PATCH/RFC [0/3]: Fix suspend and other acceleromter issues

Simon Kagstrom simon.kagstrom at
Sun Nov 16 14:25:15 CET 2008

On Sun, 16 Nov 2008 11:18:26 -0200
Werner Almesberger <werner at> wrote:

> Simon Kagstrom wrote:
> > Well, we don't set the LIR bit when configuring FF_WU_CFG. It's also
> > not used for "normal" data-ready interrupts, only with the threshold
> > stuff. So it's already cleared :-)
> Ah, that makes sense, yes. So no luck then and AOI doesn't help
> either, grr :-(

Not so fast! :-) Did a bit of testing, and setting the LIR bit when
enabling data collection with the threshold set to 1 (18mg) and at the
same time reading FF_WU_SRC in the interrupt handler makes the
interrupt load more reasonable again:

  Cpu(s): 14.2%us, 19.1%sy,  0.0%ni, 47.2%id,  0.0%wa,  0.0%hi, 19.4%si,  0.0%st
  Mem:    119968k total,   106472k used,    13496k free,     1032k buffers
  Swap:        0k total,        0k used,        0k free,    67304k cached

    PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND           
   1740 root      20   0  5176 2980 1768 S 18.1  2.5   0:36.22 python             
   1732 root      20   0  2264  844  684 S 17.1  0.7   0:27.63 dropbear

(same %si load but lower %us load than with the data ready interrupt).
So the LIR bit is probably a good idea and makes sense. Patch coming up

Thanks for the hint Werner!
// Simon

More information about the openmoko-kernel mailing list