Re: [PATCH v3 01/44] target/arm: Implement MVE VLDR/VSTR (non-widening forms)

[Claudio Fontana]

On 6/17/21 2:15 PM, Peter Maydell wrote:
> Implement the forms of the MVE VLDR and VSTR insns which perform
> non-widening loads of bytes, halfwords or words from memory into
> vector elements of the same width (encodings T5, T6, T7).

This is if I understand correctly, M-Profile only, and thus TCG-only right?

> 
> (At the moment we know for MVE and M-profile in general that
> vfp_access_check() can never return false, but we include the
> conventional return-true-on-failure check for consistency
> with non-M-profile translation code.)
> 
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
> Changes v2->v3: pass MSIZE to specify memory size,
> and advance addr by that rather than by ESIZE;
> advance addr always, not only when predication passes.
> ---
>  target/arm/{translate-mve.c => helper-mve.h} |  19 +-
>  target/arm/helper.h                          |   2 +
>  target/arm/internals.h                       |  11 ++
>  target/arm/mve.decode                        |  22 +++
>  target/arm/mve_helper.c                      | 172 +++++++++++++++++++
>  target/arm/translate-mve.c                   | 119 +++++++++++++
>  target/arm/meson.build                       |   1 +
>  7 files changed, 334 insertions(+), 12 deletions(-)
>  copy target/arm/{translate-mve.c => helper-mve.h} (61%)
>  create mode 100644 target/arm/mve_helper.c
> 
> diff --git a/target/arm/translate-mve.c b/target/arm/helper-mve.h
> similarity index 61%
> copy from target/arm/translate-mve.c
> copy to target/arm/helper-mve.h
> index e91f526a1a8..9e3b0b09afd 100644
> --- a/target/arm/translate-mve.c
> +++ b/target/arm/helper-mve.h
> @@ -1,5 +1,5 @@
>  /*
> - *  ARM translation: M-profile MVE instructions
> + *  M-profile MVE specific helper definitions
>   *
>   *  Copyright (c) 2021 Linaro, Ltd.
>   *
> @@ -16,14 +16,9 @@
>   * You should have received a copy of the GNU Lesser General Public
>   * License along with this library; if not, see 
> <http://www.gnu.org/licenses/>.
>   */
> -
> -#include "qemu/osdep.h"
> -#include "tcg/tcg-op.h"
> -#include "tcg/tcg-op-gvec.h"
> -#include "exec/exec-all.h"
> -#include "exec/gen-icount.h"
> -#include "translate.h"
> -#include "translate-a32.h"
> -
> -/* Include the generated decoder */
> -#include "decode-mve.c.inc"
> +DEF_HELPER_FLAGS_3(mve_vldrb, TCG_CALL_NO_WG, void, env, ptr, i32)
> +DEF_HELPER_FLAGS_3(mve_vldrh, TCG_CALL_NO_WG, void, env, ptr, i32)
> +DEF_HELPER_FLAGS_3(mve_vldrw, TCG_CALL_NO_WG, void, env, ptr, i32)
> +DEF_HELPER_FLAGS_3(mve_vstrb, TCG_CALL_NO_WG, void, env, ptr, i32)
> +DEF_HELPER_FLAGS_3(mve_vstrh, TCG_CALL_NO_WG, void, env, ptr, i32)
> +DEF_HELPER_FLAGS_3(mve_vstrw, TCG_CALL_NO_WG, void, env, ptr, i32)
> diff --git a/target/arm/helper.h b/target/arm/helper.h
> index dc6eb96d439..db87d7d5376 100644
> --- a/target/arm/helper.h
> +++ b/target/arm/helper.h
> @@ -1019,3 +1019,5 @@ DEF_HELPER_FLAGS_6(gvec_bfmlal_idx, TCG_CALL_NO_RWG,
>  #include "helper-a64.h"
>  #include "helper-sve.h"
>  #endif
> +
> +#include "helper-mve.h"
> diff --git a/target/arm/internals.h b/target/arm/internals.h
> index 886db56b580..3ba86e8af81 100644
> --- a/target/arm/internals.h
> +++ b/target/arm/internals.h
> @@ -1202,4 +1202,15 @@ static inline uint64_t 
> useronly_maybe_clean_ptr(uint32_t desc, uint64_t ptr)
>      return ptr;
>  }
>  
> +/* Values for M-profile PSR.ECI for MVE insns */
> +enum MVEECIState {
> +    ECI_NONE = 0, /* No completed beats */
> +    ECI_A0 = 1, /* Completed: A0 */
> +    ECI_A0A1 = 2, /* Completed: A0, A1 */
> +    /* 3 is reserved */
> +    ECI_A0A1A2 = 4, /* Completed: A0, A1, A2 */
> +    ECI_A0A1A2B0 = 5, /* Completed: A0, A1, A2, B0 */
> +    /* All other values reserved */
> +};
> +

I wonder if this should be in a MVE-specific header,
and I also wonder, when looking at code using things like ECI_A0,

what about MVE_ECI_NONE , MVE_ECI_A0, ... - would help the reader connect code 
containing these enums with the MVE feature?

Thanks,

C


>  #endif
> diff --git a/target/arm/mve.decode b/target/arm/mve.decode
> index c8492bb5763..858a161fd7e 100644
> --- a/target/arm/mve.decode
> +++ b/target/arm/mve.decode
> @@ -18,3 +18,25 @@
>  #
>  # This file is processed by scripts/decodetree.py
>  #
> +
> +%qd 22:1 13:3
> +
> +&vldr_vstr rn qd imm p a w size l
> +
> +@vldr_vstr ....... . . . . l:1 rn:4 ... ...... imm:7 &vldr_vstr qd=%qd
> +
> +# Vector loads and stores
> +
> +# Non-widening loads/stores (P=0 W=0 is 'related encoding')
> +VLDR_VSTR        1110110 0 a:1 . 1   . .... ... 111100 .......   @vldr_vstr \
> +                 size=0 p=0 w=1
> +VLDR_VSTR        1110110 0 a:1 . 1   . .... ... 111101 .......   @vldr_vstr \
> +                 size=1 p=0 w=1
> +VLDR_VSTR        1110110 0 a:1 . 1   . .... ... 111110 .......   @vldr_vstr \
> +                 size=2 p=0 w=1
> +VLDR_VSTR        1110110 1 a:1 . w:1 . .... ... 111100 .......   @vldr_vstr \
> +                 size=0 p=1
> +VLDR_VSTR        1110110 1 a:1 . w:1 . .... ... 111101 .......   @vldr_vstr \
> +                 size=1 p=1
> +VLDR_VSTR        1110110 1 a:1 . w:1 . .... ... 111110 .......   @vldr_vstr \
> +                 size=2 p=1
> diff --git a/target/arm/mve_helper.c b/target/arm/mve_helper.c
> new file mode 100644
> index 00000000000..60c61268c7c
> --- /dev/null
> +++ b/target/arm/mve_helper.c
> @@ -0,0 +1,172 @@
> +/*
> + * M-profile MVE Operations
> + *
> + * Copyright (c) 2021 Linaro, Ltd.
> + *
> + * This library is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU Lesser General Public
> + * License as published by the Free Software Foundation; either
> + * version 2.1 of the License, or (at your option) any later version.
> + *
> + * This library is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> + * Lesser General Public License for more details.
> + *
> + * You should have received a copy of the GNU Lesser General Public
> + * License along with this library; if not, see 
> <http://www.gnu.org/licenses/>.
> + */
> +
> +#include "qemu/osdep.h"
> +#include "cpu.h"
> +#include "internals.h"
> +#include "vec_internal.h"
> +#include "exec/helper-proto.h"
> +#include "exec/cpu_ldst.h"
> +#include "exec/exec-all.h"
> +
> +static uint16_t mve_element_mask(CPUARMState *env)
> +{
> +    /*
> +     * Return the mask of which elements in the MVE vector should be
> +     * updated. This is a combination of multiple things:
> +     *  (1) by default, we update every lane in the vector
> +     *  (2) VPT predication stores its state in the VPR register;
> +     *  (3) low-overhead-branch tail predication will mask out part
> +     *      the vector on the final iteration of the loop
> +     *  (4) if EPSR.ECI is set then we must execute only some beats
> +     *      of the insn
> +     * We combine all these into a 16-bit result with the same semantics
> +     * as VPR.P0: 0 to mask the lane, 1 if it is active.
> +     * 8-bit vector ops will look at all bits of the result;
> +     * 16-bit ops will look at bits 0, 2, 4, ...;
> +     * 32-bit ops will look at bits 0, 4, 8 and 12.
> +     * Compare pseudocode GetCurInstrBeat(), though that only returns
> +     * the 4-bit slice of the mask corresponding to a single beat.
> +     */
> +    uint16_t mask = FIELD_EX32(env->v7m.vpr, V7M_VPR, P0);
> +
> +    if (!(env->v7m.vpr & R_V7M_VPR_MASK01_MASK)) {
> +        mask |= 0xff;
> +    }
> +    if (!(env->v7m.vpr & R_V7M_VPR_MASK23_MASK)) {
> +        mask |= 0xff00;
> +    }
> +
> +    if (env->v7m.ltpsize < 4 &&
> +        env->regs[14] <= (1 << (4 - env->v7m.ltpsize))) {
> +        /*
> +         * Tail predication active, and this is the last loop iteration.
> +         * The element size is (1 << ltpsize), and we only want to process
> +         * loopcount elements, so we want to retain the least significant
> +         * (loopcount * esize) predicate bits and zero out bits above that.
> +         */
> +        int masklen = env->regs[14] << env->v7m.ltpsize;
> +        assert(masklen <= 16);
> +        mask &= MAKE_64BIT_MASK(0, masklen);
> +    }
> +
> +    if ((env->condexec_bits & 0xf) == 0) {
> +        /*
> +         * ECI bits indicate which beats are already executed;
> +         * we handle this by effectively predicating them out.
> +         */
> +        int eci = env->condexec_bits >> 4;
> +        switch (eci) {
> +        case ECI_NONE:
> +            break;
> +        case ECI_A0:
> +            mask &= 0xfff0;
> +            break;
> +        case ECI_A0A1:
> +            mask &= 0xff00;
> +            break;
> +        case ECI_A0A1A2:
> +        case ECI_A0A1A2B0:
> +            mask &= 0xf000;
> +            break;
> +        default:
> +            g_assert_not_reached();
> +        }
> +    }
> +
> +    return mask;
> +}
> +
> +static void mve_advance_vpt(CPUARMState *env)
> +{
> +    /* Advance the VPT and ECI state if necessary */
> +    uint32_t vpr = env->v7m.vpr;
> +    unsigned mask01, mask23;
> +
> +    if ((env->condexec_bits & 0xf) == 0) {
> +        env->condexec_bits = (env->condexec_bits == (ECI_A0A1A2B0 << 4)) ?
> +            (ECI_A0 << 4) : (ECI_NONE << 4);
> +    }
> +
> +    if (!(vpr & (R_V7M_VPR_MASK01_MASK | R_V7M_VPR_MASK23_MASK))) {
> +        /* VPT not enabled, nothing to do */
> +        return;
> +    }
> +
> +    mask01 = FIELD_EX32(vpr, V7M_VPR, MASK01);
> +    mask23 = FIELD_EX32(vpr, V7M_VPR, MASK23);
> +    if (mask01 > 8) {
> +        /* high bit set, but not 0b1000: invert the relevant half of P0 */
> +        vpr ^= 0xff;
> +    }
> +    if (mask23 > 8) {
> +        /* high bit set, but not 0b1000: invert the relevant half of P0 */
> +        vpr ^= 0xff00;
> +    }
> +    vpr = FIELD_DP32(vpr, V7M_VPR, MASK01, mask01 << 1);
> +    vpr = FIELD_DP32(vpr, V7M_VPR, MASK23, mask23 << 1);
> +    env->v7m.vpr = vpr;
> +}
> +
> +
> +#define DO_VLDR(OP, MSIZE, LDTYPE, ESIZE, TYPE)                         \
> +    void HELPER(mve_##OP)(CPUARMState *env, void *vd, uint32_t addr)    \
> +    {                                                                   \
> +        TYPE *d = vd;                                                   \
> +        uint16_t mask = mve_element_mask(env);                          \
> +        unsigned b, e;                                                  \
> +        /*                                                              \
> +         * R_SXTM allows the dest reg to become UNKNOWN for abandoned   \
> +         * beats so we don't care if we update part of the dest and     \
> +         * then take an exception.                                      \
> +         */                                                             \
> +        for (b = 0, e = 0; b < 16; b += ESIZE, e++) {                   \
> +            if (mask & (1 << b)) {                                      \
> +                d[H##ESIZE(e)] = cpu_##LDTYPE##_data_ra(env, addr, GETPC()); 
> \
> +            }                                                           \
> +            addr += MSIZE;                                              \
> +        }                                                               \
> +        mve_advance_vpt(env);                                           \
> +    }
> +
> +#define DO_VSTR(OP, MSIZE, STTYPE, ESIZE, TYPE)                         \
> +    void HELPER(mve_##OP)(CPUARMState *env, void *vd, uint32_t addr)    \
> +    {                                                                   \
> +        TYPE *d = vd;                                                   \
> +        uint16_t mask = mve_element_mask(env);                          \
> +        unsigned b, e;                                                  \
> +        for (b = 0, e = 0; b < 16; b += ESIZE, e++) {                   \
> +            if (mask & (1 << b)) {                                      \
> +                cpu_##STTYPE##_data_ra(env, addr, d[H##ESIZE(e)], GETPC()); \
> +            }                                                           \
> +            addr += MSIZE;                                              \
> +        }                                                               \
> +        mve_advance_vpt(env);                                           \
> +    }
> +
> +DO_VLDR(vldrb, 1, ldub, 1, uint8_t)
> +DO_VLDR(vldrh, 2, lduw, 2, uint16_t)
> +DO_VLDR(vldrw, 4, ldl, 4, uint32_t)
> +
> +DO_VSTR(vstrb, 1, stb, 1, uint8_t)
> +DO_VSTR(vstrh, 2, stw, 2, uint16_t)
> +DO_VSTR(vstrw, 4, stl, 4, uint32_t)
> +
> +#undef DO_VLDR
> +#undef DO_VSTR
> diff --git a/target/arm/translate-mve.c b/target/arm/translate-mve.c
> index e91f526a1a8..f98bd6d038d 100644
> --- a/target/arm/translate-mve.c
> +++ b/target/arm/translate-mve.c
> @@ -27,3 +27,122 @@
>  
>  /* Include the generated decoder */
>  #include "decode-mve.c.inc"
> +
> +typedef void MVEGenLdStFn(TCGv_ptr, TCGv_ptr, TCGv_i32);
> +
> +/* Return the offset of a Qn register (same semantics as aa32_vfp_qreg()) */
> +static inline long mve_qreg_offset(unsigned reg)
> +{
> +    return offsetof(CPUARMState, vfp.zregs[reg].d[0]);
> +}
> +
> +static TCGv_ptr mve_qreg_ptr(unsigned reg)
> +{
> +    TCGv_ptr ret = tcg_temp_new_ptr();
> +    tcg_gen_addi_ptr(ret, cpu_env, mve_qreg_offset(reg));
> +    return ret;
> +}
> +
> +static bool mve_check_qreg_bank(DisasContext *s, int qmask)
> +{
> +    /*
> +     * Check whether Qregs are in range. For v8.1M only Q0..Q7
> +     * are supported, see VFPSmallRegisterBank().
> +     */
> +    return qmask < 8;
> +}
> +
> +static bool mve_eci_check(DisasContext *s)
> +{
> +    /*
> +     * This is a beatwise insn: check that ECI is valid (not a
> +     * reserved value) and note that we are handling it.
> +     * Return true if OK, false if we generated an exception.
> +     */
> +    s->eci_handled = true;
> +    switch (s->eci) {
> +    case ECI_NONE:
> +    case ECI_A0:
> +    case ECI_A0A1:
> +    case ECI_A0A1A2:
> +    case ECI_A0A1A2B0:
> +        return true;
> +    default:
> +        /* Reserved value: INVSTATE UsageFault */
> +        gen_exception_insn(s, s->pc_curr, EXCP_INVSTATE, syn_uncategorized(),
> +                           default_exception_el(s));
> +        return false;
> +    }
> +}
> +
> +static void mve_update_eci(DisasContext *s)
> +{
> +    /*
> +     * The helper function will always update the CPUState field,
> +     * so we only need to update the DisasContext field.
> +     */
> +    if (s->eci) {
> +        s->eci = (s->eci == ECI_A0A1A2B0) ? ECI_A0 : ECI_NONE;
> +    }
> +}
> +
> +static bool do_ldst(DisasContext *s, arg_VLDR_VSTR *a, MVEGenLdStFn *fn)
> +{
> +    TCGv_i32 addr;
> +    uint32_t offset;
> +    TCGv_ptr qreg;
> +
> +    if (!dc_isar_feature(aa32_mve, s) ||
> +        !mve_check_qreg_bank(s, a->qd) ||
> +        !fn) {
> +        return false;
> +    }
> +
> +    /* CONSTRAINED UNPREDICTABLE: we choose to UNDEF */
> +    if (a->rn == 15 || (a->rn == 13 && a->w)) {
> +        return false;
> +    }
> +
> +    if (!mve_eci_check(s) || !vfp_access_check(s)) {
> +        return true;
> +    }
> +
> +    offset = a->imm << a->size;
> +    if (!a->a) {
> +        offset = -offset;
> +    }
> +    addr = load_reg(s, a->rn);
> +    if (a->p) {
> +        tcg_gen_addi_i32(addr, addr, offset);
> +    }
> +
> +    qreg = mve_qreg_ptr(a->qd);
> +    fn(cpu_env, qreg, addr);
> +    tcg_temp_free_ptr(qreg);
> +
> +    /*
> +     * Writeback always happens after the last beat of the insn,
> +     * regardless of predication
> +     */
> +    if (a->w) {
> +        if (!a->p) {
> +            tcg_gen_addi_i32(addr, addr, offset);
> +        }
> +        store_reg(s, a->rn, addr);
> +    } else {
> +        tcg_temp_free_i32(addr);
> +    }
> +    mve_update_eci(s);
> +    return true;
> +}
> +
> +static bool trans_VLDR_VSTR(DisasContext *s, arg_VLDR_VSTR *a)
> +{
> +    static MVEGenLdStFn * const ldstfns[4][2] = {
> +        { gen_helper_mve_vstrb, gen_helper_mve_vldrb },
> +        { gen_helper_mve_vstrh, gen_helper_mve_vldrh },
> +        { gen_helper_mve_vstrw, gen_helper_mve_vldrw },
> +        { NULL, NULL }
> +    };
> +    return do_ldst(s, a, ldstfns[a->size][a->l]);
> +}
> diff --git a/target/arm/meson.build b/target/arm/meson.build
> index 2b50be3f862..25a02bf2769 100644
> --- a/target/arm/meson.build
> +++ b/target/arm/meson.build
> @@ -23,6 +23,7 @@ arm_ss.add(files(
>    'helper.c',
>    'iwmmxt_helper.c',
>    'm_helper.c',
> +  'mve_helper.c',
>    'neon_helper.c',
>    'op_helper.c',
>    'tlb_helper.c',
>