Re: Gprof output in 2.21.1 different than in 2.21.53

[Tamer <nuclearpsyche at gmail dot com>]

Hi Alan,

Thanks for your reply.

What seems to have changes is the <cycle> tags. It also seems to use <cycle 1 as a whole> instead of "main".
Here is an example of before and after:

Old Version:
-----------
index % time    self  children    called     name
                                                 <spontaneous>
[1]     97.5    0.00    3.11                 main [1]
....

[2]     74.4    0.00    2.37   16384         process_one_packet [2]

New Version:
-----------
index % time    self  children    called     name
[1]     77.8    0.49    0.00  115778+34409   <cycle 1 as a whole> [1]
....

[2]     77.8    0.49    0.00  115702+1641    pkt_post_process <cycle 1> [2]

Thank you for your help.

Regards,
Tamer

On Oct 8, 2011, at 8:37 PM, Alan Modra wrote:

> On Tue, Sep 27, 2011 at 10:18:07AM -0700, Tamer wrote:
>> Hi,
>> 
>> I wrote a parser a while ago that reads gprof output and does some interpretation of the output. 
>> Today I re-installed MinGW and chose to use the latest repository packages (instead of using pre-packaged repository catalogues). Now the output of gprof is different and my parser no longer works.
>> 
>> So things were working fine on gprof 2.21.1 but then stopped working on 2.21.53.20110804
>> 
>> Is this expected? Is there a place where those changes in gprof's output are documented? I would like to know if I have to rewrite my parser or if I just happen to use a dev build which has changes that are not going GA.
>> 
> 
> You may get an answer if you tell us what changed.
> 
> -- 
> Alan Modra
> Australia Development Lab, IBM