PR/60825 Make float64x1_t in arm_neon.h a proper vector type

gcc/ChangeLog:
	PR target/60825
	* config/aarch64/aarch64.c (aarch64_simd_mangle_map): Add entry for
	V1DFmode.
	* config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_type_mode):
	add V1DFmode
	(BUILTIN_VD1): New.
	(BUILTIN_VD_RE): Remove.
	(aarch64_init_simd_builtins): Add V1DF to modes/modenames.
	(aarch64_fold_builtin): Update reinterpret patterns, df becomes v1df.
	* config/aarch64/aarch64-simd-builtins.def (create): Make a v1df
	variant but not df.
	(vreinterpretv1df*, vreinterpret*v1df): New.
	(vreinterpretdf*, vreinterpret*df): Remove.
	* config/aarch64/aarch64-simd.md (aarch64_create, aarch64_reinterpret*):
	Generate V1DFmode pattern not DFmode.
	* config/aarch64/iterators.md (VD_RE): Include V1DF, remove DF.
	(VD1): New.
	* config/aarch64/arm_neon.h (float64x1_t): typedef with gcc extensions.
	(vcreate_f64): Remove cast, use v1df builtin.
	(vcombine_f64): Remove cast, get elements with gcc vector extensions.
	(vget_low_f64, vabs_f64, vceq_f64, vceqz_f64, vcge_f64, vgfez_f64,
	vcgt_f64, vcgtz_f64, vcle_f64, vclez_f64, vclt_f64, vcltz_f64,
	vdup_n_f64, vdupq_lane_f64, vld1_f64, vld2_f64, vld3_f64, vld4_f64,
	vmov_n_f64, vst1_f64): Use gcc vector extensions.
	(vget_lane_f64, vdupd_lane_f64, vmulq_lane_f64, ): Use gcc extensions,
	add range check using __builtin_aarch64_im_lane_boundsi.
	(vfma_lane_f64, vfmad_lane_f64, vfma_laneq_f64, vfmaq_lane_f64,
	vfms_lane_f64, vfmsd_lane_f64, vfms_laneq_f64, vfmsq_lane_f64): Fix
	type signature, use gcc vector extensions.
	(vreinterpret_p8_f64, vreinterpret_p16_f64, vreinterpret_f32_f64,
	vreinterpret_f64_f32, vreinterpret_f64_p8, vreinterpret_f64_p16,
	vreinterpret_f64_s8, vreinterpret_f64_s16, vreinterpret_f64_s32,
	vreinterpret_f64_s64, vreinterpret_f64_u8, vreinterpret_f64_u16,
	vreinterpret_f64_u32, vreinterpret_f64_u64, vreinterpret_s8_f64,
	vreinterpret_s16_f64, vreinterpret_s32_f64, vreinterpret_s64_f64,
	vreinterpret_u8_f64, vreinterpret_u16_f64, vreinterpret_u32_f64,
	vreinterpret_u64_f64): Use v1df builtin not df.

gcc/testsuite/ChangeLog:
	* g++.dg/abi/mangle-neon-aarch64.C: Also test mangling of float64x1_t.
	* gcc.target/aarch64/aapcs/test_64x1_1.c: New test.
	* gcc.target/aarch64/aapcs/func-ret-64x1_1.c: New test.
	* gcc.target/aarch64/simd/ext_f64_1.c (main): Compare vector elements.
	* gcc.target/aarch64/vadd_f64.c: Rewrite with macro to use vector types.
	* gcc.target/aarch64/vsub_f64.c: Likewise.
	* gcc.target/aarch64/vdiv_f.c (INDEX*, RUN_TEST): Remove indexing scheme
	as now the same for all variants.
	* gcc.target/aarch64/vrnd_f64_1.c (compare_f64): Return float64_t not
	float64x1_t.


git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@211892 138bc75d-0d04-0410-961f-82ee72b054a4

1 files changed, 5 insertions, 2 deletions

diff --git a/gcc/config/aarch64/iterators.md b/gcc/config/aarch64/iterators.md
index 5c304bff239..3203c3da7e2 100644
--- a/gcc/config/aarch64/iterators.md
+++ b/gcc/config/aarch64/iterators.md
@@ -138,8 +138,11 @@
 ;; Double vector modes for combines.
 (define_mode_iterator VDIC [V8QI V4HI V2SI])
 
-;; Double vector modes.
-(define_mode_iterator VD_RE [V8QI V4HI V2SI DI DF V2SF])
+;; Double vector modes, inc. V1DF and the DI "vector" mode, for VREINTERPRET.
+(define_mode_iterator VD_RE [V8QI V4HI V2SI DI V1DF V2SF])
+
+;; Double vector modes inc V1DF
+(define_mode_iterator VD1 [V8QI V4HI V2SI V2SF V1DF])
 
 ;; Vector modes except double int.
 (define_mode_iterator VDQIF [V8QI V16QI V4HI V8HI V2SI V4SI V2SF V4SF V2DF])