Hi Priyanka,

I think this is the same question you asked on the list several days ago. Erik Brangs responded with advice about what to try and also about how to get help from the mailing list. For example, Erik mentioned that the NPE is likely just a side effect of the way that Jikes treats segfaults, i.e., the actual error is a segfault. He (and everyone else) weren't able to tell you the exact problem since it's hard to know what exactly the issue with your changes might be (you haven't provided any details about your changes), but his advice should be generally useful for helping you debug the problem. Could you please ask questions that follow up on feedback that's already been provided?

Thanks,
Mike

On 03/31/2014 03:46 AM, Priyanka Bhayana wrote:
Hi,
Thanks for your prompt response. I want to have the trace of the java application which is being executed by the jikes rvm, by the trace I mean all the x86 instructions alongwith the program counter, that will be executed when the application is running. I have tried to instrument the emit_opcodes methods in  Assembler.java by modifying Assembler.in and genAssembler.sh, but I am getting Null Pointer Exceptions when I am instrumenting some addressing modes. The addressing modes which are causing problems are Register- Register Index and Immediate,  So I want to ask is there any special about this addressing modes which I might be missing? And is there any other way to get the x86 instruction and program counter?


On Sun, Mar 30, 2014 at 7:59 PM, Michael Bond <mikebond@cse.ohio-state.edu> wrote:
Hi Priyanka,

The baseline compiler already adds instrumentation to count taken vs. not-taken branch frequency. Take a look at the EdgeCounts class for a starting point. The branch frequencies get passed to the opt compiler (I think BranchProfiles is the best place to look?), which also seems to be something you want to do.

I hope that helps as a starting point. For follow-up questions, it would help if you could provide more details about what you've already investigated and what exactly you're trying to do, e.g., should the trace of basic blocks be a whole-program path, or do you want to aggregate profile information bounded-length paths -- and if so, should they be acycle/intraprocedural or something else?

Cheers,
Mike


On 03/28/2014 12:48 PM, Priyanka Bhayana wrote:
Hi,
I am using jikes rvm as research infrastructure. I want to instrumentation so as to get the basic blocks frequency as well as the trace of basic blocks so that the instructions in the block can be rescheduled. I want these trace for baseline compiler.Is there any file in jikes source which can be used for this purpose? Can this thing done with jikesrvm?

--
Thanks&Regards
Priyanka


------------------------------------------------------------------------------


_______________________________________________
Jikesrvm-researchers mailing list
Jikesrvm-researchers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/jikesrvm-researchers


------------------------------------------------------------------------------

_______________________________________________
Jikesrvm-researchers mailing list
Jikesrvm-researchers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/jikesrvm-researchers




--
Thanks&Regards
Priyanka


------------------------------------------------------------------------------


_______________________________________________
Jikesrvm-researchers mailing list
Jikesrvm-researchers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/jikesrvm-researchers