[erlang-bugs] A bug in address patching in hipe_x86.c?

[Sverker Eriksson]

On 02/02/2014 08:06 PM, Yiannis Tsiouris wrote:
> Do you think that this test suffices? I'd be happy to test more
> thoroughly... Just point me to some tests/scenarios!
>

There is actually a test case that tests module upgrade with a lot of 
combinations of different function calls:

lib/kernel/test/code_SUITE:upgrade.

It only tests beam modules in its present form as it doesn't run clean 
with hipe compiled modules.

git fetch https://github.com/sverker/otp.git code_SUITE-upgrade-hipe

This commit enables hipe compiled modules and comments out all cases 
currently not working for hipe.



However, even if your patch should pass the test above, I agree with 
Mikael that it would be better if it could be solved in a different way 
as the patch does not harmonize with the other architectures.

/Sverker