CI fuzz is failing

[hugovk]

CI fuzz is failing, there's an issue in the oss-fuzz project. I can't reproduce locally, so I'm not sure what's up.

Originally posted by @wiredfool in #5697 (comment)

https://github.com/python-pillow/Pillow/actions/workflows/cifuzz.yml

[hugovk]

Logs from https://github.com/python-pillow/Pillow/runs/3517459494?check_suite_focus=true:

...
2021-09-05 11:24:20,002 - root - INFO - Removing unaffected fuzz targets.
2021-09-05 11:24:20,003 - root - DEBUG - Diffing against base_commit: 98319efcc5460e4d9a2c14e83f8fcdf082754918.
2021-09-05 11:24:20,006 - root - INFO - Diffing against 98319efcc5460e4d9a2c14e83f8fcdf082754918.
2021-09-05 11:24:20,014 - root - INFO - Files changed in PR: ['src/libImaging/FliDecode.c']
2021-09-05 11:24:20,025 - urllib3.connectionpool - DEBUG - Starting new HTTPS connection (1): storage.googleapis.com:443
2021-09-05 11:24:20,096 - urllib3.connectionpool - DEBUG - https://storage.googleapis.com:443 "GET /oss-fuzz-coverage/latest_report_info/pillow.json HTTP/1.1" 403 223
2021-09-05 11:24:20,098 - root - ERROR - Loading json from url storage.googleapis.com/oss-fuzz-coverage/latest_report_info/pillow.json failed with: Expecting value: line 1 column 1 (char 0).
2021-09-05 11:24:20,098 - root - ERROR - Could not get the coverage report json from url: storage.googleapis.com/oss-fuzz-coverage/latest_report_info/pillow.json.
2021-09-05 11:24:20,099 - root - ERROR - Could not find latest coverage report.
2021-09-05 11:24:22,577 - root - ERROR - Build check failed.
Build check: stdout: INFO: performing bad build checks for /tmp/not-out/fuzz_font
INFO: performing bad build checks for /tmp/not-out/fuzz_pillow
Retrying failed fuzz targets sequentially 1
INFO: performing bad build checks for /tmp/not-out/fuzz_pillow
BAD BUILD: /tmp/not-out/fuzz_pillow does not seem to have coverage instrumentation.
INFO: Using preloaded libfuzzer
INFO: Running with entropic power schedule (0xFF, 100).
INFO: Seed: 1337
INFO: Loaded 2 modules   (9317 inline 8-bit counters): 9019 [0x7fec52995320, 0x7fec5299765b), 298 [0x7fec526578f0, 0x7fec52657a1a), 
INFO: Loaded 2 PC tables (9317 PCs): 9019 [0x7fec52997660,0x7fec529baa10), 298 [0x7fec52657a20,0x7fec52658cc0), 
INFO: -max_len is not provided; libFuzzer will not generate inputs larger than 4096 bytes
INFO: A corpus is not provided, starting from an empty corpus
#2	INITED exec/s: 0 rss: 64Mb
ERROR: no interesting inputs were found. Is the code instrumented for coverage? Exiting.
BAD BUILD: /tmp/not-out/fuzz_pillow seems to have either startup crash or exit:
/tmp/not-out/fuzz_pillow -rss_limit_mb=2560 -timeout=25 -seed=1337 -runs=4 < /dev/null
INFO: Using preloaded libfuzzer
INFO: Running with entropic power schedule (0xFF, 100).
INFO: Seed: 1337
INFO: Loaded 2 modules   (9317 inline 8-bit counters): 9019 [0x7f2194113320, 0x7f219411565b), 298 [0x7f2193dd58f0, 0x7f2193dd5a1a), 
INFO: Loaded 2 PC tables (9317 PCs): 9019 [0x7f2194115660,0x7f2194138a10), 298 [0x7f2193dd5a20,0x7f2193dd6cc0), 
INFO: -max_len is not provided; libFuzzer will not generate inputs larger than 4096 bytes
INFO: A corpus is not provided, starting from an empty corpus
#2	INITED exec/s: 0 rss: 64Mb
ERROR: no interesting inputs were found. Is the code instrumented for coverage? Exiting.

Broken fuzz targets 1
('/tmp/not-out/fuzz_pillow', CompletedProcess(args=['bad_build_check', '/tmp/not-out/fuzz_pillow'], returncode=2, stdout=b'BAD BUILD: /tmp/not-out/fuzz_pillow does not seem to have coverage instrumentation.\nINFO: Using preloaded libfuzzer\nINFO: Running with entropic power schedule (0xFF, 100).\nINFO: Seed: 1337\nINFO: Loaded 2 modules   (9317 inline 8-bit counters): 9019 [0x7f6b4b6d8320, 0x7f6b4b6da65b), 298 [0x7f6b4b39a8f0, 0x7f6b4b39aa1a), \nINFO: Loaded 2 PC tables (9317 PCs): 9019 [0x7f6b4b6da660,0x7f6b4b6fda10), 298 [0x7f6b4b39aa20,0x7f6b4b39bcc0), \nINFO: -max_len is not provided; libFuzzer will not generate inputs larger than 4096 bytes\nINFO: A corpus is not provided, starting from an empty corpus\n#2\tINITED exec/s: 0 rss: 64Mb\nERROR: no interesting inputs were found. Is the code instrumented for coverage? Exiting.\nBAD BUILD: /tmp/not-out/fuzz_pillow seems to have either startup crash or exit:\n/tmp/not-out/fuzz_pillow -rss_limit_mb=2560 -timeout=25 -seed=1337 -runs=4 < /dev/null\nINFO: Using preloaded libfuzzer\nINFO: Running with entropic power schedule (0xFF, 100).\nINFO: Seed: 1337\nINFO: Loaded 2 modules   (9317 inline 8-bit counters): 9019 [0x7f67e9a2e320, 0x7f67e9a3065b), 298 [0x7f67e96f08f0, 0x7f67e96f0a1a), \nINFO: Loaded 2 PC tables (9317 PCs): 9019 [0x7f67e9a30660,0x7f67e9a53a10), 298 [0x7f67e96f0a20,0x7f67e96f1cc0), \nINFO: -max_len is not provided; libFuzzer will not generate inputs larger than 4096 bytes\nINFO: A corpus is not provided, starting from an empty corpus\n#2\tINITED exec/s: 0 rss: 64Mb\nERROR: no interesting inputs were found. Is the code instrumented for coverage? Exiting.\n', stderr=b''))
ERROR: 50.0% of fuzz targets seem to be broken. See the list above for a detailed information.

stderr: 

[radarhere]

@wiredfool could you provide a link to the oss-fuzz issue you mentioned?

[radarhere]

Looks like this has fixed itself.

I changed a C file, and found that CIFuzz passed - https://github.com/radarhere/Pillow/runs/3592545745
I've rerun https://github.com/python-pillow/Pillow/pull/5694/checks?check_run_id=3592548158 on #5694 and it is passing now.

[hugovk]

Thanks for checking!