Re: [ECOS] Re: STM32 ROM .bin does not start

[Bernhard Gebert <bernhard_gebert at web dot de>]

Hello Sergei,

at first: I got some progress!

> > Hm. Your quoted .ldi, .h mlt files look good to use the internal CPU's
> > resources only, but, what says arm-eabi-objdump for the final eCos ELF
> > build? Are there 0x0800_0000 and 0x2000_0000 origins only in the ELF?

As I have never used arm-eabi-objdump before, I can not answer at the moment...

> > Where are placed .data, .bss, startup stack segments, etc. You can
> > compare the dumps of the sections for working ELFs and for the got
> > ones with your tweaks those mlt files. 

Good idea, I'll do so tomorrow.

> > Of course a working JTAG engine
> > would be so helpful for digging until we can not drive a LED.

If the responsible HW engineer would not have been too stingy woth the pins, I could do so. It was me :-(


Using the USART1 is not possible, but I can flash a LED while using some 
thread-synchronisation.

What I did:
Looking at the latest _rom-ldi from CVS (currently I am still using eCos 
3.0), there is a difference I don't understand:
all the ROM areas are marked as "flash".

I simply switched "ram" to the upper half of "sram", dito with "rom" and 
"flash".

_rom.ldi
*******************************
MEMORY
{
   sram  : ORIGIN = 0x20000000, LENGTH = 0x00007D00
   flash : ORIGIN = 0x08000000, LENGTH = 0x00040000
   rom   : ORIGIN = 0x08040000, LENGTH = 0x00040000
   ram   : ORIGIN = 0x20007D00, LENGTH = 
0x00007D00-CYGNUM_HAL_COMMON_INTERRUPTS_STACK_SIZE
}
SECTIONS
{
   SECTIONS_BEGIN
   SECTION_rom_vectors (flash, 0x08000000, LMA_EQ_VMA)
   SECTION_RELOCS (flash, ALIGN (0x8), LMA_EQ_VMA)
   SECTION_text (flash, ALIGN (0x8), LMA_EQ_VMA)
   SECTION_fini (flash, ALIGN (0x8), LMA_EQ_VMA)
   SECTION_rodata (flash, ALIGN (0x8), LMA_EQ_VMA)
   SECTION_rodata1 (flash, ALIGN (0x8), LMA_EQ_VMA)
   SECTION_fixup (flash, ALIGN (0x8), LMA_EQ_VMA)
   SECTION_gcc_except_table (flash, ALIGN (0x8), LMA_EQ_VMA)
   SECTION_eh_frame (flash, ALIGN (0x8), LMA_EQ_VMA)
   SECTION_got (flash, ALIGN (0x8), LMA_EQ_VMA)
   SECTION_sram (sram, 0x20000400, FOLLOWING (.got))
   SECTION_data (ram, 0x20007D00, FOLLOWING (.sram))
   SECTION_bss (ram, ALIGN (0x8), LMA_EQ_VMA)
   CYG_LABEL_DEFN(__heap1) = ALIGN (0x8);
   SECTIONS_END
}

hal_vsr_table = 0x20000000;
hal_virtual_vector_table = hal_vsr_table + 128*4;
hal_startup_stack = 0x20000000 + 1024*64;


_rom.h
*******************************
#define CYGMEM_REGION_sram (0x20000000)
#define CYGMEM_REGION_sram_SIZE (0x00007D00)
#define CYGMEM_REGION_sram_ATTR (CYGMEM_REGION_ATTR_R | 
CYGMEM_REGION_ATTR_W)

#define CYGMEM_REGION_flash (0x08000000)
#define CYGMEM_REGION_flash_SIZE (0x00040000)
#define CYGMEM_REGION_flash_ATTR (CYGMEM_REGION_ATTR_R | 
CYGMEM_REGION_ATTR_W)

#define CYGMEM_REGION_ram (0x20007D00)
#define CYGMEM_REGION_ram_SIZE 
(0x00007D00-CYGNUM_HAL_COMMON_INTERRUPTS_STACK_SIZE)
#define CYGMEM_REGION_ram_ATTR (CYGMEM_REGION_ATTR_R | CYGMEM_REGION_ATTR_W)

#define CYGMEM_REGION_rom (0x08040000)
#define CYGMEM_REGION_rom_SIZE (0x00040000)
#define CYGMEM_REGION_rom_ATTR (CYGMEM_REGION_ATTR_R)


Now I'll try to set up a new .ldi for flash only without those work-arounds.

Nevertheless:
Can anyone explain me what the changes since 3.0 are meant for? Why is 
everything lying in the internal flash now?


--
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss