Re: [PATCH v4 2/2] EDAC: al-mc-edac: Introduce Amazon's Annapurna Labs Memory Controller EDAC

From: Marc Zyngier
Date: Fri Oct 11 2019 - 04:23:25 EST


On Thu, 10 Oct 2019 14:54:14 +0300
Talel Shenhar <talel@xxxxxxxxxx> wrote:

Hi Talel,

> The Amazon's Annapurna Labs Memory Controller EDAC supports ECC capability
> for error detection and correction (Single bit error correction, Double
> detection). This driver introduces EDAC driver for that capability.
>
> Signed-off-by: Talel Shenhar <talel@xxxxxxxxxx>
> ---
> MAINTAINERS | 7 +
> drivers/edac/Kconfig | 7 +
> drivers/edac/Makefile | 1 +
> drivers/edac/al_mc_edac.c | 358 ++++++++++++++++++++++++++++++++++++++++++++++
> 4 files changed, 373 insertions(+)
> create mode 100644 drivers/edac/al_mc_edac.c
>
> diff --git a/MAINTAINERS b/MAINTAINERS
> index 55199ef..d431e8f 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -757,6 +757,13 @@ F: drivers/tty/serial/altera_jtaguart.c
> F: include/linux/altera_uart.h
> F: include/linux/altera_jtaguart.h
>
> +AMAZON ANNAPURNA LABS MEMORY CONTROLLER EDAC
> +M: Talel Shenhar <talel@xxxxxxxxxx>
> +M: Talel Shenhar <talelshenhar@xxxxxxxxx>
> +S: Maintained
> +F: Documentation/devicetree/bindings/edac/amazon,al-mc-edac.yaml
> +F: drivers/edac/al_mc_edac.c
> +
> AMAZON ANNAPURNA LABS THERMAL MMIO DRIVER
> M: Talel Shenhar <talel@xxxxxxxxxx>
> S: Maintained
> diff --git a/drivers/edac/Kconfig b/drivers/edac/Kconfig
> index 417dad6..8c7fb73 100644
> --- a/drivers/edac/Kconfig
> +++ b/drivers/edac/Kconfig
> @@ -100,6 +100,13 @@ config EDAC_AMD64_ERROR_INJECTION
> In addition, there are two control files, inject_read and inject_write,
> which trigger the DRAM ECC Read and Write respectively.
>
> +config EDAC_AL_MC
> + tristate "Amazon's Annapurna Lab EDAC Memory Controller"
> + depends on (ARCH_ALPINE || COMPILE_TEST)
> + help
> + Support for error detection and correction for Amazon's Annapurna
> + Labs Alpine chips which allows 1 bit correction and 2 bits detection.
> +
> config EDAC_AMD76X
> tristate "AMD 76x (760, 762, 768)"
> depends on PCI && X86_32
> diff --git a/drivers/edac/Makefile b/drivers/edac/Makefile
> index d77200c..5288329 100644
> --- a/drivers/edac/Makefile
> +++ b/drivers/edac/Makefile
> @@ -22,6 +22,7 @@ obj-$(CONFIG_EDAC_GHES) += ghes_edac.o
> edac_mce_amd-y := mce_amd.o
> obj-$(CONFIG_EDAC_DECODE_MCE) += edac_mce_amd.o
>
> +obj-$(CONFIG_EDAC_AL_MC) += al_mc_edac.o
> obj-$(CONFIG_EDAC_AMD76X) += amd76x_edac.o
> obj-$(CONFIG_EDAC_CPC925) += cpc925_edac.o
> obj-$(CONFIG_EDAC_I5000) += i5000_edac.o
> diff --git a/drivers/edac/al_mc_edac.c b/drivers/edac/al_mc_edac.c
> new file mode 100644
> index 00000000..f1288c4
> --- /dev/null
> +++ b/drivers/edac/al_mc_edac.c
> @@ -0,0 +1,358 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Copyright 2019 Amazon.com, Inc. or its affiliates. All Rights Reserved.
> + */
> +#include <linux/bitfield.h>
> +#include <linux/bitops.h>
> +#include <linux/edac.h>
> +#include <linux/of_irq.h>
> +#include <linux/platform_device.h>
> +#include <linux/spinlock.h>
> +#include "edac_module.h"
> +
> +/* Registers Offset */
> +#define AL_MC_MSTR 0x00
> +#define AL_MC_ECC_CFG 0x70
> +#define AL_MC_ECC_CLEAR 0x7c
> +#define AL_MC_ECC_ERR_COUNT 0x80
> +#define AL_MC_ECC_CE_ADDR0 0x84
> +#define AL_MC_ECC_CE_ADDR1 0x88
> +#define AL_MC_ECC_UE_ADDR0 0xa4
> +#define AL_MC_ECC_UE_ADDR1 0xa8
> +#define AL_MC_ECC_CE_SYND0 0x8c
> +#define AL_MC_ECC_CE_SYND1 0x90
> +#define AL_MC_ECC_CE_SYND2 0x94
> +#define AL_MC_ECC_UE_SYND0 0xac
> +#define AL_MC_ECC_UE_SYND1 0xb0
> +#define AL_MC_ECC_UE_SYND2 0xb4
> +
> +/* Registers Fields */
> +#define AL_MC_MSTR_DEV_CFG GENMASK(31, 30)
> +#define AL_MC_MSTR_RANKS GENMASK(27, 24)
> +#define AL_MC_MSTR_DATA_BUS_WIDTH GENMASK(13, 12)
> +#define AL_MC_MSTR_DDR4 BIT(4)
> +#define AL_MC_MSTR_DDR3 BIT(0)
> +
> +#define AL_MC_ECC_CFG_SCRUB_DISABLED BIT(4)
> +#define AL_MC_ECC_CFG_ECC_MODE GENMASK(2, 0)
> +
> +#define AL_MC_ECC_CLEAR_UE_COUNT BIT(3)
> +#define AL_MC_ECC_CLEAR_CE_COUNT BIT(2)
> +#define AL_MC_ECC_CLEAR_UE_ERR BIT(1)
> +#define AL_MC_ECC_CLEAR_CE_ERR BIT(0)
> +
> +#define AL_MC_ECC_ERR_COUNT_UE GENMASK(31, 16)
> +#define AL_MC_ECC_ERR_COUNT_CE GENMASK(15, 0)
> +
> +#define AL_MC_ECC_CE_ADDR0_RANK GENMASK(25, 24)
> +#define AL_MC_ECC_CE_ADDR0_ROW GENMASK(17, 0)
> +
> +#define AL_MC_ECC_CE_ADDR1_BG GENMASK(25, 24)
> +#define AL_MC_ECC_CE_ADDR1_BANK GENMASK(18, 16)
> +#define AL_MC_ECC_CE_ADDR1_COLUMN GENMASK(11, 0)
> +
> +#define AL_MC_ECC_UE_ADDR0_RANK GENMASK(25, 24)
> +#define AL_MC_ECC_UE_ADDR0_ROW GENMASK(17, 0)
> +
> +#define AL_MC_ECC_UE_ADDR1_BG GENMASK(25, 24)
> +#define AL_MC_ECC_UE_ADDR1_BANK GENMASK(18, 16)
> +#define AL_MC_ECC_UE_ADDR1_COLUMN GENMASK(11, 0)
> +
> +/* Registers Values */
> +#define AL_MC_MSTR_DEV_CFG_X4 0
> +#define AL_MC_MSTR_DEV_CFG_X8 1
> +#define AL_MC_MSTR_DEV_CFG_X16 2
> +#define AL_MC_MSTR_DEV_CFG_X32 3
> +#define AL_MC_MSTR_RANKS_MAX 4
> +#define AL_MC_MSTR_DATA_BUS_WIDTH_X64 0
> +
> +#define DRV_NAME "al_mc_edac"
> +#define AL_MC_EDAC_MSG_MAX 256
> +#define AL_MC_EDAC_MSG(message, buffer_size, type, \
> + rank, row, bg, bank, column, syn0, syn1, syn2) \
> + snprintf(message, buffer_size, \
> + "%s rank=0x%x row=0x%x bg=0x%x bank=0x%x col=0x%x " \
> + "syn0: 0x%x syn1: 0x%x syn2: 0x%x", \
> + type == HW_EVENT_ERR_UNCORRECTED ? "UE" : "CE", \
> + rank, row, bg, bank, column, syn0, syn1, syn2)
> +
> +struct al_mc_edac {
> + void __iomem *mmio_base;
> + spinlock_t lock;
> + int irq_ce;
> + int irq_ue;
> +};
> +
> +static int al_mc_edac_handle_ce(struct mem_ctl_info *mci)
> +{
> + struct al_mc_edac *al_mc = mci->pvt_info;
> + u32 eccerrcnt, ecccaddr0, ecccaddr1, ecccsyn0, ecccsyn1, ecccsyn2, row;
> + u16 ce_count, column;
> + u8 rank, bg, bank;
> + char msg[AL_MC_EDAC_MSG_MAX];
> + unsigned long flags;
> +
> + eccerrcnt = readl_relaxed(al_mc->mmio_base + AL_MC_ECC_ERR_COUNT);
> + ce_count = FIELD_GET(AL_MC_ECC_ERR_COUNT_CE, eccerrcnt);
> + if (!ce_count)
> + return 0;
> +
> + ecccaddr0 = readl_relaxed(al_mc->mmio_base + AL_MC_ECC_CE_ADDR0);
> + ecccaddr1 = readl_relaxed(al_mc->mmio_base + AL_MC_ECC_CE_ADDR1);
> + ecccsyn0 = readl_relaxed(al_mc->mmio_base + AL_MC_ECC_UE_SYND0);
> + ecccsyn1 = readl_relaxed(al_mc->mmio_base + AL_MC_ECC_UE_SYND1);
> + ecccsyn2 = readl_relaxed(al_mc->mmio_base + AL_MC_ECC_UE_SYND2);
> +
> + writel(AL_MC_ECC_CLEAR_CE_COUNT | AL_MC_ECC_CLEAR_CE_ERR,
> + al_mc->mmio_base + AL_MC_ECC_CLEAR);

Why the non-relaxed write? All the accesses are ordered by virtue of
interacting with the same device, and I can't see anything that
requires ordering with normal memory accesses prior to this write.

This pattern appears throughout the driver, and I think they can all be
relaxed. If not, please document the reason why they can't.

> +
> + dev_dbg(mci->pdev, "eccuaddr0=0x%08x eccuaddr1=0x%08x\n",
> + ecccaddr0, ecccaddr1);
> +
> + rank = FIELD_GET(AL_MC_ECC_CE_ADDR0_RANK, ecccaddr0);
> + row = FIELD_GET(AL_MC_ECC_CE_ADDR0_ROW, ecccaddr0);
> +
> + bg = FIELD_GET(AL_MC_ECC_CE_ADDR1_BG, ecccaddr1);
> + bank = FIELD_GET(AL_MC_ECC_CE_ADDR1_BANK, ecccaddr1);
> + column = FIELD_GET(AL_MC_ECC_CE_ADDR1_COLUMN, ecccaddr1);
> +
> + AL_MC_EDAC_MSG(msg, sizeof(msg), HW_EVENT_ERR_CORRECTED,
> + rank, row, bg, bank, column,
> + ecccsyn0, ecccsyn1, ecccsyn2);
> +
> + spin_lock_irqsave(&al_mc->lock, flags);
> + edac_mc_handle_error(HW_EVENT_ERR_CORRECTED, mci,
> + ce_count, 0, 0, 0, 0, 0, -1, mci->ctl_name, msg);
> + spin_unlock_irqrestore(&al_mc->lock, flags);
> +
> + return ce_count;
> +}
> +
> +static int al_mc_edac_handle_ue(struct mem_ctl_info *mci)
> +{
> + struct al_mc_edac *al_mc = mci->pvt_info;
> + u32 eccerrcnt, eccuaddr0, eccuaddr1, eccusyn0, eccusyn1, eccusyn2, row;
> + u16 ue_count, column;
> + u8 rank, bg, bank;
> + char msg[AL_MC_EDAC_MSG_MAX];
> + unsigned long flags;
> +
> + eccerrcnt = readl_relaxed(al_mc->mmio_base + AL_MC_ECC_ERR_COUNT);
> + ue_count = FIELD_GET(AL_MC_ECC_ERR_COUNT_UE, eccerrcnt);
> + if (!ue_count)
> + return 0;
> +
> + eccuaddr0 = readl_relaxed(al_mc->mmio_base + AL_MC_ECC_UE_ADDR0);
> + eccuaddr1 = readl_relaxed(al_mc->mmio_base + AL_MC_ECC_UE_ADDR1);
> + eccusyn0 = readl_relaxed(al_mc->mmio_base + AL_MC_ECC_UE_SYND0);
> + eccusyn1 = readl_relaxed(al_mc->mmio_base + AL_MC_ECC_UE_SYND1);
> + eccusyn2 = readl_relaxed(al_mc->mmio_base + AL_MC_ECC_UE_SYND2);
> +
> + writel(AL_MC_ECC_CLEAR_UE_COUNT | AL_MC_ECC_CLEAR_UE_ERR,
> + al_mc->mmio_base + AL_MC_ECC_CLEAR);
> +
> + dev_dbg(mci->pdev, "eccuaddr0=0x%08x eccuaddr1=0x%08x\n",
> + eccuaddr0, eccuaddr1);
> +
> + rank = FIELD_GET(AL_MC_ECC_UE_ADDR0_RANK, eccuaddr0);
> + row = FIELD_GET(AL_MC_ECC_UE_ADDR0_ROW, eccuaddr0);
> +
> + bg = FIELD_GET(AL_MC_ECC_UE_ADDR1_BG, eccuaddr1);
> + bank = FIELD_GET(AL_MC_ECC_UE_ADDR1_BANK, eccuaddr1);
> + column = FIELD_GET(AL_MC_ECC_UE_ADDR1_COLUMN, eccuaddr1);
> +
> + AL_MC_EDAC_MSG(msg, sizeof(msg), HW_EVENT_ERR_UNCORRECTED,
> + rank, row, bg, bank, column,
> + eccusyn0, eccusyn1, eccusyn2);
> +
> + spin_lock_irqsave(&al_mc->lock, flags);
> + edac_mc_handle_error(HW_EVENT_ERR_UNCORRECTED, mci,
> + ue_count, 0, 0, 0, 0, 0, -1, mci->ctl_name, msg);
> + spin_unlock_irqrestore(&al_mc->lock, flags);
> +
> + return ue_count;
> +}
> +
> +static void al_mc_edac_check(struct mem_ctl_info *mci)
> +{
> + struct al_mc_edac *al_mc = mci->pvt_info;
> +
> + if (al_mc->irq_ue <= 0)
> + al_mc_edac_handle_ue(mci);
> +
> + if (al_mc->irq_ce <= 0)
> + al_mc_edac_handle_ce(mci);
> +}
> +
> +static irqreturn_t al_mc_edac_irq_handler_ue(int irq, void *info)
> +{
> + struct platform_device *pdev = info;
> + struct mem_ctl_info *mci = platform_get_drvdata(pdev);
> +
> + if (al_mc_edac_handle_ue(mci))
> + return IRQ_HANDLED;
> + return IRQ_NONE;
> +}
> +
> +static irqreturn_t al_mc_edac_irq_handler_ce(int irq, void *info)
> +{
> + struct platform_device *pdev = info;
> + struct mem_ctl_info *mci = platform_get_drvdata(pdev);
> +
> + if (al_mc_edac_handle_ce(mci))
> + return IRQ_HANDLED;
> + return IRQ_NONE;
> +}
> +
> +static enum scrub_type al_mc_edac_get_scrub_mode(void __iomem *mmio_base)
> +{
> + u32 ecccfg0;
> +
> + ecccfg0 = readl(mmio_base + AL_MC_ECC_CFG);
> +
> + if (FIELD_GET(AL_MC_ECC_CFG_SCRUB_DISABLED, ecccfg0))
> + return SCRUB_NONE;
> + else
> + return SCRUB_HW_SRC;
> +}
> +
> +static int al_mc_edac_probe(struct platform_device *pdev)
> +{
> + void __iomem *mmio_base;
> + struct edac_mc_layer layers[1];
> + struct mem_ctl_info *mci;
> + struct al_mc_edac *al_mc;
> + int ret;
> +
> + mmio_base = devm_platform_ioremap_resource(pdev, 0);
> + if (IS_ERR(mmio_base)) {
> + dev_err(&pdev->dev, "failed to ioremap memory (%ld)\n",
> + PTR_ERR(mmio_base));
> + return PTR_ERR(mmio_base);
> + }
> +
> + layers[0].type = EDAC_MC_LAYER_CHIP_SELECT;
> + layers[0].size = 1;
> + layers[0].is_virt_csrow = false;
> + mci = edac_mc_alloc(0, ARRAY_SIZE(layers), layers,
> + sizeof(struct al_mc_edac));
> + if (!mci)
> + return -ENOMEM;
> +
> + platform_set_drvdata(pdev, mci);
> + al_mc = mci->pvt_info;
> +
> + al_mc->mmio_base = mmio_base;
> +
> + al_mc->irq_ue = of_irq_get_byname(pdev->dev.of_node, "ue");
> + if (al_mc->irq_ue <= 0)
> + dev_dbg(&pdev->dev,
> + "no irq defined for ue - falling back to polling\n");
> +
> + al_mc->irq_ce = of_irq_get_byname(pdev->dev.of_node, "ce");
> + if (al_mc->irq_ce <= 0)
> + dev_dbg(&pdev->dev,
> + "no irq defined for ce - falling back to polling\n");
> +
> + if (al_mc->irq_ue <= 0 || al_mc->irq_ce <= 0)
> + edac_op_state = EDAC_OPSTATE_POLL;
> + else
> + edac_op_state = EDAC_OPSTATE_INT;
> +
> + spin_lock_init(&al_mc->lock);
> +
> + mci->edac_check = al_mc_edac_check;
> + mci->mtype_cap = MEM_FLAG_DDR3 | MEM_FLAG_DDR4;
> + mci->edac_ctl_cap = EDAC_FLAG_NONE | EDAC_FLAG_SECDED;
> + mci->edac_cap = EDAC_FLAG_SECDED;
> + mci->mod_name = DRV_NAME;
> + mci->ctl_name = "al_mc";
> + mci->pdev = &pdev->dev;
> + mci->scrub_mode = al_mc_edac_get_scrub_mode(mmio_base);
> +
> + ret = edac_mc_add_mc(mci);
> + if (ret < 0) {
> + dev_err(&pdev->dev,
> + "fail to add memory controller device (%d)\n",
> + ret);
> + goto err_add_mc;
> + }
> +
> + if (al_mc->irq_ue > 0) {
> + ret = devm_request_irq(&pdev->dev,
> + al_mc->irq_ue,
> + al_mc_edac_irq_handler_ue,
> + IRQF_SHARED,
> + pdev->name,
> + pdev);
> + if (ret != 0) {
> + dev_err(&pdev->dev,
> + "failed to request ue irq %d (%d)\n",
> + al_mc->irq_ue, ret);
> + goto err_request_irq;
> + }
> + }
> +
> + if (al_mc->irq_ce > 0) {
> + ret = devm_request_irq(&pdev->dev,
> + al_mc->irq_ce,
> + al_mc_edac_irq_handler_ce,
> + IRQF_SHARED,
> + pdev->name,
> + pdev);
> + if (ret != 0) {
> + dev_err(&pdev->dev,
> + "failed to request ce irq %d (%d)\n",
> + al_mc->irq_ce, ret);
> + goto err_request_irq;
> + }
> + }
> +
> + return 0;
> +
> +err_request_irq:
> + edac_mc_del_mc(&pdev->dev);
> +err_add_mc:
> + edac_mc_free(mci);
> +
> + return ret;
> +}
> +
> +static int al_mc_edac_remove(struct platform_device *pdev)
> +{
> + struct mem_ctl_info *mci = platform_get_drvdata(pdev);
> + struct al_mc_edac *al_mc = mci->pvt_info;
> +
> + if (al_mc->irq_ue > 0)
> + devm_free_irq(&pdev->dev, al_mc->irq_ue, pdev);
> +
> + if (al_mc->irq_ce > 0)
> + devm_free_irq(&pdev->dev, al_mc->irq_ce, pdev);
> +
> + edac_mc_del_mc(&pdev->dev);
> + edac_mc_free(mci);

nit: I always find the mix of devm_* stuff with non-devm pretty
awkward, as you end-up tracking things anyway (here, and in the error
path of the probe function).

I found the use of devm_add_action() to be pretty good at decluttering
probe and remove entry points, as you just register destructors each
time you allocate objects, and can forget any form of subsequent
tracking.

> +
> + return 0;
> +}
> +
> +static const struct of_device_id al_mc_edac_of_match[] = {
> + { .compatible = "amazon,al-mc-edac", },
> + {},
> +};
> +
> +MODULE_DEVICE_TABLE(of, al_mc_edac_of_match);
> +
> +static struct platform_driver al_mc_edac_driver = {
> + .probe = al_mc_edac_probe,
> + .remove = al_mc_edac_remove,
> + .driver = {
> + .name = DRV_NAME,
> + .of_match_table = al_mc_edac_of_match,
> + },
> +};
> +
> +module_platform_driver(al_mc_edac_driver);
> +
> +MODULE_LICENSE("GPL v2");
> +MODULE_AUTHOR("Talel Shenhar");
> +MODULE_DESCRIPTION("Amazon's Annapurna Lab's Memory Controller EDAC Driver");

Thanks,

M.
--
Jazz is not dead. It just smells funny...