* Placed C++ extern "C" blocks inside include guards.
* Added a "clean_all" rule to the Makefile to avoid recompiling libusbhsfs after each time "clean" is used.
* Added static asserts for all structs that may need it throughout the code.
* Preprocessor macros are now used to generate certificate and ticket structs.
* Added functions to deal with title info and application metadata reallocations, greatly reducing the number of references to realloc() throughout the code.
* Tweaked gamecardGetHashFileSystemContext() to not return a pointer to a dynamic context.
* Added a type field to Hash FS contexts.
* Ported the logfile handler from libusbhsfs, with some slight modifications.
* Rewrote hash file system handling from scratch. I had been wanting to do this for some months now, it's a lot better now.
* Code cleanup.
This extension was already being used - however, smRegisterService() was also being used as a fallback method.
More than a year and a half has already passed since this feature was introduced, and both SX OS and ReiNX apparently support it now, so it's a justified change.
* Renamed titleRetrieveContentMetaKeysFromDatabase() to titleGenerateTitleInfoFromStorage().
* Moved linked lists generation from titleRetrieveContentMetaKeysFromDatabase() into its own function: titleUpdateTitleInfoLinkedLists().
* Adjusted orphan title checks. It is now explicitly verified if application metadata is available before treating a title as orphan, instead of checking if the parent user application is available.
* Code cleanup in titleRefreshGameCardTitleInfo().
* titleRefreshGameCardTitleInfo() now attempts to update the application metadata pointer in orphan title entries if new application metadata was retrieved after a gamecard was inserted *and* if the orphan title count is currently non-zero.
* Updated titleRemoveGameCardTitleInfoEntries() to make it use titleUpdateTitleInfoLinkedLists().
Serves the same purpose as the modded ns-usbloader builds. Supports the full USB ABI command set.
Also fixed a small issue related to ES IPC calls not writing the total number of available rights IDs.
* usb: copy command ID and block size before moving command data within the USB transfer buffer.
* nsp_dumper_usb: now spans a background thread for the dump process, progress is now displayed, process can now be cancelled by holding B, updated to properly make use of the new usbCancelFileTransfer() behavior.
* usb_gc_dumper: updated to properly make use of the new usbCancelFileTransfer() behavior.
* usb_romfs_dumper: updated to properly make use of the new usbCancelFileTransfer() behavior.
* Updated ns-usbloader patch. Must be used on commit `8771d551a4e6fa2d645e519d504a377e34cbd730`.
Keeps alive the current USB session by informing the host device that the user cancelled an ongoing USB transfer, instead of stalling the endpoints.
Also updated the USB ABI specs once again.
* Updated usbSendCommand() to make it send the command header first, and then the command block. Makes it easier for host applications to read and parse command data.
* Let usbSendCommand() take care of handling ZLT packets if required by any command block (only SendNspHeader at this moment), as well as logging status data via usbLogStatusDetail().
* Updated USB ABI specs doc.
* nacp: updated u32 and u64 field functions to let the user choose if the provided value should be printed using decimal or hexadecimal base, fixed codestyle.
* utils: properly implement ALIGN_DOWN() macro.
* gamecard: added more comments, modified gamecardGetBundledFirmwareUpdateVersion() to fill a VersionType1 pointer instead of a u32, fallback to manually reading the gamecard certificate if fsDeviceOperatorGetGameCardDeviceCertificate() fails, updated GameCardFwVersion enum.
* title: tweaked conditions for some functions.
* usb: explicitly defined values for previously omitted struct members while initializing USB comms, imported libusb structs to make it easier to work with the USB BOS, added a few more comments and changed around some code.
* utils: tweaked the illegal filesystem character array to remove some chars that are actually allowed.
* Temporarily changed tracked libusbhsfs branch from main to dev.
* Minor improvements in ums.c/h.
* Refactored nsp_dumper_sd into nsp_dumper_stor - now capable of dumping NSPs to a mounted partition from a UMS device.
* Fixed utilsWaitForButtonPress() when no button mask is provided.
* Refactored utilsGetFreeSpaceFromFileSystemByPath() into utilsGetFileSystemStatsByPath().
* Updated hardcoded system title list to match HOS 11.0.0.
* Updated NcaKeyGeneration_Current.
* Updated button input handling to match the new pad API from libnx.
* Free space is now retrieved via statvfs() for all filesystems, including the SD card.
* Updated Makefile to reflect latest libusbhsfs changes (dev branch).
* Cancel USB file transfer if something goes wrong during a NSP transfer via USB.
* Added SD card RomFS dumper.
* Further optimizations to the *WriteNcaPatch() functions.
* Change struct naming in nso.c/h (thanks @0Liam !).
* Replace fsp-usb with libusbhsfs. ( ͡° ͜ʖ ͡°)
* Fixed possible invalid memory accesses using TitleInfo pointers from gamecard titles if the gamecard they belong to is taken out.
* Fixed invalid for loop condition in _titleGetInfoFromStorageByTitleId() if no gamecard titles have been retrieved.
* Added NSP dumper PoC (SD card only atm, single-threaded).
* Cert: replaced a wrong strcmp() with a proper strncmp().
* CNMT: added functions to update content info entries and generate/write Partition FS patches.
* NCA: encrypt key area right after removing titlekey crypto.
* NPDM/ProgramInfo: changed function names.
* NPDM: check if the NCA has been modified before attempting to patch ACID data + calculate RSA-PSS signature *after* generating the PFS patch, not before. lol
* PFS: restore name table size value before writing the header padding.
* Tik: reworked the ticket lookup algorithm. Now uses information from ticket_list.bin to properly calculate the offset to the requested ticket in ticket.bin.
* Title: changed title type strings used for filename generation.
* Updated to-do list.
* AES: moved CTR initializing/updating functions here from nca.c.
* BKTR/RomFS/PFS: check if we're dealing with a NCA with titlekey crypto and the titlekey hasn't been retrieved.
* BFTTF: use void pointers for output font data.
* Mem: Only exclude Unmapped/Uo/ThreadLocal/Reserved memory pages if dealing with FS.
* NCA: use content type context pointers inside NCA contexts to manage ContentMeta, ProgramInfo, Nacp and LegalInfo contexts.
* NCA: added 'written' bool elements to patch structs to indicate patch write completion.
* NPDM: remove unnecessary inline functions, generate PFS patch right after changing ACID data, add a pfsWriteEntryPatchToMemoryBuffer wrapper.
* PFS: added PartitionFileSystemFileContext and related functions to deal with NSP headers.
* ProgramInfo: removed unnecessary inline functions.
* Save: added commented code to dump a full system savefile - will probably use it down the road.
* Tik: added support for volatile tickets (thanks to @shchmue and @Whovian9369!), added a rights ID string representation to the Ticket struct, clear Volatile and ELicenseRequired flags in conversions to common tickets.
* Title: added a function to calculate the number of titles (current + siblings) from a TItleInfo block.
* Utils: added a function to generate a dynamically allocated path string using a prefix, a filename and a extension.
* Removed explicit offset checks throughout all the code.
* Codestyle fixes.
* Updated to-do.
* Added a function to write re-encrypted NCA and/or FS section headers. Getting ready to re-implement NSP dumping.
* Added more CNMT XML fields (thanks 0Liam).
* Updated my to-do list because I'm dummy.
* Added 'encrypted_header' members to both NcaContext and NcaFsSectionContext. In-place re-encryption isn't desirable in this case.
* Fixed FsAccessControl-related type naming for ACI0 blocks.
Also:
* Fixed support for NSOs with invalid module_name, api_info, dynstr and/or dynsym offsets/sizes.
* Updated the XML generator example to generate XML data for all available NCAs, regardless of their ID offset values.
XML generation confirmed to be working. The new algorithm faithfully reproduces the same output from legacy nxdumptool with much less overhead and memory usage.
From now on, I'll delete code from the legacy codebase directory as soon as specific features from it are rewritten... Because it's a mess to navigate.
