8266 SMC uart corruption

[Rune Torgersen]

Found more info:

The corruption happens when cpm_uart_set_termios() is called.
I have no idea how to fix it, except for putting in a udelay(10000) right before the smc mode register gets rewritten.
 

> -----Original Message-----
> From: Rune Torgersen 
> Sent: Tuesday, February 08, 2005 11:46
> To: linuxppc-embedded at ozlabs.org
> Subject: RE: 8266 SMC uart corruption
> 
> More Info:
> This seems to be related to cpm_uart_stop_tx()
> There seems to be an excessive amount of stops and starts going on
> In the following, I print an \ for each time stop_tx is 
> called, and a / for start_tx
> 
> ---------
> EXT3-fs: mounted filesystem with ordered data mode.
> VFS: Mounted root (ext3 filesystem).
> Freeing unused kernel memory: 288k init
> \/I\/T\/·\/vers\/i\/on 2\/.\/85 b\/o\/otin\/ /
> \/Acti\/v\/atin\/g\/ all\/ \/swap\/ 
> \/file\/s\//par\/t\/itio\/n\/s...\<6>Adding 1028152k swap on 
> /dev/hda2.  Priority:1 extents:1
> /       \/[  \/OK\/  ]\/
> \/Moun\/t\/ing \/p\/roc \/f\/ile \/s\/yste\/m\/...\/    \/[  
> \/OK\/  ]\/
> \/Moun\/t\/ing \/r\/oot \/f\/ile \/s\/yste\/m\/ in 
> \/r\/ead-\/o\/nly \/m\/ode.\/.\/.\/
> 
> ----------
> And so on.
> 
> If I put in a udelay(10000) at the end of cpm_uart_stop_tx, 
> all output prints OK, but slow.
> 
> 2.6.7 and prior does not use the stop at all form what I can see.
>  
> 
> > -----Original Message-----
> > From: Rune Torgersen 
> > Sent: Tuesday, February 01, 2005 08:42
> > To: linuxppc-embedded at ozlabs.org
> > Subject: 8266 SMC uart corruption
> > 
> > Hi
> > 
> > Has anybordy else seen corruption of the output from SMC1 
> > when used as serial console?
> > This only occurs in in 2.6.8 and later (when the CPM serial 
> > drivers got moved)
> > And it only occurs when init is writing to screen. 2.6.7 and 
> > earlier works perfectly.
> > It happens at all baudrates tested (9600 and 115200)
> > 
> > Anybody seen something similar and/or have a clue where to 
> > start looking for this?
> > I have looked at the serial drivers, and cannot find any 
> > differences large differences from whenthey still were in 
> > arch/ppc/8260_io to the ones in drivers/serial/cpm_uart
> > 
> > Capture:
> > 
> > Freeing unused kernel memory: 288k iεt
> > I■T: version 2.85 boot∙g
> > ...
> > I■T Entering runlevel: 3
> > 
> > 
> > These lines are the only ones that get corrupted. All other 
> > output (kernel printouts, console and startup scripts) are normal.
>