[PATCH 0/4] EBI2 bus and ethernet on MSM8660/APQ8060

[PATCH 0/4] EBI2 bus and ethernet on MSM8660/APQ8060

[Linus Walleij]

This is the second iteration of the EBI2+ethernet series for
MSM8660/APQ8060. Things are working like a charm on the APQ8060
Dragonboard and all review comments are addressed AFAICT.

Linus Walleij (4):
  soc: qcom: add EBI2 device tree bindings
  soc: qcom: add EBI2 driver
  ARM: dts: add EBI2 to the Qualcomm MSM8660 DTSI
  ARM: dts: add SMSC ethernet on the APQ8060 Dragonboard

 .../devicetree/bindings/bus/qcom,ebi2.txt          | 136 +++++++
 arch/arm/boot/dts/qcom-apq8060-dragonboard.dts     | 130 +++++++
 arch/arm/boot/dts/qcom-msm8660.dtsi                |  17 +
 drivers/bus/Kconfig                                |   7 +
 drivers/bus/Makefile                               |   1 +
 drivers/bus/qcom-ebi2.c                            | 404 +++++++++++++++++++++
 6 files changed, 695 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/bus/qcom,ebi2.txt
 create mode 100644 drivers/bus/qcom-ebi2.c

-- 
2.7.4

[PATCH 1/4 v2] soc: qcom: add EBI2 device tree bindings

[Linus Walleij]

This adds device tree bindings for the External Bus Interface 2, EBI2
to the Qualcomm SoC drivers.

Cc: devicetree-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
Signed-off-by: Linus Walleij <linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
---
ChangeLog v1->v2:
- Introduce per-SoC compatible strings for apq8060 and msm8660
- Move the binding to bus/qcom,ebi2.txt
- Add a vendor prefix for each chipselect XMEM configuration option
  "qcom,xmem-foo"
- Move all CS XMEM settings to the EBI2 node, encode the six
  different chipselects into an array of settings.
- Use ranges to specify the address windows for the six different
  chipselects.
- Alter #address-cells to <2> using the first as chipselect and the
  second as offset into the range.
- Augment the example accordingly.
---
 .../devicetree/bindings/bus/qcom,ebi2.txt          | 136 +++++++++++++++++++++
 1 file changed, 136 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/bus/qcom,ebi2.txt

diff --git a/Documentation/devicetree/bindings/bus/qcom,ebi2.txt b/Documentation/devicetree/bindings/bus/qcom,ebi2.txt
new file mode 100644
index 000000000000..4adadb65bea6
--- /dev/null
+++ b/Documentation/devicetree/bindings/bus/qcom,ebi2.txt
@@ -0,0 +1,136 @@
+Qualcomm External Bus Interface 2 (EBI2)
+
+The EBI2 contains two peripheral blocks: XMEM and LCDC. The XMEM handles any
+external memory (such as NAND or other memory-mapped peripherals) whereas
+LCDC handles LCD displays.
+
+As it says it connects devices to an external bus interface, meaning address
+lines (up to 9 address lines so can only address 1KiB external memory space),
+data lines (16 bits), OE (output enable), ADV (address valid, used on some
+NOR flash memories), WE (write enable). This on top of 6 different chip selects
+(CS0 thru CS5) so that in theory 6 different devices can be connected.
+
+Apparently this bus is clocked at 64MHz. It has dedicated pins on the package
+and the bus can only come out on these pins, however if some of the pins are
+unused they can be left unconnected or remuxed to be used as GPIO or in some
+cases other orthogonal functions as well.
+
+Also CS1 and CS2 has -A and -B signals. Why they have that is unclear to me.
+
+The chip selects have the following memory range assignments. This region of
+memory is referred to as "Chip Peripheral SS FPB0" and is 168MB big.
+
+Chip Select                     Physical address base
+CS0 GPIO134                     0x1a800000-0x1b000000 (8MB)
+CS1 GPIO39 (A) / GPIO123 (B)    0x1b000000-0x1b800000 (8MB)
+CS2 GPIO40 (A) / GPIO124 (B)    0x1b800000-0x1c000000 (8MB)
+CS3 GPIO133                     0x1d000000-0x25000000 (128 MB)
+CS4 GPIO132                     0x1c800000-0x1d000000 (8MB)
+CS5 GPIO131                     0x1c000000-0x1c800000 (8MB)
+
+The APQ8060 Qualcomm Application Processor User Guide, 80-N7150-14 Rev. A,
+August 6, 2012 contains some incomplete documentation of the EBI2.
+
+FIXME: the manual mentions "write precharge cycles" and "precharge cycles".
+We have not been able to figure out which bit fields these correspond to
+in the hardware, or what valid values exist. The current hypothesis is that
+this is something just used on the FAST chip selects and that the SLOW
+chip selects are understood fully. There is also a "byte device enable"
+flag somewhere for 8bit memories.
+
+FIXME: The chipselects have SLOW and FAST configuration registers. It's a bit
+unclear what this means, if they are mutually exclusive or can be used
+together, or if some chip selects are hardwired to be FAST and others are SLOW
+by design.
+
+The XMEM registers are totally undocumented but could be partially decoded
+because the Cypress AN49576 Antioch Westbridge apparently has suspiciously
+similar register layout, see: http://www.cypress.com/file/105771/download
+
+Required properties:
+- compatible: should be one of:
+  "qcom,msm8660-ebi2"
+  "qcom,apq8060-ebi2"
+- #address-cells: shoule be <2>: the first cell is the chipselect,
+  the second cell is the offset inside the memory range
+- #size-cells: should be <1>
+- ranges: should be set to:
+  ranges = <0 0x0 0x1a800000 0x00800000>,
+           <1 0x0 0x1b000000 0x00800000>,
+           <2 0x0 0x1b800000 0x00800000>,
+           <3 0x0 0x1d000000 0x00800000>,
+           <4 0x0 0x1c800000 0x00800000>,
+           <5 0x0 0x1c000000 0x00800000>;
+- reg: two ranges of registers: EBI2 config and XMEM config areas
+- reg-names: should be "ebi2", "xmem"
+- clocks: two clocks, EBI_2X and EBI
+- clock-names: shoule be "ebi2x", "ebi2"
+
+The following optional properties are all arrays indexed to the target
+chipselect, usually an array of six elements representing the six chipselects:
+
+Optional properties arrays for SLOW chip selects:
+- qcom,xmem-recovery-cycles: recovery cycles is the time the memory continues to
+  drive the data bus after OE is de-asserted, in order to avoid contention on
+  the data bus. They are inserted when reading one CS and switching to another
+  CS or read followed by write on the same CS. Valid values 0 thru 15. Minimum
+  value is actually 1, so a value of 0 will still yield 1 recovery cycle.
+- qcom,xmem-write-hold-cycles: write hold cycles, these are extra cycles
+  inserted after every write minimum 1. The data out is driven from the time
+  WE is asserted until CS is asserted. With a hold of 1 (value = 0), the CS
+  stays active for 1 extra cycle etc. Valid values 0 thru 15.
+- qcom,xmem-write-delta-cycles: initial latency for write cycles inserted for
+  the first write to a page or burst memory. Valid values 0 thru 255.
+- qcom,xmem-read-delta-cycles: initial latency for read cycles inserted for the
+  first read to a page or burst memory. Valid values 0 thru 255.
+- qcom,xmem-write-wait-cycles: number of wait cycles for every write access, 0=1
+  cycle. Valid values 0 thru 15.
+- qcom,xmem-read-wait-cycles: number of wait cycles for every read access, 0=1
+  cycle. Valid values 0 thru 15.
+
+Optional properties arrays for FAST chip selects:
+- qcom,xmem-address-hold-enable: this is a boolean property stating that we
+  shall hold the address for an extra cycle to meet hold time requirements
+  with ADV assertion.
+- qcom,xmem-adv-to-oe-recovery-cycles: the number of cycles elapsed before an OE
+  assertion, with respect to the cycle where ADV (address valid) is asserted.
+  2 means 2 cycles between ADV and OE. Valid values 0, 1, 2 or 3.
+- qcom,xmem-read-hold-cycles: the length in cycles of the first segment of a
+  read transfer. For a single read trandfer this will be the time from CS
+  assertion to OE assertion. Valid values 0 thru 15.
+
+Optional subnodes:
+- Nodes inside the EBI2 will be considered device nodes.
+
+Example:
+
+ebi2@1a100000 {
+	compatible = "qcom,apq8060-ebi2";
+	#address-cells = <2>;
+	#size-cells = <1>;
+	ranges = <0 0x0 0x1a800000 0x00800000>,
+		 <1 0x0 0x1b000000 0x00800000>,
+		 <2 0x0 0x1b800000 0x00800000>,
+		 <3 0x0 0x1d000000 0x00800000>,
+		 <4 0x0 0x1c800000 0x00800000>,
+		 <5 0x0 0x1c000000 0x00800000>;
+	reg = <0x1a100000 0x1000>, <0x1a110000 0x1000>;
+	reg-names = "ebi2", "xmem";
+	clocks = <&gcc EBI2_2X_CLK>, <&gcc EBI2_CLK>;
+	clock-names = "ebi2x", "ebi2";
+	/* Make sure to set up the pin control for the EBI2 */
+	pinctrl-names = "default";
+	pinctrl-0 = <&foo_ebi2_pins>;
+	qcom,xmem-recovery-cycles = <0>, <0>, <0>, <0>, <0>, <0>;
+	qcom,xmem-write-hold-cycles = <0>, <0>, <3>, <0>, <0>, <0>;
+	qcom,xmem-write-delta-cycles = <0>, <0>, <31>, <0>, <0>, <0>;
+	qcom,xmem-read-delta-cycles = <0>, <0>, <28>, <0>, <0>, <0>;
+	qcom,xmem-write-wait-cycles = <0>, <0>, <9>, <0>, <0>, <0>;
+	qcom,xmem-read-wait-cycles = <0>, <0>, <9>, <0>, <0>, <0>;
+
+ 	foo-ebi2@1b800000 {
+		compatible = "foo";
+		reg = <2 0x1b800000 0x100>;
+		(...)
+	};
+};
-- 
2.7.4

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/4 v2] soc: qcom: add EBI2 device tree bindings

[Arnd Bergmann]

On Wednesday 24 August 2016, Linus Walleij wrote:
> +
> +ebi2@1a100000 {
> +	compatible = "qcom,apq8060-ebi2";
> +	#address-cells = <2>;
> +	#size-cells = <1>;
> +	ranges = <0 0x0 0x1a800000 0x00800000>,
> +		 <1 0x0 0x1b000000 0x00800000>,
> +		 <2 0x0 0x1b800000 0x00800000>,
> +		 <3 0x0 0x1d000000 0x00800000>,
> +		 <4 0x0 0x1c800000 0x00800000>,
> +		 <5 0x0 0x1c000000 0x00800000>;
> +	reg = <0x1a100000 0x1000>, <0x1a110000 0x1000>;
> +	reg-names = "ebi2", "xmem";

...

> + 	foo-ebi2@1b800000 {
> +		compatible = "foo";
> +		reg = <2 0x1b800000 0x100>;
> +		(...)
> +	};
> +};

In this example, the addresses don't match: I assume the child
node should be listed as

	foo-ebi@2,0 {
		reg = <2 0 0x100>;
	};

instead.

	Arnd
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/4 v2] soc: qcom: add EBI2 device tree bindings

[Linus Walleij]

On Mon, Aug 29, 2016 at 3:16 PM, Arnd Bergmann <arnd-r2nGTMty4D4@public.gmane.org> wrote:
> On Wednesday 24 August 2016, Linus Walleij wrote:
>> +
>> +ebi2@1a100000 {
>> +     compatible = "qcom,apq8060-ebi2";
>> +     #address-cells = <2>;
>> +     #size-cells = <1>;
>> +     ranges = <0 0x0 0x1a800000 0x00800000>,
>> +              <1 0x0 0x1b000000 0x00800000>,
>> +              <2 0x0 0x1b800000 0x00800000>,
>> +              <3 0x0 0x1d000000 0x00800000>,
>> +              <4 0x0 0x1c800000 0x00800000>,
>> +              <5 0x0 0x1c000000 0x00800000>;
>> +     reg = <0x1a100000 0x1000>, <0x1a110000 0x1000>;
>> +     reg-names = "ebi2", "xmem";
>
> ...
>
>> +     foo-ebi2@1b800000 {
>> +             compatible = "foo";
>> +             reg = <2 0x1b800000 0x100>;
>> +             (...)
>> +     };
>> +};
>
> In this example, the addresses don't match: I assume the child
> node should be listed as
>
>         foo-ebi@2,0 {
>                 reg = <2 0 0x100>;
>         };
>
> instead.

Yeah it's a plain mistake, sorry..

Noticed it too when following up in the other thread.

Yours,
Linus Walleij
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 2/4 v2] soc: qcom: add EBI2 driver

[Linus Walleij]

This adds a driver for the Qualcomm External Bus Interface EBI2
found in the MSM8660 and APQ8060 SoCs (at least).

This was tested with the SMSC9112 ethernet on the APQ8060
Dragonboard sitting on top of the SLOW CS2.

Some of my understanding if very vague and based on guesses and
extrapolations: the documentation in APQ8060 Qualcomm Application
Processor User Guide 80-N7150-14 Rev. A describes select features but
does not document the register bit fields.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
---
ChangeLog v1->v2:
- Move the driver to drivers/bus with other external bus interfaces
  after realizing that is what it is.
- Drop the default y for MACH_MSM8660, let's deal with this through
  defconfig instead.
- Support compatible strings "qcom,msm8660-ebi2" and
  "qcom,apq8060-ebi2"
- Augmented to handle a vendor prefix in fron of every XMEM
  property "qcom,xmem-foo"
- Augment bit placement for CSn enablement: CS0, CS1, CS4 and CS5
  being two bits wide
- Remove some guesswork around register assignment after feedback
  from sboyd. (Thanks!)
- Handle the errorpath by disabling clocks in a better way.
- Break out a per-CS setup function
- Walk the children and activate the CS that are actually in use
- Parse the XMEM settings from the EBI2 node array of settings
  per-chipselect in accordance with the new bindings.
- Silence a bunch of debug prints to dev_dbg() keep one line for
  CS setup because it is helpful for quick diagnosing any problems.
---
 drivers/bus/Kconfig     |   7 +
 drivers/bus/Makefile    |   1 +
 drivers/bus/qcom-ebi2.c | 404 ++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 412 insertions(+)
 create mode 100644 drivers/bus/qcom-ebi2.c

diff --git a/drivers/bus/Kconfig b/drivers/bus/Kconfig
index 3b205e212337..952eeebe0b79 100644
--- a/drivers/bus/Kconfig
+++ b/drivers/bus/Kconfig
@@ -108,6 +108,13 @@ config OMAP_OCP2SCP
 	  OCP2SCP and in OMAP5, both USB PHY and SATA PHY is connected via
 	  OCP2SCP.
 
+config QCOM_EBI2
+	bool "Qualcomm External Bus Interface 2 (EBI2)"
+	help
+	  Say y here to enable support for the Qualcomm External Bus
+	  Interface 2, which can be used to connect things like NAND Flash,
+	  SRAM, ethernet adapters, FPGAs and LCD displays.
+
 config SIMPLE_PM_BUS
 	bool "Simple Power-Managed Bus Driver"
 	depends on OF && PM
diff --git a/drivers/bus/Makefile b/drivers/bus/Makefile
index ac84cc4348e3..c6cfa6b2606e 100644
--- a/drivers/bus/Makefile
+++ b/drivers/bus/Makefile
@@ -15,6 +15,7 @@ obj-$(CONFIG_MVEBU_MBUS) 	+= mvebu-mbus.o
 obj-$(CONFIG_OMAP_INTERCONNECT)	+= omap_l3_smx.o omap_l3_noc.o
 
 obj-$(CONFIG_OMAP_OCP2SCP)	+= omap-ocp2scp.o
+obj-$(CONFIG_QCOM_EBI2)		+= qcom-ebi2.o
 obj-$(CONFIG_SUNXI_RSB)		+= sunxi-rsb.o
 obj-$(CONFIG_SIMPLE_PM_BUS)	+= simple-pm-bus.o
 obj-$(CONFIG_TEGRA_ACONNECT)	+= tegra-aconnect.o
diff --git a/drivers/bus/qcom-ebi2.c b/drivers/bus/qcom-ebi2.c
new file mode 100644
index 000000000000..41e044fc4856
--- /dev/null
+++ b/drivers/bus/qcom-ebi2.c
@@ -0,0 +1,404 @@
+/*
+ * Qualcomm External Bus Interface 2 (EBI2) driver
+ * an older version of the Qualcomm Parallel Interface Controller (QPIC)
+ *
+ * Copyright (C) 2016 Linaro Ltd.
+ *
+ * Author: Linus Walleij <linus.walleij@linaro.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2, as
+ * published by the Free Software Foundation.
+ *
+ * See the device tree bindings for this block for more details on the
+ * hardware.
+ */
+
+#include <linux/clk.h>
+#include <linux/err.h>
+#include <linux/io.h>
+#include <linux/of.h>
+#include <linux/of_platform.h>
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/slab.h>
+#include <linux/platform_device.h>
+#include <linux/bitops.h>
+
+/*
+ * CS0, CS1, CS4 and CS5 are two bits wide, CS2 and CS3 are one bit.
+ */
+#define EBI2_CS0_ENABLE_MASK BIT(0)|BIT(1)
+#define EBI2_CS1_ENABLE_MASK BIT(2)|BIT(3)
+#define EBI2_CS2_ENABLE_MASK BIT(4)
+#define EBI2_CS3_ENABLE_MASK BIT(5)
+#define EBI2_CS4_ENABLE_MASK BIT(6)|BIT(7)
+#define EBI2_CS5_ENABLE_MASK BIT(8)|BIT(9)
+#define EBI2_CSN_MASK GENMASK(9, 0)
+
+#define EBI2_XMEM_CFG 0x0000 /* Power management etc */
+
+/*
+ * SLOW CSn CFG
+ *
+ * Bits 31-28: RECOVERY recovery cycles (0 = 1, 1 = 2 etc) this is the time the
+ *             memory continues to drive the data bus after OE is de-asserted.
+ *             Inserted when reading one CS and switching to another CS or read
+ *             followed by write on the same CS. Valid values 0 thru 15.
+ * Bits 27-24: WR_HOLD write hold cycles, these are extra cycles inserted after
+ *             every write minimum 1. The data out is driven from the time WE is
+ *             asserted until CS is asserted. With a hold of 1, the CS stays
+ *             active for 1 extra cycle etc. Valid values 0 thru 15.
+ * Bits 23-16: WR_DELTA initial latency for write cycles inserted for the first
+ *             write to a page or burst memory
+ * Bits 15-8:  RD_DELTA initial latency for read cycles inserted for the first
+ *             read to a page or burst memory
+ * Bits 7-4:   WR_WAIT number of wait cycles for every write access, 0=1 cycle
+ *             so 1 thru 16 cycles.
+ * Bits 3-0:   RD_WAIT number of wait cycles for every read access, 0=1 cycle
+ *             so 1 thru 16 cycles.
+ */
+#define EBI2_XMEM_CS0_SLOW_CFG 0x0008
+#define EBI2_XMEM_CS1_SLOW_CFG 0x000C
+#define EBI2_XMEM_CS2_SLOW_CFG 0x0010
+#define EBI2_XMEM_CS3_SLOW_CFG 0x0014
+#define EBI2_XMEM_CS4_SLOW_CFG 0x0018
+#define EBI2_XMEM_CS5_SLOW_CFG 0x001C
+
+#define EBI2_XMEM_RECOVERY_SHIFT	28
+#define EBI2_XMEM_WR_HOLD_SHIFT		24
+#define EBI2_XMEM_WR_DELTA_SHIFT	16
+#define EBI2_XMEM_RD_DELTA_SHIFT	8
+#define EBI2_XMEM_WR_WAIT_SHIFT		4
+#define EBI2_XMEM_RD_WAIT_SHIFT		0
+
+/*
+ * FAST CSn CFG
+ * Bits 31-28: ?
+ * Bits 27-24: RD_HOLD: the length in cycles of the first segment of a read
+ *             transfer. For a single read trandfer this will be the time
+ *             from CS assertion to OE assertion.
+ * Bits 18-24: ?
+ * Bits 17-16: ADV_OE_RECOVERY, the number of cycles elapsed before an OE
+ *             assertion, with respect to the cycle where ADV is asserted.
+ *             2 means 2 cycles between ADV and OE. Values 0, 1, 2 or 3.
+ * Bits 5:     ADDR_HOLD_ENA, The address is held for an extra cycle to meet
+ *             hold time requirements with ADV assertion.
+ *
+ * The manual mentions "write precharge cycles" and "precharge cycles".
+ * We have not been able to figure out which bit fields these correspond to
+ * in the hardware, or what valid values exist. The current hypothesis is that
+ * this is something just used on the FAST chip selects. There is also a "byte
+ * device enable" flag somewhere for 8bit memories.
+ */
+#define EBI2_XMEM_CS0_FAST_CFG 0x0028
+#define EBI2_XMEM_CS1_FAST_CFG 0x002C
+#define EBI2_XMEM_CS2_FAST_CFG 0x0030
+#define EBI2_XMEM_CS3_FAST_CFG 0x0034
+#define EBI2_XMEM_CS4_FAST_CFG 0x0038
+#define EBI2_XMEM_CS5_FAST_CFG 0x003C
+
+#define EBI2_XMEM_RD_HOLD_SHIFT		24
+#define EBI2_XMEM_ADV_OE_RECOVERY_SHIFT	16
+#define EBI2_XMEM_ADDR_HOLD_ENA_SHIFT	5
+
+/**
+ * struct cs_data - struct with info on a chipselect setting
+ * @enable_mask: mask to enable the chipselect in the EBI2 config
+ * @slow_cfg0: offset to XMEMC slow CS config
+ * @fast_cfg1: offset to XMEMC fast CS config
+ */
+struct cs_data {
+	u32 enable_mask;
+	u16 slow_cfg;
+	u16 fast_cfg;
+};
+
+static const struct cs_data cs_info[] = {
+	{
+		/* CS0 */
+		.enable_mask = EBI2_CS0_ENABLE_MASK,
+		.slow_cfg = EBI2_XMEM_CS0_SLOW_CFG,
+		.fast_cfg = EBI2_XMEM_CS0_FAST_CFG,
+	},
+	{
+		/* CS1 */
+		.enable_mask = EBI2_CS1_ENABLE_MASK,
+		.slow_cfg = EBI2_XMEM_CS1_SLOW_CFG,
+		.fast_cfg = EBI2_XMEM_CS1_FAST_CFG,
+	},
+	{
+		/* CS2 */
+		.enable_mask = EBI2_CS2_ENABLE_MASK,
+		.slow_cfg = EBI2_XMEM_CS2_SLOW_CFG,
+		.fast_cfg = EBI2_XMEM_CS2_FAST_CFG,
+	},
+	{
+		/* CS3 */
+		.enable_mask = EBI2_CS3_ENABLE_MASK,
+		.slow_cfg = EBI2_XMEM_CS3_SLOW_CFG,
+		.fast_cfg = EBI2_XMEM_CS3_FAST_CFG,
+	},
+	{
+		/* CS4 */
+		.enable_mask = EBI2_CS4_ENABLE_MASK,
+		.slow_cfg = EBI2_XMEM_CS4_SLOW_CFG,
+		.fast_cfg = EBI2_XMEM_CS4_FAST_CFG,
+	},
+	{
+		/* CS5 */
+		.enable_mask = EBI2_CS5_ENABLE_MASK,
+		.slow_cfg = EBI2_XMEM_CS5_SLOW_CFG,
+		.fast_cfg = EBI2_XMEM_CS5_FAST_CFG,
+	},
+};
+
+/**
+ * struct ebi2_xmem_prop - describes an XMEM config property
+ * @prop: the device tree binding name
+ * @max: maximum value for the property
+ * @slowreg: true if this property is in the SLOW CS config register
+ * else it is assumed to be in the FAST config register
+ * @shift: the bit field start in the SLOW or FAST register for this
+ * property
+ */
+struct ebi2_xmem_prop {
+	const char *prop;
+	u32 max;
+	bool slowreg;
+	u16 shift;
+};
+
+static const struct ebi2_xmem_prop xmem_props[] = {
+	{
+		.prop = "qcom,xmem-recovery-cycles",
+		.max = 15,
+		.slowreg = true,
+		.shift = EBI2_XMEM_RECOVERY_SHIFT,
+	},
+	{
+		.prop = "qcom,xmem-write-hold-cycles",
+		.max = 15,
+		.slowreg = true,
+		.shift = EBI2_XMEM_WR_HOLD_SHIFT,
+	},
+	{
+		.prop = "qcom,xmem-write-delta-cycles",
+		.max = 255,
+		.slowreg = true,
+		.shift = EBI2_XMEM_WR_DELTA_SHIFT,
+	},
+	{
+		.prop = "qcom,xmem-read-delta-cycles",
+		.max = 255,
+		.slowreg = true,
+		.shift = EBI2_XMEM_RD_DELTA_SHIFT,
+	},
+	{
+		.prop = "qcom,xmem-write-wait-cycles",
+		.max = 15,
+		.slowreg = true,
+		.shift = EBI2_XMEM_WR_WAIT_SHIFT,
+	},
+	{
+		.prop = "qcom,xmem-read-wait-cycles",
+		.max = 15,
+		.slowreg = true,
+		.shift = EBI2_XMEM_RD_WAIT_SHIFT,
+	},
+	{
+		.prop = "qcom,xmem-address-hold-enable",
+		.max = 1, /* boolean prop */
+		.slowreg = false,
+		.shift = EBI2_XMEM_ADDR_HOLD_ENA_SHIFT,
+	},
+	{
+		.prop = "qcom,xmem-adv-to-oe-recovery-cycles",
+		.max = 3,
+		.slowreg = false,
+		.shift = EBI2_XMEM_ADV_OE_RECOVERY_SHIFT,
+	},
+	{
+		.prop = "qcom,xmem-read-hold-cycles",
+		.max = 15,
+		.slowreg = false,
+		.shift = EBI2_XMEM_RD_HOLD_SHIFT,
+	},
+};
+
+static void qcom_ebi2_setup_chipselect(struct device_node *np,
+				       struct device *dev,
+				       void __iomem *ebi2_base,
+				       void __iomem *ebi2_xmem,
+				       u32 csindex)
+{
+	const struct cs_data *csd;
+	u32 slowcfg, fastcfg;
+	u32 val;
+	int ret;
+	int i;
+
+	csd = &cs_info[csindex];
+	val = readl_relaxed(ebi2_base);
+	val |= csd->enable_mask;
+	writel_relaxed(val, ebi2_base);
+	dev_dbg(dev, "enabled CS%u\n", csindex);
+
+	/* Next set up the XMEMC */
+	slowcfg = 0;
+	fastcfg = 0;
+
+	for (i = 0; i < ARRAY_SIZE(xmem_props); i++) {
+		const struct ebi2_xmem_prop *xp = &xmem_props[i];
+
+		/* All are regular u32 values */
+		ret = of_property_read_u32_index(np, xp->prop, csindex, &val);
+		if (ret) {
+			dev_dbg(dev, "could not read %s for CS%d\n",
+				xp->prop, csindex);
+			continue;
+		}
+
+		/* First check boolean props */
+		if (xp->max == 1 && val) {
+			if (xp->slowreg)
+				slowcfg |= BIT(xp->shift);
+			else
+				fastcfg |= BIT(xp->shift);
+			dev_dbg(dev, "set %s flag\n", xp->prop);
+			continue;
+		}
+
+		/* We're dealing with an u32 */
+		if (val > xp->max) {
+			dev_err(dev,
+				"too high value for %s: %u, capped at %u\n",
+				xp->prop, val, xp->max);
+			val = xp->max;
+		}
+		if (xp->slowreg)
+			slowcfg |= (val << xp->shift);
+		else
+			fastcfg |= (val << xp->shift);
+		dev_dbg(dev, "set %s to %u\n", xp->prop, val);
+	}
+
+	dev_info(dev, "CS%u: SLOW CFG 0x%08x, FAST CFG 0x%08x\n",
+		 csindex, slowcfg, fastcfg);
+
+	if (slowcfg)
+		writel_relaxed(slowcfg, ebi2_xmem + csd->slow_cfg);
+	if (fastcfg)
+		writel_relaxed(fastcfg, ebi2_xmem + csd->fast_cfg);
+}
+
+static int qcom_ebi2_probe(struct platform_device *pdev)
+{
+	struct device_node *np = pdev->dev.of_node;
+	struct device_node *child;
+	struct device *dev = &pdev->dev;
+	struct resource *res;
+	void __iomem *ebi2_base;
+	void __iomem *ebi2_xmem;
+	struct clk *ebi2xclk;
+	struct clk *ebi2clk;
+	bool have_children = false;
+	u32 val;
+	int ret;
+
+	ebi2xclk = devm_clk_get(dev, "ebi2x");
+	if (IS_ERR(ebi2xclk))
+		return PTR_ERR(ebi2xclk);
+
+	ret = clk_prepare_enable(ebi2xclk);
+	if (ret) {
+		dev_err(dev, "could not enable EBI2X clk (%d)\n", ret);
+		return ret;
+	}
+
+	ebi2clk = devm_clk_get(dev, "ebi2");
+	if (IS_ERR(ebi2clk)) {
+		ret = PTR_ERR(ebi2clk);
+		goto err_disable_2x_clk;
+	}
+
+	ret = clk_prepare_enable(ebi2clk);
+	if (ret) {
+		dev_err(dev, "could not enable EBI2 clk\n");
+		goto err_disable_2x_clk;
+	}
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	ebi2_base = devm_ioremap_resource(dev, res);
+	if (IS_ERR(ebi2_base)) {
+		ret = PTR_ERR(ebi2_base);
+		goto err_disable_clk;
+	}
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 1);
+	ebi2_xmem = devm_ioremap_resource(dev, res);
+	if (IS_ERR(ebi2_xmem)) {
+		ret = PTR_ERR(ebi2_xmem);
+		goto err_disable_clk;
+	}
+
+	/* Allegedly this turns the power save mode off */
+	writel_relaxed(0UL, ebi2_xmem + EBI2_XMEM_CFG);
+
+	/* Disable all chipselects */
+	val = readl_relaxed(ebi2_base);
+	val &= ~EBI2_CSN_MASK;
+	writel_relaxed(val, ebi2_base);
+
+	/* Walk over the child nodes and see what chipselects we use */
+	for_each_available_child_of_node(np, child) {
+		u32 csindex;
+
+		/* Figure out the chipselect */
+		ret = of_property_read_u32(child, "reg", &csindex);
+		if (ret)
+			return ret;
+
+		if (csindex > 5) {
+			dev_err(dev,
+				"invalid chipselect %u, we only support 0-5\n",
+				csindex);
+			continue;
+		}
+
+		qcom_ebi2_setup_chipselect(np,
+					   dev,
+					   ebi2_base,
+					   ebi2_xmem,
+					   csindex);
+
+		/* We have at least one child */
+		have_children = true;
+	}
+
+	if (have_children)
+		return of_platform_default_populate(np, NULL, dev);
+	return 0;
+
+err_disable_clk:
+	clk_disable_unprepare(ebi2clk);
+err_disable_2x_clk:
+	clk_disable_unprepare(ebi2xclk);
+
+	return ret;
+}
+
+static const struct of_device_id qcom_ebi2_of_match[] = {
+	{ .compatible = "qcom,msm8660-ebi2", },
+	{ .compatible = "qcom,apq8060-ebi2", },
+	{ }
+};
+
+static struct platform_driver qcom_ebi2_driver = {
+	.probe = qcom_ebi2_probe,
+	.driver = {
+		.name = "qcom-ebi2",
+		.of_match_table = qcom_ebi2_of_match,
+	},
+};
+builtin_platform_driver(qcom_ebi2_driver);
-- 
2.7.4

Re: [PATCH 2/4 v2] soc: qcom: add EBI2 driver

[Arnd Bergmann]

On Wednesday 24 August 2016, Linus Walleij wrote:

> +static void qcom_ebi2_setup_chipselect(struct device_node *np,
> +				       struct device *dev,
> +				       void __iomem *ebi2_base,
> +				       void __iomem *ebi2_xmem,
> +				       u32 csindex)
> +{
> +	const struct cs_data *csd;
> +	u32 slowcfg, fastcfg;
> +	u32 val;
> +	int ret;
> +	int i;
> +
> +	csd = &cs_info[csindex];
> +	val = readl_relaxed(ebi2_base);
> +	val |= csd->enable_mask;
> +	writel_relaxed(val, ebi2_base);
> +	dev_dbg(dev, "enabled CS%u\n", csindex);

better use readl/writel instead of *_relaxed() if it's not performance critical.
If you have a function that is worth optimizing with relaxed accessors, add a
comment about how your concluded that it's safe to do so.

> +static struct platform_driver qcom_ebi2_driver = {
> +	.probe = qcom_ebi2_probe,
> +	.driver = {
> +		.name = "qcom-ebi2",
> +		.of_match_table = qcom_ebi2_of_match,
> +	},
> +};
> +builtin_platform_driver(qcom_ebi2_driver);

Why not allow this to be a loadable module?

	Arnd

Re: [PATCH 2/4 v2] soc: qcom: add EBI2 driver

[Linus Walleij]

On Mon, Aug 29, 2016 at 3:20 PM, Arnd Bergmann <arnd@arndb.de> wrote:
> On Wednesday 24 August 2016, Linus Walleij wrote:
>
>> +static void qcom_ebi2_setup_chipselect(struct device_node *np,
>> +                                    struct device *dev,
>> +                                    void __iomem *ebi2_base,
>> +                                    void __iomem *ebi2_xmem,
>> +                                    u32 csindex)
>> +{
>> +     const struct cs_data *csd;
>> +     u32 slowcfg, fastcfg;
>> +     u32 val;
>> +     int ret;
>> +     int i;
>> +
>> +     csd = &cs_info[csindex];
>> +     val = readl_relaxed(ebi2_base);
>> +     val |= csd->enable_mask;
>> +     writel_relaxed(val, ebi2_base);
>> +     dev_dbg(dev, "enabled CS%u\n", csindex);
>
> better use readl/writel instead of *_relaxed() if it's not performance critical.
> If you have a function that is worth optimizing with relaxed accessors, add a
> comment about how your concluded that it's safe to do so.

OK changing it.

>> +static struct platform_driver qcom_ebi2_driver = {
>> +     .probe = qcom_ebi2_probe,
>> +     .driver = {
>> +             .name = "qcom-ebi2",
>> +             .of_match_table = qcom_ebi2_of_match,
>> +     },
>> +};
>> +builtin_platform_driver(qcom_ebi2_driver);
>
> Why not allow this to be a loadable module?

I don't see any point in it, it's so basic to be able to access the
MMIO devices on the platform, the plan was to select the driver
from the corresponding mach-qcom SoC type (MSM8660+APQ8060).

But I can make it a module instead, if preferred.

Yours,
Linus Walleij

Re: [PATCH 2/4 v2] soc: qcom: add EBI2 driver

[Arnd Bergmann]

On Monday 29 August 2016, Linus Walleij wrote:
> >> +static struct platform_driver qcom_ebi2_driver = {
> >> +     .probe = qcom_ebi2_probe,
> >> +     .driver = {
> >> +             .name = "qcom-ebi2",
> >> +             .of_match_table = qcom_ebi2_of_match,
> >> +     },
> >> +};
> >> +builtin_platform_driver(qcom_ebi2_driver);
> >
> > Why not allow this to be a loadable module?
> 
> I don't see any point in it, it's so basic to be able to access the
> MMIO devices on the platform, the plan was to select the driver
> from the corresponding mach-qcom SoC type (MSM8660+APQ8060).
> 
> But I can make it a module instead, if preferred.

Ok, I expected this to be only for rare optional devices such as
NOR flash or the ethernet device from your example. If this is required
for booting a system, making it a module won't help much.

	Arnd

[PATCH 3/4 v2] ARM: dts: add EBI2 to the Qualcomm MSM8660 DTSI

[Linus Walleij]

This adds the external bus interface EBI2 to the MSM8660 device
tree, albeit with status = "disabled" so that devices actually
using EBI2 can turn it on if needed.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
---
ChangeLog v1->v2:
- Use the new #address-cells = <2> for indicating the CS in the
  first address cell
- Use the ranges property properly for defining the six different
  CS address windows
---
 arch/arm/boot/dts/qcom-msm8660.dtsi | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/arch/arm/boot/dts/qcom-msm8660.dtsi b/arch/arm/boot/dts/qcom-msm8660.dtsi
index 8c65e0d82559..fd18ce414347 100644
--- a/arch/arm/boot/dts/qcom-msm8660.dtsi
+++ b/arch/arm/boot/dts/qcom-msm8660.dtsi
@@ -141,6 +141,23 @@
 			};
 		};
 
+		ebi2@1a100000 {
+			compatible = "qcom,msm8660-ebi2";
+			#address-cells = <2>;
+			#size-cells = <1>;
+			ranges = <0 0x0 0x1a800000 0x00800000>,
+				 <1 0x0 0x1b000000 0x00800000>,
+				 <2 0x0 0x1b800000 0x00800000>,
+				 <3 0x0 0x1d000000 0x00800000>,
+				 <4 0x0 0x1c800000 0x00800000>,
+				 <5 0x0 0x1c000000 0x00800000>;
+			reg = <0x1a100000 0x1000>, <0x1a110000 0x1000>;
+			reg-names = "ebi2", "xmem";
+			clocks = <&gcc EBI2_2X_CLK>, <&gcc EBI2_CLK>;
+			clock-names = "ebi2x", "ebi2";
+			status = "disabled";
+		};
+
 		qcom,ssbi@500000 {
 			compatible = "qcom,ssbi";
 			reg = <0x500000 0x1000>;
-- 
2.7.4

[PATCH 4/4 v2] ARM: dts: add SMSC ethernet on the APQ8060 Dragonboard

[Linus Walleij]

The SMSC9112 ethernet controller is connected to chip select 2
on the EBI2 bus on the APQ8060 Dragonboard. We set this up by
activating EBI2, creating a chipselect entry as a subnode, and then
putting the ethernet controller in a subnode of the chipselect.

After the chipselect is configured, the SMSC device will be
instantiated.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
---
ChangeLog v1->v2:
- Use the new bindings with the first address cell indicating
  the chipselect
- Use offset zero into the range in the EBI2 node (the range
  defines the base address of the chipselect)
- Move all the XMEM setup to arrays in the EBI2 node
---
 arch/arm/boot/dts/qcom-apq8060-dragonboard.dts | 130 +++++++++++++++++++++++++
 1 file changed, 130 insertions(+)

diff --git a/arch/arm/boot/dts/qcom-apq8060-dragonboard.dts b/arch/arm/boot/dts/qcom-apq8060-dragonboard.dts
index 0abc93e5bb00..7e09a3ac0b90 100644
--- a/arch/arm/boot/dts/qcom-apq8060-dragonboard.dts
+++ b/arch/arm/boot/dts/qcom-apq8060-dragonboard.dts
@@ -51,6 +51,29 @@
 			regulator-boot-on;
 		};
 
+		/* GPIO controlled ethernet power regulator */
+		dragon_veth: xc622a331mrg {
+			compatible = "regulator-fixed";
+			regulator-name = "XC6222A331MR-G";
+			regulator-min-microvolt = <3300000>;
+			regulator-max-microvolt = <3300000>;
+			vin-supply = <&vph>;
+			gpio = <&pm8058_gpio 40 GPIO_ACTIVE_HIGH>;
+			enable-active-high;
+			pinctrl-names = "default";
+			pinctrl-0 = <&dragon_veth_gpios>;
+			regulator-always-on;
+		};
+
+		/* VDDvario fixed regulator */
+		dragon_vario: nds332p {
+			compatible = "regulator-fixed";
+			regulator-name = "NDS332P";
+			regulator-min-microvolt = <1800000>;
+			regulator-max-microvolt = <1800000>;
+			vin-supply = <&pm8058_s3>;
+		};
+
 		/* This is a levelshifter for SDCC5 */
 		dragon_vio_txb: txb0104rgyr {
 			compatible = "regulator-fixed";
@@ -167,6 +190,41 @@
 					bias-pull-up;
 				};
 			};
+
+			dragon_ebi2_pins: ebi2 {
+				/*
+				 * Pins used by EBI2 on the Dragonboard, actually only
+				 * only CS2 is used by a real peripheral. CS0 is just
+				 * routed to a test point.
+				 */
+				mux0 {
+					/*
+					 * Pins used by EBI2 on the Dragonboard, actually only
+					 * only CS2 is used by a real peripheral. CS0 is just
+					 * routed to a test point.
+					 */
+					pins =
+					    /* "gpio39", CS1A_N this is not good to mux */
+					    "gpio40", /* CS2A_N */
+					    "gpio134"; /* CS0_N testpoint TP29 */
+					function = "ebi2cs";
+				};
+				mux1 {
+					pins =
+					    /* EBI2_ADDR_7 downto EBI2_ADDR_0 address bus */
+					    "gpio123", "gpio124", "gpio125", "gpio126",
+					    "gpio127", "gpio128", "gpio129", "gpio130",
+					    /* EBI2_DATA_15 downto EBI2_DATA_0 data bus */
+					    "gpio135", "gpio136", "gpio137", "gpio138",
+					    "gpio139", "gpio140", "gpio141", "gpio142",
+					    "gpio143", "gpio144", "gpio145", "gpio146",
+					    "gpio147", "gpio148", "gpio149", "gpio150",
+					    "gpio151", /* EBI2_OE_N */
+					    "gpio153", /* EBI2_ADV */
+					    "gpio157"; /* EBI2_WE_N */
+					function = "ebi2";
+				};
+			};
 		};
 
 		qcom,ssbi@500000 {
@@ -201,6 +259,15 @@
 				};
 
 				gpio@150 {
+					dragon_ethernet_gpios: ethernet-gpios {
+						pinconf {
+							pins = "gpio7";
+							function = "normal";
+							input-enable;
+							bias-disable;
+							power-source = <PM8058_GPIO_S3>;
+						};
+					};
 					dragon_bmp085_gpios: bmp085-gpios {
 						pinconf {
 							pins = "gpio16";
@@ -238,6 +305,15 @@
 							power-source = <PM8058_GPIO_S3>;
 						};
 					};
+					dragon_veth_gpios: veth-gpios {
+						pinconf {
+							pins = "gpio40";
+							function = "normal";
+							bias-disable;
+							drive-push-pull;
+							// power-source = <PM8058_GPIO_S3>;
+						};
+					};
 				};
 			};
 		};
@@ -283,6 +359,60 @@
 			};
 		};
 
+		ebi2@1a100000 {
+			/* The EBI2 will instantiate first, then populate its children */
+			status = "ok";
+			pinctrl-names = "default";
+			pinctrl-0 = <&dragon_ebi2_pins>;
+
+			/*
+			 * SLOW chipselect config
+			 * Delay 9 cycles (140ns@64MHz) between SMSC LAN9221
+			 * Ethernet controller reads and writes on CS2.
+			 */
+			qcom,xmem-recovery-cycles = <0>, <0>, <0>,
+						  <0>, <0>, <0>;
+			qcom,xmem-write-hold-cycles = <0>, <0>, <3>,
+						    <0>, <0>, <0>;
+			qcom,xmem-write-delta-cycles = <0>, <0>, <31>,
+						     <0>, <0>, <0>;
+			qcom,xmem-read-delta-cycles = <0>, <0>, <28>,
+						    <0>, <0>, <0>;
+			qcom,xmem-write-wait-cycles = <0>, <0>, <9>,
+						    <0>, <0>, <0>;
+			qcom,xmem-read-wait-cycles = <0>, <0>, <9>,
+						   <0>, <0>, <0>;
+
+			/*
+			 * An on-board SMSC LAN9221 chip for "debug ethernet",
+			 * which is actually just an ordinary ethernet on the
+			 * EBI2. This has a 25MHz chrystal next to it, so no
+			 * clocking is needed.
+			 */
+			ethernet-ebi2@2,1b800000 {
+				compatible = "smsc,lan9221", "smsc,lan9115";
+				reg = <2 0x0 0x100>;
+				/*
+				 * GPIO7 has interrupt 198 on the PM8058
+				 * The second interrupt is the PME interrupt
+				 * for network wakeup, connected to the TLMM.
+				 */
+				interrupts-extended = <&pmicintc 198 IRQ_TYPE_EDGE_FALLING>,
+						    <&tlmm 29 IRQ_TYPE_EDGE_RISING>;
+				reset-gpios = <&tlmm 30 GPIO_ACTIVE_LOW>;
+				vdd33a-supply = <&dragon_veth>;
+				vddvario-supply = <&dragon_vario>;
+				pinctrl-names = "default";
+				pinctrl-0 = <&dragon_ethernet_gpios>;
+				phy-mode = "mii";
+				reg-io-width = <2>;
+				smsc,force-external-phy;
+				/* IRQ on edge falling = active low */
+				smsc,irq-active-low;
+				smsc,irq-push-pull;
+			};
+		};
+
 		rpm@104000 {
 			/*
 			 * Set up of the PMIC RPM regulators for this board
-- 
2.7.4

Re: [PATCH 4/4 v2] ARM: dts: add SMSC ethernet on the APQ8060 Dragonboard

[Arnd Bergmann]

On Wednesday 24 August 2016, Linus Walleij wrote:
> +                       /*
> +                        * An on-board SMSC LAN9221 chip for "debug ethernet",
> +                        * which is actually just an ordinary ethernet on the
> +                        * EBI2. This has a 25MHz chrystal next to it, so no
> +                        * clocking is needed.
> +                        */
> +                       ethernet-ebi2@2,1b800000 {
> +                               compatible = "smsc,lan9221", "smsc,lan9115";
> +                               reg = <2 0x0 0x100>;
> +                               /*

The unit address does not match the register property here.

	Arnd

Re: [PATCH 4/4 v2] ARM: dts: add SMSC ethernet on the APQ8060 Dragonboard

[Linus Walleij]

On Mon, Aug 29, 2016 at 3:21 PM, Arnd Bergmann <arnd@arndb.de> wrote:
> On Wednesday 24 August 2016, Linus Walleij wrote:
>> +                       /*
>> +                        * An on-board SMSC LAN9221 chip for "debug ethernet",
>> +                        * which is actually just an ordinary ethernet on the
>> +                        * EBI2. This has a 25MHz chrystal next to it, so no
>> +                        * clocking is needed.
>> +                        */
>> +                       ethernet-ebi2@2,1b800000 {
>> +                               compatible = "smsc,lan9221", "smsc,lan9115";
>> +                               reg = <2 0x0 0x100>;
>> +                               /*
>
> The unit address does not match the register property here.

So the convention is: ethernet-ebi2@2,0 I take it, OK fixing!
(Also in the bindings.)

Yours,
Linus Walleij

Re: [PATCH 0/4] EBI2 bus and ethernet on MSM8660/APQ8060

[Linus Walleij]

On Wed, Aug 24, 2016 at 10:52 AM, Linus Walleij
<linus.walleij@linaro.org> wrote:

> Linus Walleij (4):
>   soc: qcom: add EBI2 device tree bindings
>   soc: qcom: add EBI2 driver

Of course I forgot to update the subject to "bus: qcom: ..." on the
first two patches. Always some glitch. Oh well, bear it in mind and
review the rest and I'll respin a final v3 in due time.

Yours,
Linus Walleij