On 09/22/2014 07:26 PM, Steven Rostedt wrote: > On Mon, 22 Sep 2014 14:32:59 +0100 > Markos Chandras <markos.chandras@xxxxxxxxxx> wrote: > >> According to Documentation/trace/ftrace-design.txt, the selfpc >> should be the return address minus the mcount overhead (8 bytes). >> This brings static trace in line with the dynamic trace regarding >> the selfpc argument to the tracing function. >> >> This also removes the magic number '8' with the proper >> MCOUNT_INSN_SIZE. > > I could also update the generic code to handle delay slots. > > -- Steve As I said to the other patch, if you want to fix the delay slots in the generic code that may be preferred indeed. On the other hand, the static tracer still needs fixing so the correct selfpc is used. I will update this patch based on the way you choose to handle delay slots in the generic code. Thanks -- markos
