Exynos SROMc configuration and Ethernet support for SMDK5410

Bindings are based on u-boot implementation, however they are stored in subnodes, providing support for more than one bank.

Since the driver now does more than suspend-resume support, dependency on CONFIG_PM is removed.

Signed-off-by Pavel Fedin <[email protected]>

---
 arch/arm/mach-exynos/Kconfig      |  2 +-
 drivers/soc/samsung/Kconfig       |  2 +-
 drivers/soc/samsung/exynos-srom.c | 42 ++++++++++++++++++++++++++++++++++++++-
 3 files changed, 43 insertions(+), 3 deletions(-)

arch/arm/mach-exynos/Kconfig +1 -1

index 83c85f5..c22dc42 100644
@@ -16,7 +16,7 @@ menuconfig ARCH_EXYNOS
 	select ARM_GIC
 	select COMMON_CLK_SAMSUNG
 	select EXYNOS_THERMAL
-	select EXYNOS_SROM if PM
+	select EXYNOS_SROM
 	select HAVE_ARM_SCU if SMP
 	select HAVE_S3C2410_I2C if I2C
 	select HAVE_S3C2410_WATCHDOG if WATCHDOG

drivers/soc/samsung/Kconfig +1 -1

index 2833b5b..ea4bc2a 100644
@@ -8,6 +8,6 @@ config SOC_SAMSUNG
 
 config EXYNOS_SROM
 	bool
-	depends on ARM && ARCH_EXYNOS && PM
+	depends on ARM && ARCH_EXYNOS
 
 endmenu

drivers/soc/samsung/exynos-srom.c +0 -0

index 57a232d..6c8c56a 100644
@@ -67,9 +67,46 @@ static struct exynos_srom_reg_dump *exynos_srom_alloc_reg_dump(
 	return rd;

drivers/soc/samsung/exynos-srom.c +15 -1

+	__raw_writel(bw, srom->reg_base + EXYNOS_SROM_BW);
+
+	__raw_writel((timing[0] << EXYNOS_SROM_BCX__PMC__SHIFT) |
+		    (timing[1] << EXYNOS_SROM_BCX__TACP__SHIFT) |
+		    (timing[2] << EXYNOS_SROM_BCX__TCAH__SHIFT) |
+		    (timing[3] << EXYNOS_SROM_BCX__TCOH__SHIFT) |
+		    (timing[4] << EXYNOS_SROM_BCX__TACC__SHIFT) |
+		    (timing[5] << EXYNOS_SROM_BCX__TCOS__SHIFT) |
+		    (timing[6] << EXYNOS_SROM_BCX__TACS__SHIFT),
+		    srom->reg_base + EXYNOS_SROM_BC0 + bank);
+
+	return 0;
+}
+
 static int exynos_srom_probe(struct platform_device *pdev)
 {
-	struct device_node *np;
+	struct device_node *np, *child;
 	struct exynos_srom *srom;
 	struct device *dev = &pdev->dev;
 

drivers/soc/samsung/exynos-srom.c +1 -0

+
 	return 0;
 }
 

Add documentation for new properties, allowing bank configuration.

Signed-off-by Pavel Fedin <[email protected]>

---
 .../bindings/arm/samsung/exynos-srom.txt           | 24 +++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

Documentation/devicetree/bindings/arm/samsung/exynos-srom.txt +0 -0

index 33886d5..9e4a40b 100644

Documentation/devicetree/bindings/arm/samsung/exynos-srom.txt +13 -0

+
+Example: basic definition, no banks are configured
+	sromc at 12570000 {
+		compatible = "samsung,exynos-srom";
+		reg = <0x12570000 0x10>;
+	};
+
+Example: SROMc with bank3 configuration
 	sromc at 12570000 {
 		compatible = "samsung,exynos-srom";
 		reg = <0x12570000 0x10>;
+		bank at 3 {
+			bank = <3>;
+			width = <2>;
+			srom-timing = <1 9 12 1 9 1 1>;
+		};
 	};

drivers/soc/samsung/exynos-srom.c static int exynos_srom_probe(struct platform_device *pdev) · line 141

Quoted 2 lines

drivers/soc/samsung/exynos-srom.c +2 -0

+	for_each_child_of_node(np, child)
+		decode_sromc(srom, child);

You ignore the return value here so bank may be not configured but device probe will return 0.

Maybe clean up and fail the probe?

Hello!

Quoted 5 lines

drivers/soc/samsung/exynos-srom.c static int exynos_srom_probe(struct platform_device *pdev) · line 141

Quoted 2 lines

drivers/soc/samsung/exynos-srom.c +2 -0

+	for_each_child_of_node(np, child)
+		decode_sromc(srom, child);

You ignore the return value here so bank may be not configured but device probe will return 0.

Yes, so that banks which are described correctly, will still be configured.

Quoted 2 lines

Maybe clean up and fail the probe?

I think it's not fatal, so cleanup is not necessary. May be dev_warn() in this case?

You missed here a lot of potential Cc. Please use get_maintainer.pl script. It *must* be sent to devicetree list.

Please send it also to DeviceTree maintainers because you are adding quite generic names for bindings so they may have interesting thoughts on this.

LKML list is also missing.

Any vendor prefix here? How generic is this?

But the system won't get any information about the failure. Device using these banks should not probe in such case.

Hello!

Quoted 8 lines

You missed here a lot of potential Cc. Please use get_maintainer.pl script. It *must* be sent to devicetree list.

Please send it also to DeviceTree maintainers because you are adding quite generic names for bindings so they may have interesting thoughts on this.

LKML list is also missing.

Ok. Next version will go there too.

Quoted 1 line

Any vendor prefix here? How generic is this?

I just don't know... Does *everything* really need a vendor prefix? How readable would that be? "compatible" property already says
that it's samsung-exynos-specific. And IMHO it's quite obvious that properties of vendor-specific device are automatically vendor-specific.
Ok, i am currently fixing up the rest and will post v4 soon, and will Cc: it to devicetree ML.

P.S. I intentionally omit the rest of the text in order to avoid overquoting, since we are already focused on some specific things.
If someone wants to get the whole picture, he/she could just browse back to the start of the thread. Or is it considered wrong approach in this particular ML?

I just don't know...

If someone wants to get the whole picture, he/she could just browse back to the start of the thread.

Quoted 3 lines

Quoted 1 line

Any vendor prefix here? How generic is this?

I just don't know... Does *everything* really need a vendor prefix? How readable would that

be? "compatible" property already says

Quoted 1 line

that it's samsung-exynos-specific. And IMHO it's quite obvious that properties of vendor-

specific device are automatically

Quoted 2 lines

vendor-specific.
Ok, i am currently fixing up the rest and will post v4 soon, and will Cc: it to devicetree

ML.

Which my reply you are referring to? You stripped part of some sentence and put it without *any context*. Just random sentence. I asked for vendor prefix in few places... srom-timing? width? And I do not remember where I used exactly these words.

Hello!

Quoted 10 lines

Quoted 3 lines

Quoted 1 line

Any vendor prefix here? How generic is this?

I just don't know... Does *everything* really need a vendor prefix? How readable would that

be? "compatible" property already says

Quoted 1 line

that it's samsung-exynos-specific. And IMHO it's quite obvious that properties of vendor-

specific device are automatically

Quoted 2 lines

vendor-specific.
Ok, i am currently fixing up the rest and will post v4 soon, and will Cc: it to devicetree

ML.

Quoted 4 lines

Which my reply you are referring to? You stripped part of some sentence and put it without *any context*. Just random sentence. I asked for vendor prefix in few places... srom-timing? width? And I do not remember where I used exactly these words.

Ok, sorry, i promise to improve. :)
Anyway, i have figured out how to add sub-devices, and heavily modified the whole thing. And indeed, vendor prefix is now very useful, so i added it to all three properties. Making v4...