tberghammer added a comment. In http://reviews.llvm.org/D16772#340950, @clayborg wrote:
> Looks good. > > A related comment about IT instructions. You need to be careful when software > single stepping through instructions that are in the middle of an ITSTATE > block. You can NOT use a 16 bit instruction that isn't a BKPT otherwise you > change your instructions. Are you guys using the BKPT trap for software > breakpoints? If not, you should not step through IT instructions that have 32 > bit thumb instructions or you will hose your program. Let me know if you need > to know more about this. We are currently using 16 bit UDF instruction and as far as I can tell from some manual testing it is working as expected both when stepping over 16 bit and when stepping over 32 bit instructions. I will look into it why we are using UDF instead of BKPT and will create a separate CL for replacing it. In http://reviews.llvm.org/D16772#341030, @omjavaid wrote: > Looks good. Was there a test failing in testsuite due to this? I detected the issue based on TestStandardUnwind but we have no test exercising this feature directly. P.S.: TestStandardUnwind still failing on some android arm devices because we don't handle the IT instruction for unwinding either. I am working on a fix for that as well. http://reviews.llvm.org/D16772 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits