Received-SPF: pass (google.com: domain of linux-kernel+bounces-120380-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.48.161 as permitted sender) client-ip=147.75.48.161;
From: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Date: Wed, 27 Mar 2024 12:05:47 +0530
Subject: [PATCH v11 1/8] PCI: dwc: ep: Fix DBI access failure for drivers
 requiring refclk from host
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit
Message-Id: <20240327-pci-dbi-rework-v11-1-6f5259f90673@linaro.org>
References: <20240327-pci-dbi-rework-v11-0-6f5259f90673@linaro.org>
In-Reply-To: <20240327-pci-dbi-rework-v11-0-6f5259f90673@linaro.org>
To: Jingoo Han <jingoohan1@gmail.com>, 
 Gustavo Pimentel <gustavo.pimentel@synopsys.com>, 
 Lorenzo Pieralisi <lpieralisi@kernel.org>, 
 =?utf-8?q?Krzysztof_Wilczy=C5=84ski?= <kw@linux.com>, 
 Rob Herring <robh@kernel.org>, Bjorn Helgaas <bhelgaas@google.com>, 
 Marek Vasut <marek.vasut+renesas@gmail.com>, 
 Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>, 
 Thierry Reding <thierry.reding@gmail.com>, 
 Jonathan Hunter <jonathanh@nvidia.com>, 
 Kishon Vijay Abraham I <kishon@ti.com>, Vidya Sagar <vidyas@nvidia.com>, 
 Vignesh Raghavendra <vigneshr@ti.com>, Richard Zhu <hongxing.zhu@nxp.com>, 
 Lucas Stach <l.stach@pengutronix.de>, Shawn Guo <shawnguo@kernel.org>, 
 Sascha Hauer <s.hauer@pengutronix.de>, 
 Pengutronix Kernel Team <kernel@pengutronix.de>, 
 Fabio Estevam <festevam@gmail.com>, NXP Linux Team <linux-imx@nxp.com>, 
 Minghuan Lian <minghuan.Lian@nxp.com>, Mingkai Hu <mingkai.hu@nxp.com>, 
 Roy Zang <roy.zang@nxp.com>, 
 Kunihiko Hayashi <hayashi.kunihiko@socionext.com>, 
 Masami Hiramatsu <mhiramat@kernel.org>, 
 Kishon Vijay Abraham I <kishon@kernel.org>, 
 Jesper Nilsson <jesper.nilsson@axis.com>, 
 Srikanth Thokala <srikanth.thokala@intel.com>, 
 Shawn Lin <shawn.lin@rock-chips.com>, Heiko Stuebner <heiko@sntech.de>
Cc: linux-pci@vger.kernel.org, linux-kernel@vger.kernel.org, 
 linux-renesas-soc@vger.kernel.org, linux-arm-msm@vger.kernel.org, 
 linux-tegra@vger.kernel.org, linux-omap@vger.kernel.org, 
 linux-arm-kernel@lists.infradead.org, linuxppc-dev@lists.ozlabs.org, 
 Niklas Cassel <cassel@kernel.org>, linux-arm-kernel@axis.com, 
 linux-rockchip@lists.infradead.org, 
 Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>, 
 Frank Li <Frank.Li@nxp.com>

The DWC glue drivers requiring an active reference clock from the PCIe host
for initializing their PCIe EP core, set a flag called 'core_init_notifier'
to let DWC driver know that these drivers need a special attention during
initialization. In these drivers, access to the hw registers (like DBI)
before receiving the active refclk from host will result in access failure
and also could cause a whole system hang.

But the current DWC EP driver doesn't honor the requirements of the drivers
setting 'core_init_notifier' flag and tries to access the DBI registers
during dw_pcie_ep_init(). This causes the system hang for glue drivers such
as Tegra194 and Qcom EP as they depend on refclk from host and have set the
above mentioned flag.

To workaround this issue, users of the affected platforms have to maintain
the dependency with the PCIe host by booting the PCIe EP after host boot.
But this won't provide a good user experience, since PCIe EP is _one_ of
the features of those platforms and it doesn't make sense to delay the
whole platform booting due to PCIe requiring active refclk.

So to fix this issue, let's move all the DBI access from
dw_pcie_ep_init() in the DWC EP driver to the dw_pcie_ep_init_complete()
API. This API will only be called by the drivers setting
'core_init_notifier' flag once refclk is received from host. For the rest
of the drivers that gets the refclk locally, this API will be called
within dw_pcie_ep_init().

Fixes: e966f7390da9 ("PCI: dwc: Refactor core initialization code for EP mode")
Co-developed-by: Vidya Sagar <vidyas@nvidia.com>
Signed-off-by: Vidya Sagar <vidyas@nvidia.com>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Niklas Cassel <cassel@kernel.org>
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
---
 drivers/pci/controller/dwc/pcie-designware-ep.c | 120 ++++++++++++++----------
 1 file changed, 71 insertions(+), 49 deletions(-)

diff --git a/drivers/pci/controller/dwc/pcie-designware-ep.c b/drivers/pci/controller/dwc/pcie-designware-ep.c
index 5befed2dc02b..c05304eabb89 100644
--- a/drivers/pci/controller/dwc/pcie-designware-ep.c
+++ b/drivers/pci/controller/dwc/pcie-designware-ep.c
@@ -602,11 +602,16 @@ static unsigned int dw_pcie_ep_find_ext_capability(struct dw_pcie *pci, int cap)
 int dw_pcie_ep_init_complete(struct dw_pcie_ep *ep)
 {
 	struct dw_pcie *pci = to_dw_pcie_from_ep(ep);
+	struct dw_pcie_ep_func *ep_func;
+	struct device *dev = pci->dev;
+	struct pci_epc *epc = ep->epc;
 	unsigned int offset, ptm_cap_base;
 	unsigned int nbars;
 	u8 hdr_type;
+	u8 func_no;
+	int i, ret;
+	void *addr;
 	u32 reg;
-	int i;
 
 	hdr_type = dw_pcie_readb_dbi(pci, PCI_HEADER_TYPE) &
 		   PCI_HEADER_TYPE_MASK;
@@ -617,6 +622,58 @@ int dw_pcie_ep_init_complete(struct dw_pcie_ep *ep)
 		return -EIO;
 	}
 
+	dw_pcie_version_detect(pci);
+
+	dw_pcie_iatu_detect(pci);
+
+	ret = dw_pcie_edma_detect(pci);
+	if (ret)
+		return ret;
+
+	if (!ep->ib_window_map) {
+		ep->ib_window_map = devm_bitmap_zalloc(dev, pci->num_ib_windows,
+						       GFP_KERNEL);
+		if (!ep->ib_window_map)
+			goto err_remove_edma;
+	}
+
+	if (!ep->ob_window_map) {
+		ep->ob_window_map = devm_bitmap_zalloc(dev, pci->num_ob_windows,
+						       GFP_KERNEL);
+		if (!ep->ob_window_map)
+			goto err_remove_edma;
+	}
+
+	if (!ep->outbound_addr) {
+		addr = devm_kcalloc(dev, pci->num_ob_windows, sizeof(phys_addr_t),
+				    GFP_KERNEL);
+		if (!addr)
+			goto err_remove_edma;
+		ep->outbound_addr = addr;
+	}
+
+	for (func_no = 0; func_no < epc->max_functions; func_no++) {
+
+		ep_func = dw_pcie_ep_get_func_from_ep(ep, func_no);
+		if (ep_func)
+			continue;
+
+		ep_func = devm_kzalloc(dev, sizeof(*ep_func), GFP_KERNEL);
+		if (!ep_func)
+			goto err_remove_edma;
+
+		ep_func->func_no = func_no;
+		ep_func->msi_cap = dw_pcie_ep_find_capability(ep, func_no,
+							      PCI_CAP_ID_MSI);
+		ep_func->msix_cap = dw_pcie_ep_find_capability(ep, func_no,
+							       PCI_CAP_ID_MSIX);
+
+		list_add_tail(&ep_func->list, &ep->func_list);
+	}
+
+	if (ep->ops->init)
+		ep->ops->init(ep);
+
 	offset = dw_pcie_ep_find_ext_capability(pci, PCI_EXT_CAP_ID_REBAR);
 	ptm_cap_base = dw_pcie_ep_find_ext_capability(pci, PCI_EXT_CAP_ID_PTM);
 
@@ -651,14 +708,17 @@ int dw_pcie_ep_init_complete(struct dw_pcie_ep *ep)
 	dw_pcie_dbi_ro_wr_dis(pci);
 
 	return 0;
+
+err_remove_edma:
+	dw_pcie_edma_remove(pci);
+
+	return ret;
 }
 EXPORT_SYMBOL_GPL(dw_pcie_ep_init_complete);
 
 int dw_pcie_ep_init(struct dw_pcie_ep *ep)
 {
 	int ret;
-	void *addr;
-	u8 func_no;
 	struct resource *res;
 	struct pci_epc *epc;
 	struct dw_pcie *pci = to_dw_pcie_from_ep(ep);
@@ -666,7 +726,6 @@ int dw_pcie_ep_init(struct dw_pcie_ep *ep)
 	struct platform_device *pdev = to_platform_device(dev);
 	struct device_node *np = dev->of_node;
 	const struct pci_epc_features *epc_features;
-	struct dw_pcie_ep_func *ep_func;
 
 	INIT_LIST_HEAD(&ep->func_list);
 
@@ -684,26 +743,6 @@ int dw_pcie_ep_init(struct dw_pcie_ep *ep)
 	if (ep->ops->pre_init)
 		ep->ops->pre_init(ep);
 
-	dw_pcie_version_detect(pci);
-
-	dw_pcie_iatu_detect(pci);
-
-	ep->ib_window_map = devm_bitmap_zalloc(dev, pci->num_ib_windows,
-					       GFP_KERNEL);
-	if (!ep->ib_window_map)
-		return -ENOMEM;
-
-	ep->ob_window_map = devm_bitmap_zalloc(dev, pci->num_ob_windows,
-					       GFP_KERNEL);
-	if (!ep->ob_window_map)
-		return -ENOMEM;
-
-	addr = devm_kcalloc(dev, pci->num_ob_windows, sizeof(phys_addr_t),
-			    GFP_KERNEL);
-	if (!addr)
-		return -ENOMEM;
-	ep->outbound_addr = addr;
-
 	epc = devm_pci_epc_create(dev, &epc_ops);
 	if (IS_ERR(epc)) {
 		dev_err(dev, "Failed to create epc device\n");
@@ -717,23 +756,6 @@ int dw_pcie_ep_init(struct dw_pcie_ep *ep)
 	if (ret < 0)
 		epc->max_functions = 1;
 
-	for (func_no = 0; func_no < epc->max_functions; func_no++) {
-		ep_func = devm_kzalloc(dev, sizeof(*ep_func), GFP_KERNEL);
-		if (!ep_func)
-			return -ENOMEM;
-
-		ep_func->func_no = func_no;
-		ep_func->msi_cap = dw_pcie_ep_find_capability(ep, func_no,
-							      PCI_CAP_ID_MSI);
-		ep_func->msix_cap = dw_pcie_ep_find_capability(ep, func_no,
-							       PCI_CAP_ID_MSIX);
-
-		list_add_tail(&ep_func->list, &ep->func_list);
-	}
-
-	if (ep->ops->init)
-		ep->ops->init(ep);
-
 	ret = pci_epc_mem_init(epc, ep->phys_base, ep->addr_size,
 			       ep->page_size);
 	if (ret < 0) {
@@ -749,25 +771,25 @@ int dw_pcie_ep_init(struct dw_pcie_ep *ep)
 		goto err_exit_epc_mem;
 	}
 
-	ret = dw_pcie_edma_detect(pci);
-	if (ret)
-		goto err_free_epc_mem;
-
 	if (ep->ops->get_features) {
 		epc_features = ep->ops->get_features(ep);
 		if (epc_features->core_init_notifier)
 			return 0;
 	}
 
+	/*
+	 * NOTE:- Avoid accessing the hardware (Ex:- DBI space) before this
+	 * step as platforms that implement 'core_init_notifier' feature may
+	 * not have the hardware ready (i.e. core initialized) for access
+	 * (Ex: tegra194). Any hardware access on such platforms result
+	 * in system hang.
+	 */
 	ret = dw_pcie_ep_init_complete(ep);
 	if (ret)
-		goto err_remove_edma;
+		goto err_free_epc_mem;
 
 	return 0;
 
-err_remove_edma:
-	dw_pcie_edma_remove(pci);
-
 err_free_epc_mem:
 	pci_epc_mem_free_addr(epc, ep->msi_mem_phys, ep->msi_mem,
 			      epc->mem->window.page_size);

-- 
2.25.1