malloc: performance improvements and bugfixes

[Torvald Riegel]

On Tue, 2016-01-26 at 09:14 -0800, Jörn Engel wrote:
> On Tue, Jan 26, 2016 at 01:22:41PM +0100, Torvald Riegel wrote:
> > 
> > Could you describe the workloads you used to select / validate these
> > changes?  While several of the items you listed might be generally
> > useful, others are certainly workload-specific, at least as far as
> > choices regarding trade-offs are concerned.
> 
> Think a single heavily multithreaded process consuming most of the
> system, then make it latency-sensitive and you are pretty much there.
> 
> I think the hugepage stuff is pretty unique to us, but most of the rest
> should apply to anything big and performance-sensitive.

How do the allocation patterns look like?  There can be big variations
in allocation frequency and size, lifetime of allocated regions,
relation between allocations and locality, etc.  Some programs allocate
most up-front, others have lots of alloc/dealloc during the lifetime of
the program.