[U-Boot-Users] [PATCH] Add support for Amstrad Delta (E3).

[Jonathan McDowell]

On Fri, Sep 01, 2006 at 01:34:52PM +0200, Ladislav Michl wrote:
> On Tue, Jun 13, 2006 at 05:59:27PM +0100, Jonathan McDowell wrote:
> > diff --git a/Makefile b/Makefile
> > index 440ab2c..659e9b1 100644
> > --- a/Makefile
> > +++ b/Makefile
> > @@ -1565,6 +1565,9 @@ netstar_config:		unconfig
> >  omap1510inn_config :	unconfig
> >  	@./mkconfig $(@:_config=) arm arm925t omap1510inn
> >  
> > +amsdelta_config :	unconfig
> > +	@./mkconfig $(@:_config=) arm arm925t amsdelta
> > +
> 
> It would be nice to keep machines sorted alphabeticaly...

Sure, that seems reasonable.
 
> >  omap5912osk_config :	unconfig
> >  	@./mkconfig $(@:_config=) arm arm926ejs omap5912osk NULL omap
> >  
> 
> [snip]
> 
> > diff --git a/cpu/arm925t/start.S b/cpu/arm925t/start.S
> > index acd7742..c2c2adb 100644
> > --- a/cpu/arm925t/start.S
> > +++ b/cpu/arm925t/start.S
> > @@ -125,6 +125,7 @@ reset:
> >  	orr	r0,r0,#0xd3
> >  	msr	cpsr,r0
> >  
> > +#ifndef CONFIG_AMSDELTA
> >  	/*
> >  	 * Set up 925T mode
> >  	 */
> > @@ -159,6 +160,7 @@ poll1:
> >  	ldrh r1, [r0]
> >  	ands r1, r1, #0x01
> >  	beq poll1
> > +#endif
> 
> Why it this needed?

Without it the boot process is incredibly slow. This section was
isolated as different between the 925 and 926 init with no obvious extra
benefit and once removed the boot process speeds up significantly
(loading the kernel from NAND is about 10 times faster, for example,
though it can be seen even just in terms of interactive u-boot use).

I don't have any other 925 platform available to determine if this can
be left out for them all, hence only disabling it for the Delta.
 
> >  	/*
> >  	 * we do sys-critical inits only at reboot,
> 
> Jonathan, you will need a bit more patience. It tooks several months to
> get patches accepted :-)

I can be patient about acceptance as long as there's some indication
they've actually been seen and will get there eventually - I may have
another board I want to port u-boot to, but there's no point if it
proves impossible to push changes upstream.

J.

-- 
/-\                             |    "Could I have an 'E', please,
|@/  Debian GNU/Linux Developer |        Bob?"  (Blockbusters)
\-                              |