[PATCH v3] malloc: Improve performance of __libc_malloc

[DJ Delorie]

Wilco Dijkstra <Wilco.Dijkstra@arm.com> writes:
>> It would be messier but faster if, when USE_TCACHE isn't defined, we
>> avoid doing the non-inlined tail call.
>
> Is it useful to keep the USE_TCACHE defines? As it is no longer experimental,
> it would be better to remove them.

I'm not opposed to it, but it's an easy way to compare with and without
tcache, in case "disabled but present" has some measurable overhead.
It's also a convenient thing to search for when looking for all
tcache-related code.

These are weak reasons to keep it, though.

It would be interesting to revisit my original with/without benchmarks,
which includes with/without fastbins, given the modern code in tcache.