ADDITIONAL SYSTEM INFORMATION :
Tested on multiple JDKs, 17, 21, 24, 25, seems to affect all versions.

A DESCRIPTION OF THE PROBLEM :
When capturing an allocation profile, the first ObjectAllocationSample events have an incorrect weight (they are including allocations from before the start of the capture).

This seems to be a regression of JDK-8307488, however after testing with several JDKs, I cannot find a version where it was actually fixed.

STEPS TO FOLLOW TO REPRODUCE THE PROBLEM :
The reproducer attached to issue JDK-8307488 still reproduces the issue: https://bugs.openjdk.org/browse/JDK-8307488

EXPECTED VERSUS ACTUAL BEHAVIOR :
EXPECTED -
Expected to not see the first ObjectAllocationSample event have a such a large weight of ~2GB; it should be much lower, as well as reporting the correct class / line where the allocation occurred.
ACTUAL -
First allocation from the profile is 2GB:

jdk.ObjectAllocationSample {
  startTime = 10:34:06.286 (2025-07-17)
  objectClass = java.lang.Object (classLoader = bootstrap)
  weight = 2.0 GB
  eventThread = "main" (javaThreadId = 3)
  stackTrace = [
    WrongAllocationWeight.main(String[]) line: 28
  ]
}

It's also attributed to the wrong line / class.

---------- BEGIN SOURCE ----------
See test case / reproducer attached to JDK-8307488: https://bugs.openjdk.org/browse/JDK-8307488
---------- END SOURCE ----------