Armada XP has support for X4 lanes, boards specify this in their serdes_cfg. During PEX init in high_speed_env_lib.c, the configuration is stored in GEN_PURP_RES_2_REG.

When enumerating PEX, subsequent interfaces of an X4 lane must be skipped. Otherwise the enumeration hangs up the board.

The way this is implemented here is not exactly beautiful, but it mimics how Marvell's BSP does it. Alternatively we could get the information using board_serdes_cfg_get(), but that won't lead to clean code, either. Especially since the ugly includes will have to be done anyway.

Signed-off-by: Phil Sutter phil@nwl.cc --- drivers/pci/pci_mvebu.c | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+)

diff --git a/drivers/pci/pci_mvebu.c b/drivers/pci/pci_mvebu.c index fd2744d..aab53f4 100644 --- a/drivers/pci/pci_mvebu.c +++ b/drivers/pci/pci_mvebu.c @@ -18,6 +18,11 @@ #include <asm/arch/soc.h> #include <linux/mbus.h>

+#if defined(CONFIG_ARMADA_XP) +#include "../ddr/marvell/axp/ddr3_init.h" +#include "../../arch/arm/mach-mvebu/serdes/axp/board_env_spec.h" +#endif + DECLARE_GLOBAL_DATA_PTR;

/* PCIe unit register offsets */ @@ -155,6 +160,16 @@ static void mvebu_get_port_lane(struct mvebu_pcie *pcie, int pex_idx, } #endif

+#if defined(CONFIG_ARMADA_XP) +static int mvebu_pex_unit_is_x4(int pex_idx) +{ + int pex_unit = pex_idx < 9 ? pex_idx >> 2 : 3; + u32 mask = (0x0f << (pex_unit * 8)); + + return (reg_read(GEN_PURP_RES_2_REG) & mask) == mask; +} +#endif + static inline bool mvebu_pcie_link_up(struct mvebu_pcie *pcie) { u32 val; @@ -419,5 +434,11 @@ void pci_init_board(void) writel(0, pcie->base + PCIE_BAR_HI_OFF(0));

bus = hose->last_busno + 1; + +#if defined(CONFIG_ARMADA_XP) + /* need to skip more for X4 links, otherwise scan will hang */ + if (mvebu_pex_unit_is_x4(i)) + i += 3; +#endif } }