Re: [PATCH v2 0/9] refcount: Improve code-gen

From: Peter Zijlstra
Date: Mon Dec 13 2021 - 11:12:05 EST


On Mon, Dec 13, 2021 at 03:42:34PM +0100, Marco Elver wrote:
> On Mon, Dec 13, 2021 at 01:20PM +0100, Peter Zijlstra wrote:
> >
> > Also, since this will conflict most mighty with Marco's kcsan work in
> > the rcu tree, I've written the below patch as a replacement for the
> > instrumentation patch in that series.
> >
> > It depends on other bits in that series so it cannot be applied here,
> > but will come in handy when trying to resolve the merge conflict.
> >
> > ---
> > Subject: locking/atomics, kcsan: Add instrumentation for barriers
> > From: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> > Date: Thu Dec 9 17:08:16 CET 2021
> >
> > Adds the required KCSAN instrumentation for barriers of atomics.
> >
> > [adapted from https://lkml.kernel.org/r/20211130114433.2580590-17-elver@xxxxxxxxxx]
> > Originally-by: Marco Elver <elver@xxxxxxxxxx>
> > Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
>
> Thanks!
>
> I wanted to test this, but can't quite figure out which branches I need
> to merge together to get it to apply. I assume -rcu + your series? Or
> the updated series not yet sent?

I think I applied the patches from your series up-to, but not including,
the atomic-instrument patch, then the first few patches of this posting
which cause the rejects and then this patch.

I've not attempted the actual git-merge or did much testing on the
results.