IDAPro7.5 Bugfixes

Bugfixes:

• BUGFIX: "bad event during undo" could occur in some cases
• BUGFIX: "find next error" could crash IDA
• BUGFIX: "ida -I1" was modifying a wrong registry key when trying to set itself as the systemwide just-in-time debugger
• BUGFIX: ARM: A64 LDARP instruction was printed with an incorrectly duplicated operand
• BUGFIX: ARM: IDA could show wrong values if instruction simplification was enabled and instrucions with shifted immediate values were present
• BUGFIX: ARM: The A64 instruction CRC32W was printed with an unnecessary .W suffix
• BUGFIX: compile_idc_snippet() could fail if the snippet was ending with a comment and no newline
• BUGFIX: cursor position in the list of xrefs to stkvars was not preserved
• BUGFIX: debugger: a malicious client could invoke commands on a password-protected debug server without a password
• BUGFIX: debugger: IDA could crash with interr 40052 when exiting while process is suspended with tracing enabled
• BUGFIX: debugger: IDA could exit with internal error 40038 if erasing a breakpoint from the process failed unexpectedly
• BUGFIX: debugger: IDA could fail to attach through GDB to a running instance of QEMU
• BUGFIX: debugger: IDA could INTERR with 64-bit GDB flags register
• BUGFIX: debugger: in rare cases IDA could crash when using Appcall in win32 debugger
• BUGFIX: debugger: ios debugger could fail to handle read/write breakpoints in multithreaded situations.
• BUGFIX: debugger: linux: the base of segment registers was calculated incorrectly in x86_64
• BUGFIX: debugger: PPC: when debugging VLE code, IDA could put breakpoints at wrong locations
• BUGFIX: debugger: values of Dn registers on ARM32 platform would not be available
• BUGFIX: debugger: when attaching to some Windows 10 systems using Windbg backend, IDA would appear to hang
• BUGFIX: debugger: win32: On Windows 7, IDA could incorrectly rebase the database if the executable was mapped into the address space a second time (can happen e.g. when displaying the icon in a File Open dialog)
• BUGFIX: decompiler: assigning to a part of a variable could be erroneously translated as assigning to the whole variable
• BUGFIX: decompiler: changed the hotkey for "global xrefs" to Ctrl-X because Shift-X does not work well in all contexts (for example, in choosers)
• BUGFIX: decompiler: decompiler could lose instructions which modified its operands
• BUGFIX: decompiler: fixed a crash on decompilation failure when COLLAPSE_LVARS=YES in hexrays.cfg
• BUGFIX: decompiler: fixed interr 52329, which could occur if a enum type was renamed after its application in the decompiler
• BUGFIX: decompiler: fixed numerous internal errors
• BUGFIX: decompiler: IDA could crash with unhandled exception on opening a database which was saved after using the decompiler
• BUGFIX: decompiler: in some cases "Cancel" button did not stop the decompilation
• BUGFIX: decompiler: interr could occur if a parenthesis was used in a variable name
• BUGFIX: decompiler: it could be required to press 'Escape' twice in order to cancel a decompilation requested by jumping to an address
• BUGFIX: decompiler: it was impossible to input the negative number for the shifted value in the "convert to struct*" dialog
• BUGFIX: decompiler: ppc instruction mulhd was decompiled incorrectly
• BUGFIX: decompiler: pressing enter at the end of the very first line of the function body would not add an empty line as it should
• BUGFIX: decompiler: renaming the same variable twice from two different pseudocode windows could cause an erroneous warning
• BUGFIX: decompiler: some forced variables were not applied correctly
• BUGFIX: decompiler: some lvar mappings would be ignored by the decompiler
• BUGFIX: decompiler: some SSE2 instructions were decompiled to wrong intrinsics
• BUGFIX: decompiler: when canceling a jump from "Pseudocode-A" to a new function, canceling decompilation could cause IDA to switch to "IDA View-A"
• BUGFIX: demangler: for old borland mode (v < 5.5) some types in template arguments were demangled incorrectly
• BUGFIX: DWARF: The DWARF plugin could complain about invalid data for some Golang binaries
• BUGFIX: DWARF: The DWARF plugin could enter an inconsistent state and bail out upon certain constructs
• BUGFIX: DWARF: The DWARF plugin could fail to parse certain constructs involving similarly-named typedefs, to various templates instantiations
• BUGFIX: DWARF: The plugin could create the same parameter multiple times, if certain (GCC) constructs were used to specify their const value
• BUGFIX: ELF: MIPS: improve handling of the special symbol "_gp_disp"
• BUGFIX: ELF: PLT stubs could be truncated and marked as no-return in some MIPS files, resulting in bad analysis
• BUGFIX: ELF: some ARM shared objects could fail to resolve external symbols (imports)
• BUGFIX: enum radix was not immediately propagated from the enum view to the local types
• BUGFIX: fixed a random interr 30143 that was occurring when attaching to a WoW64 application that was generating lots of exceptions
• BUGFIX: fixed erroneous internal error 1544 that could occur after a debugger session
• BUGFIX: gdb debuggers could interr 30044 in multithreaded situations.
• BUGFIX: GDB would not mask exceptions even if configured to do so
• BUGFIX: GDB would not respect the user's request when manually resuming after exceptions
• BUGFIX: GDB: LR was incorrectly set as instruction pointer for PPC configurations (correct register is PC)
• BUGFIX: hexview: editing undefined byte and setting its value to 0xFF, could fail to show the value properly
• BUGFIX: IDA analysis could loop indefinitely when analyzing some switch patterns produced by clang (e.g. in chrome.dll)
• BUGFIX: IDA could crash in case of a network error or if a remote GDB target did not support/report threads
• BUGFIX: IDA could crash on exit when cleaning the leaked type objects (e.g. after a decompiler error)
• BUGFIX: IDA could crash when debugger flag names were used as variables in IDC scripts
• BUGFIX: IDA could crash when loading a new database with autoanalysis in progress
• BUGFIX: IDA could crash when using watches during debugging
• BUGFIX: IDA could fail to restore some segment register areas
• BUGFIX: IDA could INTERR(40662) with C++ plugins that provide a PCF_EA_CAPABLE place_t implementation
• BUGFIX: IDA could produce a fatal error when applying a function prototype with __spoils list which included ARM64 Xnn registers