On 04.05.10 11:02:45, Linus Torvalds wrote:
> That said, I would suggest you stop using oprofile. It's a total piece of
> sh*t how it tries to match those idiotic string names to anything. That
> interface is _fundamentally_ broken, and can never work for new models,
> and by "new", I obviously mean "anything more recent than about two years"
> going by past experience.
> The "perf" tools are way more pleasant to use - they have a better
> interface, a saner setup, and they give much more useful information.
we all agree the perf architecture is better than oprofile's. But
oprofile was there when other subsystems like perf or ftrace did not
even exist yet. So oprofile did work for years and we shouldn't forget
We also agree that we do not rework oprofile to make its architecture
better, there is no sense doing this as we write perf from scratch. In
the past months I posted by far more patches for perf to lkml than for
oprofile. We want to have all features that are in oprofile also in
perf. But this cannot be done over night. Also, I like the idea having
a single pmu implementation only.
There are still oprofile users who are using oprofile, they are
sending also updates, bug reports and fixes. I don't see reasons
against this, oprofile is already in "some sort" of maintenance mode.
That said, I don't see an alternative to the current direction we are
going. We're slowly getting there but we cannot ask oprofile users to
switch to perf in a second.
Btw, your statement is actually very productive as it moves people to
start using and supporting perf sooner.
Advanced Micro Devices, Inc.
Operating System Research Center