Alastair Robertson

It looks like this affects all probe types using perf events (`profile`, `interval`, `software` and `hardware`). Ideally, our solution would work for all these probe types and not just `interval`....

I think this is a problem with pahole's DWARF -> BTF translator: https://git.kernel.org/pub/scm/devel/pahole/pahole.git/tree/btf_encoder.c#n1199 We use it to generate our test BTF data here: https://github.com/bpftrace/bpftrace/blob/3691e715736cbd128d5e8ccec45b8255361ded91/tests/data/CMakeLists.txt#L16-L22

Thanks for the explanation - that does clear up how it's intended to be used. I'm still not sure what the specific purpose is though (not that I'm saying there...

Does the `-c` command line option not work for that purpose? e.g. ``` # bpftrace my_script.bt -c ./test/run-integration-tests.sh ```

The things I'm not keen on in the proposed bpftrace mitigations are: 1. We'll have no way to determine if the running kernel has the fixes applied or is vulnerable...

> It is worth keeping in mind that the bpf lock change being made upstream does not prevent ABBA deadlocks, but only backs out of them after the code fails...

One liners wouldn't typically define an `END` probe though - in that case I'm saying we should keep the current behaviour and print out all maps. As for why the...

> Ah, I see. That will probably be possible once nested maps come into play. AFAIK, there're still problems with their support in the kernel, though. Off topic, but after...

Definitely a useful feature to have. It would be good to have this working for all data sources (BTF, DWARF, C), so I've bumped it up to a medium difficulty....

I'd also add that `$myenum` should be displayed as a string in `print($myenum)` and `@[$myenum] = $myenum`. It would be a change in behaviour, but I'd say it's more useful...