Error in mapping physical address 5084

[chandramills]

I have Starcore code that I have compiled in 10.8.3. I then used the sc100-elf2xx.exe to generate a binary from the .eld file built by CodeWarrior. I then loaded the binaries to the Starcore from the PowerPC on my b4860 EVM, and the binaries loaded and ran successfully. I have since moved the code to version 10.9.0 and updated the variables to match the version 10.9.0 variables. I used version 10.9.0's sc100-elf2xx.exe to generate binaries from the .eld files, but when I tried to load the binaries, I got the message

Error in mapping physical address 5084 to virtual address in load_dsp_image
Error in loading Dsp image StarCore

I then tried version 10.8.3's sc100-elf2xx.exe to generate binaries from the 10.9.0 .eld file, and the binaries loaded successfully. After looking at the differences between the two binaries, I found that 10.9.0's binary added extra hex values, the first 64 bits being 0x0000000000005084, the address that generated the error message. It seems that this is a bug in 10.9.0 that needs to be fixed. Any suggestions on how to move forward with 10.9.0?

[chandramills]

When I built the 10.9 code, I forgot to include the flag "-ccsr_address 0xFFEc40000". Once I did that, the DSP binaries loaded successfully, but now I think the DSPs aren't booting correctly. I get a "sleep 1, waiting for hw sem" loop.

Why would this be happening?