On 08/30/2013 03:44 AM, Tomasz Nowicki wrote:

W dniu 30.08.2013 02:07, al.stone@linaro.org pisze:

...

From: Al Stone ahs3@redhat.com

These are the FDT changes only.

Signed-off-by: Al Stone al.stone@linaro.org

arch/arm/boot/dts/exynos5250-arndale.dts | 26 ++++++++++++++++++++++++++ arch/arm/boot/dts/exynos5250-pinctrl.dtsi | 2 +- arch/arm/boot/dts/exynos5250.dtsi | 6 ++++-- 3 files changed, 31 insertions(+), 3 deletions(-)

diff --git a/arch/arm/boot/dts/exynos5250-arndale.dts b/arch/arm/boot/dts/exynos5250-arndale.dts index 3937993..95c08e7 100644 --- a/arch/arm/boot/dts/exynos5250-arndale.dts +++ b/arch/arm/boot/dts/exynos5250-arndale.dts @@ -485,4 +485,30 @@ rtc { status = "okay"; };

• leds {

•    compatible = "gpio-leds";
  

•    user1 {
  

•        label = "D12";
  

•        gpios = <&gpy6 5 1 0 0>;
  

•        linux,default-trigger = "heartbeat";
  

•    };
  

•    user2 {
  

•        label = "D13";
  

•        gpios = <&gpy6 6 1 0 0>;
  

•    };
  

•    user3 {
  

•        label = "D14";
  

•        gpios = <&gpy6 7 1 0 0>;
  

•    };
  

•    user4 {
  

•        label = "D15";
  

•        gpios = <&gpy6 4 1 0 0>;
  

•    };
  

• };
• };

Is it necessary to transform GPIO to ACPI ? Looks weird we add code to DTS files to enable ACPI :)

/me laughs at himself :-)

No, those are not needed. Those were part of some playing I was doing with the LEDs on the board and should not be in the patch. Sorry about that -- on the plus side, I've got some scripts that make nice patterns with the LEDs :).

W dniu 30.08.2013 02:07, al.stone@linaro.org pisze:

From: Al Stone ahs3@redhat.com

Correct logic where assumptions had been made when trying to configure
a single Samsung pinctrl device via ACPI.  Tested via enabling both
controllers 3 and 4 of the 4 available.

Signed-off-by: Al Stone al.stone@linaro.org

drivers/pinctrl/pinctrl-samsung.c | 53 +++++++++++++++++++++------------------ 1 file changed, 28 insertions(+), 25 deletions(-)

diff --git a/drivers/pinctrl/pinctrl-samsung.c b/drivers/pinctrl/pinctrl-samsung.c index 5817eb6..c5e3e62 100644 --- a/drivers/pinctrl/pinctrl-samsung.c +++ b/drivers/pinctrl/pinctrl-samsung.c @@ -830,7 +830,7 @@ static acpi_status samsung_acpi_read_banks(acpi_handle handle, u32 level, devm_kfree(drvdata->dev, bank->type); return status; }

• bank->pin_base = (u32)value;

• bank->pin_base = (u32)value + ctrl->nr_pins;

  status = acpi_evaluate_integer(handle, "NPIN", NULL, &value); if (ACPI_FAILURE(status)) {

@@ -842,6 +842,8 @@ static acpi_status samsung_acpi_read_banks(acpi_handle handle, u32 level, } bank->nr_pins = (u32)value; ctrl->nr_pins += bank->nr_pins;

• bank->gpio_chip.base = pin_base;

• pin_base += bank->nr_pins;

  bank->name = tag; bank->acpi_node = &drvdata->dev->acpi_node;

@@ -899,7 +901,6 @@ static int samsung_pin_ctrl_acpi_parse( dev_err(dev, "failure walking controller for banks\n"); return -EINVAL; }

• pin_base += ctrl->nr_pins;

  return 0; }

@@ -933,7 +934,7 @@ static acpi_status samsung_acpi_read_group(acpi_handle handle, u32 level, if (grp->num_pins == 0) break; }

• if (!grp || ii >= drvdata->nr_groups) {

• if (!grp && ii >= drvdata->nr_groups) { dev_err(dev, "too many pin groups defined for %s\n", (char *)name_buffer.pointer); return AE_BAD_PARAMETER;

@@ -1025,32 +1026,40 @@ static acpi_status samsung_acpi_read_group(acpi_handle handle, u32 level, grp->pins = pins;

Up to here patches are applying fine, later it failed. I was trying to apply it against "[Linaro-acpi] [PATCH 0/6 v2] First of the Arndale GPIO/pinctrl patches" patch set and I can't find the reason why... seem to be because of white spaces...

/* set up the functions */

• pmx = (struct samsung_pmx_func *)&drvdata->pmx_functions[0];
• for (ii = 0; ii < drvdata->nr_functions; ii++, pmx++) {

• if (!pmx->name)
  

• 	break;
  

• }
• if (!pmx || ii >= drvdata->nr_functions) {

• dev_err(dev, "too many pin functions defined for %s\n",
  

• len = strlen((char *)name_buffer.pointer);
• pname2 = devm_kzalloc(dev, len + FSUFFIX_LEN + 1, GFP_KERNEL);
• if (!pname2) {

• dev_err(dev, "cannot alloc function name space for %s\n",
  (char *)name_buffer.pointer);
  

  ACPI_FREE(buf.pointer); ACPI_FREE(info.pointer); devm_kfree(dev, pname);

• return AE_BAD_PARAMETER;
  

• return AE_NO_MEMORY;
  

  }
• memcpy(pname2, (char *)name_buffer.pointer, len);
• memcpy(pname2 + len, FUNCTION_SUFFIX, FSUFFIX_LEN);

• len = strlen(grp->name);
• pname2 = devm_kzalloc(dev, len + FSUFFIX_LEN + 1, GFP_KERNEL);
• if (!pname2) {

• dev_err(dev, "cannot alloc function name space for %s \n",
  

• pmx = (struct samsung_pmx_func *)drvdata->pmx_functions;
• for (ii = 0; ii < drvdata->nr_functions; ii++, pmx++) {

• if (pmx->name)
  

• 	if (strncmp(pmx->name, pname2, strlen(pname2)) == 0)
  

• 		break;
  

• }
• if (ii >= drvdata->nr_functions) {

• pmx = (struct samsung_pmx_func *)drvdata->pmx_functions;
  

• for (ii = 0; ii < drvdata->nr_functions; ii++, pmx++)
  

• 	if (!pmx->name)
  

• 		break;
  

• }
• if (ii >= drvdata->nr_functions) {

• dev_err(dev, "too many pin functions defined for %s\n",
  (char *)name_buffer.pointer);
  

  ACPI_FREE(buf.pointer); ACPI_FREE(info.pointer); devm_kfree(dev, pname);

• return AE_NO_MEMORY;
  

• devm_kfree(dev, pname2);
  

• return AE_BAD_PARAMETER;
  

  }

• memcpy(pname2, grp->name, len);

• memcpy(pname2 + len, FUNCTION_SUFFIX, FSUFFIX_LEN); pmx->name = pname2;

  ACPI_FREE(info.pointer);

@@ -1082,6 +1091,7 @@ static int samsung_pin_group_acpi_parse( return -EINVAL; } gcnt = (int)value;

• fcnt = gcnt;

  grps = devm_kzalloc(dev, gcnt * sizeof(*grps), GFP_KERNEL); if (!grps) {

@@ -1089,13 +1099,6 @@ static int samsung_pin_group_acpi_parse( return -EINVAL; }

• status = acpi_evaluate_integer(dev_handle, "NFUN", NULL, &value);
• if (ACPI_FAILURE(status)) {

• dev_err(dev, "cannot get NFUN value\n");
  

• return -EINVAL;
  

• }
• fcnt = (int)value;
• funcs = devm_kzalloc(dev, fcnt * sizeof(*funcs), GFP_KERNEL); if (!funcs) { devm_kfree(dev, grps);

Well done!

Acked-by: Tomasz Nowicki tomasz.nowicki@linaro.org