[clang] fixup! [clang] Forward `-fvalidate-ast-input-files-content` when loading AST dumps (PR #196980)

[via cfe-commits]

llvmorg-github-actions[bot] wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-clang-static-analyzer-1

Author: Arseniy Zaostrovnykh (necto)

<details>
<summary>Changes</summary>

Apparently `-t 300001010000` is not a universally valid date format.

This should fix the buildbot failure caused by #<!-- -->196298 
However, I am not sure this new format is portable either. `-t #####` at least was already used in some other tests.
One option would be to commit this and find out. Another is to go back to the python script, that I presume is more portable.

---
Full diff: https://github.com/llvm/llvm-project/pull/196980.diff


1 Files Affected:

- (modified) clang/test/Analysis/ctu/reusable-pch.c (+2-2) 


``````````diff
diff --git a/clang/test/Analysis/ctu/reusable-pch.c b/clang/test/Analysis/ctu/reusable-pch.c
index ddcccc9cf6150..34067fa74640c 100644
--- a/clang/test/Analysis/ctu/reusable-pch.c
+++ b/clang/test/Analysis/ctu/reusable-pch.c
@@ -19,8 +19,8 @@
 // Step 2b: Run with content validation - no difference.
 // RUN: %{ctu_analysis} %t/main.c -fvalidate-ast-input-files-content
 
-// Step 3: Set mtime of the source from which PCH was built to the year 3000 (way in the future).
-// RUN: touch -t 300001010000 %t/other.c
+// Step 3: Advance the modification time of the source from which the PCH was derived.
+// RUN: touch -d "+2 hours" %t/other.c
 
 // Step 4a: Run CTU using the "stale" PCH, and it should still load it and find the division by zero bug.
 // RUN: %{ctu_analysis} -fvalidate-ast-input-files-content %t/main.c

``````````

</details>


https://github.com/llvm/llvm-project/pull/196980