retpoline
retpoline
Hmm integer overflow somewhere along the way makes sense. Regardless, thanks for the fix!
I see. Perhaps a check or warning for a less-than-sane pixel size?
Thanks for taking a look. Attached the second repro file. [crash.j2k.txt](https://github.com/uclouvain/openjpeg/files/8816189/crash.j2k.txt)
Did you happen to repro in the console or in a debugger using libefence? `(gdb) set environment LD_PRELOAD=/usr/lib/libefence.so`
I'd recommend running it with efence and seeing if it repros as it works differently than ASAN and each of them may catch bugs that the other does not. It's...
Perhaps it was fixed in a previous commit, it's been a few months since these crashes were found.
The ufile link has probably expired after 3 months or so, I've uploaded here as that should work too. [draco-crashes.zip](https://github.com/google/draco/files/8578571/draco-crashes.zip) Thanks for taking a look!
Looks good! `caught exception: Array items number is out of bounds` Sure, the fuzzer used was [litefuzz](https://github.com/sec-tools/litefuzz) -- it's very straightward to use if you'd like to run it again...
@jsteemann glad I could get the conversation started! nice work.
Another workaround for now would also be using an older version of SQLAlchemy. `pip install sqlalchemy==1.3.24`