Received-SPF: pass (google.com: domain of linux-kernel+bounces-170215-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
From: Sean Anderson <sean.anderson@linux.dev>
To: Laurent Pinchart <laurent.pinchart@ideasonboard.com>,
	linux-phy@lists.infradead.org
Cc: Vinod Koul <vkoul@kernel.org>,
	linux-arm-kernel@lists.infradead.org,
	linux-kernel@vger.kernel.org,
	Michal Simek <michal.simek@amd.com>,
	Kishon Vijay Abraham I <kishon@kernel.org>,
	Sean Anderson <sean.anderson@linux.dev>
Subject: [PATCH v2 3/4] phy: zynqmp: Only wait for PLL lock "primary" instances
Date: Mon,  6 May 2024 13:01:09 -0400
Message-Id: <20240506170110.2874724-4-sean.anderson@linux.dev>
In-Reply-To: <20240506170110.2874724-1-sean.anderson@linux.dev>
References: <20240506170110.2874724-1-sean.anderson@linux.dev>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit

For PCIe and DisplayPort, the phy instance represents the controller's
logical lane. Wait for the instance 0 phy's PLL to lock as other
instances will never lock. We do this in xpsgtr_wait_pll_lock so callers
don't have to determine the correct lane themselves.

The original comment is wrong about cumulative wait times. Since we are
just polling a bit, all subsequent waiters will finish immediately.

Signed-off-by: Sean Anderson <sean.anderson@linux.dev>
---

Changes in v2:
- Move the logic for waiting on PLL lock to xpsgtr_wait_pll_lock

 drivers/phy/xilinx/phy-zynqmp.c | 30 +++++++++++++++++++++---------
 1 file changed, 21 insertions(+), 9 deletions(-)

diff --git a/drivers/phy/xilinx/phy-zynqmp.c b/drivers/phy/xilinx/phy-zynqmp.c
index 5a8f81bbeb8d..b86fe2a249a8 100644
--- a/drivers/phy/xilinx/phy-zynqmp.c
+++ b/drivers/phy/xilinx/phy-zynqmp.c
@@ -294,10 +294,30 @@ static int xpsgtr_wait_pll_lock(struct phy *phy)
 	struct xpsgtr_phy *gtr_phy = phy_get_drvdata(phy);
 	struct xpsgtr_dev *gtr_dev = gtr_phy->dev;
 	unsigned int timeout = TIMEOUT_US;
+	u8 protocol = gtr_phy->protocol;
 	int ret;
 
 	dev_dbg(gtr_dev->dev, "Waiting for PLL lock\n");
 
+	/*
+	 * For DP and PCIe, only the instance 0 PLL is used. Switch to that phy
+	 * so we wait on the right PLL.
+	 */
+	if ((protocol == ICM_PROTOCOL_DP || protocol == ICM_PROTOCOL_PCIE) &&
+	    gtr_phy->instance) {
+		int i;
+
+		for (i = 0; i < NUM_LANES; i++) {
+			gtr_phy = &gtr_dev->phys[i];
+
+			if (gtr_phy->protocol == protocol && !gtr_phy->instance)
+				goto got_phy;
+		}
+
+		return -EBUSY;
+	}
+
+got_phy:
 	while (1) {
 		u32 reg = xpsgtr_read_phy(gtr_phy, L0_PLL_STATUS_READ_1);
 
@@ -627,15 +647,7 @@ static int xpsgtr_phy_power_on(struct phy *phy)
 	/* Skip initialization if not required. */
 	if (!xpsgtr_phy_init_required(gtr_phy))
 		return ret;
-	/*
-	 * Wait for the PLL to lock. For DP, only wait on DP0 to avoid
-	 * cumulating waits for both lanes. The user is expected to initialize
-	 * lane 0 last.
-	 */
-	if (gtr_phy->protocol != ICM_PROTOCOL_DP || !gtr_phy->instance)
-		ret = xpsgtr_wait_pll_lock(phy);
-
-	return ret;
+	return xpsgtr_wait_pll_lock(phy);
 }
 
 static int xpsgtr_phy_configure(struct phy *phy, union phy_configure_opts *opts)
-- 
2.35.1.1320.gc452695387.dirty