It's big, complex and looks fragile. And it reports lots of false
positives, with big warnings to people to generally not use or make
reports about itit unless they are already experts. So it looks like
something pretty specific.
I also want to feel like it actually has been helpful.
But the _real_ issue is that I'm simply merging about 20+ trees per day,
and I want to know what I merge, and I want to feel comfy about it as I
merge it.
And kmemcheck didn't make me do that, adn I didn't feel like I had the
time to really look at it any closer. So while I might merge it before the
merge window is open, there's simply no way I'll merge it today (and
almost certainly not this week, considering that I'll be on the coast for
a long weekend).
IOW, I'm quite busy enough merging _other_ things, and handling the
fallout from those other merges (mostly the firmware discussion ;). So I'm
simply not interested in kmemcheck right now - I have enough to keep me
busy.
So kmemcheck in no way made me go: "ok, we _need_ to merge this", and at
the same time it's biggish and scary and complex enough that I felt I
needed to take the time to think about it - which I don't feel I have
right now.
The RCU thing, on the other hand, looked like a no-brainer from a merge
standpoint. However, since it had merged kmemcheck, it didn't get merged
either. I'll take a look at Ingo's second version that didn't tie the two
together.
This is why tying things together doesn't necessarily work, and why I told
Ingo that I'm not pulling the RCU thing just because I wasn't ready to
pull kmemcheck.
Linus
--
