GCC/Eclipse integration generating non-working executable?

RE: GCC/Eclipse integration generating non-working executable?

Pierre Landau — Thu, 03 Mar 2016 19:01:13 GMT

OK, after messing with stack and heap sizes, I was able to get the thing to start, and even the debugger appears to be running. Thanks for your help!

RE: GCC/Eclipse integration generating non-working executable?

Pierre Landau — Thu, 03 Mar 2016 18:39:31 GMT

Thanks! this helps explain what's going on. OTOH, the ble_app_gls demo app overflows RAM when built in GCC (apparently not so in Keil) so I'm going to have to do some tap dancing.

RE: GCC/Eclipse integration generating non-working executable?

Vidar Berg — Wed, 02 Mar 2016 11:47:46 GMT

Are you using a 16K RAM variant of the nRF51? In that case you'd have to adjust the linker settings since the default is configuration is for the 32K variant. Difference between Keil and GCC in this case is that the GCC links the call stack to top of RAM, while Keil places it just above the globals. In other words, it may still work with Keil even with wrong linker settings.

E.g., 32K ram configuration in ble_app_gls_gcc_nrf51.ld from SDK 11 :

MEMORY
{
  FLASH (rx) : ORIGIN = 0x1b000, LENGTH = 0x25000
  RAM (rwx) :  ORIGIN = 0x20001f00, LENGTH = 0x6100
}

and same for 16K:

MEMORY
{
  FLASH (rx) : ORIGIN = 0x1b000, LENGTH = 0x25000
  RAM (rwx) :  ORIGIN = 0x20001f00, LENGTH = 0x2100  
}

RE: GCC/Eclipse integration generating non-working executable?

Nguyen Hoan Hoang — Tue, 01 Mar 2016 21:00:43 GMT

Try this blog site for Eclipse + GCC It has examples source code with full Eclipse project file. Debugging and flashing info with OpenOCD. Parallel programming of nRF5x series with IDAPnRFProg without requiring to merge hex.