On Wed, 22 Jul 2026 21:22:49 GMT, Shiv Shah <[email protected]> wrote:

> serviceability/sa/ClhsdbLongConstant.java asserts an exact value for 
> VM_Version::CPU_SHA in checkForTruncation(). CPU_SHA is a feature-flag bit 
> index in vm_version_x86.hpp whose value shifts whenever a CPU feature is 
> added, so the hardcoded expectation goes stale and the test fails (it has 
> already been bumped once, and the in-file comment lists yet another value).
> 
> This check exists only to confirm the `longConstant` output was not 
> truncated, using CPU_SHA as a sentinel that appears late in the output, the 
> exact value is irrelevant to that purpose. SA reads the value correctly from 
> the VM in every case.
> 
> Change checkForTruncation() to verify the constant is present with a 
> parseable long value (checkLongValuePresent) instead of asserting a specific 
> number, so the check no longer breaks on every feature-flag change. The 
> markWord::hash_mask_in_place assertion is unchanged, since that value is 
> derived from a stable definition.
> 
> Testing: serviceability/sa/ClhsdbLongConstant.java (x64).
> 
> 
> 
> 
> 
> 
> ---------
> - [x] I confirm that I make this contribution in accordance with the [OpenJDK 
> Interim AI Policy](https://openjdk.org/legal/ai).

Why the test fails only in container/docker environment? Why no failures in jdk 
CI?

-------------

PR Comment: https://git.openjdk.org/jdk/pull/32017#issuecomment-5061569840

Reply via email to