|
|
|
|
@@ -160,3 +160,107 @@ GATE PASS
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
Only retained source edit is the information type alignment, committed as 1fe346a7. Exact counts and code bytes unchanged across every owned unit. Final DOL hash matches, no regression, no forbidden patterns, no readability warnings. All eight open functions have three fresh attempts. Four data sections remain below 100%; full data matching is not claimed.
|
|
|
|
|
|
|
|
|
|
## Round 2 baseline
|
|
|
|
|
|
|
|
|
|
HEAD 3f330a39 includes accepted #735. Origin was fetched before each unit inspection; ongoing unrelated main commits do not change assigned sources. Fresh baseline quick gate passes with data 144/1016/40/408 and exact functions 27/21/21/88. No code tuning is needed for a data-only lane. Revisit merged data ownership rather than adding source padding.
|
|
|
|
|
|
|
|
|
|
### sd_drv last merged diagnostic owner
|
|
|
|
|
|
|
|
|
|
Original lbl_8169134C owns 988 bytes starting at pool offset 0x634, but nm and source literals show 19 separate null-terminated diagnostic/format strings within that inferred aggregate. Every source byte through offset 0xa0d equals the target. Two final bytes at 0xa0e/0xa0f are linker alignment, not part of the final 54-byte error string. Target pfd_sddrv_full_format at object 0x2dbc uses relocated pool anchor r30 plus 0x9d8 immediately before OSReport. The final original string begins exactly at 0x816916f0 and ends at 0x81691726; next relocated real table st_uhs_msc_blk_func starts at 0x81691728 in the neighbouring unit. Preserve both split extents and every payload byte. Split the aggregate into its actual strings, proven by source literals, terminators and target pool-reference instructions, instead of changing the section end or fabricating padding.
|
|
|
|
|
|
|
|
|
|
Splitting all 19 real literals within the last inferred aggregate produces .data 100% and matched_data 1016 -> 3592/3592. Payload bytes and total_data are unchanged, two final zero alignment bytes remain in the same .data split. Every exact function remains exact. Quick gate PASS, zero regressions, forbidden/style 0, retail DOL hash. This is a corrected object ownership boundary supported by relocated pool-base plus 0x9d8, not a section-size reduction.
|
|
|
|
|
|
|
|
|
|
### Focus event vtable ownership
|
|
|
|
|
|
|
|
|
|
Original event vtable inferred size 0xb0 incorrectly absorbs three different compiler-emitted weak vtables and four final alignment bytes. The real event class declares four virtual methods and its compiler-generated table is 0x18. Constructor target .text relocations 0x16/0x2a point to the event table, with retained method relocations at 0x7d0/7d4/7d8/7dc. Source emits the real 0x5c PaneManager table at .data+0x7e0, 0x18 EventHandler table at +0x83c and 0x20 Interface table at +0x854. Their original retained definitions already exist at 0x816359c4, 0x81635a38 and 0x81635a50 with precisely those same sizes and weak scope. Focus make_gui_mgr target relocations at 0x2866/286e refer to the retained PaneManager table; source emits its deduplicated local weak definition. This proves separate real compiler objects, not additional event virtual slots. Split the incorrect aggregate into the real class table owners with their names, sizes and weak scope. Keep all 2168 data bytes in the same split, including discarded zero bytes and four final alignment bytes.
|
|
|
|
|
|
|
|
|
|
Focus ownership correction produces all non-text sections 100%, data 408 -> 2576/2576, exact 88 unchanged. Quick full gate PASS with zero regressions, forbidden/style 0 and correct DOL hash. No source/class/inline changes required; the existing real classes already emit the tables in exactly the original order. Original section byte counts and split extents are unchanged.
|
|
|
|
|
|
|
|
|
|
### Surface TickTimer format ownership
|
|
|
|
|
|
|
|
|
|
Correct previous-round note: the weak TickTimer report format at pool+0xd9c is real text ` : %d[ms]\n`, not zero-filled. Both raw objects contain identical 11 payload bytes, target includes one final zero alignment byte. The real inline TickTimer::report already emits its weak 11-byte string. Target RSO resolver call relocation at 0x434 references the retained TickTimer::report method. Its emitted source inline body references @STRING@ via address-half relocations at 0x4e6/0x4f6; the original has no duplicate method body. Source symbol size, literal contents and original literal terminator establish the discarded inline data owner's 11-byte extent. Add the missing weak literal owner at 0x81643aac with its real 11-byte null-terminated extent; preserve the original 3496-byte split and final alignment byte.
|
|
|
|
|
|
|
|
|
|
Surface missing weak format symbol correction produces every section 100% and matched_data 40 -> 3536/3536. Quick gate PASS, zero regressions, forbidden/style 0, correct DOL hash. Source already has its real inline TickTimer and 12-byte SurfaceManager vtable. Existing shared-header weak-zero workaround is untouched because it is unrelated to this missing ownership label and all sections now match without new source tricks.
|
|
|
|
|
|
|
|
|
|
### CDB first divergence, real file-size getter
|
|
|
|
|
|
|
|
|
|
Restore the file-size wrapper immediately before its data-size counterpart. This is supported by the existing actual inlined sequence in CDBRecordEncrypt and CDBRecordDecrypt: CDBLock, file-open diagnostic/error, CDBRecordFileGetFileSize, CDBUnlock. The backing typed function exists in the repository and original object. No invented external calls or dummy literals are added. Measure this first restoration before proceeding into the missing reduction diagnostics.
|
|
|
|
|
|
|
|
|
|
File-size restoration fixes the first divergence exactly: strings 8 and 9 now both match bytes and offsets 0x1b0/0x1ec. First divergence advances to 10 at 0x228, now the missing ReduceFileSize diagnostics. CDB original .text contains no reduction functions or bodies, original bridge/file/VF units likewise have no truncation API in source, headers, retained symbols or Ghidra export. No invented external truncation routine or placeholder error-only function will be used to emit these strings. Current source has 19 distinct diagnostics versus 47 original; missing payload remains visible and data stays 144/2640. Read-only field/keyword setters do not repair the first reduction block, so no speculative API additions are made.
|
|
|
|
|
|
|
|
|
|
File-size wrapper quick gate PASS with zero regressions and all 27 original exact functions preserved. Its first two previously misplaced diagnostics now match the original pool offsets. Restoration is supported by surviving inlined instructions and existing backing API; remaining APIs without surviving bodies are left unresolved.
|
|
|
|
|
|
|
|
|
|
### Round 2 scope and evidence audit
|
|
|
|
|
|
|
|
|
|
Repeated nm -n, objdump -t, objdump -r -j for every owned non-text section on both objects. Full output at /tmp/data-d6-r2.final-data-evidence.txt. Verified raw bytes for sd_drv, www_surface and Focus agree through the complete emitted source payload in every section. Only zero linker alignment is absent at source section ends. Their original split ranges and total_data never changed. Source's real weak class/vtable/format definitions are preserved, not replaced with zero arrays or assembly.
|
|
|
|
|
|
|
|
|
|
Remaining CDBRecord code functions CDBCryptBuffer and CDBRecordEncrypt each already have three distinct experiments in this persistent attempts log. CDBCryptBuffer retains the exact 119-instruction stream except three missing-pool offsets; repeated register/loop variants cannot create the reduction diagnostics. Encrypt has 283/284 instructions, pool offsets and stack allocation differences. Round 2 fixes the first two token-order diagnostics by a real recovered getter, but cannot recover the reduction/mutation/duplication bodies from retained instructions. No bogus truncation API, stub, unused string object, padding or ownership reduction is added for these missing bytes. Existing data total 2640 remains authoritative.
|
|
|
|
|
|
|
|
|
|
CDBRecordEncrypt | use recovered file-size wrapper in encryption | objdiff 96.616196%, instructions 284/284, diffs 185; unit code 5464, data 144, functions 27
|
|
|
|
|
|
|
|
|
|
CDBRecordEncrypt | wrapper with signed file null comparison | objdiff 97.19718%, instructions 283/284, diffs different instruction counts; unit code 5464, data 144, functions 27
|
|
|
|
|
|
|
|
|
|
CDBRecordEncrypt | wrapper with named file lifetime | objdiff 96.616196%, instructions 284/284, diffs 185; unit code 5464, data 144, functions 27
|
|
|
|
|
|
|
|
|
|
Three fresh Encrypt experiments replace the manually expanded file-size sequence with the recovered real getter. Pointer-null and named-file-lifetime variants score 96.616196%, 284/284 instructions. The signed null test proven by the original inline cmpwi retains 97.19718%, 283/284 instructions. Retain this typed getter call: a direct disassembly comparison against the previously full-gated object confirms every one of the 29 original function instruction streams is unchanged, not only the 27 exact functions. Thus the restored getter is used by live encryption code and replaces its duplicate expanded sequence. No newly invented operation or unused diagnostic object is introduced. Pool first divergence remains 10 with the first ten strings at identical offsets.
|
|
|
|
|
|
|
|
|
|
## Round 2 final clean gate
|
|
|
|
|
|
|
|
|
|
```text
|
|
|
|
|
full build: ok
|
|
|
|
|
main.dol sha1: 26116613f624061ba99c8d1a299aaa6efa85670d
|
|
|
|
|
[libs/RevoEX/src/cdb/CDBRecord] pool: DIVERGES at string 10 (mine=19 orig=47)
|
|
|
|
|
[libs/RevoEX/src/cdb/CDBRecord] objdiff: code 5464/7076 data 144/2640 functions 27/29 fuzzy 99.5483 linked code 0
|
|
|
|
|
[libs/RevoEX/src/cdb/CDBRecord] instruction-exact functions: 27/29
|
|
|
|
|
[libs/RevoEX/src/cdb/CDBRecord] section .bss size 128 match 100.0
|
|
|
|
|
[libs/RevoEX/src/cdb/CDBRecord] section .data size 2496 match 52.795387
|
|
|
|
|
[libs/RevoEX/src/cdb/CDBRecord] section .rodata size 16 match 100.0
|
|
|
|
|
[libs/RevoEX/src/cdb/CDBRecord] section .text size 7076 match 99.54833
|
|
|
|
|
[libs/RevoEX/src/cdb/CDBRecord] below 100: CDBCryptBuffer 99.97479
|
|
|
|
|
[libs/RevoEX/src/cdb/CDBRecord] below 100: CDBRecordEncrypt 97.19718
|
|
|
|
|
[libs/RevoEX/src/cdb/CDBRecord] baseline: code 5464/7076 data 144 functions 27 fuzzy 99.5472
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] pool: IDENTICAL
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] objdiff: code 8940/11760 data 3592/3592 functions 21/26 fuzzy 99.0544 linked code 0
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] instruction-exact functions: 21/26
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] section .bss size 608 match 100.0
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] section .data size 2576 match 100.0
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] section .rodata size 368 match 100.0
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] section .sbss size 40 match 100.0
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] section .text size 11760 match 99.05442
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] below 100: pfd_sddrv_init 92.326385
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] below 100: pfd_sddrv_finalize 92.63158
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] below 100: pfd_sddrv_get_total_sectors 99.5
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] below 100: pfd_sddrv_store_mbr_buf 99.75247
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] below 100: pfd_sddrv_build_fat32_mbr_bpb 94.75225
|
|
|
|
|
[libs/RVL_SDK/src/fa/driver/sd_drv] baseline: code 8940/11760 data 1016 functions 21 fuzzy 99.0544
|
|
|
|
|
[src/iplwww/www_surface] pool: IDENTICAL
|
|
|
|
|
[src/iplwww/www_surface] objdiff: code 1916/1916 data 3536/3536 functions 21/21 fuzzy 100.0000 linked code 1916
|
|
|
|
|
[src/iplwww/www_surface] instruction-exact functions: 21/21
|
|
|
|
|
[src/iplwww/www_surface] section .data size 3496 match 100.0
|
|
|
|
|
[src/iplwww/www_surface] section .sbss size 32 match 100.0
|
|
|
|
|
[src/iplwww/www_surface] section .sdata size 8 match 100.0
|
|
|
|
|
[src/iplwww/www_surface] section .text size 1916 match 100.0
|
|
|
|
|
[src/iplwww/www_surface] baseline: code 1916/1916 data 40 functions 21 fuzzy 100.0000
|
|
|
|
|
[src/scene/board/iplFocusObject] pool: IDENTICAL
|
|
|
|
|
[src/scene/board/iplFocusObject] objdiff: code 19772/20332 data 2576/2576 functions 88/89 fuzzy 99.6471 linked code 0
|
|
|
|
|
[src/scene/board/iplFocusObject] instruction-exact functions: 88/89
|
|
|
|
|
[src/scene/board/iplFocusObject] section .data size 2168 match 100.0
|
|
|
|
|
[src/scene/board/iplFocusObject] section .rodata size 176 match 100.0
|
|
|
|
|
[src/scene/board/iplFocusObject] section .sdata size 176 match 100.0
|
|
|
|
|
[src/scene/board/iplFocusObject] section .sdata2 size 56 match 100.0
|
|
|
|
|
[src/scene/board/iplFocusObject] section .text size 20332 match 99.64706
|
|
|
|
|
[src/scene/board/iplFocusObject] below 100: cmn_calc__Q33ipl5scene12focus_objectFv 87.185715
|
|
|
|
|
[src/scene/board/iplFocusObject] baseline: code 19772/20332 data 408 functions 88 fuzzy 99.6471
|
|
|
|
|
regressions vs baseline: 0
|
|
|
|
|
global matched_code_percent: 88.55724 -> 88.55724
|
|
|
|
|
global fuzzy_match_percent: 99.45033 -> 99.45033
|
|
|
|
|
global complete_code_percent: 62.78043 -> 62.78043
|
|
|
|
|
global matched_data_percent: 96.55914 -> 97.00876
|
|
|
|
|
forbidden patterns added (net, per file): 0
|
|
|
|
|
readability warnings (net, per file; must be 0 in the final result): 0
|
|
|
|
|
note: config touched: config/43U/symbols.txt (orchestrator reviews every config/symbols change)
|
|
|
|
|
GATE PASS
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
Data before -> after: CDBRecord 144 -> 144/2640, sd_drv 1016 -> 3592/3592, www_surface 40 -> 3536/3536, Focus 408 -> 2576/2576. Instruction-exact counts 27 -> 27, 21 -> 21, 21 -> 21, 88 -> 88. Matched code bytes unchanged 5464, 8940, 1916, 19772. Full clean gate passes, DOL SHA1 correct, regression/forbidden/style counts zero. No splits or total_data were changed. All three completed data units have every non-text section at 100%; CDB remains incomplete. Every original open code function has three logged attempts; Encrypt has six.
|
|
|
|
|
|