Re: grub-probe detects ext4 wronly as ext2

["Javier Martín" <lordhabbit@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 2532 bytes --]

El mié, 16-07-2008 a las 17:27 +0200, Felix Zielcke escribió:
> From: "JavierMartín" <lordhabbit@gmail.com>
> Sent: Wednesday, July 16, 2008 5:09 PM
> To: "The development of GRUB 2" <grub-devel@gnu.org>
> Subject: Re: grub-probe detects ext4 wronly as ext2
> 
> > I see the ext4 patch was checked in recently. Can the "forbid-incompat"
> > patch with the new, specific error messages be committed too then? I'm
> > submitting an updated version (i.e. against the current HEAD) because
> > new lines were added.
> >
> > PS: does the ext4 patch add support for META_BG? it should be added to
> > the list of supported incompat features then.
> 
> I don't know what this META_BG is but even the ext2 kernel driver supports 
> it [0]
Er... of course, the Linux extN implementation is the de-facto reference
implementation. Some incompat features are only used in newer versions
like ext3 and ext4, while others are added to ext2 too. I was talking
about the GRUB extN driver, which recently got patched by Bean.
> Maybe the list of flags have a bit changed so I'm so nice and give you even 
> a git link to the current ext4.h in Linus' official git tree [1]
> You'll probably mean FLEX_BG :)
All those flags make my head spin so fast I'll create a dark hole
through gravitomagnetic effects. I no longer know what the hell does
each one do T_T
> 
> I didn't take a deep look at the changes between the first patch from Bean 
> and the last one which he commited.
> But for me it's now working fine with whole / on ext4 made with the final 
> e2fsprogs 1.41 in Debian unstable,
> with flex_bg,extents and uninit_bg from the INCOMPAT list, so flex_bg and 
> uninit_bg should be added to your list which are ignored/supported
Uninit_bg is signaled (iIrc) in the superblock by a ROCOMPAT flag,
GDT_CSUM, and then in the block groups by whatever-it-is (head spinning
even faster).
> 
> Maybe it's just luck for me that it works now with uninit_bg and flex_bg, 
> the best would be if other people would test it
AFAIK, uninit_bg should work if the (readonly) GRUB reader respects the
spec and skips "invalid" block groups/inodes/whatever (those that
haven't been initialized). As I don't know what the f*** do META_BG and
FLEX_BG do, I can't tell you whether they truly work or it's just a
matter of luck: it's Bean who can tell us whether or not he implemented
support for them - I only added the "extents" flag to the supported list
on my patch, but including more is a matter of seconds.


[-- Attachment #2: Esta parte del mensaje está firmada digitalmente --]
[-- Type: application/pgp-signature, Size: 827 bytes --]