2006-02-25 01:50:32

by Chuck Ebbert

[permalink] [raw]
Subject: [patch] x86_64: don't use early_printk() during memory init

printk is working by the time this memory init message prints.
As it stands, output jumps to the top of the screen and prints
this message, then back to normal boot messages, overwriting
a line at the top.

Signed-off-by: Chuck Ebbert <[email protected]>

---

This bug was introduced 5 weeks ago and should be fixed before 2.6.16.

--- 2.6.16-rc4-64.orig/arch/x86_64/mm/init.c
+++ 2.6.16-rc4-64/arch/x86_64/mm/init.c
@@ -312,7 +312,7 @@ static void __init find_early_table_spac
table_start >>= PAGE_SHIFT;
table_end = table_start;

- early_printk("kernel direct mapping tables up to %lx @ %lx-%lx\n",
+ printk("kernel direct mapping tables up to %lx @ %lx-%lx\n",
end, table_start << PAGE_SHIFT, table_end << PAGE_SHIFT);
}

--
Chuck
"Equations are the Devil's sentences." --Stephen Colbert


2006-02-25 04:41:13

by Andi Kleen

[permalink] [raw]
Subject: Re: [patch] x86_64: don't use early_printk() during memory init

On Saturday 25 February 2006 02:45, Chuck Ebbert wrote:
> printk is working by the time this memory init message prints.
> As it stands, output jumps to the top of the screen and prints
> this message, then back to normal boot messages, overwriting
> a line at the top.

Using of early_printk here was intentional because it needs
much less infrastructure than printk and is pretty good proof
that the kernel at least started.

So please don't apply the patch.

-Andi

2006-02-25 06:25:20

by Chuck Ebbert

[permalink] [raw]
Subject: Re: [patch] x86_64: don't use early_printk() during memory init

In-Reply-To: <[email protected]>

On Sat, 25 Feb 2006 at 05:26:05 +0100, Andi Kleen wrote:

> > printk is working by the time this memory init message prints.
> > As it stands, output jumps to the top of the screen and prints
> > this message, then back to normal boot messages, overwriting
> > a line at the top.
>
> Using of early_printk here was intentional because it needs
> much less infrastructure than printk and is pretty good proof
> that the kernel at least started.

Well it made me think something had gone horribly wrong and was
scribbling over video memory. Especially since it left debris
at the end of the line... and you don't want to fix that either.

--
Chuck
"Equations are the Devil's sentences." --Stephen Colbert