[PATCH] EDAC: drivers for AMD 76x and Intel E750x, E752x

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

1 files changed, 356 insertions, 0 deletions

diff --git a/drivers/edac/amd76x_edac.c b/drivers/edac/amd76x_edac.c
new file mode 100644
index 000000000000..8e2b1295e70c
--- /dev/null
+++ b/drivers/edac/amd76x_edac.c
@@ -0,0 +1,356 @@
+/*
+ * AMD 76x Memory Controller kernel module
+ * (C) 2003 Linux Networx (http://lnxi.com)
+ * This file may be distributed under the terms of the
+ * GNU General Public License.
+ *
+ * Written by Thayne Harbaugh
+ * Based on work by Dan Hollis <goemon at anime dot net> and others.
+ *      http://www.anime.net/~goemon/linux-ecc/
+ *
+ * $Id: edac_amd76x.c,v 1.4.2.5 2005/10/05 00:43:44 dsp_llnl Exp $
+ *
+ */
+
+
+#include <linux/config.h>
+#include <linux/module.h>
+#include <linux/init.h>
+
+#include <linux/pci.h>
+#include <linux/pci_ids.h>
+
+#include <linux/slab.h>
+
+#include "edac_mc.h"
+
+
+#define AMD76X_NR_CSROWS 8
+#define AMD76X_NR_CHANS  1
+#define AMD76X_NR_DIMMS  4
+
+
+/* AMD 76x register addresses - device 0 function 0 - PCI bridge */
+#define AMD76X_ECC_MODE_STATUS  0x48    /* Mode and status of ECC (32b)
+                                         *
+                                         * 31:16 reserved
+                                         * 15:14 SERR enabled: x1=ue 1x=ce
+                                         * 13    reserved
+                                         * 12    diag: disabled, enabled
+                                         * 11:10 mode: dis, EC, ECC, ECC+scrub
+                                         *  9:8  status: x1=ue 1x=ce
+                                         *  7:4  UE cs row
+                                         *  3:0  CE cs row
+                                         */
+#define AMD76X_DRAM_MODE_STATUS 0x58    /* DRAM Mode and status (32b)
+                                         *
+                                         * 31:26 clock disable 5 - 0
+                                         * 25    SDRAM init
+                                         * 24    reserved
+                                         * 23    mode register service
+                                         * 22:21 suspend to RAM
+                                         * 20    burst refresh enable
+                                         * 19    refresh disable
+                                         * 18    reserved
+                                         * 17:16 cycles-per-refresh
+                                         * 15:8  reserved
+                                         *  7:0  x4 mode enable 7 - 0
+                                         */
+#define AMD76X_MEM_BASE_ADDR    0xC0    /* Memory base address (8 x 32b)
+                                         *
+                                         * 31:23 chip-select base
+                                         * 22:16 reserved
+                                         * 15:7  chip-select mask
+                                         *  6:3  reserved
+                                         *  2:1  address mode
+                                         *  0    chip-select enable
+                                         */
+
+
+struct amd76x_error_info {
+        u32 ecc_mode_status;
+};
+
+
+enum amd76x_chips {
+        AMD761 = 0,
+        AMD762
+};
+
+
+struct amd76x_dev_info {
+        const char *ctl_name;
+};
+
+
+static const struct amd76x_dev_info amd76x_devs[] = {
+        [AMD761] = {.ctl_name = "AMD761"},
+        [AMD762] = {.ctl_name = "AMD762"},
+};
+
+
+/**
+ *      amd76x_get_error_info   -       fetch error information
+ *      @mci: Memory controller
+ *      @info: Info to fill in
+ *
+ *      Fetch and store the AMD76x ECC status. Clear pending status
+ *      on the chip so that further errors will be reported
+ */
+
+static void amd76x_get_error_info (struct mem_ctl_info *mci,
+                                   struct amd76x_error_info *info)
+{
+        pci_read_config_dword(mci->pdev, AMD76X_ECC_MODE_STATUS,
+                                &info->ecc_mode_status);
+
+        if (info->ecc_mode_status & BIT(8))
+                pci_write_bits32(mci->pdev, AMD76X_ECC_MODE_STATUS,
+                                   (u32) BIT(8), (u32) BIT(8));
+
+        if (info->ecc_mode_status & BIT(9))
+                pci_write_bits32(mci->pdev, AMD76X_ECC_MODE_STATUS,
+                                   (u32) BIT(9), (u32) BIT(9));
+}
+
+
+/**
+ *      amd76x_process_error_info       -       Error check
+ *      @mci: Memory controller
+ *      @info: Previously fetched information from chip
+ *      @handle_errors: 1 if we should do recovery
+ *
+ *      Process the chip state and decide if an error has occurred.
+ *      A return of 1 indicates an error. Also if handle_errors is true
+ *      then attempt to handle and clean up after the error
+ */
+
+static int amd76x_process_error_info (struct mem_ctl_info *mci,
+                struct amd76x_error_info *info, int handle_errors)
+{
+        int error_found;
+        u32 row;
+
+        error_found = 0;
+
+        /*
+         *      Check for an uncorrectable error
+         */
+        if (info->ecc_mode_status & BIT(8)) {
+                error_found = 1;
+
+                if (handle_errors) {
+                        row = (info->ecc_mode_status >> 4) & 0xf;
+                        edac_mc_handle_ue(mci,
+                            mci->csrows[row].first_page, 0, row,
+                            mci->ctl_name);
+                }
+        }
+
+        /*
+         *      Check for a correctable error
+         */
+        if (info->ecc_mode_status & BIT(9)) {
+                error_found = 1;
+
+                if (handle_errors) {
+                        row = info->ecc_mode_status & 0xf;
+                        edac_mc_handle_ce(mci,
+                            mci->csrows[row].first_page, 0, 0, row, 0,
+                            mci->ctl_name);
+                }
+        }
+        return error_found;
+}
+
+/**
+ *      amd76x_check    -       Poll the controller
+ *      @mci: Memory controller
+ *
+ *      Called by the poll handlers this function reads the status
+ *      from the controller and checks for errors.
+ */
+
+static void amd76x_check(struct mem_ctl_info *mci)
+{
+        struct amd76x_error_info info;
+        debugf3("MC: " __FILE__ ": %s()\n", __func__);
+        amd76x_get_error_info(mci, &info);
+        amd76x_process_error_info(mci, &info, 1);
+}
+
+
+/**
+ *      amd76x_probe1   -       Perform set up for detected device
+ *      @pdev; PCI device detected
+ *      @dev_idx: Device type index
+ *
+ *      We have found an AMD76x and now need to set up the memory
+ *      controller status reporting. We configure and set up the
+ *      memory controller reporting and claim the device.
+ */
+
+static int amd76x_probe1(struct pci_dev *pdev, int dev_idx)
+{
+        int rc = -ENODEV;
+        int index;
+        struct mem_ctl_info *mci = NULL;
+        enum edac_type ems_modes[] = {
+                EDAC_NONE,
+                EDAC_EC,
+                EDAC_SECDED,
+                EDAC_SECDED
+        };
+        u32 ems;
+        u32 ems_mode;
+
+        debugf0("MC: " __FILE__ ": %s()\n", __func__);
+
+        pci_read_config_dword(pdev, AMD76X_ECC_MODE_STATUS, &ems);
+        ems_mode = (ems >> 10) & 0x3;
+
+        mci = edac_mc_alloc(0, AMD76X_NR_CSROWS, AMD76X_NR_CHANS);
+
+        if (mci == NULL) {
+                rc = -ENOMEM;
+                goto fail;
+        }
+
+        debugf0("MC: " __FILE__ ": %s(): mci = %p\n", __func__, mci);
+
+        mci->pdev = pci_dev_get(pdev);
+        mci->mtype_cap = MEM_FLAG_RDDR;
+
+        mci->edac_ctl_cap = EDAC_FLAG_NONE | EDAC_FLAG_EC | EDAC_FLAG_SECDED;
+        mci->edac_cap = ems_mode ?
+            (EDAC_FLAG_EC | EDAC_FLAG_SECDED) : EDAC_FLAG_NONE;
+
+        mci->mod_name = BS_MOD_STR;
+        mci->mod_ver = "$Revision: 1.4.2.5 $";
+        mci->ctl_name = amd76x_devs[dev_idx].ctl_name;
+        mci->edac_check = amd76x_check;
+        mci->ctl_page_to_phys = NULL;
+
+        for (index = 0; index < mci->nr_csrows; index++) {
+                struct csrow_info *csrow = &mci->csrows[index];
+                u32 mba;
+                u32 mba_base;
+                u32 mba_mask;
+                u32 dms;
+
+                /* find the DRAM Chip Select Base address and mask */
+                pci_read_config_dword(mci->pdev,
+                                      AMD76X_MEM_BASE_ADDR + (index * 4),
+                                      &mba);
+
+                if (!(mba & BIT(0)))
+                        continue;
+
+                mba_base = mba & 0xff800000UL;
+                mba_mask = ((mba & 0xff80) << 16) | 0x7fffffUL;
+
+                pci_read_config_dword(mci->pdev, AMD76X_DRAM_MODE_STATUS,
+                                      &dms);
+
+                csrow->first_page = mba_base >> PAGE_SHIFT;
+                csrow->nr_pages = (mba_mask + 1) >> PAGE_SHIFT;
+                csrow->last_page = csrow->first_page + csrow->nr_pages - 1;
+                csrow->page_mask = mba_mask >> PAGE_SHIFT;
+                csrow->grain = csrow->nr_pages << PAGE_SHIFT;
+                csrow->mtype = MEM_RDDR;
+                csrow->dtype = ((dms >> index) & 0x1) ? DEV_X4 : DEV_UNKNOWN;
+                csrow->edac_mode = ems_modes[ems_mode];
+        }
+
+        /* clear counters */
+        pci_write_bits32(mci->pdev, AMD76X_ECC_MODE_STATUS, (u32) (0x3 << 8),
+                         (u32) (0x3 << 8));
+
+        if (edac_mc_add_mc(mci)) {
+                debugf3("MC: " __FILE__
+                        ": %s(): failed edac_mc_add_mc()\n", __func__);
+                goto fail;
+        }
+
+        /* get this far and it's successful */
+        debugf3("MC: " __FILE__ ": %s(): success\n", __func__);
+        return 0;
+
+fail:
+        if (mci) {
+                if(mci->pdev)
+                        pci_dev_put(mci->pdev);
+                edac_mc_free(mci);
+        }
+        return rc;
+}
+
+/* returns count (>= 0), or negative on error */
+static int __devinit amd76x_init_one(struct pci_dev *pdev,
+                                     const struct pci_device_id *ent)
+{
+        debugf0("MC: " __FILE__ ": %s()\n", __func__);
+
+        /* don't need to call pci_device_enable() */
+        return amd76x_probe1(pdev, ent->driver_data);
+}
+
+
+/**
+ *      amd76x_remove_one       -       driver shutdown
+ *      @pdev: PCI device being handed back
+ *
+ *      Called when the driver is unloaded. Find the matching mci
+ *      structure for the device then delete the mci and free the
+ *      resources.
+ */
+
+static void __devexit amd76x_remove_one(struct pci_dev *pdev)
+{
+        struct mem_ctl_info *mci;
+
+        debugf0(__FILE__ ": %s()\n", __func__);
+
+        if ((mci = edac_mc_find_mci_by_pdev(pdev)) == NULL)
+                return;
+        if (edac_mc_del_mc(mci))
+                return;
+        pci_dev_put(mci->pdev);
+        edac_mc_free(mci);
+}
+
+
+static const struct pci_device_id amd76x_pci_tbl[] __devinitdata = {
+        {PCI_VEND_DEV(AMD, FE_GATE_700C), PCI_ANY_ID, PCI_ANY_ID, 0, 0,
+         AMD762},
+        {PCI_VEND_DEV(AMD, FE_GATE_700E), PCI_ANY_ID, PCI_ANY_ID, 0, 0,
+         AMD761},
+        {0,}                    /* 0 terminated list. */
+};
+
+MODULE_DEVICE_TABLE(pci, amd76x_pci_tbl);
+
+
+static struct pci_driver amd76x_driver = {
+        .name = BS_MOD_STR,
+        .probe = amd76x_init_one,
+        .remove = __devexit_p(amd76x_remove_one),
+        .id_table = amd76x_pci_tbl,
+};
+
+int __init amd76x_init(void)
+{
+        return pci_register_driver(&amd76x_driver);
+}
+
+static void __exit amd76x_exit(void)
+{
+        pci_unregister_driver(&amd76x_driver);
+}
+
+module_init(amd76x_init);
+module_exit(amd76x_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Linux Networx (http://lnxi.com) Thayne Harbaugh");
+MODULE_DESCRIPTION("MC support for AMD 76x memory controllers");