Hi Barry,
Thanks for testing,
I think this patch introduced a sequencing issue.

Just re-looking at the code, is there any reason for this to be
init_earlycall? And do a ioremap at that early stage?

Can't we move it to device_init level function "sirfsoc_mach_init()" at
which it will be safe to do a ioremap with all mm setup finished?


Thanks,
srini

On 03/06/13 03:19, Barry Song wrote:

this broke SiRF by:
Uncompressing Linux... done, booting the kernel.
[    0.000000] Booting Linux on physical CPU 0x0
[    0.000000] Linux version 3.10.0-rc2-00723-gd1a0472-dirty
(barry at barry-laptop) (gcc version 4.5.4 20110505 (prerelease)
(Ubuntu/Linaro 4.5-2011.5-csr-build) ) #35 SMP PREEMPT Mon Jun 3
09:51:59 CST 2013
[    0.000000] CPU: ARMv7 Processor [413fc090] revision 0 (ARMv7), cr=10c53c7d
[    0.000000] CPU: PIPT / VIPT nonaliasing data cache, VIPT aliasing
instruction cache
[    0.000000] Machine: Generic ATLAS6 (Flattened Device Tree), model:
CSR SiRFatlas6 Evaluation Board
[    0.000000] bootconsole [earlycon0] enabled
[    0.000000] Memory policy: ECC disabled, Data cache writealloc
[    0.000000] Unable to handle kernel NULL pointer dereference at
virtual address 00000000
[    0.000000] pgd = c0004000
[    0.000000] [00000000] *pgd=00000000
[    0.000000] Internal error: Oops: 5 [#1] PREEMPT SMP ARM
[    0.000000] Modules linked in:
[    0.000000] CPU: 0 PID: 0 Comm: swapper Not tainted
3.10.0-rc2-00723-gd1a0472-dirty #35
[    0.000000] task: c05cfb40 ti: c05c4000 task.ti: c05c4000
[    0.000000] PC is at kmem_cache_alloc+0x2c/0x124
[    0.000000] LR is at __get_vm_area_node.clone.26+0x7c/0x190