Provide a helper that 1's element I for all NF vectors. For the callers, this performs the element mask test once instead of NF times.
Signed-off-by: Richard Henderson <[email protected]> --- target/riscv/tcg/vector_helper.c | 69 ++++++++++++++++---------------- 1 file changed, 34 insertions(+), 35 deletions(-) diff --git a/target/riscv/tcg/vector_helper.c b/target/riscv/tcg/vector_helper.c index e28d8a3d9f..20a8f95efb 100644 --- a/target/riscv/tcg/vector_helper.c +++ b/target/riscv/tcg/vector_helper.c @@ -328,6 +328,15 @@ static void vext_set_tail_elems_1s(uint32_t vl, void *vd, } } +static void vext_set_nf_elems_1s(void *vd, uint32_t i, uint32_t nf, + uint32_t esz, uint32_t max_elems) +{ + for (uint32_t k = 0; k < nf; ++k) { + vext_set_elems_1s(vd, true, (i + k * max_elems) * esz, + (i + k * max_elems + 1) * esz); + } +} + /* * stride: access vector element from strided memory */ @@ -337,7 +346,6 @@ vext_ldst_stride(void *vd, void *v0, target_ulong base, target_ulong stride, vext_ldst_elem_fn_tlb *ldst_elem, uint32_t log2_esz, uintptr_t ra) { - uint32_t i, k; uint32_t nf = vext_nf(desc); uint32_t max_elems = vext_max_elems(desc, log2_esz); uint32_t esz = 1 << log2_esz; @@ -345,23 +353,21 @@ vext_ldst_stride(void *vd, void *v0, target_ulong base, target_ulong stride, VSTART_CHECK_EARLY_EXIT(env, env->vl); - for (i = env->vstart; i < env->vl; env->vstart = ++i) { - k = 0; - while (k < nf) { - if (!vm && !vext_elem_mask(v0, i)) { - /* set masked-off elements to 1s */ - vext_set_elems_1s(vd, vma, (i + k * max_elems) * esz, - (i + k * max_elems + 1) * esz); - k++; - continue; + for (uint32_t i = env->vstart; i < env->vl; env->vstart = ++i) { + if (!vm && !vext_elem_mask(v0, i)) { + if (vma) { + vext_set_nf_elems_1s(vd, i, nf, esz, max_elems); } + continue; + } + + for (uint32_t k = 0; k < nf; ++k) { target_ulong addr = base + stride * i + (k << log2_esz); ldst_elem(env, adjust_addr(env, addr), i + k * max_elems, vd, ra); - k++; } } - env->vstart = 0; + env->vstart = 0; vext_set_tail_elems_1s(env->vl, vd, desc, nf, esz, max_elems); } @@ -632,7 +638,6 @@ vext_ldst_index(void *vd, void *v0, target_ulong base, vext_ldst_elem_fn_tlb *ldst_elem, uint32_t log2_esz, uintptr_t ra) { - uint32_t i, k; uint32_t nf = vext_nf(desc); uint32_t vm = vext_vm(desc); uint32_t max_elems = vext_max_elems(desc, log2_esz); @@ -641,24 +646,21 @@ vext_ldst_index(void *vd, void *v0, target_ulong base, VSTART_CHECK_EARLY_EXIT(env, env->vl); - /* load bytes from guest memory */ - for (i = env->vstart; i < env->vl; env->vstart = ++i) { - k = 0; - while (k < nf) { - if (!vm && !vext_elem_mask(v0, i)) { - /* set masked-off elements to 1s */ - vext_set_elems_1s(vd, vma, (i + k * max_elems) * esz, - (i + k * max_elems + 1) * esz); - k++; - continue; + for (uint32_t i = env->vstart; i < env->vl; env->vstart = ++i) { + if (!vm && !vext_elem_mask(v0, i)) { + if (vma) { + vext_set_nf_elems_1s(vd, i, nf, esz, max_elems); } + continue; + } + + for (uint32_t k = 0; k < nf; ++k) { abi_ptr addr = get_index_addr(base, i, vs2) + (k << log2_esz); ldst_elem(env, adjust_addr(env, addr), i + k * max_elems, vd, ra); - k++; } } - env->vstart = 0; + env->vstart = 0; vext_set_tail_elems_1s(env->vl, vd, desc, nf, esz, max_elems); } @@ -824,25 +826,22 @@ ProbeSuccess: } } else { for (i = env->vstart; i < env->vl; i++) { - k = 0; - while (k < nf) { - if (!vext_elem_mask(v0, i)) { - /* set masked-off elements to 1s */ - vext_set_elems_1s(vd, vma, (i + k * max_elems) * esz, - (i + k * max_elems + 1) * esz); - k++; - continue; + if (!vext_elem_mask(v0, i)) { + if (vma) { + vext_set_nf_elems_1s(vd, i, nf, esz, max_elems); } + continue; + } + for (k = 0; k < nf; ++k) { addr = base + ((i * nf + k) << log2_esz); ldst_tlb(env, adjust_addr(env, addr), i + k * max_elems, vd, ra); - k++; } } } } - env->vstart = 0; + env->vstart = 0; vext_set_tail_elems_1s(env->vl, vd, desc, nf, esz, max_elems); } -- 2.43.0
