On Thu, 5 Apr 2018, Jakub Jelinek wrote:
> Hi!
>
> As mentioned in the PR, on the following testcase we cp_save_expr in order
> to avoid evaluating -1 * __builtin_expect (({ ... }), ...) twice and use
> the SAVE_EXPR in the original shift as well as in a comparison, but then
> we attempt to optim
Hi!
As mentioned in the PR, on the following testcase we cp_save_expr in order
to avoid evaluating -1 * __builtin_expect (({ ... }), ...) twice and use
the SAVE_EXPR in the original shift as well as in a comparison, but then
we attempt to optimize the comparison by grabbing parts of the SAVE_EXPR,
