Re: Fix powerTOP regression with 2.6.39-rc5

[Ingo Molnar <mingo@elte.hu>]

* Linus Torvalds <torvalds@linux-foundation.org> wrote:

> On Fri, May 6, 2011 at 1:20 PM, Steven Rostedt <rostedt@goodmis.org> wrote:
> >
> > I strongly NACK this!
> 
> Doesn't matter.
> 
> Binary compatibility is more important.

Yes, absolutely, violently agreed.

Acked-by: Ingo Molnar <mingo@elte.hu>

Steve, we had this argument again and again internally, and you still do not 
seem to understand it: viable tooling is *way* more important than the 
short-term, marginal cleanliness interests of kernel developers. We wont be 
able to merge ftrace into perf until you understand this principle ...

Arjan, Steve, i think we need to create a 'perf test' testcase for ftrace 
events as well, to catch such ABI breakages faster, hm? It took a couple of 
months for this breakage to surface and that's clearly too slow.

> And if binaries don't use the interface to parse the format (or just parse it 
> wrongly - see the fairly recent example of adding uuid's to 
> /proc/self/mountinfo), then it's a regression.
> 
> And regressions get reverted, unless there are security issues or similar 
> that makes us go "Oh Gods, we really have to break things".
> 
> I don't understand why this simple logic is so hard for some kernel 
> developers to understand. Reality matters. Your personal wishes matter NOT AT 
> ALL.

You have just summed up the main philosophical difference between perf and 
ftrace: with perf we have a "sane tooling first" approach, while ftrace is 
still the old "kernel developers first" approach.

In the past 10 years i pushed tons of instrumentation code upstream and for a 
long time the kernel-integrated ftrace approach looked like the technical best 
solution to me, but after 2 years of sane instrumentation tooling via a proper 
user-space ABI and tools/perf/ i'm not looking back.

I am strongly convinced that we need to bite the bullet and unify the two 
approaches to enable even better tooling: expose the remaining bits of tracing 
functionality not available via perf yet via the perf ABI and move it under a 
single umbrella, slowly phase out the ABI-unstable /debug/tracing/ debugfs crap 
for new features and use the strict perf ABI approach. Steve?

Thanks,

	Ingo