[cc: Cyrille]
On Fri, Feb 23, 2018 at 09:17:33AM +0800, Shawn Lin wrote:
> This patch adds support to the Rockchip PCIe controller in endpoint mode
> which currently supports up to 32 regions, and each regions should at
> least be 1MB per TRM.
>
> Signed-off-by: Shawn Lin <shawn.lin@xxxxxxxxxxxxxx>
>
> ---
>
> Changes in v2:
> - fix some error handling
>
> drivers/pci/rockchip/Kconfig | 12 +
> drivers/pci/rockchip/Makefile | 1 +
> drivers/pci/rockchip/pcie-rockchip-ep.c | 641 ++++++++++++++++++++++++++++++++
> drivers/pci/rockchip/pcie-rockchip.c | 25 +-
> drivers/pci/rockchip/pcie-rockchip.h | 92 +++++
> 5 files changed, 762 insertions(+), 9 deletions(-)
> create mode 100644 drivers/pci/rockchip/pcie-rockchip-ep.c
>
> diff --git a/drivers/pci/rockchip/Kconfig b/drivers/pci/rockchip/Kconfig
> index d655b77..668f850 100644
> --- a/drivers/pci/rockchip/Kconfig
> +++ b/drivers/pci/rockchip/Kconfig
> @@ -17,4 +17,16 @@ config PCIE_ROCKCHIP_HOST
> There is 1 internal PCIe port available to support GEN2 with
> 4 slots.
>
> +config PCIE_ROCKCHIP_EP
> + bool "Rockchip PCIe endpoint controller"
> + depends on ARCH_ROCKCHIP || COMPILE_TEST
> + depends on OF
> + depends on PCI_ENDPOINT
> + select MFD_SYSCON
> + select PCIE_ROCKCHIP
> + help
> + Say Y here if you want to support Rockchip PCIe controller in
> + endpoint mode on Rockchip SoC. There is 1 internal PCIe port
> + available to support GEN2 with 4 slots.
> +
> endmenu
> diff --git a/drivers/pci/rockchip/Makefile b/drivers/pci/rockchip/Makefile
> index 0af40db..8dce495 100644
> --- a/drivers/pci/rockchip/Makefile
> +++ b/drivers/pci/rockchip/Makefile
> @@ -1,3 +1,4 @@
> # SPDX-License-Identifier: GPL-2.0
> obj-$(CONFIG_PCIE_ROCKCHIP) += pcie-rockchip.o
> obj-$(CONFIG_PCIE_ROCKCHIP_HOST) += pcie-rockchip-host.o
> +obj-$(CONFIG_PCIE_ROCKCHIP_EP) += pcie-rockchip-ep.o
> diff --git a/drivers/pci/rockchip/pcie-rockchip-ep.c b/drivers/pci/rockchip/pcie-rockchip-ep.c
> new file mode 100644
> index 0000000..afe9fc4
> --- /dev/null
> +++ b/drivers/pci/rockchip/pcie-rockchip-ep.c
> @@ -0,0 +1,641 @@
> +// SPDX-License-Identifier: GPL-2.0+
> +/*
> + * Rockchip AXI PCIe endpoint controller driver
> + *
> + * Copyright (c) 2018 Rockchip, Inc.
> + *
> + * Author: Shawn Lin <shawn.lin@xxxxxxxxxxxxxx>
> + * Simon Xue <xxm@xxxxxxxxxxxxxx>
> + *
> + * Bits taken from Cadence endpoint controller driver
Well, "Bits" is an understatement here; anyway I think this is
a useless comment so I would remove it.
> + */
> +
> +#include <linux/kernel.h>
> +#include <linux/of.h>
> +#include <linux/pci-epc.h>
> +#include <linux/platform_device.h>
> +#include <linux/sizes.h>
> +#include <linux/delay.h>
> +#include <linux/configfs.h>
> +#include <linux/pci-epf.h>
Nit: Alphabetical order.
> +#include "pcie-rockchip.h"
> +
> +/**
> + * struct rockchip_pcie_ep - private data for PCIe endpoint controller driver
> + * @pcie: Rockchip PCIe controller
> + * @data: pointer to a 'struct rockchip_pcie_data'
> + * @ob_region_map: bitmask of mapped outbound regions
> + * @ob_addr: base addresses in the AXI bus where the outbound regions start
> + * @irq_phys_addr: base address on the AXI bus where the MSI/legacy IRQ
> + * dedicated outbound regions is mapped.
> + * @irq_cpu_addr: base address in the CPU space where a write access triggers
> + * the sending of a memory write (MSI) / normal message (legacy
> + * IRQ) TLP through the PCIe bus.
> + * @irq_pci_addr: used to save the current mapping of the MSI/legacy IRQ
> + * dedicated outbound region.
> + * @irq_pci_fn: the latest PCI function that has updated the mapping of
> + * the MSI/legacy IRQ dedicated outbound region.
> + * @irq_pending: bitmask of asserted legacy IRQs.
> + * @max_regions: maximum number of regions supported by hardware
> + */
> +struct rockchip_pcie_ep {
> + rockchip_pcie_epc pcie;
> + struct pci_epc *epc;
> + unsigned long ob_region_map;
> + phys_addr_t *ob_addr;
> + phys_addr_t irq_phys_addr;
> + void __iomem *irq_cpu_addr;
> + u64 irq_pci_addr;
> + u8 irq_pci_fn;
> + u8 irq_pending;
> + u32 max_regions;
> +};
> +
I think struct cdns_pcie_ep layout (whose content is identical)
is more sensible, follow it.
> +void rockchip_pcie_prog_ep_ob_atu(struct rockchip_pcie *rockchip, u8 fn,
> + u32 r, u32 type, u64 cpu_addr, u64 pci_addr,
> + size_t size, bool is_clr)
> +{
> + u64 sz = 1ULL << fls64(size - 1);
> + int num_pass_bits = ilog2(sz);
> + u32 addr0, addr1, desc0, desc1;
> + bool is_nor_msg = (type == AXI_WRAPPER_NOR_MSG);
> +
> + if (is_clr) {
> + rockchip_pcie_write(rockchip, 0,
> + ROCKCHIP_PCIE_AT_OB_REGION_PCI_ADDR0(r));
> + rockchip_pcie_write(rockchip, 0,
> + ROCKCHIP_PCIE_AT_OB_REGION_PCI_ADDR1(r));
> + rockchip_pcie_write(rockchip, 0,
> + ROCKCHIP_PCIE_AT_OB_REGION_DESC0(r));
> + rockchip_pcie_write(rockchip, 0,
> + ROCKCHIP_PCIE_AT_OB_REGION_DESC1(r));
> + rockchip_pcie_write(rockchip, 0,
> + ROCKCHIP_PCIE_AT_OB_REGION_CPU_ADDR0(r));
> + rockchip_pcie_write(rockchip, 0,
> + ROCKCHIP_PCIE_AT_OB_REGION_CPU_ADDR1(r));
> + return;
> + }
You can split the (is_clr) path in a separate function, say:
rockchip_pcie_clear_ep_ob_atu()
it would simplify this one.
> +
> + /* The minimal region size is 1MB */
> + if (num_pass_bits < 8)
> + num_pass_bits = 8;
> +
> + cpu_addr -= rockchip->mem_res->start;
> + addr0 = ((is_nor_msg ? 0x10 : (num_pass_bits - 1)) &
> + PCIE_CORE_OB_REGION_ADDR0_NUM_BITS) |
> + (lower_32_bits(cpu_addr) & PCIE_CORE_OB_REGION_ADDR0_LO_ADDR);
> + addr1 = upper_32_bits(is_nor_msg ? cpu_addr : pci_addr);
> + desc0 = ROCKCHIP_PCIE_AT_OB_REGION_DESC0_DEVFN(fn) | type;
> + desc1 = 0;
> +
> + if (is_nor_msg) {
> + rockchip_pcie_write(rockchip, 0,
> + ROCKCHIP_PCIE_AT_OB_REGION_PCI_ADDR0(r));
> + rockchip_pcie_write(rockchip, 0,
> + ROCKCHIP_PCIE_AT_OB_REGION_PCI_ADDR1(r));
> + rockchip_pcie_write(rockchip, desc0,
> + ROCKCHIP_PCIE_AT_OB_REGION_DESC0(r));
> + rockchip_pcie_write(rockchip, desc1,
> + ROCKCHIP_PCIE_AT_OB_REGION_DESC1(r));
> + } else {
> + /* PCI bus address region */
> + rockchip_pcie_write(rockchip, addr0,
> + ROCKCHIP_PCIE_AT_OB_REGION_PCI_ADDR0(r));
> + rockchip_pcie_write(rockchip, addr1,
> + ROCKCHIP_PCIE_AT_OB_REGION_PCI_ADDR1(r));
> + rockchip_pcie_write(rockchip, desc0,
> + ROCKCHIP_PCIE_AT_OB_REGION_DESC0(r));
> + rockchip_pcie_write(rockchip, desc1,
> + ROCKCHIP_PCIE_AT_OB_REGION_DESC1(r));
> +
> + addr0 =
> + ((num_pass_bits - 1) & PCIE_CORE_OB_REGION_ADDR0_NUM_BITS) |
> + (lower_32_bits(cpu_addr) &
> + PCIE_CORE_OB_REGION_ADDR0_LO_ADDR);
> + addr1 = upper_32_bits(cpu_addr);
> + }
> +
> + /* CPU bus address region */
> + rockchip_pcie_write(rockchip, addr0,
> + ROCKCHIP_PCIE_AT_OB_REGION_CPU_ADDR0(r));
> + rockchip_pcie_write(rockchip, addr1,
> + ROCKCHIP_PCIE_AT_OB_REGION_CPU_ADDR1(r));
> +}
> +
> +static int rockchip_pcie_ep_write_header(struct pci_epc *epc, u8 fn,
> + struct pci_epf_header *hdr)
> +{
> + struct rockchip_pcie_ep *ep = epc_get_drvdata(epc);
> + struct rockchip_pcie *rockchip = &ep->pcie;
> +
> + /* All functions share the same vendor ID with function 0 */
> + if (fn == 0) {
> + u32 vid_regs = (hdr->vendorid & GENMASK(15, 0)) |
> + (hdr->subsys_vendor_id & GENMASK(31, 16)) << 16;
> +
> + rockchip_pcie_write(rockchip, vid_regs,
> + PCIE_CORE_CONFIG_VENDOR);
> + }
> +
> + rockchip_pcie_write(rockchip, hdr->deviceid << 16,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) + PCI_VENDOR_ID);
> +
> + rockchip_pcie_write(rockchip,
> + hdr->revid |
> + hdr->progif_code << 8 |
> + hdr->subclass_code << 16 |
> + hdr->baseclass_code << 24,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) + PCI_REVISION_ID);
> + rockchip_pcie_write(rockchip, hdr->cache_line_size,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) +
> + PCI_CACHE_LINE_SIZE);
> + rockchip_pcie_write(rockchip, hdr->subsys_id << 16,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) +
> + PCI_SUBSYSTEM_VENDOR_ID);
> + rockchip_pcie_write(rockchip, hdr->interrupt_pin << 8,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) +
> + PCI_INTERRUPT_LINE);
> +
> + return 0;
> +
Useless empty line.
> +}
> +
> +static int rockchip_pcie_ep_set_bar(struct pci_epc *epc, u8 fn,
> + enum pci_barno bar, dma_addr_t bar_phys,
> + size_t size, int flags)
> +{
> + struct rockchip_pcie_ep *ep = epc_get_drvdata(epc);
> + struct rockchip_pcie *rockchip = &ep->pcie;
> + u32 addr0, addr1, reg, cfg, b, aperture, ctrl;
> + u64 sz;
> +
> + /* BAR size is 2^(aperture + 7) */
> + sz = max_t(size_t, size, MIN_EP_APERTURE);
> +
> + /*
> + * roundup_pow_of_two() returns an unsigned long, which is not suited
> + * for 64bit values.
> + */
> + sz = 1ULL << fls64(sz - 1);
> + aperture = ilog2(sz) - 7; /* 128B -> 0, 256B -> 1, 512B -> 2, ... */
> +
> + if ((flags & PCI_BASE_ADDRESS_SPACE) == PCI_BASE_ADDRESS_SPACE_IO) {
> + ctrl = ROCKCHIP_PCIE_CORE_BAR_CFG_CTRL_IO_32BITS;
> + } else {
> + bool is_prefetch = !!(flags & PCI_BASE_ADDRESS_MEM_PREFETCH);
> + bool is_64bits = sz > SZ_2G;
> +
> + if (is_64bits && (bar & 1))
> + return -EINVAL;
> +
> + if (is_64bits && is_prefetch)
> + ctrl =
> + ROCKCHIP_PCIE_CORE_BAR_CFG_CTRL_PREFETCH_MEM_64BITS;
> + else if (is_prefetch)
> + ctrl =
> + ROCKCHIP_PCIE_CORE_BAR_CFG_CTRL_PREFETCH_MEM_32BITS;
> + else if (is_64bits)
> + ctrl = ROCKCHIP_PCIE_CORE_BAR_CFG_CTRL_MEM_64BITS;
> + else
> + ctrl = ROCKCHIP_PCIE_CORE_BAR_CFG_CTRL_MEM_32BITS;
> + }
> +
> + if (bar < BAR_4) {
> + reg = ROCKCHIP_PCIE_CORE_EP_FUNC_BAR_CFG0(fn);
> + b = bar;
> + } else {
> + reg = ROCKCHIP_PCIE_CORE_EP_FUNC_BAR_CFG1(fn);
> + b = bar - BAR_4;
> + }
> +
> + addr0 = lower_32_bits(bar_phys);
> + addr1 = upper_32_bits(bar_phys);
> +
> + cfg = rockchip_pcie_read(rockchip, reg);
> + cfg &= ~(ROCKCHIP_PCIE_CORE_EP_FUNC_BAR_CFG_BAR_APERTURE_MASK(b) |
> + ROCKCHIP_PCIE_CORE_EP_FUNC_BAR_CFG_BAR_CTRL_MASK(b));
> + cfg |= (ROCKCHIP_PCIE_CORE_EP_FUNC_BAR_CFG_BAR_APERTURE(b, aperture) |
> + ROCKCHIP_PCIE_CORE_EP_FUNC_BAR_CFG_BAR_CTRL(b, ctrl));
> +
> + rockchip_pcie_write(rockchip, cfg, reg);
> + rockchip_pcie_write(rockchip, addr0,
> + ROCKCHIP_PCIE_AT_IB_EP_FUNC_BAR_ADDR0(fn, bar));
> + rockchip_pcie_write(rockchip, addr1,
> + ROCKCHIP_PCIE_AT_IB_EP_FUNC_BAR_ADDR1(fn, bar));
> +
> + return 0;
> +}
> +
> +static void rockchip_pcie_ep_clear_bar(struct pci_epc *epc, u8 fn,
> + enum pci_barno bar)
> +{
> + struct rockchip_pcie_ep *ep = epc_get_drvdata(epc);
> + struct rockchip_pcie *rockchip = &ep->pcie;
> + u32 reg, cfg, b, ctrl;
> +
> + if (bar < BAR_4) {
> + reg = ROCKCHIP_PCIE_CORE_EP_FUNC_BAR_CFG0(fn);
> + b = bar;
> + } else {
> + reg = ROCKCHIP_PCIE_CORE_EP_FUNC_BAR_CFG1(fn);
> + b = bar - BAR_4;
> + }
> +
> + ctrl = ROCKCHIP_PCIE_CORE_BAR_CFG_CTRL_DISABLED;
> + cfg = rockchip_pcie_read(rockchip, reg);
> + cfg &= ~(ROCKCHIP_PCIE_CORE_EP_FUNC_BAR_CFG_BAR_APERTURE_MASK(b) |
> + ROCKCHIP_PCIE_CORE_EP_FUNC_BAR_CFG_BAR_CTRL_MASK(b));
> + cfg |= ROCKCHIP_PCIE_CORE_EP_FUNC_BAR_CFG_BAR_CTRL(b, ctrl);
> +
> + rockchip_pcie_write(rockchip, cfg, reg);
> + rockchip_pcie_write(rockchip, 0x0,
> + ROCKCHIP_PCIE_AT_IB_EP_FUNC_BAR_ADDR0(fn, bar));
> + rockchip_pcie_write(rockchip, 0x0,
> + ROCKCHIP_PCIE_AT_IB_EP_FUNC_BAR_ADDR1(fn, bar));
> +
Useless empty line.
> +}
> +
> +static int rockchip_pcie_ep_map_addr(struct pci_epc *epc, u8 fn,
> + phys_addr_t addr, u64 pci_addr,
> + size_t size)
> +{
> + struct rockchip_pcie_ep *ep = epc_get_drvdata(epc);
> + struct rockchip_pcie *pcie = &ep->pcie;
> + u32 r;
> +
> + r = find_first_zero_bit(&ep->ob_region_map,
> + sizeof(ep->ob_region_map) * BITS_PER_LONG);
> + if (r >= ep->max_regions - 1) {
Why are we comparing for equality against (max_regions - 1) ? Aren't
we leaving a region out ?
NB: I know this code is identical to the Cadence driver but copy and
paste is not a good reason for it and if that's a bug it ought to be
fixed in both drivers.
> + dev_err(&epc->dev, "no free outbound region\n");
> + return -EINVAL;
> + }
> +
> + rockchip_pcie_prog_ep_ob_atu(pcie, fn, r, AXI_WRAPPER_MEM_WRITE, addr,
> + pci_addr, size, false);
> +
> + set_bit(r, &ep->ob_region_map);
> + ep->ob_addr[r] = addr;
> +
> + return 0;
> +
Useless empty line.
> +}
> +
> +static void rockchip_pcie_ep_unmap_addr(struct pci_epc *epc, u8 fn,
> + phys_addr_t addr)
> +{
> + struct rockchip_pcie_ep *ep = epc_get_drvdata(epc);
> + struct rockchip_pcie *rockchip = &ep->pcie;
> + u32 r;
> +
> + for (r = 0; r < ep->max_regions - 1; r++)
> + if (ep->ob_addr[r] == addr)
> + break;
See above for max_regions - 1.
> +
> + if (r == ep->max_regions - 1)
> + return;
Ditto.
> +
> + rockchip_pcie_prog_ep_ob_atu(rockchip, 0, 0, 0, 0, 0, 0, true);
> +
> + ep->ob_addr[r] = 0;
> + clear_bit(r, &ep->ob_region_map);
> +}
> +
> +static int rockchip_pcie_ep_set_msi(struct pci_epc *epc, u8 fn,
> + u8 multi_msg_cap)
> +{
> + struct rockchip_pcie_ep *ep = epc_get_drvdata(epc);
> + struct rockchip_pcie *rockchip = &ep->pcie;
> + u16 flags;
> +
> + flags = rockchip_pcie_read(rockchip,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) +
> + ROCKCHIP_PCIE_EP_MSI_CTRL_REG);
> + flags &= ~ROCKCHIP_PCIE_EP_MSI_CTRL_MMC_MASK;
> + flags |=
> + ((multi_msg_cap << 1) << ROCKCHIP_PCIE_EP_MSI_CTRL_MMC_OFFSET) |
> + PCI_MSI_FLAGS_64BIT;
> + flags &= ~ROCKCHIP_PCIE_EP_MSI_CTRL_MASK_MSI_CAP;
> + rockchip_pcie_write(rockchip, flags,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) +
> + ROCKCHIP_PCIE_EP_MSI_CTRL_REG);
> + return 0;
> +}
> +
> +static int rockchip_pcie_ep_get_msi(struct pci_epc *epc, u8 fn)
> +{
> + struct rockchip_pcie_ep *ep = epc_get_drvdata(epc);
> + struct rockchip_pcie *rockchip = &ep->pcie;
> + u16 flags;
> +
> + flags = rockchip_pcie_read(rockchip,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) +
> + ROCKCHIP_PCIE_EP_MSI_CTRL_REG);
> + if (!(flags & ROCKCHIP_PCIE_EP_MSI_CTRL_ME))
> + return -EINVAL;
> +
> + return ((flags & ROCKCHIP_PCIE_EP_MSI_CTRL_MME_MASK) >>
> + ROCKCHIP_PCIE_EP_MSI_CTRL_MME_OFFSET);
> +}
> +
> +static void rockchip_pcie_ep_assert_intx(struct rockchip_pcie_ep *ep, u8 fn,
> + u8 intx, bool is_asserted)
> +{
> + struct rockchip_pcie *rockchip = &ep->pcie;
> + u32 r = ep->max_regions - 1;
> + u32 offset;
> + u16 status;
> + u8 msg_code;
> +
> + if (unlikely(ep->irq_pci_addr != ROCKCHIP_PCIE_EP_PCI_LEGACY_IRQ_ADDR ||
> + ep->irq_pci_fn != fn)) {
> + rockchip_pcie_prog_ep_ob_atu(rockchip, fn, r,
> + AXI_WRAPPER_NOR_MSG,
> + ep->irq_phys_addr, 0, 0, false);
> + ep->irq_pci_addr = ROCKCHIP_PCIE_EP_PCI_LEGACY_IRQ_ADDR;
> + ep->irq_pci_fn = fn;
> + }
> +
> + intx &= 3;
> + if (is_asserted) {
> + ep->irq_pending |= BIT(intx);
> + msg_code = ROCKCHIP_PCIE_MSG_CODE_ASSERT_INTA + intx;
> + } else {
> + ep->irq_pending &= ~BIT(intx);
> + msg_code = ROCKCHIP_PCIE_MSG_CODE_DEASSERT_INTA + intx;
> + }
> +
> + status = rockchip_pcie_read(rockchip,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) +
> + ROCKCHIP_PCIE_EP_CMD_STATUS);
> + status &= ROCKCHIP_PCIE_EP_CMD_STATUS_IS;
> +
> + if ((status != 0) ^ (ep->irq_pending != 0)) {
> + status ^= ROCKCHIP_PCIE_EP_CMD_STATUS_IS;
> + rockchip_pcie_write(rockchip, status,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) +
> + ROCKCHIP_PCIE_EP_CMD_STATUS);
> + }
> +
> + offset =
> + ROCKCHIP_PCIE_MSG_ROUTING(ROCKCHIP_PCIE_MSG_ROUTING_LOCAL_INTX) |
> + ROCKCHIP_PCIE_MSG_CODE(msg_code) | ROCKCHIP_PCIE_MSG_NO_DATA;
> + writel(0, ep->irq_cpu_addr + offset);
> +}
> +
> +static int rockchip_pcie_ep_send_legacy_irq(struct rockchip_pcie_ep *ep, u8 fn,
> + u8 intx)
> +{
> + u16 cmd;
> +
> + cmd = rockchip_pcie_read(&ep->pcie,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) +
> + ROCKCHIP_PCIE_EP_CMD_STATUS);
> +
> + if (cmd & PCI_COMMAND_INTX_DISABLE)
> + return -EINVAL;
> +
> + rockchip_pcie_ep_assert_intx(ep, fn, intx, true);
> + mdelay(1);
Is there some HW engineering backing this mdelay() value ? That's a
question for Cadence driver too.
> + rockchip_pcie_ep_assert_intx(ep, fn, intx, false);
> + return 0;
> +}
> +
> +static int rockchip_pcie_ep_send_msi_irq(struct rockchip_pcie_ep *ep, u8 fn,
> + u8 interrupt_num)
> +{
> + struct rockchip_pcie *rockchip = &ep->pcie;
> + u16 flags, mme, data, data_mask;
> + u8 msi_count;
> + u64 pci_addr, pci_addr_mask = 0xff;
> +
> + /* Check MSI enable bit */
> + flags = rockchip_pcie_read(&ep->pcie,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) +
> + ROCKCHIP_PCIE_EP_MSI_CTRL_REG);
> + if (!(flags & ROCKCHIP_PCIE_EP_MSI_CTRL_ME))
> + return -EINVAL;
> +
> + /* Get MSI numbers from MME */
> + mme = ((flags & ROCKCHIP_PCIE_EP_MSI_CTRL_MME_MASK) >>
> + ROCKCHIP_PCIE_EP_MSI_CTRL_MME_OFFSET);
> + msi_count = 1 << mme;
> + if (!interrupt_num || interrupt_num > msi_count)
> + return -EINVAL;
> +
> + /* Set MSI private data */
> + data_mask = msi_count - 1;
> + data = rockchip_pcie_read(rockchip,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) +
> + ROCKCHIP_PCIE_EP_MSI_CTRL_REG +
> + PCI_MSI_DATA_64);
> + data = (data & ~data_mask) | ((interrupt_num - 1) & data_mask);
> +
> + /* Get MSI PCI address */
> + pci_addr = rockchip_pcie_read(rockchip,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) +
> + ROCKCHIP_PCIE_EP_MSI_CTRL_REG +
> + PCI_MSI_ADDRESS_HI);
> + pci_addr <<= 32;
> + pci_addr |= rockchip_pcie_read(rockchip,
> + ROCKCHIP_PCIE_EP_FUNC_BASE(fn) +
> + ROCKCHIP_PCIE_EP_MSI_CTRL_REG +
> + PCI_MSI_ADDRESS_LO);
> + pci_addr &= GENMASK_ULL(63, 2);
> +
> + /* Set the outbound region if needed. */
> + if (unlikely(ep->irq_pci_addr != (pci_addr & ~pci_addr_mask) ||
> + ep->irq_pci_fn != fn)) {
> + rockchip_pcie_prog_ep_ob_atu(rockchip, fn, ep->max_regions - 1,
> + AXI_WRAPPER_MEM_WRITE,
> + ep->irq_phys_addr,
> + pci_addr & ~pci_addr_mask,
> + pci_addr_mask + 1, false);
> + ep->irq_pci_addr = (pci_addr & ~pci_addr_mask);
> + ep->irq_pci_fn = fn;
> + }
> +
> + writew(data, ep->irq_cpu_addr + (pci_addr & pci_addr_mask));
> + return 0;
> +}
> +
> +
Useless empty line.
> +static int rockchip_pcie_ep_raise_irq(struct pci_epc *epc, u8 fn,
> + enum pci_epc_irq_type type,
> + u8 interrupt_num)
> +{
> + struct rockchip_pcie_ep *ep = epc_get_drvdata(epc);
> +
> + switch (type) {
> + case PCI_EPC_IRQ_LEGACY:
> + return rockchip_pcie_ep_send_legacy_irq(ep, fn, 0);
Why always INTA (question valid on Cadence driver too) ?
> + case PCI_EPC_IRQ_MSI:
> + return rockchip_pcie_ep_send_msi_irq(ep, fn, interrupt_num);
> + default:
> + return -EINVAL;
> + }
> +}
> +
> +static int rockchip_pcie_ep_start(struct pci_epc *epc)
> +{
> + struct rockchip_pcie_ep *ep = epc_get_drvdata(epc);
> + struct rockchip_pcie *rockchip = &ep->pcie;
> + struct pci_epf *epf;
> + u32 cfg;
> +
> + cfg = BIT(0);
> + list_for_each_entry(epf, &epc->pci_epf, list)
> + cfg |= BIT(epf->func_no);
> +
> + rockchip_pcie_write(rockchip, cfg, PCIE_CORE_PHY_FUNC_CFG);
> +
> + list_for_each_entry(epf, &epc->pci_epf, list)
> + pci_epf_linkup(epf);
> +
> + return 0;
> +}
> +
> +static const struct pci_epc_ops rockchip_pcie_epc_ops = {
> + .write_header = rockchip_pcie_ep_write_header,
> + .set_bar = rockchip_pcie_ep_set_bar,
> + .clear_bar = rockchip_pcie_ep_clear_bar,
> + .map_addr = rockchip_pcie_ep_map_addr,
> + .unmap_addr = rockchip_pcie_ep_unmap_addr,
> + .set_msi = rockchip_pcie_ep_set_msi,
> + .get_msi = rockchip_pcie_ep_get_msi,
> + .raise_irq = rockchip_pcie_ep_raise_irq,
> + .start = rockchip_pcie_ep_start,
> +};
> +
> +/**
> + * rockchip_pcie_parse_dt - Parse Device Tree
> + * @rockchip: PCIe port information
Missing a parameter. By the way a kerneldoc for this entry does
not seem that useful to me, you can remove it.
> + *
> + * Return: '0' on success and error value on failure
> + */
> +static int rockchip_pcie_parse_ep_dt(struct rockchip_pcie *rockchip,
> + struct rockchip_pcie_ep *ep)
> +{
> + struct device *dev = rockchip->dev;
> + int err;
> +
> + err = rockchip_pcie_parse_dt(rockchip);
> + if (err)
> + return err;
> +
> + err = rockchip_pcie_get_phys(rockchip);
> + if (err)
> + return err;
> +
> + err = of_property_read_u32(dev->of_node,
> + "rockchip,max-outbound-regions",
> + &ep->max_regions);
> + if (err < 0 || ep->max_regions > MAX_REGION_LIMIT)
> + ep->max_regions = MAX_REGION_LIMIT;
> +
> + err = of_property_read_u8(dev->of_node, "max-functions",
> + &ep->epc->max_functions);
> + if (err < 0)
> + ep->epc->max_functions = 1;
> +
> + return 0;
> +}
> +
> +static const struct of_device_id rockchip_pcie_ep_of_match[] = {
> + { .compatible = "rockchip,rk3399-pcie-ep"},
> + {},
> +};
> +
> +static int rockchip_pcie_ep_probe(struct platform_device *pdev)
> +{
> + struct device *dev = &pdev->dev;
> + struct rockchip_pcie_ep *ep;
> + rockchip_pcie_epc *rockchip;
> + struct pci_epc *epc;
> + size_t max_regions;
> + int err;
> +
> + ep = devm_kzalloc(dev, sizeof(*ep), GFP_KERNEL);
> + if (!ep)
> + return -ENOMEM;
> +
> + platform_set_drvdata(pdev, ep);
What's this used for ?
> + rockchip = &ep->pcie;
> + rockchip->is_rc = false;
> + rockchip->dev = dev;
> +
> + epc = devm_pci_epc_create(dev, &rockchip_pcie_epc_ops);
> + if (IS_ERR(epc)) {
> + dev_err(dev, "failed to create epc device\n");
> + return PTR_ERR(epc);
> + }
> +
> + ep->epc = epc;
> + epc_set_drvdata(epc, ep);
> +
> + err = rockchip_pcie_parse_ep_dt(rockchip, ep);
> + if (err)
> + return err;
> +
> + err = rockchip_pcie_enable_clocks(rockchip);
> + if (err)
> + return err;
> +
> + err = rockchip_pcie_init_port(rockchip);
> + if (err)
> + goto err_disable_clocks;
> +
> + /* Establish the link automatically */
> + rockchip_pcie_write(rockchip, PCIE_CLIENT_LINK_TRAIN_ENABLE,
> + PCIE_CLIENT_CONFIG);
> +
> + max_regions = ep->max_regions;
> + ep->ob_addr = devm_kzalloc(dev, max_regions * sizeof(*ep->ob_addr),
> + GFP_KERNEL);
> +
> + if (!ep->ob_addr) {
> + err = -ENOMEM;
> + goto err_uninit_port;
> + }
> +
> + /* Only enable function 0 by default */
> + rockchip_pcie_write(rockchip, BIT(0), PCIE_CORE_PHY_FUNC_CFG);
> +
> +
Useless empty line.
> + err = pci_epc_mem_init(epc, rockchip->mem_res->start,
> + resource_size(rockchip->mem_res));
> + if (err < 0) {
> + dev_err(dev, "failed to initialize the memory space\n");
> + goto err_uninit_port;
> + }
> +
> + ep->irq_cpu_addr = pci_epc_mem_alloc_addr(epc, &ep->irq_phys_addr,
> + SZ_128K);
> + if (!ep->irq_cpu_addr) {
> + dev_err(dev, "failed to reserve memory space for MSI\n");
> + err = -ENOMEM;
> + goto err_epc_mem_exit;
> + }
> +
> + ep->irq_pci_addr = ROCKCHIP_PCIE_EP_DUMMY_IRQ_ADDR;
> +
> + return 0;
> +err_epc_mem_exit:
> + pci_epc_mem_exit(epc);
> +err_uninit_port:
> + rockchip_pcie_deinit_phys(rockchip);
> +err_disable_clocks:
> + rockchip_pcie_disable_clocks(rockchip);
> + return err;
> +}
> +
> +static struct platform_driver rockchip_pcie_ep_driver = {
> + .driver = {
> + .name = "rockchip-pcie-ep",
> + .of_match_table = rockchip_pcie_ep_of_match,
> + },
> + .probe = rockchip_pcie_ep_probe,
> +};
> +
> +builtin_platform_driver(rockchip_pcie_ep_driver);
> diff --git a/drivers/pci/rockchip/pcie-rockchip.c b/drivers/pci/rockchip/pcie-rockchip.c
> index d03508c..1d38ad7 100644
> --- a/drivers/pci/rockchip/pcie-rockchip.c
> +++ b/drivers/pci/rockchip/pcie-rockchip.c
> @@ -29,15 +29,22 @@ int rockchip_pcie_parse_dt(struct rockchip_pcie *rockchip)
> struct resource *regs;
> int err;
>
> - regs = platform_get_resource_byname(pdev,
> - IORESOURCE_MEM,
> - "axi-base");
> - rockchip->reg_base = devm_pci_remap_cfg_resource(dev, regs);
> - if (IS_ERR(rockchip->reg_base))
> - return PTR_ERR(rockchip->reg_base);
> -
> - regs = platform_get_resource_byname(pdev,
> - IORESOURCE_MEM,
> + if (rockchip->is_rc) {
> + regs = platform_get_resource_byname(pdev,
> + IORESOURCE_MEM,
> + "axi-base");
> + rockchip->reg_base = devm_pci_remap_cfg_resource(dev, regs);
> + if (IS_ERR(rockchip->reg_base))
> + return PTR_ERR(rockchip->reg_base);
> + } else {
> + rockchip->mem_res =
> + platform_get_resource_byname(pdev, IORESOURCE_MEM,
> + "mem-base");
> + if (!rockchip->mem_res)
> + return -EINVAL;
> + }
> +
> + regs = platform_get_resource_byname(pdev, IORESOURCE_MEM,
> "apb-base");
> rockchip->apb_base = devm_ioremap_resource(dev, regs);
> if (IS_ERR(rockchip->apb_base))
> diff --git a/drivers/pci/rockchip/pcie-rockchip.h b/drivers/pci/rockchip/pcie-rockchip.h
> index af3e74f..f4bfc22 100644
> --- a/drivers/pci/rockchip/pcie-rockchip.h
> +++ b/drivers/pci/rockchip/pcie-rockchip.h
> @@ -23,6 +23,9 @@
[...]
> +typedef struct rockchip_pcie rockchip_pcie_epc;
This typedef has no justification and just obfuscates, please remove it.
Thanks,
Lorenzo
