Re: [Qemu-devel] [PATCH 1.7] timers: fix stop/cont with -icount

[Paolo Bonzini]

Il 28/10/2013 17:32, Paolo Bonzini ha scritto:
> Stop/cont commands are broken with -icount due to a deadlock.  The
> real problem is that the computation of timers_state.cpu_ticks_offset
> makes no sense with -icount enabled: we set it to an icount clock value
> in cpu_disable_ticks, and subtract a TSC (or similar, whatever
> cpu_get_real_ticks happens to return) value in cpu_enable_ticks.
> 
> The fix is simple.  timers_state.cpu_ticks_offset is only used
> together with cpu_get_real_ticks, so we can use cpu_get_real_ticks
> in cpu_disable_ticks.  There is no need to update cpu_ticks_prev
> at the time cpu_disable_ticks is called; instead, we can do it
> the next time cpu_get_ticks is called.
> 
> The change to cpu_disable_ticks is the important part of the patch.
> The rest modifies the code to always check timers_state.cpu_ticks_prev,
> even when the ticks are not advancing (i.e. the VM is stopped).  It also
> makes a similar change to cpu_get_clock_locked, so that the code remains
> similar for cpu_get_ticks and cpu_get_clock_locked.
> 
> Signed-off-by: Paolo Bonzini <address@hidden>
> ---
>  cpus.c | 42 ++++++++++++++++++++++--------------------
>  1 file changed, 22 insertions(+), 20 deletions(-)
> 
> diff --git a/cpus.c b/cpus.c
> index 398229e..c2c6864 100644
> --- a/cpus.c
> +++ b/cpus.c
> @@ -165,36 +165,38 @@ int64_t cpu_get_icount(void)
>  /* Caller must hold the BQL */
>  int64_t cpu_get_ticks(void)
>  {
> +    int64_t ticks;
> +
>      if (use_icount) {
>          return cpu_get_icount();
>      }
> -    if (!timers_state.cpu_ticks_enabled) {
> -        return timers_state.cpu_ticks_offset;
> -    } else {
> -        int64_t ticks;
> -        ticks = cpu_get_real_ticks();
> -        if (timers_state.cpu_ticks_prev > ticks) {
> -            /* Note: non increasing ticks may happen if the host uses
> -               software suspend */
> -            timers_state.cpu_ticks_offset += timers_state.cpu_ticks_prev - 
> ticks;
> -        }
> -        timers_state.cpu_ticks_prev = ticks;
> -        return ticks + timers_state.cpu_ticks_offset;
> +
> +    ticks = timers_state.cpu_ticks_offset;
> +    if (timers_state.cpu_ticks_enabled) {
> +        ticks += cpu_get_real_ticks();
> +    }
> +
> +    if (timers_state.cpu_ticks_prev > ticks) {
> +        /* Note: non increasing ticks may happen if the host uses
> +           software suspend */
> +        timers_state.cpu_ticks_offset += timers_state.cpu_ticks_prev - ticks;
> +        ticks = timers_state.cpu_ticks_prev;
>      }
> +
> +    timers_state.cpu_ticks_prev = ticks;
> +    return ticks;
>  }
>  
>  static int64_t cpu_get_clock_locked(void)
>  {
> -    int64_t ti;
> +    int64_t ticks;
>  
> -    if (!timers_state.cpu_ticks_enabled) {
> -        ti = timers_state.cpu_clock_offset;
> -    } else {
> -        ti = get_clock();
> -        ti += timers_state.cpu_clock_offset;
> +    ticks = timers_state.cpu_clock_offset;
> +    if (timers_state.cpu_ticks_enabled) {
> +        ticks += get_clock();
>      }
>  
> -    return ti;
> +    return ticks;
>  }
>  
>  /* return the host CPU monotonic timer and handle stop/restart */
> @@ -235,7 +237,7 @@ void cpu_disable_ticks(void)
>      /* Here, the really thing protected by seqlock is cpu_clock_offset. */
>      seqlock_write_lock(&timers_state.vm_clock_seqlock);
>      if (timers_state.cpu_ticks_enabled) {
> -        timers_state.cpu_ticks_offset = cpu_get_ticks();
> +        timers_state.cpu_ticks_offset += cpu_get_real_ticks();
>          timers_state.cpu_clock_offset = cpu_get_clock_locked();
>          timers_state.cpu_ticks_enabled = 0;
>      }
> 

Ping, did you miss this one?

Paolo