You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
After fault injection, there are some cases where not all the runs have trace files generated.
For some of the runs, the program may crash before outputting trace files. And whether the trace files is generated or not is random. We may not be able to reproduce the same result every time we run fault injection.
The text was updated successfully, but these errors were encountered:
Can you please provide an example program where this happens along with the input yaml files ? I'm not able to reproduce this issue on my Mac system. Thanks,
I replicated the bug with the test program factorial.c.
The steps below is how I reproduced the issue:
In instrument, I selected ret-(ReturnInst) and full trace with backward checked only. Note, if you select both backward and forward, it is very unlikely to reproduce the bug.
In profiling, I gave 10 as input. Then, in Runtimeoptions, I chose bitflip as the fault type.
After fault injection, the trace files are missing.
The contents below is the input.yaml file generated through the instrument step above:
kernelOption:
- forceRun
compileOption:
instSelMethod:
- insttype:
include:
- ret
After fault injection, there are some cases where not all the runs have trace files generated.
For some of the runs, the program may crash before outputting trace files. And whether the trace files is generated or not is random. We may not be able to reproduce the same result every time we run fault injection.
The text was updated successfully, but these errors were encountered: