[v3,00/69] target/arm: AArch64 decodetree conversion, final part

Message

Richard Henderson Dec. 11, 2024, 4:29 p.m. UTC

Finish the conversion of all aarch64 instructions to decodetree.

Changes for v3:
  - Fix decode for f16 fsqrt (vector) in patch 23, prior to conversion.
    This is the only patch without R-B.


r~


Richard Henderson (69):
  target/arm: Add section labels for "Data Processing (register)"
  target/arm: Convert UDIV, SDIV to decodetree
  target/arm: Convert LSLV, LSRV, ASRV, RORV to decodetree
  target/arm: Convert CRC32, CRC32C to decodetree
  target/arm: Convert SUBP, IRG, GMI to decodetree
  target/arm: Convert PACGA to decodetree
  target/arm: Convert RBIT, REV16, REV32, REV64 to decodetree
  target/arm: Convert CLZ, CLS to decodetree
  target/arm: Convert PAC[ID]*, AUT[ID]* to decodetree
  target/arm: Convert XPAC[ID] to decodetree
  target/arm: Convert disas_logic_reg to decodetree
  target/arm: Convert disas_add_sub_ext_reg to decodetree
  target/arm: Convert disas_add_sub_reg to decodetree
  target/arm: Convert disas_data_proc_3src to decodetree
  target/arm: Convert disas_adc_sbc to decodetree
  target/arm: Convert RMIF to decodetree
  target/arm: Convert SETF8, SETF16 to decodetree
  target/arm: Convert CCMP, CCMN to decodetree
  target/arm: Convert disas_cond_select to decodetree
  target/arm: Introduce fp_access_check_scalar_hsd
  target/arm: Introduce fp_access_check_vector_hsd
  target/arm: Convert FCMP, FCMPE, FCCMP, FCCMPE to decodetree
  target/arm: Fix decode of fp16 vector fabs, fneg, fsqrt
  target/arm: Convert FMOV, FABS, FNEG (scalar) to decodetree
  target/arm: Pass fpstatus to vfp_sqrt*
  target/arm: Remove helper_sqrt_f16
  target/arm: Convert FSQRT (scalar) to decodetree
  target/arm: Convert FRINT[NPMSAXI] (scalar) to decodetree
  target/arm: Convert BFCVT to decodetree
  target/arm: Convert FRINT{32,64}[ZX] (scalar) to decodetree
  target/arm: Convert FCVT (scalar) to decodetree
  target/arm: Convert handle_fpfpcvt to decodetree
  target/arm: Convert FJCVTZS to decodetree
  target/arm: Convert handle_fmov to decodetree
  target/arm: Convert SQABS, SQNEG to decodetree
  target/arm: Convert ABS, NEG to decodetree
  target/arm: Introduce gen_gvec_cls, gen_gvec_clz
  target/arm: Convert CLS, CLZ (vector) to decodetree
  target/arm: Introduce gen_gvec_cnt, gen_gvec_rbit
  target/arm: Convert CNT, NOT, RBIT (vector) to decodetree
  target/arm: Convert CMGT, CMGE, GMLT, GMLE, CMEQ (zero) to decodetree
  target/arm: Introduce gen_gvec_rev{16,32,64}
  target/arm: Convert handle_rev to decodetree
  target/arm: Move helper_neon_addlp_{s8,s16} to neon_helper.c
  target/arm: Introduce gen_gvec_{s,u}{add,ada}lp
  target/arm: Convert handle_2misc_pairwise to decodetree
  target/arm: Remove helper_neon_{add,sub}l_u{16,32}
  target/arm: Introduce clear_vec
  target/arm: Convert XTN, SQXTUN, SQXTN, UQXTN to decodetree
  target/arm: Convert FCVTN, BFCVTN to decodetree
  target/arm: Convert FCVTXN to decodetree
  target/arm: Convert SHLL to decodetree
  target/arm: Implement gen_gvec_fabs, gen_gvec_fneg
  target/arm: Convert FABS, FNEG (vector) to decodetree
  target/arm: Convert FSQRT (vector) to decodetree
  target/arm: Convert FRINT* (vector) to decodetree
  target/arm: Convert FCVT* (vector, integer) scalar to decodetree
  target/arm: Convert FCVT* (vector, fixed-point) scalar to decodetree
  target/arm: Convert [US]CVTF (vector, integer) scalar to decodetree
  target/arm: Convert [US]CVTF (vector, fixed-point) scalar to
    decodetree
  target/arm: Rename helper_gvec_vcvt_[hf][su] with _rz
  target/arm: Convert [US]CVTF (vector) to decodetree
  target/arm: Convert FCVTZ[SU] (vector, fixed-point) to decodetree
  target/arm: Convert FCVT* (vector, integer) to decodetree
  target/arm: Convert handle_2misc_fcmp_zero to decodetree
  target/arm: Convert FRECPE, FRECPX, FRSQRTE to decodetree
  target/arm: Introduce gen_gvec_urecpe, gen_gvec_ursqrte
  target/arm: Convert URECPE and URSQRTE to decodetree
  target/arm: Convert FCVTL to decodetree

 target/arm/helper.h             |   43 +-
 target/arm/tcg/helper-a64.h     |    7 -
 target/arm/tcg/translate.h      |   35 +
 target/arm/tcg/gengvec.c        |  369 ++
 target/arm/tcg/helper-a64.c     |  104 -
 target/arm/tcg/neon_helper.c    |  106 +-
 target/arm/tcg/translate-a64.c  | 5670 ++++++++++---------------------
 target/arm/tcg/translate-neon.c |  337 +-
 target/arm/tcg/translate-vfp.c  |    6 +-
 target/arm/tcg/vec_helper.c     |   65 +-
 target/arm/vfp_helper.c         |   16 +-
 target/arm/tcg/a64.decode       |  502 ++-
 12 files changed, 2888 insertions(+), 4372 deletions(-)

Comments

Peter Maydell Dec. 13, 2024, 1:42 p.m. UTC | #1

On Wed, 11 Dec 2024 at 16:42, Richard Henderson
<[e-mail]> wrote:
>
> Finish the conversion of all aarch64 instructions to decodetree.
>
> Changes for v3:
>   - Fix decode for f16 fsqrt (vector) in patch 23, prior to conversion.
>     This is the only patch without R-B.
>
>
> r~



Applied to target-arm.next, thanks.

-- PMM