On 2011-02-15 21:56, Marcelo Tosatti wrote:
> Note: to be applied to uq/master.
>
> In icount mode, halt emulation should take into account the nearest event
> when sleeping.
>
> Signed-off-by: Marcelo Tosatti <mtosa...@redhat.com>
> Reported-and-tested-by: "Edgar E. Iglesias" <edgar.igles...@gmail.com>
>
> diff --git a/cpus.c b/cpus.c
> index 468544c..21c3eba 100644
> --- a/cpus.c
> +++ b/cpus.c
> @@ -770,7 +770,7 @@ static void qemu_tcg_wait_io_event(void)
> CPUState *env;
>
> while (all_cpu_threads_idle()) {
> - qemu_cond_timedwait(tcg_halt_cond, &qemu_global_mutex, 1000);
> + qemu_cond_timedwait(tcg_halt_cond, &qemu_global_mutex,
> qemu_calculate_timeout());
> }
>
> qemu_mutex_unlock(&qemu_global_mutex);As we agree that timeouts are wrong here, let's check if Paolo's patches make a difference. If not, it would still be better to understand the actual issue. After trying too many voodoo patches myself, I've finally found that mutex ping pong between iothread and vcpus by starring at a trace in kernelshark. Jan
signature.asc
Description: OpenPGP digital signature
