This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: Poor scaling
Josh Stone <jistone@redhat.com> writes:
> [...] I think I'd rather see this explicit declared on the global
> itself, instead of treating tid() specially. Something like the
> proposed percpu globals:
> https://sourceware.org/bugzilla/show_bug.cgi?id=10795
Or even more like https://sourceware.org/bugzilla/show_bug.cgi?id=10796
> The challenge is how to allocate and free that per-task memory, and how
> to look up that memory from arbitrary probe context. It's possible,
> just not entirely trivial.
... and to make it more efficient than the current code! One can
imagine hooking into tracepoints to track task lifetimes, and create
some tid-indexed rcu hash tables or somesuch. Kind of like the
kretprobes data-pouches. Possible, just a SMOP. We should probably
reopen PR10796 and give it a real try.
- FChE