Blue Swirl writes: > On Wed, Nov 10, 2010 at 5:59 PM, Lluís <xscr...@gmx.net> wrote: >> So, my patch is just a matter of having all events available _only_ when >> you use a backend other than "nop".
> Then the default for simpletrace (and dtrace?) should also be disabled > initial state. If you have 1000 tracepoints automatically enabled at > start (with simpletrace, not nop), disabling all of them if you are > only interested in a few of them is difficult. That's how it is. The patch sets always ".state=0" in "simple", and "dtrace" just does not do anything unless you plug something into the probe (similarly with "ust"). The "problem" here is that when you use a backend other than "nop", all events are available, but in an inactive state. So the burden is for the person compiling to edit trace-events and disable the unwanted events. I can leave all events with "disable" keyword set, but then the problem is the other way around, where the person compiling will have to manually select the desired events :) That's the only thing I see, but there's no silver bullet here, and is still there with the current implementation. So my only objection is that "ust" with "disable" is able to wipe the event out, while in "simple" and "dtrace" there is no way to wipe any event out, plus "simple" has "early tracing" with "disable" and "dtrace" just ignores "disable"... that's what I tried to address with the patch. Summarizing, now "disable" wipes events out, and there is no way to set "early tracing" from inside trace-events. Lluis -- "And it's much the same thing with knowledge, for whenever you learn something new, the whole world becomes that much richer." -- The Princess of Pure Reason, as told by Norton Juster in The Phantom Tollbooth
