[clr-ios] Fix Vector<T> width and R2R AVX baseline mismatch on x64 iOS/tvOS simulators by kotlarmilos · Pull Request #129012 · dotnet/runtime

kotlarmilos · 2026-06-04T19:59:31Z

Description

The interpreter only supports 128-bit Vector, but on Mac Catalyst and iOS simulator x64 it was ending up with 256-bit Vector, which is what caused the NullReferenceException in Vector.get_IsHardwareAccelerated. The runtime already limits Vector to 128-bit for the full interpreter mode, but it wasn't doing so on no-JIT builds where everything is interpreted anyway, so AVX2 on x64 silently bumped it up to 256-bit. This change applies the same 128-bit limit whenever the interpreter is the only thing running code. arm64 was never affected since Vector is always 128-bit there.

Crossgen2 had the matching problem where it presumed an x86-64-v3 (AVX2) ReadyToRun baseline for the Apple x64 simulator targets, so the precompiled R2R code and the interpreter disagreed on the Vector width. It now uses the x86-64-v2 baseline for the iOS/tvOS simulators as well, matching osx and maccatalyst, so both share the same 128-bit view. The x64 simulators also run under Rosetta, which does not support AVX, so presuming it was incorrect regardless.

As defense in depth, when an eager fixup failure disables R2R on a no-JIT runtime where the image was compiled with stripped IL bodies (so there is no IL to fall back to), the runtime now fails fast with a clear message instead of running the throw stub left in place of the IL, which previously surfaced as an opaque stack overflow.

Fixes #128901
Fixes #128898

…-JIT x64 The interpreter only supports 128-bit Vector<T>, but on Mac Catalyst and iOS simulator x64 it was ending up with 256-bit Vector<T>, which caused the NullReferenceException in Vector.get_IsHardwareAccelerated. The runtime already limits Vector<T> to 128-bit for the full interpreter mode, but it wasn't doing so on no-JIT builds where everything is interpreted anyway, so AVX2 on x64 silently bumped it up to 256-bit. Apply the same 128-bit limit whenever the interpreter is the only thing running code. arm64 was never affected since Vector<T> is always 128-bit there. Related dotnet#128898 Fixes dotnet#128901 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

Pull request overview

This PR adjusts CoreCLR’s CPU feature initialization so that when execution is interpreter-only (either DOTNET_InterpMode=3 or when the runtime is built without dynamic code compilation/JIT), Vector<T> is constrained to 128-bit on x86/x64. This prevents the runtime from advertising/selecting 256-bit Vector<T> on AVX2-capable x64 machines in configurations where the interpreter can only handle 128-bit vectors.

Changes:

Detect “interpreter-only” operation as (InterpMode == 3) || !FEATURE_DYNAMIC_CODE_COMPILED.
Clamp EXTERNAL_MaxVectorTBitWidth to 128 in interpreter-only mode to prevent enabling InstructionSet_VectorT256/512.
Clamp EXTERNAL_PreferredVectorBitWidth to 128 in interpreter-only mode to keep vector-width preference consistent with interpreter constraints.

Show a summary per file

File	Description
src/coreclr/vm/codeman.cpp	Updates `EEJitManager::SetCpuInfo()` to force 128-bit `Vector<T>` sizing when the interpreter is the only execution engine (including no-JIT builds).

Copilot's findings

Files reviewed: 1/1 changed files
Comments generated: 0

dotnet-policy-service · 2026-06-04T20:06:42Z

Tagging subscribers to 'os-ios': @vitek-karas, @kotlarmilos, @steveisok, @akoeplinger
See info in area-owners.md if you want to be subscribed.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

On interpreter-only runtimes (no JIT), R2R native code is the only executable form of a method. When the image was compiled with --strip-il-bodies and an eager fixup fails (e.g. an instruction-set check), R2R is disabled and the interpreter runs the throw stub left in place of the IL, which surfaces as an opaque self-recursing stack overflow. Fail fast with a clear message instead. Runtimes with a JIT already reject stripped images at load, so this path is only reachable when there is no recovery option. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

Copilot's findings

Files reviewed: 4/4 changed files
Comments generated: 2

iOS and tvOS devices are ARM-only, so the x86/x64 branch is never reached for TargetOS.iOS/tvOS; only the simulator targets run x64 (Intel/Rosetta) and need the x86-64-v2 baseline. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

The READYTORUN_FLAG_STRIPPED_IL_BODIES flag is emitted on each component assembly's own R2R header (RewriteComponentFile), not on the composite global header, so read it from m_pHeader like IsPartial() and the existing check in ReadyToRunInfo::InitializeForModule. Reading the composite header would miss stripped IL bodies for composite components and skip the fatal check on no-JIT builds. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

Copilot's findings

Files reviewed: 4/4 changed files
Comments generated: 0 new

The first maxVectorTBitWidth check already pins Vector<T> to 128-bit, so VectorT256/VectorT512 are never set on interpreter-only runtimes. The preferredVectorBitWidth block only controls the JIT's internal preferred vector width, which is irrelevant when the interpreter is the only engine. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

Copilot's findings

Files reviewed: 4/4 changed files
Comments generated: 0 new

janvorli

LGTM modulo a nit to consider. Thank you!

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot AI review requested due to automatic review settings June 4, 2026 19:59

Copilot started reviewing on behalf of kotlarmilos June 4, 2026 19:59 View session

github-actions Bot added the area-CodeGen-Interpreter-coreclr label Jun 4, 2026

dotnet-policy-service Bot assigned kotlarmilos Jun 4, 2026

kotlarmilos changed the title ~~[clr-ios] Fix interpreter NullReferenceException with Vector<T> on no-JIT x64 (Mac Catalyst / iOS simulator)~~ [clr-ios] Fix interpreter NullReferenceException with Vector<T> on x64 ios simulator Jun 4, 2026

Keep existing Vector<T> cap comments

33fb93c

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot AI reviewed Jun 4, 2026

View reviewed changes

kotlarmilos requested review from davidwrighton, janvorli and kg June 4, 2026 20:06

kotlarmilos added this to the 11.0.0 milestone Jun 4, 2026

kotlarmilos added the os-ios Apple iOS label Jun 4, 2026

janvorli reviewed Jun 4, 2026

View reviewed changes

Comment thread src/coreclr/vm/codeman.cpp Outdated

This was referenced Jun 5, 2026

slow macOS - "##[error]The job running on agent Azure Pipelines 9 ran longer than the maximum time of 60 minutes." dotnet/dnceng#1883

Open

The Operation will be canceled. The next steps may not contain expected logs. dotnet/dnceng#3008

Open

Use 128-bit Vector<T> baseline for crossgen2 on Apple x64 targets

f75f6b7

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

kotlarmilos requested a review from MichalStrehovsky as a code owner June 5, 2026 08:49

MichalStrehovsky reviewed Jun 5, 2026

View reviewed changes

Comment thread src/coreclr/tools/Common/InstructionSetHelpers.cs Outdated

Copilot AI review requested due to automatic review settings June 5, 2026 09:28

Copilot started reviewing on behalf of kotlarmilos June 5, 2026 09:28 View session

Copilot AI reviewed Jun 5, 2026

View reviewed changes

Comment thread src/coreclr/vm/readytoruninfo.h Outdated

Comment thread src/coreclr/vm/codeman.cpp Outdated

BrzVlad reviewed Jun 5, 2026

View reviewed changes

Comment thread src/coreclr/tools/Common/InstructionSetHelpers.cs

kotlarmilos mentioned this pull request Jun 5, 2026

Stackoverflow at startup with CoreCLR on ios simulator x64 #128898

Closed

Copilot AI review requested due to automatic review settings June 5, 2026 10:22

Copilot started reviewing on behalf of kotlarmilos June 5, 2026 10:23 View session

Fix stale Vector<T> comment to match the actual condition

0cb7e20

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

kotlarmilos changed the title ~~[clr-ios] Fix interpreter NullReferenceException with Vector<T> on x64 ios simulator~~ [clr-ios] Fix Vector<T> width and R2R AVX baseline mismatch on x64 iOS/tvOS simulators Jun 5, 2026

Copilot AI reviewed Jun 5, 2026

View reviewed changes

jkotas reviewed Jun 5, 2026

View reviewed changes

Comment thread src/coreclr/vm/codeman.cpp Outdated

jkotas reviewed Jun 5, 2026

View reviewed changes

Comment thread src/coreclr/vm/codeman.cpp Outdated

kotlarmilos and others added 2 commits June 5, 2026 17:35

Avoid reading InterpMode config when overridden on no-JIT runtimes

41dd8e6

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot AI review requested due to automatic review settings June 5, 2026 15:52

Copilot started reviewing on behalf of kotlarmilos June 5, 2026 15:52 View session

Copilot AI reviewed Jun 5, 2026

View reviewed changes

build-analysis Bot mentioned this pull request Jun 5, 2026

Assertion failed: new_time >= loop->time in Component Governance Detection dotnet/dnceng#6510

Open

3 tasks

kotlarmilos requested review from BrzVlad, MichalStrehovsky, janvorli and jkotas June 8, 2026 13:17

janvorli approved these changes Jun 8, 2026

View reviewed changes

Comment thread src/coreclr/vm/readytoruninfo.h Outdated

Rename IsStrippedILBodies to HasStrippedILBodies

a7fb6b9

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

kotlarmilos enabled auto-merge (squash) June 8, 2026 14:44

build-analysis Bot mentioned this pull request Jun 8, 2026

CI build failure: error CS8032: An instance of analyzer Microsoft.NetCore.CSharp.Analyzers.... cannot be created #129031

Open

kotlarmilos merged commit fc3e8ea into dotnet:main Jun 8, 2026
113 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[clr-ios] Fix Vector<T> width and R2R AVX baseline mismatch on x64 iOS/tvOS simulators#129012

[clr-ios] Fix Vector<T> width and R2R AVX baseline mismatch on x64 iOS/tvOS simulators#129012
kotlarmilos merged 10 commits into
dotnet:mainfrom
kotlarmilos:fix-interp-vectort-128-mobile-x64

kotlarmilos commented Jun 4, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

dotnet-policy-service Bot commented Jun 4, 2026

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

janvorli left a comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Conversation

kotlarmilos commented Jun 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Copilot's findings

Uh oh!

dotnet-policy-service Bot commented Jun 4, 2026

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot's findings

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot's findings

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot's findings

Uh oh!

janvorli left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

kotlarmilos commented Jun 4, 2026 •

edited

Loading