11-11-2011 12:52 AM - edited 11-11-2011 12:52 AM
I am trying to use the embedded PPC on the ML-405 board, and created a simple system that only has a rs232 uart and some bram. I then create a helloworld c program from within sdk, and it seems to work fine in version 13.1 of the tools. However, in both 13.2 and 13.3, when I create the same simple system in XPS and try to run helloworld from SDK, nothing seems to work. The only potential clue is an error on the console that says "Unable to decode ELF start address (), assuming 0x0." This error appears in 13.2 and 13.3, but not in 13.1. Can anyone suggest a way to determine why the embedded ppc doesn't seem to respond in 13.2 and 13.3 and whether this new error message is relevant or not?
I am fairly new to the EDK and SDK tools, and may be overlooking some basic things, so please do not assume I know what I am doing. Any help would be greatly appreciated. I tried searching on this but did not find anything yet, and any links to further reading would also be appreciated.
01-06-2012 10:08 AM
In case anyone else has this problem, here is some information after getting help from xilinx tech support.
The best that we can tell, this is because I am using WIndows 7 Enterprise instead of Windows 7 Professional.
The workaround is to use the xmd console and manually send the binary to the ppc using the "dow" command.
01-12-2012 02:48 AM
Hi, John, I have this problem also. and there is no any solution after I pass this problem to xilinx last two month .
The workaround is to run your project in debug mode or in shell mode.
01-12-2012 03:50 AM
This has been highlighted and noted by the developers and the conclusion was that this happens on some machines and not on others and so there is no direct fix. Can you try another machine if you have on at your disposal?
As you stated the workaround is to use the XMD console to download and run the application directly.
Embedded Product Applications Engineer, WTS
05-30-2012 08:52 AM - edited 05-30-2012 08:56 AM
Same problem here with XP Pro and 13.2.
Console says "20:38 ERROR : Unable to decode ELF start address ( -4), assuming 0x0.".
The elf entry point is 0xfffffffc which is the location of the .boot section. Sorry guys, for me this clearly looks like an unsigned/signed code typo somewhere in the SDK/XMD domain.
Activating internal communication between SDK and XMD in the XMD console the following output can be observed:
-> xcontinue 32 0x0 -status_on_stop#
As SDK wont accept a negative start address it erroneously sets the PC to 0x0 which is the location of the critical interrupt vector (only included when the underlying system uses interrupts), and not the start of the .text section. Building and executing a default hello_world example works at first glance, as the .text section in this case starts at 0x0, but the boot sections are neither executed. Maybe someone can have another look in this issue. Its annoying.