gprof/sprof and shared libraries...again

[Braman, Paul]

...and then our RedHat contractor upgraded our version of glibc and
everything was nice and peachy like I was expecting.

At least I learned a lot from the experience.

Thanks for listening.  :)


Paul Braman
Paul_Braman@tvratings.com


> -----Original Message-----
> From: Braman, Paul 
> Sent: Thursday, April 04, 2002 4:15 PM
> To: 'binutils@sources.redhat.com'
> Subject: RE: gprof/sprof and shared libraries...again
> 
> 
> 
> Okay, I did get a little further with the hints from Andrew Cagney.
> 
> To profile the calls to my library I made sure I compiled it with -pg.
> 
>     gcc -pg -fPIC -I. -c -o prof.lo prof.c
>     gcc --shared -o libprof.so prof.lo
> 
> Now, I added the a few options to my call to sprof and seem 
> to be getting slightly closer...
> 
>     sprof --flat-profile --graph libprof.so libprof.so.profile
> 
> This returns a flat profile that contains the symbol I am 
> looking for, but the call graph still won't show it like it 
> would with profiling libc.so.6.
> 
>     Flat profile:
> 
>     Each sample counts as 0.01 seconds.
>       %   cumulative   self              self     total
>      time   seconds   seconds    calls  us/call  us/call  name
>     100.00      0.50     0.50        0     0.00           something
> 
>     index % time    self  children    called     name
> 
>                     0.00    0.00        0/0           <UNKNOWN>
>     [0]      0.0    0.00    0.00        0         
> mcount@@GLIBC_2.0 [0]
>                     0.00    0.00        0/0           
> mcount@@GLIBC_2.0 [0]
> 
> Yes, "something" is the symbol I'm looking for, and I get 
> timing on the call, but it doesn't show up in the call graph. 
>  (Note that this much only works when I *do not* compile my 
> program with profiling.  If I do compile my executable with 
> -pg, my symbol won't even show up in the flat profile above.)
> 
> So, there's something...not sure what, but something.
> 
> 
> Paul Braman
> x5273
>