Recently, Linus and all the other kernel maintainers decided to pull i386 and x86_64 into one architecture, x86.  To do this they added suffixes to each file, "_32" or "_64", depending on what architecture it was written to run on.  I can only assume that by doing this they somehow broke the boot process of Cobalts, because the kernel we build isn't getting its command line after 2.6.24.

I don't have any Cobalts here at home, so tonight I'm going to see what I can do about fixing this issue.

David Monro seems to have tracked it down...

Code:

Ugh. It looks like the way the 16bit code in the stock kernel passes the command line to the 32bit code has changed, at least for the case of the 'old-style' command line boot protocol (which I assume is what the cobalt bootloader kernel uses). It looks like the 16bit code in arch/x86/boot/main.c takes the 'old-style' command line and copies it to where the 'new-style' command line would be, and now the 32bit code no longer handles the 'old-style' case (compare arch/i386/kernel/head.S from about line 86 with arch/x86/kernel/head_32.S from about line 113).

Moving the bootloader code from the original 2.4.19C13 kernel to a new 2.6.20.1 kernel is taking longer than I would have ever expected.  The biggest issue in front of me is the fact the code was never designed to compile under GCC 4.  Add on to that macros that have been removed from the i386 architecture, and your creating a recipe for a long job.

As it is now, I don't want to make an estimate on time to completion.  Though as I progress I'll be posting updates.  Thanks for the support :-)

It's here, powermode support for GenV RaQs.  Thanks to Herps@RaQTweak.com for suggesting it, and especially thanks to Brian99 who sacrificed one of his XTRs to test it for me.

Two patch hunk errors prompted the release of linux-cobalt-2.6.20-2007021901.illogical.patch.  Also in the news, it has been requested that the functionality of /proc/cobalt/powermode be re-added.  I am going to look in to this now that I have an older kernel source which features the code.  I will not make promises about this code, because from the looks of it, it could be a bit of testing... If it even plays nice at all.

It took a little longer than I had thought, but doing this work at work means I regularly have to put the patch on hold to take care of a customer.  But I finished it, so I'm happy, as I'm sure most of you are.  I think I've got the reboot and shutdown "bug" fixed.  Also, I believe I've tracked down the source of the ACPI woes people were experiencing.  It may take a while to figure out the exact problem and fix it, but I'll try to do it quickly.

Linux Cobalt Kernel 2.6.19.2 Patch

...since I've updated this site.  Later on today I'll be posting a Cobalt patch for the current Linux kernel (2.6.19.2) and doing some clean up.

    With the release of gentoo-sources-2.6.14-r5 I moved the kernel patch over.  As there was a bit of restructuring in certain areas of the kernel my patch affects, it took a little time to get everything smoothed over.  The new patch is available via its local Project page or via FTP at this server.

    I also completed a basic install guide yesterday at the nudging of RabidDog.  The guide is meant for those familiar with Gentoo installations, and GNU/Linux in general.  Comments and criticisms are definitely welcome, provided they are constructive.

Enjoy!

    Star topologies look pretty, the tend to work... Right up until you lose a hub.  Enter the mesh, OpenMesh IRC to be exact.  While still in planning, this could be the most interesting IRC related project, dare I say it, ever.

    After hearing that WebLED wasn't functioning right on a RaQ 2, fuxxy helped me track down the proper values to shoot to the LED driver as the values are different on MIPS than they are on x86.  So after a little Makefile hackery and some defines, it should be working.  If this version works for both I'll be tossing it in to Portage.

    I finally added a link, to my server at home running ViewCVS.  You can find the link in the links section, naturally.  And hold the criticisms, I prefer to receive constructive comments.