[ECOS] Inconsistent sizing of RAM for ARM "e7t"

[ECOS] Inconsistent sizing of RAM for ARM "e7t"

[Bill Siegmund]

The brochure for the ARM Evaluator 7T
claims it has 256KB of RAM.

In the new materials for the ARM Evaluator 7T
that appeared in the CVS version of eCos on Friday,
mlt_arm_e7t_ram.h concurs, using 0x80000.
But the corresponding *.ldi and *.mlt files use 0x28000!

mlt_arm_e7t_ram.h:
#define CYGMEM_REGION_ram_SIZE (0x80000)
.
.
#define CYGMEM_SECTION_heap1_SIZE (0x80000 - (size_t) CYG_LABEL_NAME
(__heap1))


mlt_arm_e7t_ram.ldi:
MEMORY
{
    ram : ORIGIN = 0, LENGTH = 0x28000
}

mlt_arm_e7t_ram.mlt:
region ram 0 28000 0 !

Running diff and chksum on these two aberrant files
and the corresponding files for the old Rev B AEB
-- mlt_arm_aeb.ldi and mlt_arm_aeb.mlt --
confirms these "e7t" files are exact copies of the "aeb" files.

Why 0x28000 is used in the "aeb" files
when the Rev B AEB has/had only 128KB of RAM
I do not know.

Bill Siegmund
Cal-Tex Computers, Inc.
ctc-dsl@pacbell.net

Re: [ECOS] Inconsistent sizing of RAM for ARM "e7t"

[Jesper Skov]

>>>>> "Bill" == Bill Siegmund <ctc-dsl@pacbell.net> writes:

Bill> The brochure for the ARM Evaluator 7T claims it has 256KB of
Bill> RAM.

It has 512kB RAM and 512kB flash. I've updated the memory maps
accordingly. 

Bill> Why 0x28000 is used in the "aeb" files when the Rev B AEB
Bill> has/had only 128KB of RAM I do not know.

Because it doesn't map RAM to 0 but to 0x8000. I don't remember the
details, but I think it involved dragons, wizards and magic spells.

Thanks,
Jesper