Request for review 7191817: -XX:+UseSerialGC -XX:+UseLargePages crashes with SIGFPE on Mac OS X

David Holmes david.holmes at oracle.com
Wed Oct 24 20:39:16 PDT 2012


Hi Harold,

My only query is whether _ALLBSD_SOURCE is the right conditional for 
this? Exactly which platform(s) don't support this?

The reason I ask is that even the BSD sources (eg 
./os/bsd/vm/os_bsd.hpp) contain ifdefs for _ALL_BSD_SOURCE which 
indicates that some bsd builds do not in fact set _ALLBSD_SOURCE ! Which 
I find extremely confusing.

Thanks,
David

On 25/10/2012 12:22 AM, harold seigel wrote:
> Please review this updated webrev:
> http://cr.openjdk.java.net/~coleenp/bug_7191817_2/
> <http://cr.openjdk.java.net/%7Ecoleenp/bug_7191817_2/>
>
> It incorporates Dave's suggestion to use the UNSUPPORTED_OPTION macro.
>
> Thanks, Harold
>
> On 10/24/2012 5:46 AM, David Holmes wrote:
>> Hi Harold,
>>
>> We already have a macro for this:
>>
>> // Disable options not supported in this release, with a warning if they
>> // were explicitly requested on the command-line
>> #define UNSUPPORTED_OPTION(opt, description) \
>> do { \
>> if (opt) { \
>> if (FLAG_IS_CMDLINE(opt)) { \
>> warning(description " is disabled in this release."); \
>> } \
>> FLAG_SET_DEFAULT(opt, false); \
>> } \
>> } while(0)
>>
>> We use this on platforms were not all features are supported.
>>
>> David
>>
>> On 24/10/2012 5:04 AM, harold seigel wrote:
>>> Summary: Support for the -XX:+UseLargePages flag causes crashes on BSD
>>> platforms such as MacOS X. This proposed change adds code to print out a
>>> message if -XX:+UseLargePages is specified on this platform and to then
>>> ignore the flag.
>>>
>>> Open webrev at http://cr.openjdk.java.net/~coleenp/bug_7191817
>>> <http://cr.openjdk.java.net/%7Ecoleenp/bug_7191817>
>>>
>>> Bug link at http://bugs.sun.com/view_bug.do?bug_id=7191817
>>>
>>> This change was tested with JPRT and JCK tests. Also, tests were run by
>>> hand on a MacOS X machine specifying the -XX:+UseLargePages flag to make
>>> sure that the message was output and no crashes occurred.
>>>
>>> Thanks, Harold


More information about the hotspot-runtime-dev mailing list