2011/2/6 Vladimir 'φ-coder/phcoder' Serbinenko <phco...@gmail.com>

> On 02/06/2011 05:57 AM, Nate Weibley wrote:
> > On Fri, Feb 4, 2011 at 12:01 PM, <grub-devel-requ...@gnu.org
> > <mailto:grub-devel-requ...@gnu.org>> wrote:
> >
> >     Date: Fri, 04 Feb 2011 19:19:34 +0300
> >     From: Vladimir '?-coder/phcoder' Serbinenko     <phco...@gmail.com
> >     <mailto:phco...@gmail.com>>
> >     Subject: Re: AMI Aptio EFI booting problems on ASUS G73SW-A1
> >     To: grub-devel@gnu.org <mailto:grub-devel@gnu.org>
> >     Message-ID: <4d4c2716.7080...@gmail.com
> >     <mailto:4d4c2716.7080...@gmail.com>>
> >     Content-Type: text/plain; charset="utf-8"
> >
> >     On 02/04/2011 04:54 PM, Nate Weibley wrote:
> >     > Hey GRUB devs,
> >     >
> >     > I've been having some trouble getting GRUB2 to load Arch x86_64
> >     on my
> >     > ASUS G73SW-A1. Specifically, if I do not specify noefi on the
> kernel
> >     > command line in my grub.cfg everything stalls at initrd.
> >     I don't think that it gets so far as initrd and actually panics
> early.
> >     > I'm building GRUB2 from the bzr repo, so to my knowledge it
> >     should be
> >     > up-to-date.
> >     >
> >     "noefi" isn't handled at all by GRUB itself. It's just a parameter
> >     that
> >     it transfers to linux. So if it makes a difference. It's either:
> >     a) (most likely) Linux doesn't handle your EFI either because of
> >     EFI or
> >     Linux bugs
> >     b) GRUB passes incorrect EFI pointers. Considering that this
> operation
> >     is trivial, I think this is unlikely.
> >     GRUB can't do anything for (a). Well it could but not anything in
> >     a sane
> >     way. I suggest trying different Linuxes (kernels) E.g. Ubuntu,
> Debian,
> >     Red Hat, git.
> >
> >     > I substituted noefi for add_efi_memmap and added set debug=all
> >     before
> >     > initrd is used.
> >     >
> >     add_efi_memmap should do any difference at all. It just makes
> >     Linux redo
> >     some of GRUB's job.
> >     > The system stalled again, but I did get /some /debugging output. A
> >     > link to the output is below; apologies for having to take a picture
> >     > with my cell phone as the laptop does not have a usable serial port
> >     > for capture.
> >     > http://imgur.com/40lLO
> >     > I should point out that adding set debug=all appears to stall the
> >     > system at the same point even if noefi is still set for the
> >     kernel...
> >     > so perhaps this stall is not indicative of the initrd failure. I
> >     read
> >     > through the thread suggesting that there was a bug present with
> >     system
> >     > over 8GB, but it was my impression that the bug was patched.
> >     Are you sure? It may be indicative of a failure when printing debug
> >     output. Previously there was such bug when GRUB tried to use already
> >     finalised EFI services to print debug output but AFAICT it's
> >     already fixed.
> >     > If I can provide any additional info or take any additional
> >     debugging
> >     > steps, please let me know.
> >     >
> >     > --
> >     > --Nate Weibley
> >     >
> >     >
> >     > _______________________________________________
> >     > Grub-devel mailing list
> >     > Grub-devel@gnu.org <mailto:Grub-devel@gnu.org>
> >     > http://lists.gnu.org/mailman/listinfo/grub-devel
> >     >
> >
> >
> >     --
> >     Regards
> >     Vladimir 'φ-coder/phcoder' Serbinenko
> >
> > Vladimir,
> > Per your suggestion I tried other linux distros with various kernels.
> > So far none of the EFI enabled distros are working. They do work if
> > booted via BIOS though, of course. Windows 7 64bit /does / boot
> > appropriately via EFI, so it's hard to say where the fault is. If
> > Windows is booting though, it seems more likely something is going
> > wrong in the Linux kernel EFI handling, or perhaps as you say GRUB is
> > passing incorrect pointers. Either way, they all exhibit the exact
> > same behavior... the kernel is loaded, and at the point init should be
> > called, the system stalls with no debug or error message.
> On EFI Linux has no early printk other than to a serial. So you receive
> no messages unless you're connected to serial. It makes diagnosis much
> more difficult.
> --
> Regards
> Vladimir 'φ-coder/phcoder' Serbinenko
>
>
> I'll say! I was hoping there would be a way to pass messages back through
grub like some sort of virtual serial port or to virtualize this but allow
interaction with my actual system EFI. Barring that, I can't even fathom how
I'd approach the next step of debugging. AFAIK this chipset doesn't
even *support
*legacy serial, so I couldn't even open my laptop and tap into some
traces/pads to get RS-232 out. I guess the best I can do is wait and see if
future kernels adhere more closely to the pure EFI spec and hope that's what
is allowing windows to boot.




-- 
--Nate Weibley
_______________________________________________
Grub-devel mailing list
Grub-devel@gnu.org
http://lists.gnu.org/mailman/listinfo/grub-devel

Reply via email to