- Ask a related questionWhat is a related question?A related question is a question created from another question. When the related question is created, it will be automatically linked to the original question.
Tool/software: Code Composer Studio
Hello all,
I'm working on a project where I'm adding a new feature to a board which runs a tms320c6455 DSP. The DSP is loaded through an HPI interface from a PowerPC chip, and it has 64MB of DDR2 attached.
The tools I'm using are:
Code Composer Studio 3.3.81.6
BIOS version 5.31.09
Before I started the project, the software was working just fine. I could run the code on the debugger. I could flash the board with an image and the PowerPC chip would load the code to the DSP and it would run just fine. Fast forward several months to the end of the project, and I'm putting the finishing touches on the new feature. I add a few lines of code and all of a sudden the DSP won't boot from the HPI interface anymore.
Starting from the code that doesn't run, here are the things I've tried so far to fix the issue:
1. Remove a few lines of code from IRAM. - This makes the issue go away.
2. Add a few lines of code to IRAM. - This makes the issue go away.
3. Remove a few lines of code from DDR2 RAM - This makes the issue go away.
4. Add a few lines of code to DDR2 RAM. - This makes the issue go away.
5. Run the code on a debugger with CCS. - This makes the issue go away.
6. Replace the code I added which caused the issue with function calls which just basically do a "nop" but keep the code size the same. - This does not fix the issue.
7. Remove all code added in this project, replacing it with function calls which basically do a "nop" but keep the same code size. - This does not fix the issue.
8. Remove all code from executing on the DSP except for a heartbeating thread which sends an IP packet periodically (too see if the DSP is alive) and the BIOS and replacing it with function calls which basically do a "nop" but keep the same code size - This does not fix the issue.
Additionally, I've checked the boot code to see how the PLLs and EMIF are set up. They match the GEL file setup for the debugger exactly. It's very currious that the code doesn't have the issue on the debugger, but when I flash the code to the board it doesn't work. Also, since I stripped out almost all of the application code, I feel like this can't be a logic issue in my software. I've spent quite a bit of time pouring over map files where I compare builds that run correctly with builds that don't, and the biggest takeaway I've seen with this is how much the compiler moves things around with even the smallest code change. Additionally, I find it odd that I can make the issue go away if I use more or less of the IRAM or more or less of the DDR2. When I experience the failure to boot, my IRAM usage is 0x1b9fb9 and the DDR2 usage is 0xb89aa0. (from the map file.)
My guess is that something is not lined up on an appropriate boundary somewhere depending on the code size, but I have no idea how to find out what or where it is.
Any recomendations or suggestions are greatly appreciated.
