[linuxsh-dev] HP600 patch

[NIIBE Y. <gn...@ch...>]

YAEGASHI Takeshi wrote:
 > Ok, here is the patch against linux-sh-2000-06-05.diff.gz from
 > m17n.org, please review it.  I'll appreciate all your coments
 > and suggestions.

Great.  Please check it in.  Except:

 > +++ linux/arch/sh/Makefile	Tue Jun  6 17:38:52 2000

This one.  I support RMS for the name of GNU/Linux. :-)


 > diff -ruN linux-2.4.0-test1.orig/arch/sh/kernel/irq.c linux/arch/sh/kernel/irq.c
 > --- linux-2.4.0-test1.orig/arch/sh/kernel/irq.c	Mon May 22 12:55:06 2000
 > +++ linux/arch/sh/kernel/irq.c	Tue Jun  6 17:36:10 2000
 > @@ -37,6 +37,9 @@
 >  #include <asm/irq.h>
 >  #include <linux/irq.h>
 >  
 > +#ifdef CONFIG_HD64461
 > +#include <asm/hd64461.h>
 > +#endif
 >  
 >  unsigned int local_bh_count[NR_CPUS];
 >  unsigned int local_irq_count[NR_CPUS];
 > @@ -243,6 +246,18 @@
 >  		     "shlr	%0\n\t"
 >  		     "add	#-16, %0\n\t"
 >  		     :"=z" (irq));
 > +#if defined(CONFIG_HD64461)
 > +	if (irq == CONFIG_HD64461_IRQ) {
 > +		unsigned short bit=1;
 > +		unsigned short nirr = hd64461_inw(HD64461_NIRR);
 > +		unsigned short nimr = hd64461_inw(HD64461_NIMR);
 > +		nirr &= ~nimr;
 > +		for (bit = 1, irq = 0; irq < 16; bit <<= 1, irq++)
 > +			if (nirr & bit) break;
 > +		if (irq == 16) irq = CONFIG_HD64461_IRQ;
 > +		else irq += HD64461_IRQBASE;
 > +	}
 > +#endif
 >  
 >  	kstat.irqs[cpu][irq]++;
 >  	desc = irq_desc + irq;

OK.  You virtualize the HP64461_IRQ.  I think this is good approach.


 > diff -ruN linux-2.4.0-test1.orig/drivers/char/sh-sci.h linux/drivers/char/sh-sci.h
 > --- linux-2.4.0-test1.orig/drivers/char/sh-sci.h	Tue Jun  6 18:35:53 2000
 > +++ linux/drivers/char/sh-sci.h	Tue Jun  6 17:42:41 2000
 > @@ -202,7 +202,11 @@
 >   */
 >  #if defined(__sh3__)
 >  #if defined(CONFIG_CPU_SUBTYPE_SH7709)
 > +#ifdef CONFIG_SH_HP600
 > +#define PCLK           22222222
 > +#else
 >  #define PCLK           33333333
 > +#endif
 >  #else
 >  #define PCLK           14745600	/* Isn't it 15MHz? */
 >  #endif

Could you please resolve this issue with Stuart.  It seems that
his changes are not yet checked it in.

Besides, I'm not sure hd64461_{inw,outw} is worth or not.  We may enhance
the generic I/O routine to support more than two offsets.

And please make sure that let Mihai Spatar know his code is now included
(for hitfb.c).

For now, please check them in and go ahead.
--