Reduce the severity of HWASAN not being supported by sanitizer.py

[ParisMeuleman]

@jonathanmetzman noticed a spike in these errors, we are tracking the action of fixing the option handling in https://issues.chromium.org/u/2/issues/329346758, meanwhile, we can reduce the severity of the log.

I am not sure of the actual fix, the HWAsan documentation is a bit lacking. It seems we might be able to set those as an environment variable in a wrap.sh. Can/should this be done by clusterfuzz is to be determined.

Engine fuzzers are not uzing this (most of android fuzzing IIUC?)

[jonathanmetzman]

/gcbrun

[jonathanmetzman]

I don't have a great understanding of this some I'm a little worried about turning this off for the Android team's ClusterFuzz instance.

[ParisMeuleman]

I don't have a great understanding of this some I'm a little worried about turning this off for the Android team's ClusterFuzz instance.

makes sense, I guess involving @marktefftech is our best course of action?

FYI regarding the follow up, I'm testing adding the HWASAN_OPTIONS through a wrap.sh in the apk. I had some success but still need to test some more.

[marktefftech]

This logging change LGTM

As far as a long-term solution, it would be nice to follow the existing convention of setting the ADDITIONAL_<sanitizer>_OPTIONS.

This will keep things the same for hwasan as they are for other sanitizers. We can set the options via configuration, for example, using the ClusterFuzz job definitions.

Take the engine_asan template as an example, which sets ADDITIONAL_ASAN_OPTIONS = quarantine_size_mb=64.......

[ParisMeuleman]

This logging change LGTM

As far as a long-term solution, it would be nice to follow the existing convention of setting the ADDITIONAL_<sanitizer>_OPTIONS.

This will keep things the same for hwasan as they are for other sanitizers. We can set the options via configuration, for example, using the ClusterFuzz job definitions.

Take the engine_asan template as an example, which sets ADDITIONAL_ASAN_OPTIONS = quarantine_size_mb=64.......

So using the additional_hwasan_options sounds good. Our hwasan job actually had some (but mislabelled). The issue here is that we are not using engine fuzzers, which as I understand have their own way of setting options. Instead we have a blackbox fuzzer where we:

• generate files (fuzzer's job)
• serve files over HTTP[s], with port forwarding to the host
• open chrome (a hwasan build) on one of those file
• observe the result
• repeat

Opening Chrome with those hwasan options set is the challenge:

• We have the option to add a wrap.sh script in the APK, but this means the options need to be decided at build time.
• Have a wrap.sh set using adb shell setprop wrap.$PACKAGE_NAME. This seems more flexible and seems to work with my local tests.