Historically, M-profile helper functions in m_helper.c and mve_helper.c used the unaligned cpu_*_data_ra() routines to perform guest memory accesses. This meant we had no way to enforce alignment constraints when executing helper-based loads/stores. With the addition of the cpu_*_mmu() APIs, we can now combine the current MMU state with MO_ALIGN flags to build a MemOpIdx that enforces alignment at the helper level.
This patch series: - Replaces all calls to cpu_ld*_data_ra(), cpu_st*_data_ra() in the M-profile helpers (m_helper.c) and the MVE helpers (mve_helper.c) with their cpu_*_mmu() equivalents. - Leaves SME and SVE helper code untouched, as those extensions support unaligned accesses by design. With this change, all M-profile and MVE helper-based loads and stores will now correctly honor their alignment requirements. Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1154 Signed-off-by: William Kosasih <kosasihwilli...@gmail.com> William Kosasih (11): target/arm: Bring VLSTM/VLLDM helper store/load closer to the ARM pseudocode target/arm: Fix BLXNS helper store alignment checks target/arm: Fix function_return helper load alignment checks target/arm: Fix VLDR helper load alignment checks target/arm: Fix VSTR helper store alignment checks target/arm: Fix VLDR_SG helper load alignment checks target/arm: Fix VSTR_SG helper store alignment checks target/arm: Fix VLD4 helper load alignment checks target/arm: Fix VLD2 helper load alignment checks target/arm: Fix VST4 helper store alignment checks target/arm: Fix VST2 helper store alignment checks target/arm/tcg/m_helper.c | 33 +++++--- target/arm/tcg/mve_helper.c | 162 ++++++++++++++++++++++-------------- 2 files changed, 122 insertions(+), 73 deletions(-) -- 2.48.1
