[PATCH] dix: Properly detect if the other device is frozen

[Peter Hutterer]

On Tue, Oct 27, 2009 at 06:03:08PM -0700, Jeremy Huddleston wrote:
> othersFrozen was getting unset.  We should detect if it should be set
> to true rather than setting it false for a device that isn't frozen
> 
> Signed-off-by: Jeremy Huddleston <jeremyhu at freedesktop.org>
> ---
>  dix/events.c |    8 +++-----
>  1 files changed, 3 insertions(+), 5 deletions(-)
> 
> diff --git a/dix/events.c b/dix/events.c
> index d070a4c..015c2b1 100644
> --- a/dix/events.c
> +++ b/dix/events.c
> @@ -1673,7 +1673,7 @@ AllowSome(ClientPtr client,
>      thisGrabbed = grabinfo->grab && SameClient(grabinfo->grab,
> client);
>      thisSynced = FALSE;
>      otherGrabbed = FALSE;
> -    othersFrozen = TRUE;
> +    othersFrozen = FALSE;
>      grabTime = grabinfo->grabTime;
>      for (dev = inputInfo.devices; dev; dev = dev->next)
>      {
> @@ -1689,11 +1689,9 @@ AllowSome(ClientPtr client,
>  	    otherGrabbed = TRUE;
>  	    if (grabinfo->sync.other == devgrabinfo->grab)
>  		thisSynced = TRUE;
> -	    if (devgrabinfo->sync.state < FROZEN)
> -		othersFrozen = FALSE;
> +	    if (devgrabinfo->sync.state >= FROZEN)
> +		othersFrozen = TRUE;
>  	}
> -	else if (!devgrabinfo->sync.other || !SameClient(devgrabinfo->sync.other,
> client))
> -	    othersFrozen = FALSE;
>      }
>      if (!((thisGrabbed && grabinfo->sync.state >= FROZEN) ||
> thisSynced))
>  	return;
> -- 
> 1.6.5.1

I looked at this patch and I can't see any reason why that shouldn't
work. I'll leave it running on my box for a while to see if it does
anything bad, grab syncing is...complicated.

Cheers,
  Peter