Simon Cornish

Hi Steve. Sorry for not including test cases, but this was found in a regression test where yaws is only a small part. But consider this case where an embedded...

Pretty sure dialyzer is wrong about that. And also that the test failures (peer and global) are problems with the tests and not this PR.

I added a test case and fixed the conditional. Thanks for your comments.

The sample I attached to #8208 (and now it is actually attached!) includes a rudimentary benchmark. There's a lot of variance in the results but they seem to consistently be...

If something like [this](https://github.com/erlang/otp/files/14502539/uint.patch) was what you had in mind then it was about the same.

The test module & port driver didn't get attached so here they are: [sigsegv_drv.zip](https://github.com/erlang/otp/files/14488700/sigsegv_drv.zip)