Hi Sebastian,
thanks for reporting this. Is is OK to ask a few questions as I've never
seen this before on a wide range of kit that I test this on.
1. Is this failure repeatable? (I'm not sure how it occurs since there
is a SIGSEGV handler for these cases).
2. Does it fail on specific machines?
3. What CPU model is the machine it fails on?
I've tried to understand this failure, but so far I'm quite perplexed by
it, so maybe it's a CPU specific caching behavior that I have misunderstood.
Any assistance with the questions above would be most useful,
Regards,
Colin
On 13/12/2021 21:48, Sebastian Ramacher wrote:
Source: stress-ng
Version: 0.13.08-1
Severity: important
X-Debbugs-Cc: [email protected]
The autopkgtests of stress-ng fail from time to time:
| stress-ng: 20:49:12.58 info: [1091] unsuccessful run completed in 0.51s
| stress-ng: 20:49:12.58 fail: [1091] cache instance 3 corrupted bogo-ops
counter, 2 vs 0
| stress-ng: 20:49:12.58 fail: [1091] cache instance 3 hash error in bogo-ops
counter and run flag, 796547380 vs 0
| stress-ng: 20:49:12.58 fail: [1091] metrics-check: stressor metrics
corrupted, data is compromised
| cache FAILED
...
| zlib PASSED
| 42 PASSED
| 1 FAILED, cache
| 0 SKIPPED
See
https://ci.debian.net/data/autopkgtest/testing/amd64/s/stress-ng/17550292/log.gz
for a recent failure
Cheers
