[LinuxBIOS] problem about dram init - coreboot

20 Dec 2007


      Dear All,
I've configured the via cn700 dram controller, and when I did ram_check to
many regions(not all, 'cause it wastes time)
everything was ok. but after the second stage code finished running, filo
failed to be started, with the following errors:
... ...
Jumping to boot code at 0x10e5ec
entry    = 0x0010e5ec
lb_start = 0x00004000
lb_size  = 0x0001a000
adjust   = 0x1bfe2000
buffer   = 0x1bfcc000
     elf_boot_notes = 0x00015760
adjusted_boot_notes = 0x1bff7760
Unexpected Exception: 6 @ 10:1bfee821 - Halting
Code: 0 eflags: 00010006
eax: 1bfee821 ebx: 1bfe2000 ecx: 00000000 edx: 0001a000
edi: 1c000000 esi: 0001e000 ebp: 00019eec esp: 1bffbea0
the lb_mem is :
dumping lb_mem resources
entry : 0, start 0x0, size 0x1000, type 0x10
entry : 1, start 0x1000, size 0x9f000, type 0x1
entry : 2, start 0xc0000, size 0x30000, type 0x1
entry : 3, start 0xf0000, size 0x10000, type 0x10
entry : 4, start 0x100000, size 0x1bf00000, type 0x1
and the mtrr is :
Setting variable MTRR 0, base:    0MB, range:  256MB, type WB
ADDRESS_MASK_HIGH=0xf
Setting variable MTRR 1, base:  256MB, range:  128MB, type WB
ADDRESS_MASK_HIGH=0xf
Setting variable MTRR 2, base:  384MB, range:   64MB, type WB
the total memory I reported to system through ram_resource in
pci_domain_set_resource was tolmk(which is 512 * 1024) - 64 * 1024
the 64MB is for the integrated graphics card's framebuffer.
When I accidentally  reduce the system memory in ram_resource to tolmk - 2 *
64 * 1024, filo could be started, but linux kernel failed to be started.
I then used memtest to do the test, and memtest displayed many errors,
including the regions I ram_checked ok.
could anybody give me any hints? can I say ram is initialized right?
Thanks.