Bug#800384: arrayfire: random test failures

[Edmund Grimley Evans]

Source: arrayfire
Version: 3.1.2+dfsg1-1

There have been 3 attempts to build this version on the arm64 buildds
between 12:00 and 16:00 today:

https://buildd.debian.org/status/logs.php?pkg=arrayfire&arch=arm64

It was a different machine each time, but each time one of the tests
failed with a segfault, and each time it was a different test:

Attempt 1:
	 15 - Test_sort_cpu (SEGFAULT)

Attempt 2:
	 36 - Test_memory_lock_cpu (SEGFAULT)

Attempt 3:
	 37 - Test_memory_cpu (SEGFAULT)

If these failures are as random as they seem then it would probably
build eventually, with enough tries, but perhaps you want to
investigate this, or suggest how it could be investigated.

The same thing may be happening on some other architectures:

https://buildd.debian.org/status/package.php?p=arrayfire&suite=sid