From: Aristeu Rozanski <arozansk@redhat.com>
Date: Wed, 19 Dec 2007 14:23:47 -0500
Subject: [misc] edac: add support for intel 5000 mchs
Message-id: 20071219192347.GA665@redhat.com
O-Subject: [RHEL5 PATCH] EDAC: add support to Intel 5000 MCHs
Bugzilla: 249335
https://bugzilla.redhat.com/show_bug.cgi?id=249335
This patch adds EDAC support to Intel 5000 MCHs.
Don: you'll need to add CONFIG_EDAC_I5000='m' to i686 and x86_64 config
files and this patch should be applied after i3000_edac patch.
Basic tests done by Geoff Gustafson (who also tried to trigger memory
module errors but without success).
This driver is upstream.
diff --git a/drivers/edac/Kconfig b/drivers/edac/Kconfig
index ceb6ad4..9e741ca 100644
--- a/drivers/edac/Kconfig
+++ b/drivers/edac/Kconfig
@@ -90,6 +90,13 @@ config EDAC_I3000
Support for error detection and correction on the Intel
3000 and 3010 server chipsets.
+config EDAC_I5000
+ tristate "Intel 5000 chipsets (Greencreek/Blackford)"
+ depends on EDAC_MM_EDAC && PCI && X86
+ help
+ Support for error detection and correction the Intel
+ Greekcreek/Blackford chipsets.
+
config EDAC_I82860
tristate "Intel 82860"
depends on EDAC_MM_EDAC && PCI && X86_32
diff --git a/drivers/edac/Makefile b/drivers/edac/Makefile
index ff8f92c..fcb94ca 100644
--- a/drivers/edac/Makefile
+++ b/drivers/edac/Makefile
@@ -14,6 +14,7 @@ obj-$(CONFIG_EDAC_E7XXX) += e7xxx_edac.o
obj-$(CONFIG_EDAC_E752X) += e752x_edac.o
obj-$(CONFIG_EDAC_I82875P) += i82875p_edac.o
obj-$(CONFIG_EDAC_I3000) += i3000_edac.o
+obj-$(CONFIG_EDAC_I5000) += i5000_edac.o
obj-$(CONFIG_EDAC_I82860) += i82860_edac.o
obj-$(CONFIG_EDAC_K8) += k8_edac.o
obj-$(CONFIG_EDAC_R82600) += r82600_edac.o
diff --git a/drivers/edac/edac_mc.c b/drivers/edac/edac_mc.c
index c4d45c2..247e798 100644
--- a/drivers/edac/edac_mc.c
+++ b/drivers/edac/edac_mc.c
@@ -1723,6 +1723,111 @@ void edac_mc_handle_ue_no_info(struct mem_ctl_info *mci, const char *msg)
}
EXPORT_SYMBOL_GPL(edac_mc_handle_ue_no_info);
+/*************************************************************
+ * On Fully Buffered DIMM modules, this help function is
+ * called to process UE events
+ */
+void edac_mc_handle_fbd_ue(struct mem_ctl_info *mci,
+ unsigned int csrow,
+ unsigned int channela,
+ unsigned int channelb, char *msg)
+{
+ int len = EDAC_MC_LABEL_LEN * 4;
+ char labels[len + 1];
+ char *pos = labels;
+ int chars;
+
+ if (csrow >= mci->nr_csrows) {
+ /* something is wrong */
+ edac_mc_printk(mci, KERN_ERR,
+ "INTERNAL ERROR: row out of range (%d >= %d)\n",
+ csrow, mci->nr_csrows);
+ edac_mc_handle_ue_no_info(mci, "INTERNAL ERROR");
+ return;
+ }
+
+ if (channela >= mci->csrows[csrow].nr_channels) {
+ /* something is wrong */
+ edac_mc_printk(mci, KERN_ERR,
+ "INTERNAL ERROR: channel-a out of range "
+ "(%d >= %d)\n",
+ channela, mci->csrows[csrow].nr_channels);
+ edac_mc_handle_ue_no_info(mci, "INTERNAL ERROR");
+ return;
+ }
+
+ if (channelb >= mci->csrows[csrow].nr_channels) {
+ /* something is wrong */
+ edac_mc_printk(mci, KERN_ERR,
+ "INTERNAL ERROR: channel-b out of range "
+ "(%d >= %d)\n",
+ channelb, mci->csrows[csrow].nr_channels);
+ edac_mc_handle_ue_no_info(mci, "INTERNAL ERROR");
+ return;
+ }
+
+ mci->ue_count++;
+ mci->csrows[csrow].ue_count++;
+
+ /* Generate the DIMM labels from the specified channels */
+ chars = snprintf(pos, len + 1, "%s",
+ mci->csrows[csrow].channels[channela].label);
+ len -= chars;
+ pos += chars;
+ chars = snprintf(pos, len + 1, "-%s",
+ mci->csrows[csrow].channels[channelb].label);
+
+ if (log_ue)
+ edac_mc_printk(mci, KERN_EMERG,
+ "UE row %d, channel-a= %d channel-b= %d "
+ "labels \"%s\": %s\n", csrow, channela, channelb,
+ labels, msg);
+
+ if (panic_on_ue)
+ panic("UE row %d, channel-a= %d channel-b= %d "
+ "labels \"%s\": %s\n", csrow, channela,
+ channelb, labels, msg);
+}
+EXPORT_SYMBOL(edac_mc_handle_fbd_ue);
+
+/*************************************************************
+ * On Fully Buffered DIMM modules, this help function is
+ * called to process CE events
+ */
+void edac_mc_handle_fbd_ce(struct mem_ctl_info *mci,
+ unsigned int csrow, unsigned int channel, char *msg)
+{
+
+ /* Ensure boundary values */
+ if (csrow >= mci->nr_csrows) {
+ /* something is wrong */
+ edac_mc_printk(mci, KERN_ERR,
+ "INTERNAL ERROR: row out of range (%d >= %d)\n",
+ csrow, mci->nr_csrows);
+ edac_mc_handle_ce_no_info(mci, "INTERNAL ERROR");
+ return;
+ }
+ if (channel >= mci->csrows[csrow].nr_channels) {
+ /* something is wrong */
+ edac_mc_printk(mci, KERN_ERR,
+ "INTERNAL ERROR: channel out of range (%d >= %d)\n",
+ channel, mci->csrows[csrow].nr_channels);
+ edac_mc_handle_ce_no_info(mci, "INTERNAL ERROR");
+ return;
+ }
+
+ if (log_ce)
+ /* FIXME - put in DIMM location */
+ edac_mc_printk(mci, KERN_WARNING,
+ "CE row %d, channel %d, label \"%s\": %s\n",
+ csrow, channel,
+ mci->csrows[csrow].channels[channel].label, msg);
+
+ mci->ce_count++;
+ mci->csrows[csrow].ce_count++;
+ mci->csrows[csrow].channels[channel].ce_count++;
+}
+EXPORT_SYMBOL(edac_mc_handle_fbd_ce);
/*
* Iterate over all MC instances and check for ECC, et al, errors
diff --git a/drivers/edac/edac_mc.h b/drivers/edac/edac_mc.h
index 13ddfd9..fd04c04 100644
--- a/drivers/edac/edac_mc.h
+++ b/drivers/edac/edac_mc.h
@@ -448,6 +448,11 @@ extern void edac_mc_handle_ue(struct mem_ctl_info *mci,
int row, const char *msg);
extern void edac_mc_handle_ue_no_info(struct mem_ctl_info *mci,
const char *msg);
+extern void edac_mc_handle_fbd_ue(struct mem_ctl_info *mci, unsigned int csrow,
+ unsigned int channel0, unsigned int channel1,
+ char *msg);
+extern void edac_mc_handle_fbd_ce(struct mem_ctl_info *mci, unsigned int csrow,
+ unsigned int channel, char *msg);
/*
* This kmalloc's and initializes all the structures.
diff --git a/drivers/edac/i5000_edac.c b/drivers/edac/i5000_edac.c
new file mode 100644
index 0000000..9828247
--- /dev/null
+++ b/drivers/edac/i5000_edac.c
@@ -0,0 +1,1473 @@
+/*
+ * Intel 5000(P/V/X) class Memory Controllers kernel module
+ *
+ * This file may be distributed under the terms of the
+ * GNU General Public License.
+ *
+ * Written by Douglas Thompson Linux Networx (http://lnxi.com)
+ * norsk5@xmission.com
+ *
+ * This module is based on the following document:
+ *
+ * Intel 5000X Chipset Memory Controller Hub (MCH) - Datasheet
+ * http://developer.intel.com/design/chipsets/datashts/313070.htm
+ *
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/pci.h>
+#include <linux/pci_ids.h>
+#include <linux/slab.h>
+#include <asm/mmzone.h>
+#include "edac_mc.h"
+
+/*
+ * Alter this version for the I5000 module when modifications are made
+ */
+#define I5000_REVISION " Ver: 2.0.12 " __DATE__
+#define EDAC_MOD_STR "i5000_edac"
+
+#define i5000_printk(level, fmt, arg...) \
+ edac_printk(level, "i5000", fmt, ##arg)
+
+#define i5000_mc_printk(mci, level, fmt, arg...) \
+ edac_mc_chipset_printk(mci, level, "i5000", fmt, ##arg)
+
+#ifndef PCI_DEVICE_ID_INTEL_FBD_0
+#define PCI_DEVICE_ID_INTEL_FBD_0 0x25F5
+#endif
+#ifndef PCI_DEVICE_ID_INTEL_FBD_1
+#define PCI_DEVICE_ID_INTEL_FBD_1 0x25F6
+#endif
+
+/* Device 16,
+ * Function 0: System Address
+ * Function 1: Memory Branch Map, Control, Errors Register
+ * Function 2: FSB Error Registers
+ *
+ * All 3 functions of Device 16 (0,1,2) share the SAME DID
+ */
+#define PCI_DEVICE_ID_INTEL_I5000_DEV16 0x25F0
+
+/* OFFSETS for Function 0 */
+
+/* OFFSETS for Function 1 */
+#define AMBASE 0x48
+#define MAXCH 0x56
+#define MAXDIMMPERCH 0x57
+#define TOLM 0x6C
+#define REDMEMB 0x7C
+#define RED_ECC_LOCATOR(x) ((x) & 0x3FFFF)
+#define REC_ECC_LOCATOR_EVEN(x) ((x) & 0x001FF)
+#define REC_ECC_LOCATOR_ODD(x) ((x) & 0x3FE00)
+#define MIR0 0x80
+#define MIR1 0x84
+#define MIR2 0x88
+#define AMIR0 0x8C
+#define AMIR1 0x90
+#define AMIR2 0x94
+
+#define FERR_FAT_FBD 0x98
+#define NERR_FAT_FBD 0x9C
+#define EXTRACT_FBDCHAN_INDX(x) (((x)>>28) & 0x3)
+#define FERR_FAT_FBDCHAN 0x30000000
+#define FERR_FAT_M3ERR 0x00000004
+#define FERR_FAT_M2ERR 0x00000002
+#define FERR_FAT_M1ERR 0x00000001
+#define FERR_FAT_MASK (FERR_FAT_M1ERR | \
+ FERR_FAT_M2ERR | \
+ FERR_FAT_M3ERR)
+
+#define FERR_NF_FBD 0xA0
+
+/* Thermal and SPD or BFD errors */
+#define FERR_NF_M28ERR 0x01000000
+#define FERR_NF_M27ERR 0x00800000
+#define FERR_NF_M26ERR 0x00400000
+#define FERR_NF_M25ERR 0x00200000
+#define FERR_NF_M24ERR 0x00100000
+#define FERR_NF_M23ERR 0x00080000
+#define FERR_NF_M22ERR 0x00040000
+#define FERR_NF_M21ERR 0x00020000
+
+/* Correctable errors */
+#define FERR_NF_M20ERR 0x00010000
+#define FERR_NF_M19ERR 0x00008000
+#define FERR_NF_M18ERR 0x00004000
+#define FERR_NF_M17ERR 0x00002000
+
+/* Non-Retry or redundant Retry errors */
+#define FERR_NF_M16ERR 0x00001000
+#define FERR_NF_M15ERR 0x00000800
+#define FERR_NF_M14ERR 0x00000400
+#define FERR_NF_M13ERR 0x00000200
+
+/* Uncorrectable errors */
+#define FERR_NF_M12ERR 0x00000100
+#define FERR_NF_M11ERR 0x00000080
+#define FERR_NF_M10ERR 0x00000040
+#define FERR_NF_M9ERR 0x00000020
+#define FERR_NF_M8ERR 0x00000010
+#define FERR_NF_M7ERR 0x00000008
+#define FERR_NF_M6ERR 0x00000004
+#define FERR_NF_M5ERR 0x00000002
+#define FERR_NF_M4ERR 0x00000001
+
+#define FERR_NF_UNCORRECTABLE (FERR_NF_M12ERR | \
+ FERR_NF_M11ERR | \
+ FERR_NF_M10ERR | \
+ FERR_NF_M8ERR | \
+ FERR_NF_M7ERR | \
+ FERR_NF_M6ERR | \
+ FERR_NF_M5ERR | \
+ FERR_NF_M4ERR)
+#define FERR_NF_CORRECTABLE (FERR_NF_M20ERR | \
+ FERR_NF_M19ERR | \
+ FERR_NF_M18ERR | \
+ FERR_NF_M17ERR)
+#define FERR_NF_DIMM_SPARE (FERR_NF_M27ERR | \
+ FERR_NF_M28ERR)
+#define FERR_NF_THERMAL (FERR_NF_M26ERR | \
+ FERR_NF_M25ERR | \
+ FERR_NF_M24ERR | \
+ FERR_NF_M23ERR)
+#define FERR_NF_SPD_PROTOCOL (FERR_NF_M22ERR)
+#define FERR_NF_NORTH_CRC (FERR_NF_M21ERR)
+#define FERR_NF_NON_RETRY (FERR_NF_M13ERR | \
+ FERR_NF_M14ERR | \
+ FERR_NF_M15ERR)
+
+#define NERR_NF_FBD 0xA4
+#define FERR_NF_MASK (FERR_NF_UNCORRECTABLE | \
+ FERR_NF_CORRECTABLE | \
+ FERR_NF_DIMM_SPARE | \
+ FERR_NF_THERMAL | \
+ FERR_NF_SPD_PROTOCOL | \
+ FERR_NF_NORTH_CRC | \
+ FERR_NF_NON_RETRY)
+
+#define EMASK_FBD 0xA8
+#define EMASK_FBD_M28ERR 0x08000000
+#define EMASK_FBD_M27ERR 0x04000000
+#define EMASK_FBD_M26ERR 0x02000000
+#define EMASK_FBD_M25ERR 0x01000000
+#define EMASK_FBD_M24ERR 0x00800000
+#define EMASK_FBD_M23ERR 0x00400000
+#define EMASK_FBD_M22ERR 0x00200000
+#define EMASK_FBD_M21ERR 0x00100000
+#define EMASK_FBD_M20ERR 0x00080000
+#define EMASK_FBD_M19ERR 0x00040000
+#define EMASK_FBD_M18ERR 0x00020000
+#define EMASK_FBD_M17ERR 0x00010000
+
+#define EMASK_FBD_M15ERR 0x00004000
+#define EMASK_FBD_M14ERR 0x00002000
+#define EMASK_FBD_M13ERR 0x00001000
+#define EMASK_FBD_M12ERR 0x00000800
+#define EMASK_FBD_M11ERR 0x00000400
+#define EMASK_FBD_M10ERR 0x00000200
+#define EMASK_FBD_M9ERR 0x00000100
+#define EMASK_FBD_M8ERR 0x00000080
+#define EMASK_FBD_M7ERR 0x00000040
+#define EMASK_FBD_M6ERR 0x00000020
+#define EMASK_FBD_M5ERR 0x00000010
+#define EMASK_FBD_M4ERR 0x00000008
+#define EMASK_FBD_M3ERR 0x00000004
+#define EMASK_FBD_M2ERR 0x00000002
+#define EMASK_FBD_M1ERR 0x00000001
+
+#define ENABLE_EMASK_FBD_FATAL_ERRORS (EMASK_FBD_M1ERR | \
+ EMASK_FBD_M2ERR | \
+ EMASK_FBD_M3ERR)
+
+#define ENABLE_EMASK_FBD_UNCORRECTABLE (EMASK_FBD_M4ERR | \
+ EMASK_FBD_M5ERR | \
+ EMASK_FBD_M6ERR | \
+ EMASK_FBD_M7ERR | \
+ EMASK_FBD_M8ERR | \
+ EMASK_FBD_M9ERR | \
+ EMASK_FBD_M10ERR | \
+ EMASK_FBD_M11ERR | \
+ EMASK_FBD_M12ERR)
+#define ENABLE_EMASK_FBD_CORRECTABLE (EMASK_FBD_M17ERR | \
+ EMASK_FBD_M18ERR | \
+ EMASK_FBD_M19ERR | \
+ EMASK_FBD_M20ERR)
+#define ENABLE_EMASK_FBD_DIMM_SPARE (EMASK_FBD_M27ERR | \
+ EMASK_FBD_M28ERR)
+#define ENABLE_EMASK_FBD_THERMALS (EMASK_FBD_M26ERR | \
+ EMASK_FBD_M25ERR | \
+ EMASK_FBD_M24ERR | \
+ EMASK_FBD_M23ERR)
+#define ENABLE_EMASK_FBD_SPD_PROTOCOL (EMASK_FBD_M22ERR)
+#define ENABLE_EMASK_FBD_NORTH_CRC (EMASK_FBD_M21ERR)
+#define ENABLE_EMASK_FBD_NON_RETRY (EMASK_FBD_M15ERR | \
+ EMASK_FBD_M14ERR | \
+ EMASK_FBD_M13ERR)
+
+#define ENABLE_EMASK_ALL (ENABLE_EMASK_FBD_NON_RETRY | \
+ ENABLE_EMASK_FBD_NORTH_CRC | \
+ ENABLE_EMASK_FBD_SPD_PROTOCOL | \
+ ENABLE_EMASK_FBD_THERMALS | \
+ ENABLE_EMASK_FBD_DIMM_SPARE | \
+ ENABLE_EMASK_FBD_FATAL_ERRORS | \
+ ENABLE_EMASK_FBD_CORRECTABLE | \
+ ENABLE_EMASK_FBD_UNCORRECTABLE)
+
+#define ERR0_FBD 0xAC
+#define ERR1_FBD 0xB0
+#define ERR2_FBD 0xB4
+#define MCERR_FBD 0xB8
+#define NRECMEMA 0xBE
+#define NREC_BANK(x) (((x)>>12) & 0x7)
+#define NREC_RDWR(x) (((x)>>11) & 1)
+#define NREC_RANK(x) (((x)>>8) & 0x7)
+#define NRECMEMB 0xC0
+#define NREC_CAS(x) (((x)>>16) & 0xFFFFFF)
+#define NREC_RAS(x) ((x) & 0x7FFF)
+#define NRECFGLOG 0xC4
+#define NREEECFBDA 0xC8
+#define NREEECFBDB 0xCC
+#define NREEECFBDC 0xD0
+#define NREEECFBDD 0xD4
+#define NREEECFBDE 0xD8
+#define REDMEMA 0xDC
+#define RECMEMA 0xE2
+#define REC_BANK(x) (((x)>>12) & 0x7)
+#define REC_RDWR(x) (((x)>>11) & 1)
+#define REC_RANK(x) (((x)>>8) & 0x7)
+#define RECMEMB 0xE4
+#define REC_CAS(x) (((x)>>16) & 0xFFFFFF)
+#define REC_RAS(x) ((x) & 0x7FFF)
+#define RECFGLOG 0xE8
+#define RECFBDA 0xEC
+#define RECFBDB 0xF0
+#define RECFBDC 0xF4
+#define RECFBDD 0xF8
+#define RECFBDE 0xFC
+
+/* OFFSETS for Function 2 */
+
+/*
+ * Device 21,
+ * Function 0: Memory Map Branch 0
+ *
+ * Device 22,
+ * Function 0: Memory Map Branch 1
+ */
+#define PCI_DEVICE_ID_I5000_BRANCH_0 0x25F5
+#define PCI_DEVICE_ID_I5000_BRANCH_1 0x25F6
+
+#define AMB_PRESENT_0 0x64
+#define AMB_PRESENT_1 0x66
+#define MTR0 0x80
+#define MTR1 0x84
+#define MTR2 0x88
+#define MTR3 0x8C
+
+#define NUM_MTRS 4
+#define CHANNELS_PER_BRANCH (2)
+
+/* Defines to extract the vaious fields from the
+ * MTRx - Memory Technology Registers
+ */
+#define MTR_DIMMS_PRESENT(mtr) ((mtr) & (0x1 << 8))
+#define MTR_DRAM_WIDTH(mtr) ((((mtr) >> 6) & 0x1) ? 8 : 4)
+#define MTR_DRAM_BANKS(mtr) ((((mtr) >> 5) & 0x1) ? 8 : 4)
+#define MTR_DRAM_BANKS_ADDR_BITS(mtr) ((MTR_DRAM_BANKS(mtr) == 8) ? 3 : 2)
+#define MTR_DIMM_RANK(mtr) (((mtr) >> 4) & 0x1)
+#define MTR_DIMM_RANK_ADDR_BITS(mtr) (MTR_DIMM_RANK(mtr) ? 2 : 1)
+#define MTR_DIMM_ROWS(mtr) (((mtr) >> 2) & 0x3)
+#define MTR_DIMM_ROWS_ADDR_BITS(mtr) (MTR_DIMM_ROWS(mtr) + 13)
+#define MTR_DIMM_COLS(mtr) ((mtr) & 0x3)
+#define MTR_DIMM_COLS_ADDR_BITS(mtr) (MTR_DIMM_COLS(mtr) + 10)
+
+#ifdef CONFIG_EDAC_DEBUG
+static char *numrow_toString[] = {
+ "8,192 - 13 rows",
+ "16,384 - 14 rows",
+ "32,768 - 15 rows",
+ "reserved"
+};
+
+static char *numcol_toString[] = {
+ "1,024 - 10 columns",
+ "2,048 - 11 columns",
+ "4,096 - 12 columns",
+ "reserved"
+};
+#endif
+
+/* Enumeration of supported devices */
+enum i5000_chips {
+ I5000P = 0,
+ I5000V = 1, /* future */
+ I5000X = 2 /* future */
+};
+
+/* Device name and register DID (Device ID) */
+struct i5000_dev_info {
+ const char *ctl_name; /* name for this device */
+ u16 fsb_mapping_errors; /* DID for the branchmap,control */
+};
+
+/* Table of devices attributes supported by this driver */
+static const struct i5000_dev_info i5000_devs[] = {
+ [I5000P] = {
+ .ctl_name = "I5000",
+ .fsb_mapping_errors = PCI_DEVICE_ID_INTEL_I5000_DEV16,
+ },
+};
+
+struct i5000_dimm_info {
+ int megabytes; /* size, 0 means not present */
+ int dual_rank;
+};
+
+#define MAX_CHANNELS 6 /* max possible channels */
+#define MAX_CSROWS (8*2) /* max possible csrows per channel */
+
+/* driver private data structure */
+struct i5000_pvt {
+ struct pci_dev *system_address; /* 16.0 */
+ struct pci_dev *branchmap_werrors; /* 16.1 */
+ struct pci_dev *fsb_error_regs; /* 16.2 */
+ struct pci_dev *branch_0; /* 21.0 */
+ struct pci_dev *branch_1; /* 22.0 */
+
+ u16 tolm; /* top of low memory */
+ u64 ambase; /* AMB BAR */
+
+ u16 mir0, mir1, mir2;
+
+ u16 b0_mtr[NUM_MTRS]; /* Memory Technlogy Reg */
+ u16 b0_ambpresent0; /* Branch 0, Channel 0 */
+ u16 b0_ambpresent1; /* Brnach 0, Channel 1 */
+
+ u16 b1_mtr[NUM_MTRS]; /* Memory Technlogy Reg */
+ u16 b1_ambpresent0; /* Branch 1, Channel 8 */
+ u16 b1_ambpresent1; /* Branch 1, Channel 1 */
+
+ /* DIMM infomation matrix, allocating architecture maximums */
+ struct i5000_dimm_info dimm_info[MAX_CSROWS][MAX_CHANNELS];
+
+ /* Actual values for this controller */
+ int maxch; /* Max channels */
+ int maxdimmperch; /* Max DIMMs per channel */
+};
+
+/* I5000 MCH error information retrieved from Hardware */
+struct i5000_error_info {
+
+ /* These registers are always read from the MC */
+ u32 ferr_fat_fbd; /* First Errors Fatal */
+ u32 nerr_fat_fbd; /* Next Errors Fatal */
+ u32 ferr_nf_fbd; /* First Errors Non-Fatal */
+ u32 nerr_nf_fbd; /* Next Errors Non-Fatal */
+
+ /* These registers are input ONLY if there was a Recoverable Error */
+ u32 redmemb; /* Recoverable Mem Data Error log B */
+ u16 recmema; /* Recoverable Mem Error log A */
+ u32 recmemb; /* Recoverable Mem Error log B */
+
+ /* These registers are input ONLY if there was a
+ * Non-Recoverable Error */
+ u16 nrecmema; /* Non-Recoverable Mem log A */
+ u16 nrecmemb; /* Non-Recoverable Mem log B */
+
+};
+
+/*
+ * i5000_get_error_info Retrieve the hardware error information from
+ * the hardware and cache it in the 'info'
+ * structure
+ */
+static void i5000_get_error_info(struct mem_ctl_info *mci,
+ struct i5000_error_info *info)
+{
+ struct i5000_pvt *pvt;
+ u32 value;
+
+ pvt = mci->pvt_info;
+
+ /* read in the 1st FATAL error register */
+ pci_read_config_dword(pvt->branchmap_werrors, FERR_FAT_FBD, &value);
+
+ /* Mask only the bits that the doc says are valid
+ */
+ value &= (FERR_FAT_FBDCHAN | FERR_FAT_MASK);
+
+ /* If there is an error, then read in the */
+ /* NEXT FATAL error register and the Memory Error Log Register A */
+ if (value & FERR_FAT_MASK) {
+ info->ferr_fat_fbd = value;
+
+ /* harvest the various error data we need */
+ pci_read_config_dword(pvt->branchmap_werrors,
+ NERR_FAT_FBD, &info->nerr_fat_fbd);
+ pci_read_config_word(pvt->branchmap_werrors,
+ NRECMEMA, &info->nrecmema);
+ pci_read_config_word(pvt->branchmap_werrors,
+ NRECMEMB, &info->nrecmemb);
+
+ /* Clear the error bits, by writing them back */
+ pci_write_config_dword(pvt->branchmap_werrors,
+ FERR_FAT_FBD, value);
+ } else {
+ info->ferr_fat_fbd = 0;
+ info->nerr_fat_fbd = 0;
+ info->nrecmema = 0;
+ info->nrecmemb = 0;
+ }
+
+ /* read in the 1st NON-FATAL error register */
+ pci_read_config_dword(pvt->branchmap_werrors, FERR_NF_FBD, &value);
+
+ /* If there is an error, then read in the 1st NON-FATAL error
+ * register as well */
+ if (value & FERR_NF_MASK) {
+ info->ferr_nf_fbd = value;
+
+ /* harvest the various error data we need */
+ pci_read_config_dword(pvt->branchmap_werrors,
+ NERR_NF_FBD, &info->nerr_nf_fbd);
+ pci_read_config_word(pvt->branchmap_werrors,
+ RECMEMA, &info->recmema);
+ pci_read_config_dword(pvt->branchmap_werrors,
+ RECMEMB, &info->recmemb);
+ pci_read_config_dword(pvt->branchmap_werrors,
+ REDMEMB, &info->redmemb);
+
+ /* Clear the error bits, by writing them back */
+ pci_write_config_dword(pvt->branchmap_werrors,
+ FERR_NF_FBD, value);
+ } else {
+ info->ferr_nf_fbd = 0;
+ info->nerr_nf_fbd = 0;
+ info->recmema = 0;
+ info->recmemb = 0;
+ info->redmemb = 0;
+ }
+}
+
+/*
+ * i5000_process_fatal_error_info(struct mem_ctl_info *mci,
+ * struct i5000_error_info *info,
+ * int handle_errors);
+ *
+ * handle the Intel FATAL errors, if any
+ */
+static void i5000_process_fatal_error_info(struct mem_ctl_info *mci,
+ struct i5000_error_info *info,
+ int handle_errors)
+{
+ char msg[EDAC_MC_LABEL_LEN + 1 + 90];
+ u32 allErrors;
+ int branch;
+ int channel;
+ int bank;
+ int rank;
+ int rdwr;
+ int ras, cas;
+
+ /* mask off the Error bits that are possible */
+ allErrors = (info->ferr_fat_fbd & FERR_FAT_MASK);
+ if (!allErrors)
+ return; /* if no error, return now */
+
+ /* ONLY ONE of the possible error bits will be set, as per the docs */
+ i5000_mc_printk(mci, KERN_ERR,
+ "FATAL ERRORS Found!!! 1st FATAL Err Reg= 0x%x\n",
+ allErrors);
+
+ branch = EXTRACT_FBDCHAN_INDX(info->ferr_fat_fbd);
+ channel = branch;
+
+ /* Use the NON-Recoverable macros to extract data */
+ bank = NREC_BANK(info->nrecmema);
+ rank = NREC_RANK(info->nrecmema);
+ rdwr = NREC_RDWR(info->nrecmema);
+ ras = NREC_RAS(info->nrecmemb);
+ cas = NREC_CAS(info->nrecmemb);
+
+ debugf0("\t\tCSROW= %d Channels= %d,%d (Branch= %d "
+ "DRAM Bank= %d rdwr= %s ras= %d cas= %d)\n",
+ rank, channel, channel + 1, branch >> 1, bank,
+ rdwr ? "Write" : "Read", ras, cas);
+
+ /* Only 1 bit will be on */
+ if (allErrors & FERR_FAT_M1ERR) {
+ i5000_mc_printk(mci, KERN_ERR,
+ "Alert on non-redundant retry or fast "
+ "reset timeout\n");
+
+ } else if (allErrors & FERR_FAT_M2ERR) {
+ i5000_mc_printk(mci, KERN_ERR,
+ "Northbound CRC error on non-redundant "
+ "retry\n");
+
+ } else if (allErrors & FERR_FAT_M3ERR) {
+ i5000_mc_printk(mci, KERN_ERR,
+ ">Tmid Thermal event with intelligent "
+ "throttling disabled\n");
+ }
+
+ /* Form out message */
+ snprintf(msg, sizeof(msg),
+ "(Branch=%d DRAM-Bank=%d RDWR=%s RAS=%d CAS=%d "
+ "FATAL Err=0x%x)",
+ branch >> 1, bank, rdwr ? "Write" : "Read", ras, cas,
+ allErrors);
+
+ /* Call the helper to output message */
+ edac_mc_handle_fbd_ue(mci, rank, channel, channel + 1, msg);
+}
+
+/*
+ * i5000_process_fatal_error_info(struct mem_ctl_info *mci,
+ * struct i5000_error_info *info,
+ * int handle_errors);
+ *
+ * handle the Intel NON-FATAL errors, if any
+ */
+static void i5000_process_nonfatal_error_info(struct mem_ctl_info *mci,
+ struct i5000_error_info *info,
+ int handle_errors)
+{
+ char msg[EDAC_MC_LABEL_LEN + 1 + 90];
+ u32 allErrors;
+ u32 ue_errors;
+ u32 ce_errors;
+ u32 misc_errors;
+ int branch;
+ int channel;
+ int bank;
+ int rank;
+ int rdwr;
+ int ras, cas;
+
+ /* mask off the Error bits that are possible */
+ allErrors = (info->ferr_nf_fbd & FERR_NF_MASK);
+ if (!allErrors)
+ return; /* if no error, return now */
+
+ /* ONLY ONE of the possible error bits will be set, as per the docs */
+ i5000_mc_printk(mci, KERN_WARNING,
+ "NON-FATAL ERRORS Found!!! 1st NON-FATAL Err "
+ "Reg= 0x%x\n", allErrors);
+
+ ue_errors = allErrors & FERR_NF_UNCORRECTABLE;
+ if (ue_errors) {
+ debugf0("\tUncorrected bits= 0x%x\n", ue_errors);
+
+ branch = EXTRACT_FBDCHAN_INDX(info->ferr_nf_fbd);
+ channel = branch;
+ bank = NREC_BANK(info->nrecmema);
+ rank = NREC_RANK(info->nrecmema);
+ rdwr = NREC_RDWR(info->nrecmema);
+ ras = NREC_RAS(info->nrecmemb);
+ cas = NREC_CAS(info->nrecmemb);
+
+ debugf0
+ ("\t\tCSROW= %d Channels= %d,%d (Branch= %d "
+ "DRAM Bank= %d rdwr= %s ras= %d cas= %d)\n",
+ rank, channel, channel + 1, branch >> 1, bank,
+ rdwr ? "Write" : "Read", ras, cas);
+
+ /* Form out message */
+ snprintf(msg, sizeof(msg),
+ "(Branch=%d DRAM-Bank=%d RDWR=%s RAS=%d "
+ "CAS=%d, UE Err=0x%x)",
+ branch >> 1, bank, rdwr ? "Write" : "Read", ras, cas,
+ ue_errors);
+
+ /* Call the helper to output message */
+ edac_mc_handle_fbd_ue(mci, rank, channel, channel + 1, msg);
+ }
+
+ /* Check correctable errors */
+ ce_errors = allErrors & FERR_NF_CORRECTABLE;
+ if (ce_errors) {
+ debugf0("\tCorrected bits= 0x%x\n", ce_errors);
+
+ branch = EXTRACT_FBDCHAN_INDX(info->ferr_nf_fbd);
+
+ channel = 0;
+ if (REC_ECC_LOCATOR_ODD(info->redmemb))
+ channel = 1;
+
+ /* Convert channel to be based from zero, instead of
+ * from branch base of 0 */
+ channel += branch;
+
+ bank = REC_BANK(info->recmema);
+ rank = REC_RANK(info->recmema);
+ rdwr = REC_RDWR(info->recmema);
+ ras = REC_RAS(info->recmemb);
+ cas = REC_CAS(info->recmemb);
+
+ debugf0("\t\tCSROW= %d Channel= %d (Branch %d "
+ "DRAM Bank= %d rdwr= %s ras= %d cas= %d)\n",
+ rank, channel, branch >> 1, bank,
+ rdwr ? "Write" : "Read", ras, cas);
+
+ /* Form out message */
+ snprintf(msg, sizeof(msg),
+ "(Branch=%d DRAM-Bank=%d RDWR=%s RAS=%d "
+ "CAS=%d, CE Err=0x%x)", branch >> 1, bank,
+ rdwr ? "Write" : "Read", ras, cas, ce_errors);
+
+ /* Call the helper to output message */
+ edac_mc_handle_fbd_ce(mci, rank, channel, msg);
+ }
+
+ /* See if any of the thermal errors have fired */
+ misc_errors = allErrors & FERR_NF_THERMAL;
+ if (misc_errors) {
+ i5000_printk(KERN_WARNING, "\tTHERMAL Error, bits= 0x%x\n",
+ misc_errors);
+ }
+
+ /* See if any of the thermal errors have fired */
+ misc_errors = allErrors & FERR_NF_NON_RETRY;
+ if (misc_errors) {
+ i5000_printk(KERN_WARNING, "\tNON-Retry Errors, bits= 0x%x\n",
+ misc_errors);
+ }
+
+ /* See if any of the thermal errors have fired */
+ misc_errors = allErrors & FERR_NF_NORTH_CRC;
+ if (misc_errors) {
+ i5000_printk(KERN_WARNING,
+ "\tNORTHBOUND CRC Error, bits= 0x%x\n",
+ misc_errors);
+ }
+
+ /* See if any of the thermal errors have fired */
+ misc_errors = allErrors & FERR_NF_SPD_PROTOCOL;
+ if (misc_errors) {
+ i5000_printk(KERN_WARNING,
+ "\tSPD Protocol Error, bits= 0x%x\n",
+ misc_errors);
+ }
+
+ /* See if any of the thermal errors have fired */
+ misc_errors = allErrors & FERR_NF_DIMM_SPARE;
+ if (misc_errors) {
+ i5000_printk(KERN_WARNING, "\tDIMM-Spare Error, bits= 0x%x\n",
+ misc_errors);
+ }
+}
+
+/*
+ * i5000_process_error_info Process the error info that is
+ * in the 'info' structure, previously retrieved from hardware
+ */
+static void i5000_process_error_info(struct mem_ctl_info *mci,
+ struct i5000_error_info *info,
+ int handle_errors)
+{
+ /* First handle any fatal errors that occurred */
+ i5000_process_fatal_error_info(mci, info, handle_errors);
+
+ /* now handle any non-fatal errors that occurred */
+ i5000_process_nonfatal_error_info(mci, info, handle_errors);
+}
+
+/*
+ * i5000_clear_error Retrieve any error from the hardware
+ * but do NOT process that error.
+ * Used for 'clearing' out of previous errors
+ * Called by the Core module.
+ */
+static void i5000_clear_error(struct mem_ctl_info *mci)
+{
+ struct i5000_error_info info;
+
+ i5000_get_error_info(mci, &info);
+}
+
+/*
+ * i5000_check_error Retrieve and process errors reported by the
+ * hardware. Called by the Core module.
+ */
+static void i5000_check_error(struct mem_ctl_info *mci)
+{
+ struct i5000_error_info info;
+ debugf4("MC%d: " __FILE__ ": %s()\n", mci->mc_idx, __func__);
+ i5000_get_error_info(mci, &info);
+ i5000_process_error_info(mci, &info, 1);
+}
+
+/*
+ * i5000_get_devices Find and perform 'get' operation on the MCH's
+ * device/functions we want to reference for this driver
+ *
+ * Need to 'get' device 16 func 1 and func 2
+ */
+static int i5000_get_devices(struct mem_ctl_info *mci, int dev_idx)
+{
+ //const struct i5000_dev_info *i5000_dev = &i5000_devs[dev_idx];
+ struct i5000_pvt *pvt;
+ struct pci_dev *pdev;
+
+ pvt = mci->pvt_info;
+
+ /* Attempt to 'get' the MCH register we want */
+ pdev = NULL;
+ while (1) {
+ pdev = pci_get_device(PCI_VENDOR_ID_INTEL,
+ PCI_DEVICE_ID_INTEL_I5000_DEV16, pdev);
+
+ /* End of list, leave */
+ if (pdev == NULL) {
+ i5000_printk(KERN_ERR,
+ "'system address,Process Bus' "
+ "device not found:"
+ "vendor 0x%x device 0x%x FUNC 1 "
+ "(broken BIOS?)\n",
+ PCI_VENDOR_ID_INTEL,
+ PCI_DEVICE_ID_INTEL_I5000_DEV16);
+
+ return 1;
+ }
+
+ /* Scan for device 16 func 1 */
+ if (PCI_FUNC(pdev->devfn) == 1)
+ break;
+ }
+
+ pvt->branchmap_werrors = pdev;
+
+ /* Attempt to 'get' the MCH register we want */
+ pdev = NULL;
+ while (1) {
+ pdev = pci_get_device(PCI_VENDOR_ID_INTEL,
+ PCI_DEVICE_ID_INTEL_I5000_DEV16, pdev);
+
+ if (pdev == NULL) {
+ i5000_printk(KERN_ERR,
+ "MC: 'branchmap,control,errors' "
+ "device not found:"
+ "vendor 0x%x device 0x%x Func 2 "
+ "(broken BIOS?)\n",
+ PCI_VENDOR_ID_INTEL,
+ PCI_DEVICE_ID_INTEL_I5000_DEV16);
+
+ pci_dev_put(pvt->branchmap_werrors);
+ return 1;
+ }
+
+ /* Scan for device 16 func 1 */
+ if (PCI_FUNC(pdev->devfn) == 2)
+ break;
+ }
+
+ pvt->fsb_error_regs = pdev;
+
+ debugf1("System Address, processor bus- PCI Bus ID: %s %x:%x\n",
+ pci_name(pvt->system_address),
+ pvt->system_address->vendor, pvt->system_address->device);
+ debugf1("Branchmap, control and errors - PCI Bus ID: %s %x:%x\n",
+ pci_name(pvt->branchmap_werrors),
+ pvt->branchmap_werrors->vendor, pvt->branchmap_werrors->device);
+ debugf1("FSB Error Regs - PCI Bus ID: %s %x:%x\n",
+ pci_name(pvt->fsb_error_regs),
+ pvt->fsb_error_regs->vendor, pvt->fsb_error_regs->device);
+
+ pdev = NULL;
+ pdev = pci_get_device(PCI_VENDOR_ID_INTEL,
+ PCI_DEVICE_ID_I5000_BRANCH_0, pdev);
+
+ if (pdev == NULL) {
+ i5000_printk(KERN_ERR,
+ "MC: 'BRANCH 0' device not found:"
+ "vendor 0x%x device 0x%x Func 0 (broken BIOS?)\n",
+ PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_I5000_BRANCH_0);
+
+ pci_dev_put(pvt->branchmap_werrors);
+ pci_dev_put(pvt->fsb_error_regs);
+ return 1;
+ }
+
+ pvt->branch_0 = pdev;
+
+ /* If this device claims to have more than 2 channels then
+ * fetch Branch 1's information
+ */
+ if (pvt->maxch >= CHANNELS_PER_BRANCH) {
+ pdev = NULL;
+ pdev = pci_get_device(PCI_VENDOR_ID_INTEL,
+ PCI_DEVICE_ID_I5000_BRANCH_1, pdev);
+
+ if (pdev == NULL) {
+ i5000_printk(KERN_ERR,
+ "MC: 'BRANCH 1' device not found:"
+ "vendor 0x%x device 0x%x Func 0 "
+ "(broken BIOS?)\n",
+ PCI_VENDOR_ID_INTEL,
+ PCI_DEVICE_ID_I5000_BRANCH_1);
+
+ pci_dev_put(pvt->branchmap_werrors);
+ pci_dev_put(pvt->fsb_error_regs);
+ pci_dev_put(pvt->branch_0);
+ return 1;
+ }
+
+ pvt->branch_1 = pdev;
+ }
+
+ return 0;
+}
+
+/*
+ * i5000_put_devices 'put' all the devices that we have
+ * reserved via 'get'
+ */
+static void i5000_put_devices(struct mem_ctl_info *mci)
+{
+ struct i5000_pvt *pvt;
+
+ pvt = mci->pvt_info;
+
+ pci_dev_put(pvt->branchmap_werrors); /* FUNC 1 */
+ pci_dev_put(pvt->fsb_error_regs); /* FUNC 2 */
+ pci_dev_put(pvt->branch_0); /* DEV 21 */
+
+ /* Only if more than 2 channels do we release the second branch */
+ if (pvt->maxch >= CHANNELS_PER_BRANCH)
+ pci_dev_put(pvt->branch_1); /* DEV 22 */
+}
+
+/*
+ * determine_amb_resent
+ *
+ * the information is contained in NUM_MTRS different registers
+ * determineing which of the NUM_MTRS requires knowing
+ * which channel is in question
+ *
+ * 2 branches, each with 2 channels
+ * b0_ambpresent0 for channel '0'
+ * b0_ambpresent1 for channel '1'
+ * b1_ambpresent0 for channel '2'
+ * b1_ambpresent1 for channel '3'
+ */
+static int determine_amb_present_reg(struct i5000_pvt *pvt, int channel)
+{
+ int amb_present;
+
+ if (channel < CHANNELS_PER_BRANCH) {
+ if (channel & 0x1)
+ amb_present = pvt->b0_ambpresent1;
+ else
+ amb_present = pvt->b0_ambpresent0;
+ } else {
+ if (channel & 0x1)
+ amb_present = pvt->b1_ambpresent1;
+ else
+ amb_present = pvt->b1_ambpresent0;
+ }
+
+ return amb_present;
+}
+
+/*
+ * determine_mtr(pvt, csrow, channel)
+ *
+ * return the proper MTR register as determine by the csrow and channel desired
+ */
+static int determine_mtr(struct i5000_pvt *pvt, int csrow, int channel)
+{
+ int mtr;
+
+ if (channel < CHANNELS_PER_BRANCH)
+ mtr = pvt->b0_mtr[csrow >> 1];
+ else
+ mtr = pvt->b1_mtr[csrow >> 1];
+
+ return mtr;
+}
+
+/*
+ */
+static void decode_mtr(int slot_row, u16 mtr)
+{
+ int ans;
+
+ ans = MTR_DIMMS_PRESENT(mtr);
+
+ debugf2("\tMTR%d=0x%x: DIMMs are %s\n", slot_row, mtr,
+ ans ? "Present" : "NOT Present");
+ if (!ans)
+ return;
+
+ debugf2("\t\tWIDTH: x%d\n", MTR_DRAM_WIDTH(mtr));
+ debugf2("\t\tNUMBANK: %d bank(s)\n", MTR_DRAM_BANKS(mtr));
+ debugf2("\t\tNUMRANK: %s\n", MTR_DIMM_RANK(mtr) ? "double" : "single");
+ debugf2("\t\tNUMROW: %s\n", numrow_toString[MTR_DIMM_ROWS(mtr)]);
+ debugf2("\t\tNUMCOL: %s\n", numcol_toString[MTR_DIMM_COLS(mtr)]);
+}
+
+static void handle_channel(struct i5000_pvt *pvt, int csrow, int channel,
+ struct i5000_dimm_info *dinfo)
+{
+ int mtr;
+ int amb_present_reg;
+ int addrBits;
+
+ mtr = determine_mtr(pvt, csrow, channel);
+ if (MTR_DIMMS_PRESENT(mtr)) {
+ amb_present_reg = determine_amb_present_reg(pvt, channel);
+
+ /* Determine if there is a DIMM present in this DIMM slot */
+ if (amb_present_reg & (1 << (csrow >> 1))) {
+ dinfo->dual_rank = MTR_DIMM_RANK(mtr);
+
+ if (!((dinfo->dual_rank == 0) &&
+ ((csrow & 0x1) == 0x1))) {
+ /* Start with the number of bits for a Bank
+ * on the DRAM */
+ addrBits = MTR_DRAM_BANKS_ADDR_BITS(mtr);
+ /* Add thenumber of ROW bits */
+ addrBits += MTR_DIMM_ROWS_ADDR_BITS(mtr);
+ /* add the number of COLUMN bits */
+ addrBits += MTR_DIMM_COLS_ADDR_BITS(mtr);
+
+ addrBits += 6; /* add 64 bits per DIMM */
+ addrBits -= 20; /* divide by 2^^20 */
+ addrBits -= 3; /* 8 bits per bytes */
+
+ dinfo->megabytes = 1 << addrBits;
+ }
+ }
+ }
+}
+
+/*
+ * calculate_dimm_size
+ *
+ * also will output a DIMM matrix map, if debug is enabled, for viewing
+ * how the DIMMs are populated
+ */
+static void calculate_dimm_size(struct i5000_pvt *pvt)
+{
+ struct i5000_dimm_info *dinfo;
+ int csrow, max_csrows;
+ char *p, *mem_buffer;
+ int space, n;
+ int channel;
+
+ /* ================= Generate some debug output ================= */
+ space = PAGE_SIZE;
+ mem_buffer = p = kmalloc(space, GFP_KERNEL);
+ if (p == NULL) {
+ i5000_printk(KERN_ERR, "MC: %s:%s() kmalloc() failed\n",
+ __FILE__, __func__);
+ return;
+ }
+
+ n = snprintf(p, space, "\n");
+ p += n;
+ space -= n;
+
+ /* Scan all the actual CSROWS (which is # of DIMMS * 2)
+ * and calculate the information for each DIMM
+ * Start with the highest csrow first, to display it first
+ * and work toward the 0th csrow
+ */
+ max_csrows = pvt->maxdimmperch * 2;
+ for (csrow = max_csrows - 1; csrow >= 0; csrow--) {
+
+ /* on an odd csrow, first output a 'boundary' marker,
+ * then reset the message buffer */
+ if (csrow & 0x1) {
+ n = snprintf(p, space, "---------------------------"
+ "--------------------------------");
+ p += n;
+ space -= n;
+ debugf2("%s\n", mem_buffer);
+ p = mem_buffer;
+ space = PAGE_SIZE;
+ }
+ n = snprintf(p, space, "csrow %2d ", csrow);
+ p += n;
+ space -= n;
+
+ for (channel = 0; channel < pvt->maxch; channel++) {
+ dinfo = &pvt->dimm_info[csrow][channel];
+ handle_channel(pvt, csrow, channel, dinfo);
+ n = snprintf(p, space, "%4d MB | ", dinfo->megabytes);
+ p += n;
+ space -= n;
+ }
+ n = snprintf(p, space, "\n");
+ p += n;
+ space -= n;
+ }
+
+ /* Output the last bottom 'boundary' marker */
+ n = snprintf(p, space, "---------------------------"
+ "--------------------------------\n");
+ p += n;
+ space -= n;
+
+ /* now output the 'channel' labels */
+ n = snprintf(p, space, " ");
+ p += n;
+ space -= n;
+ for (channel = 0; channel < pvt->maxch; channel++) {
+ n = snprintf(p, space, "channel %d | ", channel);
+ p += n;
+ space -= n;
+ }
+ n = snprintf(p, space, "\n");
+ p += n;
+ space -= n;
+
+ /* output the last message and free buffer */
+ debugf2("%s\n", mem_buffer);
+ kfree(mem_buffer);
+}
+
+/*
+ * i5000_get_mc_regs read in the necessary registers and
+ * cache locally
+ *
+ * Fills in the private data members
+ */
+static void i5000_get_mc_regs(struct mem_ctl_info *mci)
+{
+ struct i5000_pvt *pvt;
+ u32 actual_tolm;
+ u16 limit;
+ int slot_row;
+ int maxch;
+ int maxdimmperch;
+ int way0, way1;
+
+ pvt = mci->pvt_info;
+
+ pci_read_config_dword(pvt->system_address, AMBASE,
+ (u32 *) & pvt->ambase);
+ pci_read_config_dword(pvt->system_address, AMBASE + sizeof(u32),
+ ((u32 *) & pvt->ambase) + sizeof(u32));
+
+ maxdimmperch = pvt->maxdimmperch;
+ maxch = pvt->maxch;
+
+ debugf2("AMBASE= 0x%lx MAXCH= %d MAX-DIMM-Per-CH= %d\n",
+ (long unsigned int)pvt->ambase, pvt->maxch, pvt->maxdimmperch);
+
+ /* Get the Branch Map regs */
+ pci_read_config_word(pvt->branchmap_werrors, TOLM, &pvt->tolm);
+ pvt->tolm >>= 12;
+ debugf2("\nTOLM (number of 256M regions) =%u (0x%x)\n", pvt->tolm,
+ pvt->tolm);
+
+ actual_tolm = pvt->tolm << 28;
+ debugf2("Actual TOLM byte addr=%u (0x%x)\n", actual_tolm, actual_tolm);
+
+ pci_read_config_word(pvt->branchmap_werrors, MIR0, &pvt->mir0);
+ pci_read_config_word(pvt->branchmap_werrors, MIR1, &pvt->mir1);
+ pci_read_config_word(pvt->branchmap_werrors, MIR2, &pvt->mir2);
+
+ /* Get the MIR[0-2] regs */
+ limit = (pvt->mir0 >> 4) & 0x0FFF;
+ way0 = pvt->mir0 & 0x1;
+ way1 = pvt->mir0 & 0x2;
+ debugf2("MIR0: limit= 0x%x WAY1= %u WAY0= %x\n", limit, way1, way0);
+ limit = (pvt->mir1 >> 4) & 0x0FFF;
+ way0 = pvt->mir1 & 0x1;
+ way1 = pvt->mir1 & 0x2;
+ debugf2("MIR1: limit= 0x%x WAY1= %u WAY0= %x\n", limit, way1, way0);
+ limit = (pvt->mir2 >> 4) & 0x0FFF;
+ way0 = pvt->mir2 & 0x1;
+ way1 = pvt->mir2 & 0x2;
+ debugf2("MIR2: limit= 0x%x WAY1= %u WAY0= %x\n", limit, way1, way0);
+
+ /* Get the MTR[0-3] regs */
+ for (slot_row = 0; slot_row < NUM_MTRS; slot_row++) {
+ int where = MTR0 + (slot_row * sizeof(u32));
+
+ pci_read_config_word(pvt->branch_0, where,
+ &pvt->b0_mtr[slot_row]);
+
+ debugf2("MTR%d where=0x%x B0 value=0x%x\n", slot_row, where,
+ pvt->b0_mtr[slot_row]);
+
+ if (pvt->maxch >= CHANNELS_PER_BRANCH) {
+ pci_read_config_word(pvt->branch_1, where,
+ &pvt->b1_mtr[slot_row]);
+ debugf2("MTR%d where=0x%x B1 value=0x%x\n", slot_row,
+ where, pvt->b0_mtr[slot_row]);
+ } else {
+ pvt->b1_mtr[slot_row] = 0;
+ }
+ }
+
+ /* Read and dump branch 0's MTRs */
+ debugf2("\nMemory Technology Registers:\n");
+ debugf2(" Branch 0:\n");
+ for (slot_row = 0; slot_row < NUM_MTRS; slot_row++) {
+ decode_mtr(slot_row, pvt->b0_mtr[slot_row]);
+ }
+ pci_read_config_word(pvt->branch_0, AMB_PRESENT_0,
+ &pvt->b0_ambpresent0);
+ debugf2("\t\tAMB-Branch 0-present0 0x%x:\n", pvt->b0_ambpresent0);
+ pci_read_config_word(pvt->branch_0, AMB_PRESENT_1,
+ &pvt->b0_ambpresent1);
+ debugf2("\t\tAMB-Branch 0-present1 0x%x:\n", pvt->b0_ambpresent1);
+
+ /* Only if we have 2 branchs (4 channels) */
+ if (pvt->maxch < CHANNELS_PER_BRANCH) {
+ pvt->b1_ambpresent0 = 0;
+ pvt->b1_ambpresent1 = 0;
+ } else {
+ /* Read and dump branch 1's MTRs */
+ debugf2(" Branch 1:\n");
+ for (slot_row = 0; slot_row < NUM_MTRS; slot_row++) {
+ decode_mtr(slot_row, pvt->b1_mtr[slot_row]);
+ }
+ pci_read_config_word(pvt->branch_1, AMB_PRESENT_0,
+ &pvt->b1_ambpresent0);
+ debugf2("\t\tAMB-Branch 1-present0 0x%x:\n",
+ pvt->b1_ambpresent0);
+ pci_read_config_word(pvt->branch_1, AMB_PRESENT_1,
+ &pvt->b1_ambpresent1);
+ debugf2("\t\tAMB-Branch 1-present1 0x%x:\n",
+ pvt->b1_ambpresent1);
+ }
+
+ /* Go and determine the size of each DIMM and place in an
+ * orderly matrix */
+ calculate_dimm_size(pvt);
+}
+
+/*
+ * i5000_init_csrows Initialize the 'csrows' table within
+ * the mci control structure with the
+ * addressing of memory.
+ *
+ * return:
+ * 0 success
+ * 1 no actual memory found on this MC
+ */
+static int i5000_init_csrows(struct mem_ctl_info *mci)
+{
+ struct i5000_pvt *pvt;
+ struct csrow_info *p_csrow;
+ int empty, channel_count;
+ int max_csrows;
+ int mtr;
+ int csrow_megs;
+ int channel;
+ int csrow;
+
+ pvt = mci->pvt_info;
+
+ channel_count = pvt->maxch;
+ max_csrows = pvt->maxdimmperch * 2;
+
+ empty = 1; /* Assume NO memory */
+
+ for (csrow = 0; csrow < max_csrows; csrow++) {
+ p_csrow = &mci->csrows[csrow];
+
+ p_csrow->csrow_idx = csrow;
+
+ /* use branch 0 for the basis */
+ mtr = pvt->b0_mtr[csrow >> 1];
+
+ /* if no DIMMS on this row, continue */
+ if (!MTR_DIMMS_PRESENT(mtr))
+ continue;
+
+ /* FAKE OUT VALUES, FIXME */
+ p_csrow->first_page = 0 + csrow * 20;
+ p_csrow->last_page = 9 + csrow * 20;
+ p_csrow->page_mask = 0xFFF;
+
+ p_csrow->grain = 8;
+
+ csrow_megs = 0;
+ for (channel = 0; channel < pvt->maxch; channel++) {
+ csrow_megs += pvt->dimm_info[csrow][channel].megabytes;
+ }
+
+ p_csrow->nr_pages = csrow_megs << 8;
+
+ /* Assume DDR2 for now */
+ p_csrow->mtype = MEM_FB_DDR2;
+
+ /* ask what device type on this row */
+ if (MTR_DRAM_WIDTH(mtr))
+ p_csrow->dtype = DEV_X8;
+ else
+ p_csrow->dtype = DEV_X4;
+
+ p_csrow->edac_mode = EDAC_S8ECD8ED;
+
+ empty = 0;
+ }
+
+ return empty;
+}
+
+/*
+ * i5000_enable_error_reporting
+ * Turn on the memory reporting features of the hardware
+ */
+static void i5000_enable_error_reporting(struct mem_ctl_info *mci)
+{
+ struct i5000_pvt *pvt;
+ u32 fbd_error_mask;
+
+ pvt = mci->pvt_info;
+
+ /* Read the FBD Error Mask Register */
+ pci_read_config_dword(pvt->branchmap_werrors, EMASK_FBD,
+ &fbd_error_mask);
+
+ /* Enable with a '0' */
+ fbd_error_mask &= ~(ENABLE_EMASK_ALL);
+
+ pci_write_config_dword(pvt->branchmap_werrors, EMASK_FBD,
+ fbd_error_mask);
+}
+
+/*
+ * i5000_get_dimm_and_channel_counts(pdev, &num_csrows, &num_channels)
+ *
+ * ask the device how many channels are present and how many CSROWS
+ * as well
+ */
+static void i5000_get_dimm_and_channel_counts(struct pci_dev *pdev,
+ int *num_dimms_per_channel,
+ int *num_channels)
+{
+ u8 value;
+
+ /* Need to retrieve just how many channels and dimms per channel are
+ * supported on this memory controller
+ */
+ pci_read_config_byte(pdev, MAXDIMMPERCH, &value);
+ *num_dimms_per_channel = (int)value *2;
+
+ pci_read_config_byte(pdev, MAXCH, &value);
+ *num_channels = (int)value;
+}
+
+/*
+ * i5000_probe1 Probe for ONE instance of device to see if it is
+ * present.
+ * return:
+ * 0 for FOUND a device
+ * < 0 for error code
+ */
+static int i5000_probe1(struct pci_dev *pdev, int dev_idx)
+{
+ struct mem_ctl_info *mci;
+ struct i5000_pvt *pvt;
+ int num_channels;
+ int num_dimms_per_channel;
+ int num_csrows;
+
+ debugf0("MC: " __FILE__ ": %s(), pdev bus %u dev=0x%x fn=0x%x\n",
+ __func__,
+ pdev->bus->number,
+ PCI_SLOT(pdev->devfn), PCI_FUNC(pdev->devfn));
+
+ /* We only are looking for func 0 of the set */
+ if (PCI_FUNC(pdev->devfn) != 0)
+ return -ENODEV;
+
+ /* Ask the devices for the number of CSROWS and CHANNELS so
+ * that we can calculate the memory resources, etc
+ *
+ * The Chipset will report what it can handle which will be greater
+ * or equal to what the motherboard manufacturer will implement.
+ *
+ * As we don't have a motherboard identification routine to determine
+ * actual number of slots/dimms per channel, we thus utilize the
+ * resource as specified by the chipset. Thus, we might have
+ * have more DIMMs per channel than actually on the mobo, but this
+ * allows the driver to support upto the chipset max, without
+ * some fancy mobo determination.
+ */
+ i5000_get_dimm_and_channel_counts(pdev, &num_dimms_per_channel,
+ &num_channels);
+ num_csrows = num_dimms_per_channel * 2;
+
+ debugf0("MC: %s(): Number of - Channels= %d DIMMS= %d CSROWS= %d\n",
+ __func__, num_channels, num_dimms_per_channel, num_csrows);
+
+ /* allocate a new MC control structure */
+ mci = edac_mc_alloc(sizeof(*pvt), num_csrows, num_channels);
+
+ if (mci == NULL)
+ return -ENOMEM;
+
+ debugf0("MC: " __FILE__ ": %s(): mci = %p\n", __func__, mci);
+
+ mci->dev = &pdev->dev; /* record ptr to the generic device */
+
+ pvt = mci->pvt_info;
+ pvt->system_address = pdev; /* Record this device in our private */
+ pvt->maxch = num_channels;
+ pvt->maxdimmperch = num_dimms_per_channel;
+
+ /* 'get' the pci devices we want to reserve for our use */
+ if (i5000_get_devices(mci, dev_idx))
+ goto fail0;
+
+ /* Time to get serious */
+ i5000_get_mc_regs(mci); /* retrieve the hardware registers */
+
+ mci->mtype_cap = MEM_FLAG_FB_DDR2;
+ mci->edac_ctl_cap = EDAC_FLAG_NONE;
+ mci->edac_cap = EDAC_FLAG_NONE;
+ mci->mod_name = "i5000_edac.c";
+ mci->mod_ver = I5000_REVISION;
+ mci->ctl_name = i5000_devs[dev_idx].ctl_name;
+ mci->ctl_page_to_phys = NULL;
+
+ /* Set the function pointer to an actual operation function */
+ mci->edac_check = i5000_check_error;
+
+ /* initialize the MC control structure 'csrows' table
+ * with the mapping and control information */
+ if (i5000_init_csrows(mci)) {
+ debugf0("MC: Setting mci->edac_cap to EDAC_FLAG_NONE\n"
+ " because i5000_init_csrows() returned nonzero "
+ "value\n");
+ mci->edac_cap = EDAC_FLAG_NONE; /* no csrows found */
+ } else {
+ debugf1("MC: Enable error reporting now\n");
+ i5000_enable_error_reporting(mci);
+ }
+
+ /* add this new MC control structure to EDAC's list of MCs */
+ if (edac_mc_add_mc(mci, 0)) {
+ debugf0("MC: " __FILE__
+ ": %s(): failed edac_mc_add_mc()\n", __func__);
+ /* FIXME: perhaps some code should go here that disables error
+ * reporting if we just enabled it
+ */
+ goto fail1;
+ }
+
+ i5000_clear_error(mci);
+
+ return 0;
+
+ /* Error exit unwinding stack */
+fail1:
+
+ i5000_put_devices(mci);
+
+fail0:
+ edac_mc_free(mci);
+ return -ENODEV;
+}
+
+/*
+ * i5000_init_one constructor for one instance of device
+ *
+ * returns:
+ * negative on error
+ * count (>= 0)
+ */
+static int __devinit i5000_init_one(struct pci_dev *pdev,
+ const struct pci_device_id *id)
+{
+ int rc;
+
+ debugf0("MC: " __FILE__ ": %s()\n", __func__);
+
+ /* wake up device */
+ rc = pci_enable_device(pdev);
+ if (rc == -EIO)
+ return rc;
+
+ /* now probe and enable the device */
+ return i5000_probe1(pdev, id->driver_data);
+}
+
+/*
+ * i5000_remove_one destructor for one instance of device
+ *
+ */
+static void __devexit i5000_remove_one(struct pci_dev *pdev)
+{
+ struct mem_ctl_info *mci;
+
+ debugf0(__FILE__ ": %s()\n", __func__);
+
+ if ((mci = edac_mc_del_mc(&pdev->dev)) == NULL)
+ return;
+
+ /* retrieve references to resources, and free those resources */
+ i5000_put_devices(mci);
+
+ edac_mc_free(mci);
+}
+
+/*
+ * pci_device_id table for which devices we are looking for
+ *
+ * The "E500P" device is the first device supported.
+ */
+static const struct pci_device_id i5000_pci_tbl[] __devinitdata = {
+ {PCI_DEVICE(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_I5000_DEV16),
+ .driver_data = I5000P},
+
+ {0,} /* 0 terminated list. */
+};
+
+MODULE_DEVICE_TABLE(pci, i5000_pci_tbl);
+
+/*
+ * i5000_driver pci_driver structure for this module
+ *
+ */
+static struct pci_driver i5000_driver = {
+ .name = KBUILD_BASENAME,
+ .probe = i5000_init_one,
+ .remove = __devexit_p(i5000_remove_one),
+ .id_table = i5000_pci_tbl,
+};
+
+/*
+ * i5000_init Module entry function
+ * Try to initialize this module for its devices
+ */
+static int __init i5000_init(void)
+{
+ int pci_rc;
+
+ debugf2("MC: " __FILE__ ": %s()\n", __func__);
+
+ pci_rc = pci_register_driver(&i5000_driver);
+
+ return (pci_rc < 0) ? pci_rc : 0;
+}
+
+/*
+ * i5000_exit() Module exit function
+ * Unregister the driver
+ */
+static void __exit i5000_exit(void)
+{
+ debugf2("MC: " __FILE__ ": %s()\n", __func__);
+ pci_unregister_driver(&i5000_driver);
+}
+
+module_init(i5000_init);
+module_exit(i5000_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR
+ ("Linux Networx (http://lnxi.com) Doug Thompson <norsk5@xmission.com>");
+MODULE_DESCRIPTION("MC Driver for Intel I5000 memory controllers - "
+ I5000_REVISION);
distrib
>
Scientific%20Linux
>
5x
>
x86_64
>
by-pkgid
>
27922b4260f65d317aabda37e42bbbff
>
files
>
1891
