Netconsole Problems During Boot - Any Ideas of What to Check?

[Arlie Stephens]

Hi Folks,

I'm attempting to use netconsole to get a look at boot time kernel
messages from a headless system. The goal is to get a look at things
that happen before the system is up enough to successfully log these
messages to its local log files. 

The behaviour I'm seeing is that about 12 seconds of messages are
lost, *after* netconsole is successfully enabled. Other messages
arrive normally.  The NIC driver appears to be sending the missing
messages, but they aren't making it to neighbouring
systems... probably not making it to the wire at all, but I can't
quite prove that.  

I have built netconsole and my nic driver (igb) into the kernel,
rather than leaving them as loadable modules. I'm passing the
netconsole parameters to the kernel on its command line. Netconsole is
initializing before the system would otherwise get around to setting
its IP address, so I'm supplying that address as part of the
netconsole parameters. 

I'm passing "debug" on the kernel command line to make sure that
everything in dmesg _should_ have been sent to all consoles, including
netconsole. 

I've looked at the network traffic with wireshark as well as with the
normal receiver (netcat), and checked for anything coming from the
relevant MAC address. 

I've instrumented the NIC driver, and counted calls to the xmit
routine, along with error paths; the number of calls match the number
of printks, and the error paths I instrumented are not being taken. 

I'm fresh out of ideas ;-)  Any suggestions for where to dig would be
appreciated. 

ARCH linux. Kernel 3.13 with a few patches. x86_64.


-----cut n paste from dmesg output, around the point when data gets out---
[    6.292772] input: HID 0557:2419 as
/devices/pci0000:00/0000:00:1a.0/usb1/1-1/1-1.3/1-1.3.1/1-1.3.1:1.1/input/input5
[    6.293024] hid-generic 0003:0557:2419.0004: input,hidraw3: USB HID
v1.00 Mouse\
 [HID 0557:2419] on usb-0000:00:1a.0-1.3.1/input1
[    7.064376] mpt2sas0: host_add: handle(0x0001),
sas_addr(0x5003048011879600), phys(8)

****above gets lost; below really gets onto the wire***

[   12.196218] mpt2sas0: port enable: SUCCESS
[   12.198261] scsi 0:0:0:0: Direct-Access     ATA
ST4000DM000-1F21 CC52 PQ: 0 ANSI: 6
[   12.198400] scsi 0:0:0:0: SATA: handle(0x000b),
sas_addr(0x4433221102000000), phy(2), device_name(0x5000c5006d273da6)
[   12.198527] scsi 0:0:0:0: SATA:
enclosure_logical_id(0x5003048011879600), slot(1)
[   12.198800] scsi 0:0:0:0: atapi(n), ncq(y), asyn_notify(n),

Thanks for any ideas. I'm feeling like a complete idiot. Or maybe just
someone who really doesn't like dealing with drivers ;-) 


--
Arlie

(Arlie Stephens					arlie at worldash.org)