Posted on July 06, 2015 at 04:18MCU: STM32F4 (ARM Cortex M4)
Build environment: arm-none-eabi-gcc 4.8.4 20140725
My goal is to build an image that can be run from any properly-aligned offset in internal flash.
My ideal scenario is for the code would be position-independent, but the data and bss sections to be located at fixed offsets in SRAM. Function pointers would need to be called via the GOT, of course. Unfortunately, I wasn't able to find a combination of gcc flags that would let me generate code in this way. The generated code always seemed to assume that data variables could be found at an offset relative to the program counter. This assumption fails because the code can run from a variety of locations, but the data is always at a fixed memory address.
The next best thing I found is the following combination of flags:
# Generate position independent code.
-fPIC
# Access bss via the GOT.
-mno-pic-data-is-text-relative
# GOT is not PC-relative; store GOT location in a register.
-msingle-pic-base
# Store GOT location in r9.
-mpic-register=r9
The problem I'm having with this set of build flags is that r9 seems to get clobbered when the processor transitions from thread mode to handler mode. This leads a hard fault when my exception handler attempts to read something from the GOT. I tried using r10 instead of r9, but I observed the same behavior.
So, my questions are as follows:
1. Does anyone know of a way to get gcc to generate position independent code that assumes fixed addresses for data and bss?
2. If not, any suggestions for using -msingle-pic-base safely in handler mode?
All input is greatly appreciated.
