May 11, 2014, 22:00 (0 Talkback[s])
(Other stories by Sean Michael Kerner)
Re-Imagining Linux Platforms to Meet the Needs of Cloud Service Providers
Yes, I'm aware that this is two days early. The normal schedule has
been for me to do Sunday releases, but this time around I have a
combination of travel (which would have pushed the release to Saturday
morning from the airport as is oft my wont when traveling) and the
fact that rc5 has actually already grown to be larger than rc3 or rc4
So instead of pushing it to the last minute before I board a plane and
am off-line for a week, I decided that there is absolutely zero reason
for that kind of just-under-the-wire release timing. I'd rather do a
leisurely release on a Friday afternoon than a hurried one tomorrow
morning before then disappearing for a week.
Anyway, enough explaining. rc5 is out there, and while I'd have been
happier if it had been as small as rc4 was, it seems to all be solid
fixes (famous last words). The interesting dcache list corruption I
mentioned as being pending for rc4 is in, and it would be lovely if
you have any VFS layer stress-testing that interacts with memory
pressure, but the race was tiny to begin with, and the fix actually
cleaned things up a lot and removed more lines than it added, so I
feel good about it.
Apart from that one interesting really core change (where "really
core" is defined as "an area I personally care about and muck around
with", and not meant to be a value statement in any other way ;), it
all looks boringly familiar: 55% drivers, 20% architecture updates,
and 25% misc (filesystems, core networking, VM, etc).
And while rc5 may be bigger than rc3/4 were, it's not like it is
worrying. This merge window was bigger than most, and the fact that
rc5 is then slightly bigger than most isn't something that worries me
overmuch. And since rc4 was smaller than usual, it all evens out.
But I really *will* be entirely unreachable all next week, so get your
testing in, because the -git tree will be very quiet.