[PATCH] D49560: Add support to track total counts for DebugCounter

[George Burgess IV via Phabricator via llvm-commits]

george.burgess.iv added a comment.

> As a result, I also reverted the changes to make CounterDesc part of CounterInfo.

Sorry, I'm confused. Is the goal here to ultimately print all of the counts? I don't see how that can happen if we're only `Count`ing for things which have `skip` or `stopAfter` specified on the `opt` line.

> this will save many memory when skip/count is not set, thus fixing the problem

I'm assuming you meant that the memory savings is a nice side-effect, and not the cause of the problem (which appears to be that we're now always saying that these counters are set). If not, +~300 bytes of heap crashing a test is scary. :)


Repository:
  rL LLVM

https://reviews.llvm.org/D49560