JIT: Devirtualize non-shared generic virtual methods #122023

hez2010 · 2025-11-27T17:12:36Z

Enable devirtualization support for generic virtual methods.

When we see a base method having a method instantiation, we use FindOrCreateAssociatedMethodDesc to obtain the devirted method.

Also introduced a jit knob so that it can be turned off at any time.

AOT support is not included in this PR, which needs additional work in managed type system.

Also, if we end up with an instantiating stub (i.e. a shared generic method that requires runtime lookup), we don't have the correct generic context so we need to bail out for now.

Codegen example:

public class IntProcessor : VirtualGenericClass, IVritualGenericInterface
{
    public override void Process<T>(T item)
    {
        Console.WriteLine(item.ToString());
    }
}

public interface IVritualGenericInterface
{
    void Process<T>(T item) where T : notnull;
}

public static void Test<T>(IVritualGenericInterface ifce, T item) where T : notnull
{
    ifce.Process(item);
}

public static void Test<T>(VirtualGenericClass baseClass, T item) where T : notnull
{
    baseClass.Process(item);
}

static void Test()
{
    IVritualGenericInterface i = new IntProcessor();
    Test(i, 42);

    VirtualGenericClass c = new IntProcessor();
    Test(c, 42);
}

Codegen diff:

 G_M27646_IG01:  ;; offset=0x0000
-       push     rsi
        push     rbx
-       sub      rsp, 40
+       sub      rsp, 32
-                                                ;; size=6 bbWeight=1 PerfScore 2.25
+                                                ;; size=5 bbWeight=1 PerfScore 1.25
-G_M27646_IG02:  ;; offset=0x0006
+G_M27646_IG02:  ;; offset=0x0005
-       mov      rbx, 0x7FFE0803F318      ; Program+IntProcessor
+       mov      rbx, 0x221CA429C38      ; 'System.Int32'
        mov      rcx, rbx
-       call     CORINFO_HELP_NEWSFAST
+       call     [System.Console:WriteLine(System.Object)]
-       mov      rsi, rax
+       mov      ecx, 42
-       mov      rcx, rsi
+       call     [System.Number:Int32ToDecStr(int):System.String]
-       mov      rdx, 0x7FFE0803F100      ; Program+IVritualGenericInterface
+       mov      rcx, rax
-       mov      r8, 0x7FFE0803F6A8      ; token handle
+       call     [System.Console:WriteLine(System.String)]
-       call     CORINFO_HELP_VIRTUAL_FUNC_PTR
-       mov      rcx, rsi
-       mov      edx, 42
-       call     rax
        mov      rcx, rbx
-       call     CORINFO_HELP_NEWSFAST
+       call     [System.Console:WriteLine(System.Object)]
-       mov      rbx, rax
+       mov      ecx, 42
-       mov      rcx, rbx
+       call     [System.Number:Int32ToDecStr(int):System.String]
-       mov      rdx, 0x7FFE0803EF30      ; Program+VirtualGenericClass
+       mov      rcx, rax
-       mov      r8, 0x7FFE0803F8A8      ; token handle
+       call     [System.Console:WriteLine(System.String)]
-       call     CORINFO_HELP_VIRTUAL_FUNC_PTR
-       mov      rcx, rbx
-       mov      edx, 42
-       call     rax
        nop
-                                                ;; size=109 bbWeight=1 PerfScore 14.00
+                                                ;; size=69 bbWeight=1 PerfScore 20.00
-G_M27646_IG03:  ;; offset=0x0073
+G_M27646_IG03:  ;; offset=0x004A
-       add      rsp, 40
+       add      rsp, 32
        pop      rbx
-       pop      rsi
        ret
-                                                ;; size=7 bbWeight=1 PerfScore 2.25
+                                                ;; size=6 bbWeight=1 PerfScore 1.75

Contributes to #112596

cc: @dotnet/jit-contrib

Copilot

Pull request overview

This PR enables devirtualization support for generic virtual methods in the JIT compiler, allowing calls to generic virtual methods to be devirtualized when the exact type is known at JIT time. This optimization eliminates virtual dispatch overhead and enables further optimizations like inlining.

Key changes:

Removes the assertion that previously blocked generic method devirtualization
Generalizes array interface devirtualization to support generic virtual methods by renaming wasArrayInterfaceDevirt to needsMethodContext
Adds runtime lookup support for generic method instantiation parameters

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
src/coreclr/vm/jitinterface.cpp	Removes assertion blocking generic method devirtualization and adds logic to handle generic virtual methods using `FindOrCreateAssociatedMethodDesc`
src/coreclr/inc/corinfo.h	Renames `wasArrayInterfaceDevirt` to `needsMethodContext` to generalize the field meaning
src/coreclr/jit/jitconfigvalues.h	Adds `JitEnableGenericVirtualDevirtualization` configuration flag to control the feature
src/coreclr/jit/gentree.h	Adds `IsGenericVirtual()` helper method to identify generic virtual method calls
src/coreclr/jit/gentree.cpp	Updates `IsDevirtualizationCandidate()` to include generic virtual methods (non-AOT only)
src/coreclr/jit/importercalls.cpp	Implements devirtualization logic for generic virtual methods including runtime lookup handling, updates comments and variable names, and introduces DEVIRT label for control flow
src/coreclr/jit/inline.h	Renames `arrayInterface` field to `needsMethodContext` in `InlineCandidateInfo` struct
src/coreclr/jit/indirectcalltransformer.cpp	Updates to use renamed `needsMethodContext` field
src/coreclr/tools/Common/JitInterface/CorInfoTypes.cs	Updates managed struct definition to match renamed field
src/coreclr/tools/Common/JitInterface/CorInfoImpl.cs	Updates to use renamed field in managed implementation
src/coreclr/tools/superpmi/superpmi-shared/agnostic.h	Updates SuperPMI data structure with renamed field
src/coreclr/tools/superpmi/superpmi-shared/methodcontext.cpp	Updates SuperPMI recording/replay to use renamed field

src/coreclr/jit/importercalls.cpp

hez2010 · 2025-11-28T00:52:25Z

@MihuBot

jakobbotsch · 2025-11-28T09:54:57Z

src/coreclr/jit/importercalls.cpp

+            // If we have a RUNTIMELOOKUP helper call for the method handle,
+            // we need to pass that as the inst param instead.
+            CallArg* const methHndArg  = call->gtArgs.FindWellKnownArg(WellKnownArg::RuntimeMethodHandle);
+            GenTree* const methHndNode = methHndArg != nullptr ? methHndArg->GetEarlyNode() : nullptr;
+            if (methHndNode && methHndNode->OperIs(GT_RUNTIMELOOKUP))
+            {
+                instParam = methHndNode;
+            }


Can you explain this behavior?

It mixes syntax and semantics. Nothing guarantees that a runtime lookup appears as a GT_RUNTIMELOOKUP node. It could have been stored to a local and a number of other things.

Pushed a new commit including comments to clarify this behavior. Also updated the description of the PR.

src/coreclr/vm/jitinterface.cpp

jakobbotsch · 2025-12-10T16:58:21Z

src/coreclr/jit/importercalls.cpp

+        if (isGenericVirtual)
+        {
+            JITDUMP("Shared generic virtual devirt not supported yet\n");
+            return;
+        }


The layering feels wrong to me. How can the VM side return an instantiating stub if the base method requires a runtime lookup? There is no way to create an instantiating stub at devirtualization time in that scenario.

I think it would make more sense if resolveVirtualMethod never returns an instantiation stub for GVMs.

Moved it to VM so now the VM should never return an instantiating stub.

jakobbotsch · 2025-12-11T13:21:02Z

src/coreclr/vm/jitinterface.cpp

+        pDevirtMD = pDevirtMD->FindOrCreateAssociatedMethodDesc(
+            pDevirtMD, pExactMT, pExactMT->IsValueType() && !pDevirtMD->IsStatic(), pBaseMD->GetMethodInstantiation(), false);
+
+        // We still can't handle shared generic methods because we don't have
+        // the right generic context for runtime lookup.
+        // TODO: Remove this limitation.
+        if (pDevirtMD->IsInstantiatingStub())


If I understand correctly this will sometimes create odd instantiating stubs. For example, it may create an instantiating stub that instantiates over System.__Canon if the JIT got the method desc from a runtime lookup.

I do not think we need to create any instantiation stubs at all here. Should we instead pass allowInstParam = true to FindOrCreateAssociatedMethodDesc and then change this check to check if the returned function is shared?

Also since the declaration takes BOOL I think it should be TRUE/FALSE instead.

Yeah, since we are now rejecting any shared generic methods, I think it's okay to pass allowInstParam = true.
But if we want to extend it to support shared generic methods as well, we will need to revisit it to allow it creating an instantiating stub again, and bail if it returns an instantiating stub over System.__Canon.

Resolved in babad30

Yeah, since we are now rejecting any shared generic methods, I think it's okay to pass allowInstParam = true.
But if we want to extend it to support shared generic methods as well, we will need to revisit it to allow it creating an instantiating stub again, and bail if it returns an instantiating stub over System.__Canon.

I do not think that we should need to create instantiation stubs and then have the JIT remove them again. Instead we should make CORINFO_DEVIRTUALIZATION_INFO return the info necessary to let the JIT properly compute the generic context to pass. If the base method is a runtime lookup this will be some modified version of that runtime lookup. If the base method's instantiation is statically known then it will be some constant generic context (likely exactContext that already exists).

In either case I do not think we need to create an instantiating stub to then remove it again (and in the former case we can't even do that).

hez2010 · 2025-12-11T15:43:42Z

Do you have plans to work on that part in a later PR?

@MichalStrehovsky FYI I have a local branch that brings the non-shared GVM devirt support to R2R. NAOT still needs extra work in the JIT before it can be handled. I'll open another PR after this one merged, so that at least the changes to the managed type system can be reviewed first.

src/coreclr/vm/jitinterface.cpp

jakobbotsch

LGTM apart from that. Thanks!
@davidwrighton can you take a look too?

davidwrighton · 2026-01-05T18:33:10Z

@hez2010 sorry about the delay, I got stuck in a bunch of work just before the holidays and wasn't able to do the review before vacation began.

jakobbotsch · 2026-01-06T18:55:02Z

I'm guessing the crossgen issues are related, I don't see them on other PRs. Can you see if you can reproduce them?

hez2010 · 2026-01-08T16:53:07Z

I'm guessing the crossgen issues are related, I don't see them on other PRs. Can you see if you can reproduce them?

I managed to reproduce the freebsd one on my local cross-build environment. Will investigate it soon to see whether it's related or not.

hez2010 · 2026-01-09T13:06:42Z

We hit the assertion here:

/home/i/runtime/src/coreclr/jit/importercalls.cpp:8666
Assertion failed 'call->IsVirtualStub()' in 'System.Threading.StackHelper:CallOnEmptyStack[System.__Canon,System.__Canon,System.__Canon,System.__Canon](System.Func`4[System.__Canon,System.__Canon,System.__Canon,System.__Canon],System.__Canon,System.__Canon,System.__Canon):System.__Canon' during 'Importation' (IL size 116; hash 0x7bf56612; FullOpts)

/home/i/runtime/src/coreclr/jit/importercalls.cpp:8666
Assertion failed 'call->IsVirtualStub()' in 'System.Threading.StackHelper:CallOnEmptyStack[System.__Canon,System.__Canon](System.Func`2[System.__Canon,System.__Canon],System.__Canon):System.__Canon' during 'Importation' (IL size 102; hash 0x657768d4; FullOpts)

https://github.com/hez2010/runtime/blob/40d7569852bd8cecef706532d374ab0e8f7e53f2/src/coreclr/jit/importercalls.cpp#L8664-L8670

I think GVM applies the same with virtual stub, where in R2R mode, we might see GVM calls to non-virtuals. I think we should change the assertion to call->IsVirtualStub() || call->IsGenericVirtual(). What do you think? @jakobbotsch

jakobbotsch · 2026-01-09T13:30:04Z

I think GVM applies the same with virtual stub, where in R2R mode, we might see GVM calls to non-virtuals. I think we should change the assertion to call->IsVirtualStub() || call->IsGenericVirtual(). What do you think? @jakobbotsch

Seems reasonable to me.

src/coreclr/jit/importercalls.cpp

hez2010 added 3 commits November 27, 2025 23:34

Rename wasArrayInterfaceDevirt to needsMethodContext

fbff8d6

Devirtualize generic virtual methods

729b269

Disable GVM devirt for AOT

5c15b23

Copilot AI review requested due to automatic review settings November 27, 2025 17:12

hez2010 requested a review from MichalStrehovsky as a code owner November 27, 2025 17:12

dotnet-policy-service bot added the community-contribution Indicates that the PR has been added by a community member label Nov 27, 2025

github-actions bot added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Nov 27, 2025

Copilot started reviewing on behalf of hez2010 November 27, 2025 17:13 View session

Copilot finished reviewing on behalf of hez2010 November 27, 2025 17:16

Copilot AI reviewed Nov 27, 2025

View reviewed changes

src/coreclr/jit/importercalls.cpp Outdated Show resolved Hide resolved

src/coreclr/jit/importercalls.cpp Outdated Show resolved Hide resolved

src/coreclr/jit/importercalls.cpp Outdated Show resolved Hide resolved

src/coreclr/jit/importercalls.cpp Outdated Show resolved Hide resolved

Unblock late devirt as well

6e678a5

hez2010 force-pushed the gvm-devirt-2 branch from 530e75d to 6e678a5 Compare November 27, 2025 17:26

hez2010 added 6 commits November 28, 2025 02:42

Set single-def

c488b84

Address feedbacks

0ca8b93

Merge branch 'main' into gvm-devirt-2

b396d26

Meh

0ac9d6d

JIT format

8976d4a

JIT format again

8d0845e

This was referenced Nov 27, 2025

LibraryImportGenerator.Unit.Tests crashing on linux-x64 mono interpreter #100800

Open

[8.0] GC Large array test failure #114222

Open

Check methHndArg

1d389f1

MihuBot mentioned this pull request Nov 28, 2025

[JitDiff X64] [hez2010] JIT: Devirtualize generic virtual methods MihuBot/runtime-utils#1649

Open

hez2010 marked this pull request as draft November 28, 2025 02:33

This was referenced Nov 28, 2025

CryptographicException in NonPowerOfTwoKeySizeOaepRoundtrip test #120606

Open

SIGSEGV in System.Net.Quic.Functional.Tests #121575

Closed

jakobbotsch reviewed Nov 28, 2025

View reviewed changes

hez2010 added 3 commits November 29, 2025 23:54

Rework how inst param being embedded

217c9f2

Bail out GVM devirt in managed type system instead

4b64c2e

Remove the out-dated comments

e8668a1

hez2010 mentioned this pull request Dec 10, 2025

JIT: Devirtualization for generic virtual methods #112596

Open

9 tasks

jakobbotsch reviewed Dec 10, 2025

View reviewed changes

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

jakobbotsch reviewed Dec 10, 2025

View reviewed changes

hez2010 added 2 commits December 11, 2025 21:49

Rename to needsMethodContext

ac3b74c

Never return an instantiation stub

ed99ebc

jakobbotsch reviewed Dec 11, 2025

View reviewed changes

Do not create an instantiating stub

babad30

hez2010 force-pushed the gvm-devirt-2 branch from 66f7a79 to babad30 Compare December 11, 2025 15:38

Merge branch 'main' into gvm-devirt-2

036e446

build-analysis bot mentioned this pull request Dec 11, 2025

"Over release of MOW - COM" assert in COM wrappers #122447

Closed

Copilot AI mentioned this pull request Dec 12, 2025

Disable UniqueMarshalling COM tests causing "Over release of MOW - COM" assertion #122473

Closed

jakobbotsch reviewed Dec 12, 2025

View reviewed changes

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

jakobbotsch approved these changes Dec 12, 2025

View reviewed changes

jakobbotsch requested a review from davidwrighton December 12, 2025 10:31

hez2010 added 2 commits December 12, 2025 19:56

Always set needsMethodContext and isInstantiatingStub to false

e22692c

Merge branch 'main' into gvm-devirt-2

d2af9f4

davidwrighton approved these changes Jan 5, 2026

View reviewed changes

jakobbotsch closed this Jan 6, 2026

jakobbotsch reopened this Jan 6, 2026

build-analysis bot mentioned this pull request Jan 6, 2026

iOS.Device test WorkItemExecutions #122874

Open

Merge branch 'main' into gvm-devirt-2

40d7569

jakobbotsch reviewed Jan 9, 2026

View reviewed changes

src/coreclr/jit/importercalls.cpp Outdated Show resolved Hide resolved

Address the crossgen2 assertion

7b4c2c8

jakobbotsch approved these changes Jan 9, 2026

View reviewed changes

JIT: Devirtualize non-shared generic virtual methods #122023

Are you sure you want to change the base?

JIT: Devirtualize non-shared generic virtual methods #122023

Uh oh!

Conversation

hez2010 commented Nov 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

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!

Uh oh!

hez2010 commented Nov 28, 2025

Uh oh!

jakobbotsch Nov 28, 2025

Choose a reason for hiding this comment

Uh oh!

This comment was marked as outdated.

Uh oh!

This comment was marked as outdated.

Uh oh!

hez2010 Nov 30, 2025 • 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!

jakobbotsch Dec 10, 2025 • 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!

hez2010 Dec 11, 2025

Choose a reason for hiding this comment

Uh oh!

jakobbotsch Dec 11, 2025 • 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!

jakobbotsch Dec 11, 2025 • 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!

hez2010 Dec 11, 2025 • 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!

hez2010 Dec 11, 2025

Choose a reason for hiding this comment

Uh oh!

jakobbotsch Dec 12, 2025 • 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!

hez2010 commented Dec 11, 2025

Uh oh!

Uh oh!

jakobbotsch left a comment

Choose a reason for hiding this comment

Uh oh!

davidwrighton commented Jan 5, 2026

Uh oh!

jakobbotsch commented Jan 6, 2026

Uh oh!

hez2010 commented Jan 8, 2026

Uh oh!

hez2010 commented Jan 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jakobbotsch commented Jan 9, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

hez2010 commented Nov 27, 2025 •

edited

Loading

hez2010 Nov 30, 2025 •

edited

Loading

jakobbotsch Dec 10, 2025 •

edited

Loading

jakobbotsch Dec 11, 2025 •

edited

Loading

jakobbotsch Dec 11, 2025 •

edited

Loading

hez2010 Dec 11, 2025 •

edited

Loading

jakobbotsch Dec 12, 2025 •

edited

Loading

hez2010 commented Jan 9, 2026 •

edited

Loading