This is the mail archive of the gcc-help@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Scheduling and inserting NOPs


Boris Boesler <baembel@gmx.de> writes:

>   Now I need a pass to insert the missing NOPs. The approach Ian
> mentioned with the DFA-simulator does not work, because the simulator
> detects structural hardware hazards only (see http://www.nabble.com/Re
> %3A-Question-of-the-DFA-scheduler-p603486.html), which can not occur
> in my processor. Or did the DFA change?

I didn't suggest using the DFA simulator.  I agree that that won't
work.  What I suggested was to insert the NOPs in the
TARGET_ASM_FUNCTION_PROLOGUE hook.  This is not a scheduling hook.  It
is target specific code that is run immediately before the assembly
instructions are generated.

It seems to me that I keep saying the same thing, and you keep saying
that it won't work without explaining why.  I won't reply again.

Andrew's suggestion of using TARGET_MACHINE_DEPENDENT_REORG can also
work.  I don't like that approach as much in practice because it
doesn't work in conjunction with delay slots.  On some machines delay
slots are a good way to model VLIW semantics, in which instructions
can execute in parallel.

Ian


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]