[LinuxBIOS] [PATCH] Reimplementation/fixing of CS5530/CS5530A southbridge code

[Jonathan Sturges]

Thanks for the updated 5530 code!
However, I'm having some difficulty enabling the IDE controllers when 
using this patch.  As far as I can tell, configuration parsing doesn't 
appear to work for new ideX_enable flags.

The 'register "ide0_enable" = "1"' configuration option seemed to have 
no effect on my 5530 system.  Even though both IDE controllers were 
enabled in Config.lb, when I'd test the resulting image, IDE0 would be 
disabled and IDE1 enabled.  In the end, to make it work, I changed:
"if (conf->ide0_enable) {"
to:
"if ( 1 ) {"
...in cs5530_ide.c to get IDE0 enabled.  This forces it to work.

Any ideas what would cause this?  By the way, I'm using the Eaglelion 
5BCM target for the moment, as it's nearly identical to my system.

thanks,
Jonathan


Uwe Hermann wrote:
> See patch.
> 
> This is also required to make the JUKI-511P code work AFAICT (won't
> build otherwise).
> 
> 
> Uwe.
> 
> 
> ------------------------------------------------------------------------
> 
> This is a full rewrite of all the CS5530/CS5530A code. The previous code was
> mostly undocumented, had a broken coding style, contained lots of dead 
> code and had several other problems, e.g. it enabled write access to the 
> ROM (why?), it unconditionally enabled primary/secondary IDE (which should
> have a config option) and that even _twice_ (which is um... wrong).
> 
> The new code
> 
>  - has 'ide0_enable' and 'ide1_enable' config options (which actually
>    work) to enable/disable the primary/secondary IDE interface in
>    Config.lb.
> 
>  - Does _not_ enable write access to the ROM (or is there some good
>    reason to do that? If so, it should at least have a config option).
> 
>  - Contains a bit more documentation.
> 
>  - Uses readable (and documented) #defines instead of hardcoded magic values.
> 
>  - aaand... it actually compiles ;-) Yep, that's right. The previous code
>    wouldn't even build, as it hadn't been fully ported from v1 (still used
>    v1 functions which are simply not available in v2).
>