Re: [git patches] libata fixes

[Andrew Morton <akpm@linux-foundation.org>]

On Wed, 28 Mar 2007 13:53:10 -0700 (PDT)
Linus Torvalds <torvalds@linux-foundation.org> wrote:

> 
> 
> On Wed, 28 Mar 2007, Jeff Garzik wrote:
> >
> > Jeff Garzik wrote:
> > > This disables libata ACPI, among other things.
> > 
> > If a -rc6 is possible, that would be quite nice...
> 
> Heh. I don't think -rc6 is "possible" - it's "inevitable". We have too 
> much fallout from the timer changes still outstanding. It looks people 
> finally figured out a big issue with the HPET timer, and that hopefully 
> resolves most of the remaining timer-related regressions, but yes, we most 
> definitely _will_ have an -rc6.

yup.

> Andrew, what's your feeling apart from the timer fallout?

There are two main metrics:

a) the number of bugs which Adrian is tracking.  I think this still
   exceeds 25, which is a lot.

b) the rate at which fixes are arriving.  I have accumulated 15-20
   since the last batch (40 hours ago), which is still a pretty high rate.

Based on that, we're still quite a long way from -final.

(But you know me - I'd be happy releasing 2.6.21 in July)
(Don't ask me what year I'm referring to, either)


There is another metric to look at, too: the number of fixes which are
going into 2.6.x.y.  If that fix count is high, and if those fixes fix bugs
which were not present in 2.6.x-1 then this is an indication that something
is wrong - many regressions are sneaking through the -rc process.

And I haven't run the numbers, but I get the impression that 2.6.20.x has
an unusually large number of fixes in it.