Messages in this thread | | | Date | Sat, 6 Dec 2008 03:10:33 +0100 | From | "Kay Sievers" <> | Subject | Re: Runaway loop with the current git. |
| |
On Fri, Dec 5, 2008 at 22:24, Evgeniy Polyakov <zbr@ioremap.net> wrote: > On Fri, Dec 05, 2008 at 10:17:01PM +0100, Kay Sievers (kay.sievers@vrfy.org) wrote: >> > And what's with tty drivers? If kernel freezes there even if suddenly >> > faulty userspace started to load them, where this can happen? Apparently >> > it is not sleeping userspace since console does not respond to input. >> >> 5-1 dev_t request looks like something in initramfs accesses >> /dev/console, but the driver for it is not properly >> initialized/registered that time, and the kernel module loader tries >> to load a module? The forked process might also try to access >> /dev/console again, and hence the loop? > > But why system freezes? Is this init linking/__init calling order > changes?
Can you add something like: +++ b/kernel/kmod.c @@ -108,6 +108,7 @@ int request_module(const char *fmt, ...) return -ENOMEM; }
+ printk("XXX call modprobe %s %s[%u]\n", module_name, current->comm, task_pid_nr(current));
It may show which process is looking for /dev/console and causes modprobe to run, and maybe we get an idea what's going on. It may at least show if it's a /dev/console problem.
Thanks, Kay
| |