On 26/02/2015 11:35, Paolo Bonzini wrote:
>
>
> On 25/02/2015 18:16, Leon Alrae wrote:
>> Hi,
>>
>> On 16/02/2015 18:17, Paolo Bonzini wrote:
>>> This needs to go away sooner or later, but one complication is the
>>> complex VFIO data structures that are modified in instance_finalize.
>>> Take a shortcut for now.
>>
>> It looks like this commit introduced a race condition. QEMU segfaults
>> very early in one of my tests where I boot Linux, and on one of my
>> machines it's easier to reproduce than on the others.
>
> This should be enough to fix it. If it works for you I'll send a pull
> request as soon as possible.
>
> diff --git a/cpus.c b/cpus.c
> index 1cd9867..7bed67b 100644
> --- a/cpus.c
> +++ b/cpus.c
> @@ -1115,7 +1115,7 @@ bool qemu_in_vcpu_thread(void)
>
> void qemu_mutex_lock_iothread(void)
> {
> - if (!tcg_enabled()) {
> + if (!tcg_enabled() || !first_cpu) {
> qemu_mutex_lock(&qemu_global_mutex);
> } else {
> iothread_requesting_mutex = true;
It works for me -- it fixes the crash, thanks.
FYI I hit some other problem where QEMU stops responding (deadlock?)
during a stress test quite randomly. The issue is most likely introduced
by this commit (and above fix doesn't seem to be related). I'm going to
look into that.
Leon
