Fix in-process decoding of DateTime in EventSource

[josalem]

Fixes #61563

Will require backporting to release/6.0 since this is a regression from 5.0.

@MihaZupan root caused this issue succinctly in this comment: #61563 (comment)

@tommcdon it is a bug introduced by #52092 in .NET 6.0, but only affects in-process EventListeners.

The difference is in how the serialized value (long) is de-serialized by DecodeObjects:

• In 5.0, the DateTime would be converted from a long back to a DateTime via DateTime.FromFileTimeUtc
• In 6.0, the value is interpreted as a DateTime directly (i.e. exact number of ticks)

Sadly this was not caught by tests since none of them uses DateTime with WriteEventCore - we are only testing the varargs overload.

The fix here would be to change

runtime/src/libraries/System.Private.CoreLib/src/System/Diagnostics/Tracing/EventSource.cs

Line 1844 in a25aa66

decoded = *(DateTime*)dataPointer;

to

decoded = DateTime.FromFileTimeUtc(*(long*)dataPointer);

-- MihaZupan

[ghost]

Tagging subscribers to this area: @tarekgh, @tommcdon, @pjanotti
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Fixes #61563

Will require backporting to release/6.0 since this is a regression from 5.0.

@MihaZupan root caused this issue succinctly in this comment: #61563 (comment)

@tommcdon it is a bug introduced by #52092 in .NET 6.0, but only affects in-process EventListeners.

The difference is in how the serialized value (long) is de-serialized by DecodeObjects:

• In 5.0, the DateTime would be converted from a long back to a DateTime via DateTime.FromFileTimeUtc
• In 6.0, the value is interpreted as a DateTime directly (i.e. exact number of ticks)

Sadly this was not caught by tests since none of them uses DateTime with WriteEventCore - we are only testing the varargs overload.

The fix here would be to change

runtime/src/libraries/System.Private.CoreLib/src/System/Diagnostics/Tracing/EventSource.cs

Line 1844 in a25aa66

decoded = *(DateTime*)dataPointer;

to

decoded = DateTime.FromFileTimeUtc(*(long*)dataPointer);

-- MihaZupan

Author:	josalem
Assignees:	josalem
Labels:	area-System.Diagnostics.Tracing
Milestone:	7.0.0

[noahfalk]

LGTM

[danmoseley]

Sadly this was not caught by tests

Is such a test feasible?

[josalem]

Sadly this was not caught by tests

Is such a test feasible?

Let me take a look and add one if so.

[josalem]

Test added! I'll merge on clean CI.

[josalem]

/azp run runtime

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[josalem]

The test failures are #67878.

The macOS NativeAOT job timed out, so I'll rerun it.

[josalem]

/azp run runtime

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[josalem]

Tests all seem to pass but the Build Analysis check fails with test failures that aren't shown elsewhere. Folks okay with merging anyway?

[noahfalk]

I'd merge it : )

[josalem]

/backport to release/6.0

[github-actions]

Started backporting to release/6.0: https://github.com/dotnet/runtime/actions/runs/2204546765