[DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D10 by NathanNeurotic · Pull Request #560 · NathanNeurotic/Open-PS2-Loader

NathanNeurotic · 2026-08-29T12:10:58Z

Replaces #543. GitHub could not reopen that draft after its head branch was repaired to remove an accidental CRLF-expanded publication. This PR points at the repaired branch.

What this draft contains

This PR carries the controlled Open PS2 Loader C-to-C++ parity migration through Wave D10.

Phase 0: shared C/C++ flags, C++ source selection, dependency tracking, truthful provenance, and the nbns.cpp linkage spike.
Wave A: 10 leaf/format translation units.
Wave B: 13 language/config/util/I/O/support/archive/VMC/cheat/application/favourites translation units.
Wave C1: bdmsupport, hddsupport, ethsupport, udpfssupport, mmcesupport, and vcdsupport, plus the typed ioPutSimpleAction bridge.
Wave D1: OSDHistory, fntsys, and renderman.
Wave D2: textures.
Wave D3: data-only dialogs.
Wave D4: pad and the public controller header's C-linkage guard.
Wave D5: asynchronous artwork-cache worker texcache.
Wave D6: per-game settings UI guigame, including the PADEMU import boundary and byte-preserved controller lookup table.
Wave D7: dialog runtime dia, including target-verified UIItem layout and preserved public C linkage.
Wave D8: theme renderer themes, with five C++-required type-recovery casts and target-verified theme ABI layouts.
Wave D9: no-reset ELF handoff elfldr_noreset, retaining the GNU VLA/argv lifetime and exact launch code while restoring explicit public C linkage.
Wave D10: launch/reset/VMC hub system, retaining public ABI and launch ordering while adding only strict-C++ compatibility at C-library, constness, byte-arithmetic, and legacy mutable-argv boundaries.

system.c passed its final high-risk source, ABI/layout, codegen, payload, clean-build, harness, and emulator-startup gates as Wave D10. Targeted real-console launch/reset validation remains outstanding.

Published checkpoints

Wave C1: 628366d1a5935b291ffb321290496dc092f559c4
Wave D1: f78d5b061129e9ecc36ef191610124dc538c5bc6
Wave D2: a36ff2cd828bf1846e3d630986de6eb108cb24bc
Wave D3: 54b741b52cc4ccdc730f64943345ee9450b9f87f
Wave D4: 1c9d9d1934980517d904e21f4efb87276ed519b4
Wave D5: 0bc05ed83fcbac344c50016e9f396f5aab6e15a6
Wave D6: 34c48da9f3fdb1419282bbf88c983c1773d7e89e
Wave D7: d4aab18f2f09fff80d5a446a5dfb86483f9cc14d
Wave D8: 3e89a0046aacb8c0763ded28c75a9714252c8aa2
Wave D9: cdedebab1f234011de137b4b6eb63d1af4d099b4
Wave D10: e1940efb8459bc829e223094f924f5f2d83f8c19

Each checkpoint was published as a non-forced fast-forward. Its remote Git tree exactly matches the independently validated local tree.

Validation

All accepted waves pass strict GNU++17 pinned PS2DEV compilation and a clean full release link without -fpermissive.

Wave C1 packed ELF: 2ddf7312df54c925427f843b8f85d45ddf60a53dc9a97bfbb5376d22b65dcea6.
Wave D1 packed ELF: 77571c0512a95b0eb02f76ea6f5c3e7044165e04e671238bfdaea550db5542d5; exact artifact reached the RiptOPL menu in PCSX2.
Waves D2 and D3 are byte-identical to D1 at both packed and stripped ELF levels. D2 changes only normalized textures.o; D3's data-only rename leaves all 232 normalized objects exact.
Wave D4 packed ELF: a2f8010ee48b19b7d417761bf34e5fffef3876f7c1e0ff4a7f49e03653f16474. Only normalized pad.o differs; final loadable section sizes, imports, global symbols, and all 20 measured stack frames match. The 12 changed opcodes are confined to equivalent boolean scheduling and symmetric inequality operand order. The exact artifact rendered the menu in PCSX2 and remained responsive.
Wave D5 packed ELF: 601a5c6d2cc6af50db85258413cbbfcec726689848b1e0cbd910a158db1e11ff. Only normalized texcache.o differs; loadable data sizes, imports, and all 23 measured stack frames match, while final .text is eight bytes smaller. A deeper optimized-tree comparison found the same canonical semantic-operation multisets in all four codegen-different functions; cacheGetTextureEx also retains the same 128-block optimized CFG and call multiset. The exact artifact rendered the menu in PCSX2 and remained responsive.
Wave D6 packed ELF: 8246e8f20aecabf2c8097e9064b96b6df006c52c6bce65f223e5f331840b6539. Only normalized guigame.o differs. All 24 non-code data payloads (2,177 bytes), 70 imports, 18 global symbol names, and all 36 stack frames match; final .text grows by 88 bytes. All 36 optimized functions retain exact block counts, call multisets, field-reference multisets, return counts, and switch counts; 23 opcode streams are exact and the other 13 are bounded to boolean lowering, zero-initialization, invariant-load scheduling, and register scheduling. The exact artifact rendered the menu in PCSX2 2.7.525 at 59.95 FPS and remained responsive.
Wave D7 packed ELF: e10e96ee874f5a7ea267d8ade457c4e39ece926488d37361c643cc03c263eb48. Only normalized dia.o differs; all eight non-code payloads (213 bytes), undefined symbols, and 18 public C names match. A target C/C++ probe produced an exact 120-byte UIItem layout record (sizeof=88, alignment 4, all nested offsets equal). Nineteen of 20 stack frames match; diaRenderUI grows 192 to 208 bytes while its code shrinks. Fifteen of 20 opcode streams are exact; all 20 optimized functions retain exact call, field, return, and switch summaries, with 18/20 exact block counts. Final .text shrinks by 72 bytes. The exact artifact rendered the menu in PCSX2 2.7.525 and remained responsive; the captured 119.84 FPS was turbo mode, not a performance baseline.
Wave D8 packed ELF: edb6bfcd7edf3b940a285815c064a6bb45e4acc806c5e8da49ff4a606c4c461c. Only normalized themes.o differs; all 28 non-code payloads (1,607 bytes), undefined-symbol sets, and 15 public C names match. A target C/C++ probe produced an exact 364-byte theme-layout record across 91 size/alignment/offset values. Fifty-nine of 60 stack frames match; non-recursive linkItemsListCoverElems grows from 32 to 48 bytes while its opcode stream shrinks. Forty-six of 60 opcode streams are exact; all 60 optimized functions retain exact blocks, calls, fields, assignments, conditions, returns, and switches. Final .text grows by 16 bytes. The exact artifact rendered the menu in PCSX2 v2.7.525 and remained live across two observations; the displayed 20-30 FPS / 35-50% speed was emulator mode, not a performance baseline.
Wave D9 packed and stripped artifacts are byte-identical to D8: packed SHA-256 edb6bfcd7edf3b940a285815c064a6bb45e4acc806c5e8da49ff4a606c4c461c, stripped opl.elf SHA-256 dc52485231e9c1cabb701167192a07035c033c1cd22768024bc8198f659c1eb7. All 232 normalized objects—including elfldr_noreset.o—match. sysLoadELFKeepIOP retains its exact unmangled 672-byte/168-word body and 64-byte dynamic stack record; all optimized IR summaries and a 116-byte target layout record are exact. The prior D8 PCSX2 menu smoke applies to these same bytes but did not invoke the keep-IOP handoff.
Wave D10 packed ELF SHA-256 is a6921e51c5f6773cb83c24e541b713f6e721f0d3df439b15bc3a4871a2bfc542; stripped opl.elf SHA-256 is 229985e86182346e3ef97a08c6f6f726026a2a79d2cd94c79ec0f124551268a7. Only normalized system.o differs (231/232 objects exact); final undefined symbols and .data/.rodata/.bss/.ctors/.dtors sizes match, while .text is 88 bytes smaller. Fifteen of 20 opcode streams are exact, including sysLaunchLoaderElf, sysLaunchDisc, and sysReset; all 20 retain exact call, field, return, and switch summaries, with only the flag helper structurally folded. Nineteen of 20 stack frames match, and critical launch/reset frames are exact. A target C/C++ layout record is byte-exact; 20/20 payload sections are semantically exact; and the flag helper passed 12 known plus 1,000,000 deterministic differential cases. The exact ELF reached and continued rendering the RiptOPL menu in PCSX2 v2.7.525, but no migrated launch/reset path was exercised.

The compiler-IR classifications are supporting evidence, not formal equivalence proofs. Generated evidence and binaries remain local under _audit/ and are intentionally not committed.

Hardware status

No PS2 hardware validation is claimed for these new checkpoints.

D4 still needs native DualShock input/rumble, both ports, reconnect, screen-transition edge handling, DS3/DS4 USB/Bluetooth PADEMU, and MX4SIO/MMCE SIO2-contention tests.

D5 still needs storage-backed successful/missing artwork, queue cancellation, rapid scrolling, theme-switch destruction, BGM contention, launch drain, and USB/HDD/SMB/MMCE/MX4SIO tests on real hardware.

D6 still needs per-game configuration round-trips, PADEMU pages, DS3/DS4 USB/Bluetooth behavior, PadMacro bitfield persistence, VMC creation, GSM/cheat settings, game launch, and MMCE/MX4SIO testing on real hardware.

D7 still needs actual dialog entry/navigation, keyboard and colour editors, Settings L1/R1 traversal, value save/reload, and real-PS2 controller/storage coverage.
D8 still needs built-in/custom theme load/reload, sparse fonts, custom assets, coverflow, per-device filtering, artwork/attribute-glyph/cache paths, settings background, device hotplug, theme switching, BGM/art contention, MMCE/MX4SIO, and real-PS2 testing.
D9 still needs real-PS2 Neutrino and POPSTARTER keep-IOP launches across applicable USB/HDD/BDM/SMB/UDPFS/MMCE/MX4SIO mounts, argv/path rejection cases, delayed teardown, child-loader rejection, target takeover, and return behavior.

D10 still needs real-PS2 standard, Neutrino, POPSTARTER, and keep-IOP launches; IOP reset/module reload; VMC validation/creation; PADEMU/DS34 operation; representative HDD/SMB/USB/MMCE/MX4SIO paths; failure/return behavior; and cold-boot timing.

Review note

The old src/bdmsupport.c blob was stored as CRLF because an embedded NUL made Git treat it as binary despite text=auto. Replacing that byte with textual \0 makes the renamed .cpp normalize to LF, so the aggregate diff shows line-ending churn. Review that file with whitespace ignored to see the substantive edits.

Current status

This PR remains a draft and must not be merged. The controlled source-conversion checkpoints through D10 are published, but the hardware-sensitive paths above are not hardware-proven and validation remains incomplete.

Review focus

Please review C/C++ linkage and ABI preservation, the typed function-pointer queue bridge, Bin2C pointer types, moved device-list initializers, const/signedness compatibility edits, D4 controller codegen classification, D5 artwork-worker ownership/codegen classification, D6 PADEMU linkage/controller-table/bitfield preservation, D7 dialog linkage, UIItem ABI, input flow, and the reviewed +16-byte diaRenderUI stack delta, plus D8 theme-pointer ownership, public C linkage, aggregate layouts, renderer/cache call flow, and the reviewed +16-byte linkItemsListCoverElems stack delta, plus D9 public C linkage, retained VLA/argv lifetime, exact inline-assembly/call sequence, embedded-ELF layouts, and byte-identical release proof, plus D10 DS34 C linkage, cheat-list read-only ABI boundary, launch/reset ordering, target layouts, payload parity, bounded compiler codegen, and explicit hardware-test gaps.

…e exception) Four changes, no source touched, C output byte-identical to baseline: - Shared C/C++ flag set: EE_CXXFLAGS := $(EE_CFLAGS) snapshot taken before Makefile.eeglobal folds both into the SDK defaults identically, plus the strict freestanding C++ set (-std=gnu++17 -fno-exceptions -fno-rtti -fno-threadsafe-statics -fno-use-cxa-atexit). No -fpermissive. - C++ source rule: src/*.cpp compiles with $(EE_CXX) $(EE_CXXFLAGS); generated asm/*.c stays on the C rule (generated sources out of scope). - EE_DEPS: fix broken substitution reference ($(patsubst %.o,%.d,...)); -include now actually pulls 232 .d files. - Truthful provenance: dirty check uses git status --porcelain (catches staged and untracked files, not just unstaged). Gate results (PS2DEVPINNED, fixed /work container path): - Rebuilt C objects 232/232 byte-identical to archived baseline; opl.elf/opl_stripped.elf/RIPTOPL.ELF hashes unchanged. - Null rebuild = 0 compiles; touching include/config.h recompiles exactly the 36 dependent TUs. - Macro parity: 13-flavour -dM matrix, per-flavour C/C++ diffs limited to language-inherent predefined macros; all feature macros (OPL_VERSION, PADEMU, RTL/IGS, GSM_1080P, DTL_T10000, OPLDIAG, debug/TTY) identical. - Dirty tree now versions -dirty correctly; DIRTY= override reproduces the baseline version string exactly.

First src/ translation unit compiled as strict GNU++17 under the Phase-0 plumbing. nbns.c renamed to nbns.cpp with zero content change; include/nbns.h gains the minimal extern "C" guard (owner-approved header exception) so the three public symbols keep C linkage for the untouched C caller ethsupport.c. Spike gate evidence (_audit/phase1/, PS2DEVPINNED, version string pinned to baseline): - Control reproduced: unchanged C++ compile mangles all three symbols and grows nbnsFindName's stack frame 48->64 bytes at identical 136/32/160-byte function sizes. - After conversion: nm shows unmangled nbnsInit/nbnsDeinit/nbnsFindName in both obj/nbns.o and linked opl.elf, at the SAME addresses as baseline. - The 64-byte frame persists post-fix and is explained by disassembly: C++ object-lifetime rules for the (struct nbnsFindNameResult *)RpcBuffer cast force a pointer spill across SifCallRpc. Same instruction count, same semantics; documented, not a defect. - 230/232 objects byte-identical to baseline; obj/ethsupport.o delta is debug-info only (nbns.h line shift) - disassembly and all data sections byte-identical. - opl.elf loadable section sizes identical to baseline; zero __cxa_*/_ZN/ _ZSt runtime symbols. - Wire protocol probe passes in both languages: sizeof(nbnsFindNameResult)=8, offsets 0/4; C++ probe references resolve to the unmangled C names. - Macro parity extended to PADEMU=0/GSM1080P=0 off-states: unchanged.

Converted under the Phase-0 plumbing, strict GNU++17, no -fpermissive: httpclient, ps2cnf, ioprp, xparam, gsm, atlas, artindex, folderbrowse, lz4, zso (src/*.c -> src/*.cpp, EE_OBJS stems unchanged). Edits beyond the rename, all linkage-preserving or cast-only: - Minimal extern "C" guards (nbns.h pattern) on 16 include/ headers the converted TUs consume: opl, ioman, xparam, ps2cnf, ioprp, config, util, system, renderman, pggsm, guigame, atlas, artindex, folderbrowse, hdd, hddsupport. Invisible to C compilation. - Source-local extern "C" wrappers (modules/ headers stay untouched per owner rule): httpclient.cpp wraps modules/network/common/httpclient.h; gsm.cpp wraps ee_core/include/coreconfig.h; lz4.cpp and zso.cpp wrap their text-included modules/isofs bodies so LZ4_*/ziso_* and the ziso_alloc/read_raw_data backends keep C linkage for the C callers (hddsupport.c, supportbase.c, util.c). - atlas.cpp: explicit (const char *) cast in atlasCopyData (C++ forbids the implicit const void* -> const char* conversion). No const drop. - modules/isofs/zso.c: two explicit (u8 *) casts in ziso_init (void* -> u8*). Valid C; IOP isofs.irx object byte-identical. - xparam.cpp: extern "C" on the two TU-local globals CheckSpecialDiscXParamTitle/ApplyExtraXParamTitle (not declared in xparam.h) so no mangled global symbols exist at all. Wave gate evidence (_audit/phase1/wave-a/, PS2DEVPINNED image, version string pinned to baseline REVISION=2657 CODE_ANCHOR=679dc31e): - Full pinned release build passes; per-TU C++ strict compiles pass. - Debug-stripped object hashes vs 679dc31 baseline: 224/232 identical; deltas exactly {nbns (spike), artindex, atlas, folderbrowse, gsm, httpclient, ps2cnf, zso}. ioprp/lz4/xparam byte-identical after strip. - gsm/folderbrowse stripped deltas are symbol-table/section-name only (mangled internal-linkage statics); code+rodata bytes identical. - Real codegen deltas are small register-allocation/scheduling/addressing drifts (atlasPlace walk, artIndexMayExist +8B, read_raw_data +16B, HttpSendGetRequest -8B, ps2cnfGetBootFile -16B, folderbrowse spill-order swaps). .su: every function's stack frame identical C vs C++. - ps2cnf: C++ lowers isgraph/isspace to newlib calls instead of the _ctype_ table; same semantics, link verified. - opl.elf: zero mangled global symbols, zero __cxa_/_ZN/_ZSt/_ZTI/_ZTV; .text +48 bytes total (includes the nbns spike delta); all other loadable section sizes identical. - Raw object hashes of untouched C TUs that include guarded headers move by DWARF line shift only (-gdwarf-2 -gz); debug-stripped comparison is the parity proof. - PCSX2 smoke: wave RIPTOPL.ELF boots to the menu at ~60fps/100% speed.

coderabbitai · 2026-08-29T12:11:12Z

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands.}

NathanNeurotic · 2026-08-29T12:40:21Z

@techwritescode

NathanNeurotic · 2026-08-29T18:27:05Z

Wave C1 checkpoint is now pushed at 628366d1a5935b291ffb321290496dc092f559c4.

Converted: bdmsupport, hddsupport, ethsupport, udpfssupport, mmcesupport, and vcdsupport.

Validation completed before publication:

strict GNU++17 pinned release build and final link passed;
232-vs-232 whole-tree stripped-object comparison changed only the six converted objects plus the intentional ioman.o bridge;
all original global C symbols were preserved;
moved 112-byte device-list initializers and callback relocations match their C controls;
converted-unit stack frames are unchanged;
exact packed ELF SHA-256 is 2ddf7312df54c925427f843b8f85d45ddf60a53dc9a97bfbb5376d22b65dcea6;
the exact artifact reaches the RiptOPL main menu in PCSX2 2.7.525.

This remains a draft checkpoint. No PS2 hardware validation is claimed, and elfldr_noreset.c / system.c are not included. For bdmsupport, use whitespace-ignored review because removing its historical embedded NUL also lets Git normalize the old CRLF blob to LF.

NathanNeurotic · 2026-08-30T02:56:53Z

Waves D1-D5 are now published as five separate non-forced fast-forward checkpoints.

Remote commits and exact trees:

D1 f78d5b061129e9ecc36ef191610124dc538c5bc6 — tree 957a50ada4d49c35cbc1969c2c2db60459e6bbd6
D2 a36ff2cd828bf1846e3d630986de6eb108cb24bc — tree 23da1fb0541c07613a68d601a49f587b0592a7a9
D3 54b741b52cc4ccdc730f64943345ee9450b9f87f — tree 85d32f205016dea77c6bc167aecd7144cfa8af36
D4 1c9d9d1934980517d904e21f4efb87276ed519b4 — tree 012559b9804028769a1fe2d940c0a40f8f589a44
D5 0bc05ed83fcbac344c50016e9f396f5aab6e15a6 — tree 5b6ad012232bc12a2ab95b2a1b5868d7c5d2bf33

Validation summary:

every wave passes strict GNU++17 clean pinned release build/link;
D2 and D3 packed/stripped ELFs are byte-identical to D1 (77571c0512a95b0eb02f76ea6f5c3e7044165e04e671238bfdaea550db5542d5);
D4 changes only normalized pad.o, preserves sections/imports/symbols/stack frames, and its exact ELF (a2f8010e…6474) reaches the PCSX2 menu;
D5 changes only normalized texcache.o, preserves loadable data/imports/stack frames, has an eight-byte-smaller .text, and its exact ELF (601a5c6d…e11ff) reaches the PCSX2 menu;
D5's four codegen-different functions have exact canonical optimized semantic-operation multisets; this is supporting evidence, not formal proof.

This PR remains draft / DO NOT MERGE. No PS2 hardware validation is claimed. D4 controller/PADEMU/SIO2 behavior and D5 storage-backed art/cancellation/theme-switch/launch-drain behavior still require targeted console testing. elfldr_noreset.c and system.c remain excluded.

NathanNeurotic · 2026-08-30T03:17:46Z

Wave D6 is now published as a separate non-forced fast-forward checkpoint.

Remote commit: 34c48da9f3fdb1419282bbf88c983c1773d7e89e
Exact remote/local tree: 79de42b4a24e521a8a152b1788030e69f9bca194
Scope: 99% rename of src/guigame.c to src/guigame.cpp, C linkage around DS3/DS4 imports, const-correct Bluetooth-version strings, and a byte-preserved C++17 rewrite of the indexed controller lookup initializer.

Validation completed before publication:

strict GNU++17 per-TU compile and clean pinned full release build/link passed;
231/232 normalized objects are exact; only guigame.o differs;
all 24 non-code data payloads (2,177 bytes), both 17-byte controller tables, 70 imports, 18 global symbol names, and all 36 stack frames match;
final .text grows by 88 bytes; 23/36 opcode streams are exact, and optimized block/call/field/return/switch summaries match for all 36 functions;
the remaining codegen is bounded to C-vs-C++ boolean lowering, local zero-initialization, invariant-load scheduling, and register scheduling;
exact packed ELF SHA-256 is 8246e8f20aecabf2c8097e9064b96b6df006c52c6bce65f223e5f331840b6539;
that exact artifact rendered the RiptOPL menu in PCSX2 2.7.525 at 59.95 FPS and remained responsive.

This remains draft / DO NOT MERGE. No PS2 hardware validation is claimed. Per-game config round-trips, PADEMU/DS3/DS4, PadMacro persistence, VMC, GSM/cheats, launch, MMCE, and MX4SIO still require targeted console testing. elfldr_noreset.c and system.c remain excluded.

NathanNeurotic · 2026-08-30T08:17:47Z

Wave D7 is now published as a separate non-forced fast-forward checkpoint.

Remote commit: d4aab18f2f09fff80d5a446a5dfb86483f9cc14d
Exact remote/local tree: ffa7b1cb154b036031bf3adb474ca5d1e53879db
Scope: 99% rename of src/dia.c to src/dia.cpp, with only the C++-required exact two-byte null initializer and malloc result cast.

Validation completed before publication:

strict target C/C++ per-TU probe and clean pinned full release build/link passed;
231/232 normalized objects are exact; only dia.o differs;
all eight non-code data payloads (213 bytes), undefined-symbol sets, and 18 public C symbol names match;
a target-compiled 120-byte UIItem layout record is byte-exact between C and C++ (sizeof=88, alignment 4, all nested union/callback/trailing-field offsets equal);
final .text shrinks by 72 bytes;
15/20 opcode streams are exact; all 20 optimized functions retain exact call, field-reference, return, and switch summaries, with 18/20 exact block counts;
19/20 stack frames match; diaRenderUI grows from 192 to 208 bytes (+16), explicitly reviewed and retained while its emitted code shrinks;
exact packed ELF SHA-256 is e10e96ee874f5a7ea267d8ade457c4e39ece926488d37361c643cc03c263eb48;
that exact artifact rendered the RiptOPL menu in PCSX2 2.7.525 and remained responsive. The captured 119.84 FPS was turbo mode, not a performance baseline.

This remains draft / DO NOT MERGE. The emulator smoke did not enter a dialog. Dialog navigation, keyboard/colour editors, Settings L1/R1 traversal, value save/reload, and real-PS2 controller/storage coverage still require targeted testing. elfldr_noreset.c and system.c remain excluded.

NathanNeurotic · 2026-08-30T09:22:36Z

Wave D8 is now published as a separate non-forced fast-forward checkpoint.

Remote commit: 3e89a0046aacb8c0763ded28c75a9714252c8aa2
Exact remote/local tree: 6a25a27d84f45c28e2feeaca5c29014186e07e7e
Scope: 99% rename of src/themes.c to src/themes.cpp, with one C++-required malloc result cast and four explicit void * to item_list_t * ownership-recovery casts.

Validation completed before publication:

strict target C/C++ per-TU probes and clean pinned full release build/link passed;
231/232 normalized objects are exact; only themes.o differs;
all 28 non-code payloads (1,607 bytes), undefined-symbol sets, and 15 public C symbol names match;
a target-compiled 364-byte theme-layout record is byte-exact between C and C++ across 91 size/alignment/offset values;
46/60 opcode streams are exact, while all 60 optimized functions retain exact block, call, field, assignment, condition, return, and switch summaries;
59/60 stack frames match; non-recursive linkItemsListCoverElems grows from 32 to 48 bytes (+16) while its emitted opcode stream shrinks;
final .text grows by 16 bytes;
exact packed ELF SHA-256 is edb6bfcd7edf3b940a285815c064a6bb45e4acc806c5e8da49ff4a606c4c461c;
that exact artifact rendered the RiptOPL main menu in PCSX2 v2.7.525 and remained live across two observations. The displayed 20-30 FPS / 35-50% speed was emulator mode, not a performance baseline.

This remains draft / DO NOT MERGE. No PS2 hardware validation is claimed. Built-in/custom themes, sparse fonts/assets, coverflow, per-device filtering, artwork/glyph/cache behavior, settings background, hotplug/theme switching, BGM contention, MMCE/MX4SIO, and real-PS2 behavior still require targeted testing. elfldr_noreset.c and system.c remain separately gated and excluded.

NathanNeurotic · 2026-08-30T13:16:05Z

Wave D9 is now published as a separate non-forced fast-forward checkpoint.

Remote commit: cdedebab1f234011de137b4b6eb63d1af4d099b4
Exact remote/local tree: 46aa35c890e3d4d0e672e88f9e03cf4c91802b7a
Scope: 98% rename of src/elfldr_noreset.c to src/elfldr_noreset.cpp, plus inclusion of system.h so sysLoadELFKeepIOP retains and type-checks its public C linkage. The GNU VLA and every handoff operation remain source-identical.

Validation completed before publication:

strict target C/C++ probes and clean pinned full release build/link passed without diagnostics;
all 232 normalized objects are byte-identical, including elfldr_noreset.o;
packed ELF SHA-256 remains edb6bfcd7edf3b940a285815c064a6bb45e4acc806c5e8da49ff4a606c4c461c;
stripped opl.elf SHA-256 remains dc52485231e9c1cabb701167192a07035c033c1cd22768024bc8198f659c1eb7;
sysLoadELFKeepIOP retains its exact unmangled 672-byte / 168-word body and exact 64-byte dynamic stack record;
optimized blocks, calls, fields, assignments, conditions, PHIs, returns, and switches are all exact;
a PS2-target 116-byte C/C++ layout record for pointer widths and both embedded-ELF structs is byte-exact.

The D9 packed ELF is literally the D8 ELF already observed at the PCSX2 menu, so a separate menu boot cannot distinguish this checkpoint. That smoke did not invoke the no-reset handoff.

This remains draft / DO NOT MERGE. No real-PS2 keep-IOP validation is claimed. Neutrino/POPSTARTER launches across live device mounts, argv/path rejection, delayed teardown, child-loader rejection, target takeover, and return behavior still require targeted hardware tests. system.c remains separately gated and excluded.

NathanNeurotic · 2026-08-31T19:36:53Z

Wave D10 is now published as the final high-risk source-conversion checkpoint.

Remote commit: e1940efb8459bc829e223094f924f5f2d83f8c19
Exact remote/validated-local tree: 60bb4419d522b517f3c8f9714724c4284950af0c
Scope: 99% rename of src/system.c to src/system.cpp, with only strict-C++ compatibility edits for DS34 C linkage, private table constness, legacy mutable argv boundaries, signedness, the source-audited read-only cheat-list ABI boundary, and explicit byte arithmetic.

Validation completed before publication:

strict target C/C++ probe and clean pinned full release build/link passed without candidate diagnostics or new C++ runtime symbols;
231/232 normalized objects are exact; only system.o differs;
final undefined-symbol sets and .data/.rodata/.bss/.ctors/.dtors sizes match; final .text is 88 bytes smaller;
15/20 function opcode streams are exact, including sysLaunchLoaderElf, sysLaunchDisc, and sysReset; all 20 retain exact call, field, return, and switch summaries;
19/20 stack frames match, with every critical launch/reset frame exact;
target C/C++ layouts are byte-exact; all 20 payload sections are semantically exact; the sole raw payload difference is literal ordering with the same string set;
the flag helper passed 12 known plus 1,000,000 deterministic differential cases;
exact packed ELF SHA-256: a6921e51c5f6773cb83c24e541b713f6e721f0d3df439b15bc3a4871a2bfc542;
that exact ELF reached and continued rendering the RiptOPL menu in PCSX2 v2.7.525.

This PR remains a draft and is not ready for review or merge. No reviewers were requested. No PS2 hardware validation is claimed. Standard, Neutrino, POPSTARTER, and keep-IOP launches; IOP reset/module reload; VMC; PADEMU/DS34; storage backends; failure/return behavior; and cold-boot timing still require targeted real-console testing.

…to phaseDout Brings PR #560's 12 commits into the ps2build-port branch so #560 can be closed: 18 src/*.c -> .cpp conversions (auto-merged; the ethsupport ps2ip/lwip include fix carried into ethsupport.cpp), header extern-C guards, wave-b strict cleanup. Merge fixups for the ps2build side: - ps2.yaml opl sources: 18 .c -> .cpp renames - opl cflags: add the Makefile EE_CXXFLAGS freestanding set (-std=gnu++17 -fno-exceptions -fno-rtti -fno-threadsafe-statics -fno-use-cxa-atexit) now that opl compiles C++ TUs

NathanNeurotic · 2026-09-04T19:19:19Z

Superseded by #602: all 13 commits from CPLUSPLUS/ROUND1 (wave-b strict cleanup + 11 Convert-to-C++ waves through D10) were merged into phaseDout as 2c65d5e. Nothing is lost - the full history is reachable from #602, which also carries the ps2build port on top. Post-merge build verified green: clean two-pass ps2build build, 722 steps, opl.elf 4,068,032 B + opl-packed.elf, 148/148 assets real.

NathanNeurotic and others added 5 commits August 27, 2026 09:03

refactor: continue C++ conversion wave (b)

be8801f

fix: complete wave-b strict C++ cleanup

530290b

NathanNeurotic mentioned this pull request Aug 29, 2026

C++ migration: Phase-0 plumbing, nbns spike, waves (a)-(b) #543

Closed

Convert storage and network support units to C++

628366d

NathanNeurotic changed the title ~~C++ migration: Phase-0 plumbing, nbns spike, waves (a)-(b)~~ C++ migration: Phase-0 plumbing through Wave C1 Aug 29, 2026

NathanNeurotic changed the title ~~C++ migration: Phase-0 plumbing through Wave C1~~ [DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave C1 Aug 29, 2026

NathanNeurotic added 5 commits August 29, 2026 19:51

Convert OSD, renderer, and font units to C++

f78d5b0

Convert texture loader to C++

a36ff2c

Convert dialog definitions to C++

54b741b

Convert controller input to C++

1c9d9d1

Convert artwork cache worker to C++

0bc05ed

NathanNeurotic changed the title ~~[DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave C1~~ [DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D5 Aug 30, 2026

Convert per-game settings UI to C++

34c48da

NathanNeurotic changed the title ~~[DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D5~~ [DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D6 Aug 30, 2026

Convert dialog runtime to C++

d4aab18

NathanNeurotic changed the title ~~[DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D6~~ [DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D7 Aug 30, 2026

Convert theme renderer to C++

3e89a00

NathanNeurotic changed the title ~~[DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D7~~ [DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D8 Aug 30, 2026

Convert no-reset ELF handoff to C++

cdedeba

NathanNeurotic changed the title ~~[DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D8~~ [DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D9 Aug 30, 2026

Convert system launch and reset paths to C++

e1940ef

NathanNeurotic changed the title ~~[DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D9~~ [DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D10 Aug 31, 2026

NathanNeurotic mentioned this pull request Sep 4, 2026

[DO NOT MERGE] C++ migration (waves through D10) + ps2build port #602

Draft

NathanNeurotic closed this Sep 4, 2026

Navigation Menu

[DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D10 - #560

[DO NOT MERGE] C++ migration: Phase-0 plumbing through Wave D10#560
NathanNeurotic wants to merge 16 commits into
rebuild/mainfrom
CPLUSPLUS/ROUND1

NathanNeurotic commented Aug 29, 2026 •

edited

Loading

Uh oh!

coderabbitai Bot commented Aug 29, 2026

Draft PR not reviewed

Uh oh!

NathanNeurotic commented Aug 29, 2026

Uh oh!

NathanNeurotic commented Aug 29, 2026

Uh oh!

NathanNeurotic commented Aug 30, 2026

Uh oh!

NathanNeurotic commented Aug 30, 2026

Uh oh!

NathanNeurotic commented Aug 30, 2026

Uh oh!

NathanNeurotic commented Aug 30, 2026

Uh oh!

NathanNeurotic commented Aug 30, 2026

Uh oh!

NathanNeurotic commented Aug 31, 2026

Uh oh!

NathanNeurotic commented Sep 4, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

NathanNeurotic commented Aug 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What this draft contains

Published checkpoints

Validation

Hardware status

Review note

Current status

Review focus

Uh oh!

coderabbitai Bot commented Aug 29, 2026

Draft PR not reviewed

Uh oh!

NathanNeurotic commented Aug 29, 2026

Uh oh!

NathanNeurotic commented Aug 29, 2026

Uh oh!

NathanNeurotic commented Aug 30, 2026

Uh oh!

NathanNeurotic commented Aug 30, 2026

Uh oh!

NathanNeurotic commented Aug 30, 2026

Uh oh!

NathanNeurotic commented Aug 30, 2026

Uh oh!

NathanNeurotic commented Aug 30, 2026

Uh oh!

NathanNeurotic commented Aug 31, 2026

Uh oh!

NathanNeurotic commented Sep 4, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

NathanNeurotic commented Aug 29, 2026 •

edited

Loading