RFR(S): JDK-8181503 "Can't compile hotspot with c++11"
coleen.phillimore at oracle.com
coleen.phillimore at oracle.com
Wed Jan 31 16:17:41 UTC 2018
On 1/31/18 10:35 AM, Gerard Ziemski wrote:
>> On Jan 31, 2018, at 1:46 AM, Thomas Stüfe <thomas.stuefe at gmail.com> wrote:
>>
>>
>>
>> On Wed, Jan 31, 2018 at 2:07 AM, David Holmes <david.holmes at oracle.com> wrote:
>> Hi Gerard,
>>
>> These seem okay.
>>
>> I'm a little surprised that the LogConfiguration change doesn't cause complaints about passing true/false as int ??
>>
>> And I second Coleen's query re the VMError codes - there seems to be no reason for their strange values. They were added by JDK-4965918.
>>
>>
>>
>> They look Windows-ish, like someone wanted them to mimic SEH codes - like someone took them directly from the examples here: https://msdn.microsoft.com/en-us/library/het71c37.aspx.
>>
>> They bothered us for some reason I do not really remember anymore, so in our port they long have had different values, with no adverse effects.
> Based on the comments from vmError.hpp:
>
> static int _id; // Solaris/Linux signals: 0 - SIGRTMAX
> // Windows exceptions: 0xCxxxxxxx system errors
> // 0x8xxxxxxx system warnings
>
> There seems to be ranges of errors on Windows that are used by the OS, so I’m guessing we chose a range that seemed free: https://msdn.microsoft.com/en-us/library/windows/desktop/ms681381(v=vs.85).aspx
Ah, so the field in vmError is shared with Windows error codes and unix
signals. I think your fix is fine. I'll answer Thomas's mail that
decoupling these would be a good idea, if I read his email right, but
not with this change.
Thanks,
Coleen
>
>
>> On Jan 30, 2018, at 4:53 PM, coleen.phillimore at oracle.com wrote:
>>
>>
>> Why are these defined with these odd values anyway? why not define as 0, 1 and 2? Or do they map to some Windows error values?
>>
>> thanks,
>> Coleen
> 0,1,2 seem to be taken by the OS, on Windows see: https://msdn.microsoft.com/en-us/library/windows/desktop/ms681382(v=vs.85).aspx, and Linux, BSD see: http://man7.org/linux/man-pages/man7/signal.7.html, so it seems we wanted unique error codes that could not be mistaken for system errors or signals.
>
>
> cheers
>
>
>
>
>
More information about the hotspot-runtime-dev
mailing list