grub-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: grub-efi x86_64 on macbook air?


From: step21
Subject: Re: grub-efi x86_64 on macbook air?
Date: Wed, 28 Jan 2009 18:30:24 +0100

On Wed, Jan 28, 2009 at 5:56 PM, Javier Martín <address@hidden> wrote:
> El mié, 28-01-2009 a las 17:42 +0100, step21 escribió:
>> Hey, while investigating possibilities on how it might be possible to
>> install/run something "wubi" like on macs (without having windows
>> installed) I ran into a couple of issues. I post here mainly in hope
>> that one of the folks that actually wrote the code for this or at
>> least have experience with it read this, cause most ppl on irc (while
>> being very helpful) seem to know more about the pure x86/bios/linux
>> side of things.
>>
>> Besides some (minor?) issues like that the command line
>> freezes/becomes unresponsive after a (relatively short) amount of time
>> and the fact that booting back to OS X (which everyone says should
>> work 100%) does not work at all (it complains that it can't find the
>> specified os x *efi, but it is there, and search correctly finds the
>> right drive) my biggest problem right now is the following:
>>
>> I got some kerne/initrd combo to boot (2.6.26 from debian lenny
>> netinstall iirc) but it gets to the language selection screen, first
>> thing you notice, the keyboard is dead. I supplied "init=/bin/sh" to
>> be able to read at least the last of the error messages. As it turns
>> out to me it seems that seemingly everything pci-related is
>> unavailable. Error messages are as follows:
>>
>> [time after boot]PCI No IRQ known for interrup pin C of device
>> 0000:00:1a.7. Please Try using pci=biosirq
>> [time after boot]ehci_hcd 0000:00:1a.7 Found HC with no IRQ. Check
>> BIOS/PCI 0000:00:1a.7 setup!
>> [time after boot]ehci_hcd 0000:00:1a.7 fail, -19
>> (of course time/device ids and pins (A-D) changed, but otherwise they
>> were the same I think)
>> I tried supplying pci=biosirq although I didn't think it would work,
>> and it didn't, or some solution for a similiar issue from the debian
>> wiki for the macbook (using a usb keyboard) which didn't work either.
>> The usb keyboard works for refit and grub, but not once booted into
>> linux. On some other wiki page it was suggested to supply "noapic
>> acpi=force" and maybe "irqpoll" which I tried in various combinations
>> without a change.
>>
>> The machine I use for testing this is a first generation macbook air.
>> Now while there are numerous reports about people running linux on it
>> (pretty well actually) by providing it with a legacy bios environment,
>> I didn't find a report about one running grub-efi on it successfully
>> to boot linux. The most relevant posts my searches come up with are my
>> own posts on the ubunutu forums.
>>
>> So, I'm not sure who is to blame here, or if it's maybe just my own
>> fault, but I decided that it would be nice to get the
>> opinion/solutions? of some ppl who might be more intimately familiar
>> with the workings of the mac boot firmware etc. and maybe know if this
>> actually has been tested and should work, or not.
>>
>> Thanks for an help in advance.
>
> Are you sure that Debian lenny supports booting from EFI? I'm quite
> literally talking out of my ass here, so I don't have a clue, but it
> seems that the kernel you're trying to boot is trying to use BIOS
> functionality. GRUB does _not_ include BIOS emulation, so that will
> certainly not work. In order for a Linux kernel to boot from EFI, you
> have to enable the switch in the pre-build kernel configuration.
>
> From my past knowledge (on my brother's Macbook), at least Ubuntu/x86
> kernels _do_ have the EFI switch enabled, so you might give it a try and
> check if that's what's going wrong.
>
> On the OS X and keyboard issues, I'm afraid I cannot help you. Perhaps
> our elders would be wiser?
>
> PS: you could try, however, to enable debug output when compiling your
> GRUB so that you could see if there was anything wrong when loading the
> file, like a hypothetical hfs.mod error (again, talking out of my
> non-talking orifices)
>
> _______________________________________________
> Grub-devel mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/grub-devel
>
>

Well, I didn't check specifically, but afaik the kernel has it
enabled. I tried with an ubuntu kernel before which didn't even get
that far. (just loaded kernel/initrd on commandline, but after "boot"
nothing happened)
Yes, I know grub does not supply bios emulation, but maybe the kernel
just fails to recongize that it's supposed to use efi?
On compiling kernel and grub2: I couldn't get compiling grub2 to work
on os x, even with a seperate gnu gcc, then it failed on linking and
it seems there is no gnu version of ld available for os x. (people say
it just doesn't run) so I kinda gave up. Also, which applies also to
kernel compiling, currently this is the only machine that totally
belongs to myself. I might drob wubi or something on some windows box
nobody really uses, and hope ppl don't hate me too much for that, then
I could compile grub2/my own kernel where I know I have everything
enabled. Anything specific I should be aware of if I compile grub2
with debug support? Someone on irc mentioned invoking grub_set_env()
in grub_main() or something, is that right? I know that I can set
debug=all in grub.cfg, but most of the time the output is so mch it
just floods my whole screen ...




reply via email to

[Prev in Thread] Current Thread [Next in Thread]