Fix #3108: conversion opcodes must update runtime type tag by shai-almog · Pull Request #5009 · codenameone/CodenameOne

shai-almog · 2026-05-23T04:08:50Z

Summary

Follow-up to Fix #3108: don't rely on C arg eval order in PUTFIELD/MULTIANEWARRAY #4985. That PR fixed the PUTFIELD emission for chained assignments but the NPE in ddyer's reproducer still reproduced because the conversion opcodes (I2D, I2L, F2D, F2L, L2D) updated SP[-1].data.* while leaving the runtime type tag stale.
BC_DUP2_X1 / BC_DUP2_X2 / BC_DUP_X2 dispatch on IS_DOUBLE_WORD(SP[-1].type). After PUSH_INT + I2D (data updated, tag still CN1_TYPE_INT), DUP2_X1 took the cat-1 branch, shifted SP by +2 instead of +1, and reshuffled three slots as if all single-word. The first putfield happened to read the right operands; the second read the double's bit pattern as the object pointer and crashed.
Every cat-changing conversion (and the narrowing inverses, for symmetry) now writes SP[-1].type = CN1_TYPE_*. Parallel macros in cn1_globals.h get the same treatment so the runtime header is consistent.

Reproducer

Reduced from ddyer's program in #3108:

private double boardScaleIndex, newBoardScaleIndex;
public void init() {
    boardScaleIndex = newBoardScaleIndex = Math.min(numBoardScales - 1, (int)(isi * SCALE));
}

Bytecode: invokestatic Math.min(II)I → i2d → dup2_x1 → two putfields. Pre-fix translator emits I2D without a type-tag write, so the dup macro corrupts the stack and the second putfield NPEs on iOS.

Verification

Standalone C test mirroring the BC_* macros: second putfield's target resolved to 0x4028000000000000 (= IEEE-754 bits of 12.0) pre-fix; resolves to the correct this pointer post-fix.
Re-ran the translator on ddyer's reproducer — emitted C now contains SP[-1].type = CN1_TYPE_DOUBLE; before BC_DUP2_X1.
New regression test conversionOpcodesUpdateRuntimeTypeTag covers all 12 conversion opcodes.
Full BytecodeInstructionIntegrationTest (72 tests) passes.

Test plan

mvn test -Dtest=BytecodeInstructionIntegrationTest (72/72 pass)
Translator regenerated for reproducer; emitted C shows tag write before BC_DUP2_X1
Push translator + matching cn1_globals.h to build server and confirm with ddyer that IOS miscompilation, results in nullPointerException #3108 reproducer no longer NPEs on iOS

🤖 Generated with Claude Code

Follow-up to #4985. The PUTFIELD fix there was correct but only addressed half of the chained-assignment NPE. Repro (ddyer's original program): private double boardScaleIndex, newBoardScaleIndex; boardScaleIndex = newBoardScaleIndex = Math.min(n, (int) expr); javac emits PUSH_INT (Math.min result), I2D, DUP2_X1, two PUTFIELDs. The I2D / I2L / F2D / F2L / L2D macros only wrote SP[-1].data.* and left the runtime type tag untouched. BC_DUP2_X1 / BC_DUP2_X2 / BC_DUP_X2 dispatch via IS_DOUBLE_WORD(...) on that tag, so an I2D-produced double whose tag still reads CN1_TYPE_INT sent the dup down the cat-1 branch: - SP bumped by +2 instead of +1 - three slots reshuffled as if all single-word - first putfield happened to read the right operands (lucky) - second putfield read the double's IEEE-754 bit pattern as the object reference and crashed with NPE on iOS Reproduced end-to-end with a standalone C test that mirrors the BC_* macros: post-fix the second putfield's target resolves to the expected this pointer; pre-fix it resolved to 0x4028000000000000 (= bits of 12.0). Each cat-changing conversion now rewrites SP[-1].type to the new CN1_TYPE_*. The narrowing inverses (L2I/L2F, D2I/D2F) and the pure cat-1 conversion I2F do the same for symmetry — DUP2/DUP_X2 path can hit them in less obvious bytecode sequences. I2B/I2C/I2S already preserve CN1_TYPE_INT, so they're left alone. Regression test conversionOpcodesUpdateRuntimeTypeTag asserts each of the 12 conversion opcodes emits the corresponding SP[-1].type = CN1_TYPE_* write. Full BytecodeInstructionIntegrationTest (72 tests) passes; verified the fail-on-pre-fix-code shape too. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

shai-almog · 2026-05-23T04:16:17Z

Compared 20 screenshots: 20 matched.
✅ JavaScript-port screenshot tests passed.

github-actions · 2026-05-23T04:18:08Z

✅ Continuous Quality Report

Test & Coverage

✅ Tests: 2596 total, 0 failed, 0 skipped
📊 Line coverage: 53.91% [HTML preview] [Download]
- Lowest covered classes
  - com.codename1.nfc.NdefMessage – 0.00%
  - com.codename1.nfc.NdefRecord – 0.00%
  - com.codename1.components.OtpField – 0.00%
  - com.codename1.nfc.NfcReadOptions – 0.00%
  - com.codename1.nfc.Tag – 0.00%
  - com.codename1.security.AuthenticationOptions – 0.00%
  - com.codename1.nfc.Nfc – 0.00%
  - com.codename1.nfc.ApduResponse – 0.00%
  - com.codename1.system.SimulatorHookExecutor – 0.00%
  - com.codename1.nfc.MifareClassic – 0.00%

Static Analysis

SpotBugs [Report archive]
- ✅ ByteCodeTranslator: 0 findings (no issues)
- ✅ android: 0 findings (no issues)
- ✅ codenameone-maven-plugin: 0 findings (no issues)
- ✅ core-unittests: 0 findings (no issues)
- ✅ ios: 0 findings (no issues)
✅ PMD: 0 findings (no issues) [Report archive]
✅ Checkstyle: 0 findings (no issues) [Report archive]

Generated automatically by the PR CI workflow.

shai-almog · 2026-05-23T04:40:35Z

Compared 109 screenshots: 109 matched.
✅ Native iOS screenshot tests passed.

Benchmark Results

VM Translation Time: 0 seconds
Compilation Time: 262 seconds

Build and Run Timing

Metric	Duration
Simulator Boot	64000 ms
Simulator Boot (Run)	2000 ms
App Install	16000 ms
App Launch	12000 ms
Test Execution	315000 ms

Detailed Performance Metrics

Metric	Duration
Base64 payload size	8192 bytes
Base64 benchmark iterations	6000
Base64 native encode	509.000 ms
Base64 CN1 encode	1245.000 ms
Base64 encode ratio (CN1/native)	2.446x (144.6% slower)
Base64 native decode	269.000 ms
Base64 CN1 decode	855.000 ms
Base64 decode ratio (CN1/native)	3.178x (217.8% slower)
Base64 SIMD encode	478.000 ms
Base64 encode ratio (SIMD/native)	0.939x (6.1% faster)
Base64 encode ratio (SIMD/CN1)	0.384x (61.6% faster)
Base64 SIMD decode	420.000 ms
Base64 decode ratio (SIMD/native)	1.561x (56.1% slower)
Base64 decode ratio (SIMD/CN1)	0.491x (50.9% faster)
Image encode benchmark iterations	100
Image createMask (SIMD off)	60.000 ms
Image createMask (SIMD on)	10.000 ms
Image createMask ratio (SIMD on/off)	0.167x (83.3% faster)
Image applyMask (SIMD off)	133.000 ms
Image applyMask (SIMD on)	54.000 ms
Image applyMask ratio (SIMD on/off)	0.406x (59.4% faster)
Image modifyAlpha (SIMD off)	145.000 ms
Image modifyAlpha (SIMD on)	62.000 ms
Image modifyAlpha ratio (SIMD on/off)	0.428x (57.2% faster)
Image modifyAlpha removeColor (SIMD off)	181.000 ms
Image modifyAlpha removeColor (SIMD on)	67.000 ms
Image modifyAlpha removeColor ratio (SIMD on/off)	0.370x (63.0% faster)
Image PNG encode (SIMD off)	1113.000 ms
Image PNG encode (SIMD on)	1849.000 ms
Image PNG encode ratio (SIMD on/off)	1.661x (66.1% slower)
Image JPEG encode	916.000 ms

shai-almog · 2026-05-23T04:44:32Z

Compared 110 screenshots: 110 matched.
✅ Native iOS Metal screenshot tests passed.

Benchmark Results

VM Translation Time: 0 seconds
Compilation Time: 376 seconds

Build and Run Timing

Metric	Duration
Simulator Boot	102000 ms
Simulator Boot (Run)	1000 ms
App Install	16000 ms
App Launch	12000 ms
Test Execution	304000 ms

Detailed Performance Metrics

Metric	Duration
Base64 payload size	8192 bytes
Base64 benchmark iterations	6000
Base64 native encode	824.000 ms
Base64 CN1 encode	1704.000 ms
Base64 encode ratio (CN1/native)	2.068x (106.8% slower)
Base64 native decode	543.000 ms
Base64 CN1 decode	1331.000 ms
Base64 decode ratio (CN1/native)	2.451x (145.1% slower)
Base64 SIMD encode	748.000 ms
Base64 encode ratio (SIMD/native)	0.908x (9.2% faster)
Base64 encode ratio (SIMD/CN1)	0.439x (56.1% faster)
Base64 SIMD decode	773.000 ms
Base64 decode ratio (SIMD/native)	1.424x (42.4% slower)
Base64 decode ratio (SIMD/CN1)	0.581x (41.9% faster)
Image encode benchmark iterations	100
Image createMask (SIMD off)	61.000 ms
Image createMask (SIMD on)	20.000 ms
Image createMask ratio (SIMD on/off)	0.328x (67.2% faster)
Image applyMask (SIMD off)	365.000 ms
Image applyMask (SIMD on)	107.000 ms
Image applyMask ratio (SIMD on/off)	0.293x (70.7% faster)
Image modifyAlpha (SIMD off)	152.000 ms
Image modifyAlpha (SIMD on)	58.000 ms
Image modifyAlpha ratio (SIMD on/off)	0.382x (61.8% faster)
Image modifyAlpha removeColor (SIMD off)	226.000 ms
Image modifyAlpha removeColor (SIMD on)	143.000 ms
Image modifyAlpha removeColor ratio (SIMD on/off)	0.633x (36.7% faster)
Image PNG encode (SIMD off)	1398.000 ms
Image PNG encode (SIMD on)	1031.000 ms
Image PNG encode ratio (SIMD on/off)	0.737x (26.3% faster)
Image JPEG encode	538.000 ms

github-actions · 2026-05-23T04:56:29Z

✅ ByteCodeTranslator Quality Report

Test & Coverage

✅ Tests: 648 total, 0 failed, 2 skipped

Benchmark Results

Execution Time: 9838 ms
Hotspots (Top 20 sampled methods):
- 21.94% java.lang.String.indexOf (377 samples)
- 21.19% com.codename1.tools.translator.Parser.isMethodUsed (364 samples)
- 18.04% java.util.ArrayList.indexOf (310 samples)
- 5.36% com.codename1.tools.translator.BytecodeMethod.addToConstantPool (92 samples)
- 3.96% java.lang.Object.hashCode (68 samples)
- 2.68% java.lang.System.identityHashCode (46 samples)
- 2.39% com.codename1.tools.translator.ByteCodeClass.calcUsedByNative (41 samples)
- 1.92% com.codename1.tools.translator.ByteCodeClass.updateAllDependencies (33 samples)
- 1.92% com.codename1.tools.translator.ByteCodeClass.markDependent (33 samples)
- 1.46% com.codename1.tools.translator.Parser.generateClassAndMethodIndexHeader (25 samples)
- 1.28% com.codename1.tools.translator.Parser.getClassByName (22 samples)
- 1.11% com.codename1.tools.translator.Parser.cullMethods (19 samples)
- 0.81% com.codename1.tools.translator.BytecodeMethod.appendMethodSignatureSuffixFromDesc (14 samples)
- 0.81% java.lang.StringBuilder.append (14 samples)
- 0.76% com.codename1.tools.translator.BytecodeMethod.isMethodUsedByNative (13 samples)
- 0.70% java.lang.StringCoding.encode (12 samples)
- 0.64% java.util.TreeMap.getEntry (11 samples)
- 0.64% com.codename1.tools.translator.BytecodeMethod.optimize (11 samples)
- 0.64% com.codename1.tools.translator.BytecodeMethod.appendCMethodPrefix (11 samples)
- 0.58% java.io.FileOutputStream.writeBytes (10 samples)
⚠️ Coverage report not generated.

Static Analysis

✅ SpotBugs: no findings (report was not generated by the build).
⚠️ PMD report not generated.
⚠️ Checkstyle report not generated.

Generated automatically by the PR CI workflow.

shai-almog linked an issue May 23, 2026 that may be closed by this pull request

IOS miscompilation, results in nullPointerException #3108

Closed

shai-almog merged commit a8f8671 into master May 23, 2026
20 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix #3108: conversion opcodes must update runtime type tag#5009

Fix #3108: conversion opcodes must update runtime type tag#5009
shai-almog merged 1 commit into
masterfrom
fix-3108-conversion-type-tag

shai-almog commented May 23, 2026

Uh oh!

shai-almog commented May 23, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented May 23, 2026

Uh oh!

shai-almog commented May 23, 2026 •

edited

Loading

Uh oh!

shai-almog commented May 23, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented May 23, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

shai-almog commented May 23, 2026

Summary

Reproducer

Verification

Test plan

Uh oh!

shai-almog commented May 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented May 23, 2026

✅ Continuous Quality Report

Test & Coverage

Static Analysis

Uh oh!

shai-almog commented May 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Benchmark Results

Build and Run Timing

Detailed Performance Metrics

Uh oh!

shai-almog commented May 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Benchmark Results

Build and Run Timing

Detailed Performance Metrics

Uh oh!

github-actions Bot commented May 23, 2026

✅ ByteCodeTranslator Quality Report

Test & Coverage

Benchmark Results

Static Analysis

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

shai-almog commented May 23, 2026 •

edited

Loading

shai-almog commented May 23, 2026 •

edited

Loading

shai-almog commented May 23, 2026 •

edited

Loading