Re: Fix VR_ANTI_RANGE handling in intersect_range_with_nonzero_bits (PR 84321)

2018-02-13 Thread Jeff Law
On 02/13/2018 08:26 AM, Richard Biener wrote: > On Mon, Feb 12, 2018 at 4:29 PM, Richard Sandiford > wrote: >> VR_ANTI_RANGE is basically a union of two ranges, and although >> intersect_range_with_nonzero_bits had code to deal with the upper >> one being empty, it

Re: Fix VR_ANTI_RANGE handling in intersect_range_with_nonzero_bits (PR 84321)

2018-02-13 Thread Richard Biener
On Mon, Feb 12, 2018 at 4:29 PM, Richard Sandiford wrote: > VR_ANTI_RANGE is basically a union of two ranges, and although > intersect_range_with_nonzero_bits had code to deal with the upper > one being empty, it didn't handle the lower one being empty. > There were

Fix VR_ANTI_RANGE handling in intersect_range_with_nonzero_bits (PR 84321)

2018-02-12 Thread Richard Sandiford
VR_ANTI_RANGE is basically a union of two ranges, and although intersect_range_with_nonzero_bits had code to deal with the upper one being empty, it didn't handle the lower one being empty. There were also some off-by-one errors. This patch rewrites the code in a hopefully clearer way. Tested on