diff options
author | Borislav Petkov <bp@suse.de> | 2020-12-15 09:18:44 +0100 |
---|---|---|
committer | Borislav Petkov <bp@suse.de> | 2020-12-28 19:36:25 +0100 |
commit | 61810096de3c3ec977c71dbb7e00447d70045163 (patch) | |
tree | 9f35f9923d3fe5700cb9cef496399f8f84e9f0ec | |
parent | 2a28ceef00bac65d6bb1757002f742806837e100 (diff) | |
download | linux-61810096de3c3ec977c71dbb7e00447d70045163.tar.gz linux-61810096de3c3ec977c71dbb7e00447d70045163.tar.bz2 linux-61810096de3c3ec977c71dbb7e00447d70045163.zip |
EDAC/amd64: Merge error injection sysfs facilities
Merge them into the main driver and put them inside an EDAC_DEBUG
ifdeffery to simplify the driver and have all debugging/injection stuff
behind a debug build-time switch.
No functional changes.
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Yazen Ghannam <yazen.ghannam@amd.com>
Link: https://lkml.kernel.org/r/20201215110517.5215-2-bp@alien8.de
-rw-r--r-- | drivers/edac/Kconfig | 7 | ||||
-rw-r--r-- | drivers/edac/Makefile | 6 | ||||
-rw-r--r-- | drivers/edac/amd64_edac.c | 235 | ||||
-rw-r--r-- | drivers/edac/amd64_edac.h | 4 | ||||
-rw-r--r-- | drivers/edac/amd64_edac_inj.c | 235 |
5 files changed, 235 insertions, 252 deletions
diff --git a/drivers/edac/Kconfig b/drivers/edac/Kconfig index 81c42664f21b..47953b06d6c8 100644 --- a/drivers/edac/Kconfig +++ b/drivers/edac/Kconfig @@ -81,10 +81,9 @@ config EDAC_AMD64 Support for error detection and correction of DRAM ECC errors on the AMD64 families (>= K8) of memory controllers. -config EDAC_AMD64_ERROR_INJECTION - bool "Sysfs HW Error injection facilities" - depends on EDAC_AMD64 - help + When EDAC_DEBUG is enabled, hardware error injection facilities + through sysfs are available: + Recent Opterons (Family 10h and later) provide for Memory Error Injection into the ECC detection circuits. The amd64_edac module allows the operator/user to inject Uncorrectable and Correctable diff --git a/drivers/edac/Makefile b/drivers/edac/Makefile index 1c70cdcf2b15..2d1641a27a28 100644 --- a/drivers/edac/Makefile +++ b/drivers/edac/Makefile @@ -44,11 +44,7 @@ obj-$(CONFIG_EDAC_IE31200) += ie31200_edac.o obj-$(CONFIG_EDAC_X38) += x38_edac.o obj-$(CONFIG_EDAC_I82860) += i82860_edac.o obj-$(CONFIG_EDAC_R82600) += r82600_edac.o - -amd64_edac_mod-y := amd64_edac.o -amd64_edac_mod-$(CONFIG_EDAC_AMD64_ERROR_INJECTION) += amd64_edac_inj.o - -obj-$(CONFIG_EDAC_AMD64) += amd64_edac_mod.o +obj-$(CONFIG_EDAC_AMD64) += amd64_edac.o obj-$(CONFIG_EDAC_PASEMI) += pasemi_edac.o diff --git a/drivers/edac/amd64_edac.c b/drivers/edac/amd64_edac.c index b00dea78541d..d55f8ef2240c 100644 --- a/drivers/edac/amd64_edac.c +++ b/drivers/edac/amd64_edac.c @@ -607,8 +607,237 @@ static struct attribute *dbg_attrs[] = { static const struct attribute_group dbg_group = { .attrs = dbg_attrs, }; -#endif /* CONFIG_EDAC_DEBUG */ +static ssize_t inject_section_show(struct device *dev, + struct device_attribute *mattr, char *buf) +{ + struct mem_ctl_info *mci = to_mci(dev); + struct amd64_pvt *pvt = mci->pvt_info; + return sprintf(buf, "0x%x\n", pvt->injection.section); +} + +/* + * store error injection section value which refers to one of 4 16-byte sections + * within a 64-byte cacheline + * + * range: 0..3 + */ +static ssize_t inject_section_store(struct device *dev, + struct device_attribute *mattr, + const char *data, size_t count) +{ + struct mem_ctl_info *mci = to_mci(dev); + struct amd64_pvt *pvt = mci->pvt_info; + unsigned long value; + int ret; + + ret = kstrtoul(data, 10, &value); + if (ret < 0) + return ret; + + if (value > 3) { + amd64_warn("%s: invalid section 0x%lx\n", __func__, value); + return -EINVAL; + } + + pvt->injection.section = (u32) value; + return count; +} + +static ssize_t inject_word_show(struct device *dev, + struct device_attribute *mattr, char *buf) +{ + struct mem_ctl_info *mci = to_mci(dev); + struct amd64_pvt *pvt = mci->pvt_info; + return sprintf(buf, "0x%x\n", pvt->injection.word); +} + +/* + * store error injection word value which refers to one of 9 16-bit word of the + * 16-byte (128-bit + ECC bits) section + * + * range: 0..8 + */ +static ssize_t inject_word_store(struct device *dev, + struct device_attribute *mattr, + const char *data, size_t count) +{ + struct mem_ctl_info *mci = to_mci(dev); + struct amd64_pvt *pvt = mci->pvt_info; + unsigned long value; + int ret; + + ret = kstrtoul(data, 10, &value); + if (ret < 0) + return ret; + + if (value > 8) { + amd64_warn("%s: invalid word 0x%lx\n", __func__, value); + return -EINVAL; + } + + pvt->injection.word = (u32) value; + return count; +} + +static ssize_t inject_ecc_vector_show(struct device *dev, + struct device_attribute *mattr, + char *buf) +{ + struct mem_ctl_info *mci = to_mci(dev); + struct amd64_pvt *pvt = mci->pvt_info; + return sprintf(buf, "0x%x\n", pvt->injection.bit_map); +} + +/* + * store 16 bit error injection vector which enables injecting errors to the + * corresponding bit within the error injection word above. When used during a + * DRAM ECC read, it holds the contents of the of the DRAM ECC bits. + */ +static ssize_t inject_ecc_vector_store(struct device *dev, + struct device_attribute *mattr, + const char *data, size_t count) +{ + struct mem_ctl_info *mci = to_mci(dev); + struct amd64_pvt *pvt = mci->pvt_info; + unsigned long value; + int ret; + + ret = kstrtoul(data, 16, &value); + if (ret < 0) + return ret; + + if (value & 0xFFFF0000) { + amd64_warn("%s: invalid EccVector: 0x%lx\n", __func__, value); + return -EINVAL; + } + + pvt->injection.bit_map = (u32) value; + return count; +} + +/* + * Do a DRAM ECC read. Assemble staged values in the pvt area, format into + * fields needed by the injection registers and read the NB Array Data Port. + */ +static ssize_t inject_read_store(struct device *dev, + struct device_attribute *mattr, + const char *data, size_t count) +{ + struct mem_ctl_info *mci = to_mci(dev); + struct amd64_pvt *pvt = mci->pvt_info; + unsigned long value; + u32 section, word_bits; + int ret; + + ret = kstrtoul(data, 10, &value); + if (ret < 0) + return ret; + + /* Form value to choose 16-byte section of cacheline */ + section = F10_NB_ARRAY_DRAM | SET_NB_ARRAY_ADDR(pvt->injection.section); + + amd64_write_pci_cfg(pvt->F3, F10_NB_ARRAY_ADDR, section); + + word_bits = SET_NB_DRAM_INJECTION_READ(pvt->injection); + + /* Issue 'word' and 'bit' along with the READ request */ + amd64_write_pci_cfg(pvt->F3, F10_NB_ARRAY_DATA, word_bits); + + edac_dbg(0, "section=0x%x word_bits=0x%x\n", section, word_bits); + + return count; +} + +/* + * Do a DRAM ECC write. Assemble staged values in the pvt area and format into + * fields needed by the injection registers. + */ +static ssize_t inject_write_store(struct device *dev, + struct device_attribute *mattr, + const char *data, size_t count) +{ + struct mem_ctl_info *mci = to_mci(dev); + struct amd64_pvt *pvt = mci->pvt_info; + u32 section, word_bits, tmp; + unsigned long value; + int ret; + + ret = kstrtoul(data, 10, &value); + if (ret < 0) + return ret; + + /* Form value to choose 16-byte section of cacheline */ + section = F10_NB_ARRAY_DRAM | SET_NB_ARRAY_ADDR(pvt->injection.section); + + amd64_write_pci_cfg(pvt->F3, F10_NB_ARRAY_ADDR, section); + + word_bits = SET_NB_DRAM_INJECTION_WRITE(pvt->injection); + + pr_notice_once("Don't forget to decrease MCE polling interval in\n" + "/sys/bus/machinecheck/devices/machinecheck<CPUNUM>/check_interval\n" + "so that you can get the error report faster.\n"); + + on_each_cpu(disable_caches, NULL, 1); + + /* Issue 'word' and 'bit' along with the READ request */ + amd64_write_pci_cfg(pvt->F3, F10_NB_ARRAY_DATA, word_bits); + + retry: + /* wait until injection happens */ + amd64_read_pci_cfg(pvt->F3, F10_NB_ARRAY_DATA, &tmp); + if (tmp & F10_NB_ARR_ECC_WR_REQ) { + cpu_relax(); + goto retry; + } + + on_each_cpu(enable_caches, NULL, 1); + + edac_dbg(0, "section=0x%x word_bits=0x%x\n", section, word_bits); + + return count; +} + +/* + * update NUM_INJ_ATTRS in case you add new members + */ + +static DEVICE_ATTR(inject_section, S_IRUGO | S_IWUSR, + inject_section_show, inject_section_store); +static DEVICE_ATTR(inject_word, S_IRUGO | S_IWUSR, + inject_word_show, inject_word_store); +static DEVICE_ATTR(inject_ecc_vector, S_IRUGO | S_IWUSR, + inject_ecc_vector_show, inject_ecc_vector_store); +static DEVICE_ATTR(inject_write, S_IWUSR, + NULL, inject_write_store); +static DEVICE_ATTR(inject_read, S_IWUSR, + NULL, inject_read_store); + +static struct attribute *inj_attrs[] = { + &dev_attr_inject_section.attr, + &dev_attr_inject_word.attr, + &dev_attr_inject_ecc_vector.attr, + &dev_attr_inject_write.attr, + &dev_attr_inject_read.attr, + NULL +}; + +static umode_t inj_is_visible(struct kobject *kobj, struct attribute *attr, int idx) +{ + struct device *dev = kobj_to_dev(kobj); + struct mem_ctl_info *mci = container_of(dev, struct mem_ctl_info, dev); + struct amd64_pvt *pvt = mci->pvt_info; + + if (pvt->fam < 0x10) + return 0; + return attr->mode; +} + +static const struct attribute_group inj_group = { + .attrs = inj_attrs, + .is_visible = inj_is_visible, +}; +#endif /* CONFIG_EDAC_DEBUG */ /* * Return the DramAddr that the SysAddr given by @sys_addr maps to. It is @@ -3469,9 +3698,7 @@ static struct amd64_family_type *per_family_init(struct amd64_pvt *pvt) static const struct attribute_group *amd64_edac_attr_groups[] = { #ifdef CONFIG_EDAC_DEBUG &dbg_group, -#endif -#ifdef CONFIG_EDAC_AMD64_ERROR_INJECTION - &amd64_edac_inj_group, + &inj_group, #endif NULL }; diff --git a/drivers/edac/amd64_edac.h b/drivers/edac/amd64_edac.h index c072ccd3e7e2..85aa820bc165 100644 --- a/drivers/edac/amd64_edac.h +++ b/drivers/edac/amd64_edac.h @@ -462,10 +462,6 @@ struct ecc_settings { } flags; }; -#ifdef CONFIG_EDAC_AMD64_ERROR_INJECTION -extern const struct attribute_group amd64_edac_inj_group; -#endif - /* * Each of the PCI Device IDs types have their own set of hardware accessor * functions and per device encoding/decoding logic. diff --git a/drivers/edac/amd64_edac_inj.c b/drivers/edac/amd64_edac_inj.c deleted file mode 100644 index d96d6116f0fb..000000000000 --- a/drivers/edac/amd64_edac_inj.c +++ /dev/null @@ -1,235 +0,0 @@ -// SPDX-License-Identifier: GPL-2.0 -#include "amd64_edac.h" - -static ssize_t amd64_inject_section_show(struct device *dev, - struct device_attribute *mattr, - char *buf) -{ - struct mem_ctl_info *mci = to_mci(dev); - struct amd64_pvt *pvt = mci->pvt_info; - return sprintf(buf, "0x%x\n", pvt->injection.section); -} - -/* - * store error injection section value which refers to one of 4 16-byte sections - * within a 64-byte cacheline - * - * range: 0..3 - */ -static ssize_t amd64_inject_section_store(struct device *dev, - struct device_attribute *mattr, - const char *data, size_t count) -{ - struct mem_ctl_info *mci = to_mci(dev); - struct amd64_pvt *pvt = mci->pvt_info; - unsigned long value; - int ret; - - ret = kstrtoul(data, 10, &value); - if (ret < 0) - return ret; - - if (value > 3) { - amd64_warn("%s: invalid section 0x%lx\n", __func__, value); - return -EINVAL; - } - - pvt->injection.section = (u32) value; - return count; -} - -static ssize_t amd64_inject_word_show(struct device *dev, - struct device_attribute *mattr, - char *buf) -{ - struct mem_ctl_info *mci = to_mci(dev); - struct amd64_pvt *pvt = mci->pvt_info; - return sprintf(buf, "0x%x\n", pvt->injection.word); -} - -/* - * store error injection word value which refers to one of 9 16-bit word of the - * 16-byte (128-bit + ECC bits) section - * - * range: 0..8 - */ -static ssize_t amd64_inject_word_store(struct device *dev, - struct device_attribute *mattr, - const char *data, size_t count) -{ - struct mem_ctl_info *mci = to_mci(dev); - struct amd64_pvt *pvt = mci->pvt_info; - unsigned long value; - int ret; - - ret = kstrtoul(data, 10, &value); - if (ret < 0) - return ret; - - if (value > 8) { - amd64_warn("%s: invalid word 0x%lx\n", __func__, value); - return -EINVAL; - } - - pvt->injection.word = (u32) value; - return count; -} - -static ssize_t amd64_inject_ecc_vector_show(struct device *dev, - struct device_attribute *mattr, - char *buf) -{ - struct mem_ctl_info *mci = to_mci(dev); - struct amd64_pvt *pvt = mci->pvt_info; - return sprintf(buf, "0x%x\n", pvt->injection.bit_map); -} - -/* - * store 16 bit error injection vector which enables injecting errors to the - * corresponding bit within the error injection word above. When used during a - * DRAM ECC read, it holds the contents of the of the DRAM ECC bits. - */ -static ssize_t amd64_inject_ecc_vector_store(struct device *dev, - struct device_attribute *mattr, - const char *data, size_t count) -{ - struct mem_ctl_info *mci = to_mci(dev); - struct amd64_pvt *pvt = mci->pvt_info; - unsigned long value; - int ret; - - ret = kstrtoul(data, 16, &value); - if (ret < 0) - return ret; - - if (value & 0xFFFF0000) { - amd64_warn("%s: invalid EccVector: 0x%lx\n", __func__, value); - return -EINVAL; - } - - pvt->injection.bit_map = (u32) value; - return count; -} - -/* - * Do a DRAM ECC read. Assemble staged values in the pvt area, format into - * fields needed by the injection registers and read the NB Array Data Port. - */ -static ssize_t amd64_inject_read_store(struct device *dev, - struct device_attribute *mattr, - const char *data, size_t count) -{ - struct mem_ctl_info *mci = to_mci(dev); - struct amd64_pvt *pvt = mci->pvt_info; - unsigned long value; - u32 section, word_bits; - int ret; - - ret = kstrtoul(data, 10, &value); - if (ret < 0) - return ret; - - /* Form value to choose 16-byte section of cacheline */ - section = F10_NB_ARRAY_DRAM | SET_NB_ARRAY_ADDR(pvt->injection.section); - - amd64_write_pci_cfg(pvt->F3, F10_NB_ARRAY_ADDR, section); - - word_bits = SET_NB_DRAM_INJECTION_READ(pvt->injection); - - /* Issue 'word' and 'bit' along with the READ request */ - amd64_write_pci_cfg(pvt->F3, F10_NB_ARRAY_DATA, word_bits); - - edac_dbg(0, "section=0x%x word_bits=0x%x\n", section, word_bits); - - return count; -} - -/* - * Do a DRAM ECC write. Assemble staged values in the pvt area and format into - * fields needed by the injection registers. - */ -static ssize_t amd64_inject_write_store(struct device *dev, - struct device_attribute *mattr, - const char *data, size_t count) -{ - struct mem_ctl_info *mci = to_mci(dev); - struct amd64_pvt *pvt = mci->pvt_info; - u32 section, word_bits, tmp; - unsigned long value; - int ret; - - ret = kstrtoul(data, 10, &value); - if (ret < 0) - return ret; - - /* Form value to choose 16-byte section of cacheline */ - section = F10_NB_ARRAY_DRAM | SET_NB_ARRAY_ADDR(pvt->injection.section); - - amd64_write_pci_cfg(pvt->F3, F10_NB_ARRAY_ADDR, section); - - word_bits = SET_NB_DRAM_INJECTION_WRITE(pvt->injection); - - pr_notice_once("Don't forget to decrease MCE polling interval in\n" - "/sys/bus/machinecheck/devices/machinecheck<CPUNUM>/check_interval\n" - "so that you can get the error report faster.\n"); - - on_each_cpu(disable_caches, NULL, 1); - - /* Issue 'word' and 'bit' along with the READ request */ - amd64_write_pci_cfg(pvt->F3, F10_NB_ARRAY_DATA, word_bits); - - retry: - /* wait until injection happens */ - amd64_read_pci_cfg(pvt->F3, F10_NB_ARRAY_DATA, &tmp); - if (tmp & F10_NB_ARR_ECC_WR_REQ) { - cpu_relax(); - goto retry; - } - - on_each_cpu(enable_caches, NULL, 1); - - edac_dbg(0, "section=0x%x word_bits=0x%x\n", section, word_bits); - - return count; -} - -/* - * update NUM_INJ_ATTRS in case you add new members - */ - -static DEVICE_ATTR(inject_section, S_IRUGO | S_IWUSR, - amd64_inject_section_show, amd64_inject_section_store); -static DEVICE_ATTR(inject_word, S_IRUGO | S_IWUSR, - amd64_inject_word_show, amd64_inject_word_store); -static DEVICE_ATTR(inject_ecc_vector, S_IRUGO | S_IWUSR, - amd64_inject_ecc_vector_show, amd64_inject_ecc_vector_store); -static DEVICE_ATTR(inject_write, S_IWUSR, - NULL, amd64_inject_write_store); -static DEVICE_ATTR(inject_read, S_IWUSR, - NULL, amd64_inject_read_store); - -static struct attribute *amd64_edac_inj_attrs[] = { - &dev_attr_inject_section.attr, - &dev_attr_inject_word.attr, - &dev_attr_inject_ecc_vector.attr, - &dev_attr_inject_write.attr, - &dev_attr_inject_read.attr, - NULL -}; - -static umode_t amd64_edac_inj_is_visible(struct kobject *kobj, - struct attribute *attr, int idx) -{ - struct device *dev = kobj_to_dev(kobj); - struct mem_ctl_info *mci = container_of(dev, struct mem_ctl_info, dev); - struct amd64_pvt *pvt = mci->pvt_info; - - if (pvt->fam < 0x10) - return 0; - return attr->mode; -} - -const struct attribute_group amd64_edac_inj_group = { - .attrs = amd64_edac_inj_attrs, - .is_visible = amd64_edac_inj_is_visible, -}; |