Hi Gurumurthy,

GurumurthyGowdar wrote:

Dear All, i am using MPC8280 in our board which has 512MB of SDRAM i.e. from micron.

when i give command mtest to check memory from U-boot, machine check and program exception is coming after 40 - 60mins later it hangs.

i have checked the SDRAM configuration settings everything is fine as per the device datasheet.

Good. I would still be suspicious that this is a possible problem until you understand what is going on.

please i want to know where is the problem, what is the cause for this exception. why memory fails after 40-50mins.

That is quite a while before failure. It looks like you are running u-boot RAM tests which are not all that stressful (running linux is very stressful). Based on your listing below, I count 24 passes through memory. Not a whole lot of passes before failure.

error output is as attached:-

U-Boot 1.3.1 (Jul 27 2009 - 17:27:10)

MPC8260 Reset Status: External Soft, External Hard

MPC8260 Clock Configuration

• Bus-to-Core Mult 3x, VCO Div 4, 60x Bus Freq 16-50 , Core Freq 50-150

^^^^^^^^^^^^^^^^^ Bus freq 16-50MHz, core freq 50-150MHz, but...

• dfbrg 1, corecnf 0x10, busdf 4, cpmdf 1, plldf 0, pllmf 4, pcidf 3
• vco_out 500000000, scc_clk 125000000, brg_clk 31250000
• cpu_clk 300000000, cpm_clk 250000000, bus_clk 100000000

...this says your CPU clock is 300MHz and bus clock is 100MHz. Is your HRCW (other configuration?) correct for the clocking you are using???

I think your first task is to figure out if this is correct and fix it if it isn't.

CPU: MPC8260 (HiP7 Rev 14, Mask 1.0 1K49M) at 300 MHz Board: MPC8280 Based Remote IO CARD.SLN Technologies Pvt Ltd. I2C: ready DRAM: 512 MB FLASH: ## Unknown FLASH on Bank 0 - Size = 0x00000000 = 0 MB 256 MB *** Warning - bad CRC, using default environment

In: serial Out: serial Err: serial Net: FCC2 ETHERNET RIO->

RIO->

RIO->mtest 0x2000 0x1ff00000

Pattern 00000000 Writing... Reading... Pattern FFFFFFFF Writing... Reading...

[snip]

Pattern 0000000B Writing... Reading... Pattern FFFFFFF4 Writing... Reading...

Mem error @ 0x1C593D38: found F8E9B800, expected F8E9B8A6 Mem error @ 0x1D0037F8: found F8BFF900, expected F8BFF9F6 Mem error @ 0x1D582E78: found F8A9FC00, expected F8A9FC56 Mem error @ 0x1D80BF68: found F89FD800, expected F89FD81A

^^ Last byte is 0x00 when it fails... implies the LSbyte of the memory isn't responding. I would be suspicious that you have a signal integrity problem, especially suspicious of the control signals (OE*?) on the LSbyte of your memory subsystem - either the memories themselves or any buffers you have in the data path.

If you CPU configuration (above) is wrong, that could very well cause timing issues with the SDRAM. Otherwise I would use a high speed scope with SHORT leads and look at your control signals. Comparing the upper byte signals with the LSbyte may be informative.

Since it takes a while to fail, it may be heat related (not that likely, however).

Machine check in kernel mode. Caused by (from msr): regs 1ffa1cd0 Unknown values in msr NIP: 1FFE0048 XER: 20000000 LR: 1FFE0054 REGS: 1ffa1cd0 TRAP: 0200 DAR: 00000000 MSR: 0000b000 EE: 1 PR: 0 FP: 1 ME: 1 IR/DR: 00

GPR00: 1FFE0054 1FFA1DC0 EFFFFFFF 1FFF3308 1D8D0A18 F1398ADC F89CC56E 00000030 GPR08: 00000041 F0000088 00000000 F0000091 1FFA1B78 FFF7FEFB 1FFFE000 1FFF44B8 GPR16: 1FFF6E08 00000000 00000000 00000000 1FFF32D0 1FFF32F8 1FFF32FC 1FFF3308 GPR24: FFFFFFFF 00002000 1FF00000 FFFFFFF4 7F9CC213 1FFA1F6C 1FFFE7F4 1D8D0A1C Call backtrace: machine check

Ouch, bad. grep says that is coming from ./cpu/mpc8260/traps.c: MachineCheckException().

do help me out in this regard,

with thanks and regards, gurumurthy

Good luck, gvb