This is the mail archive of the ecos-discuss@sources.redhat.com mailing list for the eCos project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

RE: ROM and hardware vectors for ARM.



	-----Original Message-----
	From:	Gary Thomas [SMTP:gthomas@ecoscentric.com]
	Sent:	17. oktober 2002 16:19
	To:	Erik Aagaard Knudsen
	Cc:	'ecos-discuss@sources.redhat.com'
	Subject:	Re: [ECOS] ROM and hardware vectors for ARM.

	On Thu, 2002-10-17 at 08:12, Erik Aagaard Knudsen wrote:
	>> It is our inpression that in order to be able to debug an
application on an
	>> ARM-target, the ROM-redboot must 
	>> 
	>> 1. Boot from adr 0x0 and initialize the CPU/Memory
	>> 2. Remap the system so that the boot ROM is removed from addr.
0x0 and
	>> replaced by RAM.
	>> 3. Interrupt vectors is then initialized in this new RAM at addr.
0x0
	>> 
	>> What can we do if we do not have the possibility of remapping
(swapping
	>> ROM/RAM addresses)?

	>Why can't you?  What's your processor (core)?

	The CPU we are using is the S3C44B0X from samsung. (AMR7TDMI) It
does not have this functionality as a feature. It has internal RAM located
from address 0x10000000 so this is very far from the vectors. We could make
changes to the hardware, so that some external pins could switch CS0,CS1
pins using some external logic. Is this the way a normal eCos hardware
platform must be designed ? 

	>> 
	>> The secondary vector table is located in the vectors section (
0x20 up). Can

	>This would take some work.  The vectors assume that the secondary
table 
	>can be reached with a PC-relative instruction, which limits how far
away
	>the table can be - probably out of reach if it's in a different
section
	>of memory.

	>> this table without too much fuss be moved to RAM instead (f.ex.
into the
	>> fixed vectors section)? This would ofcourse mean that we have to
initialize
	>> these RAM locations with the correct pointers, but it would mean
that these
	>> vector pointers could be located anywhere in the memory space,
hense
	>> removing the need for hardware remapping logic.
	>> 
	>> Are there other problems, which we have not identified?
	>> 

	>Yes, plenty of them.  Grant Edwards did this a long time ago,
	>he might be able to give some insight.

	-- 
	------------------------------------------------------------
	Gary Thomas                  |
	eCosCentric, Ltd.            |  
	+1 (970) 229-1963            |  eCos & RedBoot experts
	gthomas@ecoscentric.com      |
	http://www.ecoscentric.com/  |
	------------------------------------------------------------

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


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]