Fix source_file bug in mono_ppdb_lookup_location_internal

[bholmes]

document-records affect sequence points that follow in the byte stream, not points previously read.

When reading the document-record we may already be at the target offset. Do not update docname local if we have reached the end target already otherwise the wrong source_file will be reported for the requested offset

[bholmes]

@thaystg Can you think of a way to add a test for this? (I have verified the fix in Unity)

Our bug was from a class that had the ctor method body defined in one file and variables with initializers in another.
The last variable initialization line file name would be reported incorrectly.

[thaystg]

"The last variable initialization line file name would be reported incorrectly." does it mean that you could not step in the correct line, right?

[thaystg]

If this is the case you can create a test case on wasm debugger tests.

runtime/src/mono/browser/debugger/DebuggerTestSuite/BreakpointTests.cs

Lines 135 to 136 in 5a2d151

	public async Task CreateGoodBreakpointAndHit()
	{

Like this one, trying to add a breakpoint at the last variable initialization line.

Please let me know if you need any help to do it.

[bholmes]

It affects the Code Coverage tool

https://docs.unity3d.com/ScriptReference/TestTools.Coverage.GetStatsForAllCoveredMethods.html

I think you can see this issue, even though there is not much info https://issuetracker.unity3d.com/issues/coverage-api-returns-wrong-file-path-for-fields-in-partial-class-that-are-declared-with-an-initial-value

[bholmes]

@thaystg I have not been able to add a test to the BreakpointTests. As soon as I add a field with initializer to a partial class in the debugger-test2.cs file, I can no longer hit any breakpoints inside of (or any methods called by) the .ctor of the same partial class in debugger-test.cs. This is both with and without my changes. I am going to give up on this.

Also any reason to be concerned about the 'Build browser-wasm linux Release Mono_DebuggerTests_chrome' failure?

[thaystg]

@ilonatommy if you have any available time, can you try to create a test case for this one?

[vargaz]

This seems to have caused a regression:
#97342

[thaystg]

Let's revert this and backport the PR to net9 preview1?

[unity-nikos]

@thaystg @vargaz @bholmes could this be the case in trunk/LTS too? As the same PR landed there here.

[bholmes]

@thaystg @vargaz @bholmes could this be the case in trunk/LTS too? As the same PR landed there here.

Unity 1897 was patched with Unity-Technologies/mono#1941