wasm2c: Fix warning on signed/unsigned comparison

[shravanrn]

No description provided.

[keithw]

lgtm -- dupe of #2351 (I have the same question I had there but I assume the answer is basically "it happens on a newer compiler")

[shravanrn]

lgtm -- dupe of #2351 (I have the same question I had there but I assume the answer is basically "it happens on a newer compiler")

I think it's more the case that this code is executed only if there is a custom "grow_failed" handler. So, we don't see the warning in this CI, but it shows up on the Firefox build. I'll update the wasm2c_rlbox test to integrate a custom grow failed handler when I get the chance to make sure this code is exercised.

Separately, any idea if there is something wrong with the CI runner? I'm trying to track down the CI crash but it seems like the crashes aren't related to this commit?

[sbc100]

Interesting, binaryen just ran into some santizer-only failures that were unrelated to any changes to binaryen itself. @kripken does this looks like the same issue?

[kripken]

Hmm, it could be! Same symptoms, which are the ASan build just exits with an error, no message, nothing else. We also saw a TSan error which actually did contain an error message.

WebAssembly/binaryen#6396 has my investigation and fixes in Binaryen. Basically my best guess is that the github actions bots changed something, and recent LLVM versions are needed to handle it. Clang 17 is enough for ASan, but clang 18 is needed to also fix TSan if you want that as well.

[keithw]

We're having this problem on a lot of our students' machines too. The recent update to Ubuntu 23.10 (to kernel 6.5.0-25-generic #25-Ubuntu) seems to have broken a lot of the ASan output from g++ 13.2 and clang++ 16.0.6. Apparently there's some ASLR-related change? Haven't searched for more yet. :-(

[shravanrn]

Hmm... not sure what is the best way to proceed. Is this is something that is affecting all PRs. If yes, would it be reasonable to disable the failing CI builds for now? Also, happy to wait to land if someone is working on fixing the CI already

[shravanrn]

Testing a fix per
actions/runner#3207 (comment)

[shravanrn]

Looks like that fix worked. Will spin the CI workaround out into its own PR