Soft freeze for the glibc-2.43 release

[Florian Weimer]

* Wilco Dijkstra:

> Hi Florian,
>
>>>> I plan to post my batched free changes.  Do they reach parity with 32?
>>>> Average tcache consumption should be less because we are not filling
>>>> it as aggressively.
>>>
>>> Yes with tcache_count=32 the geomean of omnetpp and xalancbmk is within 0.1%.
>>
>> Sorry, and this was with batched free?
>
> Yes that is batched free vs non-batched at tcache_count=32. Both
> options are also practically the same with mxfast=0. Basically it's
> only bad for performance if tcache size is small since it makes the
> effective size even smaller.

Thanks.  I posted my patches for batched frees:

  [PATCH 0/3] malloc: batched frees
  <https://inbox.sourceware.org/libc-alpha/cover.1766051058.git.fweimer@redhat.com/T/>

I added a Tested-by: in your name, hope this is okay.

Thanks,
Florian