Re: New filesystem for Linux

[Mikulas Patocka <mikulas@artax.karlin.mff.cuni.cz>]

On Fri, 3 Nov 2006, Nikita Danilov wrote:

> Mikulas Patocka writes:
> > > Mikulas Patocka <mikulas@artax.karlin.mff.cuni.cz> writes:
> > >
> > >> new method to keep data consistent in case of crashes (instead
> > >> of journaling),
> > >
> > > What is that method?
> >
> > Some tricks to avoid journal --- see
> > http://artax.karlin.mff.cuni.cz/~mikulas/spadfs/download/INTERNALS
> >
> > --- unlike journaling it survives only 65536 crashes :)
>
> What happens when hard-linked file is accessed, and it is found that
> last fnode (one in fixed_fnode_block), has wrong "crash count"?
>
> Nikita.

Fixed fnode block contains (txc,cc) pair describing which fnode and nlink 
count is valid. --- currently two fnodes are superflous (there could be 
just one), they are reserved for the possibility to atomically modify 
extended attributes --- but there is no code currently that does it.

The fnodes live on their own with their own (txc,cc) pair --- it is a bit 
confusing to have pair on both fixed_fnode_block and fnode --- the reason 
is that the code for handling fnodes in directories can be reused to 
handle fnodes in fixed_fnode_blocks and I can avoid many
if (is_fnode_fixed()) branches.

If the fnode in fixed_fnode_block has invalid crash count, 
fixed_fnode_block's (cc,txc) pair should never point to it. Or did it 
happen to you?

Mikulas
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/