[Dev] [RFC] Tizen system rollback

Łukasz Stelmach l.stelmach at samsung.com
Mon Dec 2 15:21:24 GMT 2013


It was <2013-12-02 pon 14:11>, when Artem Bityutskiy wrote:
> On Mon, 2013-12-02 at 13:17 +0100, Łukasz Stelmach wrote:
>> >> integration of snapshots with file-system code *may* (it is yet to
>> be
>> >> measured) bring better performance. It would also allow for
>> snapshots
>> >> for /boot (p. 11).
>> >
>> >       How? Teach bootloader to deal with snapshots?
>> 
>> Teach it btrfs. As far as I know btrfs snapshots are simply
>> alternative
>> root directories (or subvolumes? Is there anyone wiser than me to
>> help?).  All you have to do is to tell the boot-loader which one to
>> "mount".
>
> I wonder, what is the boot-loader(s) we discuss?

We talk about "a boot-loader" ;-)

> I am aware of 2 major types of boot-loaders.
>
> 1. simple ones, which do not know any FS, they simply read the kernel
> from a special place where it is stored in a simple raw format. E.g.,
> this may be a special partition, or a separate flash. There may be some
> trivial format which allows for storing several kernels/initrd, may be
> changing the boot parameters, etc.

Yet another option is to provide a list of blocks to read like lilo(1)
does.

> 2. complex ones, which support file-systems.
>
> Which ones do we discuss? Dow we want to bind Tizen architecture to
> either of these? Or we want to be agnostic?
>
> A couple of random thoughts...
>
> Type 2 bootloaders have own disadvantages. E.g., the FS driver teds to
> quickly get out-of-date there. There is more complexity when we consider
> corrupted FS handling.

In case of boot-loaders this isn't that bad because you need a read-only
support and on-disk format does not change that often.

> It would be cool if the Aliaksei's architecture could be generalized so
> that it would allow for btrfs as a special case...

This would be cool indeed.

-- 
Łukasz Stelmach
Samsung R&D Institute Poland
Samsung Electronics
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 489 bytes
Desc: not available
URL: <http://lists.tizen.org/pipermail/dev/attachments/20131202/71001782/attachment.sig>


More information about the Dev mailing list