On 28/5/26 21:36, Pierrick Bouvier wrote:
> On 5/27/2026 1:13 PM, Philippe Mathieu-Daudé wrote:
>> v11:
>> - rewrite some patches while reviewing for mainstream integration
>>
>> v10 cover from Djordje:
>>
>> Hi all,
>>
>> This series adds big-endian data support for RISC-V system emulation.
>>
>> Following the discussion on previous versions, this series models
>> fixed-endian harts, not a mixed-endian implementation where software can
>> toggle MBE/SBE/UBE at runtime. The reset value of MBE/SBE/UBE is selected
>> from the CPU configuration. These bits remain outside the
>> mstatus/mstatush/sstatus writable masks, so they are effectively hardwired
>> to 0 for the default little-endian CPUs and to 1 when `big-endian=on`.
>>
>> The user-visible `big-endian` CPU property is exposed only after the data
>> path, boot path, and page-table walk support are in place, so the series
>> remains bisectable.
>>
>> Djordje Todorovic (6):
>> target/riscv: Implement runtime data endianness via MSTATUS bits
>> target/riscv: Fix page table walk endianness for big-endian harts
>> target/riscv: Add big-endian CPU configuration field and reset logic
>> hw/riscv/boot: Honour data endianness
>> target/riscv: Expose and document the CPU 'big-endian' property
>> tests/functional: Add RISC-V endianness test
>>
>> Philippe Mathieu-Daudé (6):
>> target/riscv: Initialize DisasContext::mo_endian once
>> target/riscv: Assert PTE.N/PBMT bits are not expected on RV32
>> target/riscv: De-indent some code in get_physical_address()
>> target/riscv: Remove target_ulong use in get_physical_address()
>> hw/riscv/boot: Rewrite setup_rom_reset_vec() using load/store API
>> hw/riscv/boot: Replace cpu_to_le32() -> const_le32()
>>
>> docs/system/target-riscv.rst | 29 ++++++
>> target/riscv/cpu.h | 1 +
>> target/riscv/internals.h | 29 ++++--
>> target/riscv/cpu_cfg_fields.h.inc | 1 +
>> hw/riscv/boot.c | 93 ++++++++++---------
>> target/riscv/cpu.c | 15 ++-
>> target/riscv/cpu_helper.c | 44 +++++----
>> target/riscv/tcg/tcg-cpu.c | 2 +
>> target/riscv/translate.c | 22 ++---
>> target/riscv/insn_trans/trans_rva.c.inc | 4 +-
>> target/riscv/insn_trans/trans_rvd.c.inc | 4 +-
>> target/riscv/insn_trans/trans_rvf.c.inc | 4 +-
>> target/riscv/insn_trans/trans_rvi.c.inc | 8 +-
>> target/riscv/insn_trans/trans_rvzacas.c.inc | 4 +-
>> target/riscv/insn_trans/trans_rvzalasr.c.inc | 4 +-
>> target/riscv/insn_trans/trans_rvzce.c.inc | 4 +-
>> target/riscv/insn_trans/trans_rvzfh.c.inc | 4 +-
>> target/riscv/insn_trans/trans_rvzicfiss.c.inc | 4 +-
>> target/riscv/insn_trans/trans_xmips.c.inc | 8 +-
>> target/riscv/insn_trans/trans_xthead.c.inc | 16 ++--
>> target/riscv/insn_trans/trans_zilsd.c.inc | 4 +-
>> tests/functional/riscv64/meson.build | 1 +
>> tests/functional/riscv64/test_endianness.py | 57 ++++++++++++
>> 23 files changed, 240 insertions(+), 122 deletions(-)
>> create mode 100644 tests/functional/riscv64/test_endianness.py
>>
>
> Seems there are some build failures:
> https://github.com/p-b-o/qemu-ci/actions/runs/26537039673
IIUC this branch was applied /before/ commit c2444a426ec ("Merge
tag 'single-binary-20260527'). Now it depends on Alistair riscv
queue so I need to wait before respinning.
>
> Regards,
> Pierrick