Search:

Here is the experience of Jens Axboe with rMBP :
link

Hi,

I had the same problem on a MacBook Pro with 12.04 64-bit.
I needed to select "Try before install" instead of "Install", so that I could open a Terminal and type "ps -efH", and I saw that...

"hfs-bless" does exist on Linux in the "mactel-boot" package :
https://bugzilla.redhat.com/show_bug.cgi?id=755093

metatech

grub.efi needs to be on a HFS+ partition, but the real Linux partition can be on a ext4 partition.

You can "bless" the "grub.efi" to avoid using rEFIt.
Type the following command in a MacOS terminal :


sudo bless --folder=/Volumes/something --file=/Volumes/something/efi/grub/grub.efi --setBoot

Very interesting !
Could this trick also be used to re-enable the Intel card in BIOS mode ?

metatech

fooblablah,

What happens if you comment out "fix_video" and use "lvds_channels=2" (all other parameters being equal) ? Does that work without freezing ?
If yes, what happens if you uncomment...

Hi fooblablabla,
Thanks for your test !
Something is not clear from your description : does the picture on the screen look distorted or is it all black ?
Thanks,
metatech

wilfriedd,

Thanks a lot for your test !
Apparently, the PCI ID of Intel CPU have several variants for the same generation.
Please replace 0x01168086 by 0x01268086 in the patch and that should...

Have you tried the grub2 patch to avoid the need for the LVDS kernel patch ?
http://ubuntuforums.org/showpost.php?p=11198339&postcount=605

Some people still had the problem with Ubuntu 11.10 :
https://bugs.launchpad.net/ubuntu-release-notes/+bug/774089/comments/133

Let's hope it's fixed for next year's release...

Hello,

I think I found an alternative way to setup the "LVDS channels" in grub instead of using the "Implement manual override of LVDS single/dual channel mode" kernel patch, to avoid the need to...

You are most likely hitting this well-known bug :
https://bugs.launchpad.net/efibootmgr/+bug/774089
You can try to reinstall your firmware (see instructions in thread).

In BIOS mode, the interface is by default in legacy IDE mode.
In EFI mode, the interface is by default in SATA AHCI mode.
But the following is a workaround : ...

The general rule is that the 9600M only can be used in BIOS mode and the 9400M and the 9600M can both be used in EFI mode.

This rule is independent of the Ubuntu version.

It surely is.

Can you add "echo" between all lines to see which is the last line displaying an output ?
Can you post the resulting grub.cfg, so that it's easy to see which sequence is executed ?...

I am running out of ideas…

Which is the last "inb" command whose output is displayed ?

Have you tried without all the "gfxterm" stuff ?

Thanks.

I meant something like this :
outb 0x728 1
inb 0x728
outb 0x710 2
inb 0x710
outb 0x740 2
inb 0x740
outb 0x750 0
inb 0x750

Yes the logs might some clues, especially the line looking like the following :


efifb: dmi detected MacBookPro5,3 - framebuffer at d0010000 (1440x900, stride 8192)


Can you add the following...

Hello,

The magic numbers of the gmux looks the same (mainly 0x750). Anyway, they have been the same since the first dual-graphics laptops in 2009 (except brightness), so this is most probably not...

Hello,

Xserve machines do not have a BIOS compatibility layer.
Your machine must be be booted in EFI mode, hence the error.
You could try "fakebios" in grub.cfg if you are lucky.

The "magic" gmux hex values (eg 0x750) can be found in the log under MacOS X by enabling trace on the gmux driver :
Add "agclog=10000 agcdebug=4294967295" to...

Hi,

Recent Mac models (from 2010 and higher) are reported not to work "out of the box" with the Linux kernel in EFI mode.
3 possibilities to make it work :
- Use the kernel v3.0rc7 (or higher),...

Grub2 v1.99 is absolutely required for UEFI booting (with the "newreloc" feature), or you risk memory corruptions because of memory overwriting.
Please see...

Hi Sloth,

Thanks for the great news...
Does it mean the "LVDS dual channel mode" is not needed either to use the Intel graphic card ?

metatech