[android] Disable a couple Reflection tests and fix an install path #82620
Conversation
The tests broke on the community Android CI since swiftlang#82325, and I just noticed the install issue when cross-compiling Testing with a freshly-built compiler, which I'd never done before. Also, fix the NDK path shown in the CMake output.
finagolfin
requested review from
etcwilde,
justice-adams-apple,
shahmishal and
slavapestov
as code owners
|
@swift-ci please smoke test |
|
@swift-ci smoke test windows |
| @@ -10,6 +10,7 @@ | |||
|
|
|||
| // FIXME: rdar://127796117 | |||
| // UNSUPPORTED: OS=linux-gnu && CPU=aarch64 | |||
| // XFAIL: OS=linux-android | |||
There was a problem hiding this comment.
Do you have the failure handy? I don't know if I'm allowed to copy the failure from the radar, but I can confirm if it's the same as the issue making linux-gnu aarch64 unsupported. If they are the same issue, it's probably worth keeping everything on the same line so that they get cleaned up together and I can make a note in the radar.
There was a problem hiding this comment.
@adrian-prantl says both are passing on linux AArch64 now. 🤷
There was a problem hiding this comment.
Huh. I wonder if I should update my checkout too? I've got a crash with the same stack trace on x86_64 FreeBSD. ¯\_(ツ)_/¯
Edit: Nope, confirmed that those changes didn't fix my issue. ELF section headers are still being read wrong.
| @@ -5,6 +5,7 @@ | |||
| // | |||
| // Temporarily disable on AArch64 Linux (rdar://88451721) | |||
| // UNSUPPORTED: OS=linux-gnu && CPU=aarch64 | |||
| // XFAIL: OS=linux-android | |||
There was a problem hiding this comment.
Same comment, do you have the failure handy? The radar is for a different failure mode, but I can try to confirm that it's the same issue.
|
@swift-ci smoke test windows |
|
Merging since this won't break anything, since I'm the only one using that new install path flag. |
|
Can you please provide the failure that you were seeing? Ideally a stack trace. For this one, can you also include which linker you were using. I'd like to try to keep some record somewhere so that it might get fixed at some point, or at least we know why it's disabled. Thanks. If you're using |
I reported my results on the pull that broke it, which I applied locally to confirm that it caused the regression, thought you might have scrolled up and seen that when I linked Adrian's comments here. You can examine it yourself on the community Android CI. We only use lld for Android. No segfaults for Android, just different output. I applied your ELF pull natively on Android, made no difference, so it looks like Android and FreeBSD are failing for different reasons. |
…wiftlang#82620) The tests broke on the community Android CI since swiftlang#82325, and I just noticed the install issue when cross-compiling Testing with a freshly-built compiler, which I'd never done before. Also, fix the NDK path shown in the CMake output.
The tests broke on the community Android CI since #82325, and I just noticed the install issue when cross-compiling Testing with a freshly-built compiler, which I'd never done before. Also, fix the NDK path shown in the CMake output.