=?utf-8?Q?Re[2]=3A_[Gc]_Test_failures, _hangs, _crashes_with_2011-04-25_CVS_checkout?=

Ivan Maidanski ivmai at mail.ru
Wed May 11 13:59:45 PDT 2011


Hello Peter,

Could find out why is one of the threads blocked in lock() syscall (instead of sigsuspend)?
It could be any of the following reasons:
- SIG_SUSPEND was not delivered to that thread (eg. because it is not in the set of allowed signals);
- SIG_SUSPEND was not send by the thread that is doing "stop-the-world";
- GC_suspend_handler was invoked in that thread but was existed without waiting while GC_world_is_stopped.
regards.

Tue, 10 May 2011 20:26:29 -0500 "Peter O'Gorman" <gc at mlists.thewrittenword.com>:
> On Mon, May 09, 2011 at 12:17:39AM +0400, Ivan Maidanski wrote:
> > Hi Peter,
> > 
> 
> Hi Ivan,
> 
> Thanks for looking into this.
> 
> > It seems one suspend signal is lost. Try with environment variable
> GC_RETRY_SIGNALS=1.
> > If this solves the problem then we should turn on retry mode for these
> targets by default.
> 
> It didn't solve the problem, it still hung.
> 
> Peter
> -- 
> Peter O'Gorman
> pogma at thewrittenword.com



More information about the Gc mailing list