Re: [Valgrind-users] valgrind and timers..

[John R. <jreiser@BitWagon.com>]

> ... my problem is that since valgrind makes things much slower my
> application is not behaving the same under valgrind than under the real
> operating system...

This shows that the application is not robust enough to run under
heavy load.  Valgrind gave you a bonus by enlarging your test coverage.
If the application is that sensitive to maximum response time, then
it ought to detect poor conditions, and either adjust or fail explicitly.

> I wonder if valgrind has some solution to this.. for it is a very
> complex task to readapt the timers in the application itself.. 

All serious timer-driven code multiplies each timer delay by the
value of an environment variable such as TIMER_DELAY_EXPANSION_FACTOR.
This can even be done using LD_PRELOAD of a shared library that
contains a subroutine which re-defines the low-level timer call.

--