#2 Mainboard detection at runtime

closed
General (7)
5
2003-03-04
2002-12-17
No

Hi,
The current way of having to pick a mainboard at compile time is
not very nice for setting up many different machines (can't just
copy around binaries), and it's very bad for Linux distributors (we
can't possibly know in advance what board we'll be run on).

I've added BIOS detection to the code, and made it pick the right
board etc. at runtime - I've also adapted guess to this way of
doing things.

Hope you find it useful.

Discussion

  • Sergei Haller

    Sergei Haller - 2002-12-17

    Logged In: YES
    user_id=76018

    high!

    I'd be more than happy to adapt such a feature. A while ago
    I started
    to write a config file parser (see readconf.c in CVS).
    Unfortunately I didn't
    have much time in the last few months. I'm looking forward
    to see your
    patch and the way you implemented it. But there's no
    attachment on the
    SF page.

    bye,
    Sergei.

     
  • Bernhard Rosenkraenzer

    Logged In: YES
    user_id=15538

    Hmm... guess sourceforge doesn't like Konqueror ATM. :/
    Re-Attaching using mozilla

     
  • Bernhard Rosenkraenzer

    Logged In: YES
    user_id=15538

    Ok, seems to be there now.

     
  • Bernhard Rosenkraenzer

    Logged In: YES
    user_id=15538

    Btw, this is relative to yesterday's CVS... Since you added
    new boards today, it doesn't apply to the current version.

    Unless you beat me to it, I'll attach a port to newer CVS
    later (but not right away, got some other work to do)

     
  • Bernhard Rosenkraenzer

    Logged In: YES
    user_id=15538

    Adding new version - synced with CVS as of a couple of
    minutes ago, and I've added a couple of mainboards I had
    floating around as well (Asus A7S333, SiS 630S, Chaintec
    CT-7AJA-2, etc.)

     
  • Nobody/Anonymous

    Logged In: NO

    Didn't take a deep look yet.

    in bios.c I'd like some more comments and at all places,
    where an error condition arises (all the returns), to print
    the reason (at least
    when run with the -d option) (at very least a comment, what
    happened here) -- since I'm not convinient with /dev/mem and
    the adresses you read, I can't do that myself.

    on my Biostar M6VCG biosinfo returns all (null) values.
    (so nvram-wakeup and guess result both in a segmentation fault)

    The other thing is more philosophical: a runtime detection
    of the
    board is a very nice thing. But if the board is not
    supported yet, one still has to recompile the program to get
    it to work.

    What I had in mind is a config file, so one only has to
    compile the program once and to edit the config file after that.

    The best solution would be probably to try to detect the
    board at
    runtime and if it doesn't work, switch back to a config file
    (which would be placed in /etc ?).

     
  • Sergei Haller

    Sergei Haller - 2002-12-17

    Logged In: YES
    user_id=76018

    ... Didn't realise that I wasn't logged in ...

     
  • Sergei Haller

    Sergei Haller - 2003-03-04
    • status: open --> closed
     
  • Sergei Haller

    Sergei Haller - 2003-03-04

    Logged In: YES
    user_id=76018

    this patch is gone into 0.90 release

     

Log in to post a comment.

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:





No, thanks