Profiling floating point emulation (fp-bit)

Profiling floating point emulation (fp-bit)

[Peter Barada]

I'm building for a target that doesn't have hardware FP support, so it
has to include fp-bit.c to get its support.

The application I'm building does a lot of floating point, so I'm
trying to profile the floating point emulation which is not built with
profiling.

Anybody have any ideas on how to get a libgcc1_p.a that has profiling
turned on (and properly installed)?

-- 
Peter Barada                            pbarada@wavemark.com
Wizard                                  781-270-7098 x226
WaveMark Technologies, Inc.             781-270-0193 (fax)

"Real men know that you should never attempt to accomplish with words
what you can do with a flame thrower" --Bruce Ferstein

Re: Profiling floating point emulation (fp-bit)

[Jim Wilson]

An easy solution is to just install two copies of EGCS, one of which
had libgcc.a compiled with -pg (or -p).  You should be able to just add the
appropriate option to LIBGCC2_CFLAGS to make this happen.

If you want a more elegant solution, then you could handle -pg as multilib
options, which would cause every library to be build 2 times as often as it
does now, with half of them being build with profiling and the other half
without.  Then gcc should automatically use the profiled libgcc.a when you
use -pg.

Jim

Re: Profiling floating point emulation (fp-bit)

[Jeffrey A Law]

  In message <199803140229.VAA08783@miacomet.wavemark.com>you write:
  > 
  > I'm building for a target that doesn't have hardware FP support, so it
  > has to include fp-bit.c to get its support.
  > 
  > The application I'm building does a lot of floating point, so I'm
  > trying to profile the floating point emulation which is not built with
  > profiling.
  > 
  > Anybody have any ideas on how to get a libgcc1_p.a that has profiling
  > turned on (and properly installed)?
You could add -p/-pg to LIBGCC2_CFLAGS in your Makefile.

jeff