[U-Boot] [PATCH V7 1/3] Initial support for Marvell Orion5x SoC

[Albert ARIBAUD]

Hi Prafulla,

Le 10/06/2010 21:36, Prafulla Wadaskar a écrit :

 >> +/* Display device and revision IDs.
 >> + * This function must cover all known device/revision
 >> + * combinations, not only the one for which u-boot is
 >> + * compiled; this way, one can identify actual HW in
 >> + * case of a mismatch.
 >> + */

 >> +     if (dev == MV88F5181_DEV_ID) {
 >
 > Pls see comments at the end

 >> +     } else if (dev == MV88F6183_DEV_ID) {
 >
 > Pls see comments at the end

 >> +/*
 >> + * The following definitions are intended for identifying
 >> + * the real device and revision on which u-boot is running
 >> + * even if it was compiled only for a specific one. Thus,
 >> + * these constants must not be considered chip-specific.
 >> + */
 >> +
 >> +/* Orion-1 (88F5181) and Orion-VoIP (88F5181L) */
 >> +#define MV88F5181_DEV_ID        0x5181
 >> +#define MV88F5181_REV_B1        3
 >> +#define MV88F5181L_REV_A0       8
 >> +#define MV88F5181L_REV_A1       9
 >> +/* Orion-NAS (88F5182) */
 >> +#define MV88F5182_DEV_ID        0x5182
 >> +#define MV88F5182_REV_A2        2
 >> +/* Orion-2 (88F5281) */
 >> +#define MV88F5281_DEV_ID        0x5281
 >> +#define MV88F5281_REV_D0        4
 >> +#define MV88F5281_REV_D1        5
 >> +#define MV88F5281_REV_D2        6
 >> +/* Orion-1-90 (88F6183) */
 >> +#define MV88F6183_DEV_ID        0x6183
 >> +#define MV88F6183_REV_B0        3
 >> +
 >
 > Pls refer comments at the end.

 > Small request-
 > As per this definition only 5182 is supported and tested,
 > It would be more logical to remove 5181 and 6183 specific code,
 > We can always add it as it is required, at this moment it looks like 
a dead code.

This was raised and discussed before:

<http://article.gmane.org/gmane.comp.boot-loaders.u-boot/73579/match=orion5x+device+revision>

For any other purpose than identification of the actual SoC, I would 
have agreed about removing anything not related to 5182. But here, 
precisely, the code is trying to print the identity of the *actual* SoC 
it is being run on, as opposed to that of the *assumed* SoC it was 
compiled for.

Thus here the code must define and test for as many device and revision 
IDs as possible so as to give the most accurate information possible.

This is epxlicitely stated in the comments before the identification 
function and the device and revision defines.

 > also moving 5281 specific code to mv88f5182.h will make this patch 
clean and complete.

If you're talking about the device / variant IDs, I don't think they 
should move for the reason above; if that's something else, can you be 
more precise?

 > Rest everything seems to be okay

Ok. If we can agree about the identification function, then I'll post a 
final V8 patch with GPIO/MPP enabled.

 > Regards..
 > Prafulla . .

Cheers!

Amicalement,
-- 
Albert.