allow using shims for aarch64_be#5075
Conversation
|
Thank you for contributing to Miri! A reviewer will take a look at your PR, typically within a week or two. |
| // We currently only test on little-endian, but big-endian aarch64 does exist. | ||
| return shims::aarch64::EvalContextExt::emulate_aarch64_intrinsic( | ||
| this, link_name, abi, args, dest, |
There was a problem hiding this comment.
So... are the implementations correct on BE? Do we have any way to check that?
There was a problem hiding this comment.
So... are the implementations correct on BE?
Unclear. I believe stdarch is kinda broken actually. Which is weird because we run the whole test suite etc but I think we've messed up loads and stores.
When I run the neon shim test file with qemu it falls over on the first assert. The test code translated to C works just fine, so stdarch is doing something wrong.
There was a problem hiding this comment.
In that case I'd prefer to wait until we have confidence in our implementation, before enabling the intrinsics again. Or is that somehow making it harder to fix stdarch?
rustbot
added
S-waiting-on-author
|
Reminder, once the PR becomes ready for a review, use |
3 participants
Related to #5065, it's fine to just allow shims on
aarch64_be.I've also removed some transmutes from the tests, they are very likely to cause endianness issues.