Re: [PATCH,HURD] Fix grub-probe with userland partition support

[Vladimir 'φ-coder/phcoder' Serbinenko]

On 24.04.2012 19:13, Samuel Thibault wrote:
> Vladimir 'φ-coder/phcoder' Serbinenko, le Tue 24 Apr 2012 15:19:25 +0200, a 
> écrit :
>> On 24.04.2012 14:42, Samuel Thibault wrote:
>>> Vladimir 'φ-coder/phcoder' Serbinenko, le Tue 24 Apr 2012 11:55:33 +0200, a 
>>> écrit :
>>>> They have iso9660 spanning through
>>>> the whole disk but all of the disk other than the first sector is
>>>> in some kind of partition table to avoid it being accidentally
>>>> overwritten.  So even though the file itself is inside a partition, we
>>>> want the whole disk.
>>> The partition itself can not be mounted?
>> No, since it contains FS structures at wrong offset.
> Ok.
BTW sometimes partition can be mounted but contains another FS which
references the same file by another name or not at all.
>>> 0-sized indeed poses problems.
>>>
>>>> I'm surprised that Hurd doesn't offer a way to just ask "What does this
>>>> filesystem translator consume?"
>>> Because the whole point of the Hurd is to let the user have access
>>> to more powerful ways.  A file can reside inside an iso file, which
>>> is stored in an ext2fs, which is stored in a file,
>> So much GRUB can handle.
> But how to express that to GRUB? grub_guess_root_devices only returns
> a series of alternative paths.  See below.
By just giving the file in question.
But it currently always suppose that the file will be for VM and not
host use.
>> We don't handle loopback automatically right now since it's not clear
>> whether it's a loopback for VM or loopback used on host.
> You'd need to know which files to open anyway. You don't want to browse
> all filesystems for that :) So the OS-specific function has to tell you.
Yes.
>> As for the second, we're limited to what GRUB can do and so it won't be
>> possible to have /boot on translator from hyperspace.
> Sure.  But it can be something expressed in a complex way by the user,
> which can actually be reached by GRUB.  That said, as I said earlier, we
> can ask the user to refrain himself when it's about /boot.
We can add more translators handling as-needed. But this problem isn't
Hurd-specific. A simple GNU/Linux example is unionfs: currently we can't
traverse it even if it refers to something easy. Trouble is that
unionfs/translator reference "foo" can point to different files
depending on circumstances and we need to follow this changes if they
happen behind the grub-mkconfig back (e.g. if unionfs had kernel just in
branch1 and then it was moved to branch2 we would need to handle this
change without rerunning grub-mkconfig).
>>> But we generally don't want to impose any syntax here, it could actually
>>> be
>>>
>>> /opt/my/own/translator xyz
>>>
>>> I guess we'll have to impose some syntax anyway for whatever contains
>>> /boot, so that grub can open it itself.
>> There should be a standartised way to get this information for any
>> conventional FS, otherwise it makes porting programs which use this
>> information much more difficult and in most cases results in dirty
>> workarounds.
> So far, I've mostly seen GRUB really needing that information.
I suppose that databases would want to know this for optimisations.

-- 
Regards
Vladimir 'φ-coder/phcoder' Serbinenko

signature.asc
Description: OpenPGP digital signature