[llvm-dev] XRay threshold (bug?)
Dean Michael Berris via llvm-dev
llvm-dev at lists.llvm.org
Thu Jun 8 04:41:02 PDT 2017
Yes I like option 1 as well, if size() isn't actually the machine instruction count. My only concern might be the cost of counting the instructions, but if it's only when XRay instrumentation is on then this should be fine.
Let me have a look at the patch.
Sent from my iPhone
> On 8 Jun 2017, at 9:26 pm, Serge Rogatch via llvm-dev <llvm-dev at lists.llvm.org> wrote:
> Hi Dean,
> I've noticed that XRay in llvm\lib\CodeGen\XRayInstrumentation.cpp compares its threshold against Function::size() . However, Function::size() returns the number of basic blocks (as I understand, such as cycle bodies, if/else bodies, switch-case bodies, etc.), rather than the number of instructions.
> Was your intent to count the instructions instead? The name of the parameter -fxray-instruction-threshold=N suggests this, as well as XRay documentation at http://llvm.org/docs/XRay.html .
> If so, I see two options:
> 1. Count the number of MachineInstr`s in MachineFunction : this gives better estimate for the number of assembly instructions on the target. So a user can check in disassembly that the threshold works more or less correctly.
> 2. Count the number of Instruction`s in a Function : AFAIK, this gives correct number of IR instructions, which the user can check in IR listing. However, this number may be far (several times for small functions) from the number of assembly instructions finally emitted.
> My team is in favor of option 1 (proposed patch here https://reviews.llvm.org/D34027 ), because we think that having the closer estimate for the number of assembly instructions emitted is more important than to have a clear definition of the metric.
> What do you think?
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the llvm-dev