[9] RFR of 8080589: (fs) FileChannel.force should use fcntl(F_FULLFSYNC) instead of fsync on OS X
Brian Burkhalter
brian.burkhalter at oracle.com
Tue May 19 14:07:14 UTC 2015
On May 18, 2015, at 7:14 PM, Robert Muir <rcmuir at gmail.com> wrote:
> On Mon, May 18, 2015 at 9:24 PM, Brian Burkhalter
> <brian.burkhalter at oracle.com> wrote:
>>
>> +#ifdef MACOSX
>> + result = fcntl(fd, F_FULLFSYNC);
>> +#else
>
> According to fcntl(2) on OS X, this may not be implemented for all filesystems.
>
> Should code handle failure/ENOTSUP as suggested here?
> https://community.oracle.com/thread/453477
> I see similar code in other open-source databases (but typically not
> checking errno for anything specific).
Thanks for pointing this out. I will look into it further and update the patch as needed.
Brian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/nio-dev/attachments/20150519/ffbdea0c/attachment.html>
More information about the nio-dev
mailing list