[PATCH] move PHY fine tune to devicetree.cb

[Rudolf Marek]

Hello,

Following patch moves the PHY fine tune settings into devicetree.cb. Suited for 
AMD SB7xx. Please change the default settings for Gigabyte and Jetway.

Signed-off-by: Rudolf Marek

Untested I can test on monday evening.

Thanks,
Rudolf

[Stefan Reinauer]

On 8/22/10 10:37 PM, Rudolf Marek wrote:
> Index: src/mainboard/amd/mahogany/devicetree.cb
>
> --- src/mainboard/amd/mahogany/devicetree.cb.orig	2010-08-22 11:54:20.000000000 +0200
> +++ src/mainboard/amd/mahogany/devicetree.cb	2010-08-22 12:12:45.000000000 +0200
> @@ -110,6 +110,19 @@
>    				register "sata0_enable" = "1"
>    				register "boot_switch_sata_ide" = "0"	# 0: boot from SATA. 1: IDE
>    				register "hda_viddid" = "0x10ec0882"
> 			# SATA PHY fine tune, depends on PCB layout!
> 			register "sphy15_p0" = "0x01b48017"
> 			register "sphy15_p1" = "0x01b48019"
> 			register "sphy15_p2" = "0x01b48016"
> 			register "sphy15_p3" = "0x01b48016"
> 			register "sphy15_p4" = "0x01b48016"
> 			register "sphy15_p5" = "0x01b48016"
> 			register "sphy30_p0" = "0xa09a"
> 			register "sphy30_p1" = "0xa09f"
> 			register "sphy30_p2" = "0xa07a"
> 			register "sphy30_p3" = "0xa07a"
> 			register "sphy30_p4" = "0xa07a"
> 			register "sphy30_p5" = "0xa07a"
> 	end	#southbridge/amd/sb700
> end #  device pci 18.0

Shouldn't such values rather go into the mainboard specific code (ie an
array in mainboard,c) instead of the device tree?

[Patrick Georgi]

Am 22.08.2010 23:38, schrieb Stefan Reinauer:
> Shouldn't such values rather go into the mainboard specific code (ie an
> array in mainboard,c) instead of the device tree?

Depends on what we define the purpose of the registers in the devicetree
to be.
Some of the things we currently stuff in registers might be better put
into CMOS ("ide or sata" decisions, for example), so that's a more
general question than just for these values.

Patrick

[Carl-Daniel Hailfinger]

On 22.08.2010 23:41, Patrick Georgi wrote:
> Am 22.08.2010 23:38, schrieb Stefan Reinauer:
> > Shouldn't such values rather go into the mainboard specific code (ie an
> > array in mainboard,c) instead of the device tree?
>
> Depends on what we define the purpose of the registers in the devicetree
> to be.
> Some of the things we currently stuff in registers might be better put
> into CMOS ("ide or sata" decisions, for example), so that's a more
> general question than just for these values.

If we don't store mainboard specific settings for generic (chipset,
superio) hardware in the device tree, what settings should be in the
device tree? None?

Regards,
Carl-Daniel

[Stefan Reinauer]

On 8/23/10 12:24 AM, Carl-Daniel Hailfinger wrote:
> On 22.08.2010 23:41, Patrick Georgi wrote:
> > Am 22.08.2010 23:38, schrieb Stefan Reinauer:
> > > Shouldn't such values rather go into the mainboard specific code (ie an
> > > array in mainboard,c) instead of the device tree?
> >
> > Depends on what we define the purpose of the registers in the devicetree
> > to be.
> > Some of the things we currently stuff in registers might be better put
> > into CMOS ("ide or sata" decisions, for example), so that's a more
> > general question than just for these values.
>
> If we don't store mainboard specific settings for generic (chipset,
> superio) hardware in the device tree, what settings should be in the
> device tree? None?

higher level information, not just register dumps.