SELFish documentation

Start here

• README - SELFish documentation

Guide

• BUILDING - Building SELFish
• features/README - SELFish Features & Packaging Recipes
• features/eboot - Signed Executables (eboot)
• features/library - Taking it as a library
• features/pkg - Retail Packages (PKG)
• features/reading - Reading a file
• features/title - Title Directories
• features/user-guide - SELFish User Guide
• features/writing - Producing something the hardware will load

Reference

• GLOSSARY - Glossary: the formats
• USER_GUIDE - SELFish User Guide & Packaging Cookbook
• backlog/01-the-vendor-dynamic-table-done-both - 1. The vendor dynamic table - done, both directions
• backlog/02-section-headers-and-symbol-tables-done - 2. Section headers and symbol tables - done
• backlog/03-param-sfo-and-param-json-done - 3. param.sfo and param.json - done
• backlog/04-writing-the-filesystem-done-all-three - 4. Writing the filesystem - done, all three layers
• backlog/05-writing-a-package-done-image-included - 5. Writing a package - done, image included
• backlog/06-the-linker-script-done - 6. The linker script - done
• backlog/07-platform-abi-declarations-deferred-with - 7. Platform ABI declarations - deferred with a condition
• backlog/08-not-planned - Not planned
• backlog/_preamble - Backlog
• decisions/D001-a-fourth-repository-holding-formats-and - D001 - A fourth repository, holding formats and nothing that knows what a consumer is for
• decisions/D002-the-generation-is-a-type-with-no-default - D002 - The generation is a type with no Default
• decisions/D003-not-an-sdk-was-the-wrong-line-the-line - D003 - "Not an SDK" was the wrong line. The line is knowledge against runtime
• decisions/D004-one-import-hash-pinned-by-389-pairs - D004 - One import hash, pinned by 389 pairs somebody else produced
• decisions/D005-oops-is-additive-never-required - D005 - OOPS is additive, never required
• decisions/D006-entry-answers-for-itself-because-the - D006 - Entry answers for itself, because the first version of it could not be called
• decisions/D007-the-container-is-byte-identical-to-the - D007 - The container is byte-identical to the implementation it replaces
• decisions/D008-the-package-s-outer-layer-first-because - D008 - The package's outer layer first, because it needs no cryptography
• decisions/D009-the-crypto-chain-is-unchanged-between - D009 - The crypto chain is unchanged between generations, and that is a finding
• decisions/D010-every-container-in-three-current - D010 - Every container in three current-generation packages uses the previous generation's magic
• decisions/D011-the-image-offset-is-a-header-field-and - D011 - The image offset is a header field, and the evidence that it was not could not have failed
• decisions/D012-what-a-package-writer-still-needs-and - D012 - What a package writer still needs, and what it does not
• decisions/D013-two-vendor-tag-ranges-and-each-side - D013 - Two vendor tag ranges, and each side documented only the one it uses
• decisions/D014-reassembly-is-the-inverse-of-building - D014 - Reassembly is the inverse of building, and both live together
• decisions/D015-selfish-elf-depends-on-selfish-nid - D015 - selfish-elf depends on selfish-nid, which reorders the spine
• decisions/D016-an-import-s-library-is-looked-up-by-its - D016 - An import's library is looked up by its id, not by its position in the table
• decisions/D017-relocations-are-read-and-censused-here - D017 - Relocations are read and censused here; applying them stays with the consumer
• decisions/D018-an-unrecognised-relocation-type-gets-no - D018 - An unrecognised relocation type gets no name
• decisions/D019-the-param-sfo-alignment-rule-came-from - D019 - The PARAM.SFO alignment rule came from a source and was refuted by eleven files
• decisions/D020-unterminated-text-is-a-separate-value - D020 - Unterminated text is a separate value variant, not a flag
• decisions/D021-there-is-a-second-package-magic-and-it - D021 - There is a second package magic, and it is named and refused rather than guessed at
• decisions/D022-package-writing-stays-blocked-and-the - D022 - Package writing stays blocked, and the source that looked like it might unblock it does not
• decisions/D023-sections-are-a-separate-module-from - D023 - Sections are a separate module from dynamic, because they answer a different question
• decisions/D024-the-linker-script-moves-here-and-its - D024 - The linker script moves here, and its constants are tested against the crate's
• decisions/D025-the-dynamic-table-writer-moves-here-and - D025 - The dynamic-table writer moves here, and the manifest stays with the caller
• decisions/D026-the-end-to-end-test-is-the-point-of - D026 - The end-to-end test is the point of putting both halves in one crate
• decisions/D027-filesystem-writing-is-blocked-and-now - D027 - Filesystem writing is blocked, and now measured rather than asserted
• decisions/D028-the-header-identity-is-stamped-here-too - D028 - The header identity is stamped here too, and e_type is a parameter rather than a constant
• decisions/D029-the-first-consumer-migrated-and-the - D029 - The first consumer migrated, and the migration was the best review this repository has had
• decisions/D030-the-two-halves-check-each-other-where - D030 - The two halves check each other where material is unavailable
• decisions/D031-the-command-line-tool-does-not-panic - D031 - The command-line tool does not panic when a pipe closes
• decisions/D032-this-repository-gets-a-ci-workflow-and - D032 - This repository gets a CI workflow, and obSCEne's is marked as unable to pass
• decisions/D033-the-six-unknown-package-entries-all - D033 - The six unknown package entries all vary between titles, so there is no constant to fall back on
• decisions/D034-two-package-entries-were-derived-rather - D034 - Two package entries were derived rather than cited, and there is a command that re-derives them
• decisions/D035-a-derivation-ships-with-the-command - D035 - A derivation ships with the command that re-runs it
• decisions/D036-entry-0x80-is-a-digest-manifest-over - D036 - Entry 0x80 is a digest manifest over named things, and two of its slots are established
• decisions/D037-a-package-writer-that-refuses-to-invent - D037 - A package writer that refuses to invent the entries nothing established explains
• decisions/D038-what-the-cryptographic-entries-are-not - D038 - What the cryptographic entries are not, recorded so nobody re-runs the searches
• decisions/D039-liborbispkg-named-all-six-and-the - D039 - LibOrbisPkg named all six, and the derivations were right
• decisions/D040-playgo-chunk-sha-is-solved-and-material - D040 - PLAYGO_CHUNK_SHA is solved, and material settled what the source left open
• decisions/D041-the-licence-entries-are-stored - D041 - The licence entries are stored encrypted, so naming them was not enough
• decisions/D042-the-filesystem-superblock-was-never-a - D042 - The filesystem superblock was never a wall - it was 95 unnamed bytes
• decisions/D043-naming-a-field-is-not-the-same-as-being - D043 - Naming a field is not the same as being able to write one
• decisions/D044-the-licence-entries-were-never-unknown - D044 - The licence entries were never unknown - a thirty-two-byte record, not three fields
• decisions/D045-understood-is-not-written - D045 - Understood is not written
• decisions/D046-the-licence-structure-is-measured - D046 - The licence structure is measured; producing one needs two keys this repository does not hold
• decisions/D047-a-licence-is-built-from-scratch-and-a - D047 - A licence is built from scratch, and a real one is reproduced byte for byte
• decisions/D048-pack-now-demands-only-what-a-title-is - D048 - pack now demands only what a title is, not what a package is
• decisions/D049-the-filesystem-is-built-in-two-halves - D049 - The filesystem is built in two halves, because a package has two filesystems built to different rules
• decisions/D050-pfsc-does-not-compress-and-that-is-the - D050 - PFSC does not compress, and that is the format rather than a shortcut
• decisions/D051-the-two-ways-of-writing-a-1-near-the - D051 - The two ways of writing a 1 near the end of the superblock are two fields, not one
• decisions/D052-a-block-signature-is-an-hmac-under-a - D052 - A block signature is an HMAC under a key the builder computes
• decisions/D053-signature-ordering-is-the-correctness - D053 - Signature ordering is the correctness argument, so it is written as one sequence
• decisions/D054-the-key-blobs-are-computed-and-the - D054 - The key blobs are computed, and the proof is that they come out byte-identical to real packages
• decisions/D055-the-passcode-reaches-further-than-the - D055 - The passcode reaches further than the key blobs, and a test keyed a package differently to find out
• decisions/D056-the-header-past-0x410-was-entirely-zero - D056 - The header past 0x410 was entirely zero, so nothing could be mounted
• decisions/D057-the-flat-path-table-is-real-now-and-it - D057 - The flat path table is real now, and it was built precisely because nothing here reads it
• decisions/D058-selfish-image-exists-because-another - D058 - selfish image exists because another session asked for it in a script comment
• decisions/D059-a-param-sfo-is-a-format-and-belongs - D059 - A param.sfo is a format and belongs here. An icon is a picture and does not
• decisions/D060-the-default-icon-is-selfish-s-own-mark - D060 - The default icon is selfish's own mark, reversing D059
• decisions/D061-the-param-sfo-field-set-was-the - D061 - The param.sfo field set was the previous generation's, and two homebrew packages said so
• decisions/D062-a-second-param-sfo-implementation-and - D062 - A second PARAM.SFO implementation, and why it happened
• decisions/D063-one-directory-entry-writer-because-the - D063 - One directory-entry writer, because the copy had the rule without the reasoning
• decisions/D064-a-literal-nul-byte-in-a-source-file - D064 - A literal NUL byte in a source file made it invisible to grep, which hid a duplicate constant
• decisions/D065-cargo-doc-was-failing-and-the-gate-did - D065 - cargo doc was failing, and the gate did not build docs
• decisions/D066-documentation-drifts-in-the-direction - D066 - Documentation drifts in the direction of claiming less than the code does
• decisions/D067-supporting-the-current-generation-is - D067 - Supporting the current generation is three-quarters done and blocked on one oracle
• decisions/D068-the-logo-is-one-file-committed-and-the - D068 - The logo is one file, committed, and the drawn mark is gone
• decisions/D070-the-cache-size-is-a-ceiling-measured - D070 - The cache size is a ceiling measured from the image, not a constant
• decisions/D071-d070-s-rule-was-wrong-and-the-corrected - D071 - D070's rule was wrong, and the corrected one cannot be tested by avoiding it
• decisions/D072-the-inner-filesystem-s-root-pointed-its - D072 - The inner filesystem's root pointed its parent at the super root
• decisions/D073-the-tool-converts-a-supplied-icon - D073 - The tool converts a supplied icon, rather than asking four projects to export one
• decisions/D074-an-executable-declares-no-export - D074 - An executable declares no export library, which frees library id zero
• decisions/D075-a-container-entry-s-memsz-is-its-data-s - D075 - A container entry's memsz is its data's size, not the segment's memory size
• decisions/D076-the-dynamic-table-lives-at-the-tail-of - D076 - The dynamic table lives at the tail of the vendor segment, not in the image
• decisions/D077-the-vendor-segment-begins-with-a - D077 - The vendor segment begins with a fingerprint region, and leaving it out moves everything
• decisions/D078-a-bundled-library-is-a-third-layout-and - D078 - A bundled library is a third layout, and it needs a third linker script
• decisions/D079-dt-sce-original-filename-is-required - D079 - DT_SCE_ORIGINAL_FILENAME is required, and holds the module's own name
• decisions/D080-the-keystone-is-derived-not-supplied - D080 - The keystone is derived, not supplied, and every package this crate built was missing one
• decisions/D081-param-json-does-not-belong-in-a-package - D081 - param.json does not belong in a package, and its presence there was mixing two routes
• decisions/D082-a-package-cannot-produce-a-current - D082 - A package cannot produce a current-generation title, so native is a second delivery route rather than a second package format
• decisions/D083-a-comment-in-the-keyset-file-broke-the - D083 - A comment in the keyset file broke the licence, because the reader matched prose
• decisions/D084-a-real-container-can-be-audited-against - D084 - A real container can be audited against the format table - the oracle step, as a command
• decisions/D085-getting-a-real-current-generation-self - D085 - Getting a real current-generation SELF is a measurement obSCEne runs, not a file this repo holds - and the sandbox makes it conditional
• decisions/D086-confirm-the-format-on-the-console-and - D086 - Confirm the format on the console and report the verdict - do not carry the bytes off
• decisions/D087-native-ps5-title-manifests-param-json - D087 - Native PS5 title manifests (param.json) and differential measurement over fake-signed packaging
• decisions/D088-libkernel-vaddrs-example-exports-as - D088 - libkernel_vaddrs example: exports as name vaddr, and why the names are not ours
• decisions/D089-naming-a-hash-backwards-is-a-search - D089 - Naming a hash backwards is a search over somebody else's vocabulary, so the vocabulary is an argument
• decisions/D090-the-unterminated-sfo-format-is-a - D090 - The unterminated SFO format is a length, not a promise of text, and reading it as text failed whole files
• decisions/D091-a-writer-must-refuse-where-a-reader - D091 - A writer must refuse where a reader may guess, and this crate had them the wrong way round
• decisions/D092-the-container-that-matched-this - D092 - The container that matched this table was one this table could have written
• decisions/D093-the-audit-reports-what-kind-of - D093 - The audit reports what kind of container it just agreed with, before it reports how many rows agreed
• decisions/D094-the-audit-checks-the-tail-it-pins - D094 - The audit checks the tail it pins, and reports it apart from the header
• decisions/D095-a-second-reader-found-three-and-one - D095 - A second reader found three disagreements, and only one of them was about the format
• decisions/D096-two-answers-to-the-symbol-count-and - D096 - Two answers to the symbol count, and the divergence is kept rather than reconciled
• decisions/D097-the-wrap-default-was-justified-by-a - D097 - The wrap default was justified by a population counted in the wrong place, and the default is right anyway
• decisions/D098-measured-vendor-values-are-recorded - D098 - Measured vendor values are recorded against the rows they refute, never as a profile a writer could reach for
• decisions/D099-the-inner-size-was-never-unknown - D099 - The inner size was never unknown, and entry 0x1001 is computed like every other derivable entry
• decisions/D100-stamp-and-wrap-retire-into-the-pipeline - D100 - stamp and wrap retire into the pipeline; --format prx, and --privilege/--sdk as pipeline options
• decisions/D101-a-module-built-here-cannot-carry-a-weak-import - D101 - A module built here cannot carry a weak undefined import, by construction; the forcing to GLOBAL stays
• decisions/D102-title-metadata-options-on-the-pipeline - D102 - The title metadata a title carries gets pipeline spellings; --root does not
• decisions/D103-the-agc-shader-container-and-where-it-may-come-from - D103 - The AGC shader container is admissible as a format, but only the part a citable source derives - not the vendor-binary part, and not because hardware accepts it
• decisions/D104-selfish-cli-depends-on-nothing-outside-this-repository-again - D104 - selfish-cli depends on nothing outside this repository again: the commit stamp is local git, the logging was dead weight
• decisions/_preamble - Decisions
• worklog/001-the-repository-exists-and-the-spine-is - The repository exists, and the spine is laid
• worklog/002-the-generation-split-made - The generation split, made unrepresentable rather than tested for
• worklog/003-the-hash-and-an-objection-that-turned - The hash, and an objection that turned out to argue the other way
• worklog/004-the-executable-format-and-the-container - The executable format, and the container in both directions
• worklog/005-the-package-s-outer-container - The package's outer container
• worklog/006-the-key-derivation - The key derivation
• worklog/007-the-filesystem-and-the-whole-chain-end - The filesystem, and the whole chain end to end
• worklog/008-a-command-line-tool-and-an-audit-that - A command-line tool, and an audit that found the real gap
• worklog/009-the-dynamic-table-and-the-whole-stack - The dynamic table, and the whole stack on genuine material
• worklog/010-the-symbol-table-and-imports-resolved - The symbol table, and imports resolved to names
• worklog/011-relocations-and-the-check-that-closed - Relocations, and the check that closed the loop
• worklog/012-param-sfo-and-param-json-and-eleven - PARAM.SFO and param.json, and eleven files that disagreed with the source
• worklog/013-two-package-magics-and-a-source-that - Two package magics, and a source that was not the one we needed
• worklog/014-section-headers-and-the-link-time - Section headers and the link-time symbol table
• worklog/015-the-linker-script-and-a-test-that - The linker script, and a test that actually links
• worklog/016-the-dynamic-table-writer-and-the-loop - The dynamic-table writer, and the loop closing
• worklog/017-sizing-the-filesystem-writing-gap - Sizing the filesystem-writing gap instead of starting it
• worklog/018-running-the-finished-thing-and-finding - Running the finished thing, and finding the last missing field
• worklog/019-the-first-migration - The first migration
• worklog/020-closing-three-gaps-the-migration-opened - Closing three gaps the migration opened
• worklog/021-crunching-the-unknown-package-entries - Crunching the unknown package entries
• worklog/022-package-writing-built-up-to-the-wall - Package writing, built up to the wall
• worklog/023-the-source-and-what-it-did-and-did-not - The source, and what it did and did not settle
• worklog/024-the-superblock-was-never-a-wall - The superblock was never a wall
• worklog/025-licences-built-and-proven - Licences, built and proven
• worklog/026-sections-dynamic-symbols-reading-dynsym - Sections::dynamic_symbols - reading .dynsym
• worklog/027-the-filesystem-written - The filesystem, written
• worklog/028-a-package-obscene-can-build-and-what-is - A package obSCEne can build, and what is still placeholder
• worklog/029-the-reader-s-refusals-tested - The reader's refusals, tested
• worklog/030-package-header-sensible-defaults - Package header: sensible defaults instead of zeros (found on hardware)
• worklog/031-the-package-header-is-integrity - The package header is integrity-protected, and that is the wall (measured on hardware)
• worklog/032-a-fake-package-through-three-console - A fake package through three console rejection stages (measured on hardware)
• worklog/033-a-hardcoded-cache-size-makes-a-small - A hardcoded cache size makes a small package unmountable
• worklog/034-auditing-the-other-constants-found-a - Auditing the other constants found a second one that was never constant
• worklog/035-the-indirect-block-looked-guilty-and-is - The indirect block looked guilty and is not, which sets up the next test
• worklog/036-the-inner-filesystem-mounts-it-was-the - The inner filesystem mounts. It was the root's parent all along.
• worklog/037-an-eboot-is-not-the-module-this-project - An eboot is not the module this project builds, in three separate ways
• worklog/038-the-eboot-executes-it-is-a-process-now - The eboot executes. It is a process now, and it dies on a syscall.
• worklog/039-five-refusals-each-naming-the-next-an - Five refusals, each naming the next: an eboot walked into a console's loader
• worklog/040-a-package-built-here-installs-mounts - A package built here installs, mounts, loads and executes on hardware
• worklog/041-a-third-linker-script-and-the-tag-that - A third linker script, and the tag that was named but never written
• worklog/042-libkernel-vaddrs-example - 2026-08-31 - libkernel_vaddrs example
• worklog/043-native-ps5-title-generation-obscene - 2026-08-31 - Native PS5 Title Generation & obSCEne Self-Resolver Oracle
• worklog/044-three-unnameable-imports-and-one-was - 2026-09-09 - Three unnameable imports, and one of them was never a name
• worklog/045-account-id-was-not-text-and-the-reader - 2026-09-09 - ACCOUNT_ID was not text, and the reader refused the file rather than the key
• worklog/046-the-same-function-twice-to-opposite - 2026-09-09 - The same function twice, written to opposite standards
• worklog/047-the-table-was-right-and-the-limit-was - 2026-09-09 - The table was right, and the limit was the wrong shape
• worklog/048-a-label-in-a-candidate-list-is-not - 2026-09-09 - A label in a candidate list is not provenance
• worklog/049-the-confirmation-was-circular-and-the - 2026-09-09 - The confirmation was circular, and the entry that caught it had just written the rule
• worklog/050-the-lesson-put-in-the-tool-and-the - 2026-09-09 - The lesson put in the tool, and the reversal made airtight
• worklog/051-the-tail-held-and-the-instinct-was - 2026-09-09 - The tail held, and the instinct about it was wrong the other way
• worklog/052-the-second-reader-and-the-error-that - 2026-09-09 - The second reader, and the error that pointed at the wrong thing
• worklog/053-two-readers-agree-on-every-field-and - 2026-09-09 - Two readers agree on every field, and the third failure of the day was agreeing
• worklog/054-thirty-two-containers-and-a-confound - 2026-09-09 - Thirty-two containers, a clean split, and a confound I had already glossed
• worklog/055-the-symbol-level-agreed-and-the - 2026-09-09 - The symbol level agreed, and the question about it found something
• worklog/056-the-accusation-in-the-suffix-file-was - 2026-09-09 - The accusation in the suffix file was made from a label
• worklog/057-a-title-id-in-a-citation-read-as-a-fact - 2026-09-09 - A title id in a citation, read as a fact
• worklog/058-the-audit-closed-and-the-test-that - 2026-09-09 - The audit closed, and the test that makes it cheap next time
• worklog/059-the-magic-settled-by-the-row-nobody - 2026-09-09 - The magic, settled by the row nobody asked for
• worklog/060-both-magics-are-accepted-and-the - 2026-09-09 - Both magics are accepted, and the cross-census already contained our own file
• worklog/061-a-green-gate-that-was-never-checked - 2026-09-10 - A green gate that was never checked, and a commit that took somebody else's work
• worklog/062-one-vocabulary-and-the-orphaned - 2026-09-10 - One vocabulary, and an attribute that outlived its variant
• worklog/063-a-guard-that-covered-one-of-three - 2026-09-10 - A guard that covered one of three, and five copies of one table
• worklog/064-half-of-a-refusal-was-never-a-refusal - 2026-09-10 - Half of a refusal was never a refusal
• worklog/065-the-question-i-filed-was-answered-in - 2026-09-10 - The question I filed was answered in a file I had already cited
• worklog/066-the-last-two-verbs-and-a-paid-value-nothing-wrote - 2026-09-10 - The last two verbs, and a paid value nothing wrote
• worklog/067-the-pipeline-learns-what-a-title-carries - 2026-09-10 - The pipeline learns what a title carries, and an empty content id that keyed nothing
• worklog/068-a-review-caught-icon-copied-not-converted - 2026-09-10 - A review caught --icon copied, not converted - and my evidence that it was
• worklog/069-the-shader-container-was-citable-after-all - 2026-09-10 - The shader container was citable after all, once I looked past the first two emulators
• worklog/070-a-cli-verb-so-a-c-consumer-can-reach-the-shader-builder - 2026-09-11 - A CLI verb, so a C consumer can reach the shader builder
• worklog/071-pixel-and-vertex-stages-and-the-constants-i-would-not-ship - 2026-09-11 - Pixel and vertex stages, the hardware that confirmed the header, and the constants I would not ship
• worklog/072-selfish-cli-stands-alone-again - 2026-09-11 - selfish-cli stands alone again: the local commit stamp, and the logging that logged nothing
• worklog/_preamble - Worklog

Project memory

• BACKLOG - Backlog
• DECISIONS - Decisions
• WORKLOG - Worklog