Re: [PATCH 1/2] tools, perf: Add a precise event qualifier v2

[Vince Weaver]

I hate having to justify why breaking the ABI is unacceptable.


On Mon, 22 Jul 2013, Andi Kleen wrote:

> On Tue, Jul 23, 2013 at 01:40:40AM -0400, Vince Weaver wrote:
> > On Mon, 22 Jul 2013, Andi Kleen wrote:
> > 
> > > From: Andi Kleen <ak@xxxxxxxxxxxxxxx>
> > > 
> > > Add a precise qualifier, like cpu/event=0x3c,precise=1/
> > 
> > So you're adding this to "events/" but not to "format/"?
> 
> Fair point, but note that precise_ip is a bitfield. Since there is no
> way to express that currently the parser would need to special case it anyways.

It breaks the ABI.  The events/* sysfs files are documented as only 
holding values for the bitfields described in format/*.

The perf_event ABI is there for all tools, not to just be a thin shim
for the perf tool.

Even if the maintainers for whatever are going to just ignore the ABI, 
then fine, but then I'd like to see a patch also updating the sysfs
ABI documentation and a patch to the perf_event_open.2 manpage would be 
nice as well.

> > This will break various existing programs.
> 
> Like what?

Why does it matter?  It breaks multiple programs I work on.

A major one is the trinity fuzzer, which parses perf_event sysfs to 
construct valid events.

The code won't segfault because I wrote the parser to be robust, but it 
will spit a warning when invalid junk is put in the sysfs events
files, and I'll get e-mails about it spamming trinity runs with warning 
messages within hours of such a commit hitting linus-git.

I'm a bit grumpy about this because I just finished fixing the fallout 
from your last time breaking the ABI a few weeks ago when your broken code 
started dumping non-hex fields into the sysfs event strings.  I've learned 
now that I have to go over your patches with a fine-tooth code because you 
have no regard for the ABI.

Vince


--
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/