The sfrander pointer analysis gives an incorrect (NoAlias) result, which generates a smaller SCC and gets distributed through the LoopDistribution pass. This, in the end, results in a segfault of the test LoopDistribution5. This change disables —sfrander so that all LoopDistribution5 regression tests can pass.
This bug has also been reported to SVF here for help.
The
sfranderpointer analysis gives an incorrect (NoAlias) result, which generates a smaller SCC and gets distributed through theLoopDistributionpass. This, in the end, results in a segfault of the testLoopDistribution5. This change disables—sfranderso that allLoopDistribution5regression tests can pass.This bug has also been reported to SVF here for help.