Re: Debian kernel regression, was Re: Modernizing a Macintosh LC III
On Thu, Nov 21, 2013 at 10:38 AM, Finn Thain <fthain@telegraphics.com.au> wrote:
> On Thu, 21 Nov 2013, Michael Schmitz wrote:
>
>> Serial or early fb console looks the only way to debug this.
>
> It hangs at "K", which means that none of the messages that would normally
> appear on the early console have appeared yet.
There's not that much happening anymore in head.S:
putc 'K'
subl %a6,%a6 /* clear a6 for gdb */
/*
* The new 64bit printf support requires an early exception initialization.
*/
jbsr base_trap_init
/* jump to the kernel start */
putc '\n'
leds 0x55
jbsr start_kernel
I guess we can't see whether the '\n' was printed?
It would only be visible on a serial console, or when it has to scroll the
screen when reaching the bottom.
And after that we're in C land.
> Based on the first screen shot posted by Patrick, the first couple of
> lines to appear should have been,
>
> [ 0.000000 ] Linux version 3.12...
> [ 0.000000 ] m68k_parse_bootinfo...
> [ 0.000000 ] bootconsole [early0] enabled
>
> Problem is, I've no idea why none of this shows up. Is it because the hang
> happened before the bootconsole was enabled? Maybe even before before the
> first printk ("Linux version 3.12...")?
>
> Is there a way to get console messages given that the bootconsole might
> not be enabled yet? Otherwise, I still favour "git bisect".
>
> Though I suppose we could go fishing based on the interrupt hypothesis,
> and somehow inhibit the normal interrupt initialisation, so that the
> bootconsole might start. Not sure where that leads though.
When debugging things like this, I usually end up adding debug prints to
init/main.c:start_kernel() initially, and see what happens...
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
Reply to: