This is the mail archive of the
mailing list for the glibc project.
Re: malloc cache/trace preview
- From: Florian Weimer <fweimer at redhat dot com>
- To: Szabolcs Nagy <szabolcs dot nagy at arm dot com>
- Cc: DJ Delorie <dj at redhat dot com>, libc-alpha at sourceware dot org, nd at arm dot com
- Date: Wed, 10 Feb 2016 14:27:26 +0100
- Subject: Re: malloc cache/trace preview
- Authentication-results: sourceware.org; auth=none
- References: <201602092251 dot u19MpO5s027665 at greed dot delorie dot com> <56BB2D1B dot 8000804 at arm dot com> <56BB31B7 dot 30203 at redhat dot com> <56BB3A9E dot 3010801 at arm dot com>
On 02/10/2016 02:26 PM, Szabolcs Nagy wrote:
> On 10/02/16 12:48, Florian Weimer wrote:
>> On 02/10/2016 01:29 PM, Szabolcs Nagy wrote:
>>> i think the sbrk in the ctor of the trace tool is not ok,
>>> if there are malloc calls using brk before that ctor.
>>> (makes the malloc heap non-contiguous.)
>> malloc has code to compensate for that. Does it work?
> without looking into it, my guess is that the 'compensation'
> has significant effect on malloc behaviour.
Ah, that is true. Good point.