[Top] [All Lists]

Interrupt handlers

Subject: Interrupt handlers
From: Ralf Baechle <>
Date: Sun, 21 Sep 1997 09:48:57 +0200 (MET DST)
Hi all,

some notes about the interrupt handlers.  The interrupt handling code we're
currently using is still mostly the first generation code.  Weaknesses
of the old code include:

 - Examining the bits in a loop this hurts especially on the R4000/R4200/R4400
   where each taken branch involves a two cycle penalty.
 - As a side effect of scanning in a loop we had a fixed priority.  That
   was ok for what the code was originally written for, but everybody
   (including me ...) shamelessly copied the code without fixing that ever.
   fixing that ever.
 - As side effect of the way normal and "SA_INTERRUPT" interrupts were
   being handled we had a extra indirection over a jump table.

Now after a little overhaul the interrupt handlers for the eval board I'm
just hacking is a lot faster and smaller; as a nice side effect they have
become _very_ trivial, too.  The cleanup of the "fast" interrupts also
affected the generic part of the interrupt handling.  I tried to do the
required modifications for the other systems code but without the hardware
at hand I of course could not test it.  So please check the code for your


<Prev in Thread] Current Thread [Next in Thread>
  • Interrupt handlers, Ralf Baechle <=