In message 20040105064054.54073.qmail@web15201.mail.bjs.yahoo.com you wrote:
Breakpoint 6, dram_size (mamr_value=546328576, base=0x0, maxsize=67108864) at RPXlite.c:199 (gdb) d b (gdb) step (gdb) step (gdb) step (gdb) step (gdb) step (gdb) step (gdb) step
Instead of using DDD where nobody can see what's going on it would have been much more useful to use plain GDB so we could see at which line this happened. Also, setting "d/i $pc" is usually very useful.
Program received signal SIGSTOP, Stopped (signal). 0x00001000 in ?? () (gdb)
By DDD,I could see the code executed to the following cycle and hanged after one cycle.At the same time,cnt and maxsize equal to 16777216 and 67108864 respectively.
for (cnt = maxsize/sizeof(long); cnt > 0; cnt >>= 1) { addr = base + cnt; /* pointer arith! */ save[i++] = *addr; *addr = ~cnt; }
Really? With just 8 steps I would think you should not be able to perform the init part of the function plus two full loops through the loop.
How big is your SDRAM_MAX_SIZE #defined? And how big is your actual RAM?
I set SDRAM_MAX_SIZE as 0x04000000 according to two MT48LC16M16A2TG£75 on LITE_DW.
So it seems the problem hapens when you actually access the RAM for the first time. The previous cycles through the test loop were still outside the available RAM area (testing at 64 MB and 32 MB);
Thank you for your notes.Actually,I cannot help suspect that address multiplexing on LITE_DW makes the problems.I notice that A6mpc is connected to A12sd but A8mpc and A7mpc linking BA0sd and BA1sd on LITE_DW.Some documents from Micron Technical Notes,howerever, say the MSB for the SDRAM are the bank address lines BA1 and BA0.I am a little confused by these authoritative reference.Which one is better for practice?Is it related with above phenomenon?
Sorry, but I don't know anything about the RPXlite hardware.
Best regards,
Wolfgang Denk