Ubuntu
linux-source-2.6.15 package

kernel logging confused after "No module loaded"

Bug #25550 reported by Jörg Höhle on 2005-11-09

Affects		Status	Importance	Assigned to	Milestone
	linux-source-2.6.15 (Ubuntu)	Invalid	Medium	Ben Collins

Bug Description

Hi,

After my update from Hoary to Breezy, I observed a repeated pattern in my kern.log.
Every time after "No module symbols loaded", the next line is partially garbled.

Here are some examples:

Nov 9 13:43:31 localhost kernel: No module symbols loaded - kernel modules not
enabled.
Nov 9 13:43:31 localhost kernel: .617000] Checking if this processor honours
the WP bit even in supervisor mode... Ok.
^^^^^^^^
Nov 9 17:08:51 localhost kernel: No module symbols loaded - kernel modules not
enabled.
Nov 9 17:08:51 localhost kernel: 000] Mount-cache hash table entries: 512
^^^^
Nov 9 17:56:25 localhost kernel: No module symbols loaded - kernel modules not
enabled.
Nov 9 17:56:25 localhost kernel: nitialized
(the latter may be a typo in kernel sources?)

Nov 9 15:21:18 localhost kernel: No module symbols loaded - kernel modules not
enabled.
Nov 9 15:21:18 localhost kernel:

Nov 9 15:38:12 localhost kernel: Inspecting /boot/System.map-2.6.12-9-686
Nov 9 15:38:12 localhost kernel: Loaded 28225 symbols from
/boot/System.map-2.6.12-9-686.
Nov 9 15:38:12 localhost kernel: Symbols match kernel version 2.6.12.
Nov 9 15:38:12 localhost kernel: No module symbols loaded - kernel modules not
enabled.
Nov 9 15:38:12 localhost kernel: 00000000 00000000
Nov 9 15:38:12 localhost kernel: [4294667.385000] Intel machine check
architecture supported.

BTW, I don't understand "kernel modules not enabled" since the system seems very
well able to load kernel modules on demand, e.g. iptables. What's the actual
meaning of this message?

The machine is a Fujitsu/Siemens laptop Lifebook C1110.

Regards,
Jörg Höhle

Revision history for this message

Jörg Höhle (joerg-cyril-hoehle) wrote on 2005-11-09:

It just occured to me that the relevant point seems not to be "after modules
loaded message", but instead before the first time when kernel want to start
using the new (vs. Hoary) [44444444.5555555] format.

Nov 9 16:32:37 localhost kernel: Symbols match kernel version 2.6.12.
Nov 9 16:32:37 localhost kernel: No module symbols loaded - kernel modules not
enabled.
Nov 9 16:32:37 localhost kernel: 0 00000000
Nov 9 16:32:37 localhost kernel: [4294669.634000] CPU: L1 I cache: 32K, L1 D
cache: 32K
Nov 9 16:32:37 localhost kernel: [4294669.634000] CPU: L2 cache: 2048K

Using linux-image-2.6.12-9.23-686

Revision history for this message

Ben Collins (ben-collins) wrote on 2005-11-09:

This can't be helped. The new "format" is what comes out of the kernel. The
problem is one of timing. Sometimes a kernel module will print something like:

printk(KERN_INFO "foobar: Checking blah blah blah...");
/* do something */
printk("done.\n");

After the first printk, and before the next, the code can be interrupted by
another process or interrupt that also spits out a printk, which pretty much
overwrites in the middle of the line.

The other thing is that the klogd is started _well_ after the kernel starts
printing the messages, and must first consume the entire kmsg buffer.

So, this isn't an issue that is easily fixes, nor do I consider it broken really.

Revision history for this message

Jörg Höhle (joerg-cyril-hoehle) wrote on 2005-11-10:

Ben wrote:
>klogd is started _well_ after the kernel starts
>printing the messages, and must first consume the entire kmsg buffer.

Good to know.
Since the pattern is repeatable and happens once at the start, I now wonder
whether the kmsg buffer is too small for the current amount of messages,
therefore it just starts printing contents of a rolled-over ring buffer, i.e.
somewhere amid a line?

Increasing the kernel buffer size could then solve that issue and not loose
messages. I'll see if I can find a place where this is configured, without
building a custom kernel