Re: Serial init / BRG

The lockup is only the shell.  I echo from ttyCPM0 console and it
"locks" then I echo from an ssh session and the kernel messages still
output from the ttyCPM0 console.

NOTE: I found a severe issue that seems to have unlocked ttyCPM2.  BUT
ttyCPM1 still hangs.  (a while back, we commented out the out_be in
cpm_setbrg - oops (console debug left over))

with address / value in cpm_setbrg

echo test > /dev/ttyCPM1 # * note still hang *
CPM uart[1]:startup - port->irq 00000028
CPM uart[1]:startup - port->irq 40
cpm_setbrg:107 - brg:0 baud:9600
cpm_setbrg:108 - addr:e00119f0 value:00010140

echo test > /dev/ttyCPM2  * note doesn't hang *
CPM uart[2]:startup - port->irq 0000002b
CPM uart[2]:startup - port->irq 43
cpm_setbrg:107 - brg:3 baud:9600
cpm_setbrg:108 - addr:e00119fc value:00010140

On 10/16/07, Scott Wood [off-list ref] wrote:

Alan Bennett wrote:

quoted

I also verified that when I attempt to echo to the ports that I reach
the appropriate driver
i.e.
echo "hello" > /dev/ttyCPM1
   CPM uart[1]:startup - port->irq 40
echo "hello" > /dev/ttyCPM2
   CPM uart[2]:startup - port->irq 43
but brg's stay 0x0 and my shell's lock up something fierce.

Hmm...  try putting a debugging printk in cpm_setbrg with the address
and value it's writing.

Is it just the shell that locks up, or the console (i.e. can you still
get kernel messages)?

quoted

(note: my tree is still based off of one you provided a few weeks
back.  have the 82xx / ep8248e changes we went through been pushed
into linus' git tree?)

It should all be in Linus's tree except the actual ep8248e file, which
didn't make it this time around due to interdependencies on netdev.
Note that there are some small device tree changes from the tree you're
currently using (in particular, muram is now in a node of its own).

-Scott