[ixp1200] Execution Time Measurement in IXP1200 via gettimeofday

[Jo Ueyama]

Hi All,
    I am trying to measure the execution time (via C/C++ gettimeofday) 
of applications running on the StrongARM but I am not being able to 
measure it properly. Basically, in 10/20% of the executions,  
gettimeofday is outputing time (in miliseconds) that is out of order 
(i.e. sometimes it outputs time lower than the previous one resulting in 
a negative time difference). Let me mention that the same code runs 
nicely on the PC.
    Has anyone ever come across such situation? How do you all normally 
measure execution time in the IXP1200 board?
    Thanks a lot in anticipation.

Regards,
-Jo'