Btw, here is another interesting paper about post-dominators and control
dependence:
I think a great outcome of your internship would be some precise
documentation regarding the guarantees the LLVM dominators give --
possibly also considering classic and weak control dependence and the
difference between loop-dominance and dominance.
Please keep me up-to-date with your work. This is really work that has
long been overdue!
It seems the names “inloop” and “next” are not used.
My bad, I almost always use the tool with -view-cfg and didn’t notice that BB names were different. I fixed that and now when you just run the dominators tool on a module it should work fine. It now also supports reading .ll files.
Interesting. Probably it is just a testing/debugging tool, but maybe
clarifying this in the output may avoid confusion for others who try
your changes.
Yep, I briefly mentioned that in the first line of the tool’s source file, but I should probably add a note in some more visible place. Sorry for making it confusing.
We could even have a tool that records the dom-tree modification
requests, and dumps a corresponding IR-file.
Obviously, this is just an idea.
I’ll think about it.
Btw, here is another interesting paper about post-dominators and control
dependence:
>
> I tried to use a bitcode file directly:
>
> ...
>
> It seems the names "inloop" and "next" are not used.
My bad, I almost always use the tool with -view-cfg and didn't notice
that
BB names were different. I fixed that and now when you just run the
dominators tool on a module it should work fine. It now also supports
reading .ll files.
Interesting. Probably it is just a testing/debugging tool, but maybe
> clarifying this in the output may avoid confusion for others who try
> your changes.
Yep, I briefly mentioned that in the first line of the tool's source
file,
but I should probably add a note in some more visible place. Sorry for
making it confusing.
Great. It now works as expected. I still don't have an option to compute
post-dominators (I assume it is not yet implemented), but otherwise the
tools now does what I expect. Thanks for improving it!