Joakim Tjernlund wrote:
> On Tue, 2006-11-28 at 16:16 -0600, Timur Tabi wrote:
>> Joakim Tjernlund wrote:
>>> No, see attached patch(s)
>> Ah, I see.
>>> Not tested in your tree as I don't use that one (yet)
>> Git didn't like your patches, for some reason, so I had to apply them by hand,
>> but everything seems to be okay. I will apply them to our tree for Wolfgang's
> While I am at it, I would also like to see this in u-boot
> We use I2C as HRCW since we wan't to haw our flash reset connetced to
> HRESET, otherwise you might be unable to boot if the flash is in non
> read array mode when the board resets.
> We also need to have the version info in the begining of the flash so
> we can identify what version of u-boot we have installed.
> From 89b60f21af0d04959d93ccb70fd781c8aba9e66c Mon Sep 17 00:00:00 2001
> From: Joakim Tjernlund <Joakim.Tjernlund@...>
> Date: Tue, 28 Nov 2006 23:42:31 +0100
> Subject: [PATCH] Make HRCW and version info in data segment configurable.
> Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund@...>
> cpu/mpc83xx/start.S | 28 +++++++++++++++++-----------
> 1 files changed, 17 insertions(+), 11 deletions(-)
> diff --git a/cpu/mpc83xx/start.S b/cpu/mpc83xx/start.S
> index 0f27bb6..44bca26 100644
> --- a/cpu/mpc83xx/start.S
> +++ b/cpu/mpc83xx/start.S
> @@ -77,20 +77,12 @@
> - * Version string - must be in data segment because MPC83xx uses the
> - * first 256 bytes for the Hard Reset Configuration Word table (see
> - * below). Similarly, can't have the U-Boot Magic Number as the first
> - * thing in the image - don't know how this will affect the image tools,
> - * but I guess I'll find out soon.
> + * MPC83xx can use the first 0x40 bytes for the Hard Reset Configuration Word
> + * table (see below) if so configured.
> - .data
> - .globl version_string
> - .ascii U_BOOT_VERSION
> - .ascii " (", __DATE__, " - ", __TIME__, ")"
> - .ascii " ", CONFIG_IDENT_STRING, "\0"
> +#ifndef CFG_HRCW_IN_I2C_EEPROM
> #define _HRCW_TABLE_ENTRY(w) \
> .fill 8,1,(((w)>>24)&0xff); \
> .fill 8,1,(((w)>>16)&0xff); \
> @@ -99,7 +91,21 @@ version_string:
> + * Version string - May be in data segment if one wants to reserve the
> + * space left to address 0x100 for future expansion of HRCW bytes.
> + */
> +#ifdef CFG_VERSION_STRING_IN_DATA
> + .data
> + .long 0x27051956 /* U-Boot Magic Number */
> + .globl version_string
> + .ascii U_BOOT_VERSION
> + .ascii " (", __DATE__, " - ", __TIME__, ")"
> + .ascii " ", CONFIG_IDENT_STRING, "\0"
> #ifndef CONFIG_DEFAULT_IMMR
> #error CONFIG_DEFAULT_IMMR must be defined
I don't believe you want to do this: you are removing the HRCW entirely
from flash if you have it in I2C (CFG_HRCW_IN_I2C_EEPROM). Unless I'm
missing something (possible), I don't see why you even need a
configuration option CFG_HRCW_IN_I2C_EEPROM.
* You can chose to ignore the flash-HRCW and use the I2C one simply by
pulling a config pin up/down on power up (if I understand the manuals
correctly), so you can store your HRCW in both places and chose which
you use via a hardware configuration jumper. This is a Very Good
Thing[tm]. Suppressing the flash version simply because you chose not
to use it is not something _I_ want to do on my 8360.
* If you put your HRCW in I2C _and_ in flash, you will be able to boot
via the flash HRCW if your I2C version gets hosed. If you don't, you
will be hosed.
* Having a redundant HRCW in flash is essentially free: it costs you 64
bytes of otherwise unused memory (well, other than the version string
which you moved there, but there should be room for both).
My suggestion for putting the u-boot version string in a fixed place in
flash was to put it _after_ the HRCW (e.g. at offset 0x40). 0x60 (96)
bytes should be plenty for the version string. Instead, you've unwisely
(IMHO) _replaced_ the HRCW. Browsing some of the other PowerPC CPU
types, the ones that can appear to put the version string at the very
start of flash. This implies putting the version string in flash at
offset 0x40 is OK (I was concerned about relocation fixup issues with
pointers to flash vs. RAM after relocation).
On a related topic, the comment at the head of start.S says
* Version string - must be in data segment because MPC83xx uses the
* first 256 bytes for the Hard Reset Configuration Word table (see
* below). Similarly, can't have the U-Boot Magic Number as the first
* thing in the image - don't know how this will affect the image tools,
* but I guess I'll find out soon.
I see the comment was cut'n'pasted from the 8260 start.S. Actually only
the first 64 (0x40) bytes of memory are used for the HRCW, _not_ all 256
Looks like I (or somebody) needs to submit a clean up patch for that
comment for the 8260 and 8360. Sigh. I just _had_ to look. Curiosity
will be the death of me yet. ;-)