Hello,

I am having the exact same problem. It doesn't return to start_here label after executing RFI instruction. (linux-2.4.20)

linux-2.4.20/arch/ppc/kernel/start.S ==>

turn_on_mmu: mfmsr r0 ori r0,r0,MSR_DR|MSR_IR mtspr SRR1,r0 lis r0,start_here@h ori r0,r0,start_here@l mtspr SRR0,r0 SYNC RFI /* enables MMU */

I have 128Meg or SDRAM on a MPC8260 board. The initialization sequence in U-BOOT is as follows.

void initdram() { #define CFG_SDRAM_BASE 0x0 #define CFG_IMMR 0x60000000

int i; int size; int cpu_type=0; volatile immap_t *immap = (immap_t *)CFG_IMMR; volatile memctl8260_t *memctl = &immap->im_memctl; volatile uchar c = 0; volatile uchar *ramaddr = (uchar *)(CFG_SDRAM_BASE) + 0x110;

/* Set OR1 and OR2 values */ memctl->memc_or1 = 0xf8002940; memctl->memc_br1 = 0x00000041;

/* * Quote from 8260 UM (10.4.2 SDRAM Power-On Initialization, 10-35): * * "At system reset, initialization software must set up the * programmable parameters in the memory controller banks registers * (ORx, BRx, P/LSDMR). After all memory parameters are configured, * system software should execute the following initialization sequence * for each SDRAM device. * * 1. Issue a PRECHARGE-ALL-BANKS command * 2. Issue eight CBR REFRESH commands * 3. Issue a MODE-SET command to initialize the mode register * * The initial commands are executed by setting P/LSDMR[OP] and * accessing the SDRAM with a single-byte transaction." * * The appropriate BRx/ORx registers have already been set when we * get here. The SDRAM can be accessed at the address CFG_SDRAM_BASE. */

memctl->memc_psrt = psrt; memctl->memc_mptpr = mptpr;;

memctl->memc_psdmr = (ulong) psdmr | PSDMR_OP_PREA; *ramaddr = c;

memctl->memc_psdmr = (ulong) psdmr | PSDMR_OP_CBRR; for (i = 0; i < 8; i++) *ramaddr = c;

memctl->memc_psdmr = (ulong) psdmr | PSDMR_OP_MRW; *ramaddr = c;

memctl->memc_psdmr = (ulong) psdmr | PSDMR_OP_NORM | PSDMR_RFEN; *ramaddr = c;

I have turned off Data cache in U-BOOT. If I enabel D-CACHE, the code hangs in U-BOOT itself.

Any help is very much appreciated.

Regards, Hebbar.

\ OFE8EA1A87.9A7B8A06-ONC1256D86.0028950A-C1256D86.002D5CE3@se.wavetek.com you wrote:

I've debugged and found that the board hangs in start_here function and doesn't reach the early_init function.

Because the problem appears on only one of my boards out of 16, I suspect a hardware problem, and because it occurs just after mmu initialisation and caches enabling, I suspect more especially sdram accesses.

I want to know if anybody has encountered a similar problem and if the diagnostic was the same.

This is a standard problem, which can show up in many forms. My bet is that your SDRAM initialization sequence is incomplete / broken. Remember that it is NOT sufficient to set up the mmeory controller, you must also intialize the SDRAM chips themself.

Best regards,

Wolfgang Denk

-- Software Engineering: Embedded and Realtime Systems, Embedded Linux Phone: (+49)-8142-4596-87 Fax: (+49)-8142-4596-88 Email: wd@denx.de "What is wanted is not the will to believe, but the will to find out, which is the exact opposite." - Bertrand Russell, _Sceptical_Essays_, 1928

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/