Hello all,
I'm trying to measure the cost in execution time of address translation
in an ARM o3 FS system.
I replaced the translateTiming call in lsq.cc with a translateFunctional
call. My assumption is that the
translateFunctional has the same functionality as translateTiming but it
excludes the calculation of delays.
After this change, it seems that the simulation does not progress and
ends up to an infinite loop in doSimLoop().
I observe the same issue when I replace the timing call with the
translateAtomic (with simple caches instead of ruby).
Is this behavior expected?
Any ideas or other workarounds on this issue would be appreciated.
I attach a debug trace file and a stack trace from gdb.
Thank you,
Sotiris
Hello all,
I'm trying to measure the cost in execution time of address translation
in an ARM o3 FS system.
I replaced the translateTiming call in lsq.cc with a translateFunctional
call. My assumption is that the
translateFunctional has the same functionality as translateTiming but it
excludes the calculation of delays.
After this change, it seems that the simulation does not progress and
ends up to an infinite loop in doSimLoop().
I observe the same issue when I replace the timing call with the
translateAtomic (with simple caches instead of ruby).
Is this behavior expected?
Any ideas or other workarounds on this issue would be appreciated.
I attach a debug trace file and a stack trace from gdb.
Thank you,
Sotiris