From mboxrd@z Thu Jan  1 00:00:00 1970
From: Stefano Babic <sbabic@denx.de>
Date: Tue, 16 Nov 2010 10:34:44 +0100
Subject: [U-Boot] [PATCH v2 1/1] imx: Get fec mac address from fuse
In-Reply-To: <AANLkTi=NZurD4NkjCC6+v0A5rKmXoG6tpFZQqhkyFmFH@mail.gmail.com>
References: <1289705187-11350-1-git-send-email-r64343@freescale.com>	<4CE0155D.8090800@denx.de>	<AANLkTimXrZANNdroChBFpOhu13VUWXsBdOdBUUn5fBYj@mail.gmail.com>	<20101116065304.9D24214EA7E@gemini.denx.de>
	<AANLkTi=NZurD4NkjCC6+v0A5rKmXoG6tpFZQqhkyFmFH@mail.gmail.com>
Message-ID: <4CE25034.4090403@denx.de>
List-Id: <u-boot.lists.denx.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: u-boot@lists.denx.de

On 11/16/2010 08:16 AM, Jason Liu wrote:
>>> struct iim_regs {
>>>         u32 stat;
>>>         u32 statm;
>>>         u32 err;
>>>         u32 emask;
>>>         u32 fctl;
>>>         u32 ua;
>>>         u32 la;
>>>         u32 sdat;
>>>         u32 prev;
>>>         u32 srev;
>>>         u32 preg_p;
>>>         u32 scs0;
>>>         u32 scs1;
>>>         u32 scs2;
>>>         u32 scs3;
>>>         u32 res0[0x1f1];
>>>         u32 iim_bank_area0_1[9];
>>>         u32 mac_addr[6];
>>>         u32 iim_bank_area0_2[0x11];
>>>         u32 res1[0xe0];
>>>         u32 iim_bank_area1[0x20];
>>>         u32 res2[0xe0];
>>>         u32 iim_bank_area2[0x20];
>>>         u32 res3[0xe0];
>>>         u32 iim_bank_area3[0x20];
>>>         u32 res4[0xe0];
>>> };
>>
>> This struct is supposed to describe some register layout defined by
>> the hardware - but does the hardware know anything about teh special
>> meaning "MAC address" of this specific field, or is this an
>> interpretation that we just define as we like?

The MAC address is stored at a specific address, that is different in
each i.MX implementation. There are a lot of fields the hardware is
aware of.

> 
> This is defined according to the IC spec and it should be hw layout.
> We can read such kind of fuse just like read register.  But currently, uboot
> require that all the register array should be included into one
> structure and not
> use the *(offset + base_addr) method, this is also stefano suggestion. But for
> fuse support, there are many fuses but we only use FEC_MAC fuse
> currently, this is why
> I ask how to handle it better in uboot.
> 
> Stefano, what's your comments?

I would like to propose a structure to better clarify the internal
layout. Let me explain and check if this can be suitable for you. We
could use a union to define the different layouts for the fuse banks,
such as (the example is for the i.MX51):

typedef union fuse_bank {
	struct {
		u32 fuse_lock;
		u32 osc_jtag;
		u32 reserved0;
		u32 bt
		.....
		u32 mac_addr[6];
		....
		u32 reserved_filled[..]; /* to fill the 0x80-0xFF*/
	};
	/* Now the layout for the other fuse banks */
	struct {
		....
	};
	/*
	 * If we do not want to set now the layout, we can distinguish
	 * only between real register and reserved addresses
	 * as you already did
	 */
	 struct {
		u32 fuse_regs[0x20];
		u32 reserved[0xe0];
	}
}

And in your structure you can have something like this:

        ....
        u32 scs3;
        u32 res0[0x1f1];
	fuse_bank fuses[4];
}

You do not need to set the layout for the fuse banks 1-3, if you do not
want, but the structure is prepared if someone else will add functions
to manage the fuses.

What do you think about it ?

Best regards,
Stefano Babic

-- 
=====================================================================
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: +49-8142-66989-0 Fax: +49-8142-66989-80  Email: office at denx.de
=====================================================================