Re: [PATCH] Fix ICE with FRE devirtualization (PR middle-end/77259)

2016-08-17 Thread Richard Biener
On Wed, 17 Aug 2016, Jakub Jelinek wrote: > On Wed, Aug 17, 2016 at 10:03:28AM +0200, Richard Biener wrote: > > On Tue, 16 Aug 2016, Jakub Jelinek wrote: > > > > > Hi! > > > > > > The FRE devirtualization unlike gimple-fold or other places would > > > transform > > > some method call with TREE_

Re: [PATCH] Fix ICE with FRE devirtualization (PR middle-end/77259)

2016-08-17 Thread Jakub Jelinek
On Wed, Aug 17, 2016 at 10:03:28AM +0200, Richard Biener wrote: > On Tue, 16 Aug 2016, Jakub Jelinek wrote: > > > Hi! > > > > The FRE devirtualization unlike gimple-fold or other places would transform > > some method call with TREE_ADDRESSABLE lhs into __builtin_unreachable call > > with the sam

Re: [PATCH] Fix ICE with FRE devirtualization (PR middle-end/77259)

2016-08-17 Thread Richard Biener
On Tue, 16 Aug 2016, Jakub Jelinek wrote: > Hi! > > The FRE devirtualization unlike gimple-fold or other places would transform > some method call with TREE_ADDRESSABLE lhs into __builtin_unreachable call > with the same lhs, which is invalid (__builtin_unreachable returns void). > Also, gimple_c

[PATCH] Fix ICE with FRE devirtualization (PR middle-end/77259)

2016-08-16 Thread Jakub Jelinek
Hi! The FRE devirtualization unlike gimple-fold or other places would transform some method call with TREE_ADDRESSABLE lhs into __builtin_unreachable call with the same lhs, which is invalid (__builtin_unreachable returns void). Also, gimple_call_fntype has not been adjusted in these cases. Fixed