On Wed, Feb 16, 2011 at 10:13 AM, Sascha Hauer s.hauer@pengutronix.dewrote:
On Wed, Feb 16, 2011 at 09:37:47AM +0100, Yong Shen wrote:
Hi Sascha,
local_irq_disable();do_gettimeofday(&before);plat_lpc = __raw_readl(MXC_CORTEXA8_PLAT_LPC) &~(MXC_CORTEXA8_PLAT_LPC_DSM);One thing that strikes me here is the fact that this code can probably run on i.MX53 aswell, right? It's only that these registers have different addresses on i.MX53. The MXC_ prefix is therefore not a good idea. Switching this to MX51_ and having an additional MX53_ register leads to code duplication. This shows that it's a bad idea to code fixed addresses in the code. We should go for base + offset instead so that this code will have a better start on i.MX53. This of course needs changes in the current crm_regs.h and probably in the i.MX51/53 clock code.
Yes, for mx53, it is similar. But for the case you are talking about, is it easier that we keep MXC_ prefix in this file and define MXC_ to MX51 or MX53 in crm_regs.h
according
to which board is running?
I don't understand. How can we 'define' (which is compile time) to something depending on the board (which is runtime)?
I ignored the goal is one image running on multiple SOCs.
In addition, registers for this code are not in one section, which means many BASEx + offset there, if I understand right. Do you have a sample
for
'base + offset' case? since mx53 just came in, I am not sure about such case.
Forget it. I just realized that more or less by accident the virtual addresses for the i.MX51 and i.MX53 are the same.
So, the conclusion is: still using MXC_ prefix in this period. right? Correct me.
Yong
Sascha
-- Pengutronix e.K. | | Industrial Linux Solutions | http://www.pengutronix.de/ | Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 | Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |