5:08 a.m.
As you all might know, it is currently impossible to boot the default
install of F16 on Macs. It does not seem to matter if you use boot
loaders such as reFit and Bootcamp or remove MacOS completely and try
to boot. Apparently, if you somehow (how to use it is not very clear
from the "Common F16 bugs" page,
http://fedoraproject.org/wiki/Common_F16_bugs#Incorrect_partition_type_as...)
use the nogpt kernel parameter you might be able to install F16.
There are a couple of bugs open regarding this,
https://bugzilla.redhat.com/show_bug.cgi?id=752548 and
https://bugzilla.redhat.com/show_bug.cgi?id=746895 are two examples.
It seems like the boot procedure has changed quite a bit from F15 to
F16, and not just the GRUB2 change. I am sure these were changed for
the best of reasons, but it is very frustrating to not be able to use
F16 on hardware that could run previous Fedoras very well. Also, there
are a lot of Macs out in the wild, and by excluding them we shut out a
large amount of (possible users).
So what could be done about this? Would it be enough to have a visible
nopgt option in Anaconda? Should we do special "Boot test days"? Or is
something else needed?
/Andreas
12:20 p.m.
On Fri, 2011-12-02 at 12:08 +0100, Andreas Tunek wrote:
As you all might know, it is currently impossible to boot the
default
install of F16 on Macs.
I'm not entirely sure that's correct. We had several reports of
successful EFI installs of F16 in the relevant bug reports at RC stage.
What are you basing this apparently authoritative declaration on?
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
12:48 p.m.
2011/12/2 Adam Williamson <awilliam(a)redhat.com>:
On Fri, 2011-12-02 at 12:08 +0100, Andreas Tunek wrote:
> As you all might know, it is currently impossible to boot the default
> install of F16 on Macs.
I'm not entirely sure that's correct. We had several reports of
successful EFI installs of F16 in the relevant bug reports at RC stage.
What are you basing this apparently authoritative declaration on?
--
As far as I know, you can only install if you use the nopgt kernel
parameter or change the partitions after the install. I could of
course be wrong, but nothing I found contradicted these facts.
On my own iMac, when I tried to do preupgrade from the working F15 to
F16 the system would not boot into F16. I the tried to install the
live CD and I could not get that to boot either. I tried both using
reFit and using the whole disk.
Has anyone managed to install F16 on a Mac without using any extra
options such as nopgt or changing partion flags after install?
/Andreas
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
--
devel mailing list
devel(a)lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
12:55 p.m.
On Fri, 2011-12-02 at 19:48 +0100, Andreas Tunek wrote:
2011/12/2 Adam Williamson <awilliam(a)redhat.com>:
> On Fri, 2011-12-02 at 12:08 +0100, Andreas Tunek wrote:
>> As you all might know, it is currently impossible to boot the default
>> install of F16 on Macs.
>
> I'm not entirely sure that's correct. We had several reports of
> successful EFI installs of F16 in the relevant bug reports at RC stage.
>
> What are you basing this apparently authoritative declaration on?
> --
As far as I know, you can only install if you use the nopgt kernel
parameter or change the partitions after the install. I could of
course be wrong, but nothing I found contradicted these facts.
On my own iMac, when I tried to do preupgrade from the working F15 to
F16 the system would not boot into F16. I the tried to install the
live CD and I could not get that to boot either. I tried both using
reFit and using the whole disk.
Has anyone managed to install F16 on a Mac without using any extra
options such as nopgt or changing partion flags after install?
Probably the guy who has the best handle on it is Chris Murphy, though I
don't know if he follows this list. He did quite a lot of testing of F16
on Macs and seems to have the best understanding of how exactly Apple's
Very Very Special firmware works. See
https://bugzilla.redhat.com/show_bug.cgi?id=752548#c4 .
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
1:54 p.m.
On 12/02/2011 10:20 AM, Adam Williamson wrote:
On Fri, 2011-12-02 at 12:08 +0100, Andreas Tunek wrote:
> As you all might know, it is currently impossible to boot the default
> install of F16 on Macs.
I'm not entirely sure that's correct. We had several reports
of
successful EFI installs of F16 in the relevant bug reports at RC stage.
What are you basing this apparently authoritative declaration on?
https://bugzilla.redhat.com/show_bug.cgi?id=751147
The Fedora 16 install DVD x86_64 won't even boot on my MacBookPro.
--
2:02 p.m.
On Fri, 2011-12-02 at 11:54 -0800, John Reiser wrote:
On 12/02/2011 10:20 AM, Adam Williamson wrote:
> On Fri, 2011-12-02 at 12:08 +0100, Andreas Tunek wrote:
>> As you all might know, it is currently impossible to boot the default
>> install of F16 on Macs.
> I'm not entirely sure that's correct. We had several reports of
> successful EFI installs of F16 in the relevant bug reports at RC stage.
>
> What are you basing this apparently authoritative declaration on?
https://bugzilla.redhat.com/show_bug.cgi?id=751147
The Fedora 16 install DVD x86_64 won't even boot on my MacBookPro.
efibootmgr is entirely the wrong component for that bug, FWIW.
reassigning.
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
2:21 p.m.
On Fri, 2011-12-02 at 11:54 -0800, John Reiser wrote:
On 12/02/2011 10:20 AM, Adam Williamson wrote:
> On Fri, 2011-12-02 at 12:08 +0100, Andreas Tunek wrote:
>> As you all might know, it is currently impossible to boot the default
>> install of F16 on Macs.
> I'm not entirely sure that's correct. We had several reports of
> successful EFI installs of F16 in the relevant bug reports at RC stage.
>
> What are you basing this apparently authoritative declaration on?
https://bugzilla.redhat.com/show_bug.cgi?id=751147
The Fedora 16 install DVD x86_64 won't even boot on my MacBookPro.
We explicitly don't support native EFI installs on Macs, at the request
of the anaconda team, as Apple's EFI implementation is hideous. Having
said that, I'm pretty sure at least a few people reported success on
Macs with the changes to grub2 between Beta and Final, but I'm having
trouble digging up the references.
Macs are always a puzzle and a mess, don't get me wrong. But I didn't
think the situation was as black and white as 'you cannot install F16 on
a Mac'.
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
3:39 p.m.
On Fri, Dec 02, 2011 at 12:21:49PM -0800, Adam Williamson wrote:
We explicitly don't support native EFI installs on Macs, at the
request
of the anaconda team, as Apple's EFI implementation is hideous. Having
said that, I'm pretty sure at least a few people reported success on
Macs with the changes to grub2 between Beta and Final, but I'm having
trouble digging up the references.
Macs are always a puzzle and a mess, don't get me wrong. But I didn't
think the situation was as black and white as 'you cannot install F16 on
a Mac'.
Macs won't do legacy boots off gpt partitions.
--
Matthew Garrett | mjg59(a)srcf.ucam.org
4:12 p.m.
On Fri, 2011-12-02 at 21:39 +0000, Matthew Garrett wrote:
On Fri, Dec 02, 2011 at 12:21:49PM -0800, Adam Williamson wrote:
> We explicitly don't support native EFI installs on Macs, at the request
> of the anaconda team, as Apple's EFI implementation is hideous. Having
> said that, I'm pretty sure at least a few people reported success on
> Macs with the changes to grub2 between Beta and Final, but I'm having
> trouble digging up the references.
>
> Macs are always a puzzle and a mess, don't get me wrong. But I didn't
> think the situation was as black and white as 'you cannot install F16 on
> a Mac'.
Macs won't do legacy boots off gpt partitions.
Okay, so if you have one of the Macs we can't do EFI boot on, the
situation is indeed 'you must install via BIOS emulation and use 'nogpt'
parameter', it seems.
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
11:19 a.m.
2011/12/2 Adam Williamson <awilliam(a)redhat.com>:
On Fri, 2011-12-02 at 21:39 +0000, Matthew Garrett wrote:
> On Fri, Dec 02, 2011 at 12:21:49PM -0800, Adam Williamson wrote:
> > We explicitly don't support native EFI installs on Macs, at the request
> > of the anaconda team, as Apple's EFI implementation is hideous. Having
> > said that, I'm pretty sure at least a few people reported success on
> > Macs with the changes to grub2 between Beta and Final, but I'm having
> > trouble digging up the references.
> >
> > Macs are always a puzzle and a mess, don't get me wrong. But I didn't
> > think the situation was as black and white as 'you cannot install F16 on
> > a Mac'.
>
> Macs won't do legacy boots off gpt partitions.
Okay, so if you have one of the Macs we can't do EFI boot on, the
situation is indeed 'you must install via BIOS emulation and use 'nogpt'
parameter', it seems.
--
Yes, and the question becomes, can we make this easier in F17?
/Andreas
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
--
devel mailing list
devel(a)lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
12:49 p.m.
On Sat, 2011-12-03 at 17:54 +0000, Matthew Garrett wrote:
On Sat, Dec 03, 2011 at 06:19:27PM +0100, Andreas Tunek wrote:
> Yes, and the question becomes, can we make this easier in F17?
Yes, we can do EFI installs.
We could also blacklist all Apples to use MS-DOS partitions if you do a
BIOS install. It's in fact trivially easy - given where the blacklist is
in anaconda's code, it naturally applies only to BIOS installs, not EFI
installs.
So really all we'd need to do is add a single line to the Lenovo
blacklist, most probably. And if Apples can't boot GPT disks via BIOS
and Apple has no intention of fixing this, as seems to be the case, it
probably makes sense to do so.
Anaconda team, WDYT? Should I open a bug report to have the blacklist
extended?
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
1:16 p.m.
On Mon, Dec 05, 2011 at 10:49:39AM -0800, Adam Williamson wrote:
On Sat, 2011-12-03 at 17:54 +0000, Matthew Garrett wrote:
> On Sat, Dec 03, 2011 at 06:19:27PM +0100, Andreas Tunek wrote:
>
> > Yes, and the question becomes, can we make this easier in F17?
>
> Yes, we can do EFI installs.
We could also blacklist all Apples to use MS-DOS partitions if you do a
BIOS install. It's in fact trivially easy - given where the blacklist is
in anaconda's code, it naturally applies only to BIOS installs, not EFI
installs.
That would be a bad idea, given that Apples use GPT. You'd end up with
an inconsistent disklabel.
--
Matthew Garrett | mjg59(a)srcf.ucam.org
2:52 p.m.
On Mon, 2011-12-05 at 19:16 +0000, Matthew Garrett wrote:
On Mon, Dec 05, 2011 at 10:49:39AM -0800, Adam Williamson wrote:
> On Sat, 2011-12-03 at 17:54 +0000, Matthew Garrett wrote:
> > On Sat, Dec 03, 2011 at 06:19:27PM +0100, Andreas Tunek wrote:
> >
> > > Yes, and the question becomes, can we make this easier in F17?
> >
> > Yes, we can do EFI installs.
>
> We could also blacklist all Apples to use MS-DOS partitions if you do a
> BIOS install. It's in fact trivially easy - given where the blacklist is
> in anaconda's code, it naturally applies only to BIOS installs, not EFI
> installs.
That would be a bad idea, given that Apples use GPT. You'd end up with
an inconsistent disklabel.
Well, it only applies when Fedora would be rewriting the disk label
anyway, and it only does *that* if you're formatting the entire disk. If
you're not formatting the entire disk, no matter what hardware you're
installing to or what any blacklists say or anything like that, anaconda
is not going to write a new disk label. It will just do its best to work
with whatever's already there. It only writes a new disk label (and
follows the logic paths to determine what type that disk label will be)
when you're entirely formatting the disk.
As discussed on IRC, though, if we think doing BIOS installs on Macs is
fundamentally a bad idea, we should just write anaconda to stop /
heavily discourage you from doing that instead.
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
6:14 p.m.
On Fri, 2011-12-02 at 12:08 +0100, Andreas Tunek wrote:
As you all might know, it is currently impossible to boot the
default
install of F16 on Macs. It does not seem to matter if you use boot
loaders such as reFit and Bootcamp or remove MacOS completely and try
to boot. Apparently, if you somehow (how to use it is not very clear
from the "Common F16 bugs" page,
http://fedoraproject.org/wiki/Common_F16_bugs#Incorrect_partition_type_as...)
use the nogpt kernel parameter you might be able to install F16.
There are a couple of bugs open regarding this,
https://bugzilla.redhat.com/show_bug.cgi?id=752548 and
https://bugzilla.redhat.com/show_bug.cgi?id=746895 are two examples.
It seems like the boot procedure has changed quite a bit from F15 to
F16, and not just the GRUB2 change. I am sure these were changed for
the best of reasons, but it is very frustrating to not be able to use
F16 on hardware that could run previous Fedoras very well. Also, there
are a lot of Macs out in the wild, and by excluding them we shut out a
large amount of (possible users).
So what could be done about this? Would it be enough to have a visible
nopgt option in Anaconda? Should we do special "Boot test days"? Or is
something else needed?
So, revisiting this: right now it seems like the aim for F17 is to focus
on native EFI boot and install on Mac, not installation using BIOS
compatibility mode. We should aim to fix any remaining issues with
booting Macs via EFI and test throughout F17 cycle to ensure it works.
In the past we have not supported EFI boot on Macs because the Apple
implementation of EFI is rather non-standard, but apparently just trying
to work with Apple's EFI quirks is our best shot at getting reliable
installs on Macs in future.
That's as I understand it from this morning's #anaconda discussion,
anyway.
As far as F16 goes, my current understanding is that it is possible to
install F16 onto some Macs via native EFI boot; however, we do have
problems booting some other Macs natively via EFI with F16. If you try
to do an F16 install in BIOS compatibility mode it may work and may not,
much like previous releases, but if you do it in such a way that Fedora
gets to format an entire disk, it will write a GPT disk label and that
certainly won't work for BIOS compatibility booting. I'm honestly not
sure of all the implications of the 'hybrid GPT' approach Macs can use
for dual-booting OS X (via EFI) and Windows (via BIOS compatibility)
with all this, I think that's where a lot of the complexity we see in
real-world Mac install cases comes in; one of the advantages of making
EFI our 'officially supported' way to install on Macs in future is we
don't have to worry about that, I guess.
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
1:32 p.m.
2011/12/6 Adam Williamson <awilliam(a)redhat.com>:
On Fri, 2011-12-02 at 12:08 +0100, Andreas Tunek wrote:
> As you all might know, it is currently impossible to boot the default
> install of F16 on Macs. It does not seem to matter if you use boot
> loaders such as reFit and Bootcamp or remove MacOS completely and try
> to boot. Apparently, if you somehow (how to use it is not very clear
> from the "Common F16 bugs" page,
>
http://fedoraproject.org/wiki/Common_F16_bugs#Incorrect_partition_type_as...)
> use the nogpt kernel parameter you might be able to install F16.
>
> There are a couple of bugs open regarding this,
> https://bugzilla.redhat.com/show_bug.cgi?id=752548 and
> https://bugzilla.redhat.com/show_bug.cgi?id=746895 are two examples.
>
> It seems like the boot procedure has changed quite a bit from F15 to
> F16, and not just the GRUB2 change. I am sure these were changed for
> the best of reasons, but it is very frustrating to not be able to use
> F16 on hardware that could run previous Fedoras very well. Also, there
> are a lot of Macs out in the wild, and by excluding them we shut out a
> large amount of (possible users).
>
> So what could be done about this? Would it be enough to have a visible
> nopgt option in Anaconda? Should we do special "Boot test days"? Or is
> something else needed?
So, revisiting this: right now it seems like the aim for F17 is to focus
on native EFI boot and install on Mac, not installation using BIOS
compatibility mode. We should aim to fix any remaining issues with
booting Macs via EFI and test throughout F17 cycle to ensure it works.
Will it be possible to do dualboot (MacOS + Fedora) and tripleboot
(MacOS + Windows + Fedora) with this method?
/Andreas
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
--
devel mailing list
devel(a)lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
2:20 p.m.
On Tue, 2011-12-06 at 20:32 +0100, Andreas Tunek wrote:
>> It seems like the boot procedure has changed quite a bit
from F15 to
>> F16, and not just the GRUB2 change. I am sure these were changed for
>> the best of reasons, but it is very frustrating to not be able to use
>> F16 on hardware that could run previous Fedoras very well. Also, there
>> are a lot of Macs out in the wild, and by excluding them we shut out a
>> large amount of (possible users).
>>
>> So what could be done about this? Would it be enough to have a visible
>> nopgt option in Anaconda? Should we do special "Boot test days"? Or
is
>> something else needed?
>
> So, revisiting this: right now it seems like the aim for F17 is to focus
> on native EFI boot and install on Mac, not installation using BIOS
> compatibility mode. We should aim to fix any remaining issues with
> booting Macs via EFI and test throughout F17 cycle to ensure it works.
>
Will it be possible to do dualboot (MacOS + Fedora) and tripleboot
(MacOS + Windows + Fedora) with this method?
If you do an EFI install of Windows (as I believe at least Win7 is
capable of), I'd say almost certainly 'yes'.
If you want to use the hybrid MBR thingy, which as I understand it
allows OS X (and hopefully Fedora) to see the disk as a GPT disk and
Windows to see it as an MBR disk, and do a BIOS emulation install of
Windows alongside EFI installs of OS X and Fedora...er, the answer I
guess is 'I really don't know'. Someone else may. I don't know the
status of this at present and I don't know if, if it doesn't work right
now, we can do anything to fix it.
(I'm considering buying a cheap second-hand Mac just so I can poke at
this mess personally...)
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net
4550
days inactive
4554
days old
16 comments
4 participants
participants (4)
-
Adam Williamson -
Andreas Tunek -
John Reiser -
Matthew Garrett