On 2026-09-13 6:00, Richard Henderson wrote:
On 9/2/26 05:20, Philippe Mathieu-Daudé wrote:
LoongArch, M68K, MicroBlaze, and OpenRISC interrupt handlers
don't require external BQL protection, so move them from
do_interrupt_locked() to do_interrupt().
Signed-off-by: Philippe Mathieu-Daudé <[email protected]>
---
target/loongarch/tcg/tcg_cpu.c | 2 +-
target/m68k/cpu.c | 2 +-
target/microblaze/cpu.c | 2 +-
target/or1k/cpu.c | 2 +-
4 files changed, 4 insertions(+), 4 deletions(-)
How did you determine that these do not require locking?
E.g. why does alpha require locking when loongarch does not?
I couldn't find explicit BQL use in these, but now I see why this
poor audit isn't good enough. Better to convert all targets to use
explicit BQL locking (next patch), then remove unnecessary locking
one target at a time, with good justification in commit description.
