The GFileInfo modification time test assumed that the difference between
a modification timestamp in seconds and in microseconds must be greater
than 0.
Mathmatically, there's a one-in-a-million chance that it will be 0. It
turns out that one-in-a-million chances happen approximately once every
fortnight on the autobuilder.
[ YOCTO 14373 ]
(From OE-Core rev: 7d2bd897fdbcf5215bf2287cb96eb6b58803f08c)
Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
486a35a674