[PATCH] [BasicAA] Fix incorrect scale & offset initializer

[hfinkel at anl.gov]

It would be much better to have a test case that directly tests the functionality in question, and makes sure the answer is right (instead of just ensuring that it does not crash). Could you extract from your test case a test that you run with -aa-eval? (It seems like this should be relatively easy; if not, please say so.)


http://reviews.llvm.org/D9813

EMAIL PREFERENCES
  http://reviews.llvm.org/settings/panel/emailpreferences/