From 9caa6732f9aafd8335adda6e9a57239d1fed40cb Mon Sep 17 00:00:00 2001 From: Alan Modra Date: Thu, 29 Nov 2018 15:33:01 +1030 Subject: [PATCH] [RS6000] PR11848 rs6000_emit_move long double split This split is disabled for power7 and up, so we don't often see its bad effects. However, on a powerpc-linux compiler (which defaults to PPC750 judging from rs6000/sysv4.h) we see long double ld1 (void) { return 1.0L; } compiled with -msoft-float -O2 -S resulting in ld1: li 3,0 li 4,0 mr 6,4 mr 5,3 li 4,0 lis 3,0x3ff0 blr Things go awry in init-regs, with the TFmode reg being initialized to zero on seeing the subreg from the split. (And that initialization itself is split by rs6000_emit_move!) Later passes apparently don't clean up the rubbish. Since the split was added for Darwin (as the comment says), let's get rid of it on other targets. * config/rs6000/rs6000.c (rs6000_emit_move): Disable long double split for targets other than Darwin. From-SVN: r266610 --- gcc/ChangeLog | 7 ++++++- gcc/config/rs6000/rs6000.c | 4 ++-- 2 files changed, 8 insertions(+), 3 deletions(-) diff --git a/gcc/ChangeLog b/gcc/ChangeLog index 011fe4c60ac..36b1b2baa8b 100644 --- a/gcc/ChangeLog +++ b/gcc/ChangeLog @@ -1,3 +1,8 @@ +2018-11-29 Alan Modra + + * config/rs6000/rs6000.c (rs6000_emit_move): Disable long + double split for targets other than Darwin. + 2018-11-29 Alan Modra * config/rs6000/rs6000.md (movdi_internal64): Correct lengths. @@ -692,7 +697,7 @@ 2018-11-23 Christoph Muellner - * config/aarch64/aarch64-cores.def: Define emag. + * config/aarch64/aarch64-cores.def: Define emag. * config/aarch64/aarch64-tune.md: Regenerated with emag. * config/aarch64/aarch64.c (emag_tunings): New struct. * doc/invoke.texi: Document mtune value. diff --git a/gcc/config/rs6000/rs6000.c b/gcc/config/rs6000/rs6000.c index 93d58a2855f..d986207a0c6 100644 --- a/gcc/config/rs6000/rs6000.c +++ b/gcc/config/rs6000/rs6000.c @@ -9967,8 +9967,8 @@ rs6000_emit_move (rtx dest, rtx source, machine_mode mode) /* 128-bit constant floating-point values on Darwin should really be loaded as two parts. However, this premature splitting is a problem when DFmode values can go into Altivec registers. */ - if (FLOAT128_IBM_P (mode) && !reg_addr[DFmode].scalar_in_vmx_p - && GET_CODE (operands[1]) == CONST_DOUBLE) + if (TARGET_MACHO && CONST_DOUBLE_P (operands[1]) && FLOAT128_IBM_P (mode) + && !reg_addr[DFmode].scalar_in_vmx_p) { rs6000_emit_move (simplify_gen_subreg (DFmode, operands[0], mode, 0), simplify_gen_subreg (DFmode, operands[1], mode, 0), -- 2.30.2