在 2022/5/12 下午4:55, ~eopxd 写道:
According to v-spec, mask agnostic behavior can be either kept as undisturbed or set elements' bits to all 1s. To distinguish the difference of mask policies, QEMU should be able to simulate the mask agnostic behavior as "set mask elements' bits to all 1s". There are multiple possibility for agnostic elements according to v-spec. The main intent of this patch-set tries to add option that can distinguish between mask policies. Setting agnostic elements to all 1s allows QEMU to express this. The following instructions that are always unmasked and not affected: - Vector add-with-carry and subtract-with-borrow instructions - Vector merge and move instructions - Vector reduction instructions - Vector mask-register logical instructions - `vcompress` This patch set is based on v17 of patch set "Add tail agnostic behavior for rvv instructions". Based on: <165234397852.32492.120314973852405009...@git.sr.ht> v2 updates: - Rebase upon changes of the tail agnostic patch-set - Minor change for vector load/store instructions v3 updates: - Rebase upon changes of the tail agnostic patch-set - Fix coding style, add missing space - Trigger `vma` when encountering vector load instructions and not in vector stores
Instead of add is_load in the helper functions, maybe it's better to just not set vta and vma for
vector store instructions in trans_rvv. Regards, Weiwei Li
Yueh-Ting (eop) Chen (9): target/riscv: rvv: Add mask agnostic for vv instructions target/riscv: rvv: Add mask agnostic for vector load / store instructions target/riscv: rvv: Add mask agnostic for vx instructions target/riscv: rvv: Add mask agnostic for vector integer shift instructions target/riscv: rvv: Add mask agnostic for vector integer comparison instructions target/riscv: rvv: Add mask agnostic for vector fix-point arithmetic instructions target/riscv: rvv: Add mask agnostic for vector floating-point instructions target/riscv: rvv: Add mask agnostic for vector mask instructions target/riscv: rvv: Add mask agnostic for vector permutation instructions eopXD (1): target/riscv: rvv: Add option 'rvv_ma_all_1s' to enable optional mask agnostic behavior target/riscv/cpu.c | 1 + target/riscv/cpu.h | 2 + target/riscv/cpu_helper.c | 2 + target/riscv/insn_trans/trans_rvv.c.inc | 32 +++++ target/riscv/internals.h | 5 +- target/riscv/translate.c | 2 + target/riscv/vector_helper.c | 152 ++++++++++++++++++++---- 7 files changed, 171 insertions(+), 25 deletions(-)