Re: sv_grow() and malloc()

On 23 September 2014 13:49, Dave Mitchell <davem@iabyn.com> wrote:

> (this is spin-off from Reini's suggested work to tweak sv_grow()
> allocations).
>
> I've been thinking about about all our heuristics for growing (and
> over-growing) strings in sv_grow(), and how it's likely to interact with
> the OS's malloc() library.
>

At least one OS (sun maybe?) supports an introspection
routine Perl_safesysmalloc_size which exposes the canonical form of this
information.

Use of this routine in sv.c caused problems with the new COW logic and the
readline stuff so I disabled it. It can be re-enabled by defining:

PERL_UNWARANTED_CHUMMINESS_WITH_MALLOC

I do not recommend doing this, it confuses the COW logic quite a bit.

Yves

-- 
perl -Mre=debug -e "/just|another|perl|hacker/"

• sv_grow() and malloc() by Dave Mitchell
• Re: sv_grow() and malloc() by demerphq
• Re: sv_grow() and malloc() by Dave Mitchell
• Re: sv_grow() and malloc() by Father Chrysostomos
• Re: sv_grow() and malloc() by Aaron Crane
• Re: sv_grow() and malloc() by H.Merijn Brand
• Re: sv_grow() and malloc() by Tony Cook
• Re: sv_grow() and malloc() by Dave Mitchell
• Re: sv_grow() and malloc() by Steve Hay
• Re: sv_grow() and malloc() by Andy Dougherty
• Re: sv_grow() and malloc() by Craig A. Berry
• Re: sv_grow() and malloc() by Andy Dougherty
• Re: sv_grow() and malloc() by Aaron Crane
• Re: sv_grow() and malloc() by Father Chrysostomos
• Re: sv_grow() and malloc() by Dave Mitchell