[clr-ios] Fix SIGSEGV in open virtual delegate dispatch with interpreter by kotlarmilos · Pull Request #126199 · dotnet/runtime

kotlarmilos · 2026-03-27T13:09:01Z

Description

Fix a crash when the interpreter invokes a compiled open virtual delegate target.

When the interpreter's call delegate handler detects an open virtual delegate whose resolved target is compiled (targetIp == NULL), the code previously fell through to a generic path that called GetTarget() on the delegate. For open delegates, GetTarget() returns null, which corrupted the argument slot and caused a SIGSEGV.

The fix adds a dedicated path for open virtual delegates with compiled targets. It uses memmove to shift the call arguments down by one slot (removing the delegate object) while preserving 16-byte alignment for V128 arguments, then jumps to CALL_INTERP_METHOD to invoke the resolved targetMethod directly. The same shifting is applied for interpreted targets in the non-tail-call path. The alignment-preserving shift logic is extracted into a ShiftDelegateCallArgs() helper to avoid duplication.

Additionally, this re-enables test suites previously excluded under #124325 (except System.Runtime.Serialization.Xml.Tests, which hits a native stack overflow unrelated to this fix).

dotnet-policy-service · 2026-03-27T13:10:07Z

Tagging subscribers to this area: @BrzVlad, @janvorli, @kg
See info in area-owners.md if you want to be subscribed.

Copilot

Pull request overview

Fixes an interpreter crash on iOS when invoking compiled open virtual delegate targets, and updates AppleMobile/CoreCLR libraries test selection.

Changes:

Handle open virtual delegates whose resolved target has no interpreter bytecode (targetIp == NULL) by skipping the delegate argument slot and directly dispatching via CALL_INTERP_METHOD.
Remove several AppleMobile/CoreCLR test project exclusions from src/libraries/tests.proj.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
src/libraries/tests.proj	Removes previously-added AppleMobile/CoreCLR test exclusions (re-enables several Serialization test projects).
src/coreclr/vm/interpexec.cpp	Adds a NULL-`targetIp` open-virtual delegate path that avoids `GetTarget()` and dispatches the resolved `MethodDesc` directly.

src/libraries/tests.proj

src/coreclr/vm/interpexec.cpp

kotlarmilos · 2026-03-30T20:27:35Z

/azp run runtime-ioslike

azure-pipelines · 2026-03-30T20:27:52Z

Azure Pipelines successfully started running 1 pipeline(s).

Wrap the GetTarget/NonVirtualEntry2MethodDesc block in braces so the CALL_DELEGATE_INVOKE goto can legally jump past the OBJECTREF variable initialization. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

src/coreclr/vm/interpexec.cpp

Extract the duplicated memmove logic for removing the delegate object from call arguments into a static ShiftDelegateCallArgs() helper. Re-add System.Runtime.Serialization.Xml.Tests exclusion (dotnet#124325). Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 3 comments.

Comments suppressed due to low confidence (1)

src/libraries/Common/tests/System/Runtime/Serialization/Utils.cs:380

AssemblyDependencyResolver is unsupported on several platforms (notably Android/iOS/tvOS/Browser). This helper now special-cases Browser + AppleMobile, but it can still attempt to construct/use AssemblyDependencyResolver on Android (and potentially WASI), which would throw PlatformNotSupportedException if this helper is used there.

Consider switching these guards to PlatformDetection.IsMobile (as used in other tests) or explicitly excluding all platforms where AssemblyDependencyResolver is unsupported.

        public TestAssemblyLoadContext(string name, bool isCollectible, string mainAssemblyToLoadPath = null) : base(name, isCollectible)
        {
            if (!PlatformDetection.IsBrowser && !PlatformDetection.IsAppleMobile)
                _resolver = new AssemblyDependencyResolver(mainAssemblyToLoadPath ?? Assembly.GetExecutingAssembly().Location);
        }

        protected override Assembly Load(AssemblyName name)
        {
            if (PlatformDetection.IsBrowser || PlatformDetection.IsAppleMobile)
            {
                return base.Load(name);
            }

            string assemblyPath = _resolver.ResolveAssemblyToPath(name);
            if (assemblyPath != null)

src/coreclr/vm/interpexec.cpp

src/libraries/System.Runtime.Serialization.Xml/tests/DataContractSerializer.cs

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

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

Copilot

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.

src/coreclr/vm/interpexec.cpp

- Fix ShiftDelegateCallArgs memmove length to use totalArgsSize - firstAlignedDstOffset instead of totalArgsSize - sizeOfArgsUpto16ByteAlignment to avoid reading past actual argument data when pre-alignment size is not 16-byte aligned. - Extend INTOP_CALLDELEGATE_TAIL to carry sizeOfArgsUpto16ByteAlignment and targetArgsSize (oplength 4 -> 6) so the open virtual compiled target path works correctly for tail calls. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

kotlarmilos · 2026-04-09T20:27:16Z

/azp run runtime-ioslike

azure-pipelines · 2026-04-09T20:27:34Z

Azure Pipelines successfully started running 1 pipeline(s).

src/coreclr/vm/interpexec.cpp

janvorli · 2026-04-10T14:17:22Z

src/coreclr/interpreter/compiler.cpp

+                        targetArgsSize = ALIGN_UP_TO(targetArgsSize, argAlignment);
+                        targetArgsSize += size;
+                    }
+                    if (!found16ByteAligned)


Why is this needed now that you loop over all the arguments? It seems that in case no 16 byte aligned arg was found, the sizeOfArgsUpto16ByteAlignment would already be equal to the targetArgsSize.

Removed. When no 16-byte aligned arg is found, the loop already accumulates all sizes into sizeOfArgsUpto16ByteAlignment, so it equals targetArgsSize at the end.

src/coreclr/vm/interpexec.cpp

janvorli · 2026-04-10T14:24:56Z

src/coreclr/interpreter/inc/intops.def

-OPDEF(INTOP_CALLDELEGATE, "call.delegate", 5, 1, 1, InterpOpMethodHandle)
-OPDEF(INTOP_CALLDELEGATE_TAIL, "call.delegate.tail", 4, 1, 1, InterpOpMethodHandle)
+OPDEF(INTOP_CALLDELEGATE, "call.delegate", 6, 1, 1, InterpOpMethodHandle)
+OPDEF(INTOP_CALLDELEGATE_TAIL, "call.delegate.tail", 6, 1, 1, InterpOpMethodHandle)


Why are we expanding the INTOP_CALLDELEGATE_TAIL to 6 too?

Reverted. Tail calls do not need the size metadata. The interpreted path uses pTargetMethod->argsSize, and the compiled open virtual path now skips the delegate slot with callArgsOffset += INTERP_STACK_SLOT_SIZE, matching the interpreted tail path.

Actually it is needed since the callArgsOffset += INTERP_STACK_SLOT_SIZE approach breaks 16-byte alignment for V128 arguments (as Vlad noted in #126199 (comment)). Both tail and non-tail paths now use ShiftDelegateCallArgs which needs the size metadata.

I really meant to ask why the call.delegate.tail size grows by 2, I would expect it to grow by 1.

…check - Revert INTOP_CALLDELEGATE_TAIL to oplength 4; tail calls use the simpler callArgsOffset += INTERP_STACK_SLOT_SIZE approach matching the interpreted tail path. already makes sizeOfArgsUpto16ByteAlignment == targetArgsSize when no V128 arg exists. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

- Expand INTOP_CALLDELEGATE_TAIL to oplength 6 with size metadata - Use ShiftDelegateCallArgs in interpreted tail path instead of callArgsOffset + INTERP_STACK_SLOT_SIZE which breaks 16-byte alignment - Use ShiftDelegateCallArgs unconditionally in compiled open virtual path Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.

Comments suppressed due to low confidence (1)

src/coreclr/vm/interpexec.cpp:3224

In the open-virtual + compiled-target path, the tailcall case adjusts callArgsOffset by one slot and then jumps to CALL_INTERP_METHOD. If CALL_INTERP_METHOD ends up invoking a compiled method (targetIp == NULL), the call stub expects the argument block to follow interpreter stack alignment rules from offset 0 (including 16-byte alignment for V128). Bumping the base pointer by 8 can violate those assumptions. Consider either applying the same alignment-preserving shift for this tailcall case (which likely means carrying the size metadata on INTOP_CALLDELEGATE_TAIL too), or falling back to the normal delegate invoke/shuffle-thunk path for tail calls when the resolved target is compiled.


                    if ((targetMethod = NonVirtualEntry2MethodDesc(targetAddress)) != NULL)
                    {
                        // In this case targetMethod holds a pointer to the MethodDesc that will be called by using targetMethodObj as
                        // the this pointer. This may be the final method (in the case of instance method delegates), or it may be a

src/libraries/tests.proj

src/coreclr/vm/interpexec.cpp

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

Fix SIGSEGV in open virtual delegate dispatch with interpreter

25dde3a

Copilot AI review requested due to automatic review settings March 27, 2026 13:09

kotlarmilos requested review from BrzVlad, janvorli and kg as code owners March 27, 2026 13:09

kotlarmilos self-assigned this Mar 27, 2026

kotlarmilos added this to the 11.0.0 milestone Mar 27, 2026

github-actions bot added the area-CodeGen-Interpreter-coreclr label Mar 27, 2026

kotlarmilos added the os-ios Apple iOS label Mar 27, 2026

Copilot started reviewing on behalf of kotlarmilos March 27, 2026 13:09 View session

Copilot AI reviewed Mar 27, 2026

View reviewed changes

src/libraries/tests.proj Show resolved Hide resolved

src/coreclr/vm/interpexec.cpp Show resolved Hide resolved

BrzVlad reviewed Mar 27, 2026

View reviewed changes

src/coreclr/vm/interpexec.cpp Outdated Show resolved Hide resolved

Refactor delegate invocation handling in interpreter

80a07a4

This was referenced Mar 30, 2026

Unable to pull image from mcr.microsoft.com #117164

Open

XHarness package install failure on iOS due to devicectl NSPOSIXErrorDomain error 49 #123796

Open

Copilot AI review requested due to automatic review settings March 31, 2026 09:08

Copilot started reviewing on behalf of kotlarmilos March 31, 2026 09:09 View session

Copilot AI reviewed Mar 31, 2026

View reviewed changes

src/coreclr/vm/interpexec.cpp Show resolved Hide resolved

Copilot AI review requested due to automatic review settings April 8, 2026 08:19

Copilot started reviewing on behalf of kotlarmilos April 8, 2026 08:20 View session

kotlarmilos force-pushed the bugfix/clr-interpreter-open-delegate branch from 2f8a225 to 6b0cbc0 Compare April 8, 2026 08:21

Copilot AI reviewed Apr 8, 2026

View reviewed changes

src/coreclr/vm/interpexec.cpp Outdated Show resolved Hide resolved

src/libraries/System.Runtime.Serialization.Xml/tests/DataContractSerializer.cs Show resolved Hide resolved

src/libraries/System.Runtime.Serialization.Xml/tests/DataContractSerializer.cs Outdated Show resolved Hide resolved

Update src/coreclr/vm/interpexec.cpp

1648fca

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

Copilot AI review requested due to automatic review settings April 8, 2026 08:36

Copilot started reviewing on behalf of kotlarmilos April 8, 2026 08:38 View session

Address review: fix pointer style and add MacCatalyst to skip list

6e408b7

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

Copilot AI reviewed Apr 8, 2026

View reviewed changes

src/coreclr/vm/interpexec.cpp Show resolved Hide resolved

src/coreclr/vm/interpexec.cpp Outdated Show resolved Hide resolved

This was referenced Apr 8, 2026

Android arm32 device not found (armeabi-v7a architecture unavailable) #125440

Open

System.Net.NameResolution.Tests DNS failures: Name or service not known #126641

Open

kotlarmilos requested review from BrzVlad and jkotas April 9, 2026 10:48

BrzVlad approved these changes Apr 9, 2026

View reviewed changes

janvorli reviewed Apr 10, 2026

View reviewed changes

src/coreclr/vm/interpexec.cpp Show resolved Hide resolved

janvorli reviewed Apr 10, 2026

View reviewed changes

Copilot AI review requested due to automatic review settings April 10, 2026 16:07

Copilot started reviewing on behalf of kotlarmilos April 10, 2026 16:09 View session

Copilot AI reviewed Apr 10, 2026

View reviewed changes

src/libraries/tests.proj Outdated Show resolved Hide resolved

src/coreclr/vm/interpexec.cpp Show resolved Hide resolved

Update tracking issue to dotnet#124344 in tests.proj

54ad681

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

Copilot AI mentioned this pull request Apr 11, 2026

Fix WasmTestOnChrome: exclude browser/wasi from gen-debug-dump-docs.py helix post-command #126789

Open

1 task

Conversation

kotlarmilos commented Mar 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Uh oh!

dotnet-policy-service bot commented Mar 27, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

kotlarmilos commented Mar 30, 2026

Uh oh!

azure-pipelines bot commented Mar 30, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

kotlarmilos commented Apr 9, 2026

Uh oh!

azure-pipelines bot commented Apr 9, 2026

Uh oh!

Uh oh!

janvorli Apr 10, 2026

Choose a reason for hiding this comment

Uh oh!

kotlarmilos Apr 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

janvorli Apr 10, 2026

Choose a reason for hiding this comment

Uh oh!

kotlarmilos Apr 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kotlarmilos Apr 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

janvorli Apr 10, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

kotlarmilos commented Mar 27, 2026 •

edited

Loading

kotlarmilos Apr 10, 2026 •

edited

Loading

kotlarmilos Apr 10, 2026 •

edited

Loading

kotlarmilos Apr 10, 2026 •

edited

Loading