Thus spake Benjamin Herrenschmidt (benh@kernel.crashing.org):

On Mon, 2009-09-28 at 18:21 -0700, Rex Feany wrote:

quoted

quoted

It's going to be hard for me to get that "right" since I don't really
know what's going on with the core here, but I suppose if we get it
moving along with extra tlb invalidations, that should be "good enough"
until somebody who really knows what's going on comes up with possibly
a better fix.

I've tried sticking tlbil_va() in those places, nothing seems to help.
In some cases userspace is slow, in other cases userspace is faster and
unstable: sometimes commands hang, sometimes I am able to ctrl-c and
and kill it, sometimes I get other strange crashes or falures (so far no
kernel oopses though).

And you are positive that with 2.6.31 and your other patch, it works
both fast and stable ? This is strange... the code should be mostly
identical. I'll have a second look and see if I can get you a patch that
reproduce -exactly- the behaviour of 2.6.31 plus your patch.

The difference is night and day - with 2.6.31 I can boot into single
user mode, I can run our custom software (I left it running over night
with very simple test script - no crashes).

With the top of the tree I can sometimes boot into a shell, and if it
isn't crashing or hanging on boot it runs very slow (10+ seconds to do
anything, if I am lucky).

thanks!
/rex