[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 --- Comment #21 from Bernd Edlinger --- Author: edlinger Date: Thu Aug 4 13:23:36 2016 New Revision: 239123 URL: https://gcc.gnu.org/viewcvs?rev=239123=gcc=rev Log: 016-08-04 Bernd Edlinger PR

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 --- Comment #20 from Andreas Schwab --- Works again.

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 --- Comment #19 from Bernd Edlinger --- Created attachment 39036 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=39036=edit combined patch Hi, I've updated the patch again based on Segher's comments. So that's what I'd like to post on

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 --- Comment #18 from Andreas Schwab --- With the patches in #16 and #17 bootstrap was successful and the isl testsuite passes.

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 --- Comment #17 from Bernd Edlinger --- And here an alternative patch for the other paradoxical subreg bug: Index: cse.c === --- cse.c (revision 238891) +++ cse.c

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 --- Comment #16 from Bernd Edlinger --- Alternative patch, that looks like it fixes this issue (but not the other bug). Note: it generates one instruction more than the first patch, thus paradoxical subregs trigger at least one other bug, and

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 --- Comment #15 from Bernd Edlinger --- patch posted here: https://gcc.gnu.org/ml/gcc-patches/2016-07/msg01742.html

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 James Greenhalgh changed: What|Removed |Added CC||zsojka at seznam dot cz --- Comment

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 --- Comment #13 from Tamar Christina --- Also successfully bootstrapped on an aarch64 juno board and no regressions.

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 --- Comment #12 from Andreas Schwab --- Sucessfully bootstrapped on aarch64_ilp32 and the ISL testsuite passes.

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 Tamar Christina changed: What|Removed |Added CC||tamar.christina at arm dot com ---

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 --- Comment #10 from Bernd Edlinger --- (In reply to James Greenhalgh from comment #2) > > So I have two questions. > > First, where did the DImode paradoxical subreg come from in the first place, > and why wasn't it a zero-extend? > >

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 Bernd Edlinger changed: What|Removed |Added CC||bernd.edlinger at hotmail dot de ---

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 Andrew Pinski changed: What|Removed |Added Keywords||ra --- Comment #8 from Andrew Pinski

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 Richard Biener changed: What|Removed |Added Priority|P3 |P2

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 --- Comment #7 from Andrew Pinski --- Ok, the one which I was looking into did the correct thing for reload which is why I was confused. Anyways reload is where the problem is. It is doing a store of SI mode but then loading it via DI mode

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 --- Comment #6 from Andrew Pinski --- (In reply to Andrew Pinski from comment #5) > Is an invalid combine. GCC must be thinking the upper bits of reg 419 (aka > 487) was zero for some reason. No that is fine. In fact reg 487 was done using

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 --- Comment #5 from Andrew Pinski --- That is the following (I am using ubuntu's 5.3.1 in 1604 right now): Trying 944, 945 -> 946: Successfully matched this instruction: (set (reg:DI 1 x1) (ior:DI (ashift:DI (reg/f:DI 110 [ obj1$4 ])

[Bug rtl-optimization/71779] [5/6/7 regression] isl miscompiled with -mabi=ilp32

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71779 Andrew Pinski changed: What|Removed |Added Component|target |rtl-optimization --- Comment #4 from