Re: ASRock A785GMH/128M fails to find devices when in RAID mode and disks are present

[Xen]

Andrei Borzenkov schreef op 31-07-2016 7:02:
> You still provided no evidence that any bootloader sees disks in BIOS 
> on
> this system.

How can I test? Is it relevant? How do I find out?

Would Windows installation on a RAID disk constitute a test?


> This is completely unrelated issue and could be one fixed by
> 3bca85b4184f74995a7cc2791e432173fde26d34.

Oh, thanks, I will compile. Ubuntu grub on this system is ancient (in 
general it is).

Look, I created a repo for my pvinstall fork: :P.

https://github.com/drydenp/grub2-pvinstall

Okay, I will stop bugging you now.

I can at least confirm that the Git version of Grub2 also won't find any 
devices although it has a nicer boot prompt ;-).

Something friendlier.

Just incredible that the 16.04 Ubuntu version still includes patches 
from yourself (ie. fc535b32b9fe0cd84213724ffba43b98e2da6ab4) as an 
additional patch they apply to the base version, which means that the 
base version is how old? That patch was from februari 2015.

I guess if you have a lot of patches to apply but still. The number of 
patches is not greater than 100 and they should be able to handle that.

No less than 41 of those are patches from upstream, backported, either 
from Ubuntu or Debian devs themselves, or clean from Git.

The oldest of which is from January 2014. *Shakes head*. So the used 
version is even older than that. Of the 41 backported patches 8 were 
from Colin Watson. So... there are 59 in there that are unique to 
Ubuntu/Debian, and 44 they won't need to repeat. 8 out of 67 were 
accepted by upstream first. They basically - given that release - only 
have to repeat 84% of their original work as 16% was back then already 
accepted upstream.

Releasing a new (upstream) version then would only require 59 patches to 
be re-applied, possibly less. 27 of those still cleanly apply to git and 
32 fail.

Given that they are probably independent of each other, if not in 
concept, then at least technically, that means only 48% of the original 
work needs to be repeated or adjusted/fixed, and 40% needs no work. Of 
the unique patches, only roughly half needs to be repeated.

58 out of 212 total hunks fail and 91 hunks succeed of all the failed 
patches, meaning that only 39% of hunks do not apply of the failed 
patches.

:P.

Anyway. I gotta get going.

5 of the supposedly 32 failed patches are actually "reverse patches" 
(forgot to check) and apply cleanly, so were accepted after release. 2 
or 3 fail in part. So the numbers go down to at least 27 failed patches 
instead of 32, making the total numbers 27 that apply cleanly and 27 
that fail.

Some of the above statistics contain failed hunks that were actually 
reverse hunks that succeeded, so the number of total failed hunks goes 
down to at least 30 or less. Anyway, no time.