On Wed, 2004-12-01 at 12:05 -0800, Michael wrote:
> > Hmmm,
> > I too intergrated Lothar's changes but still the same problem,
> > even with
> > using it directly as you said above.
> > When I look at my syslog, I see the correct value being written
> > to the
> > data port, but when I read the register I only get 0x00002edf :-(
> > On the other hand, HCLSTHRESH gets the right value...
> > 
> 
> Hrm..  It must be getting overwritten.
No it is not! The register gets written only once during hc_start() of
ohci-hcd.c when periodic_reinit() is called. And although the value of
0x27782edf is sent to the data port, the register shows 0x00002edf!

> The power-on-reset value of HCLSTHRESH is the one I'm using.  If
> you are as well, it may either never get set, or just re-set with
> the same value.

Hmm, let's see: The power-on-reset value of HcLSThreshold, bits
LST[10:0] is 0x628, so I might as well not be writing it at all. But
then the power-on-reset values of HcFmInterval are:
FSMPS[14:0] = 0x0
FI[13:0] = 0x2edf
So I think it is more likely that I never get to write the register at
all! But it seems unthinkable to my humble mind that every register (16
or 32 bit) I read or write behaves correctly except those two... How can
this happen? The command and data ports are being addressed correctly
and I can see the correct value being sent to the register, something
like:
A>[EMAIL PROTECTED]
D>[EMAIL PROTECTED]

-- 
Dimitris Lampridis <[EMAIL PROTECTED]>

Attachment: signature.asc
Description: This is a digitally signed message part

Reply via email to