qemu-riscv
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH v2] target/riscv: fix shifts shamt value for rv128c


From: Frédéric Pétrot
Subject: Re: [PATCH v2] target/riscv: fix shifts shamt value for rv128c
Date: Mon, 11 Jul 2022 21:54:27 +0200
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101 Thunderbird/91.11.0

Hi Richard,

Le 11/07/2022 à 06:44, Richard Henderson a écrit :
On Sun, 10 July 2022, 16:36 Frédéric Pétrot, <frederic.petrot@univ-grenoble-alpes.fr <mailto:frederic.petrot@univ-grenoble-alpes.fr>> wrote:

    For rv128c shifts, a shamt of 0 is a shamt of 64, while for rv32c/rv64c
    it stays 0 and is a hint instruction that does not change processor state.
    For rv128c right shifts, the 6-bit shamt is in addition sign extended to
    7 bits.


Um, what says that? First, it makes no sense — while some systems define negative shifts as opposite direction, riscv is not one of them.

  This is here (I just pulled riscv-isa-manual and recompiled to be right on
  the page): Volume I: RISC-V Unprivileged ISA V20191214-draft (same branch as
  the one you indicate), page 119 & 120.
  This applies *only* to the compressed instructions c.srli and c.srai, not to
  their uncompressed version (the 0 becoming 64 also applies to c.slli).
  "Sign extension" is duplication of the 6th bit on the 7th bit only, and the
  top bits are zeroed, thus leading to shamts of 1-31, 64, 96-127.

Second, the 20220707 draft spec does not agree with you:

   Shifts by an immediate (SLLI/SRLI/SRAI)
   are now encoded using the low 7 bits of
   the I-immediate.... >
Nothing about that sentence says "signed".

  Agreed, on the non compressed insns, but the compressed ones have a 6-bit
  shamt only as visible on page 18.6 page 125. The explanation for rv128 shifts
  is further detailed in the emphasized paragraph on top of page 120.

  Frédéric


r~


    Signed-off-by: Frédéric Pétrot <frederic.petrot@univ-grenoble-alpes.fr
    <mailto:frederic.petrot@univ-grenoble-alpes.fr>>
    ---
      target/riscv/insn16.decode |  7 ++++---
      disas/riscv.c              | 27 +++++++++++++++++++++------
      target/riscv/translate.c   | 20 ++++++++++++++++++--
      3 files changed, 43 insertions(+), 11 deletions(-)

    diff --git a/target/riscv/insn16.decode b/target/riscv/insn16.decode
    index 02c8f61b48..ccfe59f294 100644
    --- a/target/riscv/insn16.decode
    +++ b/target/riscv/insn16.decode
    @@ -31,7 +31,8 @@
      %imm_cb        12:s1 5:2 2:1 10:2 3:2 !function=ex_shift_1
      %imm_cj        12:s1 8:1 9:2 6:1 7:1 2:1 11:1 3:3 !function=ex_shift_1

    -%shimm_6bit   12:1 2:5               !function=ex_rvc_shifti
    +%shlimm_6bit  12:1 2:5               !function=ex_rvc_shiftli
    +%shrimm_6bit  12:1 2:5               !function=ex_rvc_shiftri
      %uimm_6bit_lq 2:4 12:1 6:1           !function=ex_shift_4
      %uimm_6bit_ld 2:3 12:1 5:2           !function=ex_shift_3
      %uimm_6bit_lw 2:2 12:1 4:3           !function=ex_shift_2
    @@ -82,9 +83,9 @@
      @c_addi16sp     ... .  ..... ..... .. &i imm=%imm_addi16sp rs1=2 rd=2

      @c_shift        ... . .. ... ..... .. \
    -                &shift rd=%rs1_3 rs1=%rs1_3 shamt=%shimm_6bit
    +                &shift rd=%rs1_3 rs1=%rs1_3 shamt=%shrimm_6bit
      @c_shift2       ... . .. ... ..... .. \
    -                &shift rd=%rd rs1=%rd shamt=%shimm_6bit
    +                &shift rd=%rd rs1=%rd shamt=%shlimm_6bit

      @c_andi         ... . .. ... ..... .. &i imm=%imm_ci rs1=%rs1_3 rd=%rs1_3

    diff --git a/disas/riscv.c b/disas/riscv.c
    index 7af6afc8fa..489c2ae5e8 100644
    --- a/disas/riscv.c
    +++ b/disas/riscv.c
    @@ -2402,10 +2402,25 @@ static int32_t operand_sbimm12(rv_inst inst)
              ((inst << 56) >> 63) << 11;
      }

    -static uint32_t operand_cimmsh6(rv_inst inst)
    +static uint32_t operand_cimmshl6(rv_inst inst, rv_isa isa)
      {
    -    return ((inst << 51) >> 63) << 5 |
    +    int imm = ((inst << 51) >> 63) << 5 |
              (inst << 57) >> 59;
    +    if (isa == rv128) {
    +        imm = imm ? imm : 64;
    +    }
    +    return imm;
    +}
    +
    +static uint32_t operand_cimmshr6(rv_inst inst, rv_isa isa)
    +{
    +    int imm = ((inst << 51) >> 63) << 5 |
    +        (inst << 57) >> 59;
    +    if (isa == rv128) {
    +        imm = imm | (imm & 32) << 1;
    +        imm = imm ? imm : 64;
    +    }
    +    return imm;
      }

      static int32_t operand_cimmi(rv_inst inst)
    @@ -2529,7 +2544,7 @@ static uint32_t operand_rnum(rv_inst inst)

      /* decode operands */

    -static void decode_inst_operands(rv_decode *dec)
    +static void decode_inst_operands(rv_decode *dec, rv_isa isa)
      {
          rv_inst inst = dec->inst;
          dec->codec = opcode_data[dec->op].codec;
    @@ -2652,7 +2667,7 @@ static void decode_inst_operands(rv_decode *dec)
          case rv_codec_cb_sh6:
              dec->rd = dec->rs1 = operand_crs1rdq(inst) + 8;
              dec->rs2 = rv_ireg_zero;
    -        dec->imm = operand_cimmsh6(inst);
    +        dec->imm = operand_cimmshr6(inst, isa);
              break;
          case rv_codec_ci:
              dec->rd = dec->rs1 = operand_crs1rd(inst);
    @@ -2667,7 +2682,7 @@ static void decode_inst_operands(rv_decode *dec)
          case rv_codec_ci_sh6:
              dec->rd = dec->rs1 = operand_crs1rd(inst);
              dec->rs2 = rv_ireg_zero;
    -        dec->imm = operand_cimmsh6(inst);
    +        dec->imm = operand_cimmshl6(inst, isa);
              break;
          case rv_codec_ci_16sp:
              dec->rd = rv_ireg_sp;
    @@ -3193,7 +3208,7 @@ disasm_inst(char *buf, size_t buflen, rv_isa isa,
    uint64_t pc, rv_inst inst)
          dec.pc = pc;
          dec.inst = inst;
          decode_inst_opcode(&dec, isa);
    -    decode_inst_operands(&dec);
    +    decode_inst_operands(&dec, isa);
          decode_inst_decompress(&dec, isa);
          decode_inst_lift_pseudo(&dec);
          format_inst(buf, buflen, 16, &dec);
    diff --git a/target/riscv/translate.c b/target/riscv/translate.c
    index 63b04e8a94..d7c82a9c81 100644
    --- a/target/riscv/translate.c
    +++ b/target/riscv/translate.c
    @@ -705,10 +705,26 @@ static int ex_rvc_register(DisasContext *ctx, int reg)
          return 8 + reg;
      }

    -static int ex_rvc_shifti(DisasContext *ctx, int imm)
    +static int ex_rvc_shiftli(DisasContext *ctx, int imm)
      {
          /* For RV128 a shamt of 0 means a shift by 64. */
    -    return imm ? imm : 64;
    +    if (get_ol(ctx) == MXL_RV128) {
    +        imm = imm ? imm : 64;
    +    }
    +    return imm;
    +}
    +
    +static int ex_rvc_shiftri(DisasContext *ctx, int imm)
    +{
    +    /*
    +     * For RV128 a shamt of 0 means a shift by 64, furthermore, for right
    +     * shifts, the shamt is sign-extended.
    +     */
    +    if (get_ol(ctx) == MXL_RV128) {
    +        imm = imm | (imm & 32) << 1;
    +        imm = imm ? imm : 64;
    +    }
    +    return imm;
      }

      /* Include the auto-generated decoder for 32 bit insn */
-- 2.36.1



--
+---------------------------------------------------------------------------+
| Frédéric Pétrot,                            Pr. Grenoble INP-Ensimag/TIMA |
| Mob/Pho: +33 6 74 57 99 65/+33 4 76 57 48 70      Ad augusta  per angusta |
| http://tima.univ-grenoble-alpes.fr frederic.petrot@univ-grenoble-alpes.fr |
+---------------------------------------------------------------------------+



reply via email to

[Prev in Thread] Current Thread [Next in Thread]