A patch propagation was missed. The following patch should fix the
problem. The MPSC driver gets initialized multiple times, it appears,
and the driver has to ignore all but the first.
-corey
Prevent the console from being enabled twice.
Signed-off-by: Corey Minyard [EMAIL PROTECTED]
I backported the mpsc driver from 2.6.23 to 2.6.21 (drivers/serial/mpsc.c)
and it works fine now, there's not much change between the two in the diff,
but enough to make it works.
/jonathan
On Nov 12, 2007 11:18 AM, Corey Minyard [EMAIL PROTECTED] wrote:
Not sure. 2.6.23 boots up on a 74xx
Am I gonna run in any trouble backporting the serial-marvel-poll stuff from
2.6.23 patches?
It's not included with 2.6.21, only the 8250.
It doesn't seem to be fine, I backported all the changes against
drivers/serial/mpsc.c in 2.6.21, and when the kernel boots, after the ttyMM1
init it just
Is it coming from that patch?
http://osdir.com/ml/linux.kernel.openipmi/2006-06/msg3.html
On Nov 9, 2007 3:58 PM, Corey Minyard [EMAIL PROTECTED] wrote:
I'm not really sure. That actually hasn't been tested in its current
form, the patch was ported from something else. So it very well
No, that's not it. It's an internal thing. You are having trouble with
the console rework. I'll try to review the patches to make sure
everything is there. But maybe you'll trace it down before then. I'm
pretty sure everything is there.
-corey
Jonathan Fournier wrote:
Is it coming from
Without any marvel-poll patches, kernel loads fine but when I try to load
the ipmi_serial it says ttyMM1 doesn't exists (which I know does), with the
marvel-poll 2.6.23 patches backported to 2.6.21, I get that error right
after the ttyMM0/ttyMM1 (i.e. ttyMM0 at blah using IRQ(xx)...) init lines
I can't tell why yet, but it seems to be related with this
.cons= MPSC_CONSOLE,
Before it was set in the struct uart_driver mpsc_reg right away, now it's
waiting a bit in mpsc_polled_init() and set it at the end just before doing
uart_register_polled()
If I put back on the way .cons was
I'm not really sure here. There was a driver that worked with this
board, but it had a lot of problems. Looking at that driver and this
data, the only thing I can see is that the other driver terminated the
message with a 0x0d, not a 0x0a. It may be best to send both, so I'm
attaching a
Jonathan Fournier wrote:
I have a 2.6.21 kernel and the all the patches for it from SF v39.1.
Could it clash with the kernel softdog if enabled? I think there's
another option for other watchdog to register their minor number to
it? I'm not really familiar with that thing.
Yes, softdog
Damn, it was that! It's now working like the other boards.
Thanks for your time,
/jonathan
On Nov 7, 2007 9:34 AM, Corey Minyard [EMAIL PROTECTED] wrote:
I'm not really sure here. There was a driver that worked with this
board, but it had a lot of problems. Looking at that driver and this
I assume you tried the ppInvIANA option, too.
There is a debug option in the ipmi_serial module, try setting debug=7
and generate output.
I think I can get access to a 7101, I'll try it out. This hasn't been
tested on a 7101 yet.
-corey
Jonathan Fournier wrote:
Hi,
I'm using the
ppInvIANA? I only used the options that were in the Documentation/IPMI.txt
once patched.
Is it the same option than pp / nopp ?
I think the 7101 is using an old BMR-100 from Pigeon Point (C5470) in
TerminalMode
/jonathan
On Nov 6, 2007 3:38 PM, Corey Minyard [EMAIL PROTECTED] wrote:
I assume
BTW, I couldn't get access to a 7101, but if you can give me access I
could hack on it a little.
-corey
Jonathan Fournier wrote:
I added the patch, it's not complaining that the option doesn't
exists, bit it stills time out...
How do I enable debug=7 ?
/jonathan
On Nov 6, 2007 4:17 PM,
It's a standard module option. If it's a module, you just do debug=7
on the modprobe/insmod line. You can set it using ipmi_serial.debug=y
on the kernel command line if it's compiled into the kernel. You can
dynamically change it using echo 7
/sys/modules/ipmi_serial/parameters/debug
I added the patch, it's not complaining that the option doesn't exists, bit
it stills time out...
How do I enable debug=7 ?
/jonathan
On Nov 6, 2007 4:17 PM, Corey Minyard [EMAIL PROTECTED] wrote:
Jonathan Fournier wrote:
I get this:
[EMAIL PROTECTED]:/root /etc/init.d/ipmi start
Jonathan Fournier wrote:
I get this:
[EMAIL PROTECTED]:/root /etc/init.d/ipmi start
Starting ipmi drivers: Nov 6 20:29:00 localhost kernel: ipmi message
handler version 39.1
Nov 6 20:29:00 localhost kernel: IPMI Serial System Interface driver
Nov 6 20:29:00 localhost kernel: ipmi_serial:
I get this:
[EMAIL PROTECTED]:/root /etc/init.d/ipmi start
Starting ipmi drivers: Nov 6 20:29:00 localhost kernel: ipmi message
handler version 39.1
Nov 6 20:29:00 localhost kernel: IPMI Serial System Interface driver
Nov 6 20:29:00 localhost kernel: ipmi_serial: Registering TerminalMode
codec
Did you look in the debug log? There are various settings in
/proc/sys/kernel/printk (I think) that suppress logs of various levels
to the kernel, and all kinds of syslog settings. These will all come
out as debug logs.
-corey
Jonathan Fournier wrote:
Even with debug=7, I don't see more
Even with debug=7, I don't see more outputs from /var/log/messages ?
On Nov 6, 2007 5:14 PM, Corey Minyard [EMAIL PROTECTED] wrote:
BTW, I couldn't get access to a 7101, but if you can give me access I
could hack on it a little.
-corey
Jonathan Fournier wrote:
I added the patch, it's not
I'm getting this with debug:
ipmi message handler version 39.1
IPMI Serial System Interface driver
ipmi_serial: Registering TerminalMode codec
ipmi_serial(ttyS4): Found a matching serial port
ipmi_serial(ttyS4): Outgoing data: 5b([) 31(1) 38(8) 30(0) 30(0) 30(0) 31(1)
5d(]) 0a( )
20 matches
Mail list logo
