Comment 5 for bug 1421389

Hi Terry,

Thanks for confirming. In my particular case, I am running on a thumb only target (Cortex-M3), so my case would certainly be improved by adding logic to set the thumb bit to function addresses.

Like I said earlier, I don't have a testcase I can provide on this issue, but I can say that there are no linker symbols involved here that could understandably cause an issue like this. In my case, the object files are providing the symbols in such a way that the linker *should* be able to determine how to handle this correctly.

I can provide readelf/objdump output if needed. Let me know what commands you'd like output from and I can provide it tomorrow. Also, If you'd like me to test out a patch of the proposed logic change, I'd be happy to do so.