RFR(XXS): 8007312: null check signal semaphore in os::signal_notify windows

[David Holmes]

On 2/02/2013 12:49 AM, Markus Grönlund wrote:
> Thanks David,
>
> Ok, so based on our discussions, we'll narrow this to only do a null check on the signal semaphore handle on Windows.
>
> I have updated the webrev, here:
> http://cr.openjdk.java.net/~mgronlun/8007312/webrev02/

Looks okay to me.

I'll try to test out what happens on Solaris/Linux if you manage to 
sneak in a very early ctrl-C.

Thanks,
David

>
> Also I changed the bug description to reflect the updated information (to the updated subject in the mail above):
> http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=8007312
>
>
> With this change we don't trap on the assert when pressing ctrl-c if the hotspot signal handler is not setup.
>
> Thanks a lot for your help David, have a great weekend.
>
> Cheers
> Markus
>
>
>
> -----Original Message-----
> From: David Holmes
> Sent: den 1 februari 2013 14:41
> To: Markus Grönlund
> Cc: hotspot-dev at openjdk.java.net
> Subject: Re: RFR(XXS): 8007312: Signal Dispatcher thread to start and register ctrl-break handler before TRACE_INITIALIZE
>
> Which brings us back to simply checking for not being initialized.
>
> David
>
>>>>>
>>>>>
>>>>>
>>>>> (windows analysis)
>>>>>
>>>>>
>>>>>
>>>>> Pressing ctrl-c before Hotspot signal/console handler has been registered actually asserts/stops the VM (which to the user appears like a crash) on non-product builds.
>>>>>
>>>>>
>>>>>
>>>>> Before Hotspot registers its own jvm!consoleHandler with kernel32!CtrlRoutine, the C runtime default msvcr100!ctrlevent_capture is implicitly used - this calls back into jvm!UserHandler, which forwards into os::signal_notify() which uses uninitialized variables.
>>>>>
>>>>>
>>>>>
>>>>> // on Windows this creates the following issue when closing a NULL
>>>>> handle to a  semaphore
>>>>>
>>>>>
>>>>>
>>>>> void os::signal_notify(int signal_number) {
>>>>>
>>>>>        BOOL ret;
>>>>>
>>>>>
>>>>>        Atomic::inc(&pending_signals[signal_number]);
>>>>>
>>>>>        ret = ::ReleaseSemaphore(sig_sem, 1, NULL);<<--- call
>>>>> ReleaseSemaphore on global handle sig_sem which has not been
>>>>> setup/created yet == is NULL (is created in os::signal_init_pd())
>>>>>
>>>>>        assert(ret != 0, "ReleaseSemaphore() failed");<<-- assert
>>>>> traps here (GetLastError() == 0xc0000008 - An invalid HANDLE was
>>>>> specified)
>>>>>
>>>>>
>>>>>
>>>>> }
>>>>
>>>>>
>>>>>
>>>>> Thanks
>>>>>
>>>>> Markus