On Wed, Mar 04, 2026 at 12:09:58PM +0800, Jay Chang wrote:
> Reset irq_overflow_left to 0 before setting up a new timer. Without
> this fix, a stale irq_overflow_left value from a previous timer setup
> could cause incorrect timer behavior.
> 
> Signed-off-by: Jay Chang <[email protected]>
> Reviewed-by: Frank Chang <[email protected]>
> ---
>  hw/riscv/riscv-iommu-hpm.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/hw/riscv/riscv-iommu-hpm.c b/hw/riscv/riscv-iommu-hpm.c
> index c5034bff79..e8d284ac8b 100644
> --- a/hw/riscv/riscv-iommu-hpm.c
> +++ b/hw/riscv/riscv-iommu-hpm.c
> @@ -228,6 +228,7 @@ static void hpm_setup_timer(RISCVIOMMUState *s, uint64_t 
> value)
>      }
>  
>      overflow_at = (uint64_t)qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + 
> overflow_ns;
> +    s->irq_overflow_left = 0;
>  
Looks good. Without this reset, if a previous call set
irq_overflow_left (overflow_at > INT64_MAX) and the current
call does NOT overflow, the stale value persists and the
two-phase timer fires incorrectly.

Reviewed-by: Chao Liu <[email protected]>

Best regards,
Chao Liu
>      if (overflow_at > INT64_MAX) {
>          s->irq_overflow_left = overflow_at - INT64_MAX;
> -- 
> 2.48.1
> 

Reply via email to