This patch addresses the concern the Segher raised in the original submission of the patch to fix PR target/93569. In addition to checking for D*-form addresses in the traditional Altivec registers, this patch also checks for D*-form addresses for vectors in the traditional floating point registers. Neither one of these address forms were allowed before ISA 3.0 (power9).
I have done bootstraps on both little and big endian Linux 64-bit systems, and there were no regressions for this change. Can I check this patch into the master branch? https://gcc.gnu.org/ml/gcc-patches/2020-02/msg00387.html 2020-02-06 Michael Meissner <meiss...@linux.ibm.com> PR target/93569 * config/rs6000/rs6000.c (reg_to_non_prefixed): Before ISA 3.0 we only had X-FORM (reg+reg) addressing for vectors. Also before ISA 3.0, we only had X-FORM addressing for scalars in the traditional Altivec registers. --- /tmp/VQDg8p_rs6000.c 2020-02-06 11:55:27.509363545 -0500 +++ gcc/config/rs6000/rs6000.c 2020-02-06 11:54:28.461531334 -0500 @@ -24923,7 +24923,8 @@ reg_to_non_prefixed (rtx reg, machine_mo unsigned size = GET_MODE_SIZE (mode); /* FPR registers use D-mode for scalars, and DQ-mode for vectors, IEEE - 128-bit floating point, and 128-bit integers. */ + 128-bit floating point, and 128-bit integers. Before power9, only indexed + addressing was available for vectors. */ if (FP_REGNO_P (r)) { if (mode == SFmode || size == 8 || FLOAT128_2REG_P (mode)) @@ -24936,16 +24937,20 @@ reg_to_non_prefixed (rtx reg, machine_mo && (VECTOR_MODE_P (mode) || FLOAT128_VECTOR_P (mode) || mode == TImode || mode == CTImode)) - return NON_PREFIXED_DQ; + return (TARGET_P9_VECTOR) ? NON_PREFIXED_DQ : NON_PREFIXED_X; else return NON_PREFIXED_DEFAULT; } /* Altivec registers use DS-mode for scalars, and DQ-mode for vectors, IEEE - 128-bit floating point, and 128-bit integers. */ + 128-bit floating point, and 128-bit integers. Before power9, only indexed + addressing was available. */ else if (ALTIVEC_REGNO_P (r)) { + if (!TARGET_P9_VECTOR) + return NON_PREFIXED_X; + if (mode == SFmode || size == 8 || FLOAT128_2REG_P (mode)) return NON_PREFIXED_DS; -- Michael Meissner, IBM IBM, M/S 2506R, 550 King Street, Littleton, MA 01460-6245, USA email: meiss...@linux.ibm.com, phone: +1 (978) 899-4797
