This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [Patch] Remove use of sbrk
- From: Tristan Gingold <gingold at adacore dot com>
- To: binutils <binutils at sourceware dot org>
- Cc: Simon Marchi <simon dot marchi at ericsson dot com>
- Date: Mon, 3 Jul 2017 15:50:14 +0200
- Subject: Re: [Patch] Remove use of sbrk
- Authentication-results: sourceware.org; auth=none
- References: <aa8937f6-e578-9a94-0a36-d393de182ea8@ericsson.com>
> On 25 Jun 2017, at 13:48, Simon Marchi <simon.marchi@ericsson.com> wrote:
>
>> Hi,
>>
>> this patch removes the reports of data size based on sbrk.
>>
>> The rational is:
>> * I got warnings (and thus build failure unless --disable-werror) on Darwin, as sbrk(2) is
>> marked as deprecated. (I agree that this point is very minor).
>> * I seriously doubt about the accuracy of memory usage as reported by sbrk(2) as malloc doesn't
>> allocate all its memory with sbrk. I also doubt that this is often used by developers as
>> there are much more advanced tools to investigate performance issues.
>>
>> Any comments ? Ok for trunk ?
>>
>> Tristan.
>
> Hi Tristan,
>
> I was looking into this issue and found that you tackled this a while ago. You patch looks acceptable
> to me, and nobody opposed to it, so I was wondering what was the status. Did it simply fall through
> the cracks?
>
> On macOS, it doesn't even give results that make sense:
>
> ./nm-new: data size -140731891278888
I was not sure about a consensus on this point. Maybe worth asking directly:
Is someone against suppressing sbrk() calls ?
Tristan.