[U-Boot-Users] [PATCH/review] Blackfin: use on-chip syscontrol() to reset

[Mike Frysinger]

On Wednesday 04 June 2008, Detlev Zundel wrote:
> > On Sunday 01 June 2008, Wolfgang Denk wrote:
> >> In message vapier at gentoo.org you wrote:
> >> > When available, use the on-chip syscontrol() rom function to perform a
> >> > system reset rather than our own local method.
> >> > -		asm("jump (%0);" : : "a" (bfin_reset));
> >> > +		if (0 /*BOOTROM_CAPS_SYSCONTROL -- current silicon is broken*/)
> >> > +			syscontrol(SYSCTRL_SOFTRESET, NULL, NULL);
> >> > +		else
> >> > +			asm("jump (%0);" : : "a" (bfin_reset));
> >> >  }
> >>
> >> Your comment suggests you change something, but actually you;re just
> >> adding dead code.
> >
> > it's dead code because the current silicon is broken.  i know the code
> > works as ive tested it in scenarios where the anomaly isnt triggered, it
> > just cant be enabled until fixed hardware is released.  it's much easier
> > to go back and fix 1 if statement then have to carry this patch somewhere
> > it can got lost and i need to waste time and rewrite it.
>
> Hm.  If the current - available - silicon is broken, then even if it is
> fixed in the future, you will have to support both versions, right?  So
> in the end you will need to have a discriminating factor on what to use.
>
> So if you introduced some variable or bit in a variable (a define
> doesn't look promising as I guess you want to support both broken and
> non-broken silicon with a single image), you could have correct code now
> and in the future - and no dead code complaints.

there is already a Blackfin anomaly framework in place for this.  we're just 
awaiting for the anomaly to be formalized.  that is what i meant when i 
said "just update the if line".

> PowerPC Linux btw. does quite a lot of such "cpu_feature" stuff to
> dynamically accomodate lots of different silicon.

we dont do dynamic stuff like powerpc
-mike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 827 bytes
Desc: This is a digitally signed message part.
Url : http://lists.denx.de/pipermail/u-boot/attachments/20080604/14d68c7f/attachment.pgp