All of lore.kernel.org
 help / color / mirror / Atom feed
From: Stephen Kou <stephen@hyarros.com>
To: The development of GNU GRUB <grub-devel@gnu.org>
Subject: Re: UEFI Boot with Grub-Experimental
Date: Fri, 02 Jul 2010 15:30:30 -0700	[thread overview]
Message-ID: <4C2E6886.2040108@hyarros.com> (raw)
In-Reply-To: <alpine.DEB.2.00.1007022033140.3494@birdy>

  Most likely the disk enumeration is different than what you have in 
the grub list.

boot into a working system, plug your disk with the linux in, and then 
do sudo blkid

get the UUID of your root filesystem, and then put it as 
root=UUID=000000-0000-0000-0000-000000000 instead of /dev/sda4.

Hopefully that will work for you.

--Stephen

On 7/2/2010 11:46 AM, Reynald Lercier wrote:
> Hi Stephen,
>
> The situation is a little bit better in my side too...
>
> I recompile the 2.6.34 kernel following the instructions in 
> Documentation/x86/x86_64/uefi.txt, from the kernel linux package.
> Especially, I defined "CONFIG_FRAMEBUFFER_CONSOLE=y" int the .config 
> file instead of "CONFIG_FRAMEBUFFER_CONSOLE=m".
>
> (Do you have a difference in your side between 2.6.32 and 2.6.35 RC in 
> view of this variable ?)
>
> Kernel now boots, with the usual logs (YES !), with the noefi option,
> but I get a crash when the kernel try to access to the root device...
>
> Maybe,
> http://ubuntuforums.org/showthread.php?t=995704&highlight=framebuffer&page=112 
>
> can help.
>
> -reynald
>
> On Fri, 2 Jul 2010, stephen@hyarros.com wrote:
>
>>
>> Hi Reynold,
>>
>> I've made some progress past you recently!  I think there are bugs in
>> the Linux Kernel now that must be overcome by the folks over at kernel
>> development.  Did you try the 'noefi' kernl boot flag?  It allowed me to
>> get past the hang and my system would actually boot.
>>
>> Also, I'm using the new 2.6.35 RC kernel (the ubuntu-flavoured one,
>> acutally) -- i get a weird framebuffer hang (my screen turns orange) if
>> I use the kernel included in lucid (2.6.32).  I've yet to try 2.6.34;
>> let us know how it goes for you.
>>
>> -stephen
>>
>> On Fri, 2 Jul 2010 16:14:53 +0200 (CEST), Reynald Lercier
>> <reynald.lercier@m4x.org> wrote:
>>> I observed a similar beahaviour as stephen in my side.
>>>
>>> Especially I had no dificulty at all to switch to a 1680x1050x32
>>> graphic mode (following Piscium's advices) in the grub2 interface with
>>> what follows in grub.cfg
>>>
>>>     insmod efi_gop
>>>     insmod font
>>>     loadfont (hd1,gpt4)/usr/share/grub/unicode.pf2
>>>     insmod gfxterm
>>>     set gfxmode=1680x1050x32
>>>     set gfxpayload=1680x1050x32
>>>     terminal_output gfxterm
>>>
>>>     insmod part_gpt
>>>     insmod ext2
>>>
>>>     set timeout=10
>>>
>>>     menuentry "Linux (with fakebios + mbp62hd)" {
>>>           search  -s -f  /boot/vmlinuz-2.6.34-custom
>>>           fakebios
>>>           linux /boot/vmlinuz-2.6.34-custom root=/dev/sda4 video=efifb
>>>           initrd /boot/initrd.img-2.6.34-custom
>>>     }
>>>
>>> When the power is on, rEFIt first runs, which then, when I select the
>>> corresponding icon, runs EFI grub2 (that I put in the EFI fat first
>>> partition), which in turn should boot a linux kernel. But, the laptop
>>> always hangs here, in the boot comand (black screen, no message).
>>>
>>> I tried several linux kernel, the lucid ubuntu's one
>>> (2.6.32-22-generic) and a recompiled 2.6.34-custom too. But, the
>>> laptop still always hangs at the boot up (black screen, no message).
>>> Unfortunataly, absolutely no output in any file of the /var/log
>>> directory too.
>>>
>>> Just to test further, I tried to patch the
>>> "linux-2.6.34/drivers/video/efifb.c" linux kernel file too, by adding
>>> information about my hardware (framebuffer address, stride, etc... of
>>> the GT330M nvidia crad, all obtained from rEFIt commands: pci,
>>> devices, dh, etc.) in the dmi_list[] variable as follows
>>>
>>>     static struct efifb_dmi_info {
>>>         char *optname;
>>>         unsigned long base;
>>>         int stride;
>>>         int width;
>>>         int height;
>>>     } dmi_list[] = {
>>>             ...
>>>     [M_MBP_6_2_HD] = { "mbp62hd", 0x90030000, 2048 * 4, 1680, 1050 },
>>>     [M_UNKNOWN] = { NULL, 0, 0, 0, 0 }
>>>     };
>>>
>>> I modified acordingly enum definitionw (for M_MBP_6_2_HD) and
>>> __initdata dmi_system_table.
>>>
>>> But,
>>>     linux /boot/vmlinuz-2.6.34-custom root=/dev/sda4 
>>> video=efifb:mbp62hd
>>> still behaves badly (black screen).
>>>
>>> Any idea about how I could understand further what's going on ?
>>>
>>>
>>> On Thu, 1 Jul 2010, stephen@hyarros.com wrote:
>>>
>>>>
>>>> Ah, yes -- the gfxpayload variable works.  But now after it
>>>> successfully initializes the video mode, it hangs at the boot with 
>>>> the _
>>>> (no output yet at all from kernel).
>>>>
>>>> I'm trying to boot ubuntu lucid (2.6.32-22-generic).  The EFI boot
>>>> works on some machines, one others it doesn't.
>>>>
>>>> Thanks
>>>>
>>>> On Thu, 1 Jul 2010 11:33:17 -0700, Seth Goldberg
>>>> <seth.goldberg@oracle.com> wrote:
>>>>> Which kernel or os are you loading?  Did you try setting the
>>>>> gfxpayload env var to 1024x768?
>>>>>
>>>>>   --S
>>>>>
>>>>> On Jul 1, 2010, at 10:54 AM, <stephen@hyarros.com> wrote:
>>>>>
>>>>>> Something I forgot to mention that's important -- (sorry for the 
>>>>>> spam)
>>>>>> -- GRUB tries to initalize with 800x600 regardless of what 
>>>>>> $gfxmode is
>>>>>> set to.
>>>>>>
>>>>>> set gfxmode=1024x768
>>>>>>
>>>>>> will still result in GRUB trying to initalize the video as 800x600
>>>>>> after the 'boot' command is issued.
>>>>>>
>>>>>> -stephen
>>>>>>
>>>>>> On Thu, 01 Jul 2010 10:49:59 -0700, <stephen@hyarros.com> wrote:
>>>>>>> Hi everyone,
>>>>>>>
>>>>>>> I've had some interesting discoveries / success with this 
>>>>>>> problem in
>>>>>>> the past couple of days.  Where I am I have several machines to 
>>>>>>> try out.
>>>>>>> On some of the machines, it works; while on others, it doesn't.  
>>>>>>> I'm
>>>>>>> pretty sure this all has to do with the video modes now.
>>>>>>>
>>>>>>> On my laptop (which also supports UEFI), there is only one video 
>>>>>>> mode
>>>>>>> supported as reported by efi_video_modes: 1024x768.  However, 
>>>>>>> when GRUB
>>>>>>> is booting, it calls grub_video_set_mode with the string 
>>>>>>> "800x600".  It
>>>>>>> then fails to initialize the GOP adapter (which reports it only 
>>>>>>> supports
>>>>>>> 1024x768).  Then it complains that no suitable mode is found, 
>>>>>>> and tries
>>>>>>> to boot nayways without a video mode set.
>>>>>>>
>>>>>>> Does anyone know why it would be trying to boot as 800x600 only 
>>>>>>> and not
>>>>>>> the 1024?
>>>>>>>
>>>>>>> I'll be looking into the code more, but thought I'd let those 
>>>>>>> who are
>>>>>>> interested know.
>>>>>>>
>>>>>>> -stephen
>>>>>>>
>>>>>>> On Thu, 1 Jul 2010 08:16:34 +0200 (CEST), Reynald Lercier
>>>>>>> <reynald.lercier@m4x.org> wrote:
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> I encounter very similar problemes on a my macbook pro 15', a 
>>>>>>>> MBP 6,2.
>>>>>>>>
>>>>>>>> (I need full EFI booting on this machine in order to use under 
>>>>>>>> linux
>>>>>>>> the INTEL graphic card, instead of the NVIDIA GT330M one, and 
>>>>>>>> finally
>>>>>>>> increase a lot the battery run time)
>>>>>>>>
>>>>>>>>
>>>>>>>> In my case efi_video_info returns
>>>>>>>>
>>>>>>>> GOP info:
>>>>>>>> List of video modes:
>>>>>>>> 0: 1680 x 1050, BGRA8, scan line 1680
>>>>>>>> Current mode: 0
>>>>>>>>
>>>>>>>> Same question, what to do now with this ?
>>>>>>>>
>>>>>>>> Thanks.
>>>>>>>>
>>>>>>>> On Wed, 30 Jun 2010, stephen@hyarros.com wrote:
>>>>>>>>
>>>>>>>>> Hi,
>>>>>>>>>
>>>>>>>>> Thanks for the response.
>>>>>>>>>
>>>>>>>>> After trying terminal_output, the computer screen would simply 
>>>>>>>>> go black
>>>>>>>>> and the machine would hang (the numlock key would not respond) 
>>>>>>>>> after the
>>>>>>>>> terminal_output gfx command was executed; this would happen 
>>>>>>>>> regardless
>>>>>>>>> of whether or not set gfxmode was called before.
>>>>>>>>>
>>>>>>>>> I also have just tried the efi_video_info patch; the system 
>>>>>>>>> reports:
>>>>>>>>>
>>>>>>>>> GOP info:
>>>>>>>>> List of video modes:
>>>>>>>>> 0: 1024 x 768, bitonly, scan line 1024
>>>>>>>>> Current mode: 0
>>>>>>>>>
>>>>>>>>> Do i need to pass this information on to the kernel somehow?
>>>>>>>>>
>>>>>>>>> Thanks.
>>>>>>>>>
>>>>>>>>> On Wed, 30 Jun 2010 10:40:31 +0100, Colin Watson 
>>>>>>>>> <cjwatson@ubuntu.com>
>>>>>>>>> wrote:
>>>>>>>>>> On Wed, Jun 30, 2010 at 01:54:36AM -0700, stephen@hyarros.com 
>>>>>>>>>> wrote:
>>>>>>>>>>> After having no luck using the grub-efi-amd64 package in 
>>>>>>>>>>> ubuntu, or the
>>>>>>>>>>> grub trunk, I've started trying to compile my own grub and 
>>>>>>>>>>> getting it to
>>>>>>>>>>> boot on a new Intel motherboard which supports EFI.  I've 
>>>>>>>>>>> not been able
>>>>>>>>>>> to get any output yet from the acutal linux kernel; usually 
>>>>>>>>>>> the system
>>>>>>>>>>> will simply hang after the boot menu option is selected, or 
>>>>>>>>>>> the 'boot'
>>>>>>>>>>> command is issued from the grub command line.
>>>>>>>>>>>
>>>>>>>>>>> Currently the farthest I've gotten is using the grub command 
>>>>>>>>>>> line and
>>>>>>>>>>> typing in the following commands:
>>>>>>>>>>>
>>>>>>>>>>> insmod efi_gop # no impact on result
>>>>>>>>>>> insmod ext2
>>>>>>>>>>> insmod part_gpt
>>>>>>>>>>>
>>>>>>>>>>> set root=(hd0,gpt3)
>>>>>>>>>>> fakeroot # optional, no impact on result
>>>>>>>>>>
>>>>>>>>>> I guess that should be 'fakebios'.
>>>>>>>>>>
>>>>>>>>>>> error: no suitable mode found
>>>>>>>>>>
>>>>>>>>>> After 'insmod efi_gop', could you try 'insmod gfxterm' and then
>>>>>>>>>> 'terminal_output gfxterm', and see what happens?  Before the
>>>>>>>>>> terminal_output command, you can also use 'set gfxmode=MODE' 
>>>>>>>>>> (e.g. 'set
>>>>>>>>>> gfxmode=1024x768') to change its mode selection.  gfxterm can 
>>>>>>>>>> help
>>>>>>>>>> matters here, as that way you have a working video mode that 
>>>>>>>>>> the kernel
>>>>>>>>>> can be told to inherit, rather than having to probe its own.
>>>>>>>>>>
>>>>>>>>>> Unfortunately right now it's hard to get debugging 
>>>>>>>>>> information on EFI
>>>>>>>>>> video modes.  Since you're building your own GRUB anyway, 
>>>>>>>>>> though, you
>>>>>>>>>> could try this patch against trunk:
>>>>>>>>>>
>>>>>>>>>>  http://people.canonical.com/~cjwatson/tmp/grub-efivideoinfo.patch 
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> That will give you an 'efi_video_info' command, which should 
>>>>>>>>>> dump out
>>>>>>>>>> the available GOP modes, and might be useful to get a 
>>>>>>>>>> slightly better
>>>>>>>>>> idea of what's going on.
>>>>>>>>>>
>>>>>>>>>>> booting however
>>>>>>>>>>> _
>>>>>>>>>>>
>>>>>>>>>>> And then nothing else happens.
>>>>>>>>>>
>>>>>>>>>> It's possible that the kernel may have booted successfully, 
>>>>>>>>>> but that you
>>>>>>>>>> simply don't have a working console.  It would be useful to 
>>>>>>>>>> try pinging
>>>>>>>>>> the machine to test that.
>>>>>>>>>>
>>>>>>>>>>> I've also tried newreloc, but I don't think this has 
>>>>>>>>>>> anything to do with
>>>>>>>>>>> relocations.
>>>>>>>>>>
>>>>>>>>>> Agreed.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> Grub-devel mailing list
>>>>>>>> Grub-devel@gnu.org
>>>>>>>> http://lists.gnu.org/mailman/listinfo/grub-devel
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Grub-devel mailing list
>>>>>>> Grub-devel@gnu.org
>>>>>>> http://lists.gnu.org/mailman/listinfo/grub-devel
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> Grub-devel mailing list
>>>>>> Grub-devel@gnu.org
>>>>>> http://lists.gnu.org/mailman/listinfo/grub-devel
>>>>>
>>>>> _______________________________________________
>>>>> Grub-devel mailing list
>>>>> Grub-devel@gnu.org
>>>>> http://lists.gnu.org/mailman/listinfo/grub-devel
>>>>
>>>>
>>>> _______________________________________________
>>>> Grub-devel mailing list
>>>> Grub-devel@gnu.org
>>>> http://lists.gnu.org/mailman/listinfo/grub-devel
>>>>
>>>
>>>
>>> _______________________________________________
>>> Grub-devel mailing list
>>> Grub-devel@gnu.org
>>> http://lists.gnu.org/mailman/listinfo/grub-devel
>>
>>
>> _______________________________________________
>> Grub-devel mailing list
>> Grub-devel@gnu.org
>> http://lists.gnu.org/mailman/listinfo/grub-devel
>>
>
>
> _______________________________________________
> Grub-devel mailing list
> Grub-devel@gnu.org
> http://lists.gnu.org/mailman/listinfo/grub-devel



  reply	other threads:[~2010-07-02 22:31 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-06-30  8:54 UEFI Boot with Grub-Experimental stephen
2010-06-30  9:40 ` Colin Watson
2010-06-30 17:10   ` stephen
2010-07-01  6:16     ` Reynald Lercier
2010-07-01 17:49       ` stephen
2010-07-01 17:54         ` stephen
2010-07-01 18:33           ` Seth Goldberg
2010-07-01 18:43             ` stephen
2010-07-02 14:14               ` Reynald Lercier
2010-07-02 17:16                 ` stephen
2010-07-02 18:46                   ` Reynald Lercier
2010-07-02 22:30                     ` Stephen Kou [this message]
2010-07-07 23:49                   ` stephen
2010-07-08  0:11                     ` Vladimir 'φ-coder/phcoder' Serbinenko
2010-07-08  1:39                       ` Stephen Kou
2010-07-01 19:09           ` Piscium
2010-09-05 20:14     ` Vladimir 'φ-coder/phcoder' Serbinenko
2012-03-04  1:06     ` Vladimir 'φ-coder/phcoder' Serbinenko

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4C2E6886.2040108@hyarros.com \
    --to=stephen@hyarros.com \
    --cc=grub-devel@gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.