[U-Boot] [PATCH v2] pci: pcie_imx: Fix hang on mx6qp

Bin Meng bmeng.cn at gmail.com
Tue Oct 13 15:41:03 CEST 2015


Hi Fabio,

On Tue, Oct 13, 2015 at 9:37 PM, Fabio Estevam
<fabio.estevam at freescale.com> wrote:
> PCI driver currently hangs on mx6qp.
>
> Toggle the reset bit with the appropriate timings to fix the issue.
>
> Based on the FSL kernel driver implementation.
>
> Signed-off-by: Fabio Estevam <fabio.estevam at freescale.com>
> ---
> Changes since v1:
> - Rearrange the code to avoid explicit delay (Marek)
>
>  arch/arm/include/asm/arch-mx6/iomux.h | 2 ++
>  drivers/pci/pcie_imx.c                | 9 +++++++++
>  2 files changed, 11 insertions(+)
>
> diff --git a/arch/arm/include/asm/arch-mx6/iomux.h b/arch/arm/include/asm/arch-mx6/iomux.h
> index 9b3a91f..e3ef59f 100644
> --- a/arch/arm/include/asm/arch-mx6/iomux.h
> +++ b/arch/arm/include/asm/arch-mx6/iomux.h
> @@ -18,6 +18,8 @@
>  #define IOMUXC_GPR1_REF_SSP_EN                 (1 << 16)
>  #define IOMUXC_GPR1_TEST_POWERDOWN             (1 << 18)
>
> +#define IOMUXC_GPR1_PCIE_SW_RST                (1 << 29)
> +
>  /*
>   * IOMUXC_GPR5 bit fields
>   */
> diff --git a/drivers/pci/pcie_imx.c b/drivers/pci/pcie_imx.c
> index 1568f20..40f2e6a 100644
> --- a/drivers/pci/pcie_imx.c
> +++ b/drivers/pci/pcie_imx.c
> @@ -19,6 +19,7 @@
>  #include <asm/io.h>
>  #include <linux/sizes.h>
>  #include <errno.h>
> +#include <asm/arch/sys_proto.h>
>
>  #define PCI_ACCESS_READ  0
>  #define PCI_ACCESS_WRITE 1
> @@ -430,6 +431,10 @@ static int imx_pcie_write_config(struct pci_controller *hose, pci_dev_t d,
>  static int imx6_pcie_assert_core_reset(void)
>  {
>         struct iomuxc *iomuxc_regs = (struct iomuxc *)IOMUXC_BASE_ADDR;
> +
> +       if (is_mx6dqp())
> +               setbits_le32(&iomuxc_regs->gpr[1], IOMUXC_GPR1_PCIE_SW_RST);
> +
>  #if defined(CONFIG_MX6SX)
>         struct gpc *gpc_regs = (struct gpc *)GPC_BASE_ADDR;
>
> @@ -536,6 +541,10 @@ static int imx6_pcie_deassert_core_reset(void)
>
>         enable_pcie_clock();
>
> +

nits: this blank line can be avoided, no?

> +       if (is_mx6dqp())
> +               clrbits_le32(&iomuxc_regs->gpr[1], IOMUXC_GPR1_PCIE_SW_RST);
> +
>         /*
>          * Wait for the clock to settle a bit, when the clock are sourced
>          * from the CPU, we need about 30 ms to settle.
> --

Regards,
Bin


More information about the U-Boot mailing list