          Nate

          Comments
          1. By markus () markus@openbsd.org on mailto:markus@openbsd.org

            strange, i never had problems with various
            CF writers (cardbus) and different CFs, but
            i mostly use SanDisk. only thing you need
            to make sure you use the same CHS mapping as
            the soekris bios, when you are creating the
            bootblock. newer soekris bioses print
            the mapping, so this gets trivial.

        2. By Anonymous Coward () on

          For the reader, we use a sandisk pcmcia cf reader. We have a usb sandisk imagemate that also works, but it gets weird after writing a few cards and the box has to be rebooted.

          For the CF, we use 32Mb Memorex or 128Mb "Mr. Flash" cards, depending on the application.

    4. By mirabile () mirabile@bsdcow.net on http://MirBSD.BSDadvocacy.org/

      Weird. This (not being able to handle multiblock
      requests during [CD-ROM] boot) seems to be the
      reason for a large number of "bad magic" during
      CD-ROM install I've seen in the last couple of
      weeks. The BIOSes of many notebooks are broken in
      that respect, too.

      You are making me go through my rewritten boot
      code again, you know that?

      I'm unclear about how to handle it. Holding
      the SHIFT key while booting enables single-block
      transfers? [does this work on soekris, do they
      have the normal BIOS functions and are these
      actually functional? please drop me a mail]

      A different solution would be to go for single-
      block transfers after bad magic.

      The easiest solution would be a different boot
      sector, but...

      Comments
      1. By mirabile () mirabile@bsdcow.net on http://mirbsd.bsdadvocacy.org/active/cvsweb.cgi/sr

        Okay, I went through these, and I hope the result
        works for all of you.

        See this posting to misc@ for reference:

        http://marc.theaimsgroup.com/?m=106348283112951

    5. By Wim () wim@kd85.com on http://soekris.kd85.com

      Wel, in any case, the same problem is still present with the CF booting, I hope somebody will
      finally make a sector read instead of byte read modification to the bootloader...

      The most reliable CF I found so far is the Kingston CF/64-s (not the CF/64) and the Kingston CF/32.

      The 2.02 bootblocks still work fine on those:
      Using Drive: 0 Partition: 3
      reading boot......
      probing: pc0 com0 com1 pci mem[639K 63M a20=on]
      disk: hd0+
      >> OpenBSD/i386 BOOT 2.02
      switching console to com0
      com0: changing speed to 19200 baud
      com0: change your terminal to match!
      com0: will change speed in 5 seconds....
      com0: console is at 19200 baud
      boot>
      booting hd0a:/bsd:
      4462484|
      +838332 [58+204144/
      +181627-
      ]=0x56c714
      entry point at 0x100120
      [ using 386244 bytes of bsd ELF symbol table ]
      Copyright (c) 1982, 1986, 1989, 1991, 1993
      The Regents of the University of California. All rights reserved.
      Copyright (c) 1995-2003 OpenBSD. All rights reserved. http://www.OpenBSD.org

  3. By Jeffrey () on

    I'm just curious if you really *have* to use the snapshots to generate valuable results for developers. The reason I ask is because dial-up simply does not allow me to spend hours on the phone downloadng...

    This is especially true when I have perfectly good source trees to build from, and `make build' takes an hour or two. I never do snapshots for this very reason. Granted, I'll happily do snapshots when it is required... For example when i386 went ELF =)

    So, can anyone give some insight as to the relative value of build vs. snapshot download..?
    Of course, testing packages over ports is perfectly understandable though!

    Comments
    1. By Anonymous Coward () on

      i think its just for testing that the install process goes smoothly, eg for things like cross device hard links that you probably wouldnt pick up from a make build etc etc.

      sometimes i install a snapshot over make build just for that nice clean fresh install feeling. who says we dont know how to party?

    2. By tedu () on

      snapshots on ftp are not always built from the cvs sources available.

      Comments
      1. By Anonymous Coward () on

        to expand on this. Snapshots can be cvs + non-committed patches.

Credits

Copyright © - Daniel Hartmeier. All rights reserved. Articles and comments are copyright their respective authors, submission implies license to publish on this web site. Contents of the archive prior to as well as images and HTML templates were copied from the fabulous original deadly.org with Jose's and Jim's kind permission. This journal runs as CGI with httpd(8) on OpenBSD, the source code is BSD licensed. undeadly \Un*dead"ly\, a. Not subject to death; immortal. [Obs.]