Florian Behrens
01-11-2005, 06:09 AM
Hi,
has anybody yet tried to get Linux 2.6.14 running on a MPC8349E?
Compilation works fine using the default config (mpc834x_sys_defconfig)
but after booting the kernel with U-Boot 1.1.3 it starves after the
line 'Uncompressing Kernel Image ... OK'. Using my debugger I can debug
up to the RFI instruction in turn_on_mmu (head.S). Probably a problem
setting up the MMU correctly?
I'm not sure if this is relevant, but I'm using the default DIP switch
settings in the Getting Started doc.
Any hints appreciated,
Florian.
(By the way: Why does 'make ... vmlinux' and 'make ... uImage' works
fine, but 'make ... <without target>' not?)
prabhat_singh@satyam.com
07-11-2005, 10:25 PM
Hi,
I feel your RAM is not working in "BURST" Mode thats why kernel
is crashing.
So try to increase the instruction accessing interval or delay by
register setting .
Thanks
Florian Behrens
12-11-2005, 06:36 AM
Hi,
thanks for your help, but I rather think that it's a problem with MMU
setup (probably connected to wrong DIP switch/RCW settings?).
After the mentioned RFI instruction, the CPU jumps to start_here
somewhere in segment 0xC000xxxx, which is quite reasonable.
Unfortunately there is no code, which is still in segment 0x0. Why? Is
there any documentation out there (except from the commenting in the
code) which explains, what head.S does exactly?
Did anybody of you get Linux running on MPC834x? Did you use the
standard configuration and DIP switch settings?
Thanks in advance,
Florian.
prabhat_singh@satyam.com schrieb:
> Hi,
> I feel your RAM is not working in "BURST" Mode thats why kernel
> is crashing.
> So try to increase the instruction accessing interval or delay by
> register setting .
>
> Thanks
Tauno Voipio
13-11-2005, 06:11 AM
Florian Behrens wrote:
> Hi,
>
> thanks for your help, but I rather think that it's a problem with MMU
> setup (probably connected to wrong DIP switch/RCW settings?).
>
> After the mentioned RFI instruction, the CPU jumps to start_here
> somewhere in segment 0xC000xxxx, which is quite reasonable.
> Unfortunately there is no code, which is still in segment 0x0. Why? Is
> there any documentation out there (except from the commenting in the
> code) which explains, what head.S does exactly?
>
> Did anybody of you get Linux running on MPC834x? Did you use the
> standard configuration and DIP switch settings?
The kernel should map the loaded and uncompressed kernel
code starting from virtual address 0xc0000000 before
jumping in.
Maybe the MMU is not enabled in time.
The standard method is to map the kernel into two blocks
of virtual addresses: an unity mapping and a mapping to
the final virtual address, then enable the MMU, jump
to the final virtual location and unmap the unity mapping.
--
Tauno Voipio
tauno voipio (at) iki fi