RFR: 8226816: add UserHandler calls to event log

[Baesken, Matthias]

Hi David,  thanks for looking into this .

> 
> If you add this then we don't need distinct POSIX and non-POSIX versions
> - the existing os::Posix::get_signal_name etc could all be hoisted into
> os.cpp and the os class - no?
>

Should I go for this ?
The coding is still a little different   (e.g. is_valid_signal (.. ) call  in os_posix )  but I think it could be done without much trouble (maybe with a few small ifdefs ) .


> That's a LinuxThreads anachronism which has been copied, unnecessarily
> into the other OS implementations. I will file a RFE to get rid of it.

Good catch !

Best regards, Matthias

> 
> Hi Matthias,
> 
> On 27/06/2019 6:56 pm, Baesken, Matthias wrote:
> > Hello, please review  the following  small  patch .
> > It adds  event logging to  the UserHandler  (user signal handler) calls .
> 
> That seems reasonable.
> 
> > (additionally  it adds   a function   os::win32::get_signal_name
> > to get  signal names for signal numbers ; this is similar to what we already
> had for posix ).
> 
> If you add this then we don't need distinct POSIX and non-POSIX versions
> - the existing os::Posix::get_signal_name etc could all be hoisted into
> os.cpp and the os class - no?
> 
> Aside: I spotted this in UserHandler:
> 
>      // 4511530 - sem_post is serialized and handled by the manager
> thread. When
>      // the program is interrupted by Ctrl-C, SIGINT is sent to every
> thread. We
>      // don't want to flood the manager thread with sem_post requests.
>      if (sig == SIGINT && Atomic::add(1, &sigint_count) > 1)
>        return;
> 
> That's a LinuxThreads anachronism which has been copied, unnecessarily
> into the other OS implementations. I will file a RFE to get rid of it.
> 
> Thanks,
> David
> 
> >
> >
> > Bug/webrev :
> >
> > https://bugs.openjdk.java.net/browse/JDK-8226816
> >
> > http://cr.openjdk.java.net/~mbaesken/webrevs/8226816.0/
> >
> > Thanks, Matthias
> >