Benjamin Davis wrote:
> Errata 9 of
> http://e-www.motorola.com/files/32bit/doc/errata/MPC7455CE.pdf
> talks about some issues with QACK,
Okay.
> I further noticed how in arch/ppc/platform/pmac_feature.c, in
> probe_uninorth:
> if (uninorth_rev < 0x11) {
> actrl = UN_IN(UNI_N_ARB_CTRL) & ~UNI_N_ARB_CTRL_QACK_DELAY_MASK;
> actrl |= ((uninorth_rev < 3) ? UNI_N_ARB_CTRL_QACK_DELAY105 :
> UNI_N_ARB_CTRL_QACK_DELAY) <<
> UNI_N_ARB_CTRL_QACK_DELAY_SHIFT;
> UN_OUT(UNI_N_ARB_CTRL, actrl);
>
> It would appear to me that a QACK delay is being set by communication
> with the uninorth. I will say now though this is my first experiance
> with the kernel.
> I was kinda thinking they might have gotten around the errata by
> extending the delay of the QACK. This QACK is the system saying it is
> time to get out of nap, and snoop.
I don't see any reference in the 64260 manual that lets you control
this. Are you sure that you need to nap/snooze? IOW, are you sure
you're not solving a problem you don't have?
Mark
** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
