Outdated jemalloc in CURRENT

[Li-Wen Hsu]

On Sat, Apr 18, 2020 at 11:22 PM Steve Kargl
<sgk at troutmask.apl.washington.edu> wrote:
>
> On Sat, Apr 18, 2020 at 05:01:08PM +0200, Gordon Bergling wrote:
> >
> > On Sat, Apr 18, 2020 at 07:20:03AM -0700, Steve Kargl wrote:
> >> On Sat, Apr 18, 2020 at 03:05:25PM +0300, nonameless at ukr.net wrote:
> >>> Hi everyone!
> >>>
> >>> As I see, CURRENT still uses outdated jemalloc 5.1.0 with some
> >>> performance regressions that was fixed in 5.2.1.
> >>>
> >>> Are there some issues that blocking update jemalloc to recent version?
> >>>
> >>
> >> ------------------------------------------------------------------------
> >> r354606 | jasone | 2019-11-10 21:06:49 -0800 (Sun, 10 Nov 2019) | 4 lines
> >>
> >> Revert r354605: Update jemalloc to version 5.2.1.
> >>
> >> Compilation fails for non-llvm-based platforms.
> >>
> >> ------------------------------------------------------------------------
> >> r354605 | jasone | 2019-11-10 19:27:14 -0800 (Sun, 10 Nov 2019) | 2 lines
> >>
> >> Update jemalloc to version 5.2.1.
> >>
> >
> > I am not sure, that this info is correct. As far as I remember the
> > update for jemalloc was reverted due to build problems, on some
> > architecture. An updated revision has still to be commited to -CURRENT.
> >
>
> Those two commits confirm your memory.  5.2.1 was committed
> in r354605.  5.2.1 was reverted with r354606 where the reason
> for reverting is stated.  As 5.2.1 has not been re-committed
> and there is nothing in reviews.freebsd.org for review, one
> may expect the reason in r354606 still stands.
>
> PS: Please, do not top-post.
> PPS: Please, wrap your messages to something less than 80 characters.

I think it's fine to (and hope we can) update it again as the previous
error was on outdated gcc which is not existing anymore:

https://lists.freebsd.org/pipermail/svn-src-all/2020-March/195013.html

Li-Wen