What's new in IDA 7.4.191112

Processor Modules:

M16C: improved jumptable recognition for M32C and R32C

Installer:

Installer: improved support for HiDPI displays in installers

Installer: ignore known unusable/buggy Python 3.x installs on Windows

Debugger:

DEBUGGER: android: introduced IDA_LIBC_PATH envvar to specify the path of libc.in a nonstandard location (e.g. in Android 10)

Plugins:

DSCU: allow __auth_stubs sections to be loaded individually from the dyldcache

Kernel/Misc:

goodname.cfg: improved std::basic_string patterns

BUGFIXES:

BUGFIX: 6502: cross-references to 16-bit addresses could be truncated to 8 bits

BUGFIX: debugger: IDA could appear to hang when deleting debugger segments after terminating a debugged process.

BUGFIX: debugger: mac, linux debug servers could crash when using --on-broken-connection-keep-session (-k) switch.

BUGFIX: DOS: IDA could crash after rebasing databases created for DOS .SYS files

BUGFIX: DSCU: "load libname.dylib" context menu item could be unavailable for some valid addresses

BUGFIX: hexrays: 32-bit magic divisions could be incorrectly recognized on arm64 and ppc64

BUGFIX: hexrays: fixed interr 50340

BUGFIX: hexrays: IDA could crash if no type was selected in "Convert to struct *" dialog, but OK was clicked

BUGFIX: hexrays: it was impossible to change the representation of some constants

BUGFIX: hexrays: op_uses_x() was returning false for 'cot_sizeof'

BUGFIX: hexrays: some switch cases would be wrongly printed as unsigned values

BUGFIX: hexrays: UDT return type could cause INTERR and __spoil<...> attribute was not supported for user-defined calls

BUGFIX: hexrays: wrong handling of big register arglocs on BE plaftorm: valid fastcall functions could be erroneously converted to usercall

BUGFIX: IDA could crash when trying to use license borrowing functionality without opening a file first

BUGFIX: IDA could fail with "out of memory" if something was wrong with the list of trusted idbs

BUGFIX: IDAPython could crash at startup on OSX when used with Python 3.6.

BUGFIX: IDAPython: ida_hexrays.cblock_t.insert() was unusable

BUGFIX: IDAPython: ida_kernwin.UI_Hooks.preprocess_action() wouldn't allow inhibiting the action

BUGFIX: IDAPython: ida_nalt.retrieve_input_file_md5() & ida_nalt.retrieve_input_file_sha256() were returning invalid data

BUGFIX: IDAPython: inspect.stack() could fail on recent versions of Python3 (e.g., Python 3.7.4)

BUGFIX: IDAPython: re-instantiating a chooser a second time could cause some context menu items to appear twice

BUGFIX: IDAPython: when returning an inappropriate type for ida_kernwin.Chooser2.OnGetLine() (e.g., returning a 'map()' in Python3, which produces a map iterable object and not a list), we would receive a cryptic message.

BUGFIX: installer: Python 2.7 installation could fail if path to temp directory contained non-latin characters.

BUGFIX: Lumina: IDA could submit the "File Name:" comment at the beginning of the IDB, as part of the first function's comments.

BUGFIX: M16C: ADD.L #imm, SP instructions could update SP with wrong value

BUGFIX: M16C: added missing DIV?.L and MUL?.L M32C instructions

BUGFIX: pc: sysret was not marked as an instruction that stops execution

BUGFIX: PPC: some VLE switches were not recognized if they used a register other than r0 for loading jump table elements

BUGFIX: Pressing F5 could cause a confusing "conflicting shortcuts" message to be printed in some situations

BUGFIX: ui/qt: it was impossible to change the font for the new type editor

BUGFIX: ui/qt: navigating some forms containing a text area using <Tab> and <Shift+Tab>, was broken

BUGFIX: ui/qt: the "Output window" was not understanding the 'r' (carriage return) character, making Python's 'tqdm' unusable

BUGFIX: UI: it was impossible to add the same file in "Recent scripts" in two IDA instances running simultaneously

BUGFIX: UI: IDA's UI could be sluggish on OSX when using an external display.

BUGFIX: UI: list of patched bytes would be empty when patching a rebased program

BUGFIX: ui: the script snippets could fail highlighting properly when Unicode was involved

BUGFIX: unnamed types could not be imported in structure/enum views