[PATCH] D30691: [analyzer] Support for naive cross translational unit analysis
Daniel Krupp via Phabricator via cfe-commits
cfe-commits at lists.llvm.org
Mon Jun 19 10:26:32 PDT 2017
dkrupp added a comment.
> It would be best to just add the scan-build-py support to the tree, especially, since the new scrips are not tested.
OK. We will update this patch with the scan-build-py changes and remove the ctu-build.py and ctu-analyze.py scripts.
> I am curious which optimization strategies you considered. An idea that @NoQ came up with was to serialize only the most used translation units. Another idea is to choose the TUs that a particular file has most dependencies on and only inline functions from those.
Both of these strategies could work in practice, we did not try them. We implemented the two extremes: serialize all TUs/don't serialize any of the TUs. Both of them could can be useful in practice as is (depending if the user is cpu/memory/disk space bound). I think we could try with the above suggested optimizations as an incremental improvement (and keep this initial patch as simple as possible).
> What mode would you prefer? Would you pay the 20%-30% in speed but reduce the huge disk consumption? That might be a good option, especially, if you have not exhausted the ideas on how to optimize.
In this initial version I think we should keep the serializing mode. We just measured that the heap consumption of the non-serializing mode and it seems to be ~50% larger. Probably because the serializing mode only loads those AST fragments from the disk that is imported. But I can imagine that some user still want to use the non-serializing version which is not using extra disk space. So we will add the non-serializing mode in a next patch as an Analyzer option first (which we can turn into default behaviour later on). OK?
> I see some changes to the compiler proper, such as ASTImporter, ASTContext, SourceManager. We should split those changes into a separate patch and ask for review from people working on those components. You can point back to this patch, which would contain the analyzer related changes, and state that the other patch is blocking this work.
Allright, we will do that.
So is it OK to proceed like this?
More information about the cfe-commits