The newest ACPI kernel patch from Intel fixes a MAJOR problem with ACPI on many "legacy-free" machines that don't have a real BIOS, like the Toshiba Satellite 5005-S507. By default all IRQs for PNP devices are 0 and this causes things not only to fail but in my case it caused the PCMCIA drivers to completely lock up my machine. This is fixed by: acpica-linux-20020404.tar.gz Avaliable at: http://developer.intel.com/technology/iapc/acpi/downloads.htm Not these drivers are not currently compatible with ACPID, which is designed for the older driver, I believe. Sean
I've informed Alan Cox of this problem. We'll see what he says. I'd rather have Alan look into incorporating the right fix (if he hasn't already) into his kernels than using these ACPI patches since they can be problematic to get working correctly, even after you fix all the rejects.
I've contacted Alan Cox about this problem, and he is not aware of it and interested in getting it fixed. Where did you get this information from (regarding the solution)? Can you verify that it works? If so, I'd like to put you in contact with Alan so that he can get this ACPI issue fixed in his kernel.
Well, I've been using Linux before the 1.0 release days, but am not really a developer. I'm more on the System Administration side. I have a Toshiba Satellite 5005-S507 laptop and have been traversing some of the linux mailing lists for the 5005 series laptops because of the "legacy free" nature of these machines and the difficulty people have had setting them up under Linux. There are a few general resources out there including: http://mobilix.org/toshiba_s5005_s504.html http://rooster.stanford.edu/~ben/toshiba/ The mailing list is currently in the process of moving, but I could pass that information along as well. What seems to happen is that even with the current ACPI support compiled into the kernel all device IRQs are left as 0 and the kernel suggests via warning messages that I use "pci=biosirq" (I believe..this is from memory....I am reinstalling my system as I write this)....Using this in lilo casues a kernel panic on boot. Not using this means none of my devices have an IRQ and things fail to work. I did some investigation and found out about Intel's work with ACPI for the x86 platform and decided to give their ACPI patch for Linux a try. I applied the patch (had to hand merge three or four .rej patch files), recompiled the kernel and voila! it worked like a charm. I'm not sure that all the ACPI is working as even the Intel work is still "in progress", but their most recent patch seems to handle assigning the IRQs great amoung other basic things. ACPID however doesn't seem to work against their patch although they do point to it from their web page. The PMTOOLS package does seem to work fine however. On the mailing list at least one other person has mentioned success with this method, so I'm pretty confidant that something has been updated to work correctly for these laptops (and likely others) that was missing before. I would be happy to talk to Alan directly. He can contact me via email the quickest at spkane@genomatica.com.
Note: I have tried the newest gentoo-sources kernel and the ACPI in there still fails to work for me. As a matter of fact it casues kernel panics, which did not happen in the past. I used the Intel ACPI patches again and my system became very usable (although it took a bit of work to merge in the patches with the already heavily patched gentoo-sources). Sean
Hi Sean, I have a 6Mb patch (gentoo-sources with the new ACPI code added) for you to test. I have only tested compilation, which is working. What's the best way to get this patch to you? Will email work or shall I put it up on a Web site somewhere?
You can email it to me and I'll give it a run. Sean
You can get a test patch to fix this from http://www.gentoo.org/~drobbins/ ... apply it to stock 2.4.18 sources and let me know how/if ACPI works.
There are two patched in there both for 2.4.19. Which one is the one you want me to try, now that I have the machine in a testable state. I am currently running with a hand-patched version of gentoo-sources-2.4.19-r1. Thanks, Sean
This fix is now in our -r4 kernel. You should try that one.
I'll go ahead and mark this as fixed. The new kernels do boot on my laptop and the machine is useable. I think there are still some issues with IRQs not being setup properly on all devices although I am not sure about this, and it is likely just an issue with the current development state of the ACPI work. Thanks for the help. I'll continue to test some of the newer ACPI patches as they are released and see how they improve. Sean
