Skip to contentSkip to conversation

am33xx: cpsw: default to ethernet hwaddr from efuse if not defined in dt

PATCH Superseded Peter Korsgaard 7 participants 15 messages Jan 17, 2013 – Sep 6, 2013 2 messages not in the archive

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 am33xx: cpsw: default to ethernet hwaddr from efuse if not defined in dt

When booting with CONFIG_ARM_APPENDED_DTB (either because of using an old U-Boot, not wanting the hassle of 2 files or when using Falcon fast boot mode in U-Boot), nothing updates the ethernet hwaddr specified for the CPSW slaves, causing the driver to use a random hwaddr, which is some times troublesome.

The am33xx has unique ethernet hwaddrs programmed in the efuse, so it makes more sense to default to these rather than random ones. Add a fixup step which adds mac-address dt properties using the efuse addresses if the DTB didn't contain valid ones.

---
Changes since v1:
- Use omap_arch_initcall as pointed out by Tony

 arch/arm/mach-omap2/Makefile      |    3 ++
 arch/arm/mach-omap2/am33xx-cpsw.c |   94 +++++++++++++++++++++++++++++++++++++
 arch/arm/mach-omap2/control.h     |    4 ++
 3 files changed, 101 insertions(+)
 create mode 100644 arch/arm/mach-omap2/am33xx-cpsw.c
arch/arm/mach-omap2/Makefile +3 -0
index 65fb6fb..54fb2ee 100644
@@ -298,4 +298,7 @@ endif
 emac-$(CONFIG_TI_DAVINCI_EMAC)		:= am35xx-emac.o
 obj-y					+= $(emac-m) $(emac-y)
 
+cpsw-$(CONFIG_TI_CPSW)			:= am33xx-cpsw.o
+obj-y					+= $(cpsw-m) $(cpsw-y)
+
 obj-y					+= common-board-devices.o twl-common.o dss-common.o
arch/arm/mach-omap2/am33xx-cpsw.c new file +94 -0

94 lines not on this page · raw message

arch/arm/mach-omap2/control.h +4 -0
index e6c3281..266d512 100644
@@ -352,6 +352,10 @@
 /* AM33XX CONTROL_STATUS register */
 #define AM33XX_CONTROL_STATUS		0x040
 #define AM33XX_CONTROL_SEC_CLK_CTRL	0x1bc
+#define AM33XX_CONTROL_MAC_ID0_LOW	0x630
+#define AM33XX_CONTROL_MAC_ID0_HIGH	0x634
+#define AM33XX_CONTROL_MAC_ID1_LOW	0x638
+#define AM33XX_CONTROL_MAC_ID1_HIGH	0x63c
 
 /* AM33XX CONTROL_STATUS bitfields (partial) */
 #define AM33XX_CONTROL_STATUS_SYSBOOT1_SHIFT		22
Signature · 1 line
1.7.10.4

A message that is not in the archive (sent off-list or not yet received): 1358461134-13452-1-git-send-email-jacmet-OfajU3CKLf1/[email protected]

Acked-by
#2 lore raw
Quoted 6 lines · Peter Korsgaard

When booting with CONFIG_ARM_APPENDED_DTB (either because of using an old U-Boot, not wanting the hassle of 2 files or when using Falcon fast boot mode in U-Boot), nothing updates the ethernet hwaddr specified for the CPSW slaves, causing the driver to use a random hwaddr, which is some times troublesome.

Quoted 7 lines

The am33xx has unique ethernet hwaddrs programmed in the efuse, so it makes more sense to default to these rather than random ones. Add a fixup step which adds mac-address dt properties using the efuse addresses if the DTB didn't contain valid ones.

This implementation looks fine.

Regards
Mugunthan V N

A message that is not in the archive (sent off-list or not yet received): [email protected]

#4 lore raw
Quoted 15 lines · Mugunthan V N
Quoted 13 lines · Peter Korsgaard

When booting with CONFIG_ARM_APPENDED_DTB (either because of using an old U-Boot, not wanting the hassle of 2 files or when using Falcon fast boot mode in U-Boot), nothing updates the ethernet hwaddr specified for the CPSW slaves, causing the driver to use a random hwaddr, which is some times
troublesome.

The am33xx has unique ethernet hwaddrs programmed in the efuse, so it makes
more sense to default to these rather than random ones. Add a fixup step which adds mac-address dt properties using the efuse addresses if the DTB didn't contain valid ones.

Quoted 9 lines
Quoted 3 lines

This implementation looks fine.

Regards
Mugunthan V N

Thanks,
My approach for cpsw driver is posted here:
https://patchwork.kernel.org/patch/1966481/
I added DT option mac-address-source because there is still possiblility to have mac address set by the u-boot in the future. Beaglebone have special eeprom on board, were you can have stored user MAC and u-boot can read eeprom
data. Maybe some users will want this feature. Also in cpsw dt binding documentation, there is mention /* Filled in by U-Boot */. For that purpose,
code I proposed can easily deal with any driver exteds.

Reading CPU mac address registers with (my approach) or without (your approach) DT should be more discussed. Also your way seems to be fine.

_______________________________________________
devicetree-discuss mailing list devicetree-discuss-uLR06cmDAlY/[email protected]
https://lists.ozlabs.org/listinfo/devicetree-discuss

Tested-by
#5 lore raw
In reply to Mugunthan V N
Quoted 12 lines · Mugunthan V N
Quoted 11 lines · Peter Korsgaard

When booting with CONFIG_ARM_APPENDED_DTB (either because of using an old U-Boot, not wanting the hassle of 2 files or when using Falcon fast boot mode in U-Boot), nothing updates the ethernet hwaddr specified for the CPSW slaves, causing the driver to use a random hwaddr, which is some times troublesome.

The am33xx has unique ethernet hwaddrs programmed in the efuse, so it makes more sense to default to these rather than random ones. Add a fixup step which adds mac-address dt properties using the efuse addresses if the DTB didn't contain valid ones.

Quoted 5 lines
Quoted 2 lines

This implementation looks fine.

#6 lore raw
Quoted 13 lines · Mark Jackson
Quoted 12 lines · Mugunthan V N
Quoted 11 lines · Peter Korsgaard

When booting with CONFIG_ARM_APPENDED_DTB (either because of using an old U-Boot, not wanting the hassle of 2 files or when using Falcon fast boot mode in U-Boot), nothing updates the ethernet hwaddr specified for the CPSW slaves, causing the driver to use a random hwaddr, which is some times troublesome.

The am33xx has unique ethernet hwaddrs programmed in the efuse, so it makes more sense to default to these rather than random ones. Add a fixup step which adds mac-address dt properties using the efuse addresses if the DTB didn't contain valid ones.

Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

This implementation looks fine.

Is this ever going to be put into the mainline code ?

#7 lore raw
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Mark" == Mark Jackson <[email protected]> writes:

Hi,

Quoted 13 lines · Mark Jackson
Quoted 13 lines · Mugunthan V N
Quoted 12 lines · Peter Korsgaard
Quoted 11 lines

When booting with CONFIG_ARM_APPENDED_DTB (either because of using an old U-Boot, not wanting the hassle of 2 files or when using Falcon fast boot mode in U-Boot), nothing updates the ethernet hwaddr specified for the CPSW slaves, causing the driver to use a random hwaddr, which is some times troublesome.

The am33xx has unique ethernet hwaddrs programmed in the efuse, so it makes more sense to default to these rather than random ones. Add a fixup step which adds mac-address dt properties using the efuse addresses if the DTB didn't contain valid ones.

Quoted 6 lines
Quoted 6 lines
Quoted 4 lines
Quoted 1 line

This implementation looks fine.

Mark> Is this ever going to be put into the mainline code ?

Good question. Tony, could you please pick this up? It has been pending since January and has a number of acks.

Do you want me to resend?

Signature · 1 line
Bye, Peter Korsgaard
Tested-by
#8 lore raw
Quoted 23 lines · Peter Korsgaard
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Mark" == Mark Jackson <[email protected]> writes:

Hi,

Quoted 18 lines · Mark Jackson
Quoted 18 lines · Mugunthan V N
Quoted 16 lines · Peter Korsgaard
Quoted 6 lines

When booting with CONFIG_ARM_APPENDED_DTB (either because of using an old U-Boot, not wanting the hassle of 2 files or when using Falcon fast boot mode in U-Boot), nothing updates the ethernet hwaddr specified for the CPSW slaves, causing the driver to use a random hwaddr, which is some times troublesome.

Quoted 6 lines

The am33xx has unique ethernet hwaddrs programmed in the efuse, so it makes more sense to default to these rather than random ones. Add a fixup step which adds mac-address dt properties using the efuse addresses if the DTB didn't contain valid ones.

This implementation looks fine.

Quoted 8 lines
Quoted 1 line
Quoted 1 line

Mark> Is this ever going to be put into the mainline code ?

Good question. Tony, could you please pick this up? It has been pending since January and has a number of acks.

Do you want me to resend?

Also working nicely here on 3.11.

Kevin or Olof: can you apply? Seems to be continuing no response after Ack back in January.

-Matt

#9 lore raw
Quoted 31 lines · Matt Porter
Quoted 30 lines · Peter Korsgaard
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Mark" == Mark Jackson <[email protected]> writes:

Hi,

Quoted 13 lines · Mark Jackson
Quoted 13 lines · Mugunthan V N
Quoted 12 lines · Peter Korsgaard
Quoted 11 lines

When booting with CONFIG_ARM_APPENDED_DTB (either because of using an old U-Boot, not wanting the hassle of 2 files or when using Falcon fast boot mode in U-Boot), nothing updates the ethernet hwaddr specified for the CPSW slaves, causing the driver to use a random hwaddr, which is some times troublesome.

The am33xx has unique ethernet hwaddrs programmed in the efuse, so it makes more sense to default to these rather than random ones. Add a fixup step which adds mac-address dt properties using the efuse addresses if the DTB didn't contain valid ones.

Quoted 6 lines
Quoted 6 lines
Quoted 4 lines
Quoted 1 line

This implementation looks fine.

Mark> Is this ever going to be put into the mainline code ?

Good question. Tony, could you please pick this up? It has been pending since January and has a number of acks.

Quoted 8 lines
Quoted 1 line

Do you want me to resend?

Also working nicely here on 3.11.

Kevin or Olof: can you apply? Seems to be continuing no response after Ack back in January.

I have no idea what the patch is, it's no longer in the email. Can you resend it, please?

-Olof

#10 lore raw
Quoted 6 lines
Quoted 6 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Olof" == Olof Johansson <[email protected]> writes:

Kevin or Olof: can you apply? Seems to be continuing no response after Ack back in January.

Olof> I have no idea what the patch is, it's no longer in the email. Can you Olof> resend it, please?

Sure, one moment.

Signature · 1 line
Bye, Peter Korsgaard
#11 lore raw

PATCH RESEND 1/1 am33xx: cpsw: default to ethernet hwaddr from efuse if not defined in dt

In reply to Olof Johansson

When booting with CONFIG_ARM_APPENDED_DTB (either because of using an old U-Boot, not wanting the hassle of 2 files or when using Falcon fast boot mode in U-Boot), nothing updates the ethernet hwaddr specified for the CPSW slaves, causing the driver to use a random hwaddr, which is some times troublesome.

The am33xx has unique ethernet hwaddrs programmed in the efuse, so it makes more sense to default to these rather than random ones. Add a fixup step which adds mac-address dt properties using the efuse addresses if the DTB didn't contain valid ones.

---
Resent without changes.

 arch/arm/mach-omap2/Makefile      |    3 ++
 arch/arm/mach-omap2/am33xx-cpsw.c |   94 +++++++++++++++++++++++++++++++++++++
 arch/arm/mach-omap2/control.h     |    4 ++
 3 files changed, 101 insertions(+)
 create mode 100644 arch/arm/mach-omap2/am33xx-cpsw.c
arch/arm/mach-omap2/Makefile +3 -0
index afb457c..0afee42 100644
@@ -305,4 +305,7 @@ endif
 emac-$(CONFIG_TI_DAVINCI_EMAC)		:= am35xx-emac.o
 obj-y					+= $(emac-m) $(emac-y)
 
+cpsw-$(CONFIG_TI_CPSW)			:= am33xx-cpsw.o
+obj-y					+= $(cpsw-m) $(cpsw-y)
+
 obj-y					+= common-board-devices.o twl-common.o dss-common.o
arch/arm/mach-omap2/am33xx-cpsw.c new file +94 -0

94 lines not on this page · raw message

arch/arm/mach-omap2/control.h +4 -0
index f7d7c2e..5d684b7 100644
@@ -352,6 +352,10 @@
 /* AM33XX CONTROL_STATUS register */
 #define AM33XX_CONTROL_STATUS		0x040
 #define AM33XX_CONTROL_SEC_CLK_CTRL	0x1bc
+#define AM33XX_CONTROL_MAC_ID0_LOW	0x630
+#define AM33XX_CONTROL_MAC_ID0_HIGH	0x634
+#define AM33XX_CONTROL_MAC_ID1_LOW	0x638
+#define AM33XX_CONTROL_MAC_ID1_HIGH	0x63c
 
 /* AM33XX CONTROL_STATUS bitfields (partial) */
 #define AM33XX_CONTROL_STATUS_SYSBOOT1_SHIFT		22
Signature · 1 line
1.7.10.4
#12 lore raw
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Peter" == Peter Korsgaard <[email protected]> writes:

Peter> When booting with CONFIG_ARM_APPENDED_DTB (either because of using an old Peter> U-Boot, not wanting the hassle of 2 files or when using Falcon fast boot Peter> mode in U-Boot), nothing updates the ethernet hwaddr specified for the Peter> CPSW slaves, causing the driver to use a random hwaddr, which is some times Peter> troublesome.

Peter> The am33xx has unique ethernet hwaddrs programmed in the efuse, so it makes Peter> more sense to default to these rather than random ones. Add a fixup step Peter> which adds mac-address dt properties using the efuse addresses if the DTB Peter> didn't contain valid ones.

Peter> Signed-off-by: Peter Korsgaard <[email protected]> Peter> Acked-by: Mugunthan V N <[email protected]> Peter> Tested-by: Mark Jackson <[email protected]> Peter> Tested-by: Koen Kooi <[email protected]> Peter> Tested-by: Matt Porter <[email protected]> Peter> ---
Peter> Resent without changes.

Argh, this was supposed to have been a resend of v2. Will resend that one instead.

Signature · 1 line
Bye, Peter Korsgaard
#13 lore raw

PATCH RESEND v2 1/1 am33xx: cpsw: default to ethernet hwaddr from efuse if not defined in dt

In reply to Peter Korsgaard

When booting with CONFIG_ARM_APPENDED_DTB (either because of using an old U-Boot, not wanting the hassle of 2 files or when using Falcon fast boot mode in U-Boot), nothing updates the ethernet hwaddr specified for the CPSW slaves, causing the driver to use a random hwaddr, which is some times troublesome.

The am33xx has unique ethernet hwaddrs programmed in the efuse, so it makes more sense to default to these rather than random ones. Add a fixup step which adds mac-address dt properties using the efuse addresses if the DTB didn't contain valid ones.

---
Changes since v1:
- Use omap_arch_initcall as pointed out by Tony

 arch/arm/mach-omap2/Makefile      |    3 ++
 arch/arm/mach-omap2/am33xx-cpsw.c |   94 +++++++++++++++++++++++++++++++++++++
 arch/arm/mach-omap2/control.h     |    4 ++
 3 files changed, 101 insertions(+)
 create mode 100644 arch/arm/mach-omap2/am33xx-cpsw.c
arch/arm/mach-omap2/Makefile +3 -0
index afb457c..0afee42 100644
@@ -305,4 +305,7 @@ endif
 emac-$(CONFIG_TI_DAVINCI_EMAC)		:= am35xx-emac.o
 obj-y					+= $(emac-m) $(emac-y)
 
+cpsw-$(CONFIG_TI_CPSW)			:= am33xx-cpsw.o
+obj-y					+= $(cpsw-m) $(cpsw-y)
+
 obj-y					+= common-board-devices.o twl-common.o dss-common.o
arch/arm/mach-omap2/am33xx-cpsw.c new file +94 -0

94 lines not on this page · raw message

arch/arm/mach-omap2/control.h +4 -0
index f7d7c2e..5d684b7 100644
@@ -352,6 +352,10 @@
 /* AM33XX CONTROL_STATUS register */
 #define AM33XX_CONTROL_STATUS		0x040
 #define AM33XX_CONTROL_SEC_CLK_CTRL	0x1bc
+#define AM33XX_CONTROL_MAC_ID0_LOW	0x630
+#define AM33XX_CONTROL_MAC_ID0_HIGH	0x634
+#define AM33XX_CONTROL_MAC_ID1_LOW	0x638
+#define AM33XX_CONTROL_MAC_ID1_HIGH	0x63c
 
 /* AM33XX CONTROL_STATUS bitfields (partial) */
 #define AM33XX_CONTROL_STATUS_SYSBOOT1_SHIFT		22
Signature · 1 line
1.7.10.4
#14 lore raw

Hi,

Quoted 70 lines · Peter Korsgaard

When booting with CONFIG_ARM_APPENDED_DTB (either because of using an old U-Boot, not wanting the hassle of 2 files or when using Falcon fast boot mode in U-Boot), nothing updates the ethernet hwaddr specified for the CPSW slaves, causing the driver to use a random hwaddr, which is some times troublesome.

The am33xx has unique ethernet hwaddrs programmed in the efuse, so it makes more sense to default to these rather than random ones. Add a fixup step which adds mac-address dt properties using the efuse addresses if the DTB didn't contain valid ones.

---

arch/arm/mach-omap2/am33xx-cpsw.c new file +47 -0
new file mode 100644
index 0000000..eec29a4
@@ -0,0 +1,94 @@
+/*
+ * am335x specific cpsw dt fixups
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation version 2.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ */
+
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/etherdevice.h>
+#include <linux/of.h>
+#include <linux/of_net.h>
+
+#include "soc.h"
+#include "control.h"
+
+/**
+ * am33xx_dt_cpsw_set_mac_from_efuse - Add mac-address property using
+ * ethernet hwaddr from efuse
+ * @np:		Pointer to the cpsw slave to set mac address of
+ * @idx:	Mac address index to use from efuse
+ */
+static void am33xx_dt_cpsw_set_mac_from_efuse(struct device_node *np, int idx)
+{
+	struct property *prop;
+	u32 lo, hi;
+	u8 *mac;
+
+	switch (idx) {
+	case 0:
+		lo = omap_ctrl_readl(AM33XX_CONTROL_MAC_ID0_LOW);
+		hi = omap_ctrl_readl(AM33XX_CONTROL_MAC_ID0_HIGH);
+		break;
+
+	case 1:
+		lo = omap_ctrl_readl(AM33XX_CONTROL_MAC_ID1_LOW);
+		hi = omap_ctrl_readl(AM33XX_CONTROL_MAC_ID1_HIGH);
+		break;
+
+	default:
+		pr_err("cpsw.%d: too many slaves found\n", idx);

arch/arm/mach-omap2/am33xx-cpsw.c line 57

Quoted 10 lines
arch/arm/mach-omap2/am33xx-cpsw.c new file +10 -0
+		return;
+	}
+
+	prop = kzalloc(sizeof(*prop) + ETH_ALEN, GFP_KERNEL);
+	if (!prop)
+		return;
+
+	prop->value = prop + 1;
+	prop->length = ETH_ALEN;
+	prop->name = kstrdup("mac-address", GFP_KERNEL);

Hmm. Do we want this to be mac-address or local-mac-address? If it's mac-address, then this will override anything set by u-boot (by means of priority in of_net). I don't think that's desireable. So it should probably set local-mac-address instead.

Quoted 5 lines
arch/arm/mach-omap2/am33xx-cpsw.c new file +5 -0
+	if (!prop->name) {
+		kfree(prop);
+		return;
+	}
+

arch/arm/mach-omap2/am33xx-cpsw.c line 72

Quoted 10 lines
arch/arm/mach-omap2/am33xx-cpsw.c new file +10 -0
+	mac = prop->value;
+
+	mac[0] = hi;
+	mac[1] = hi >> 8;
+	mac[2] = hi >> 16;
+	mac[3] = hi >> 24;
+	mac[4] = lo;
+	mac[5] = lo >> 8;
+
+	of_update_property(np, prop);

mach-mxs does this too, I wonder if it's worth creating a small helper for it.

Doesn't have to be done as part of this patch but it's still worth doing.

Quoted 4 lines
arch/arm/mach-omap2/am33xx-cpsw.c new file +4 -0
+
+	pr_info("cpsw.%d: No hwaddr in dt. Using %pM from efuse\n", idx, mac);
+}
+

arch/arm/mach-omap2/am33xx-cpsw.c line 86

Quoted 10 lines
arch/arm/mach-omap2/am33xx-cpsw.c new file +10 -0
+static int __init am33xx_dt_cpsw_mac_fixup(void)
+{
+	struct device_node *np, *slave;
+	int idx = 0;
+
+	if (!soc_is_am33xx())
+		return -ENODEV;
+
+	for_each_compatible_node(np, NULL, "ti,cpsw")
+		for_each_node_by_name(slave, "slave") {

The binding doesn't enforce "slave" node names for the subnodes, so you should probably just iterate over children. Right now they are named slave@0 and slave@1, but lack reg properties (and a notion of what said reg property would symbolize).

arch/arm/mach-omap2/am33xx-cpsw.c line 89

Quoted 3 lines
arch/arm/mach-omap2/am33xx-cpsw.c new file +3 -0
+			if (!of_get_mac_address(slave))
+				am33xx_dt_cpsw_set_mac_from_efuse(slave, idx);
+			idx++;

You're assuming that you get the children in order here, which is not guaranteed.

Maybe best is to adjust the binding, to make "reg" mean something (i.e. the MAC index for this hardware) and use the <reg> property of the childnode to tell which efuse to read.

-Olof

#15 lore raw
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Olof" == Olof Johansson <[email protected]> writes:

Hi Olof,

Thanks for the review!

Quoted 27 lines
Quoted 27 lines
arch/arm/mach-omap2/am33xx-cpsw.c new file +27 -0
+/**
+ * am33xx_dt_cpsw_set_mac_from_efuse - Add mac-address property using
+ * ethernet hwaddr from efuse
+ * @np:		Pointer to the cpsw slave to set mac address of
+ * @idx:	Mac address index to use from efuse
+ */
+static void am33xx_dt_cpsw_set_mac_from_efuse(struct device_node *np, int idx)
+{
+	struct property *prop;
+	u32 lo, hi;
+	u8 *mac;
+
+	switch (idx) {
+	case 0:
+		lo = omap_ctrl_readl(AM33XX_CONTROL_MAC_ID0_LOW);
+		hi = omap_ctrl_readl(AM33XX_CONTROL_MAC_ID0_HIGH);
+		break;
+
+	case 1:
+		lo = omap_ctrl_readl(AM33XX_CONTROL_MAC_ID1_LOW);
+		hi = omap_ctrl_readl(AM33XX_CONTROL_MAC_ID1_HIGH);
+		break;
+
+	default:
+		pr_err("cpsw.%d: too many slaves found\n", idx);
+		return;
+	}
Quoted 8 lines

arch/arm/mach-omap2/am33xx-cpsw.c line 57

Quoted 8 lines
arch/arm/mach-omap2/am33xx-cpsw.c new file +8 -0
+
+	prop = kzalloc(sizeof(*prop) + ETH_ALEN, GFP_KERNEL);
+	if (!prop)
+		return;
+
+	prop->value = prop + 1;
+	prop->length = ETH_ALEN;
+	prop->name = kstrdup("mac-address", GFP_KERNEL);

Olof> Hmm. Do we want this to be mac-address or local-mac-address? If it's Olof> mac-address, then this will override anything set by u-boot (by means Olof> of priority in of_net). I don't think that's desireable. So it should Olof> probably set local-mac-address instead.

It doesn't really matter as this only gets called if the DTB doesn't contain any valid mac-address / local-mac-address properties (the if (!of_get_mac_address(slave)) check).

Quoted 7 lines
Quoted 7 lines
arch/arm/mach-omap2/am33xx-cpsw.c new file +7 -0
+	if (!prop->name) {
+		kfree(prop);
+		return;
+	}
+
+	mac = prop->value;
+
Quoted 8 lines

arch/arm/mach-omap2/am33xx-cpsw.c line 72

Quoted 8 lines
arch/arm/mach-omap2/am33xx-cpsw.c new file +8 -0
+	mac[0] = hi;
+	mac[1] = hi >> 8;
+	mac[2] = hi >> 16;
+	mac[3] = hi >> 24;
+	mac[4] = lo;
+	mac[5] = lo >> 8;
+
+	of_update_property(np, prop);

Olof> mach-mxs does this too, I wonder if it's worth creating a small helper for it.

Yeah, guess where I got the inspiration to do this from? ;)

Olof> Doesn't have to be done as part of this patch but it's still worth doing.

Ok, I'll send a patch adding of_set_mac_address() and change am335x / mxs to use it once this is in.

Quoted 6 lines
Quoted 6 lines
arch/arm/mach-omap2/am33xx-cpsw.c new file +6 -0
+
+	pr_info("cpsw.%d: No hwaddr in dt. Using %pM from efuse\n", idx, mac);
+}
+
+static int __init am33xx_dt_cpsw_mac_fixup(void)
+{
Quoted 8 lines

arch/arm/mach-omap2/am33xx-cpsw.c line 86

Quoted 8 lines
arch/arm/mach-omap2/am33xx-cpsw.c new file +8 -0
+	struct device_node *np, *slave;
+	int idx = 0;
+
+	if (!soc_is_am33xx())
+		return -ENODEV;
+
+	for_each_compatible_node(np, NULL, "ti,cpsw")
+		for_each_node_by_name(slave, "slave") {

Olof> The binding doesn't enforce "slave" node names for the subnodes, so you Olof> should probably just iterate over children. Right now they are named Olof> slave@0 and slave@1, but lack reg properties (and a notion of what said Olof> reg property would symbolize).

True. This just follows what the cpsw driver does. I would prefer to change the driver, this file, am33xx.dtsi and the bindings documentation in one go rather than doing something else than the driver here.

Quoted 3 lines

arch/arm/mach-omap2/am33xx-cpsw.c line 89

Quoted 3 lines
arch/arm/mach-omap2/am33xx-cpsw.c new file +3 -0
+			if (!of_get_mac_address(slave))
+				am33xx_dt_cpsw_set_mac_from_efuse(slave, idx);
+			idx++;

Olof> You're assuming that you get the children in order here, which is not Olof> guaranteed.

Again, this is in line with cpsw.c. Once we add a reg property we can use that as idx.

Olof> Maybe best is to adjust the binding, to make "reg" mean something Olof> (i.e. the MAC index for this hardware) and use the <reg> property Olof> of the childnode to tell which efuse to read.

Indeed.

Signature · 1 line
Bye, Peter Korsgaard
Tested-by
#16 lore raw
In reply to Mark Jackson

Op 8 jul. 2013, om 14:42 heeft Mark Jackson <[email protected]> het volgende geschreven:

Quoted 13 lines · Mugunthan V N
Quoted 12 lines · Peter Korsgaard
Quoted 11 lines

When booting with CONFIG_ARM_APPENDED_DTB (either because of using an old U-Boot, not wanting the hassle of 2 files or when using Falcon fast boot mode in U-Boot), nothing updates the ethernet hwaddr specified for the CPSW slaves, causing the driver to use a random hwaddr, which is some times troublesome.

The am33xx has unique ethernet hwaddrs programmed in the efuse, so it makes more sense to default to these rather than random ones. Add a fixup step which adds mac-address dt properties using the efuse addresses if the DTB didn't contain valid ones.

Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

This implementation looks fine.