Add linux aarch64 target to ci test#3217
Conversation
5049277 to
b2c3bb3
Compare
|
Not sure what to do about the failure: |
ac7e42b to
ebb148b
Compare
|
This needs a rebase but I think the failure is legit. Is it crashing on the valist in that test? If you can reproduce locally, it'd be great to see what bit of the test is failing. |
ebb148b to
330b518
Compare
|
I reproduced and got a stacktrace locally: How to reproduce, something like this: |
|
I tried running it manually but it seems to succeed, @emilio how can I trigger the panic by running bindgen? edit: Managed to reproduce edit2: Attached debug output edit3: I have never worked on bindgen, so I probably won't be able to figure out what could be wrong here. |
|
Ok, to reproduce that on x86_64 you can: It seems bindgen can't reason about the builtin va_list there: https://github.com/Blizzard/clang/blob/6c67f57680c38ccd146786841eeac753ad45bf0f/lib/AST/ASTContext.cpp#L6099 As a workaround for now, you can probably ifdef the tests using va_list and update expectations or so. |
|
Please do file an issue about the wrap-static-fns not dealing with valists in aarch64 |
1dbee4b to
ccc363a
Compare
|
Fixed the failed CI and rebased Maybe it is a bit excessive to run all the combinations on |
emilio
left a comment
There was a problem hiding this comment.
Looks good. If we see CI taking too long we can restrict in a follow-up.
Would you mind filing an issue for the valist bit that was broken and reference it from the code where we disable that test?
| .arg("-DUSE_VA_HEADER"); | ||
|
|
||
| #[cfg(all(target_arch = "aarch64", target_os = "linux"))] | ||
| { |
There was a problem hiding this comment.
Can you point to a follow-up issue for this?
5a8d5dd to
22ede02
Compare
I guess it should be as simple as this now that aarch64 runners have been added.