On 03/17/2016 01:48 PM, Daniel Schwierzeck wrote:
2016-03-17 13:20 GMT+01:00 Marek Vasut marex@denx.de:
On 03/17/2016 05:02 AM, Wills Wang wrote:
On Thursday, March 17, 2016 11:44 AM, Marek Vasut wrote:
On 03/17/2016 04:39 AM, Wills Wang wrote:
On Thursday, March 17, 2016 05:35 AM, Marek Vasut wrote:
On 03/16/2016 09:59 AM, Wills Wang wrote: > This patch add some common code for QCA/Atheros ath79 SOCs such as > DDR tuning, chip reset and CPU detection. > > Signed-off-by: Wills Wang wills.wang@live.com > --- > > Changes in v8: > - Use setbits_be32 > - Use lookup-table instead of big switch statement for CPU detection > Good stuff, minor nits below.
[...]
> +phys_size_t initdram(int board_type) > +{ > + ddr_tap_tuning(); Is the DDR tap tuning needed on all AR7xxx/AR9xxx systems ?
Yes, it's for optimizing DDR timing according to hardware. Sometimes, the hard code value is not ideal.
AR934x doesn't seem to need this.
I seem the following code in u-boot from Atheros LSDK: #ifdef CONFIG_AP123 ath_ddr_tap_cal();
Where can I get the LSDK ? I'd like to take a look and add it into my ar934x support patch.
> + return get_ram_size((void *)KSEG1, SZ_256M); > +} > diff --git a/arch/mips/mach-ath79/include/mach/ar71xx_regs.h > b/arch/mips/mach-ath79/include/mach/ar71xx_regs.h > new file mode 100644 > index 0000000..893dedc > --- /dev/null > +++ b/arch/mips/mach-ath79/include/mach/ar71xx_regs.h > @@ -0,0 +1,1184 @@ > +/* > + * Atheros AR71XX/AR724X/AR913X SoC register definitions > + * > + * Copyright (C) 2015-2016 Wills Wang wills.wang@live.com > + * Copyright (C) 2010-2011 Jaiganesh Narayanan > jnarayanan@atheros.com > + * Copyright (C) 2008-2010 Gabor Juhos juhosg@openwrt.org > + * Copyright (C) 2008 Imre Kaloz kaloz@openwrt.org > + * > + * SPDX-License-Identifier: GPL-2.0+ > + */ > + > +#ifndef __ASM_MACH_AR71XX_REGS_H > +#define __ASM_MACH_AR71XX_REGS_H > + > +#ifndef __ASSEMBLY__ > +#include <linux/bitops.h> > +#else > +#ifndef BIT > +#define BIT(nr) (1 << (nr)) This should really go into some common header.
This header is also included by some assembly code, but the BIT macro in linux/bitops.h isn't compatiable with assembler.
Because of the 1UL in it ?
Yes, there are also a few other C keywords and syntax, such as in-line function.
Hmmmmm, it pains me to see such duplication, but I have no better idea how to deal with it without introducing ifdefs.
actually such register definitions files are obsolete because register base adresses and sizes are taken from device-tree. The only exceptions are early startup or assembly code, but then you only need to add the definitions whose are needed by a driver. Thus I suggest you remove all the crap from this file you don't need.
And put the remaining bits into drivers themselves , right ?
But still, this doesn't help us with the BIT() macro redefinition problem.