Linux Kernel 6.0 CXL Core Regs.c 详解

前言

CXL 是一个比较新的技术，所以我研究的内核源码是选了当前比较新的内核版本 linux 6.0。打算将内核关于 CXL 的驱动进行解析一遍，一步一步慢慢来。

在阅读之前，希望读者能有一定的 PCIe 基础知识，精力有限，不能把所有知识点都能说的很详细，需要一定的基础才能理解，同时，希望在学习的过程中，手边能有 PCIe 5.0 Spec 以及 CXL 2.0 Spec，以便随时查看，当然，我也会尽量把重点的部分截图在博文中。

最后，如果有问题请留言讨论。

Ref

《PCI_Express_Base_5.0r1.0》
《CXL Specification_rev2p0_ver1p0_2020Oct26》

正文

涉及的部分寄存器，Ref CXL 2.0 Spec ：

1. CXL Subsystem Component Register Ranges

请添加图片描述
2. CXL.cache and CXL.mem Architectural Register Header

请添加图片描述
3. CXL HDM Decoder Capability Header

请添加图片描述

4. CXL HDM Decoder Capability Structure

请添加图片描述


// linux-6.0\drivers\cxl\core\regs.c

// SPDX-License-Identifier: GPL-2.0-only
/* Copyright(c) 2020 Intel Corporation. */
#include <linux/io-64-nonatomic-lo-hi.h>
#include <linux/device.h>
#include <linux/slab.h>
#include <linux/pci.h>
#include <cxlmem.h>
#include <cxlpci.h>

/**
 * DOC: cxl registers
 *
 * CXL device capabilities are enumerated by PCI DVSEC (Designated
 * Vendor-specific) and / or descriptors provided by platform firmware.
 * They can be defined as a set like the device and component registers
 * mandated by CXL Section 8.1.12.2 Memory Device PCIe Capabilities and
 * Extended Capabilities, or they can be individual capabilities
 * appended to bridged and endpoint devices.
 *
 * Provide common infrastructure for enumerating and mapping these
 * discrete capabilities.
 */

/**
 * cxl_probe_component_regs() - Detect CXL Component register blocks
 * @dev: Host device of the @base mapping
 * @base: Mapping containing the HDM Decoder Capability Header
 * @map: Map object describing the register block information found
 *
 * See CXL 2.0 8.2.4 Component Register Layout and Definition
 * See CXL 2.0 8.2.5.5 CXL Device Register Interface
 *
 * Probe for component register information and return it in map object.
 */

// 探测 CXL 组件寄存器块, 找到设备中的 HDM Decoder 寄存器并记录位置和大小
// dev : Host 设备
// base : 包含 HDM 解码能力头的映射基地址，就是寄存器块所在位置的基地址，已映射后的虚拟地址
// map : 描述发现的寄存器块信息的对象
// Ref CXL 2.0 Spec 8.1.9 Register Locator DVSEC
//                  8.2.5 CXL.cache and CXL.mem Registers
void cxl_probe_component_regs(struct device *dev, void __iomem *base,
			      struct cxl_component_reg_map *map)
{
	int cap, cap_count;
	u32 cap_array;

	*map = (struct cxl_component_reg_map) { 0 };

	/*
	 * CXL.cache and CXL.mem registers are at offset 0x1000 as defined in
	 * CXL 2.0 8.2.4 Table 141. 如上图
	 */
    // CXL_CM_OFFSET 0x1000
    // CXL_CM_OFFSET 偏移处为 .cache 和 .mem 的寄存器位置
	base += CXL_CM_OFFSET;

	cap_array = readl(base + CXL_CM_CAP_HDR_OFFSET);

    // 见上图 CXL.cache and CXL.mem Architectural Register Header
    // 或Ref CXL 2.0 8.2.5.1 CXL Capability Header Register
    // CXL Capability Header Register 的 CXL_Capability_ID (0:15) 必须为 1，不为 1 报错
	if (FIELD_GET(CXL_CM_CAP_HDR_ID_MASK, cap_array) != CM_CAP_HDR_CAP_ID) {
		dev_err(dev,
			"Couldn't locate the CXL.cache and CXL.mem capability array header.\n");
		return;
	}

	/* It's assumed that future versions will be backward compatible */
    // 见上图 CXL.cache and CXL.mem Architectural Register Header
    // 或Ref CXL 2.0 8.2.5.1 CXL Capability Header Register
    // 31:24 Array_Size 定义了存在的元素数目
	cap_count = FIELD_GET(CXL_CM_CAP_HDR_ARRAY_SIZE_MASK, cap_array);

    // 遍历，每个元素 4 个字节，首个DWORD 为头，略过
	for (cap = 1; cap <= cap_count; cap++) {
		void __iomem *register_block;
		u32 hdr;
		int decoder_cnt;
		u16 cap_id, offset;
		u32 length;
        
		hdr = readl(base + cap * 0x4);
        // 读 Capability Header id 与偏移
		cap_id = FIELD_GET(CXL_CM_CAP_HDR_ID_MASK, hdr);
		offset = FIELD_GET(CXL_CM_CAP_PTR_MASK, hdr);
		register_block = base + offset;
        // Ref CXL 2.0 8.2.5.5 CXL HDM Decoder Capability Header
        // 或上图 CXL HDM Decoder Capability Header
        // CXL_CM_CAP_CAP_ID_HDM == 0x5
        // 只处理 HDM Decoder Registers
		switch (cap_id) {
		case CXL_CM_CAP_CAP_ID_HDM:
			dev_dbg(dev, "found HDM decoder capability (0x%x)\n",
				offset);

			hdr = readl(register_block);
            // Ref 上图 CXL HDM Decoder Capability Structure
            // or CXL 2.0 8.2.5.12 CXL HDM Decoder Capability Structure
			decoder_cnt = cxl_hdm_decoder_count(hdr);
			length = 0x20 * decoder_cnt + 0x10;
            // 计算长度以及偏移并记录
			map->hdm_decoder.valid = true;
			map->hdm_decoder.offset = CXL_CM_OFFSET + offset;
			map->hdm_decoder.size = length;
			break;
		default:
			dev_dbg(dev, "Unknown CM cap ID: %d (0x%x)\n", cap_id,
				offset);
			break;
		}
	}
}
EXPORT_SYMBOL_NS_GPL(cxl_probe_component_regs, CXL);

5. CXL Device Capability Header Register

请添加图片描述

/**
 * cxl_probe_device_regs() - Detect CXL Device register blocks
 * @dev: Host device of the @base mapping
 * @base: Mapping of CXL 2.0 8.2.8 CXL Device Register Interface
 * @map: Map object describing the register block information found
 *
 * Probe for device register information and return it in map object.
 */

// 探测 CXL 设备寄存器块，记录位置及大小
// dev : Host 设备
// base : 寄存器块所在位置的基地址，已映射后的虚拟地址
// map : 描述发现的寄存器块信息的对象
void cxl_probe_device_regs(struct device *dev, void __iomem *base,
			   struct cxl_device_reg_map *map)
{
	int cap, cap_count;
	u64 cap_array;

	*map = (struct cxl_device_reg_map){ 0 };
    // 由上图 CXL Subsystem Component Register Ranges 知
    // CXL Device register blocks 偏移为 0 == CXLDEV_CAP_ARRAY_OFFSET
    // Ref CXL 2.0 8.2.8 CXL Device Register Interface
	cap_array = readq(base + CXLDEV_CAP_ARRAY_OFFSET);

    // Cap ID 必须为 0 == CXLDEV_CAP_ARRAY_CAP_ID
    // Ref CXL 2.0 CXL Device Capabilities Array Register : Capability ID
	if (FIELD_GET(CXLDEV_CAP_ARRAY_ID_MASK, cap_array) !=
	    CXLDEV_CAP_ARRAY_CAP_ID)
		return;
    // 同样，要获得元素 Capabilities 数量
	cap_count = FIELD_GET(CXLDEV_CAP_ARRAY_COUNT_MASK, cap_array);
    // 进行遍历，从 1开始，跳过头 1 DWORD
	for (cap = 1; cap <= cap_count; cap++) {
		u32 offset, length;
		u16 cap_id;

        // 每个 Capabilities 长度 16字节 
        // Ref CXL 2.0 8.2.8.1 CXL Device Capabilities Array Register : Capabilities Count
        // 获取 cap_id, offset, lenth
        // Ref CXL 2.0 8.2.8.2 CXL Device Capability Header Register
        // or 上图 CXL Device Capability Header Register

		cap_id = FIELD_GET(CXLDEV_CAP_HDR_CAP_ID_MASK,
				   readl(base + cap * 0x10));
		offset = readl(base + cap * 0x10 + 0x4);
		length = readl(base + cap * 0x10 + 0x8);
        
        // 根据 id 做不同记录处理
        // 不过都是记录偏移和大小以及生效标志
        // Ref CXL 2.0 8.2.8.2.1 CXL Device Capabilities
		switch (cap_id) {
		case CXLDEV_CAP_CAP_ID_DEVICE_STATUS:
			dev_dbg(dev, "found Status capability (0x%x)\n", offset);

			map->status.valid = true;
			map->status.offset = offset;
			map->status.size = length;
			break;
		case CXLDEV_CAP_CAP_ID_PRIMARY_MAILBOX:
			dev_dbg(dev, "found Mailbox capability (0x%x)\n", offset);
			map->mbox.valid = true;
			map->mbox.offset = offset;
			map->mbox.size = length;
			break;
		case CXLDEV_CAP_CAP_ID_SECONDARY_MAILBOX:
			dev_dbg(dev, "found Secondary Mailbox capability (0x%x)\n", offset);
			break;
		case CXLDEV_CAP_CAP_ID_MEMDEV:
			dev_dbg(dev, "found Memory Device capability (0x%x)\n", offset);
			map->memdev.valid = true;
			map->memdev.offset = offset;
			map->memdev.size = length;
			break;
		default:
			if (cap_id >= 0x8000)
				dev_dbg(dev, "Vendor cap ID: %#x offset: %#x\n", cap_id, offset);
			else
				dev_dbg(dev, "Unknown cap ID: %#x offset: %#x\n", cap_id, offset);
			break;
		}
	}
}
EXPORT_SYMBOL_NS_GPL(cxl_probe_device_regs, CXL);

// 地址映射
// 将PCI域地址（或直接理解为物理地址） addr 映射为虚拟地址
void __iomem *devm_cxl_iomap_block(struct device *dev, resource_size_t addr,
				   resource_size_t length)
{
	void __iomem *ret_val;
	struct resource *res;
    // 申请内存区域，告诉内核这块区域我用了，别人在用的话也会失败
	res = devm_request_mem_region(dev, addr, length, dev_name(dev));
	if (!res) {
		resource_size_t end = addr + length - 1;

		dev_err(dev, "Failed to request region %pa-%pa\n", &addr, &end);
		return NULL;
	}
    // 建立 io 映射，返沪虚拟基地址 ret_val
	ret_val = devm_ioremap(dev, addr, length);
	if (!ret_val)
		dev_err(dev, "Failed to map region %pr\n", res);

	return ret_val;
}

// 映射组件寄存器，HDM Decoder 寄存器块
int cxl_map_component_regs(struct pci_dev *pdev,
			   struct cxl_component_regs *regs,
			   struct cxl_register_map *map)
{
	struct device *dev = &pdev->dev;
	resource_size_t phys_addr;
	resource_size_t length;

    // 计算寄存器块所在的bar 空间以及偏移，计算基地址（物理）
	phys_addr = pci_resource_start(pdev, map->barno);
	phys_addr += map->block_offset;

    // 计算 hdm decoder 的基地址（物理）
	phys_addr += map->component_map.hdm_decoder.offset;
	length = map->component_map.hdm_decoder.size;
    // 调用上函数进行映射，虚拟基地址保存在 hdm_decoder 中
	regs->hdm_decoder = devm_cxl_iomap_block(dev, phys_addr, length);
	if (!regs->hdm_decoder)
		return -ENOMEM;

	return 0;
}
EXPORT_SYMBOL_NS_GPL(cxl_map_component_regs, CXL);

// 映射设备寄存器，.io 寄存器块
int cxl_map_device_regs(struct pci_dev *pdev,
			struct cxl_device_regs *regs,
			struct cxl_register_map *map)
{
	struct device *dev = &pdev->dev;
	resource_size_t phys_addr;

    // 计算寄存器块所在的bar 空间以及偏移，计算基地址（物理）
	phys_addr = pci_resource_start(pdev, map->barno);
	phys_addr += map->block_offset;

    // 如果 status 有效， 这个在上面函数中会根据情况置有效
	if (map->device_map.status.valid) {
		resource_size_t addr;
		resource_size_t length;
        // 计算寄存器具体基地址，以及长度
		addr = phys_addr + map->device_map.status.offset;
		length = map->device_map.status.size;
        // 建立映射，返回虚拟基地址，保存，以下相同
		regs->status = devm_cxl_iomap_block(dev, addr, length);
		if (!regs->status)
			return -ENOMEM;
	}
    // 如果 mbox 有效， 这个在上面函数中会根据情况置有效
	if (map->device_map.mbox.valid) {
		resource_size_t addr;
		resource_size_t length;

		addr = phys_addr + map->device_map.mbox.offset;
		length = map->device_map.mbox.size;
		regs->mbox = devm_cxl_iomap_block(dev, addr, length);
		if (!regs->mbox)
			return -ENOMEM;
	}
    // 如果 memdev 有效， 这个在上面函数中会根据情况置有效
	if (map->device_map.memdev.valid) {
		resource_size_t addr;
		resource_size_t length;

		addr = phys_addr + map->device_map.memdev.offset;
		length = map->device_map.memdev.size;
		regs->memdev = devm_cxl_iomap_block(dev, addr, length);
		if (!regs->memdev)
			return -ENOMEM;
	}

	return 0;
}
EXPORT_SYMBOL_NS_GPL(cxl_map_device_regs, CXL);

6. Register Offset Low

请添加图片描述

// 解码寄存器块
// Ref 上图 Register Offset Low
// or CXL 2.0 SPec 8.1.9.1 Register Offset Low

static void cxl_decode_regblock(u32 reg_lo, u32 reg_hi,
				struct cxl_register_map *map)
{
    // reg hi代表偏移的高位，与 reg low 31：16 结合
	map->block_offset = ((u64)reg_hi << 32) |
			    (reg_lo & CXL_DVSEC_REG_LOCATOR_BLOCK_OFF_LOW_MASK);
    // 2:0 表示寄存器在哪一个 bar : 0 - 5
	map->barno = FIELD_GET(CXL_DVSEC_REG_LOCATOR_BIR_MASK, reg_lo);
    // 表示寄存器的类型 15:8
	map->reg_type = FIELD_GET(CXL_DVSEC_REG_LOCATOR_BLOCK_ID_MASK, reg_lo);
}

7. Register Locator DVSEC - Header

请添加图片描述

8. Register Locator DVSEC with 3 Register Block Entries

请添加图片描述

/**
 * cxl_find_regblock() - Locate register blocks by type
 * @pdev: The CXL PCI device to enumerate.
 * @type: Register Block Indicator id
 * @map: Enumeration output, clobbered on error
 *
 * Return: 0 if register block enumerated, negative error code otherwise
 *
 * A CXL DVSEC may point to one or more register blocks, search for them
 * by @type.
 */

// 通过 Type 定位寄存器块
int cxl_find_regblock(struct pci_dev *pdev, enum cxl_regloc_type type,
		      struct cxl_register_map *map)
{
	u32 regloc_size, regblocks;
	int regloc, i;

	map->block_offset = U64_MAX;
    // 在 PCIe capabilities 中寻找 VENDOR_ID 为 PCI_DVSEC_VENDOR_ID_CXL == 0x23
    // 0x23 表示此  capabilities 是 DVSEC 
    // 然后根据 DVSEC ID 判断是哪一个 DVSEC 类型
    // CXL_DVSEC_REG_LOCATOR == 8 表示寄存器定位
    // Ref CXL 2.0 Table 124. CXL DVSEC ID Assignment
    // or 上面的图 Table 124
	regloc = pci_find_dvsec_capability(pdev, PCI_DVSEC_VENDOR_ID_CXL,
					   CXL_DVSEC_REG_LOCATOR);
	if (!regloc)
		return -ENXIO;

    // Ref 上图 Register Locator DVSEC - Header
    // or CXL 2.0 Table 131. Register Locator DVSEC - Header
    // 获取 DVSEC Length 31:20
	pci_read_config_dword(pdev, regloc + PCI_DVSEC_HEADER1, &regloc_size);
	regloc_size = FIELD_GET(PCI_DVSEC_HEADER1_LENGTH_MASK, regloc_size);

    // CXL_DVSEC_REG_LOCATOR_BLOCK1_OFFSET 0xc
    // 每个寄存器块用 8 字节, 根据总大小，计算寄存器块的个数
    // Refer 上图 Register Locator DVSEC with 3 Register Block Entries ，标黄部分
    // or CXL 2.0 Figure 131. Register Locator DVSEC with 3 Register Block Entries

	regloc += CXL_DVSEC_REG_LOCATOR_BLOCK1_OFFSET;
	regblocks = (regloc_size - CXL_DVSEC_REG_LOCATOR_BLOCK1_OFFSET) / 8;

    // 对寄存器遍历，比较寄存器类型，匹配返回 0
	for (i = 0; i < regblocks; i++, regloc += 8) {
		u32 reg_lo, reg_hi;

		pci_read_config_dword(pdev, regloc, &reg_lo);
		pci_read_config_dword(pdev, regloc + 4, &reg_hi);

		cxl_decode_regblock(reg_lo, reg_hi, map);

		if (map->reg_type == type)
			return 0;
	}

	map->block_offset = U64_MAX;
	return -ENODEV;
}
EXPORT_SYMBOL_NS_GPL(cxl_find_regblock, CXL);