This is the mail archive of the
mailing list for the Archer project.
Re: ptrace improvement: PTRACE_O_INHERIT
> I meant, I do not agree that it never makes sense to trace, say, a
> single thread from the thread group. But since we are talking about
> gdb my noncompliance is off-topic.
We are also talking about what makes sense for a kernel interface, so it's
certainly apropos to share our thoughts on the general subject. Please
elaborate on why you think this makes sense. It's certainly true that it
makes sense to have your tracing details different for one thread than for
others (stopping on more events, etc.) but that's quite a different thing
from saying it really makes sense to have a debugging interface structured
to consider tracing an individual thread as unrelated to tracing all the
threads in the same process, as Linux's ptrace is today. I can't really
imagine what you have in mind, so please tell me.