Re: Fix some x86 peepholes vs. the red-zone

2016-04-15 Thread Bernd Schmidt
On 04/15/2016 10:32 AM, Uros Bizjak wrote: This fixes possible wrong code with a tricky failure mode, so OK now. Thanks. [...] While changing this part, it can be rewritten using dg-additional options, e.g.: /* { dg-options "-Os -fomit-frame-pointer -fasynchronous-unwind-tables

Re: Fix some x86 peepholes vs. the red-zone

2016-04-14 Thread Jeff Law
On 04/12/2016 05:57 AM, Bernd Schmidt wrote: With some changes I was working on, I produced a situation where one of the x86 peepholes made an incorrect transformation. In the prologue expansion code, we have /* When using red zone we may start register saving before allocating the stack

Fix some x86 peepholes vs. the red-zone

2016-04-12 Thread Bernd Schmidt
With some changes I was working on, I produced a situation where one of the x86 peepholes made an incorrect transformation. In the prologue expansion code, we have /* When using red zone we may start register saving before allocating the stack frame saving one cycle of the prologue.