Re: H8/300 target support patch

[Yoshinori Sato <ysato@users.sourceforge.jp>]

At Sun, 15 Feb 2004 15:40:56 +0100,
Bartlomiej Zolnierkiewicz wrote:
> 
> On Sunday 15 of February 2004 07:05, Yoshinori Sato wrote:
> > Hello.
> > I develop H8/300 support of linux kernel.
> >
> > I made a patch to use IDE driver with a H8/300 target.
> > I want to do merge to source if possible.
> >
> > How will about it?
> 
> Hi,
> 
> + *  reference schematic is http://www.linet.gr.jp/mituiwa/h8/h8osv3/hddprog/hdd.png
> 
> this link is dead
>
> +#include <linux/mm.h>
> +#include <linux/interrupt.h>
> +#include <linux/blkdev.h>
> 
> why do you include mm.h and blkdev.h?

It is code in a thing of 2.4.x
I seem to have left that deleted it when shifted to new driver.
 
> +static void _outsw(unsigned long addr, void *buf, u32 len)
> +{
> +	unsigned volatile short *ap = (unsigned volatile short *)addr;
> +	unsigned short *bp = (unsigned short *)buf;
> +	unsigned short d;
> +	while(len--) {
> +		d = *bp++;
> +		*ap = (d << 8) | (d >> 8);
> +	}
> +}
> +
> +static void _insw(unsigned long addr, void *buf, u32 len)
> +{
> +	unsigned volatile short *ap = (unsigned volatile short *)addr;
> +	unsigned short *bp = (unsigned short *)buf;
> +	unsigned short d;
> +	while(len--) {
> +		d = *ap;
> +		*bp++ = (d << 8) | (d >> 8);
> +	}
> +}
> 
> why you don't use outsw() and insw() from include/asm-h8300/io.h?

Because bus turns over, insw/outsw is not usable.

> +	ide_setup_ports(&hw, IDE_BASE, (int *)offsets,
> +                             IDE_CTRL, 0, NULL,IDE_IRQ);
> 
> ide_setup_ports() is deprecated, please do not use it.
> 
> +		hwif->mmio  = 0;
> 
> Your driver is doing MMIO, it should request and release resources itself
> and set hwif->mmio to 2.
> 
> +void __init h8300_ide_print_resource(char *name, hw_regs_t *hw)
> +{
> +	printk("%s at 0x%08x-0x%08x,0x%08x on irq %d", name,
> +		(unsigned int)hw->io_ports[IDE_DATA_OFFSET],
> +		(unsigned int)hw->io_ports[IDE_DATA_OFFSET]+15,
> +		(unsigned int)hw->io_ports[IDE_CONTROL_OFFSET],
> +		hw->irq);
> +}
> 
> Is this necessary (this is a reason for extra #ifdefs in ide-probe.c)?
> Can't we do what m68k is doing (print only address and IRQ) for now?

I can use it.
But there is the problem that resource to use cannot output justly.
I think that solved if　can define a part of +7 with #define.
 
> Both drivers use same IO ops so they can be moved to some include file
> or both drivers can be merged (preferred solution) - they are for different
> boards and can't work together (plus MAX_HWIFS is defined to 1 for h8300).
>
> Here is patch (not tested) addressing some of these issues...

I identified operation.
MMIO and print_resource examine a correction method.

Thank you.

-- 
Yoshinori Sato
<ysato@users.sourceforge.jp>