Hi,
I face some issues with debugging a simple LED application for my STM32 stamp module (steitec.net/product_info.php?l…12-Stamp&products_id=143&).
My setup: I use the CodeSourcery toolchain, along with Eclipse Helios (CDT) and MingW in a Windows 7 (64 bit) environment.
My J-Link is pretty new: Firmware: J-Link ARM V8 compiled Nov 19 2010 11:55:23 and the Segger toolchain has version 4.20p. The init file - based on page 21 of the Jlink GDB Server documentation - is:

When starting the gdb server, the log says:

J-Link found 2 JTAG devices, Total IRLen = 9
JTAG ID: 0x3BA00477 (Cortex-M3)

When starting the debugging in Eclipse, the following log output is generated:

Connected to 127.0.0.1
Reading all registers
Read 4 bytes @ address 0x00000000 (Data = 0x20004FFF)
Select auto JTAG speed (1000 kHz)
Select flash device: STM32F103RB
Flash breakpoints enabled
Flash download enabled
Downloading 484 bytes @ address 0x08000000
Downloading 5736 bytes @ address 0x080001E4
Downloading 40 bytes @ address 0x0800184C
Writing register (PC = 0xE4010008)
ERROR: No matching RAMCode found (11713201)
Failed to download RAMCode.
Please check your flash settings!
Connection to debugger closed !

Obviously, there are some issues with the download.
Furthermore, the GDB-Debugger throws an error window stating "Can not read register 27 (CONTROL) while CPU is running".
I would like to add that the download of the related main.hex file via serial interface works without any problems.
Attached is the zip archive with main.elf and the linker script.

Can anybody help me?

Cheers
Christoph

Hi again,
it seems that I found the solution: in another thread (check it out, if you understand german: mikrocontroller.net/topic/170920)
I found the hint to add:

in my gdb init file. Now the download is processed without issues. My understanding is that now the debugger just avoids accessing the RAM during download..

Unfortunately, I had to realize that the download did not happen.My impression is:
The connection has been established, data has been sent to target and a verification took place. But the data has not been written.As I had flashed the target via STM Flashloader and serial port beforehand, the verification read the already flashed data.

When I changed the sources and compiled again, the GDB-Server showed:

Connected to 127.0.0.1
Reading all registers
Read 4 bytes @ address 0x00000000 (Data = 0x20004FFF)
Select auto JTAG speed (1000 kHz)
Target endianess set to "little endian"
Select flash device: STM32F103RB
Flash breakpoints enabled
Flash download enabled
Reading all registers
Downloading 484 bytes @ address 0x08000000
Downloading 5736 bytes @ address 0x080001E4
Downloading 40 bytes @ address 0x0800184C
Writing register (PC = 0x080001E4)
ERROR: Verification failed @ address 0x08000000
Connection to debugger closed !

What could be the reason? Is it possible that the cabling is not correct? As my stamp has no JTAG connection, I had to solder and wire a solution on my breadboard.
BTW, the error also happens if the JTAG speed remains at 5 kHz.
Cheers
christoph

I rechecked the JTAG-cabling and tested the consequences when removing single cables. Everytime, the GDB-Server showed errors to identify the target. Therefore, I assume that the cabling is not the root cause.

Success!
The Flash could be written when using the right device type (so simple... ). My stamp module has a STM32F103RET6 with 512K Flash. The board description @ steitec.net (see first post) mentioned a STM32F103RBT6- even though this is definitely wrong, as B indicates a 128K Flash. After editing this letter, it worked!
So final setup with arm-none-eabi-gdb.exe is as follows: