Hacker News new | ask | show | jobs
by zeratax 2 days ago
> Clang relies on address layout for ordering things

I would consider that a bug tbh
2 comments
j2kun 2 days ago
What is kind of annoying is that the author jumps to "I hate compilers" instead of "I will report/help fix this bug upstream."
link
chiyc 2 days ago
I don't get the sense they hate compilers at all. The writing describes work they seem to love doing. It's just clickbait.

And it may not have crossed their mind that the clang behavior is a bug after finding a workaround. I'd also assume compilers do things "no mere mortal can fully comprehend on their own".

link
xena 2 days ago
This might be the first time in my career I have genuinely found a compiler bug. I've been operating under the axiom of "don't assume it's a compiler bug, assume you're fucking it up somehow". When you get to the point that disabling ASLR makes it consistent intra-host I think I've won the right to at least suspect a compiler bug is at play.

I'll go file it upstream after work today.

link
ammar2 2 days ago
If you feel like increasing your power as per your post, this is a somewhat decent first LLVM issue, take a look at WebAssemblyCFGStackify.cpp :)

llvm/test/CodeGen/WebAssembly/cfg-stackify-eh.ll and friends are existing tests that you can kinda mangle if you want to get a good reproducer.

Also take a look at https://discourse.llvm.org/t/reverse-iteration-bots/72224

Otherwise, happy to put my reproducer/patch on the bug after you file it!

link
xena 2 days ago
I'm gonna have to file the bug without a minimal reproduction case. The issue seems to be those try_table blocks getting nondeterministically reordered at link time (is it using machine pointers for iteration order?). Sadly I'm observing this with a local checkout of binaryen, so it may take a while for you to find the minimal reproduction case.
link
xena 1 day ago
I filed the bug: https://github.com/llvm/llvm-project/issues/204883
link
StilesCrisis 2 days ago
If reproducible builds is a stated goal for Clang? I'm not sure that it is. If so, absolutely a bug.
link
ammar2 2 days ago
Not sure if it's a stated goal somewhere official but there's been plenty of fixes of the years moving stuff to be deterministic, e.g https://github.com/llvm/llvm-project/commit/cdbde3aacc1260a7...

The internal programming guide also says which collections to use for deterministic iteration order: https://llvm.org/docs/ProgrammersManual.html#llvm-adt-setvec...

So definitely a bug here.

link