On Tue, 28 Jun 2022 at 05:36, Richard Henderson
<richard.hender...@linaro.org> wrote:
>
> Signed-off-by: Richard Henderson <richard.hender...@linaro.org>
> ---
> v4: Drop restrict.
> ---
> target/arm/helper-sme.h | 5 +++
> target/arm/sme.decode | 11 +++++
> target/arm/sme_helper.c | 90 ++++++++++++++++++++++++++++++++++++++
> target/arm/translate-sme.c | 31 +++++++++++++
> 4 files changed, 137 insertions(+)
> #undef DO_ST
> +
> +void HELPER(sme_addha_s)(void *vzda, void *vzn, void *vpn,
> + void *vpm, uint32_t desc)
> +{
> + intptr_t row, col, oprsz = simd_oprsz(desc) / 4;
> + uint64_t *pn = vpn, *pm = vpm;
> + uint32_t *zda = vzda, *zn = vzn;
> +
> + for (row = 0; row < oprsz; ) {
> + uint64_t pa = pn[row >> 4];
> + do {
> + if (pa & 1) {
> + for (col = 0; col < oprsz; ) {
> + uint64_t pb = pm[col >> 4];
> + do {
> + if (pb & 1) {
> + zda[row * sizeof(ARMVectorReg) + col] += zn[col];
> + }
> + pb >>= 4;
> + } while (++col & 15);
> + }
> + }
> + pa >>= 4;
> + } while (++row & 15);
> + }
> +}
> +
> +void HELPER(sme_addha_d)(void *vzda, void *vzn, void *vpn,
> + void *vpm, uint32_t desc)
> +{
> + intptr_t row, col, oprsz = simd_oprsz(desc) / 8;
> + uint8_t *pn = vpn, *pm = vpm;
> + uint64_t *zda = vzda, *zn = vzn;
> +
> + for (row = 0; row < oprsz; ++row) {
> + if (pn[H1(row)] & 1) {
> + for (col = 0; col < oprsz; ++col) {
> + if (pm[H1(col)] & 1) {
> + zda[row * sizeof(ARMVectorReg) + col] += zn[col];
> + }
> + }
> + }
> + }
> +}
These array index calculations look wrong again?
Should be 'row * (sizeof(ARMVectorReg) / 8) + col' or equivalent,
I think.
> +
> +void HELPER(sme_addva_s)(void *vzda, void *vzn, void *vpn,
> + void *vpm, uint32_t desc)
> +{
> + intptr_t row, col, oprsz = simd_oprsz(desc) / 4;
> + uint64_t *pn = vpn, *pm = vpm;
> + uint32_t *zda = vzda, *zn = vzn;
> +
> + for (row = 0; row < oprsz; ) {
> + uint64_t pa = pn[row >> 4];
> + do {
> + if (pa & 1) {
> + uint32_t zn_row = zn[row];
> + for (col = 0; col < oprsz; ) {
> + uint64_t pb = pm[col >> 4];
> + do {
> + if (pb & 1) {
> + zda[row * sizeof(ARMVectorReg) + col] += zn_row;
> + }
> + pb >>= 4;
> + } while (++col & 15);
> + }
> + }
> + pa >>= 4;
> + } while (++row & 15);
> + }
> +}
> +
> +void HELPER(sme_addva_d)(void *vzda, void *vzn, void *vpn,
> + void *vpm, uint32_t desc)
> +{
> + intptr_t row, col, oprsz = simd_oprsz(desc) / 8;
> + uint8_t *pn = vpn, *pm = vpm;
> + uint64_t *zda = vzda, *zn = vzn;
> +
> + for (row = 0; row < oprsz; ++row) {
> + if (pn[H1(row)] & 1) {
> + uint64_t zn_row = zn[row];
> + for (col = 0; col < oprsz; ++col) {
> + if (pm[H1(col)] & 1) {
> + zda[row * sizeof(ARMVectorReg) + col] += zn_row;
> + }
> + }
> + }
> + }
> +}
-- PMM
