Updating the XFail on PowerPC to debug mode for layout_stride/assert.conversion.pass.cpp

[maryammo]

Adding the debug hardening modes into PowerPC target to check the assertion messages, to fix the PPC rehl bot. This is needed as this commit changed the assertion to trap in production hardening modes.

[llvmbot]

@llvm/pr-subscribers-libcxx

Author: Maryam Moghadas (maryammo)

Changes

…conversion.pass.cpp

---

Full diff: https://github.com/llvm/llvm-project/pull/79169.diff

1 Files Affected:

• (modified) libcxx/test/libcxx/containers/views/mdspan/layout_stride/assert.conversion.pass.cpp (+1-1)

diff --git a/libcxx/test/libcxx/containers/views/mdspan/layout_stride/assert.conversion.pass.cpp b/libcxx/test/libcxx/containers/views/mdspan/layout_stride/assert.conversion.pass.cpp
index 81f6321ef519cd8..fbd225b587749c2 100644
--- a/libcxx/test/libcxx/containers/views/mdspan/layout_stride/assert.conversion.pass.cpp
+++ b/libcxx/test/libcxx/containers/views/mdspan/layout_stride/assert.conversion.pass.cpp
@@ -10,7 +10,7 @@
 // UNSUPPORTED: c++03, c++11, c++14, c++17, c++20
 // UNSUPPORTED: libcpp-hardening-mode=none
 // XFAIL: libcpp-hardening-mode=debug && availability-verbose_abort-missing
-// XFAIL: target=powerpc{{.*}}le-unknown-linux-gnu
+// XFAIL: libcpp-hardening-mode=debug && target=powerpc{{.*}}le-unknown-linux-gnu
 
 // <mdspan>
 

[kamaub]

This LGTM and is in agreement with the offending commits author's comment, thank you.
Please correct the title and PR description pre-commit so it goes in formatted normally.

[ldionne]

What is special about PPC in the debug mode? I don't understand why only that test needs something special in that case, too.

[var-const]

@ldionne The root cause is that these tests fail with a different assertion than the expected one on PowerPC (I don't know why). These tests are run in the extensive mode, so they used to be XFAILed before but started unexpectedly passing now that we no longer check for assertion messages in modes other than the debug. I recommended still XFAIL'ing them in the debug mode, hence the new annotation.

[ldionne]

It looks to me like the root cause must be something specific to PowerPC, like the size of a type or its signedness differing from mainstream platforms. Or maybe the fact that this bot is Little Endian. Either way, we really should figure out the underlying issue(s) here instead of just blindly XFAILing them like we did in 503dbe7.

[maryammo]

I have opened this issue to follow up with the root cause of PowerPC assertion differences for those test cases.
The purpose of this PR is to make the rhel bot green.

[ldionne]

@maryammo Please uncheck the "keep my email private" checkbox in the Github settings. It is causing your patches to be committed under an anonymous email address, and we want to steer away from that.

[maryammo]

@maryammo Please uncheck the "keep my email private" checkbox in the Github settings. It is causing your patches to be committed under an anonymous email address, and we want to steer away from that.

Done, Thanks.