Request for review (M): 7132070: Use a mach_port_t as the OSThread thread_id rather than pthread_t on BSD/OSX

[Paul Hohensee]

Thanks!

Paul

On 2/15/12 2:25 PM, Staffan Larsen wrote:
> On 15 feb 2012, at 20:03, Paul Hohensee wrote:
>
>> On 2/15/12 1:58 PM, Staffan Larsen wrote:
>>> On 15 feb 2012, at 19:49, Paul Hohensee wrote:
>>>
>>>> Also,
>>>>
>>>> You can move the #include of mach.h into globalDefinitions_gcc.hpp,
>>>> in the #ifdef __APPLE__ section.  That way, it won't cause a compile-
>>>> time error on non-osx platforms that don't have it.
>>> I dislike adding more stuff to globalDefinitions since that gets pulled in all over the place. I'd prefer to have #includes only where they are needed.
>> globalDefinitions_gcc.hpp is where these sorts of #includes currently live,
>> so I'd go with present practice rather than doing something new.  If
>> you want to put the include of mach.h in os_bsd.inline.hpp, then you
>> should do something similar with all the platform-dependent includes
>> in globalDefinitions_gcc.hpp.
> I'd actually like to have it in osThread_bsd.hpp, but that isn't possible. I'll put in gloabalDefinitions_gcc.hpp.
>
>
>>>> The declaration of _thread_id, thread_id() and set_thread_id() in os_bsd.hpp
>>>> can be put under a #ifdef __APPLE__, vis.,
>>>>
>>>> #ifdef _ALLBSD_SOURCE
>>>> #ifdef __APPLE__
>>>>   thread_t _thread_id;
>>>> #else
>>>>   pthread_t _thread_id;
>>>> #endif
>>>> #endif
>>>>
>>>> The uses of mach_thread_self() in os_bsd.cpp can be similarly conditioned.
>>>>
>>>> If there's a thread_t defined by non-osx bsd implementations, then
>>>> you don't need predication in os_bsd.hpp, though you'd still need it
>>>> in os_bsd.cpp.  I'm assuming there's no mach_thread_self() on non-osx
>>>> bsd platforms.
>>> I could do that, but the question is how much work we should put into not trying to break something that we can't test anyway. There's no way, even if I do this that I can verify that it works.
>> Even so, if we ever need to make hotspot work on non-osx bsd platforms, we'll
>> at least have markers to guide the work.
> OK - I'll try to fix things up.
>
> /Staffan
>
>> Paul
>>
>>> /Staffan
>>>
>>>> Thanks,
>>>>
>>>> Paul
>>>>
>>>> On 2/15/12 1:33 PM, Paul Hohensee wrote:
>>>>> Imo we should at least try to make non-osx bsd builds work, since
>>>>> the original code did work for non-osx builds.  This change doesn't
>>>>> do that.
>>>>>
>>>>> In globalDefinitions_gcc.hpp, if you keep the lines
>>>>>
>>>>> #undef ELF_AC
>>>>> #undef EFL_ID
>>>>>
>>>>> then you don't have to change vm_version_x86.hpp.
>>>>>
>>>>> Paul
>>>>>
>>>>> On 2/15/12 10:16 AM, Daniel D. Daugherty wrote:
>>>>>> The _ALLBSD_SOURCE symbol is defined by the HotSpot Makefile infrastructure.
>>>>>> It is used to identify code specific to the BSD family of OSes.
>>>>>> The __APPLE__ symbol is defined by the Apple compiler(s) and it is used to
>>>>>> identify code specific to MacOS X.
>>>>>>
>>>>>> Typically you'll see something like:
>>>>>>
>>>>>> #ifdef _ALLBSD_SOURCE
>>>>>>
>>>>>> <code that works on all BSDs>
>>>>>>
>>>>>> #ifdef __APPLE__
>>>>>> <code specific to MacOS X>
>>>>>> #else
>>>>>> <code for other BSDs>
>>>>>> #endif // __APPLE__
>>>>>> #endif // _ALLBSD_SOURCE
>>>>>>
>>>>>> As for building on non-MacOS X BSDs, that would be nice, but we
>>>>>> don't have the infrastructure to do it.
>>>>>>
>>>>>> Dan
>>>>>>
>>>>>> On 2/15/12 6:57 AM, Mikael Gerdin wrote:
>>>>>>> Hi Staffan,
>>>>>>>
>>>>>>> It looks like you're adding Mac-specific stuff like thread_t and calls to ::mach_thread_self() inside _ALLBSD_SOURCE #ifdefs, are you sure this won't break BSD builds?
>>>>>>> Does the OSX compiler define _ALLBSD_SOURCE or is that for (free|net|open)bsd?
>>>>>>> It's too bad we don't do regular builds on any of the BSDs, otherwise this would have been easier to figure out.
>>>>>>>
>>>>>>> /Mikael
>>>>>>>
>>>>>>>
>>>>>>> On 2012-02-15 11:29, Staffan Larsen wrote:
>>>>>>>> Please review the following change:
>>>>>>>>
>>>>>>>> Bug: http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=7132070
>>>>>>>>
>>>>>>>> Webrev: http://cr.openjdk.java.net/~sla/7132070/webrev.00/
>>>>>>>>
>>>>>>>> This changes the value returned by OSThread::thread_id() and
>>>>>>>> os::current_thread_id() on macosx to return the mach thread_t instead of
>>>>>>>> pthread_t. There is a separate method OSThread:pthread_id() that returns
>>>>>>>> the pthread_t.
>>>>>>>>
>>>>>>>> The reason for this change is both that JFR would like a 4 byte value
>>>>>>>> for thread id, and that SA requires access to the thread_t.
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>> /Staffan