summaryrefslogtreecommitdiffstats
path: root/target/linux/lantiq/files/arch/mips/pci/pcie-lantiq.c
diff options
context:
space:
mode:
Diffstat (limited to 'target/linux/lantiq/files/arch/mips/pci/pcie-lantiq.c')
-rw-r--r--target/linux/lantiq/files/arch/mips/pci/pcie-lantiq.c1146
1 files changed, 1146 insertions, 0 deletions
diff --git a/target/linux/lantiq/files/arch/mips/pci/pcie-lantiq.c b/target/linux/lantiq/files/arch/mips/pci/pcie-lantiq.c
new file mode 100644
index 000000000..1df55b5c0
--- /dev/null
+++ b/target/linux/lantiq/files/arch/mips/pci/pcie-lantiq.c
@@ -0,0 +1,1146 @@
+#include <linux/types.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/delay.h>
+#include <linux/mm.h>
+#include <asm/paccess.h>
+#include <linux/pci.h>
+#include <linux/pci_regs.h>
+#include <linux/platform_device.h>
+
+#define CONFIG_IFX_PCIE_1ST_CORE
+
+#include "pcie-lantiq.h"
+
+#define IFX_PCIE_IR (INT_NUM_IM4_IRL0 + 25)
+#define IFX_PCIE_INTA (INT_NUM_IM4_IRL0 + 8)
+#define IFX_PCIE_INTB (INT_NUM_IM4_IRL0 + 9)
+#define IFX_PCIE_INTC (INT_NUM_IM4_IRL0 + 10)
+#define IFX_PCIE_INTD (INT_NUM_IM4_IRL0 + 11)
+#define MS(_v, _f) (((_v) & (_f)) >> _f##_S)
+#define SM(_v, _f) (((_v) << _f##_S) & (_f))
+#define IFX_REG_SET_BIT(_f, _r) \
+ IFX_REG_W32((IFX_REG_R32((_r)) &~ (_f)) | (_f), (_r))
+#define IFX_PCIE_LTSSM_ENABLE_TIMEOUT 10
+#define IFX_PCIE_PHY_LINK_UP_TIMEOUT 1000
+#define IFX_PCIE_PHY_LOOP_CNT 5
+
+static DEFINE_SPINLOCK(ifx_pcie_lock);
+
+int pcibios_1st_host_bus_nr(void);
+
+unsigned int g_pcie_debug_flag = PCIE_MSG_ANY & (~PCIE_MSG_CFG);
+
+static ifx_pcie_irq_t pcie_irqs[IFX_PCIE_CORE_NR] = {
+ {
+ .ir_irq = {
+ .irq = IFX_PCIE_IR,
+ .name = "ifx_pcie_rc0",
+ },
+
+ .legacy_irq = {
+ {
+ .irq_bit = PCIE_IRN_INTA,
+ .irq = IFX_PCIE_INTA,
+ },
+ {
+ .irq_bit = PCIE_IRN_INTB,
+ .irq = IFX_PCIE_INTB,
+ },
+ {
+ .irq_bit = PCIE_IRN_INTC,
+ .irq = IFX_PCIE_INTC,
+ },
+ {
+ .irq_bit = PCIE_IRN_INTD,
+ .irq = IFX_PCIE_INTD,
+ },
+ },
+ },
+};
+
+static inline int pcie_ltssm_enable(int pcie_port)
+{
+ int i;
+
+ IFX_REG_W32(PCIE_RC_CCR_LTSSM_ENABLE, PCIE_RC_CCR(pcie_port)); /* Enable LTSSM */
+
+ /* Wait for the link to come up */
+ for (i = 0; i < IFX_PCIE_LTSSM_ENABLE_TIMEOUT; i++) {
+ if (!(IFX_REG_R32(PCIE_LCTLSTS(pcie_port)) & PCIE_LCTLSTS_RETRAIN_PENDING)) {
+ break;
+ }
+ udelay(10);
+ }
+ if (i >= IFX_PCIE_LTSSM_ENABLE_TIMEOUT) {
+ IFX_PCIE_PRINT(PCIE_MSG_INIT, "%s link timeout!!!!!\n", __func__);
+ return -1;
+ }
+ return 0;
+}
+
+static inline void pcie_status_register_clear(int pcie_port)
+{
+ IFX_REG_W32(0, PCIE_RC_DR(pcie_port));
+ IFX_REG_W32(0, PCIE_PCICMDSTS(pcie_port));
+ IFX_REG_W32(0, PCIE_DCTLSTS(pcie_port));
+ IFX_REG_W32(0, PCIE_LCTLSTS(pcie_port));
+ IFX_REG_W32(0, PCIE_SLCTLSTS(pcie_port));
+ IFX_REG_W32(0, PCIE_RSTS(pcie_port));
+ IFX_REG_W32(0, PCIE_UES_R(pcie_port));
+ IFX_REG_W32(0, PCIE_UEMR(pcie_port));
+ IFX_REG_W32(0, PCIE_UESR(pcie_port));
+ IFX_REG_W32(0, PCIE_CESR(pcie_port));
+ IFX_REG_W32(0, PCIE_CEMR(pcie_port));
+ IFX_REG_W32(0, PCIE_RESR(pcie_port));
+ IFX_REG_W32(0, PCIE_PVCCRSR(pcie_port));
+ IFX_REG_W32(0, PCIE_VC0_RSR0(pcie_port));
+ IFX_REG_W32(0, PCIE_TPFCS(pcie_port));
+ IFX_REG_W32(0, PCIE_TNPFCS(pcie_port));
+ IFX_REG_W32(0, PCIE_TCFCS(pcie_port));
+ IFX_REG_W32(0, PCIE_QSR(pcie_port));
+ IFX_REG_W32(0, PCIE_IOBLSECS(pcie_port));
+}
+
+static inline int ifx_pcie_link_up(int pcie_port)
+{
+ return (IFX_REG_R32(PCIE_PHY_SR(pcie_port)) & PCIE_PHY_SR_PHY_LINK_UP) ? 1 : 0;
+}
+
+static inline void pcie_mem_io_setup(int pcie_port)
+{
+ unsigned int reg;
+ /*
+ * BAR[0:1] readonly register
+ * RC contains only minimal BARs for packets mapped to this device
+ * Mem/IO filters defines a range of memory occupied by memory mapped IO devices that
+ * reside on the downstream side fo the bridge.
+ */
+ reg = SM((PCIE_MEM_PHY_PORT_TO_END(pcie_port) >> 20), PCIE_MBML_MEM_LIMIT_ADDR)
+ | SM((PCIE_MEM_PHY_PORT_TO_BASE(pcie_port) >> 20), PCIE_MBML_MEM_BASE_ADDR);
+ IFX_REG_W32(reg, PCIE_MBML(pcie_port));
+
+ /* PCIe_PBML, same as MBML */
+ IFX_REG_W32(IFX_REG_R32(PCIE_MBML(pcie_port)), PCIE_PMBL(pcie_port));
+
+ /* IO Address Range */
+ reg = SM((PCIE_IO_PHY_PORT_TO_END(pcie_port) >> 12), PCIE_IOBLSECS_IO_LIMIT_ADDR)
+ | SM((PCIE_IO_PHY_PORT_TO_BASE(pcie_port) >> 12), PCIE_IOBLSECS_IO_BASE_ADDR);
+ reg |= PCIE_IOBLSECS_32BIT_IO_ADDR;
+ IFX_REG_W32(reg, PCIE_IOBLSECS(pcie_port));
+
+ reg = SM((PCIE_IO_PHY_PORT_TO_END(pcie_port) >> 16), PCIE_IO_BANDL_UPPER_16BIT_IO_LIMIT)
+ | SM((PCIE_IO_PHY_PORT_TO_BASE(pcie_port) >> 16), PCIE_IO_BANDL_UPPER_16BIT_IO_BASE);
+ IFX_REG_W32(reg, PCIE_IO_BANDL(pcie_port));
+}
+
+static inline void pcie_msi_setup(int pcie_port)
+{
+ unsigned int reg;
+
+ /* XXX, MSI stuff should only apply to EP */
+ /* MSI Capability: Only enable 32-bit addresses */
+ reg = IFX_REG_R32(PCIE_MCAPR(pcie_port));
+ reg &= ~PCIE_MCAPR_ADDR64_CAP;
+ reg |= PCIE_MCAPR_MSI_ENABLE;
+
+ /* Disable multiple message */
+ reg &= ~(PCIE_MCAPR_MULTI_MSG_CAP | PCIE_MCAPR_MULTI_MSG_ENABLE);
+ IFX_REG_W32(reg, PCIE_MCAPR(pcie_port));
+}
+
+static inline void pcie_pm_setup(int pcie_port)
+{
+ unsigned int reg;
+
+ /* Enable PME, Soft reset enabled */
+ reg = IFX_REG_R32(PCIE_PM_CSR(pcie_port));
+ reg |= PCIE_PM_CSR_PME_ENABLE | PCIE_PM_CSR_SW_RST;
+ IFX_REG_W32(reg, PCIE_PM_CSR(pcie_port));
+}
+
+static inline void pcie_bus_setup(int pcie_port)
+{
+ unsigned int reg;
+
+ reg = SM(0, PCIE_BNR_PRIMARY_BUS_NUM) | SM(1, PCIE_PNR_SECONDARY_BUS_NUM) | SM(0xFF, PCIE_PNR_SUB_BUS_NUM);
+ IFX_REG_W32(reg, PCIE_BNR(pcie_port));
+}
+
+static inline void pcie_device_setup(int pcie_port)
+{
+ unsigned int reg;
+
+ /* Device capability register, set up Maximum payload size */
+ reg = IFX_REG_R32(PCIE_DCAP(pcie_port));
+ reg |= PCIE_DCAP_ROLE_BASE_ERR_REPORT;
+ reg |= SM(PCIE_MAX_PAYLOAD_128, PCIE_DCAP_MAX_PAYLOAD_SIZE);
+
+ /* Only available for EP */
+ reg &= ~(PCIE_DCAP_EP_L0S_LATENCY | PCIE_DCAP_EP_L1_LATENCY);
+ IFX_REG_W32(reg, PCIE_DCAP(pcie_port));
+
+ /* Device control and status register */
+ /* Set Maximum Read Request size for the device as a Requestor */
+ reg = IFX_REG_R32(PCIE_DCTLSTS(pcie_port));
+
+ /*
+ * Request size can be larger than the MPS used, but the completions returned
+ * for the read will be bounded by the MPS size.
+ * In our system, Max request size depends on AHB burst size. It is 64 bytes.
+ * but we set it as 128 as minimum one.
+ */
+ reg |= SM(PCIE_MAX_PAYLOAD_128, PCIE_DCTLSTS_MAX_READ_SIZE)
+ | SM(PCIE_MAX_PAYLOAD_128, PCIE_DCTLSTS_MAX_PAYLOAD_SIZE);
+
+ /* Enable relaxed ordering, no snoop, and all kinds of errors */
+ reg |= PCIE_DCTLSTS_RELAXED_ORDERING_EN | PCIE_DCTLSTS_ERR_EN | PCIE_DCTLSTS_NO_SNOOP_EN;
+
+ IFX_REG_W32(reg, PCIE_DCTLSTS(pcie_port));
+}
+
+static inline void pcie_link_setup(int pcie_port)
+{
+ unsigned int reg;
+
+ /*
+ * XXX, Link capability register, bit 18 for EP CLKREQ# dynamic clock management for L1, L2/3 CPM
+ * L0s is reported during link training via TS1 order set by N_FTS
+ */
+ reg = IFX_REG_R32(PCIE_LCAP(pcie_port));
+ reg &= ~PCIE_LCAP_L0S_EIXT_LATENCY;
+ reg |= SM(3, PCIE_LCAP_L0S_EIXT_LATENCY);
+ IFX_REG_W32(reg, PCIE_LCAP(pcie_port));
+
+ /* Link control and status register */
+ reg = IFX_REG_R32(PCIE_LCTLSTS(pcie_port));
+
+ /* Link Enable, ASPM enabled */
+ reg &= ~PCIE_LCTLSTS_LINK_DISABLE;
+
+#ifdef CONFIG_PCIEASPM
+ /*
+ * We use the same physical reference clock that the platform provides on the connector
+ * It paved the way for ASPM to calculate the new exit Latency
+ */
+ reg |= PCIE_LCTLSTS_SLOT_CLK_CFG;
+ reg |= PCIE_LCTLSTS_COM_CLK_CFG;
+ /*
+ * We should disable ASPM by default except that we have dedicated power management support
+ * Enable ASPM will cause the system hangup/instability, performance degration
+ */
+ reg |= PCIE_LCTLSTS_ASPM_ENABLE;
+#else
+ reg &= ~PCIE_LCTLSTS_ASPM_ENABLE;
+#endif /* CONFIG_PCIEASPM */
+
+ /*
+ * The maximum size of any completion with data packet is bounded by the MPS setting
+ * in device control register
+ */
+ /* RCB may cause multiple split transactions, two options available, we use 64 byte RCB */
+ reg &= ~ PCIE_LCTLSTS_RCB128;
+ IFX_REG_W32(reg, PCIE_LCTLSTS(pcie_port));
+}
+
+static inline void pcie_error_setup(int pcie_port)
+{
+ unsigned int reg;
+
+ /*
+ * Forward ERR_COR, ERR_NONFATAL, ERR_FATAL to the backbone
+ * Poisoned write TLPs and completions indicating poisoned TLPs will set the PCIe_PCICMDSTS.MDPE
+ */
+ reg = IFX_REG_R32(PCIE_INTRBCTRL(pcie_port));
+ reg |= PCIE_INTRBCTRL_SERR_ENABLE | PCIE_INTRBCTRL_PARITY_ERR_RESP_ENABLE;
+
+ IFX_REG_W32(reg, PCIE_INTRBCTRL(pcie_port));
+
+ /* Uncorrectable Error Mask Register, Unmask <enable> all bits in PCIE_UESR */
+ reg = IFX_REG_R32(PCIE_UEMR(pcie_port));
+ reg &= ~PCIE_ALL_UNCORRECTABLE_ERR;
+ IFX_REG_W32(reg, PCIE_UEMR(pcie_port));
+
+ /* Uncorrectable Error Severity Register, ALL errors are FATAL */
+ IFX_REG_W32(PCIE_ALL_UNCORRECTABLE_ERR, PCIE_UESR(pcie_port));
+
+ /* Correctable Error Mask Register, unmask <enable> all bits */
+ reg = IFX_REG_R32(PCIE_CEMR(pcie_port));
+ reg &= ~PCIE_CORRECTABLE_ERR;
+ IFX_REG_W32(reg, PCIE_CEMR(pcie_port));
+
+ /* Advanced Error Capabilities and Control Registr */
+ reg = IFX_REG_R32(PCIE_AECCR(pcie_port));
+ reg |= PCIE_AECCR_ECRC_CHECK_EN | PCIE_AECCR_ECRC_GEN_EN;
+ IFX_REG_W32(reg, PCIE_AECCR(pcie_port));
+
+ /* Root Error Command Register, Report all types of errors */
+ reg = IFX_REG_R32(PCIE_RECR(pcie_port));
+ reg |= PCIE_RECR_ERR_REPORT_EN;
+ IFX_REG_W32(reg, PCIE_RECR(pcie_port));
+
+ /* Clear the Root status register */
+ reg = IFX_REG_R32(PCIE_RESR(pcie_port));
+ IFX_REG_W32(reg, PCIE_RESR(pcie_port));
+}
+
+static inline void pcie_root_setup(int pcie_port)
+{
+ unsigned int reg;
+
+ /* Root control and capabilities register */
+ reg = IFX_REG_R32(PCIE_RCTLCAP(pcie_port));
+ reg |= PCIE_RCTLCAP_SERR_ENABLE | PCIE_RCTLCAP_PME_INT_EN;
+ IFX_REG_W32(reg, PCIE_RCTLCAP(pcie_port));
+}
+
+static inline void pcie_vc_setup(int pcie_port)
+{
+ unsigned int reg;
+
+ /* Port VC Capability Register 2 */
+ reg = IFX_REG_R32(PCIE_PVC2(pcie_port));
+ reg &= ~PCIE_PVC2_VC_ARB_WRR;
+ reg |= PCIE_PVC2_VC_ARB_16P_FIXED_WRR;
+ IFX_REG_W32(reg, PCIE_PVC2(pcie_port));
+
+ /* VC0 Resource Capability Register */
+ reg = IFX_REG_R32(PCIE_VC0_RC(pcie_port));
+ reg &= ~PCIE_VC0_RC_REJECT_SNOOP;
+ IFX_REG_W32(reg, PCIE_VC0_RC(pcie_port));
+}
+
+static inline void pcie_port_logic_setup(int pcie_port)
+{
+ unsigned int reg;
+
+ /* FTS number, default 12, increase to 63, may increase time from/to L0s to L0 */
+ reg = IFX_REG_R32(PCIE_AFR(pcie_port));
+ reg &= ~(PCIE_AFR_FTS_NUM | PCIE_AFR_COM_FTS_NUM);
+ reg |= SM(PCIE_AFR_FTS_NUM_DEFAULT, PCIE_AFR_FTS_NUM)
+ | SM(PCIE_AFR_FTS_NUM_DEFAULT, PCIE_AFR_COM_FTS_NUM);
+ /* L0s and L1 entry latency */
+ reg &= ~(PCIE_AFR_L0S_ENTRY_LATENCY | PCIE_AFR_L1_ENTRY_LATENCY);
+ reg |= SM(PCIE_AFR_L0S_ENTRY_LATENCY_DEFAULT, PCIE_AFR_L0S_ENTRY_LATENCY)
+ | SM(PCIE_AFR_L1_ENTRY_LATENCY_DEFAULT, PCIE_AFR_L1_ENTRY_LATENCY);
+ IFX_REG_W32(reg, PCIE_AFR(pcie_port));
+
+ /* Port Link Control Register */
+ reg = IFX_REG_R32(PCIE_PLCR(pcie_port));
+ reg |= PCIE_PLCR_DLL_LINK_EN; /* Enable the DLL link */
+ IFX_REG_W32(reg, PCIE_PLCR(pcie_port));
+
+ /* Lane Skew Register */
+ reg = IFX_REG_R32(PCIE_LSR(pcie_port));
+ /* Enable ACK/NACK and FC */
+ reg &= ~(PCIE_LSR_ACKNAK_DISABLE | PCIE_LSR_FC_DISABLE);
+ IFX_REG_W32(reg, PCIE_LSR(pcie_port));
+
+ /* Symbol Timer Register and Filter Mask Register 1 */
+ reg = IFX_REG_R32(PCIE_STRFMR(pcie_port));
+
+ /* Default SKP interval is very accurate already, 5us */
+ /* Enable IO/CFG transaction */
+ reg |= PCIE_STRFMR_RX_CFG_TRANS_ENABLE | PCIE_STRFMR_RX_IO_TRANS_ENABLE;
+ /* Disable FC WDT */
+ reg &= ~PCIE_STRFMR_FC_WDT_DISABLE;
+ IFX_REG_W32(reg, PCIE_STRFMR(pcie_port));
+
+ /* Filter Masker Register 2 */
+ reg = IFX_REG_R32(PCIE_FMR2(pcie_port));
+ reg |= PCIE_FMR2_VENDOR_MSG1_PASSED_TO_TRGT1 | PCIE_FMR2_VENDOR_MSG0_PASSED_TO_TRGT1;
+ IFX_REG_W32(reg, PCIE_FMR2(pcie_port));
+
+ /* VC0 Completion Receive Queue Control Register */
+ reg = IFX_REG_R32(PCIE_VC0_CRQCR(pcie_port));
+ reg &= ~PCIE_VC0_CRQCR_CPL_TLP_QUEUE_MODE;
+ reg |= SM(PCIE_VC0_TLP_QUEUE_MODE_BYPASS, PCIE_VC0_CRQCR_CPL_TLP_QUEUE_MODE);
+ IFX_REG_W32(reg, PCIE_VC0_CRQCR(pcie_port));
+}
+
+static inline void pcie_rc_cfg_reg_setup(int pcie_port)
+{
+ /* diable ltssm */
+ IFX_REG_W32(0, PCIE_RC_CCR(pcie_port));
+
+ pcie_mem_io_setup(pcie_port);
+ pcie_msi_setup(pcie_port);
+ pcie_pm_setup(pcie_port);
+ pcie_bus_setup(pcie_port);
+ pcie_device_setup(pcie_port);
+ pcie_link_setup(pcie_port);
+ pcie_error_setup(pcie_port);
+ pcie_root_setup(pcie_port);
+ pcie_vc_setup(pcie_port);
+ pcie_port_logic_setup(pcie_port);
+}
+
+static int ifx_pcie_wait_phy_link_up(int pcie_port)
+{
+ int i;
+
+ /* Wait for PHY link is up */
+ for (i = 0; i < IFX_PCIE_PHY_LINK_UP_TIMEOUT; i++) {
+ if (ifx_pcie_link_up(pcie_port)) {
+ break;
+ }
+ udelay(100);
+ }
+ if (i >= IFX_PCIE_PHY_LINK_UP_TIMEOUT) {
+ printk(KERN_ERR "%s timeout\n", __func__);
+ return -1;
+ }
+
+ /* Check data link up or not */
+ if (!(IFX_REG_R32(PCIE_RC_DR(pcie_port)) & PCIE_RC_DR_DLL_UP)) {
+ printk(KERN_ERR "%s DLL link is still down\n", __func__);
+ return -1;
+ }
+
+ /* Check Data link active or not */
+ if (!(IFX_REG_R32(PCIE_LCTLSTS(pcie_port)) & PCIE_LCTLSTS_DLL_ACTIVE)) {
+ printk(KERN_ERR "%s DLL is not active\n", __func__);
+ return -1;
+ }
+ return 0;
+}
+
+static inline int pcie_app_loigc_setup(int pcie_port)
+{
+ IFX_REG_W32(PCIE_AHB_CTRL_BUS_ERROR_SUPPRESS, PCIE_AHB_CTRL(pcie_port));
+
+ /* Pull PCIe EP out of reset */
+ pcie_device_rst_deassert(pcie_port);
+
+ /* Start LTSSM training between RC and EP */
+ pcie_ltssm_enable(pcie_port);
+
+ /* Check PHY status after enabling LTSSM */
+ if (ifx_pcie_wait_phy_link_up(pcie_port) != 0) {
+ return -1;
+ }
+ return 0;
+}
+
+/*
+ * Must be done after ltssm due to based on negotiated link
+ * width and payload size
+ * Update the Replay Time Limit. Empirically, some PCIe
+ * devices take a little longer to respond than expected under
+ * load. As a workaround for this we configure the Replay Time
+ * Limit to the value expected for a 512 byte MPS instead of
+ * our actual 128 byte MPS. The numbers below are directly
+ * from the PCIe spec table 3-4/5.
+ */
+static inline void pcie_replay_time_update(int pcie_port)
+{
+ unsigned int reg;
+ int nlw;
+ int rtl;
+
+ reg = IFX_REG_R32(PCIE_LCTLSTS(pcie_port));
+
+ nlw = MS(reg, PCIE_LCTLSTS_NEGOTIATED_LINK_WIDTH);
+ switch (nlw) {
+ case PCIE_MAX_LENGTH_WIDTH_X1:
+ rtl = 1677;
+ break;
+ case PCIE_MAX_LENGTH_WIDTH_X2:
+ rtl = 867;
+ break;
+ case PCIE_MAX_LENGTH_WIDTH_X4:
+ rtl = 462;
+ break;
+ case PCIE_MAX_LENGTH_WIDTH_X8:
+ rtl = 258;
+ break;
+ default:
+ rtl = 1677;
+ break;
+ }
+ reg = IFX_REG_R32(PCIE_ALTRT(pcie_port));
+ reg &= ~PCIE_ALTRT_REPLAY_TIME_LIMIT;
+ reg |= SM(rtl, PCIE_ALTRT_REPLAY_TIME_LIMIT);
+ IFX_REG_W32(reg, PCIE_ALTRT(pcie_port));
+
+ IFX_PCIE_PRINT(PCIE_MSG_REG, "%s PCIE_ALTRT 0x%08x\n",
+ __func__, IFX_REG_R32(PCIE_ALTRT(pcie_port)));
+}
+
+/*
+ * Table 359 Enhanced Configuration Address Mapping1)
+ * 1) This table is defined in Table 7-1, page 341, PCI Express Base Specification v1.1
+ * Memory Address PCI Express Configuration Space
+ * A[(20+n-1):20] Bus Number 1 < n < 8
+ * A[19:15] Device Number
+ * A[14:12] Function Number
+ * A[11:8] Extended Register Number
+ * A[7:2] Register Number
+ * A[1:0] Along with size of the access, used to generate Byte Enables
+ * For VR9, only the address bits [22:0] are mapped to the configuration space:
+ * . Address bits [22:20] select the target bus (1-of-8)1)
+ * . Address bits [19:15] select the target device (1-of-32) on the bus
+ * . Address bits [14:12] select the target function (1-of-8) within the device.
+ * . Address bits [11:2] selects the target dword (1-of-1024) within the selected function.s configuration space
+ * . Address bits [1:0] define the start byte location within the selected dword.
+ */
+static inline unsigned int pcie_bus_addr(u8 bus_num, u16 devfn, int where)
+{
+ unsigned int addr;
+ u8 bus;
+
+ if (!bus_num) {
+ /* type 0 */
+ addr = ((PCI_SLOT(devfn) & 0x1F) << 15) | ((PCI_FUNC(devfn) & 0x7) << 12) | ((where & 0xFFF)& ~3);
+ } else {
+ bus = bus_num;
+ /* type 1, only support 8 buses */
+ addr = ((bus & 0x7) << 20) | ((PCI_SLOT(devfn) & 0x1F) << 15) |
+ ((PCI_FUNC(devfn) & 0x7) << 12) | ((where & 0xFFF) & ~3);
+ }
+ IFX_PCIE_PRINT(PCIE_MSG_CFG, "%s: bus addr : %02x:%02x.%01x/%02x, addr=%08x\n",
+ __func__, bus_num, PCI_SLOT(devfn), PCI_FUNC(devfn), where, addr);
+ return addr;
+}
+
+static int pcie_valid_config(int pcie_port, int bus, int dev)
+{
+ /* RC itself */
+ if ((bus == 0) && (dev == 0))
+ return 1;
+
+ /* No physical link */
+ if (!ifx_pcie_link_up(pcie_port))
+ return 0;
+
+ /* Bus zero only has RC itself
+ * XXX, check if EP will be integrated
+ */
+ if ((bus == 0) && (dev != 0))
+ return 0;
+
+ /* Maximum 8 buses supported for VRX */
+ if (bus > 9)
+ return 0;
+
+ /*
+ * PCIe is PtP link, one bus only supports only one device
+ * except bus zero and PCIe switch which is virtual bus device
+ * The following two conditions really depends on the system design
+ * and attached the device.
+ * XXX, how about more new switch
+ */
+ if ((bus == 1) && (dev != 0))
+ return 0;
+
+ if ((bus >= 3) && (dev != 0))
+ return 0;
+ return 1;
+}
+
+static inline unsigned int ifx_pcie_cfg_rd(int pcie_port, unsigned int reg)
+{
+ return IFX_REG_R32((volatile unsigned int *)(PCIE_CFG_PORT_TO_BASE(pcie_port) + reg));
+}
+
+static inline void ifx_pcie_cfg_wr(int pcie_port, unsigned int reg, unsigned int val)
+{
+ IFX_REG_W32( val, (volatile unsigned int *)(PCIE_CFG_PORT_TO_BASE(pcie_port) + reg));
+}
+
+static inline unsigned int ifx_pcie_rc_cfg_rd(int pcie_port, unsigned int reg)
+{
+ return IFX_REG_R32((volatile unsigned int *)(PCIE_RC_PORT_TO_BASE(pcie_port) + reg));
+}
+
+static inline void ifx_pcie_rc_cfg_wr(int pcie_port, unsigned int reg, unsigned int val)
+{
+ IFX_REG_W32(val, (volatile unsigned int *)(PCIE_RC_PORT_TO_BASE(pcie_port) + reg));
+}
+
+unsigned int ifx_pcie_bus_enum_read_hack(int where, unsigned int value)
+{
+ unsigned int tvalue = value;
+
+ if (where == PCI_PRIMARY_BUS) {
+ u8 primary, secondary, subordinate;
+
+ primary = tvalue & 0xFF;
+ secondary = (tvalue >> 8) & 0xFF;
+ subordinate = (tvalue >> 16) & 0xFF;
+ primary += pcibios_1st_host_bus_nr();
+ secondary += pcibios_1st_host_bus_nr();
+ subordinate += pcibios_1st_host_bus_nr();
+ tvalue = (tvalue & 0xFF000000) | (unsigned int)primary | (unsigned int)(secondary << 8) | (unsigned int)(subordinate << 16);
+ }
+ return tvalue;
+}
+
+unsigned int ifx_pcie_bus_enum_write_hack(int where, unsigned int value)
+{
+ unsigned int tvalue = value;
+
+ if (where == PCI_PRIMARY_BUS) {
+ u8 primary, secondary, subordinate;
+
+ primary = tvalue & 0xFF;
+ secondary = (tvalue >> 8) & 0xFF;
+ subordinate = (tvalue >> 16) & 0xFF;
+ if (primary > 0 && primary != 0xFF)
+ primary -= pcibios_1st_host_bus_nr();
+ if (secondary > 0 && secondary != 0xFF)
+ secondary -= pcibios_1st_host_bus_nr();
+ if (subordinate > 0 && subordinate != 0xFF)
+ subordinate -= pcibios_1st_host_bus_nr();
+ tvalue = (tvalue & 0xFF000000) | (unsigned int)primary | (unsigned int)(secondary << 8) | (unsigned int)(subordinate << 16);
+ } else if (where == PCI_SUBORDINATE_BUS) {
+ u8 subordinate = tvalue & 0xFF;
+ subordinate = subordinate > 0 ? subordinate - pcibios_1st_host_bus_nr() : 0;
+ tvalue = subordinate;
+ }
+ return tvalue;
+}
+
+/**
+ * \fn static int ifx_pcie_read_config(struct pci_bus *bus, unsigned int devfn,
+ * int where, int size, unsigned int *value)
+ * \brief Read a value from configuration space
+ *
+ * \param[in] bus Pointer to pci bus
+ * \param[in] devfn PCI device function number
+ * \param[in] where PCI register number
+ * \param[in] size Register read size
+ * \param[out] value Pointer to return value
+ * \return PCIBIOS_BAD_REGISTER_NUMBER Invalid register number
+ * \return PCIBIOS_FUNC_NOT_SUPPORTED PCI function not supported
+ * \return PCIBIOS_DEVICE_NOT_FOUND PCI device not found
+ * \return PCIBIOS_SUCCESSFUL OK
+ * \ingroup IFX_PCIE_OS
+ */
+static int ifx_pcie_read_config(struct pci_bus *bus, unsigned int devfn, int where, int size, unsigned int *value)
+{
+ unsigned int data = 0;
+ int bus_number = bus->number;
+ static const unsigned int mask[8] = {0, 0xff, 0xffff, 0, 0xffffffff, 0, 0, 0};
+ int ret = PCIBIOS_SUCCESSFUL;
+ struct ifx_pci_controller *ctrl = bus->sysdata;
+ int pcie_port = ctrl->port;
+
+ if (unlikely(size != 1 && size != 2 && size != 4)){
+ ret = PCIBIOS_BAD_REGISTER_NUMBER;
+ goto out;
+ }
+
+ /* Make sure the address is aligned to natural boundary */
+ if (unlikely(((size - 1) & where))) {
+ ret = PCIBIOS_BAD_REGISTER_NUMBER;
+ goto out;
+ }
+
+ /*
+ * If we are second controller, we have to cheat OS so that it assume
+ * its bus number starts from 0 in host controller
+ */
+ bus_number = ifx_pcie_bus_nr_deduct(bus_number, pcie_port);
+
+ /*
+ * We need to force the bus number to be zero on the root
+ * bus. Linux numbers the 2nd root bus to start after all
+ * busses on root 0.
+ */
+ if (bus->parent == NULL)
+ bus_number = 0;
+
+ /*
+ * PCIe only has a single device connected to it. It is
+ * always device ID 0. Don't bother doing reads for other
+ * device IDs on the first segment.
+ */
+ if ((bus_number == 0) && (PCI_SLOT(devfn) != 0)) {
+ ret = PCIBIOS_FUNC_NOT_SUPPORTED;
+ goto out;
+ }
+
+ if (pcie_valid_config(pcie_port, bus_number, PCI_SLOT(devfn)) == 0) {
+ *value = 0xffffffff;
+ ret = PCIBIOS_DEVICE_NOT_FOUND;
+ goto out;
+ }
+
+ IFX_PCIE_PRINT(PCIE_MSG_READ_CFG, "%s: %02x:%02x.%01x/%02x:%01d\n", __func__, bus_number,
+ PCI_SLOT(devfn), PCI_FUNC(devfn), where, size);
+
+ PCIE_IRQ_LOCK(ifx_pcie_lock);
+ if (bus_number == 0) { /* RC itself */
+ unsigned int t;
+
+ t = (where & ~3);
+ data = ifx_pcie_rc_cfg_rd(pcie_port, t);
+ IFX_PCIE_PRINT(PCIE_MSG_READ_CFG, "%s: rd local cfg, offset:%08x, data:%08x\n",
+ __func__, t, data);
+ } else {
+ unsigned int addr = pcie_bus_addr(bus_number, devfn, where);
+
+ data = ifx_pcie_cfg_rd(pcie_port, addr);
+ if (pcie_port == IFX_PCIE_PORT0) {
+#ifdef CONFIG_IFX_PCIE_HW_SWAP
+ data = le32_to_cpu(data);
+#endif /* CONFIG_IFX_PCIE_HW_SWAP */
+ } else {
+#ifdef CONFIG_IFX_PCIE1_HW_SWAP
+ data = le32_to_cpu(data);
+#endif /* CONFIG_IFX_PCIE_HW_SWAP */
+ }
+ }
+ /* To get a correct PCI topology, we have to restore the bus number to OS */
+ data = ifx_pcie_bus_enum_hack(bus, devfn, where, data, pcie_port, 1);
+
+ PCIE_IRQ_UNLOCK(ifx_pcie_lock);
+ IFX_PCIE_PRINT(PCIE_MSG_READ_CFG, "%s: read config: data=%08x raw=%08x\n",
+ __func__, (data >> (8 * (where & 3))) & mask[size & 7], data);
+
+ *value = (data >> (8 * (where & 3))) & mask[size & 7];
+out:
+ return ret;
+}
+
+static unsigned int ifx_pcie_size_to_value(int where, int size, unsigned int data, unsigned int value)
+{
+ unsigned int shift;
+ unsigned int tdata = data;
+
+ switch (size) {
+ case 1:
+ shift = (where & 0x3) << 3;
+ tdata &= ~(0xffU << shift);
+ tdata |= ((value & 0xffU) << shift);
+ break;
+ case 2:
+ shift = (where & 3) << 3;
+ tdata &= ~(0xffffU << shift);
+ tdata |= ((value & 0xffffU) << shift);
+ break;
+ case 4:
+ tdata = value;
+ break;
+ }
+ return tdata;
+}
+
+/**
+ * \fn static static int ifx_pcie_write_config(struct pci_bus *bus, unsigned int devfn,
+ * int where, int size, unsigned int value)
+ * \brief Write a value to PCI configuration space
+ *
+ * \param[in] bus Pointer to pci bus
+ * \param[in] devfn PCI device function number
+ * \param[in] where PCI register number
+ * \param[in] size The register size to be written
+ * \param[in] value The valule to be written
+ * \return PCIBIOS_BAD_REGISTER_NUMBER Invalid register number
+ * \return PCIBIOS_DEVICE_NOT_FOUND PCI device not found
+ * \return PCIBIOS_SUCCESSFUL OK
+ * \ingroup IFX_PCIE_OS
+ */
+static int ifx_pcie_write_config(struct pci_bus *bus, unsigned int devfn, int where, int size, unsigned int value)
+{
+ int bus_number = bus->number;
+ int ret = PCIBIOS_SUCCESSFUL;
+ struct ifx_pci_controller *ctrl = bus->sysdata;
+ int pcie_port = ctrl->port;
+ unsigned int tvalue = value;
+ unsigned int data;
+
+ /* Make sure the address is aligned to natural boundary */
+ if (unlikely(((size - 1) & where))) {
+ ret = PCIBIOS_BAD_REGISTER_NUMBER;
+ goto out;
+ }
+ /*
+ * If we are second controller, we have to cheat OS so that it assume
+ * its bus number starts from 0 in host controller
+ */
+ bus_number = ifx_pcie_bus_nr_deduct(bus_number, pcie_port);
+
+ /*
+ * We need to force the bus number to be zero on the root
+ * bus. Linux numbers the 2nd root bus to start after all
+ * busses on root 0.
+ */
+ if (bus->parent == NULL)
+ bus_number = 0;
+
+ if (pcie_valid_config(pcie_port, bus_number, PCI_SLOT(devfn)) == 0) {
+ ret = PCIBIOS_DEVICE_NOT_FOUND;
+ goto out;
+ }
+
+ IFX_PCIE_PRINT(PCIE_MSG_WRITE_CFG, "%s: %02x:%02x.%01x/%02x:%01d value=%08x\n", __func__,
+ bus_number, PCI_SLOT(devfn), PCI_FUNC(devfn), where, size, value);
+
+ /* XXX, some PCIe device may need some delay */
+ PCIE_IRQ_LOCK(ifx_pcie_lock);
+
+ /*
+ * To configure the correct bus topology using native way, we have to cheat Os so that
+ * it can configure the PCIe hardware correctly.
+ */
+ tvalue = ifx_pcie_bus_enum_hack(bus, devfn, where, value, pcie_port, 0);
+
+ if (bus_number == 0) { /* RC itself */
+ unsigned int t;
+
+ t = (where & ~3);
+ IFX_PCIE_PRINT(PCIE_MSG_WRITE_CFG,"%s: wr local cfg, offset:%08x, fill:%08x\n", __func__, t, value);
+ data = ifx_pcie_rc_cfg_rd(pcie_port, t);
+ IFX_PCIE_PRINT(PCIE_MSG_WRITE_CFG,"%s: rd local cfg, offset:%08x, data:%08x\n", __func__, t, data);
+
+ data = ifx_pcie_size_to_value(where, size, data, tvalue);
+
+ IFX_PCIE_PRINT(PCIE_MSG_WRITE_CFG,"%s: wr local cfg, offset:%08x, value:%08x\n", __func__, t, data);
+ ifx_pcie_rc_cfg_wr(pcie_port, t, data);
+ IFX_PCIE_PRINT(PCIE_MSG_WRITE_CFG,"%s: rd local cfg, offset:%08x, value:%08x\n",
+ __func__, t, ifx_pcie_rc_cfg_rd(pcie_port, t));
+ } else {
+ unsigned int addr = pcie_bus_addr(bus_number, devfn, where);
+
+ IFX_PCIE_PRINT(PCIE_MSG_WRITE_CFG,"%s: wr cfg, offset:%08x, fill:%08x\n", __func__, addr, value);
+ data = ifx_pcie_cfg_rd(pcie_port, addr);
+ if (pcie_port == IFX_PCIE_PORT0) {
+#ifdef CONFIG_IFX_PCIE_HW_SWAP
+ data = le32_to_cpu(data);
+#endif /* CONFIG_IFX_PCIE_HW_SWAP */
+ } else {
+#ifdef CONFIG_IFX_PCIE1_HW_SWAP
+ data = le32_to_cpu(data);
+#endif /* CONFIG_IFX_PCIE_HW_SWAP */
+ }
+ IFX_PCIE_PRINT(PCIE_MSG_WRITE_CFG,"%s: rd cfg, offset:%08x, data:%08x\n", __func__, addr, data);
+
+ data = ifx_pcie_size_to_value(where, size, data, tvalue);
+ if (pcie_port == IFX_PCIE_PORT0) {
+#ifdef CONFIG_IFX_PCIE_HW_SWAP
+ data = cpu_to_le32(data);
+#endif /* CONFIG_IFX_PCIE_HW_SWAP */
+ } else {
+#ifdef CONFIG_IFX_PCIE1_HW_SWAP
+ data = cpu_to_le32(data);
+#endif /* CONFIG_IFX_PCIE_HW_SWAP */
+ }
+ IFX_PCIE_PRINT(PCIE_MSG_WRITE_CFG, "%s: wr cfg, offset:%08x, value:%08x\n", __func__, addr, data);
+ ifx_pcie_cfg_wr(pcie_port, addr, data);
+ IFX_PCIE_PRINT(PCIE_MSG_WRITE_CFG, "%s: rd cfg, offset:%08x, value:%08x\n",
+ __func__, addr, ifx_pcie_cfg_rd(pcie_port, addr));
+ }
+ PCIE_IRQ_UNLOCK(ifx_pcie_lock);
+out:
+ return ret;
+}
+
+static struct resource ifx_pcie_io_resource = {
+ .name = "PCIe0 I/O space",
+ .start = PCIE_IO_PHY_BASE,
+ .end = PCIE_IO_PHY_END,
+ .flags = IORESOURCE_IO,
+};
+
+static struct resource ifx_pcie_mem_resource = {
+ .name = "PCIe0 Memory space",
+ .start = PCIE_MEM_PHY_BASE,
+ .end = PCIE_MEM_PHY_END,
+ .flags = IORESOURCE_MEM,
+};
+
+static struct pci_ops ifx_pcie_ops = {
+ .read = ifx_pcie_read_config,
+ .write = ifx_pcie_write_config,
+};
+
+static struct ifx_pci_controller ifx_pcie_controller[IFX_PCIE_CORE_NR] = {
+ {
+ .pcic = {
+ .pci_ops = &ifx_pcie_ops,
+ .mem_resource = &ifx_pcie_mem_resource,
+ .io_resource = &ifx_pcie_io_resource,
+ },
+ .port = IFX_PCIE_PORT0,
+ },
+};
+
+static inline void pcie_core_int_clear_all(int pcie_port)
+{
+ unsigned int reg;
+ reg = IFX_REG_R32(PCIE_IRNCR(pcie_port));
+ reg &= PCIE_RC_CORE_COMBINED_INT;
+ IFX_REG_W32(reg, PCIE_IRNCR(pcie_port));
+}
+
+static irqreturn_t pcie_rc_core_isr(int irq, void *dev_id)
+{
+ struct ifx_pci_controller *ctrl = (struct ifx_pci_controller *)dev_id;
+ int pcie_port = ctrl->port;
+
+ IFX_PCIE_PRINT(PCIE_MSG_ISR, "PCIe RC error intr %d\n", irq);
+ pcie_core_int_clear_all(pcie_port);
+ return IRQ_HANDLED;
+}
+
+static int pcie_rc_core_int_init(int pcie_port)
+{
+ int ret;
+
+ /* Enable core interrupt */
+ IFX_REG_SET_BIT(PCIE_RC_CORE_COMBINED_INT, PCIE_IRNEN(pcie_port));
+
+ /* Clear it first */
+ IFX_REG_SET_BIT(PCIE_RC_CORE_COMBINED_INT, PCIE_IRNCR(pcie_port));
+ ret = request_irq(pcie_irqs[pcie_port].ir_irq.irq, pcie_rc_core_isr, IRQF_DISABLED,
+ pcie_irqs[pcie_port].ir_irq.name, &ifx_pcie_controller[pcie_port]);
+ if (ret)
+ printk(KERN_ERR "%s request irq %d failed\n", __func__, IFX_PCIE_IR);
+
+ return ret;
+}
+
+int ifx_pcie_bios_map_irq(const struct pci_dev *dev, u8 slot, u8 pin)
+{
+ unsigned int irq_bit = 0;
+ int irq = 0;
+ struct ifx_pci_controller *ctrl = dev->bus->sysdata;
+ int pcie_port = ctrl->port;
+
+ IFX_PCIE_PRINT(PCIE_MSG_FIXUP, "%s port %d dev %s slot %d pin %d \n", __func__, pcie_port, pci_name(dev), slot, pin);
+
+ if ((pin == PCIE_LEGACY_DISABLE) || (pin > PCIE_LEGACY_INT_MAX)) {
+ printk(KERN_WARNING "WARNING: dev %s: invalid interrupt pin %d\n", pci_name(dev), pin);
+ return -1;
+ }
+ /* Pin index so minus one */
+ irq_bit = pcie_irqs[pcie_port].legacy_irq[pin - 1].irq_bit;
+ irq = pcie_irqs[pcie_port].legacy_irq[pin - 1].irq;
+ IFX_REG_SET_BIT(irq_bit, PCIE_IRNEN(pcie_port));
+ IFX_REG_SET_BIT(irq_bit, PCIE_IRNCR(pcie_port));
+ IFX_PCIE_PRINT(PCIE_MSG_FIXUP, "%s dev %s irq %d assigned\n", __func__, pci_name(dev), irq);
+ return irq;
+}
+
+/**
+ * \fn int ifx_pcie_bios_plat_dev_init(struct pci_dev *dev)
+ * \brief Called to perform platform specific PCI setup
+ *
+ * \param[in] dev The Linux PCI device structure for the device to map
+ * \return OK
+ * \ingroup IFX_PCIE_OS
+ */
+int ifx_pcie_bios_plat_dev_init(struct pci_dev *dev)
+{
+ u16 config;
+ unsigned int dconfig;
+ int pos;
+ /* Enable reporting System errors and parity errors on all devices */
+ /* Enable parity checking and error reporting */
+ pci_read_config_word(dev, PCI_COMMAND, &config);
+ config |= PCI_COMMAND_PARITY | PCI_COMMAND_SERR /*| PCI_COMMAND_INVALIDATE |
+ PCI_COMMAND_FAST_BACK*/;
+ pci_write_config_word(dev, PCI_COMMAND, config);
+
+ if (dev->subordinate) {
+ /* Set latency timers on sub bridges */
+ pci_write_config_byte(dev, PCI_SEC_LATENCY_TIMER, 0x40); /* XXX, */
+ /* More bridge error detection */
+ pci_read_config_word(dev, PCI_BRIDGE_CONTROL, &config);
+ config |= PCI_BRIDGE_CTL_PARITY | PCI_BRIDGE_CTL_SERR;
+ pci_write_config_word(dev, PCI_BRIDGE_CONTROL, config);
+ }
+ /* Enable the PCIe normal error reporting */
+ pos = pci_find_capability(dev, PCI_CAP_ID_EXP);
+ if (pos) {
+ /* Disable system error generation in response to error messages */
+ pci_read_config_word(dev, pos + PCI_EXP_RTCTL, &config);
+ config &= ~(PCI_EXP_RTCTL_SECEE | PCI_EXP_RTCTL_SENFEE | PCI_EXP_RTCTL_SEFEE);
+ pci_write_config_word(dev, pos + PCI_EXP_RTCTL, config);
+
+ /* Clear PCIE Capability's Device Status */
+ pci_read_config_word(dev, pos + PCI_EXP_DEVSTA, &config);
+ pci_write_config_word(dev, pos + PCI_EXP_DEVSTA, config);
+
+ /* Update Device Control */
+ pci_read_config_word(dev, pos + PCI_EXP_DEVCTL, &config);
+ /* Correctable Error Reporting */
+ config |= PCI_EXP_DEVCTL_CERE;
+ /* Non-Fatal Error Reporting */
+ config |= PCI_EXP_DEVCTL_NFERE;
+ /* Fatal Error Reporting */
+ config |= PCI_EXP_DEVCTL_FERE;
+ /* Unsupported Request */
+ config |= PCI_EXP_DEVCTL_URRE;
+ pci_write_config_word(dev, pos + PCI_EXP_DEVCTL, config);
+ }
+
+ /* Find the Advanced Error Reporting capability */
+ pos = pci_find_ext_capability(dev, PCI_EXT_CAP_ID_ERR);
+ if (pos) {
+ /* Clear Uncorrectable Error Status */
+ pci_read_config_dword(dev, pos + PCI_ERR_UNCOR_STATUS, &dconfig);
+ pci_write_config_dword(dev, pos + PCI_ERR_UNCOR_STATUS, dconfig);
+ /* Enable reporting of all uncorrectable errors */
+ /* Uncorrectable Error Mask - turned on bits disable errors */
+ pci_write_config_dword(dev, pos + PCI_ERR_UNCOR_MASK, 0);
+ /*
+ * Leave severity at HW default. This only controls if
+ * errors are reported as uncorrectable or
+ * correctable, not if the error is reported.
+ */
+ /* PCI_ERR_UNCOR_SEVER - Uncorrectable Error Severity */
+ /* Clear Correctable Error Status */
+ pci_read_config_dword(dev, pos + PCI_ERR_COR_STATUS, &dconfig);
+ pci_write_config_dword(dev, pos + PCI_ERR_COR_STATUS, dconfig);
+ /* Enable reporting of all correctable errors */
+ /* Correctable Error Mask - turned on bits disable errors */
+ pci_write_config_dword(dev, pos + PCI_ERR_COR_MASK, 0);
+ /* Advanced Error Capabilities */
+ pci_read_config_dword(dev, pos + PCI_ERR_CAP, &dconfig);
+ /* ECRC Generation Enable */
+ if (dconfig & PCI_ERR_CAP_ECRC_GENC)
+ dconfig |= PCI_ERR_CAP_ECRC_GENE;
+ /* ECRC Check Enable */
+ if (dconfig & PCI_ERR_CAP_ECRC_CHKC)
+ dconfig |= PCI_ERR_CAP_ECRC_CHKE;
+ pci_write_config_dword(dev, pos + PCI_ERR_CAP, dconfig);
+
+ /* PCI_ERR_HEADER_LOG - Header Log Register (16 bytes) */
+ /* Enable Root Port's interrupt in response to error messages */
+ pci_write_config_dword(dev, pos + PCI_ERR_ROOT_COMMAND,
+ PCI_ERR_ROOT_CMD_COR_EN |
+ PCI_ERR_ROOT_CMD_NONFATAL_EN |
+ PCI_ERR_ROOT_CMD_FATAL_EN);
+ /* Clear the Root status register */
+ pci_read_config_dword(dev, pos + PCI_ERR_ROOT_STATUS, &dconfig);
+ pci_write_config_dword(dev, pos + PCI_ERR_ROOT_STATUS, dconfig);
+ }
+ /* WAR, only 128 MRRS is supported, force all EPs to support this value */
+ pcie_set_readrq(dev, 128);
+ return 0;
+}
+
+static void pcie_phy_rst(int pcie_port)
+{
+ pcie_phy_rst_assert(pcie_port);
+ pcie_phy_rst_deassert(pcie_port);
+ /* Make sure PHY PLL is stable */
+ udelay(20);
+}
+
+static int pcie_rc_initialize(int pcie_port)
+{
+ int i;
+
+ pcie_rcu_endian_setup(pcie_port);
+
+ pcie_ep_gpio_rst_init(pcie_port);
+
+ /*
+ * XXX, PCIe elastic buffer bug will cause not to be detected. One more
+ * reset PCIe PHY will solve this issue
+ */
+ for (i = 0; i < IFX_PCIE_PHY_LOOP_CNT; i++) {
+ /* Disable PCIe PHY Analog part for sanity check */
+ pcie_phy_pmu_disable(pcie_port);
+ pcie_phy_rst(pcie_port);
+ /* PCIe Core reset enabled, low active, sw programmed */
+ pcie_core_rst_assert(pcie_port);
+ /* Put PCIe EP in reset status */
+ pcie_device_rst_assert(pcie_port);
+ /* PCI PHY & Core reset disabled, high active, sw programmed */
+ pcie_core_rst_deassert(pcie_port);
+ /* Already in a quiet state, program PLL, enable PHY, check ready bit */
+ pcie_phy_clock_mode_setup(pcie_port);
+ /* Enable PCIe PHY and Clock */
+ pcie_core_pmu_setup(pcie_port);
+ /* Clear status registers */
+ pcie_status_register_clear(pcie_port);
+#ifdef CONFIG_PCI_MSI
+ pcie_msi_init(pcie_port);
+#endif /* CONFIG_PCI_MSI */
+ pcie_rc_cfg_reg_setup(pcie_port);
+
+ /* Once link is up, break out */
+ if (pcie_app_loigc_setup(pcie_port) == 0)
+ break;
+ }
+ if (i >= IFX_PCIE_PHY_LOOP_CNT) {
+ printk(KERN_ERR "%s link up failed!!!!!\n", __func__);
+ return -EIO;
+ }
+ /* NB, don't increase ACK/NACK timer timeout value, which will cause a lot of COR errors */
+ pcie_replay_time_update(pcie_port);
+ return 0;
+}
+
+static int inline ifx_pcie_startup_port_nr(void)
+{
+ int pcie_port = IFX_PCIE_PORT0;
+
+ pcie_port = IFX_PCIE_PORT0;
+ return pcie_port;
+}
+
+/**
+ * \fn static int __init ifx_pcie_bios_init(void)
+ * \brief Initialize the IFX PCIe controllers
+ *
+ * \return -EIO PCIe PHY link is not up
+ * \return -ENOMEM Configuration/IO space failed to map
+ * \return 0 OK
+ * \ingroup IFX_PCIE_OS
+ */
+extern int (*ltqpci_plat_arch_init)(struct pci_dev *dev);
+extern int (*ltqpci_map_irq)(const struct pci_dev *dev, u8 slot, u8 pin);
+static int __devinit ltq_pcie_probe(struct platform_device *pdev)
+{
+ char ver_str[128] = {0};
+ void __iomem *io_map_base;
+ int pcie_port;
+ int startup_port;
+ ltqpci_map_irq = ifx_pcie_bios_map_irq;
+ ltqpci_plat_arch_init = ifx_pcie_bios_plat_dev_init;
+ /* Enable AHB Master/ Slave */
+ pcie_ahb_pmu_setup();
+
+ startup_port = ifx_pcie_startup_port_nr();
+
+ ltq_gpio_request(&pdev->dev, IFX_PCIE_GPIO_RESET, 0, 1, "pcie-reset");
+
+ for (pcie_port = startup_port; pcie_port < IFX_PCIE_CORE_NR; pcie_port++){
+ if (pcie_rc_initialize(pcie_port) == 0) {
+ /* Otherwise, warning will pop up */
+ io_map_base = ioremap(PCIE_IO_PHY_PORT_TO_BASE(pcie_port), PCIE_IO_SIZE);
+ if (io_map_base == NULL)
+ return -ENOMEM;
+ ifx_pcie_controller[pcie_port].pcic.io_map_base = (unsigned long)io_map_base;
+ register_pci_controller(&ifx_pcie_controller[pcie_port].pcic);
+ /* XXX, clear error status */
+ pcie_rc_core_int_init(pcie_port);
+ }
+ }
+
+ printk(KERN_INFO "%s", ver_str);
+return 0;
+}
+
+static struct platform_driver ltq_pcie_driver = {
+ .probe = ltq_pcie_probe,
+ .driver = {
+ .name = "pcie-xway",
+ .owner = THIS_MODULE,
+ },
+};
+
+int __init pciebios_init(void)
+{
+ return platform_driver_register(&ltq_pcie_driver);
+}
+
+arch_initcall(pciebios_init);