[Top] [All Lists]

Re: 2.6.14 - problem with malta

To: "Kishore K" <>, "Ralf Baechle" <>
Subject: Re: 2.6.14 - problem with malta
From: "Kevin D. Kissell" <>
Date: Sat, 25 Mar 2006 16:41:19 +0100
Cc: <>
Original-recipient: rfc822;
References: <> <> <>
See below...
----- Original Message -----
From: Kishore K
Sent: Saturday, March 25, 2006 10:22 AM
Subject: Re: 2.6.14 - problem with malta

On 3/24/06, Ralf Baechle <> wrote:
On Fri, Mar 24, 2006 at 08:06:43PM +0530, Kishore K wrote:

> hi
> I am trying to bring up the malta board (MIPS 4kec), using the 2.6.14 kernel
> downloaded from linux-mips. The kernel is built with malta_defconfig located
> in arch/mips/configs. After loading this kernel, board halts after printing
> "Linux Started
> Config serial console: console=ttyS0, 38400n8r"
> Kernel is built with the tool chain based on gcc 3.3.6, binutils
> .

Note I generally don't test with HJ.Lu's toolchain at all.  It may be
broken for MIPS, I don't know.  Also due to known bugs in older toolchains
the minimal requirement is binutils 2.16; the vanilla FSF version from will do.

(2.15 will core dump for some configurations but if it doesn't die, it seems
to do just fine)

Thanks for the reply. I observed the same problem  even when compiled with tool chain based on gcc 3.4.4 binutils 2.15.97 and uclibc- 0.9.27. I 'll try with binutils 2.16. Please note that my malta is based on 4kc but not 4kec.
Ah.  That makes a difference.  In your earlier mssage, you said you had a 4KEc.
The malta_defconfig file by default configures the kernel to be built assuming
MIPS32 Release 2.  The 4KEc is Release 2 compatible, but the older 4Kc and
5K cores are not.   You need to tweak the configuration such that
CONFIG_CPU_MIPS32_R1=y but CONFIG_CPU_MIPS32_R2 is not set
(it's the other way around in the malta_defconfig in the linux-mips git),
and, to be on the safe side, such that CONFIG_SYS_HAS_CPU_MIPS32_R2
isn't set.
Your failure was so early that I was wondering what on earth could have
nailed you, but if the kernel was built for Release 2, you would have had
"DI" (disable interrupts) instructions very early on, and these could have
caused a fatal early exception on a Release 1 part.
            Kevin K.
<Prev in Thread] Current Thread [Next in Thread>