[llvm-dev] [ThinLTO] Suggestions on how to traverse SCCs in the Module Summary Index bottom up

[Charles Saternos via llvm-dev]

I just realized that I've been misunderstanding this (I think). If I'm
understanding it now, the entry node for the graph should actually be the
graph's root (not a bottom node), and that the scc_iterator works its way
down. So for the ModuleSummary, I should find the main function and return
that (or whatever function is the entry point for the entire program).

If anyone has suggestions for the patch, I'd still be interested in hearing
them.

- Charles

On Wed, Aug 9, 2017 at 10:21 AM, Charles Saternos <
charles.saternos at gmail.com> wrote:

> Hey all,
>
> I'm working on adding function attribute propagation to function summaries
> in the ThinLTO index, and have run into some trouble with ensuring
> bottom-up traversal when finding the SCCs in the call graph.
>
> I'm basing my implementation for the GraphTraits for the
> ModuleSummaryIndex off the GraphTraits<CallGraph *> implementation (
> http://llvm-cs.pcc.me.uk/include/llvm/Analysis/CallGraph.h#407). In the
> GraphTrait<CallGraph *> definition, the getEntryNode function returns the
> external calling node (http://llvm-cs.pcc.me.uk/include/llvm/Analysis/
> CallGraph.h#450), which I assume is supposed to be at the bottom of the
> callgraph. Would doing same for the ModuleSummaryIndex ensure the
> scc_iterator traverses bottom up (if that even makes sense)?
>
> Rather than returning a dummy FunctionSummary that's empty for external
> functions (which is what I'm doing right now), would it make more sense to
> have one node that represents all external FunctionSummaries? If I used
> this FunctionSummary as the entry point for the scc_iterator traversal of
> the ModuleSummaryIndex callgraph would the traversal be bottom up?
>
> Here's my patch in its current state: https://reviews.llvm.org/D36311
>
> Thanks,
> Charles
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170809/fe64a0f6/attachment.html>