ARM_SP == 13, ARM_LR == 14, and ARM_PC == 15, which is not the case
* updated CMakeLists to include building arm regression test
* added explicit casts for 64 bit visual studio 2012 build to get around
truncation warnings from size_t conversion
- This mode is for the so-called MIPS "N64" ABI; it has nothing to do with the Nintendo 64 game platform.
- N64, O64, et al. are just different ABIs for the 64-bit MIPS architecture, so we replace CS_MODE_N64 with the existing CS_MODE_64
* fixed and added a test for a thumb-2 invalid sequence that was incorrectly allowed before these changes (pop.w with sp argument included)
* fixed and added a test for a blx from thumb to ARM that had its immediate argument incorrect (misaligned)
* eliminated some warnings by explicitly casting so I could turn on
treat warnings as errors locally
General notes:
* probably worth turning on treat all warnings as errors in the msvc project files, had a subtle bug that resulted from a missing declaration causing differences in dll and static compilation modes
( code was working incorrectly in dll form because of missing declaration in arch/ARM/ARMMapping.h for new function ARM_blx_to_arm_mode. Something about the linking was confusing ld when making the dll, and the resulting offsets were wonky (e.g. the added ble test would show up as #0x1fc instead of #0x1fe like it should have )
* the invalid pop was being treated as a soft fail which then gets coerced
to a success because it is != MCDisassembler_Fail in Thumb_getInstruction
what are the semantics of a soft fail? Maybe we should be able to set up
whether or not we want a soft fail to be a real fail in the csh struct?
Nguyen Anh Quynh
flyingsymbols
Jay Oster
kratolp
Yegor Derevenets