OpenBSD Journal

[h2k9]: Hardware Hackathon Summary

Contributed by jason on from the suspending-disbelief-and-laptops dept.

Here is the story about h2k9 (the 2009 Hardware Hackathon) and some of the results of that week.

Pedro Almeida tells the story about how the h2k9 came to be and how it went down.

Since 2006, I've been running the logistics for the hardware hackathons. h2k6 was a bit stressful for me since I didn't know how these kind of events used to work, and I didn't want anything to go on the wrong direction. At the time, some developers help was crucial to make the right choices.

I want to thank the Physics Department of the University of Coimbra, particularly the Computational Physics Center and to the Advanced Computation Laboratory, who gently have hosted us since h2k6.

Please read on for the rest of the story:

openbsdcoffemachine

The preparation of such events is becoming easier every year. On the 20th of November, all was prepared to host another hackathon, as you can confirm on this image.

This year, in a no particular order, the following developers attended to the hackathon: Takuya Asada (syuu@), Paul Irofti (pirofti@), Claudio Jeker (claudio@), Dariusz Swiderski (dms@), Theo de Raadt (deraadt@), Henning Brauer (henning@), Jonathan Gray (jsg@), Miod Vallat (miod@), Owain Ainsworth (oga@), Matthieu Herrb (matthieu@), Kenneth R Westerback (krw@), Mike Larkin (mlarkin@), Mark Kettenis (kettenis@), Yojiro Uo (yuo@) and Joel Sing (jsing@).

All week developers worked hard in and outside the hack room, and occasionally they had some coffee brainstorms.

coffeebrainstorm

Commits start flowing (even on Friday!) and here is a small list of the work done this year. I apologize if something is missing. The information is grouped by similar interests or work.

syuu@ and jsing@ have been working together on improving mips64 support. They have been working on getting SMP support on mips64, using an SGI Octane as a first target. This experience will sure help in making it easier to SMP working on more platforms.

pirofti@ and mlarkin@ did a huge work on APCI. The work they did, was to get ACPI suspend and resume working. Yehey!

hackroom01

dms@ have been working the Intel EP80579 SoC, on the recent added gcu(4) and improving the PowerPC support, so we will be able to run OpenBSD on some off the new/interesting RouterBoard boards. He also fixed a number of other em(4) variants.

deraadt@ and claudio@ started their work preparing the move and moving the (R.I.P.) brconfig functionality into ifconfig. This was magnificent because now we have all the network configuration functionality on one tool only. They also fixed the bridge(4) code so that it correctly captures and accounts packets in all cases (be it physical Ethernet, vether(4) or gif(4)).

hackroom03

claudio@ also fixed a few drivers (msk(4) and ep(4)) and made bnx(4) work on big-endian and non 4k page-sized systems like sparc64. Then he started to look into dynamic TX ring sizing. Bringing the MCLGETI() idea to the transmit side.

deraadt@, as usual, was flying around almost of the work done by the other developers, making suggestions and corrections, and had spent a lot of time on ACPI un-suspend.

henning@ has been in many places in the network stack, especially in the routines on the input side, improved pf(4) and improved his Portuguese. Now he can order any number of coffees or beer, from one to seven!

hackroom02

jsg@ make amd64 SpeedStep work when we don't know the bus clock and have the tables from ACPI, probe for modes on SATA controllers, made CF behind SATA work and some SSDs faster, looked at suspending/resuming pciide and ACHI, added a bunch of USB devices and started looking at ninja ata32 cf CardBus support. He also spent some time tracking down codegen bugs in pcc as well.

hackroom04

oga@ made an initial port of the DRM code needed to make R600 and R700 Radeon chipsets work with 2 acceleration and XVideo support under XD. His most time consuming work was to hack on DRM memory management support for the inteldrm drivers (using the GEM (graphics execution manager) model that was written at Intel, with some tweaks). This work has the potential to make X on those chips a LOT more robust as well as allowing for more OpenGL support and faster acceleration.

matthieu@, as usual, worked on improving xenocara, our preferred X11 distribution. He worked mostly on the xf86-input-ws driver for touchscreens and other wscons input devices, adding support for XInput device properties. This allowed matthieu@ to rewrite xtsscale(1), the calibration program for touchsreeens to be a lot more useful (no restart of X needed).

lemotelaptop

krw@ was working on fixing a problem with a particular disk on siop, so the SCSI midlayer rewrite by dlg@ can go back in. The fix, eliminating the use of ordered tag queuing, will probably fix other disks on siop too.

miod@ worked on porting OpenBSD to the Lemote Yeelong netbook, based upon the Chinese Loongson 2F Mips processor clone.

kettenis@ worked on suspend/resume, adding missing bits to some drivers, making the lid work, and preparing the way for SMP machines. He also worked on improving support for hot-plugging PCI devices on machines where the firmware doesn't set up PCI-PCI bridges properly.

temphum

With miod@ he made the ps2-behind-usb keyboard used in Naturetech sparc64 laptops work. Several models can be considered fully supported now.

h2k9isover

yuo@ added add NetMos NM9901, new umsm type modems, and released the uthum(4) driver, which supports real cheap temperature and higrometer USB sensors. He has also spent substantial time researching usb un-suspend not working on some machines, and improved usb stack in ACPI suspend/resume.

Luis Pinto, Rodolfo Gouveia and myself (Pedro Almeida), drank a lot of coffee so the h2k9 T-Shirt makes more sense now!

Well, Rodolfo start making the port of gpodder and Luis got the accelerometer calibration working on IBM/Lenovo laptops, so the only real coffee slacker was me!

Now h2k9 is over and we hope to host more hackathons on the future. See you in h2k10!

(Comments are closed)


Comments
  1. By Brynet (Brynet) brynet@gmail.com on

    Are there more pictures? :-)

    Comments
    1. By ers (ers) on -

      > Are there more pictures? :-)

      Yes, where are the pictures?
      Not only for h2k9 but also other hackathons?
      Past editions were available but for some time now, there's nothing...

  2. By Venture37 (venture37) on http://www.geeklan.co.uk

    "dms@ have been working on the Intel EP80579 SoC"

    s/have/has

    "deraadt@, as usual, was flying around almost of the work done by the other developers"

    s/almost/most ????


    Well done guys!

    Comments
    1. By Edd Barrett (edd) on http://theunixzoo.co.uk

      Great work guys. The network card in my SGI box now works :)

  3. By Nick Rivera (nrivera) nckrvr@gmail.com on

    "pirofti@ and mlarkin@ did a huge work on APCI"

    Atmospheric Pressure Chemical Ionization?
    I sure it must be ACPI.
    Thanks guys for another great bunch of work.

    Comments
    1. By Markus Peloquin (incripshin) on http://cs.wisc.edu/~markus

      > "pirofti@ and mlarkin@ did a huge work on APCI"
      >
      > Atmospheric Pressure Chemical Ionization?
      > I sure it must be ACPI.
      > Thanks guys for another great bunch of work.

      But it says ACHI. More likely AHCI (advanced host controller interface)? I don't *really* know what these things are, so context doesn't help me.

  4. By Benoît Chesneau (benoitc) bchesneau@gmail.com on http://benoitc.im

    I see some mac on pictures. Are they nvidia machines ? Which driver do you use for it then under X11 ?

    Comments
    1. By Brynet (Brynet) on

      > I see some mac on pictures. Are they nvidia machines ? Which driver do you use for it then under X11 ?

      I doubt any of them would intentionally buy machines with 'nvidia' graphics, but if they did.. there is only 1 driver they could use.

      xf86-video-nv(4).

      Obviously.

      Comments
      1. By Anonymous Coward (anon) on

        > > I see some mac on pictures. Are they nvidia machines ? Which driver do you use for it then under X11 ?
        >
        > I doubt any of them would intentionally buy machines with 'nvidia' graphics, but if they did.. there is only 1 driver they could use.
        >
        > xf86-video-nv(4).
        >
        > Obviously.

        Or vesa, or maybe wsfb if it's a powerpc Mac.

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.]