added regression test for issue #702
Nasty workaround for aquynh/capstone#702
DBG+GUI: nasty workaround for aquynh/capstone#702
If you want a cleaner fix, you can proceed to these modifications in the file X86GenDisassemblerTables.inc:
It will change the operand specifier to the proper one, where the operands are inverted.
Sorry for not providing the DIFF, it is not possible for me right now 😊
@aquynh what are those files generated from? Perhaps it's better to change the source since the _reduce.inc has different constants for those instructions...
It seems those files are generated by llvm-tblgen, and the bug was fixed in llvm r233686
x86: Fix the operand encoding in the test instruction, issue #702
fixed now, please confirm.
x86: Fix the operand encoding in the test instruction for reduce set,…
… issue #702
Works correctly now, thanks a lot!
in r2 we had this issue opened for more than a year. now we can close it too. thanks, any will in releasing capstone during 2017 or should distros and users be forced to use the code from git one year more?
yes, the plan is to merge https://github.com/aquynh/capstone/tree/encoding into "next", then we can go for v4.0. however, this would break all the bindings, that is my concern now.