RE: [PATCH] ARM: at91: board-dt-sama5: add phy_fixup to override NAND_Tree which improperly strapp-in during the reset period.

From: Yang, Wenyou
Date: Tue Dec 16 2014 - 20:21:44 EST


> -----Original Message-----
> From: Florian Fainelli [mailto:f.fainelli@xxxxxxxxx]
> Sent: Wednesday, December 17, 2014 3:10 AM
> To: Ferre, Nicolas; Yang, Wenyou
> Cc: linux@xxxxxxxxxxxxxxxx; linux-arm-kernel@xxxxxxxxxxxxxxxxxxx; linux-
> kernel@xxxxxxxxxxxxxxx
> Subject: Re: [PATCH] ARM: at91: board-dt-sama5: add phy_fixup to override
> NAND_Tree which improperly strapp-in during the reset period.
>
> On 16/12/14 02:20, Nicolas Ferre wrote:
> > Le 11/12/2014 08:31, Wenyou Yang a Ãcrit :
> >> Appearance: On some SAMA5D4EK boards, after power up, the Eth1 doesn't
> work.
> >>
> >> Reason: The PIOE2 pin is connected to the NAND_Tree# of KSZ8081, But
> >> it outputs LOW during the reset period, which cause the NAND_Tree# enabled.
> >>
> >> Add phy_fixup() to disable NAND_Tree by overriding the Operation Mode
> >> Strap Override register(i.e. Register 16h) to clear the NAND_Tree bit.
> >>
> >> Signed-off-by: Wenyou Yang <wenyou.yang@xxxxxxxxx>
> >
> > It seems correct to me.
>
> Just one minor nit, see below
>
> >
> > Acked-by: Nicolas Ferre <nicolas.ferre@xxxxxxxxx>
> >
> > Anyway, I add Florian in the loop as he is the one who proposed this
> > solution. So it would have been good to have his feeling about the
> > implementation...
> >
> > Bye,
> >
> >> ---
> >> arch/arm/mach-at91/board-dt-sama5.c | 18 ++++++++++++++++++
> >> 1 file changed, 18 insertions(+)
> >>
> >> diff --git a/arch/arm/mach-at91/board-dt-sama5.c
> >> b/arch/arm/mach-at91/board-dt-sama5.c
> >> index 8fb9ef5..97f7367 100644
> >> --- a/arch/arm/mach-at91/board-dt-sama5.c
> >> +++ b/arch/arm/mach-at91/board-dt-sama5.c
> >> @@ -17,6 +17,7 @@
> >> #include <linux/of_platform.h>
> >> #include <linux/phy.h>
> >> #include <linux/clk-provider.h>
> >> +#include <linux/phy.h>
> >>
> >> #include <asm/setup.h>
> >> #include <asm/irq.h>
> >> @@ -26,8 +27,25 @@
> >>
> >> #include "generic.h"
> >>
> >> +static int ksz8081_phy_fixup(struct phy_device *phy) {
> >> + int value;
> >> +
> >> + value = phy_read(phy, 0x16);
> >> + value &= ~0x20;
> >> + phy_write(phy, 0x16, value);
> >> +
> >> + return 0;
> >> +}
> >> +
> >> static void __init sama5_dt_device_init(void) {
> >> + if (of_machine_is_compatible("atmel,sama5d4ek") &&
> >> + IS_ENABLED(CONFIG_PHYLIB)) {
> >> + phy_register_fixup_for_id("fc028000.etherne:00",
> >> + ksz8081_phy_fixup);
>
> Matching the bus id certainly works, but I would rather match on your specific PHY
> device 32-bits OUI to make sure that if you ever change the PHY on a different
> design, this is not inadvertently writing to a register that is not micrel specific?
I don't think so.
Under the board compatible limitation, matching the bus id is enough for this case reason described above.

Moreover the PHY OUI is default value on the sama5d4ek.
MII_PHYSID1.15:0: 0x0022
MII_PHYSID2.15:10: 000101b

>
> >> + }
> >> +
> >> of_platform_populate(NULL, of_default_bus_match_table, NULL, NULL);
> >> }
> >>
> >>
> >
> >

Best Regards,
Wenyou Yang

N‹§²æ¸›yú²X¬¶ÇvØ–)Þ{.nlj·¥Š{±‘êX§¶›¡Ü}©ž²ÆzÚj:+v‰¨¾«‘êZ+€Êzf£¢·hšˆ§~†­†Ûÿû®w¥¢¸?™¨è&¢)ßf”ùy§m…á«a¶Úÿ 0¶ìå