Use inline arrays in GCMemoryInfoData.

[teo-tsirpanis]

Removes two uses of MemoryMarshal.CreateReadOnlySpan.

[azure-pipelines]

Azure Pipelines:
Successfully started running 3 pipeline(s).
13 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[dotnet-policy-service]

Tagging subscribers to this area: @dotnet/area-system-runtime
See info in area-owners.md if you want to be subscribed.

[Copilot]

Pull request overview

This PR updates GCMemoryInfoData to use inline arrays (InlineArray5<GCGenerationInfo> and InlineArray2<TimeSpan>) in order to avoid MemoryMarshal.CreateReadOnlySpan when exposing generation info and pause durations via ReadOnlySpan<T>.

Changes:

• Replaces the per-element backing fields for generation info and pause durations with InlineArray5<T> / InlineArray2<T> fields.
• Updates GCMemoryInfo.PauseDurations and GCMemoryInfo.GenerationInfo to return spans directly from the inline-array fields.
• Adds using System.Runtime.CompilerServices; for inline-array types.

[teo-tsirpanis]

@jkotas is there a way that we could make the DEFINE_FIELD_U declarations work with InlineArray?

runtime/src/coreclr/vm/corelib.h

Lines 1445 to 1451 in 815a631

	DEFINE_FIELD_U(_pauseDuration0, GCMemoryInfoData, pauseDuration0)
	DEFINE_FIELD_U(_pauseDuration1, GCMemoryInfoData, pauseDuration1)
	DEFINE_FIELD_U(_generationInfo0, GCMemoryInfoData, generationInfo0)
	DEFINE_FIELD_U(_generationInfo1, GCMemoryInfoData, generationInfo1)
	DEFINE_FIELD_U(_generationInfo2, GCMemoryInfoData, generationInfo2)
	DEFINE_FIELD_U(_generationInfo3, GCMemoryInfoData, generationInfo3)
	DEFINE_FIELD_U(_generationInfo4, GCMemoryInfoData, generationInfo4)

[azure-pipelines]

Azure Pipelines:
Successfully started running 3 pipeline(s).
13 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[jkotas]

is there a way that we could make the DEFINE_FIELD_U declarations work with InlineArray?

Try changing the C++ definition of GCMemoryInfoData to use GCGenerationInfo generationInfo[5]; and the corelib.h to DEFINE_FIELD_U(_generationInfo, GCMemoryInfoData, generationInfo)

[teo-tsirpanis]

Done, thanks. I first thought I had to encode the managed field's type in the declaration, but turns out that's not the case.

[Copilot]

Pull request overview

Copilot reviewed 5 out of 5 changed files in this pull request and generated no new comments.

Suppressed comments (3)

src/libraries/System.Private.CoreLib/src/System/GCMemoryInfo.cs:191

• If GCMemoryInfoData keeps span accessors (to avoid exposing writable inline-array fields), this property should return the accessor rather than the backing field.

        public ReadOnlySpan<GCGenerationInfo> GenerationInfo => _data._generationInfo;

src/libraries/System.Private.CoreLib/src/System/GCMemoryInfo.cs:75

• These inline-array fields were previously private (via the individual element fields). Making them internal widens the writable surface area for what appears to be runtime-filled layout data, and a repo-wide search shows no other managed code needs direct field access. Consider keeping the fields private and exposing internal ReadOnlySpan<T> accessors (still avoiding MemoryMarshal.CreateReadOnlySpan).

        internal InlineArray5<GCGenerationInfo> _generationInfo;
        internal InlineArray2<TimeSpan> _pauseDurations;

src/libraries/System.Private.CoreLib/src/System/GCMemoryInfo.cs:181

• If GCMemoryInfoData keeps span accessors (to avoid exposing writable inline-array fields), this property should return the accessor rather than the backing field.

This issue also appears on line 191 of the same file.

        public ReadOnlySpan<TimeSpan> PauseDurations => _data._pauseDurations;

[teo-tsirpanis]

CI is about to get green.

[jkotas]

/ba-g Build Monitor Helix Jobs stuck

[jkotas]

Thanks