Re: [PATCH V3 2/2] ata: ahci: ceva: Update the driver to support xilinx GT phy
From: Michal Simek <hidden>
Date: 2021-03-03 17:11:36
Also in:
linux-ide, lkml
On 2/8/21 7:03 PM, Piyush Mehta wrote:
quoted hunk ↗ jump to hunk
SATA controller used in Xilinx ZynqMP platform uses xilinx GT phy which has 4 GT lanes and can be used by 4 peripherals at a time. SATA controller uses 1 GT phy lane among the 4 GT lanes. To configure the GT lane for the SATA controller, the below sequence is expected. 1. Assert the SATA controller reset. 2. Configure the xilinx GT phy lane for SATA controller (phy_init). 3. De-assert the SATA controller reset. 4. Wait for PLL of the GT lane used by SATA to be locked (phy_power_on). The ahci_platform_enable_resources() by default does the phy_init() and phy_power_on() but the default sequence doesn't work with Xilinx platforms. Because of this reason, updated the driver to support the new sequence. Added cevapriv->rst check, for backward compatibility with the older sequence. If the reset controller is not available, then the SATA controller will configure with the older sequences. Signed-off-by: Piyush Mehta <redacted> --- drivers/ata/ahci_ceva.c | 43 ++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 40 insertions(+), 3 deletions(-)diff --git a/drivers/ata/ahci_ceva.c b/drivers/ata/ahci_ceva.c index b10fd4c..b980218 100644 --- a/drivers/ata/ahci_ceva.c +++ b/drivers/ata/ahci_ceva.c@@ -12,6 +12,7 @@ #include <linux/module.h> #include <linux/of_device.h> #include <linux/platform_device.h> +#include <linux/reset.h> #include "ahci.h" /* Vendor Specific Register Offsets */@@ -87,6 +88,7 @@ struct ceva_ahci_priv { u32 axicc; bool is_cci_enabled; int flags; + struct reset_control *rst; }; static unsigned int ceva_ahci_read_id(struct ata_device *dev,@@ -202,13 +204,48 @@ static int ceva_ahci_probe(struct platform_device *pdev) cevapriv->ahci_pdev = pdev; + cevapriv->rst = devm_reset_control_get_optional_exclusive(&pdev->dev, + NULL); + if (IS_ERR(cevapriv->rst)) { + if (PTR_ERR(cevapriv->rst) != -EPROBE_DEFER) + dev_err(&pdev->dev, "failed to get reset: %ld\n", + PTR_ERR(cevapriv->rst)); + }
nit: This can be handled via dev_err_probe() to simplify this logic here.
It was added by a787e5400a1c ("driver core: add device probe log
helper") but up to Jens if he wants this to be fixed.
+
hpriv = ahci_platform_get_resources(pdev, 0);
if (IS_ERR(hpriv))
return PTR_ERR(hpriv);
- rc = ahci_platform_enable_resources(hpriv);
- if (rc)
- return rc;
+ if (!cevapriv->rst) {
+ rc = ahci_platform_enable_resources(hpriv);
+ if (rc)
+ return rc;
+ } else {
+ int i;
+
+ rc = ahci_platform_enable_clks(hpriv);
+ if (rc)
+ return rc;
+ /* Assert the controller reset */
+ reset_control_assert(cevapriv->rst);
+
+ for (i = 0; i < hpriv->nports; i++) {
+ rc = phy_init(hpriv->phys[i]);
+ if (rc)
+ return rc;
+ }
+
+ /* De-assert the controller reset */
+ reset_control_deassert(cevapriv->rst);
+
+ for (i = 0; i < hpriv->nports; i++) {
+ rc = phy_power_on(hpriv->phys[i]);
+ if (rc) {
+ phy_exit(hpriv->phys[i]);
+ return rc;
+ }
+ }
+ }
if (of_property_read_bool(np, "ceva,broken-gen2"))
cevapriv->flags = CEVA_FLAG_BROKEN_GEN2;Acked-by: Michal Simek <redacted> Jens: Can you please take a look at this patch if you see any other issue? Thanks, Michal