diff options
Diffstat (limited to 'target/arm/neon-dp.decode')
| -rw-r--r-- | target/arm/neon-dp.decode | 18 |
1 files changed, 11 insertions, 7 deletions
diff --git a/target/arm/neon-dp.decode b/target/arm/neon-dp.decode index 1e9e859291..51aa0f0819 100644 --- a/target/arm/neon-dp.decode +++ b/target/arm/neon-dp.decode @@ -45,11 +45,16 @@ @3same_q0 .... ... . . . size:2 .... .... .... . 0 . . .... \ &3same vm=%vm_dp vn=%vn_dp vd=%vd_dp q=0 -# For FP insns the high bit of 'size' is used as part of opcode decode -@3same_fp .... ... . . . . size:1 .... .... .... . q:1 . . .... \ - &3same vm=%vm_dp vn=%vn_dp vd=%vd_dp -@3same_fp_q0 .... ... . . . . size:1 .... .... .... . 0 . . .... \ - &3same vm=%vm_dp vn=%vn_dp vd=%vd_dp q=0 +# For FP insns the high bit of 'size' is used as part of opcode decode, +# and the 'size' bit is 0 for 32-bit float and 1 for 16-bit float. +# This converts this encoding to the same MO_8/16/32/64 values that the +# integer neon insns use. +%3same_fp_size 20:1 !function=neon_3same_fp_size + +@3same_fp .... ... . . . . . .... .... .... . q:1 . . .... \ + &3same vm=%vm_dp vn=%vn_dp vd=%vd_dp size=%3same_fp_size +@3same_fp_q0 .... ... . . . . . .... .... .... . 0 . . .... \ + &3same vm=%vm_dp vn=%vn_dp vd=%vd_dp q=0 size=%3same_fp_size VHADD_S_3s 1111 001 0 0 . .. .... .... 0000 . . . 0 .... @3same VHADD_U_3s 1111 001 1 0 . .. .... .... 0000 . . . 0 .... @3same @@ -251,9 +256,8 @@ VMINNM_fp_3s 1111 001 1 0 . 1 . .... .... 1111 ... 1 .... @3same_fp @2reg_shll_b .... ... . . . 001 shift:3 .... .... 0 . . . .... \ &2reg_shift vm=%vm_dp vd=%vd_dp size=0 q=0 -# We use size=0 for fp32 and size=1 for fp16 to match the 3-same encodings. @2reg_vcvt .... ... . . . 1 ..... .... .... . q:1 . . .... \ - &2reg_shift vm=%vm_dp vd=%vd_dp size=0 shift=%neon_rshift_i5 + &2reg_shift vm=%vm_dp vd=%vd_dp size=2 shift=%neon_rshift_i5 @2reg_vcvt_f16 .... ... . . . 11 .... .... .... . q:1 . . .... \ &2reg_shift vm=%vm_dp vd=%vd_dp size=1 shift=%neon_rshift_i4 |