Hi Naoto, Good point. Here’s an updated version: http://cr.openjdk.java.net/~bpb/8249703/webrev.01/. The number of blocks is not available via a dedicated API call and so has to be calculated.
Thanks, Brian > On Aug 7, 2020, at 9:58 AM, naoto.s...@oracle.com wrote: > > Looks fine, Brian. > > I might add the condition if the block number is odd. > > Naoto > > On 8/7/20 8:33 AM, Brian Burkhalter wrote: >>> On Jul 31, 2020, at 10:42 AM, Brian Burkhalter >>> <brian.burkhal...@oracle.com> wrote: >>> >>> On macOS, the number of 1024 byte blocks appears to be incorrectly >>> calculated by 'df' using integer division by 2 of the number of 512 byte >>> blocks, resulting in a size smaller than the actual value when the number >>> of blocks is odd. >>> >>> Possibly instead of having the IS_MAC and IS_WIN constants it should use >>> the equivalent methods in jdk.test.lib.Platform? >>> >>> Thanks, >>> >>> Brian >>> >>> https://bugs.openjdk.java.net/browse/JDK-8249703 >>> <https://bugs.openjdk.java.net/browse/JDK-8249703> >>> http://cr.openjdk.java.net/~bpb/8249703/webrev.00/ >>> <http://cr.openjdk.java.net/~bpb/8249703/webrev.00/>
