This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [gcov] Merge arc counts
- From: Jan Hubicka <jh at suse dot cz>
- To: Nathan Sidwell <nathan at codesourcery dot com>
- Cc: gcc-patches at gcc dot gnu dot org, mark at codesourcery dot com, jh at suse dot cz
- Date: Sat, 10 Aug 2002 03:44:50 +0200
- Subject: Re: [gcov] Merge arc counts
- References: <3D5442F8.3D635B90@codesourcery.com>
> Hi,
> this patch changes __bb_exit_func to merge arc counts into the .da files,
> rather than appending a new record for each execution. This has the
> advantage of not eating your disk space. I also tidied up some unused
> globals and header files which appear left over from -a coverage days.
>
> Merging like this does affect the value of the maximum count
> across the whole program. The value in the .da file is now that
> seen after accumulating all the data. Previously the value used
> by predict.c was the sum of the maxima on each program run.
>
> booted & tested on i686-pc-linux-gnu, ok?
It looks OK for me, but I can not approve it.
The purpose for appending records is that the data can be analysed in
more sophisphicated ways (there are several methods of merging the
counts in more sensitive way dicussed in the literature or it may be
usefull to have the data on how much the program behaviour changes from
one run to annother. At the moment the such data are unused and in
number of applications they can be even in future, so it definitly makes
sense to have the merging at least as an option.
Please try to keep the "old" way functional so we can experiment with it
later easilly.
Honza
>
> nathan
> --
> Dr Nathan Sidwell :: http://www.codesourcery.com :: CodeSourcery LLC
> 'But that's a lie.' - 'Yes it is. What's your point?'