Enable feature stubprecode dynamic helpers again #115746

davidwrighton · 2025-05-19T21:56:12Z

Use the barrier approach to make this safe. Adding an address dependency got into lots of details around possibly needing an entire new stub type, and several paths through this code already depend on the concept of having a barrier here. To support modifying the stubs without inserting lots of magic number, rework how stubs are identified to just do a complete compare of the entire assembly stub in a way which will be easily to generalize to the cDAC.

In addition, make changes to make the FixupPrecode path safe from the same memory ordering concerns. Since the FixupPrecode always goes through a specified entrypoint, we can slide the barrier into only the slow path as long as we ensure that the Target field written with a VolatileStore after the MethodDesc and PrecodeFixupThunk fields are set. To ensure that the barrier on the writer side has happened before with hit the barrier on the reader, we ensure that the Target field is initialized into an infinite loop before the code associated with the FixupPrecode is mapped into the process.

Fixes #113810

… stream of assembly data instead of the current approach.

…e DAC. cDAC isn't done yet, but it should be simpler than the confusion that is today's implementation.

dotnet-policy-service · 2025-05-19T21:56:59Z

Tagging subscribers to this area: @mangod9
See info in area-owners.md if you want to be subscribed.

…STUBPRECODE_DYNAMIC_HELPERS_again

…gain

davidwrighton · 2025-05-21T20:32:59Z

/azp list

azure-pipelines · 2025-05-21T20:33:05Z

CI/CD Pipelines for this repository: runtime-coreclr outerloop runtime-coreclr jitstress runtime-coreclr jitstressregs runtime-coreclr jitstress2-jitstressregs runtime-coreclr gcstress0x3-gcstress0xc runtime-coreclr gcstress-extra runtime-coreclr r2r-extra runtime-coreclr jitstress-isas-x86 runtime-coreclr jitstress-isas-arm runtime-coreclr jitstressregs-x86 runtime-coreclr libraries-jitstressregs runtime-coreclr libraries-jitstress2-jitstressregs runtime-coreclr r2r runtime-coreclr runincontext runtime-coreclr crossgen2 runtime-libraries-coreclr outerloop runtime-libraries-coreclr outerloop-windows runtime-libraries-coreclr outerloop-linux runtime-libraries-coreclr outerloop-osx runtime runtime-libraries enterprise-linux runtime-libraries stress-http runtime-libraries stress-ssl runtime-dev-innerloop runtime-coreclr crossgen2 outerloop coreclr-release-outerloop-nightly runtime-coreclr crossgen2-composite runtime-jit-experimental runtime-coreclr libraries-jitstress dotnet-linker-tests runtime-coreclr ilasm runtime-coreclr crossgen2-composite gcstress runtime-coreclr pgo runtime-coreclr libraries-pgo Antigen runtime-community Fuzzlyn runtime-coreclr superpmi-replay runtime-wasm runtime-coreclr superpmi-diffs runtime-coreclr superpmi-asmdiffs-checked-release runtime-extra-platforms jit-cfg runtime-wasm-perf runtime-llvm runtime-coreclr jitstress-random runtime-coreclr libraries-jitstress-random runtime-android-grpc-client-tests runtime-wasm-non-libtests runtime-android runtime-androidemulator runtime-ioslike runtime-ioslikesimulator runtime-linuxbionic runtime-maccatalyst runtime-coreclr pgostress runtime-coreclr jitstress-isas-avx512 runtime-libraries-mono outerloop runtime-sanitized runtime-wasm-dbgtests runtime-wasm-optional runtime-nativeaot-outerloop runtime-coreclr superpmi-collect-test runtime-diagnostics

davidwrighton · 2025-05-21T20:33:20Z

/azp run runtime-diagnostics

azure-pipelines · 2025-05-21T20:33:34Z

Azure Pipelines successfully started running 1 pipeline(s).

Copilot

Pull Request Overview

This PR adds support for dynamic helper stub precodes (version 3) by switching to a full-byte-pattern comparison approach and integrates it into both the managed contracts and native runtime metadata.

Introduces PrecodeStubs_3_Impl with ReadBytesAndCompare for stub/fixup detection and updates the factory to return version 3.
Extends PrecodeMachineDescriptor (native) with full stub/fixup byte arrays and ignore masks, and initializes them in CDacPlatformMetadata.
Updates allocation and detection in precode.h/.cpp, VM startup, DAC tables, and ARM64 thunk templates to use the new patterns.

Reviewed Changes

Copilot reviewed 22 out of 22 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
PrecodeStubs_3.cs	New v3 contract API using full-byte comparisons
PrecodeStubs_2.cs, PrecodeStubs_1.cs	Ensure v2/v1 chaining still works
PrecodeStubsFactory.cs	Register version 3 in factory switch
readytoruninfo.cpp, loaderallocator.hpp	Switch dynamic helper allocation to `AllocStub()`
precode.h, precode.cpp	Remove magic offsets, add `Is*ByASM_DAC`, use full-byte compare
cdacplatformmetadata.{hpp,cpp}, dacvars.h, datadescriptor.h	Add new DAC fields & initialization for byte-pattern data
arm64 thunktemplates.*	Insert memory-barrier (`dmb ishld`) for stubs
contracts.jsonc	Bump PrecodeStubs contract version to 3
dactable.cpp, enummem.cpp	Include new platform metadata header
clrfeatures.cmake	Enable `FEATURE_STUBPRECODE_DYNAMIC_HELPERS` on AMD64/ARM64
PrecodeStubs.md	Document v3 byte-pattern approach

src/coreclr/vm/precode.h

docs/design/datacontracts/PrecodeStubs.md

Copilot · 2025-05-21T21:52:06Z

src/coreclr/vm/cdacplatformmetadata.cpp

 #endif
 }

+void CDacPlatformMetadata::InitPrecodes()
+{
+    PrecodeMachineDescriptor::Init(&(&g_cdacPlatformMetadata)->precode);


The expression &(&g_cdacPlatformMetadata)->precode is overly verbose. It can be simplified to &g_cdacPlatformMetadata.precode for clarity.

Suggested change

PrecodeMachineDescriptor::Init(&(&g_cdacPlatformMetadata)->precode);

PrecodeMachineDescriptor::Init(&g_cdacPlatformMetadata.precode);

src/coreclr/vm/precode.cpp

Co-authored-by: Copilot <[email protected]>

max-charlamb · 2025-05-21T22:37:45Z

/azp run runtime-diagnostics

azure-pipelines · 2025-05-21T22:37:55Z

Azure Pipelines successfully started running 1 pipeline(s).

jkotas · 2025-05-21T22:49:23Z

Any numbers for the worst-case perf regressions that we expect to get from this change?

jkotas · 2025-05-21T22:52:20Z

src/coreclr/vm/arm64/thunktemplates.S

        ldr x10, DATA_SLOT(StubPrecode, Target)
        ldr x12, DATA_SLOT(StubPrecode, SecretParam)
        br x10
+        brk     0xf000      // Stubs need to be 24-byte in size to allow for the data to be 3 pointers
+        brk     0xf000      // Stubs need to be 24-byte in size to allow for the data to be 3 pointers
    LEAF_END_MARKED StubPrecodeCode\STUB_PAGE_SIZE

    LEAF_ENTRY FixupPrecodeCode\STUB_PAGE_SIZE


Why can we skip the barrier for FixupPrecode?

jkotas · 2025-05-21T22:56:01Z

src/coreclr/vm/arm64/thunktemplates.S

@@ -120,9 +120,12 @@ LEAF_END_MARKED CallCountingStubCodeTemplate, _TEXT
    .irp STUB_PAGE_SIZE, 16384, 32768, 65536


Does FEATURE_MAP_THUNKS_FROM_IMAGE above need the same change?

Sigh, I missed that part of the merge. Yes.

jkotas · 2025-05-21T23:16:21Z

src/coreclr/vm/arm64/thunktemplates.S

@@ -120,9 +120,12 @@ LEAF_END_MARKED CallCountingStubCodeTemplate, _TEXT
    .irp STUB_PAGE_SIZE, 16384, 32768, 65536

    LEAF_ENTRY StubPrecodeCode\STUB_PAGE_SIZE
+        dmb ishld


Do we need a counter-part barrier on the writer sides (e.g. in DynamicHelperFixup)?

Yes. I had written... or at least thought about writing it. I'll have that put together soon.

janvorli · 2025-05-23T09:14:03Z

src/coreclr/vm/arm64/thunktemplates.S

@@ -30,11 +30,11 @@
    IN_PAGE_INDEX = 0
    .rept STUB_PRECODE_NUM_THUNKS_PER_MAPPING

+    dmb ishld


Do we need a separate barrier instruction? Would it be possible to use ldar instruction instead of the ldr to achieve the same thing?

It would need to be an ldar that loads the address of the stub, not an ldar in the stub.

…STUBPRECODE_DYNAMIC_HELPERS_again

…gain

risc-vv · 2025-06-09T08:58:19Z

@dotnet/samsung Could you please take a look? These changes may be related to riscv64.

risc-vv · 2025-06-10T10:32:51Z

RISC-V Release-CLR-QEMU: 0 / 9117 (0.00%)

=======================
      passed: 0
      failed: 9102
     skipped: 597
      killed: 15
------------------------
 TOTAL tests: 9714
VIRTUAL time: 24h 53min 38s 624ms
   REAL time: 23min 57s 66ms
=======================

report.xml, report.md, failures.xml, testclr_details.tar.zst

RISC-V Release-FX-QEMU: 0 / 262 (0.00%)

=======================
      passed: 0
      failed: 0
     skipped: 14
      killed: 262
------------------------
 TOTAL tests: 276
VIRTUAL time: 41min 36s 290ms
   REAL time: 41s 89ms
=======================

report.xml, report.md, failures.xml, testclr_details.tar.zst

RISC-V Release-CLR-VF2: 0 / 9116 (0.00%)

=======================
      passed: 0
      failed: 9101
     skipped: 597
      killed: 15
------------------------
 TOTAL tests: 9713
VIRTUAL time: 3h 3min 3s 72ms
   REAL time: 15min 30s 637ms
=======================

report.xml, report.md, failures.xml, testclr_details.tar.zst

RISC-V Release-FX-VF2: 0 / 262 (0.00%)

=======================
      passed: 0
      failed: 0
     skipped: 14
      killed: 262
------------------------
 TOTAL tests: 276
VIRTUAL time: 2min 50s 17ms
   REAL time: 17s 873ms
=======================

report.xml, report.md, failures.xml, testclr_details.tar.zst

Build information and commands

GIT: 2af00efe83a013db336ade6535e04c1c29865176
CI: 985b32219c9d1164ea1a09421e4f004672ee8c85
REPO: dotnet/runtime
BRANCH: main
CONFIG: Release
LIB_CONFIG: Release

sirntar · 2025-06-11T07:13:47Z

After running tests on riscv64, everything ended up with SEGFAULT :(
I'll attach gdb and see why

src/coreclr/vm/riscv64/thunktemplates.S

Co-authored-by: Tomasz Sowiński <[email protected]>

risc-vv · 2025-06-11T19:10:56Z

RISC-V Release-CLR-QEMU: 0 / 9119 (0.00%)

=======================
      passed: 0
      failed: 9104
     skipped: 597
      killed: 15
------------------------
 TOTAL tests: 9716
VIRTUAL time: 25h 3min 24s 295ms
   REAL time: 24min 6s 408ms
=======================

report.xml, report.md, failures.xml, testclr_details.tar.zst

RISC-V Release-FX-QEMU: 0 / 262 (0.00%)

=======================
      passed: 0
      failed: 0
     skipped: 14
      killed: 262
------------------------
 TOTAL tests: 276
VIRTUAL time: 42min 9s 648ms
   REAL time: 41s 489ms
=======================

report.xml, report.md, failures.xml, testclr_details.tar.zst

RISC-V Release-CLR-VF2: 0 / 9119 (0.00%)

=======================
      passed: 0
      failed: 9104
     skipped: 597
      killed: 15
------------------------
 TOTAL tests: 9716
VIRTUAL time: 3h 9min 3s 891ms
   REAL time: 16min 31s 660ms
=======================

report.xml, report.md, failures.xml, testclr_details.tar.zst

RISC-V Release-FX-VF2: 0 / 262 (0.00%)

=======================
      passed: 0
      failed: 0
     skipped: 14
      killed: 262
------------------------
 TOTAL tests: 276
VIRTUAL time: 1min 29s 526ms
   REAL time: 11s 879ms
=======================

report.xml, report.md, failures.xml, testclr_details.tar.zst

Build information and commands

GIT: 54b350141755a5f5ac3bb5d6a3f914c8a92404d5
CI: 8fec31ecd11a1e652234ca5056ed447368abd635
REPO: dotnet/runtime
BRANCH: main
CONFIG: Release
LIB_CONFIG: Release

sirntar

Removing fences from src/coreclr/vm/riscv64/thunktemplates.S will fix problems on rv64, but it's only a temporary solution. I'm looking for the reason why it doesn't work, but at the moment after jumping in CallDescrWorkerInternal to an invalid pTarget gdb loses track

tomeksowi

Try this, fixing 5 more load offsets got the tests running on my side

tomeksowi · 2025-06-12T14:13:26Z

src/coreclr/vm/riscv64/thunktemplates.S

+    fence r,rw
    auipc t1, 0x4
    ld t2, (StubPrecodeData__SecretParam)(t1)
    ld t1, (StubPrecodeData__Target)(t1)


Suggested change

fence r,rw

auipc t1, 0x4

ld t2, (StubPrecodeData__SecretParam)(t1)

ld t1, (StubPrecodeData__Target)(t1)

fence r,rw

auipc t1, 0x4

ld t2, (StubPrecodeData__SecretParam - 0x4)(t1)

ld t1, (StubPrecodeData__Target - 0x4)(t1)

tomeksowi · 2025-06-12T14:14:33Z

src/coreclr/vm/riscv64/thunktemplates.S

+    fence r,rw
    auipc t2, 0x4
    ld t3, (CallCountingStubData__RemainingCallCountCell)(t2)


Suggested change

fence r,rw

auipc t2, 0x4

ld t3, (CallCountingStubData__RemainingCallCountCell)(t2)

fence r,rw

auipc t2, 0x4

ld t3, (CallCountingStubData__RemainingCallCountCell - 0x4)(t2)

Please also subtract 0x4 from the two analogous loads at lines 35 and 38, (GH suggestions don't reach there)

sirntar · 2025-06-13T08:58:42Z

coreclr tests passed, ~~I'm now checking corefx, but it looks like we have a solution:~~

src/coreclr/vm/riscv64/thunktemplates.S

LEAF_ENTRY StubPrecodeCode
    fence r, rw
    auipc t1, 0x4
    ld t2, (StubPrecodeData__SecretParam - 0x4)(t1)
    ld t1, (StubPrecodeData__Target - 0x4)(t1)
    jr t1
LEAF_END_MARKED StubPrecodeCode

LEAF_ENTRY FixupPrecodeCode
    auipc t2, 0x4
    ld t2, (FixupPrecodeData__Target)(t2)
    c.jr t2

    fence r, rw
    auipc t2, 0x4
    ld t1, (FixupPrecodeData__PrecodeFixupThunk - 0xe)(t2)
    ld t2, (FixupPrecodeData__MethodDesc - 0xe)(t2)
    jr t1
LEAF_END_MARKED FixupPrecodeCode

LEAF_ENTRY CallCountingStubCode
    fence r, rw
    auipc t2, 0x4
    ld t3, (CallCountingStubData__RemainingCallCountCell - 0x4)(t2)
    lh t1, 0(t3)
    addiw t1, t1, -1
    sh t1, 0(t3)
    beq t1, zero, LOCAL_LABEL(CountReachedZero)
    ld t1, (CallCountingStubData__TargetForMethod - 0x4)(t2)
    jr t1
LOCAL_LABEL(CountReachedZero):
    ld t1, (CallCountingStubData__TargetForThresholdReached - 0x4)(t2)
    jr t1
LEAF_END_MARKED CallCountingStubCode

EDIT: all corefx tests reported stack overflow. I'm not certain if this is still only rv64 problem

davidwrighton added 2 commits May 16, 2025 23:52

Re-enable, with a dmb, and change the stub examination logic to use a…

c342e64

… stream of assembly data instead of the current approach.

Add the new logic for handling the disasm comparison in the DAC to th…

226cc59

…e DAC. cDAC isn't done yet, but it should be simpler than the confusion that is today's implementation.

github-actions bot added the area-VM-coreclr label May 19, 2025

dotnet-policy-service bot assigned davidwrighton May 19, 2025

davidwrighton added 5 commits May 19, 2025 16:39

Merge branch 'main' of github.com:dotnet/runtime into enable_FEATURE_…

5388a3d

…STUBPRECODE_DYNAMIC_HELPERS_again

More updates to latest

36e0ec1

Add cDAC support

279a458

Merge branch 'main' of github.com:dotnet/runtime into enable_FEATURE_…

eca7820

…STUBPRECODE_DYNAMIC_HELPERS_again

Adjust to having the stub templates back

3233cd5

build-analysis bot mentioned this pull request May 21, 2025

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

Open

3 tasks

davidwrighton mentioned this pull request May 21, 2025

Assert failure '!CREATE_CHECK_STRING(bSmallObjectHeapPtr || bLargeObjectHeapPtr)' #111922

Open

Merge branch 'main' into enable_FEATURE_STUBPRECODE_DYNAMIC_HELPERS_a…

8557dea

…gain

davidwrighton requested review from jkotas and max-charlamb May 21, 2025 21:49

davidwrighton marked this pull request as ready for review May 21, 2025 21:49

Copilot AI review requested due to automatic review settings May 21, 2025 21:49

Copilot AI reviewed May 21, 2025

View reviewed changes

davidwrighton and others added 4 commits May 21, 2025 15:19

Update src/coreclr/vm/precode.h

6c07b89

Co-authored-by: Copilot <[email protected]>

Update docs/design/datacontracts/PrecodeStubs.md

43e29b9

Co-authored-by: Copilot <[email protected]>

Apply suggestions from code review

721be9e

Co-authored-by: Copilot <[email protected]>

Issue issue around code copilot found

fca72ac

jkotas reviewed May 21, 2025

View reviewed changes

build-analysis bot mentioned this pull request May 22, 2025

System.Net.Quic.Tests.MsQuicTests.WriteTests failed with "System.Net.Quic.QuicException : The connection timed out from inactivity." #105177

Open

davidwrighton added 2 commits May 22, 2025 16:58

Changes to make StubPrecode have the write barrier on the writer side

051e071

Add memory ordering protection for FixupPrecode

59324fd

janvorli reviewed May 23, 2025

View reviewed changes

Merge branch 'main' of github.com:dotnet/runtime into enable_FEATURE_…

9f86bc1

…STUBPRECODE_DYNAMIC_HELPERS_again

davidwrighton marked this pull request as draft May 23, 2025 16:43

Fix build on various platforms, and reduce incorrect assert

870e7cf

build-analysis bot mentioned this pull request May 27, 2025

System.OperationCanceledException : The operation was canceled. dotnet/dnceng#5278

Open

3 tasks

davidwrighton added 4 commits May 27, 2025 15:49

Fix assertions

bc9bb86

Fix build break

07df9b1

Fix copy/paste thinko

7421c04

Fix WASM build

f2c941d

build-analysis bot mentioned this pull request May 29, 2025

Test failure: baseservices/threading/regressions/115178/115178/115178.cmd #116060

Open

Merge branch 'main' into enable_FEATURE_STUBPRECODE_DYNAMIC_HELPERS_a…

2af00ef

…gain

davidwrighton marked this pull request as ready for review June 2, 2025 16:48

tomeksowi approved these changes Jun 9, 2025

View reviewed changes

tomeksowi reviewed Jun 11, 2025

View reviewed changes

src/coreclr/vm/riscv64/thunktemplates.S Outdated Show resolved Hide resolved

Update src/coreclr/vm/riscv64/thunktemplates.S

54b3501

Co-authored-by: Tomasz Sowiński <[email protected]>

sirntar suggested changes Jun 12, 2025

View reviewed changes

tomeksowi reviewed Jun 12, 2025

View reviewed changes

	PrecodeMachineDescriptor::Init(&(&g_cdacPlatformMetadata)->precode);
	PrecodeMachineDescriptor::Init(&g_cdacPlatformMetadata.precode);

		@@ -120,9 +120,12 @@ LEAF_END_MARKED CallCountingStubCodeTemplate, _TEXT
		.irp STUB_PAGE_SIZE, 16384, 32768, 65536

Enable feature stubprecode dynamic helpers again #115746

Are you sure you want to change the base?

Enable feature stubprecode dynamic helpers again #115746

Conversation

davidwrighton commented May 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dotnet-policy-service bot commented May 19, 2025

Uh oh!

davidwrighton commented May 21, 2025

Uh oh!

azure-pipelines bot commented May 21, 2025

Uh oh!

davidwrighton commented May 21, 2025

Uh oh!

azure-pipelines bot commented May 21, 2025

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!

Copilot AI May 21, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

max-charlamb commented May 21, 2025

Uh oh!

azure-pipelines bot commented May 21, 2025

Uh oh!

jkotas commented May 21, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jkotas May 21, 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!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

risc-vv commented Jun 9, 2025

Uh oh!

risc-vv commented Jun 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sirntar commented Jun 11, 2025

Uh oh!

Uh oh!

risc-vv commented Jun 11, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sirntar left a comment

Choose a reason for hiding this comment

Uh oh!

tomeksowi left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sirntar commented Jun 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

davidwrighton commented May 19, 2025 •

edited

Loading

jkotas commented May 21, 2025 •

edited

Loading

jkotas May 21, 2025 •

edited

Loading

risc-vv commented Jun 10, 2025 •

edited

Loading

risc-vv commented Jun 11, 2025 •

edited

Loading

sirntar commented Jun 13, 2025 •

edited

Loading