Config. tips sought: Serial-console garbles boot messages.

Config. tips sought: Serial-console garbles boot messages.

[Ali Bahar]

Hi all,

would someone please be kind enough to point me in the right direction
for this! 

Symptom:

- serial console grables all the boot messages, until just at/before
the log-in prompt. Then, if I reboot, the 'shutdown' messages continue
to be seen correctly until boot-up begins again.
Sample output is provided further down.

I expect that something basic is missing here, but this has already
sucked up too much of my time! So I'll be very grateful for any
pointers as to where I should look.

Additional info is provided below.

Thanks much,
Ali


Context: 

- serial console is set up between two boxes. 
- The boxes' on-board COM ports work fine in every respect.
- a new PCI card provides additional COM ports for one box. Serial
  console does not work with this. This is the problem.
- After extensive testing, the culprit hardware has been confirmed to
be the new PCI card. The cables and the peer COM ports have been as
verified as is sane to do! Ditto the kermit/minicom/cu configuration.

- the "serial" driver (8250_pci.c) mentions this card as supported,
but I do not know the extent of its supports. Its initialization
function, pci_netmos_init(), _is_ in System.map -- so it's built in!


Additional info:

- the PCI card provides 2 COM ports. It is
  03:02.0 Serial controller: NetMos Technology PCI 9865 Multi-I/O Controller
  (For the benefit of googlers, it is an Uptech/Upmost UT300 RS-232
  PCI card. Chip: MOSCHIP MCS9865. 16C550 UART.)
  It is a high-speed port, but I run it at 115200, and have even tried
  9600 baud.

- the kernel is a very recent 2.6.39-rc5-wl, wireless-testing.git.
- the kernel command-line includes
  console=ttyS1,115200  console=tty0

- getty does not work on this. Nor does any input.
  (I'd take this observation with a tiny grain of salt, because IIRC
  there was one scenario in which it may have worked.)

- /proc/tty/driver/serial shows no rx, but its tx count requires more
  testing by me.

- Remote-Serial-Console-HOWTO, Serial-HOWTO and
  Documentation/serial-console.txt have not been fruitful, so far.



Sample output:

The garbled 'text' is the entire set of boot messages!

[ 1105.856980] ehci_hcd 0000:00:1a.7: PCI INT C disabled
[ 1105.917787] PM: Calling i8259A_shutdown+0x0/0x10
[ 1105.973273] Restarting system.
[ 1106.009954] machine restart
�u6�'&'-D�'6DL5$�t�$���Nť�4.䞎��UN�4%T����U5�%'�4e$ħNť�5.D��Nť���E46�e$Į�%7�4e$
<U+0527>6�d,\-���eMtnVm&6d�m&%7e�漍t$֞�&$M'6�$<�M&DD
'6Ď�&$��&6�Fͤ�,T�'6�6v~�&7~�-&E�m&%��&6dFM�n&��'������4T-�T
                                                          �6.�fD�&��e$^V>�V�>���O<$Pf$.����f�%7�F%w�$mm<���Υ����~�'n�GEd�.��d�7v�6f�Vf�6f�6D�E�d�FNF�6V&$d'4�d�ƥ�54�6�^e�4�$��%���t��d�%���F/��d=d��>�Dd$��lN'&v-�/�|-���/�^OĶ6D]6��$D�>M4�}<&m'�o/D$mm|g�>4���Dmm<Ǥ�$w47�o�$�V$Lm��|=tDdD��dD'�$$�tf=V�l�6�w�D���D�,�5���t�D�d>4.�nDt��d$�t$D�$�
&l$�mT$.F�mTDt��d�d4�t4�d.T��D��<�e%/$%/&��/4�n.Tm���eD
DDT�����%�<�D��<�g'?T%-�&eDMm<ť�d$MDt��$%T>�%�>Ĕd�mL.6.$%&\mm<���DDmt4���ż<�$t\�.dod<�g'/t=�f�.4�D��T%tv%�%vF�f%�<$�<%��,���$tv�'�$�4%T�E$�E46�E$���%�
                                                                      �4,䔧Nť��4.䄮�%7�E$<U+E3AE>���4,䦌�����46�E$Į�%7�5dd���%'�4e$�E46�5d$���6��t����ΧLť�T%.DEM%D�dt4V榴���'�F�6L�4'�4�D֬��ml'$݆�fL�L��'�-�F-&�-ĴM5�fn'&�D�D��l
                                                       Nm����������[
12.184356] sshd (1529): /proc/1529/oom_adj is deprecated, please use
/proc/1529/oom_score_adj instead.
[   13.930906] r8169 0000:02:00.0: eth0: link up
[   13.982514] ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready

--
To unsubscribe from this list: send the line "unsubscribe linux-serial" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Config. tips sought: Serial-console garbles boot messages.

[Ali Bahar]

On Wed, May 04, 2011 at 12:21:06PM +0800, Ali Bahar wrote:

> - serial console grables all the boot messages, until just at/before
> the log-in prompt. Then, if I reboot, the 'shutdown' messages continue
> to be seen correctly until boot-up begins again.

It should go without saying that the baud rates etc have been set to
match on both ends.

  Dev. Host                  Target's /proc/cmdline
  115200                     console=ttyS1,115200 console=tty0

Partity etc match, too.