Re: [RFC] [PATCH 0/7] reiser4: batch discard support (FITRIM ioctl): initial implementation.

[Ivan Shapovalov <intelfx100@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1345 bytes --]

On Sunday 17 August 2014 at 18:37:11, Dušan Čolić wrote:	
> On Aug 17, 2014 6:18 PM, "Ivan Shapovalov" <intelfx100@gmail.com> wrote:
> >
> > On Sunday 17 August 2014 at 18:14:10, Dušan Čolić wrote:
> > > [...]
> > > >
> > > > Points I'm really uncertain of:
> > > > - grabbing fixed amount of space (new BA_SOME_SPACE flag, grabs 25% of
> > > disk space)
> > >
> > > Maybe grabbing 25% of remaining empty disk space would be better?
> >
> > What do you mean by "remaining"? 25% is recalculated on each iteration.
> > However, the reserved and allocated disk space is reclaimed at the end
> > of each iteration, so the amount really stays the same.
> >
> I meant not to grab 25% of whole partition size but grab amount of disk
> space equal to 25% of empty partition space so if we have only 10% of free
> space left we grab only 2.5% of it per iteration so to not get ENOSPACE if
> we used 25% of whole disk size and grabbed whole remaining disk space.

Ah. I've meant "25% of free disk space at the moment of discarding".
It's already done the correct way, I've just misworded it.

-- 
Ivan Shapovalov / intelfx /

> 
> > --
> > Ivan Shapovalov / intelfx /
> >
> > >
> > > > - creation of empty atoms (reiser4_create_atom())
> > > > - handling of empty atoms in commit_current_atom()
> > > >
> > > [...]

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 213 bytes --]