Re: [RFC][PATCH 0/5] Introduce checks for preemptable code forthis_cpu_read/write()

[Peter Zijlstra]

On Tue, 2011-09-20 at 11:10 -0500, Christoph Lameter wrote:
> On Tue, 20 Sep 2011, Steven Rostedt wrote:
> 
> > I really mean all other users of this_cpu_*(), including the cmpxchg and
> > friends, still need to have preemption disabled.
> 
> This is argument against the basic design of this_cpu_ops. They were
> designed to avoid having to disable preemption for single operations on
> per cpu data. I think this shows a basic misunderstanding of what you are
> dealing with.

But part of that design is that its impossible to verify the
correctness. This is the part we object to and you keep avoiding.

There is a reason smp_processor_id() warns if its called in a
preemptible context, all the this_cpu wankery doesn't. It doesn't
provide a single useful debug feature and in places is designed so that
its impossible.

Seriously, how can you defend this shitpile with a straight face? Sure
it make slub go faster, but who gives a flying fuck if it brings the
rest of the kernel to its knees.


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/