suspend/gpu_lockup_after_suspend would fail with previous GPU lockup event

[Po-Hsu Lin]

Found this issue on recent SRU, on 201208-11587.
If GPU lockup occurred once before, this test would fail even you re-test it.

The reason for this is the parser we use, which will find every gpu hung pattern, no matter when did it happened.

def check_gpu(log=None):
    if not log:
        log = '/var/log/kern.log'
    with open(log, 'rb') as f:
        if re.findall(r'gpu\s+hung', str(f.read()), flags=re.I):
            print("GPU hung Detected")
            return 1

You could run the gpu_test script with the attached log to see what would happen