On 5/24/25 7:40 AM, Richard Henderson wrote:
When we moved TLB_MMIO and TLB_DISCARD_WRITE to TLB_SLOW_FLAGS_MASK, we failed to update atomic_mmu_lookup to properly reconstruct flags.Fixes: 24b5e0fdb543 ("include/exec: Move TLB_MMIO, TLB_DISCARD_WRITE to slow flags") Reported-by: Jonathan Cameron <jonathan.came...@huawei.com> Signed-off-by: Richard Henderson <richard.hender...@linaro.org> --- accel/tcg/cputlb.c | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c index 5f6d7c601c..86d0deb08c 100644 --- a/accel/tcg/cputlb.c +++ b/accel/tcg/cputlb.c @@ -1871,8 +1871,12 @@ static void *atomic_mmu_lookup(CPUState *cpu, vaddr addr, MemOpIdx oi, goto stop_the_world; }- /* Collect tlb flags for read. */+ /* Finish collecting tlb flags for both read and write. */ + full = &cpu->neg.tlb.d[mmu_idx].fulltlb[index]; tlb_addr |= tlbe->addr_read; + tlb_addr &= TLB_FLAGS_MASK & ~TLB_FORCE_SLOW; + tlb_addr |= full->slow_flags[MMU_DATA_STORE]; + tlb_addr |= full->slow_flags[MMU_DATA_LOAD];
[...] Looks good to me. Reviewed-by: Pierrick Bouvier <pierrick.bouv...@linaro.org>
