Installing CentOS on a MacBook Pro

[Howard S Shubs]

On a mid-2009 MacBook Pro (MBP) with MacOS X 10.8 installed on the
internal drive, I tried to install CentOS on an external drive. The
first time I did it, I had the external drive configured with a GUID
partition map. While the CentOS dvd booted and installed the OS, the
MBP turned off during the "BONG" while attempting to boot.

Today, I initialized the external drive with a MBR partition map.
Again, everything looked good until I tried to boot afterwards. It got
VERY weird there, with a _ prompt in the upper left corner of the
monitor, but got no further.

Is there a way to do this? If so, what am I missing? The external
drive is connected directly via USB, and is a 300GB device.

[Ross Maloney]

On 04/27/2013 06:13 AM, Michael Vilain wrote:
> In article<howard-275B0C....@news.individual.net>,

> The USB part is the weird oddity. Used to be that PPC Macs (and later
> Intel) could boot Firewire drives. Then Intel Macs came out with GUID
> and the ability to boot USB disks if they had a proper boot block (e.g.
> Windows). The firmware to boot USB can in Intel Macs.
>
> I haven't tried it, but change that external drive to Firewire and see
> if that changes anything. Post your progress.
>

It might be worth having a look at the article maclinux.pdf on
http://www.yenolam.com/writings. In the 2011 releases of their software
in particular, Apple appears to have attempted to block using Linux on
their hardware. That article describes how to install native Linux (in
that case Ubuntu) on a 2011 mac mini without disturbing OS X and
particularly not changing the firmware. I believe the same approach
will work on later hardware/software releases.

Ross

[Howard S Shubs]

In article <BcWdnT0H4Ntqt-bM...@westnet.com.au>,

Ross Maloney <rmat...@iinet.net.au> wrote:

> It might be worth having a look at the article maclinux.pdf on
> http://www.yenolam.com/writings. In the 2011 releases of their software
> in particular, Apple appears to have attempted to block using Linux on
> their hardware. That article describes how to install native Linux (in
> that case Ubuntu) on a 2011 mac mini without disturbing OS X and
> particularly not changing the firmware. I believe the same approach
> will work on later hardware/software releases.

No permissions at that URL. When I provided the document name, got the
dreaded 404.

[Thad Floryan]

This page is OK:

http://www.yenolam.com/writings/macnative.php

and a PDF (139kB, 12 pages) with all the instructions is here:

http://www.yenolam.com/writings/maclinux-1.0.pdf


Thad

[Howard S Shubs]

In article <vilain-E13D5D....@news.individual.net>,

Michael Vilain <vil...@NOspamcop.net> wrote:

> I haven't tried it, but change that external drive to Firewire and see
> if that changes anything. Post your progress.

No joy. Same symptoms.

[Howard S Shubs]

In article <517B3E4...@thadlabs.com>,

Thad Floryan <th...@thadlabs.com> wrote:

> This page is OK:
>
> http://www.yenolam.com/writings/macnative.php
>
> and a PDF (139kB, 12 pages) with all the instructions is here:
>
> http://www.yenolam.com/writings/maclinux-1.0.pdf

Got it now. Reading the document. Thanks.

[Thad Floryan]

You're very welcome!

Please report back how that works, I'm sure many others would
like to know.

Thad

[Howard S Shubs]

In article <howard-1B5787....@news.individual.net>,

Howard S Shubs <how...@shubs.net> wrote:

No joy. The device is still external, and that appears to not work.
Now, if I were to put another laptop drive in the machine and only
install Linux on it, then swap disks around in the laptop, that might
work. But it's a tad invasive.

[Howard S Shubs]

In article <klfp9d$ov5$1...@dont-email.me>,

Thad Floryan <th...@thadlabs.com> wrote:

> On 4/26/2013 9:42 PM, Howard S Shubs wrote:
> > In article <517B3E4...@thadlabs.com>,
> > Thad Floryan <th...@thadlabs.com> wrote:
> >
> >> This page is OK:
> >>
> >> http://www.yenolam.com/writings/macnative.php
> >>
> >> and a PDF (139kB, 12 pages) with all the instructions is here:
> >>
> >> http://www.yenolam.com/writings/maclinux-1.0.pdf
> >
> > Got it now. Reading the document. Thanks.
>
> You're very welcome!
>
> Please report back how that works, I'm sure many others would
> like to know.

They were dealing with two internal drives without an optical drive.
I'm on a laptop with one internal drive and an optical drive, so I can
boot the CentOS DVD. It's just that the resulting system on an external
drive, doesn't boot regardless of whether I connect it via FW or USB.
And I know Windows won't boot off an external USB drive. Don't know
about whether it'd boot via FW.

Anyway, I've figured another way to accomplish what I need. I'm
connecting to Red Hat Live Labs. That uses the "nx" client, which runs
on Linux... and Windows. I have Windows 7 running in VMware Fusion, and
I've been able to connect to Live Labs, though I don't think the
performance is as good as when it's running native on Linux.

[Tauno Voipio]

MacBook Pro boots from an external USB drive. This computer
would be a brick without. I restored it from a Time Machine
backup from an external USB drive.

The disk needs to have a GUID Partition Table, and there must
be a boot partition formatted as MacOs Extended (Journaled)
format, which will be shown in firmware boot menu (push alt
during power-up).

My preferred way of running Linux and Windows on this box
is VmWare Fusion. It handles half a dozen of Linuxes, Windows
XP and Solaris without a hitch.

--

Tauno Voipio

[Ross Maloney]

Thanks, Thad. for correcting the error.

Ross

[Ross Maloney]

The point is, if the firmware is capable of booting from the device,
then how is the device prepared to enable Linux to be booted off it.
That is the point of the article. Yes, the mac mini had two hard disks
which could be set up as boot devices. And that is what was done. The
approach in the article is in getting around the GUID Partition Table,
etc. etc that are different to good-old BIOS what the Macs do not use.
It looks as if the Mac will also boot from USB. I have not tried using
USB as the boot target, but the approach in the article should apply to
setting up the USB drive as well. Just think about what is involved in
the boot process. I cannot see any reason for the necessity of having
MacOS format if MacOS is not booted. The boot firmware is specific to
the hardware not the software being booted.

I have seen the VmWare approach. But why use emulation if a native
approach can be used? It is foreign to me to put another layer of
software between my programs and the hardware - it somewhat defeats the
purpose of having fast hardware.

Ross

[Robert Heller]

Unless the firmware expects a partitular partition table / filesystem on the
boot device. It sounds like the boot firmware in Macs is more like Grub than
the boot firmware in PCs. That is, the PC BIOS boot loads 512 bytes and
expects it to be 8086 machine code (or something like that) -- PC BIOSs don't
really have any sort of filesystem support. The Mac firmware apprears to
expect that the disk has a certain kind of partition table with a partitular
file system. I guess it 'mounts' this file system and loads a program / kernel
from this file system and fires that up. This is not really any different from
what RHEL does on generic PCs: a /boot file system with the kernel/initrd and
the rest of the disk formatted using LVM, with logical volumes (swap, root,
etc.) -- grub cannot manage LVM, but can manage ext2/3/4. The kernel, once
fired up, can deal with LVM. Basically, you use a 2 stage boot load process.
It sounds like you do one or two things:

1) Have a /boot partition & file system (MacOS file system) with the kernel
and initrd, with whatever config file(s) needed by the firmware to load and
start the kernel and initrd. Then there are additional partitions /
filesystems for root, swap, etc.

2) Have a MacOS file system with a Linux boot loader (ala Syslinux). The
Mac boot loader loads this program, which in turn looks for an *ext2/3/4*
/boot (or root) file system and loads the kernel and initrd from there and
fires that up.

>
> I have seen the VmWare approach. But why use emulation if a native
> approach can be used? It is foreign to me to put another layer of
> software between my programs and the hardware - it somewhat defeats the
> purpose of having fast hardware.
>
> Ross
>

--
Robert Heller -- 978-544-6933 / hel...@deepsoft.com
Deepwoods Software -- http://www.deepsoft.com/
() ascii ribbon campaign -- against html e-mail
/\ www.asciiribbon.org -- against proprietary attachments

[Richard Kettlewell]

Robert Heller <hel...@deepsoft.com> writes:
> Unless the firmware expects a partitular partition table / filesystem
> on the boot device. It sounds like the boot firmware in Macs is more
> like Grub than the boot firmware in PCs. That is, the PC BIOS boot
> loads 512 bytes and expects it to be 8086 machine code (or something
> like that) -- PC BIOSs don't really have any sort of filesystem
> support. The Mac firmware apprears to expect that the disk has a
> certain kind of partition table with a partitular file system.

Macs have used EFI for years now:
http://en.wikipedia.org/wiki/Unified_Extensible_Firmware_Interface

PCs are starting catch up with this.

--
http://www.greenend.org.uk/rjk/

[Warren Oates]

In article <87txms8...@araminta.anjou.terraraq.org.uk>,

And there's an app for that:

http://www.rodsbooks.com/refind/

which is known to work, and is a "reworking" of

http://refit.sourceforge.net/
--
Where's the Vangelis music?
Pris' tongue is sticking out in in the wide shot after Batty has kissed her.
They have put back more tits into the Zhora dressing room scene.
-- notes for Blade Runner

[Ross Maloney]

On 04/27/2013 10:08 PM, Warren Oates wrote:
> In article <87txms8...@araminta.anjou.terraraq.org.uk>,
> Richard Kettlewell <r...@greenend.org.uk> wrote:
>
>> Robert Heller <hel...@deepsoft.com> writes:
>> > Unless the firmware expects a partitular partition table / filesystem
>> > on the boot device. It sounds like the boot firmware in Macs is more
>> > like Grub than the boot firmware in PCs. That is, the PC BIOS boot
>> > loads 512 bytes and expects it to be 8086 machine code (or something
>> > like that) -- PC BIOSs don't really have any sort of filesystem
>> > support. The Mac firmware apprears to expect that the disk has a
>> > certain kind of partition table with a partitular file system.
>>
>> Macs have used EFI for years now:
>> http://en.wikipedia.org/wiki/Unified_Extensible_Firmware_Interface
>>
>> PCs are starting catch up with this.
>
> And there's an app for that:
>
> http://www.rodsbooks.com/refind/
>
> which is known to work, and is a "reworking" of
>
> http://refit.sourceforge.net/

My initial approach when trying to get Linux to load on the Mac was
through 'refit'. It didn't work. It may have worked on Macs prior to
2011, but not on the 2011 releases. They appeared to be continually
chasing changes being made on the Mac to EFI. I was also uneasy about
the approach they were taking of modifying bits and pieces. I have not
tried 'refind'.

When Ubuntu loads on the Mac it comes up with a Grubie 2 menu which
works to boot Linux. The menu items which appear there to boot OS X
don't work. To boot OS X I power up the Mac with my finger on the
Option (Alt) keyboard key to get a standard Mac menu. So, the Mac menu
will not boot Linux and the Linux (Grup) menu will not boot OS X. But
by having both menus, either operating system can be booted. I think
this is a good reason not to tamper with the Mac boot itself.

Ross

[Kenny McCormack]

In article <howard-E2A684....@news.individual.net>,

Howard S Shubs <how...@shubs.net> wrote:

...

>No joy. The device is still external, and that appears to not work.
>Now, if I were to put another laptop drive in the machine and only
>install Linux on it, then swap disks around in the laptop, that might
>work. But it's a tad invasive.

Can I ask what is the actual, underlying point/purpose to this?

It seems to me that it could be one of 3 things:

1) Solving a specific, personal problem. I.e., there is something you are
trying to accomplish, which is very specific to the sort of work you do
and/or your hardware and/or software config - that makes you want to do this
thing. The assumption is, of course, that very few of us (other than you)
would have this specific need.

2) Just because it is there (like why does a man climb a mountain...). The
intellectual challenge of it. And, of course, because Apple tries so hard
to prevent people from doing it.

3) A third reason that I thought of, but can no longer remember.

My point, my reason for posting this flam-ish post, is that if you want to
run Linux, do it on regular x86 hardware. It (regular x86 hardware) is so
much cheaper, faster, and better supported for Linux.

--
The motto of the GOP "base": You can't *be* a billionaire, but at least you
can vote like one.

[Howard S Shubs]

In article <kli6bc$r7k$1...@news.xmission.com>,

gaz...@shell.xmission.com (Kenny McCormack) wrote:

> Can I ask what is the actual, underlying point/purpose to this?

Saving me the trouble of bringing two computers, one of which is a
full-height tower, on a trip.

> It seems to me that it could be one of 3 things:
>
> 1) Solving a specific, personal problem. I.e., there is something you are
> trying to accomplish, which is very specific to the sort of work you do
> and/or your hardware and/or software config - that makes you want to do this
> thing. The assumption is, of course, that very few of us (other than you)
> would have this specific need.

> My point, my reason for posting this flam-ish post, is that if you want to
> run Linux, do it on regular x86 hardware. It (regular x86 hardware) is so
> much cheaper, faster, and better supported for Linux.

Don't have any other than my full-height tower, which I want to avoid
bringing. The reason is elsewhere in the thread.

[Paul Sture]

In article <howard-E68D33....@news.individual.net>,

Howard S Shubs <how...@shubs.net> wrote:

> In article <kli6bc$r7k$1...@news.xmission.com>,
> gaz...@shell.xmission.com (Kenny McCormack) wrote:
>
> > Can I ask what is the actual, underlying point/purpose to this?
>
> Saving me the trouble of bringing two computers, one of which is a
> full-height tower, on a trip.

A virtual machine appears to be the easiest option. Given the other
comments up thread my fear would be that a dual boot system could get
borked by an update during a trip when I didn't have other systems
handy to help in a recovery.

--
Paul Sture

[Warren Oates]

In article <wNGdnXog5NQXFOHM...@westnet.com.au>,

Ross Maloney <rmat...@iinet.net.au> wrote:

>
> When Ubuntu loads on the Mac it comes up with a Grubie 2 menu which
> works to boot Linux. The menu items which appear there to boot OS X
> don't work. To boot OS X I power up the Mac with my finger on the
> Option (Alt) keyboard key to get a standard Mac menu. So, the Mac menu
> will not boot Linux and the Linux (Grup) menu will not boot OS X. But
> by having both menus, either operating system can be booted. I think
> this is a good reason not to tamper with the Mac boot itself.

Sounds like a good way to operate. I wouldn't mess with it.

[Howard S Shubs]

In article <nospam-4CD6E3....@news.chingola.ch>,

Paul Sture <nos...@sture.ch> wrote:

> A virtual machine appears to be the easiest option. Given the other
> comments up thread my fear would be that a dual boot system could get
> borked by an update during a trip when I didn't have other systems
> handy to help in a recovery.

Well, the VM arrangement is working, so that's good. So far, I've tried
it with Windows 7, which worked. Now I'm setting up a CentOS 6.4 system
like my tower machine, to see if it would do the job and be easier on
resources.

The funny thing is that I upgraded the tower machine to kernel 3.8.8
last week. Then I was offered an upgrade by the Software Upgrade tool
to a later version of the 2.6 kernel. I turned that down, since I
doubted it would detect the newer kernel since I installed it via "make"
rather than via an rpm. Now I'm doing the same thing with the VM
system. It has been building the kernel for the last few hours.

I don't know if I could've just copied the kernel and associated files
from the tower, but since I'm not sure of all the associated files, I
figured I'd play it safe. Also, I noticed that kernel.org is now up to
3.8.10. :-> Hey, I know 3.8.8 works. No need to be at the
bleeding/cutting edge, especially when the change notes on 3.8.10 say
it's for adding a routine nothing is using yet, and the 3.8.9 change
notes didn't mean much to me.

The only change I'm making on the VM's kernel is to use xz compression
instead of bzip2. If it doesn't work, I'll change it back. xz is new
to me, so I should look it up anyway.