On Mon, Aug 2, 2021 at 2:10 PM Bartosz Golaszewski
[off-list ref] wrote:

On Sat, Jul 17, 2021 at 12:17 AM Robert Marko [off-list ref] wrote:

quoted

Delta TN4810M uses a similar CPLD GPIO expander
like the TN48M, but it has pins for 48 SFP+ ports,
making a total of 192 pins.
It also provides the TX fault pins which the TN48M
does not.

Only TX disable pins like on the TN48M are output
ones.

Thankfully, regmap GPIO allows for the driver to be
easily extended to support the TN4810M.

Note that this patch depends on the following series:
https://patchwork.ozlabs.org/project/linux-gpio/list/?series=247538

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
---
 drivers/gpio/gpio-tn48m.c | 56 ++++++++++++++++++++++++++++++++++++---
 1 file changed, 52 insertions(+), 4 deletions(-)

diff --git a/drivers/gpio/gpio-tn48m.c b/drivers/gpio/gpio-tn48m.c
index b12a6b4bc4b3..e429e7ade941 100644
--- a/drivers/gpio/gpio-tn48m.c
+++ b/drivers/gpio/gpio-tn48m.c

@@ -19,6 +19,10 @@ enum tn48m_gpio_type {
        TN48M_SFP_TX_DISABLE = 1,
        TN48M_SFP_PRESENT,
        TN48M_SFP_LOS,
+       TN4810M_SFP_TX_DISABLE,
+       TN4810M_SFP_TX_FAULT,
+       TN4810M_SFP_PRESENT,
+       TN4810M_SFP_LOS,
 };

 static int tn48m_gpio_probe(struct platform_device *pdev)

@@ -46,17 +50,36 @@ static int tn48m_gpio_probe(struct platform_device *pdev)

        config.regmap = regmap;
        config.parent = &pdev->dev;
-       config.ngpio = 4;
+       config.ngpio_per_reg = 8;

        switch (type) {
        case TN48M_SFP_TX_DISABLE:
                config.reg_set_base = base;
+               config.ngpio = 4;
                break;
        case TN48M_SFP_PRESENT:
                config.reg_dat_base = base;
+               config.ngpio = 4;
                break;
        case TN48M_SFP_LOS:
                config.reg_dat_base = base;
+               config.ngpio = 4;
+               break;
+       case TN4810M_SFP_TX_DISABLE:
+               config.reg_set_base = base;
+               config.ngpio = 48;
+               break;
+       case TN4810M_SFP_TX_FAULT:
+               config.reg_dat_base = base;
+               config.ngpio = 48;
+               break;
+       case TN4810M_SFP_PRESENT:
+               config.reg_dat_base = base;
+               config.ngpio = 48;
+               break;
+       case TN4810M_SFP_LOS:
+               config.reg_dat_base = base;
+               config.ngpio = 48;
                break;
        default:
                dev_err(&pdev->dev, "unknown type %d\n", type);

@@ -67,9 +90,34 @@ static int tn48m_gpio_probe(struct platform_device *pdev)
 }

 static const struct of_device_id tn48m_gpio_of_match[] = {
-       { .compatible = "delta,tn48m-gpio-sfp-tx-disable", .data = (void *)TN48M_SFP_TX_DISABLE },
-       { .compatible = "delta,tn48m-gpio-sfp-present", .data = (void *)TN48M_SFP_PRESENT },
-       { .compatible = "delta,tn48m-gpio-sfp-los", .data = (void *)TN48M_SFP_LOS },
+       {
+               .compatible = "delta,tn48m-gpio-sfp-tx-disable",
+               .data = (void *)TN48M_SFP_TX_DISABLE
+       },
+       {
+               .compatible = "delta,tn48m-gpio-sfp-present",
+               .data = (void *)TN48M_SFP_PRESENT
+       },
+       {
+               .compatible = "delta,tn48m-gpio-sfp-los",
+               .data = (void *)TN48M_SFP_LOS
+       },
+       {
+               .compatible = "delta,tn4810m-gpio-sfp-tx-disable",
+               .data = (void *)TN4810M_SFP_TX_DISABLE
+       },
+       {
+               .compatible = "delta,tn4810m-gpio-sfp-tx-fault",
+               .data = (void *)TN4810M_SFP_TX_FAULT
+       },
+       {
+               .compatible = "delta,tn4810m-gpio-sfp-present",
+               .data = (void *)TN4810M_SFP_PRESENT
+       },
+       {
+               .compatible = "delta,tn4810m-gpio-sfp-los",
+               .data = (void *)TN4810M_SFP_LOS
+       },
        { }
 };
 MODULE_DEVICE_TABLE(of, tn48m_gpio_of_match);
--

2.31.1

This looks good to me. I suppose the other patches are going in
through the MFD tree. I don't see anything that can fail here at
build-time - can you confirm that I can pick these patches up
separately for v5.15?

Bartosz

Scratch that, I now saw Linus' comment about the special purpose pins
under the other series. Let's clear that up first.

Bart