raymond yang wrote:
Hi All,
My board got problems while running m_test, any idea,
Looks like the high order two bytes mismatched, while low part correct.
Does this mean we failed to configure the SDRAM or due to hardwarer issue?
thanks!
Best Regards Raymond
=> mtest Pattern 00000000 Writing... Reading... Mem error @ 0x00110000: found 00044000, expected 00004000
Mem error @ 0x00118000: found 40246000, expected 00006000
Mem error @ 0x001A0000: found 00268000, expected 00028000
Mem error @ 0x001B0000: found 0006C000, expected 0002C000
Mem error @ 0x001F8000: found 0427E000, expected 0003E000
Mem error @ 0x00200000: found CC320000, expected 00040000
Mem error @ 0x00208000: found C4322000, expected 00042000
Hi Raymond,
Suspicious: only one error per instance, and that on a 4KB boundary. Sounds like Something Bad[tm] is happening on a page boundary (e.g. page close, new page open). For the common hardware faults of shorted address and/or data lines, you would get a whole burst of errors when the affected lines are exercised.
Questions: * What is your memory configuration (x8, x16, x32)? What is the SDRAM page size? * Are these the only errors found? * Can you use the command line to write to the error addresses and read back the proper values? If not, what do you get? Use various patterns, are the upper two bytes random or related to the lower two bytes?
Inquiring minds want to know... gvb