diff options
Diffstat (limited to 'drivers/edac/edac_mc.c')
| -rw-r--r-- | drivers/edac/edac_mc.c | 2349 |
1 files changed, 913 insertions, 1436 deletions
diff --git a/drivers/edac/edac_mc.c b/drivers/edac/edac_mc.c index 4bde30bb3be..2c694b5297c 100644 --- a/drivers/edac/edac_mc.c +++ b/drivers/edac/edac_mc.c @@ -25,1325 +25,620 @@ #include <linux/jiffies.h> #include <linux/spinlock.h> #include <linux/list.h> -#include <linux/sysdev.h> #include <linux/ctype.h> -#include <linux/kthread.h> +#include <linux/edac.h> +#include <linux/bitops.h> #include <asm/uaccess.h> #include <asm/page.h> #include <asm/edac.h> -#include "edac_mc.h" +#include "edac_core.h" +#include "edac_module.h" -#define EDAC_MC_VERSION "Ver: 2.0.1 " __DATE__ - - -#ifdef CONFIG_EDAC_DEBUG -/* Values of 0 to 4 will generate output */ -int edac_debug_level = 1; -EXPORT_SYMBOL_GPL(edac_debug_level); -#endif - -/* EDAC Controls, setable by module parameter, and sysfs */ -static int log_ue = 1; -static int log_ce = 1; -static int panic_on_ue; -static int poll_msec = 1000; +#define CREATE_TRACE_POINTS +#define TRACE_INCLUDE_PATH ../../include/ras +#include <ras/ras_event.h> /* lock to memory controller's control array */ -static DECLARE_MUTEX(mem_ctls_mutex); -static struct list_head mc_devices = LIST_HEAD_INIT(mc_devices); - -static struct task_struct *edac_thread; - -#ifdef CONFIG_PCI -static int check_pci_parity = 0; /* default YES check PCI parity */ -static int panic_on_pci_parity; /* default no panic on PCI Parity */ -static atomic_t pci_parity_count = ATOMIC_INIT(0); - -static struct kobject edac_pci_kobj; /* /sys/devices/system/edac/pci */ -static struct completion edac_pci_kobj_complete; -#endif /* CONFIG_PCI */ - -/* START sysfs data and methods */ - - -static const char *mem_types[] = { - [MEM_EMPTY] = "Empty", - [MEM_RESERVED] = "Reserved", - [MEM_UNKNOWN] = "Unknown", - [MEM_FPM] = "FPM", - [MEM_EDO] = "EDO", - [MEM_BEDO] = "BEDO", - [MEM_SDR] = "Unbuffered-SDR", - [MEM_RDR] = "Registered-SDR", - [MEM_DDR] = "Unbuffered-DDR", - [MEM_RDDR] = "Registered-DDR", - [MEM_RMBS] = "RMBS" -}; - -static const char *dev_types[] = { - [DEV_UNKNOWN] = "Unknown", - [DEV_X1] = "x1", - [DEV_X2] = "x2", - [DEV_X4] = "x4", - [DEV_X8] = "x8", - [DEV_X16] = "x16", - [DEV_X32] = "x32", - [DEV_X64] = "x64" -}; - -static const char *edac_caps[] = { - [EDAC_UNKNOWN] = "Unknown", - [EDAC_NONE] = "None", - [EDAC_RESERVED] = "Reserved", - [EDAC_PARITY] = "PARITY", - [EDAC_EC] = "EC", - [EDAC_SECDED] = "SECDED", - [EDAC_S2ECD2ED] = "S2ECD2ED", - [EDAC_S4ECD4ED] = "S4ECD4ED", - [EDAC_S8ECD8ED] = "S8ECD8ED", - [EDAC_S16ECD16ED] = "S16ECD16ED" -}; - -/* sysfs object: /sys/devices/system/edac */ -static struct sysdev_class edac_class = { - set_kset_name("edac"), -}; - -/* sysfs object: - * /sys/devices/system/edac/mc - */ -static struct kobject edac_memctrl_kobj; - -/* We use these to wait for the reference counts on edac_memctrl_kobj and - * edac_pci_kobj to reach 0. - */ -static struct completion edac_memctrl_kobj_complete; +static DEFINE_MUTEX(mem_ctls_mutex); +static LIST_HEAD(mc_devices); /* - * /sys/devices/system/edac/mc; - * data structures and methods + * Used to lock EDAC MC to just one module, avoiding two drivers e. g. + * apei/ghes and i7core_edac to be used at the same time. */ -static ssize_t memctrl_int_show(void *ptr, char *buffer) -{ - int *value = (int*) ptr; - return sprintf(buffer, "%u\n", *value); -} +static void const *edac_mc_owner; -static ssize_t memctrl_int_store(void *ptr, const char *buffer, size_t count) +static struct bus_type mc_bus[EDAC_MAX_MCS]; + +unsigned edac_dimm_info_location(struct dimm_info *dimm, char *buf, + unsigned len) { - int *value = (int*) ptr; + struct mem_ctl_info *mci = dimm->mci; + int i, n, count = 0; + char *p = buf; - if (isdigit(*buffer)) - *value = simple_strtoul(buffer, NULL, 0); + for (i = 0; i < mci->n_layers; i++) { + n = snprintf(p, len, "%s %d ", + edac_layer_name[mci->layers[i].type], + dimm->location[i]); + p += n; + len -= n; + count += n; + if (!len) + break; + } return count; } -struct memctrl_dev_attribute { - struct attribute attr; - void *value; - ssize_t (*show)(void *,char *); - ssize_t (*store)(void *, const char *, size_t); -}; +#ifdef CONFIG_EDAC_DEBUG -/* Set of show/store abstract level functions for memory control object */ -static ssize_t memctrl_dev_show(struct kobject *kobj, - struct attribute *attr, char *buffer) +static void edac_mc_dump_channel(struct rank_info *chan) { - struct memctrl_dev_attribute *memctrl_dev; - memctrl_dev = (struct memctrl_dev_attribute*)attr; - - if (memctrl_dev->show) - return memctrl_dev->show(memctrl_dev->value, buffer); - - return -EIO; + edac_dbg(4, " channel->chan_idx = %d\n", chan->chan_idx); + edac_dbg(4, " channel = %p\n", chan); + edac_dbg(4, " channel->csrow = %p\n", chan->csrow); + edac_dbg(4, " channel->dimm = %p\n", chan->dimm); } -static ssize_t memctrl_dev_store(struct kobject *kobj, struct attribute *attr, - const char *buffer, size_t count) +static void edac_mc_dump_dimm(struct dimm_info *dimm, int number) { - struct memctrl_dev_attribute *memctrl_dev; - memctrl_dev = (struct memctrl_dev_attribute*)attr; + char location[80]; - if (memctrl_dev->store) - return memctrl_dev->store(memctrl_dev->value, buffer, count); + edac_dimm_info_location(dimm, location, sizeof(location)); - return -EIO; + edac_dbg(4, "%s%i: %smapped as virtual row %d, chan %d\n", + dimm->mci->csbased ? "rank" : "dimm", + number, location, dimm->csrow, dimm->cschannel); + edac_dbg(4, " dimm = %p\n", dimm); + edac_dbg(4, " dimm->label = '%s'\n", dimm->label); + edac_dbg(4, " dimm->nr_pages = 0x%x\n", dimm->nr_pages); + edac_dbg(4, " dimm->grain = %d\n", dimm->grain); + edac_dbg(4, " dimm->nr_pages = 0x%x\n", dimm->nr_pages); } -static struct sysfs_ops memctrlfs_ops = { - .show = memctrl_dev_show, - .store = memctrl_dev_store -}; - -#define MEMCTRL_ATTR(_name,_mode,_show,_store) \ -struct memctrl_dev_attribute attr_##_name = { \ - .attr = {.name = __stringify(_name), .mode = _mode }, \ - .value = &_name, \ - .show = _show, \ - .store = _store, \ -}; - -#define MEMCTRL_STRING_ATTR(_name,_data,_mode,_show,_store) \ -struct memctrl_dev_attribute attr_##_name = { \ - .attr = {.name = __stringify(_name), .mode = _mode }, \ - .value = _data, \ - .show = _show, \ - .store = _store, \ -}; - -/* csrow<id> control files */ -MEMCTRL_ATTR(panic_on_ue,S_IRUGO|S_IWUSR,memctrl_int_show,memctrl_int_store); -MEMCTRL_ATTR(log_ue,S_IRUGO|S_IWUSR,memctrl_int_show,memctrl_int_store); -MEMCTRL_ATTR(log_ce,S_IRUGO|S_IWUSR,memctrl_int_show,memctrl_int_store); -MEMCTRL_ATTR(poll_msec,S_IRUGO|S_IWUSR,memctrl_int_show,memctrl_int_store); - -/* Base Attributes of the memory ECC object */ -static struct memctrl_dev_attribute *memctrl_attr[] = { - &attr_panic_on_ue, - &attr_log_ue, - &attr_log_ce, - &attr_poll_msec, - NULL, -}; - -/* Main MC kobject release() function */ -static void edac_memctrl_master_release(struct kobject *kobj) +static void edac_mc_dump_csrow(struct csrow_info *csrow) { - debugf1("%s()\n", __func__); - complete(&edac_memctrl_kobj_complete); + edac_dbg(4, "csrow->csrow_idx = %d\n", csrow->csrow_idx); + edac_dbg(4, " csrow = %p\n", csrow); + edac_dbg(4, " csrow->first_page = 0x%lx\n", csrow->first_page); + edac_dbg(4, " csrow->last_page = 0x%lx\n", csrow->last_page); + edac_dbg(4, " csrow->page_mask = 0x%lx\n", csrow->page_mask); + edac_dbg(4, " csrow->nr_channels = %d\n", csrow->nr_channels); + edac_dbg(4, " csrow->channels = %p\n", csrow->channels); + edac_dbg(4, " csrow->mci = %p\n", csrow->mci); } -static struct kobj_type ktype_memctrl = { - .release = edac_memctrl_master_release, - .sysfs_ops = &memctrlfs_ops, - .default_attrs = (struct attribute **) memctrl_attr, -}; - -/* Initialize the main sysfs entries for edac: - * /sys/devices/system/edac - * - * and children - * - * Return: 0 SUCCESS - * !0 FAILURE - */ -static int edac_sysfs_memctrl_setup(void) +static void edac_mc_dump_mci(struct mem_ctl_info *mci) { - int err=0; - - debugf1("%s()\n", __func__); - - /* create the /sys/devices/system/edac directory */ - err = sysdev_class_register(&edac_class); - - if (!err) { - /* Init the MC's kobject */ - memset(&edac_memctrl_kobj, 0, sizeof (edac_memctrl_kobj)); - edac_memctrl_kobj.parent = &edac_class.kset.kobj; - edac_memctrl_kobj.ktype = &ktype_memctrl; - - /* generate sysfs "..../edac/mc" */ - err = kobject_set_name(&edac_memctrl_kobj,"mc"); - - if (!err) { - /* FIXME: maybe new sysdev_create_subdir() */ - err = kobject_register(&edac_memctrl_kobj); - - if (err) - debugf1("Failed to register '.../edac/mc'\n"); - else - debugf1("Registered '.../edac/mc' kobject\n"); - } - } else - debugf1("%s() error=%d\n", __func__, err); - - return err; + edac_dbg(3, "\tmci = %p\n", mci); + edac_dbg(3, "\tmci->mtype_cap = %lx\n", mci->mtype_cap); + edac_dbg(3, "\tmci->edac_ctl_cap = %lx\n", mci->edac_ctl_cap); + edac_dbg(3, "\tmci->edac_cap = %lx\n", mci->edac_cap); + edac_dbg(4, "\tmci->edac_check = %p\n", mci->edac_check); + edac_dbg(3, "\tmci->nr_csrows = %d, csrows = %p\n", + mci->nr_csrows, mci->csrows); + edac_dbg(3, "\tmci->nr_dimms = %d, dimms = %p\n", + mci->tot_dimms, mci->dimms); + edac_dbg(3, "\tdev = %p\n", mci->pdev); + edac_dbg(3, "\tmod_name:ctl_name = %s:%s\n", + mci->mod_name, mci->ctl_name); + edac_dbg(3, "\tpvt_info = %p\n\n", mci->pvt_info); } +#endif /* CONFIG_EDAC_DEBUG */ + /* - * MC teardown: - * the '..../edac/mc' kobject followed by '..../edac' itself + * keep those in sync with the enum mem_type */ -static void edac_sysfs_memctrl_teardown(void) -{ - debugf0("MC: " __FILE__ ": %s()\n", __func__); - - /* Unregister the MC's kobject and wait for reference count to reach - * 0. - */ - init_completion(&edac_memctrl_kobj_complete); - kobject_unregister(&edac_memctrl_kobj); - wait_for_completion(&edac_memctrl_kobj_complete); - - /* Unregister the 'edac' object */ - sysdev_class_unregister(&edac_class); -} - -#ifdef CONFIG_PCI -static ssize_t edac_pci_int_show(void *ptr, char *buffer) -{ - int *value = ptr; - return sprintf(buffer,"%d\n",*value); -} - -static ssize_t edac_pci_int_store(void *ptr, const char *buffer, size_t count) -{ - int *value = ptr; - - if (isdigit(*buffer)) - *value = simple_strtoul(buffer,NULL,0); - - return count; -} - -struct edac_pci_dev_attribute { - struct attribute attr; - void *value; - ssize_t (*show)(void *,char *); - ssize_t (*store)(void *, const char *,size_t); -}; - -/* Set of show/store abstract level functions for PCI Parity object */ -static ssize_t edac_pci_dev_show(struct kobject *kobj, struct attribute *attr, - char *buffer) -{ - struct edac_pci_dev_attribute *edac_pci_dev; - edac_pci_dev= (struct edac_pci_dev_attribute*)attr; - - if (edac_pci_dev->show) - return edac_pci_dev->show(edac_pci_dev->value, buffer); - return -EIO; -} - -static ssize_t edac_pci_dev_store(struct kobject *kobj, - struct attribute *attr, const char *buffer, size_t count) -{ - struct edac_pci_dev_attribute *edac_pci_dev; - edac_pci_dev= (struct edac_pci_dev_attribute*)attr; - - if (edac_pci_dev->show) - return edac_pci_dev->store(edac_pci_dev->value, buffer, count); - return -EIO; -} - -static struct sysfs_ops edac_pci_sysfs_ops = { - .show = edac_pci_dev_show, - .store = edac_pci_dev_store -}; - -#define EDAC_PCI_ATTR(_name,_mode,_show,_store) \ -struct edac_pci_dev_attribute edac_pci_attr_##_name = { \ - .attr = {.name = __stringify(_name), .mode = _mode }, \ - .value = &_name, \ - .show = _show, \ - .store = _store, \ -}; - -#define EDAC_PCI_STRING_ATTR(_name,_data,_mode,_show,_store) \ -struct edac_pci_dev_attribute edac_pci_attr_##_name = { \ - .attr = {.name = __stringify(_name), .mode = _mode }, \ - .value = _data, \ - .show = _show, \ - .store = _store, \ -}; - -/* PCI Parity control files */ -EDAC_PCI_ATTR(check_pci_parity, S_IRUGO|S_IWUSR, edac_pci_int_show, - edac_pci_int_store); -EDAC_PCI_ATTR(panic_on_pci_parity, S_IRUGO|S_IWUSR, edac_pci_int_show, - edac_pci_int_store); -EDAC_PCI_ATTR(pci_parity_count, S_IRUGO, edac_pci_int_show, NULL); - -/* Base Attributes of the memory ECC object */ -static struct edac_pci_dev_attribute *edac_pci_attr[] = { - &edac_pci_attr_check_pci_parity, - &edac_pci_attr_panic_on_pci_parity, - &edac_pci_attr_pci_parity_count, - NULL, -}; - -/* No memory to release */ -static void edac_pci_release(struct kobject *kobj) -{ - debugf1("%s()\n", __func__); - complete(&edac_pci_kobj_complete); -} - -static struct kobj_type ktype_edac_pci = { - .release = edac_pci_release, - .sysfs_ops = &edac_pci_sysfs_ops, - .default_attrs = (struct attribute **) edac_pci_attr, +const char *edac_mem_types[] = { + "Empty csrow", + "Reserved csrow type", + "Unknown csrow type", + "Fast page mode RAM", + "Extended data out RAM", + "Burst Extended data out RAM", + "Single data rate SDRAM", + "Registered single data rate SDRAM", + "Double data rate SDRAM", + "Registered Double data rate SDRAM", + "Rambus DRAM", + "Unbuffered DDR2 RAM", + "Fully buffered DDR2", + "Registered DDR2 RAM", + "Rambus XDR", + "Unbuffered DDR3 RAM", + "Registered DDR3 RAM", }; +EXPORT_SYMBOL_GPL(edac_mem_types); /** - * edac_sysfs_pci_setup() + * edac_align_ptr - Prepares the pointer offsets for a single-shot allocation + * @p: pointer to a pointer with the memory offset to be used. At + * return, this will be incremented to point to the next offset + * @size: Size of the data structure to be reserved + * @n_elems: Number of elements that should be reserved + * + * If 'size' is a constant, the compiler will optimize this whole function + * down to either a no-op or the addition of a constant to the value of '*p'. * + * The 'p' pointer is absolutely needed to keep the proper advancing + * further in memory to the proper offsets when allocating the struct along + * with its embedded structs, as edac_device_alloc_ctl_info() does it + * above, for example. + * + * At return, the pointer 'p' will be incremented to be used on a next call + * to this function. */ -static int edac_sysfs_pci_setup(void) +void *edac_align_ptr(void **p, unsigned size, int n_elems) { - int err; - - debugf1("%s()\n", __func__); - - memset(&edac_pci_kobj, 0, sizeof(edac_pci_kobj)); - edac_pci_kobj.parent = &edac_class.kset.kobj; - edac_pci_kobj.ktype = &ktype_edac_pci; - err = kobject_set_name(&edac_pci_kobj, "pci"); - - if (!err) { - /* Instanstiate the csrow object */ - /* FIXME: maybe new sysdev_create_subdir() */ - err = kobject_register(&edac_pci_kobj); - - if (err) - debugf1("Failed to register '.../edac/pci'\n"); - else - debugf1("Registered '.../edac/pci' kobject\n"); - } - - return err; -} - -static void edac_sysfs_pci_teardown(void) -{ - debugf0("%s()\n", __func__); - init_completion(&edac_pci_kobj_complete); - kobject_unregister(&edac_pci_kobj); - wait_for_completion(&edac_pci_kobj_complete); -} - - -static u16 get_pci_parity_status(struct pci_dev *dev, int secondary) -{ - int where; - u16 status; + unsigned align, r; + void *ptr = *p; - where = secondary ? PCI_SEC_STATUS : PCI_STATUS; - pci_read_config_word(dev, where, &status); + *p += size * n_elems; - /* If we get back 0xFFFF then we must suspect that the card has been - * pulled but the Linux PCI layer has not yet finished cleaning up. - * We don't want to report on such devices + /* + * 'p' can possibly be an unaligned item X such that sizeof(X) is + * 'size'. Adjust 'p' so that its alignment is at least as + * stringent as what the compiler would provide for X and return + * the aligned result. + * Here we assume that the alignment of a "long long" is the most + * stringent alignment that the compiler will ever provide by default. + * As far as I know, this is a reasonable assumption. */ + if (size > sizeof(long)) + align = sizeof(long long); + else if (size > sizeof(int)) + align = sizeof(long); + else if (size > sizeof(short)) + align = sizeof(int); + else if (size > sizeof(char)) + align = sizeof(short); + else + return (char *)ptr; - if (status == 0xFFFF) { - u32 sanity; - - pci_read_config_dword(dev, 0, &sanity); - - if (sanity == 0xFFFFFFFF) - return 0; - } + r = (unsigned long)p % align; - status &= PCI_STATUS_DETECTED_PARITY | PCI_STATUS_SIG_SYSTEM_ERROR | - PCI_STATUS_PARITY; + if (r == 0) + return (char *)ptr; - if (status) - /* reset only the bits we are interested in */ - pci_write_config_word(dev, where, status); + *p += align - r; - return status; + return (void *)(((unsigned long)ptr) + align - r); } -typedef void (*pci_parity_check_fn_t) (struct pci_dev *dev); - -/* Clear any PCI parity errors logged by this device. */ -static void edac_pci_dev_parity_clear(struct pci_dev *dev) +static void _edac_mc_free(struct mem_ctl_info *mci) { - u8 header_type; - - get_pci_parity_status(dev, 0); - - /* read the device TYPE, looking for bridges */ - pci_read_config_byte(dev, PCI_HEADER_TYPE, &header_type); + int i, chn, row; + struct csrow_info *csr; + const unsigned int tot_dimms = mci->tot_dimms; + const unsigned int tot_channels = mci->num_cschannel; + const unsigned int tot_csrows = mci->nr_csrows; - if ((header_type & 0x7F) == PCI_HEADER_TYPE_BRIDGE) - get_pci_parity_status(dev, 1); -} - -/* - * PCI Parity polling - * - */ -static void edac_pci_dev_parity_test(struct pci_dev *dev) -{ - u16 status; - u8 header_type; - - /* read the STATUS register on this device - */ - status = get_pci_parity_status(dev, 0); - - debugf2("PCI STATUS= 0x%04x %s\n", status, dev->dev.bus_id ); - - /* check the status reg for errors */ - if (status) { - if (status & (PCI_STATUS_SIG_SYSTEM_ERROR)) - edac_printk(KERN_CRIT, EDAC_PCI, - "Signaled System Error on %s\n", - pci_name(dev)); - - if (status & (PCI_STATUS_PARITY)) { - edac_printk(KERN_CRIT, EDAC_PCI, - "Master Data Parity Error on %s\n", - pci_name(dev)); - - atomic_inc(&pci_parity_count); - } - - if (status & (PCI_STATUS_DETECTED_PARITY)) { - edac_printk(KERN_CRIT, EDAC_PCI, - "Detected Parity Error on %s\n", - pci_name(dev)); - - atomic_inc(&pci_parity_count); - } + if (mci->dimms) { + for (i = 0; i < tot_dimms; i++) + kfree(mci->dimms[i]); + kfree(mci->dimms); } - - /* read the device TYPE, looking for bridges */ - pci_read_config_byte(dev, PCI_HEADER_TYPE, &header_type); - - debugf2("PCI HEADER TYPE= 0x%02x %s\n", header_type, dev->dev.bus_id ); - - if ((header_type & 0x7F) == PCI_HEADER_TYPE_BRIDGE) { - /* On bridges, need to examine secondary status register */ - status = get_pci_parity_status(dev, 1); - - debugf2("PCI SEC_STATUS= 0x%04x %s\n", - status, dev->dev.bus_id ); - - /* check the secondary status reg for errors */ - if (status) { - if (status & (PCI_STATUS_SIG_SYSTEM_ERROR)) - edac_printk(KERN_CRIT, EDAC_PCI, "Bridge " - "Signaled System Error on %s\n", - pci_name(dev)); - - if (status & (PCI_STATUS_PARITY)) { - edac_printk(KERN_CRIT, EDAC_PCI, "Bridge " - "Master Data Parity Error on " - "%s\n", pci_name(dev)); - - atomic_inc(&pci_parity_count); - } - - if (status & (PCI_STATUS_DETECTED_PARITY)) { - edac_printk(KERN_CRIT, EDAC_PCI, "Bridge " - "Detected Parity Error on %s\n", - pci_name(dev)); - - atomic_inc(&pci_parity_count); + if (mci->csrows) { + for (row = 0; row < tot_csrows; row++) { + csr = mci->csrows[row]; + if (csr) { + if (csr->channels) { + for (chn = 0; chn < tot_channels; chn++) + kfree(csr->channels[chn]); + kfree(csr->channels); + } + kfree(csr); } } + kfree(mci->csrows); } + kfree(mci); } -/* - * pci_dev parity list iterator - * Scan the PCI device list for one iteration, looking for SERRORs - * Master Parity ERRORS or Parity ERRORs on primary or secondary devices +/** + * edac_mc_alloc: Allocate and partially fill a struct mem_ctl_info structure + * @mc_num: Memory controller number + * @n_layers: Number of MC hierarchy layers + * layers: Describes each layer as seen by the Memory Controller + * @size_pvt: size of private storage needed + * + * + * Everything is kmalloc'ed as one big chunk - more efficient. + * Only can be used if all structures have the same lifetime - otherwise + * you have to allocate and initialize your own structures. + * + * Use edac_mc_free() to free mc structures allocated by this function. + * + * NOTE: drivers handle multi-rank memories in different ways: in some + * drivers, one multi-rank memory stick is mapped as one entry, while, in + * others, a single multi-rank memory stick would be mapped into several + * entries. Currently, this function will allocate multiple struct dimm_info + * on such scenarios, as grouping the multiple ranks require drivers change. + * + * Returns: + * On failure: NULL + * On success: struct mem_ctl_info pointer */ -static inline void edac_pci_dev_parity_iterator(pci_parity_check_fn_t fn) +struct mem_ctl_info *edac_mc_alloc(unsigned mc_num, + unsigned n_layers, + struct edac_mc_layer *layers, + unsigned sz_pvt) { - struct pci_dev *dev = NULL; - - /* request for kernel access to the next PCI device, if any, - * and while we are looking at it have its reference count - * bumped until we are done with it + struct mem_ctl_info *mci; + struct edac_mc_layer *layer; + struct csrow_info *csr; + struct rank_info *chan; + struct dimm_info *dimm; + u32 *ce_per_layer[EDAC_MAX_LAYERS], *ue_per_layer[EDAC_MAX_LAYERS]; + unsigned pos[EDAC_MAX_LAYERS]; + unsigned size, tot_dimms = 1, count = 1; + unsigned tot_csrows = 1, tot_channels = 1, tot_errcount = 0; + void *pvt, *p, *ptr = NULL; + int i, j, row, chn, n, len, off; + bool per_rank = false; + + BUG_ON(n_layers > EDAC_MAX_LAYERS || n_layers == 0); + /* + * Calculate the total amount of dimms and csrows/cschannels while + * in the old API emulation mode */ - while((dev = pci_get_device(PCI_ANY_ID, PCI_ANY_ID, dev)) != NULL) { - fn(dev); - } -} - -static void do_pci_parity_check(void) -{ - unsigned long flags; - int before_count; - - debugf3("%s()\n", __func__); - - if (!check_pci_parity) - return; - - before_count = atomic_read(&pci_parity_count); + for (i = 0; i < n_layers; i++) { + tot_dimms *= layers[i].size; + if (layers[i].is_virt_csrow) + tot_csrows *= layers[i].size; + else + tot_channels *= layers[i].size; - /* scan all PCI devices looking for a Parity Error on devices and - * bridges - */ - local_irq_save(flags); - edac_pci_dev_parity_iterator(edac_pci_dev_parity_test); - local_irq_restore(flags); - - /* Only if operator has selected panic on PCI Error */ - if (panic_on_pci_parity) { - /* If the count is different 'after' from 'before' */ - if (before_count != atomic_read(&pci_parity_count)) - panic("EDAC: PCI Parity Error"); + if (layers[i].type == EDAC_MC_LAYER_CHIP_SELECT) + per_rank = true; } -} -static inline void clear_pci_parity_errors(void) -{ - /* Clear any PCI bus parity errors that devices initially have logged - * in their registers. + /* Figure out the offsets of the various items from the start of an mc + * structure. We want the alignment of each item to be at least as + * stringent as what the compiler would provide if we could simply + * hardcode everything into a single struct. */ - edac_pci_dev_parity_iterator(edac_pci_dev_parity_clear); -} - -#else /* CONFIG_PCI */ - -/* pre-process these away */ -#define do_pci_parity_check() -#define clear_pci_parity_errors() -#define edac_sysfs_pci_teardown() -#define edac_sysfs_pci_setup() (0) - -#endif /* CONFIG_PCI */ - -/* EDAC sysfs CSROW data structures and methods - */ - -/* Set of more default csrow<id> attribute show/store functions */ -static ssize_t csrow_ue_count_show(struct csrow_info *csrow, char *data, int private) -{ - return sprintf(data,"%u\n", csrow->ue_count); -} - -static ssize_t csrow_ce_count_show(struct csrow_info *csrow, char *data, int private) -{ - return sprintf(data,"%u\n", csrow->ce_count); -} - -static ssize_t csrow_size_show(struct csrow_info *csrow, char *data, int private) -{ - return sprintf(data,"%u\n", PAGES_TO_MiB(csrow->nr_pages)); -} - -static ssize_t csrow_mem_type_show(struct csrow_info *csrow, char *data, int private) -{ - return sprintf(data,"%s\n", mem_types[csrow->mtype]); -} - -static ssize_t csrow_dev_type_show(struct csrow_info *csrow, char *data, int private) -{ - return sprintf(data,"%s\n", dev_types[csrow->dtype]); -} - -static ssize_t csrow_edac_mode_show(struct csrow_info *csrow, char *data, int private) -{ - return sprintf(data,"%s\n", edac_caps[csrow->edac_mode]); -} - -/* show/store functions for DIMM Label attributes */ -static ssize_t channel_dimm_label_show(struct csrow_info *csrow, - char *data, int channel) -{ - return snprintf(data, EDAC_MC_LABEL_LEN,"%s", - csrow->channels[channel].label); -} - -static ssize_t channel_dimm_label_store(struct csrow_info *csrow, - const char *data, - size_t count, - int channel) -{ - ssize_t max_size = 0; - - max_size = min((ssize_t)count,(ssize_t)EDAC_MC_LABEL_LEN-1); - strncpy(csrow->channels[channel].label, data, max_size); - csrow->channels[channel].label[max_size] = '\0'; - - return max_size; -} - -/* show function for dynamic chX_ce_count attribute */ -static ssize_t channel_ce_count_show(struct csrow_info *csrow, - char *data, - int channel) -{ - return sprintf(data, "%u\n", csrow->channels[channel].ce_count); -} - -/* csrow specific attribute structure */ -struct csrowdev_attribute { - struct attribute attr; - ssize_t (*show)(struct csrow_info *,char *,int); - ssize_t (*store)(struct csrow_info *, const char *,size_t,int); - int private; -}; - -#define to_csrow(k) container_of(k, struct csrow_info, kobj) -#define to_csrowdev_attr(a) container_of(a, struct csrowdev_attribute, attr) - -/* Set of show/store higher level functions for default csrow attributes */ -static ssize_t csrowdev_show(struct kobject *kobj, - struct attribute *attr, - char *buffer) -{ - struct csrow_info *csrow = to_csrow(kobj); - struct csrowdev_attribute *csrowdev_attr = to_csrowdev_attr(attr); - - if (csrowdev_attr->show) - return csrowdev_attr->show(csrow, - buffer, - csrowdev_attr->private); - return -EIO; -} - -static ssize_t csrowdev_store(struct kobject *kobj, struct attribute *attr, - const char *buffer, size_t count) -{ - struct csrow_info *csrow = to_csrow(kobj); - struct csrowdev_attribute * csrowdev_attr = to_csrowdev_attr(attr); - - if (csrowdev_attr->store) - return csrowdev_attr->store(csrow, - buffer, - count, - csrowdev_attr->private); - return -EIO; -} - -static struct sysfs_ops csrowfs_ops = { - .show = csrowdev_show, - .store = csrowdev_store -}; - -#define CSROWDEV_ATTR(_name,_mode,_show,_store,_private) \ -struct csrowdev_attribute attr_##_name = { \ - .attr = {.name = __stringify(_name), .mode = _mode }, \ - .show = _show, \ - .store = _store, \ - .private = _private, \ -}; - -/* default cwrow<id>/attribute files */ -CSROWDEV_ATTR(size_mb,S_IRUGO,csrow_size_show,NULL,0); -CSROWDEV_ATTR(dev_type,S_IRUGO,csrow_dev_type_show,NULL,0); -CSROWDEV_ATTR(mem_type,S_IRUGO,csrow_mem_type_show,NULL,0); -CSROWDEV_ATTR(edac_mode,S_IRUGO,csrow_edac_mode_show,NULL,0); -CSROWDEV_ATTR(ue_count,S_IRUGO,csrow_ue_count_show,NULL,0); -CSROWDEV_ATTR(ce_count,S_IRUGO,csrow_ce_count_show,NULL,0); - -/* default attributes of the CSROW<id> object */ -static struct csrowdev_attribute *default_csrow_attr[] = { - &attr_dev_type, - &attr_mem_type, - &attr_edac_mode, - &attr_size_mb, - &attr_ue_count, - &attr_ce_count, - NULL, -}; - - -/* possible dynamic channel DIMM Label attribute files */ -CSROWDEV_ATTR(ch0_dimm_label,S_IRUGO|S_IWUSR, - channel_dimm_label_show, - channel_dimm_label_store, - 0 ); -CSROWDEV_ATTR(ch1_dimm_label,S_IRUGO|S_IWUSR, - channel_dimm_label_show, - channel_dimm_label_store, - 1 ); -CSROWDEV_ATTR(ch2_dimm_label,S_IRUGO|S_IWUSR, - channel_dimm_label_show, - channel_dimm_label_store, - 2 ); -CSROWDEV_ATTR(ch3_dimm_label,S_IRUGO|S_IWUSR, - channel_dimm_label_show, - channel_dimm_label_store, - 3 ); -CSROWDEV_ATTR(ch4_dimm_label,S_IRUGO|S_IWUSR, - channel_dimm_label_show, - channel_dimm_label_store, - 4 ); -CSROWDEV_ATTR(ch5_dimm_label,S_IRUGO|S_IWUSR, - channel_dimm_label_show, - channel_dimm_label_store, - 5 ); - -/* Total possible dynamic DIMM Label attribute file table */ -static struct csrowdev_attribute *dynamic_csrow_dimm_attr[] = { - &attr_ch0_dimm_label, - &attr_ch1_dimm_label, - &attr_ch2_dimm_label, - &attr_ch3_dimm_label, - &attr_ch4_dimm_label, - &attr_ch5_dimm_label -}; - -/* possible dynamic channel ce_count attribute files */ -CSROWDEV_ATTR(ch0_ce_count,S_IRUGO|S_IWUSR, - channel_ce_count_show, - NULL, - 0 ); -CSROWDEV_ATTR(ch1_ce_count,S_IRUGO|S_IWUSR, - channel_ce_count_show, - NULL, - 1 ); -CSROWDEV_ATTR(ch2_ce_count,S_IRUGO|S_IWUSR, - channel_ce_count_show, - NULL, - 2 ); -CSROWDEV_ATTR(ch3_ce_count,S_IRUGO|S_IWUSR, - channel_ce_count_show, - NULL, - 3 ); -CSROWDEV_ATTR(ch4_ce_count,S_IRUGO|S_IWUSR, - channel_ce_count_show, - NULL, - 4 ); -CSROWDEV_ATTR(ch5_ce_count,S_IRUGO|S_IWUSR, - channel_ce_count_show, - NULL, - 5 ); - -/* Total possible dynamic ce_count attribute file table */ -static struct csrowdev_attribute *dynamic_csrow_ce_count_attr[] = { - &attr_ch0_ce_count, - &attr_ch1_ce_count, - &attr_ch2_ce_count, - &attr_ch3_ce_count, - &attr_ch4_ce_count, - &attr_ch5_ce_count -}; - - -#define EDAC_NR_CHANNELS 6 - -/* Create dynamic CHANNEL files, indexed by 'chan', under specifed CSROW */ -static int edac_create_channel_files(struct kobject *kobj, int chan) -{ - int err=-ENODEV; - - if (chan >= EDAC_NR_CHANNELS) - return err; - - /* create the DIMM label attribute file */ - err = sysfs_create_file(kobj, - (struct attribute *) dynamic_csrow_dimm_attr[chan]); - - if (!err) { - /* create the CE Count attribute file */ - err = sysfs_create_file(kobj, - (struct attribute *) dynamic_csrow_ce_count_attr[chan]); - } else { - debugf1("%s() dimm labels and ce_count files created", __func__); + mci = edac_align_ptr(&ptr, sizeof(*mci), 1); + layer = edac_align_ptr(&ptr, sizeof(*layer), n_layers); + for (i = 0; i < n_layers; i++) { + count *= layers[i].size; + edac_dbg(4, "errcount layer %d size %d\n", i, count); + ce_per_layer[i] = edac_align_ptr(&ptr, sizeof(u32), count); + ue_per_layer[i] = edac_align_ptr(&ptr, sizeof(u32), count); + tot_errcount += 2 * count; } - return err; -} - -/* No memory to release for this kobj */ -static void edac_csrow_instance_release(struct kobject *kobj) -{ - struct csrow_info *cs; + edac_dbg(4, "allocating %d error counters\n", tot_errcount); + pvt = edac_align_ptr(&ptr, sz_pvt, 1); + size = ((unsigned long)pvt) + sz_pvt; - cs = container_of(kobj, struct csrow_info, kobj); - complete(&cs->kobj_complete); -} + edac_dbg(1, "allocating %u bytes for mci data (%d %s, %d csrows/channels)\n", + size, + tot_dimms, + per_rank ? "ranks" : "dimms", + tot_csrows * tot_channels); -/* the kobj_type instance for a CSROW */ -static struct kobj_type ktype_csrow = { - .release = edac_csrow_instance_release, - .sysfs_ops = &csrowfs_ops, - .default_attrs = (struct attribute **) default_csrow_attr, -}; + mci = kzalloc(size, GFP_KERNEL); + if (mci == NULL) + return NULL; -/* Create a CSROW object under specifed edac_mc_device */ -static int edac_create_csrow_object( - struct kobject *edac_mci_kobj, - struct csrow_info *csrow, - int index) -{ - int err = 0; - int chan; + /* Adjust pointers so they point within the memory we just allocated + * rather than an imaginary chunk of memory located at address 0. + */ + layer = (struct edac_mc_layer *)(((char *)mci) + ((unsigned long)layer)); + for (i = 0; i < n_layers; i++) { + mci->ce_per_layer[i] = (u32 *)((char *)mci + ((unsigned long)ce_per_layer[i])); + mci->ue_per_layer[i] = (u32 *)((char *)mci + ((unsigned long)ue_per_layer[i])); + } + pvt = sz_pvt ? (((char *)mci) + ((unsigned long)pvt)) : NULL; - memset(&csrow->kobj, 0, sizeof(csrow->kobj)); + /* setup index and various internal pointers */ + mci->mc_idx = mc_num; + mci->tot_dimms = tot_dimms; + mci->pvt_info = pvt; + mci->n_layers = n_layers; + mci->layers = layer; + memcpy(mci->layers, layers, sizeof(*layer) * n_layers); + mci->nr_csrows = tot_csrows; + mci->num_cschannel = tot_channels; + mci->csbased = per_rank; - /* generate ..../edac/mc/mc<id>/csrow<index> */ + /* + * Alocate and fill the csrow/channels structs + */ + mci->csrows = kcalloc(tot_csrows, sizeof(*mci->csrows), GFP_KERNEL); + if (!mci->csrows) + goto error; + for (row = 0; row < tot_csrows; row++) { + csr = kzalloc(sizeof(**mci->csrows), GFP_KERNEL); + if (!csr) + goto error; + mci->csrows[row] = csr; + csr->csrow_idx = row; + csr->mci = mci; + csr->nr_channels = tot_channels; + csr->channels = kcalloc(tot_channels, sizeof(*csr->channels), + GFP_KERNEL); + if (!csr->channels) + goto error; + + for (chn = 0; chn < tot_channels; chn++) { + chan = kzalloc(sizeof(**csr->channels), GFP_KERNEL); + if (!chan) + goto error; + csr->channels[chn] = chan; + chan->chan_idx = chn; + chan->csrow = csr; + } + } - csrow->kobj.parent = edac_mci_kobj; - csrow->kobj.ktype = &ktype_csrow; + /* + * Allocate and fill the dimm structs + */ + mci->dimms = kcalloc(tot_dimms, sizeof(*mci->dimms), GFP_KERNEL); + if (!mci->dimms) + goto error; + + memset(&pos, 0, sizeof(pos)); + row = 0; + chn = 0; + for (i = 0; i < tot_dimms; i++) { + chan = mci->csrows[row]->channels[chn]; + off = EDAC_DIMM_OFF(layer, n_layers, pos[0], pos[1], pos[2]); + if (off < 0 || off >= tot_dimms) { + edac_mc_printk(mci, KERN_ERR, "EDAC core bug: EDAC_DIMM_OFF is trying to do an illegal data access\n"); + goto error; + } - /* name this instance of csrow<id> */ - err = kobject_set_name(&csrow->kobj,"csrow%d",index); - if (err) - goto error_exit; + dimm = kzalloc(sizeof(**mci->dimms), GFP_KERNEL); + if (!dimm) + goto error; + mci->dimms[off] = dimm; + dimm->mci = mci; - /* Instanstiate the csrow object */ - err = kobject_register(&csrow->kobj); - if (!err) { - /* Create the dyanmic attribute files on this csrow, - * namely, the DIMM labels and the channel ce_count + /* + * Copy DIMM location and initialize it. */ - for (chan = 0; chan < csrow->nr_channels; chan++) { - err = edac_create_channel_files(&csrow->kobj,chan); - if (err) + len = sizeof(dimm->label); + p = dimm->label; + n = snprintf(p, len, "mc#%u", mc_num); + p += n; + len -= n; + for (j = 0; j < n_layers; j++) { + n = snprintf(p, len, "%s#%u", + edac_layer_name[layers[j].type], + pos[j]); + p += n; + len -= n; + dimm->location[j] = pos[j]; + + if (len <= 0) break; } - } -error_exit: - return err; -} - -/* default sysfs methods and data structures for the main MCI kobject */ - -static ssize_t mci_reset_counters_store(struct mem_ctl_info *mci, - const char *data, size_t count) -{ - int row, chan; - - mci->ue_noinfo_count = 0; - mci->ce_noinfo_count = 0; - mci->ue_count = 0; - mci->ce_count = 0; - - for (row = 0; row < mci->nr_csrows; row++) { - struct csrow_info *ri = &mci->csrows[row]; - - ri->ue_count = 0; - ri->ce_count = 0; + /* Link it to the csrows old API data */ + chan->dimm = dimm; + dimm->csrow = row; + dimm->cschannel = chn; + + /* Increment csrow location */ + if (layers[0].is_virt_csrow) { + chn++; + if (chn == tot_channels) { + chn = 0; + row++; + } + } else { + row++; + if (row == tot_csrows) { + row = 0; + chn++; + } + } - for (chan = 0; chan < ri->nr_channels; chan++) - ri->channels[chan].ce_count = 0; + /* Increment dimm location */ + for (j = n_layers - 1; j >= 0; j--) { + pos[j]++; + if (pos[j] < layers[j].size) + break; + pos[j] = 0; + } } - mci->start_time = jiffies; - return count; -} - -/* default attribute files for the MCI object */ -static ssize_t mci_ue_count_show(struct mem_ctl_info *mci, char *data) -{ - return sprintf(data,"%d\n", mci->ue_count); -} - -static ssize_t mci_ce_count_show(struct mem_ctl_info *mci, char *data) -{ - return sprintf(data,"%d\n", mci->ce_count); -} + mci->op_state = OP_ALLOC; -static ssize_t mci_ce_noinfo_show(struct mem_ctl_info *mci, char *data) -{ - return sprintf(data,"%d\n", mci->ce_noinfo_count); -} - -static ssize_t mci_ue_noinfo_show(struct mem_ctl_info *mci, char *data) -{ - return sprintf(data,"%d\n", mci->ue_noinfo_count); -} + return mci; -static ssize_t mci_seconds_show(struct mem_ctl_info *mci, char *data) -{ - return sprintf(data,"%ld\n", (jiffies - mci->start_time) / HZ); -} +error: + _edac_mc_free(mci); -static ssize_t mci_ctl_name_show(struct mem_ctl_info *mci, char *data) -{ - return sprintf(data,"%s\n", mci->ctl_name); + return NULL; } +EXPORT_SYMBOL_GPL(edac_mc_alloc); -static ssize_t mci_size_mb_show(struct mem_ctl_info *mci, char *data) +/** + * edac_mc_free + * 'Free' a previously allocated 'mci' structure + * @mci: pointer to a struct mem_ctl_info structure + */ +void edac_mc_free(struct mem_ctl_info *mci) { - int total_pages, csrow_idx; + edac_dbg(1, "\n"); - for (total_pages = csrow_idx = 0; csrow_idx < mci->nr_csrows; - csrow_idx++) { - struct csrow_info *csrow = &mci->csrows[csrow_idx]; - - if (!csrow->nr_pages) - continue; - - total_pages += csrow->nr_pages; + /* If we're not yet registered with sysfs free only what was allocated + * in edac_mc_alloc(). + */ + if (!device_is_registered(&mci->dev)) { + _edac_mc_free(mci); + return; } - return sprintf(data,"%u\n", PAGES_TO_MiB(total_pages)); -} - -struct mcidev_attribute { - struct attribute attr; - ssize_t (*show)(struct mem_ctl_info *,char *); - ssize_t (*store)(struct mem_ctl_info *, const char *,size_t); -}; - -#define to_mci(k) container_of(k, struct mem_ctl_info, edac_mci_kobj) -#define to_mcidev_attr(a) container_of(a, struct mcidev_attribute, attr) - -/* MCI show/store functions for top most object */ -static ssize_t mcidev_show(struct kobject *kobj, struct attribute *attr, - char *buffer) -{ - struct mem_ctl_info *mem_ctl_info = to_mci(kobj); - struct mcidev_attribute * mcidev_attr = to_mcidev_attr(attr); - - if (mcidev_attr->show) - return mcidev_attr->show(mem_ctl_info, buffer); - - return -EIO; -} - -static ssize_t mcidev_store(struct kobject *kobj, struct attribute *attr, - const char *buffer, size_t count) -{ - struct mem_ctl_info *mem_ctl_info = to_mci(kobj); - struct mcidev_attribute * mcidev_attr = to_mcidev_attr(attr); - - if (mcidev_attr->store) - return mcidev_attr->store(mem_ctl_info, buffer, count); - - return -EIO; + /* the mci instance is freed here, when the sysfs object is dropped */ + edac_unregister_sysfs(mci); } +EXPORT_SYMBOL_GPL(edac_mc_free); -static struct sysfs_ops mci_ops = { - .show = mcidev_show, - .store = mcidev_store -}; -#define MCIDEV_ATTR(_name,_mode,_show,_store) \ -struct mcidev_attribute mci_attr_##_name = { \ - .attr = {.name = __stringify(_name), .mode = _mode }, \ - .show = _show, \ - .store = _store, \ -}; - -/* default Control file */ -MCIDEV_ATTR(reset_counters,S_IWUSR,NULL,mci_reset_counters_store); - -/* default Attribute files */ -MCIDEV_ATTR(mc_name,S_IRUGO,mci_ctl_name_show,NULL); -MCIDEV_ATTR(size_mb,S_IRUGO,mci_size_mb_show,NULL); -MCIDEV_ATTR(seconds_since_reset,S_IRUGO,mci_seconds_show,NULL); -MCIDEV_ATTR(ue_noinfo_count,S_IRUGO,mci_ue_noinfo_show,NULL); -MCIDEV_ATTR(ce_noinfo_count,S_IRUGO,mci_ce_noinfo_show,NULL); -MCIDEV_ATTR(ue_count,S_IRUGO,mci_ue_count_show,NULL); -MCIDEV_ATTR(ce_count,S_IRUGO,mci_ce_count_show,NULL); - -static struct mcidev_attribute *mci_attr[] = { - &mci_attr_reset_counters, - &mci_attr_mc_name, - &mci_attr_size_mb, - &mci_attr_seconds_since_reset, - &mci_attr_ue_noinfo_count, - &mci_attr_ce_noinfo_count, - &mci_attr_ue_count, - &mci_attr_ce_count, - NULL -}; - -/* - * Release of a MC controlling instance +/** + * find_mci_by_dev + * + * scan list of controllers looking for the one that manages + * the 'dev' device + * @dev: pointer to a struct device related with the MCI */ -static void edac_mci_instance_release(struct kobject *kobj) +struct mem_ctl_info *find_mci_by_dev(struct device *dev) { struct mem_ctl_info *mci; + struct list_head *item; - mci = to_mci(kobj); - debugf0("%s() idx=%d\n", __func__, mci->mc_idx); - complete(&mci->kobj_complete); -} + edac_dbg(3, "\n"); -static struct kobj_type ktype_mci = { - .release = edac_mci_instance_release, - .sysfs_ops = &mci_ops, - .default_attrs = (struct attribute **) mci_attr, -}; + list_for_each(item, &mc_devices) { + mci = list_entry(item, struct mem_ctl_info, link); + if (mci->pdev == dev) + return mci; + } -#define EDAC_DEVICE_SYMLINK "device" + return NULL; +} +EXPORT_SYMBOL_GPL(find_mci_by_dev); /* - * Create a new Memory Controller kobject instance, - * mc<id> under the 'mc' directory - * - * Return: - * 0 Success - * !0 Failure + * handler for EDAC to check if NMI type handler has asserted interrupt */ -static int edac_create_sysfs_mci_device(struct mem_ctl_info *mci) +static int edac_mc_assert_error_check_and_clear(void) { - int i; - int err; - struct csrow_info *csrow; - struct kobject *edac_mci_kobj=&mci->edac_mci_kobj; - - debugf0("%s() idx=%d\n", __func__, mci->mc_idx); - memset(edac_mci_kobj, 0, sizeof(*edac_mci_kobj)); - - /* set the name of the mc<id> object */ - err = kobject_set_name(edac_mci_kobj,"mc%d",mci->mc_idx); - if (err) - return err; - - /* link to our parent the '..../edac/mc' object */ - edac_mci_kobj->parent = &edac_memctrl_kobj; - edac_mci_kobj->ktype = &ktype_mci; - - /* register the mc<id> kobject */ - err = kobject_register(edac_mci_kobj); - if (err) - return err; - - /* create a symlink for the device */ - err = sysfs_create_link(edac_mci_kobj, &mci->dev->kobj, - EDAC_DEVICE_SYMLINK); - if (err) - goto fail0; + int old_state; - /* Make directories for each CSROW object - * under the mc<id> kobject - */ - for (i = 0; i < mci->nr_csrows; i++) { - csrow = &mci->csrows[i]; + if (edac_op_state == EDAC_OPSTATE_POLL) + return 1; - /* Only expose populated CSROWs */ - if (csrow->nr_pages > 0) { - err = edac_create_csrow_object(edac_mci_kobj,csrow,i); - if (err) - goto fail1; - } - } - - return 0; - - /* CSROW error: backout what has already been registered, */ -fail1: - for ( i--; i >= 0; i--) { - if (csrow->nr_pages > 0) { - init_completion(&csrow->kobj_complete); - kobject_unregister(&mci->csrows[i].kobj); - wait_for_completion(&csrow->kobj_complete); - } - } + old_state = edac_err_assert; + edac_err_assert = 0; -fail0: - init_completion(&mci->kobj_complete); - kobject_unregister(edac_mci_kobj); - wait_for_completion(&mci->kobj_complete); - return err; + return old_state; } /* - * remove a Memory Controller instance + * edac_mc_workq_function + * performs the operation scheduled by a workq request */ -static void edac_remove_sysfs_mci_device(struct mem_ctl_info *mci) +static void edac_mc_workq_function(struct work_struct *work_req) { - int i; + struct delayed_work *d_work = to_delayed_work(work_req); + struct mem_ctl_info *mci = to_edac_mem_ctl_work(d_work); - debugf0("%s()\n", __func__); + mutex_lock(&mem_ctls_mutex); - /* remove all csrow kobjects */ - for (i = 0; i < mci->nr_csrows; i++) { - if (mci->csrows[i].nr_pages > 0) { - init_completion(&mci->csrows[i].kobj_complete); - kobject_unregister(&mci->csrows[i].kobj); - wait_for_completion(&mci->csrows[i].kobj_complete); - } + /* if this control struct has movd to offline state, we are done */ + if (mci->op_state == OP_OFFLINE) { + mutex_unlock(&mem_ctls_mutex); + return; } - sysfs_remove_link(&mci->edac_mci_kobj, EDAC_DEVICE_SYMLINK); - init_completion(&mci->kobj_complete); - kobject_unregister(&mci->edac_mci_kobj); - wait_for_completion(&mci->kobj_complete); -} - -/* END OF sysfs data and methods */ - -#ifdef CONFIG_EDAC_DEBUG - -void edac_mc_dump_channel(struct channel_info *chan) -{ - debugf4("\tchannel = %p\n", chan); - debugf4("\tchannel->chan_idx = %d\n", chan->chan_idx); - debugf4("\tchannel->ce_count = %d\n", chan->ce_count); - debugf4("\tchannel->label = '%s'\n", chan->label); - debugf4("\tchannel->csrow = %p\n\n", chan->csrow); -} -EXPORT_SYMBOL_GPL(edac_mc_dump_channel); + /* Only poll controllers that are running polled and have a check */ + if (edac_mc_assert_error_check_and_clear() && (mci->edac_check != NULL)) + mci->edac_check(mci); -void edac_mc_dump_csrow(struct csrow_info *csrow) -{ - debugf4("\tcsrow = %p\n", csrow); - debugf4("\tcsrow->csrow_idx = %d\n", csrow->csrow_idx); - debugf4("\tcsrow->first_page = 0x%lx\n", - csrow->first_page); - debugf4("\tcsrow->last_page = 0x%lx\n", csrow->last_page); - debugf4("\tcsrow->page_mask = 0x%lx\n", csrow->page_mask); - debugf4("\tcsrow->nr_pages = 0x%x\n", csrow->nr_pages); - debugf4("\tcsrow->nr_channels = %d\n", - csrow->nr_channels); - debugf4("\tcsrow->channels = %p\n", csrow->channels); - debugf4("\tcsrow->mci = %p\n\n", csrow->mci); -} -EXPORT_SYMBOL_GPL(edac_mc_dump_csrow); + mutex_unlock(&mem_ctls_mutex); -void edac_mc_dump_mci(struct mem_ctl_info *mci) -{ - debugf3("\tmci = %p\n", mci); - debugf3("\tmci->mtype_cap = %lx\n", mci->mtype_cap); - debugf3("\tmci->edac_ctl_cap = %lx\n", mci->edac_ctl_cap); - debugf3("\tmci->edac_cap = %lx\n", mci->edac_cap); - debugf4("\tmci->edac_check = %p\n", mci->edac_check); - debugf3("\tmci->nr_csrows = %d, csrows = %p\n", - mci->nr_csrows, mci->csrows); - debugf3("\tdev = %p\n", mci->dev); - debugf3("\tmod_name:ctl_name = %s:%s\n", - mci->mod_name, mci->ctl_name); - debugf3("\tpvt_info = %p\n\n", mci->pvt_info); + /* Reschedule */ + queue_delayed_work(edac_workqueue, &mci->work, + msecs_to_jiffies(edac_mc_get_poll_msec())); } -EXPORT_SYMBOL_GPL(edac_mc_dump_mci); - -#endif /* CONFIG_EDAC_DEBUG */ -/* 'ptr' points to a possibly unaligned item X such that sizeof(X) is 'size'. - * Adjust 'ptr' so that its alignment is at least as stringent as what the - * compiler would provide for X and return the aligned result. +/* + * edac_mc_workq_setup + * initialize a workq item for this mci + * passing in the new delay period in msec * - * If 'size' is a constant, the compiler will optimize this whole function - * down to either a no-op or the addition of a constant to the value of 'ptr'. + * locking model: + * + * called with the mem_ctls_mutex held */ -static inline char * align_ptr(void *ptr, unsigned size) +static void edac_mc_workq_setup(struct mem_ctl_info *mci, unsigned msec, + bool init) { - unsigned align, r; + edac_dbg(0, "\n"); - /* Here we assume that the alignment of a "long long" is the most - * stringent alignment that the compiler will ever provide by default. - * As far as I know, this is a reasonable assumption. - */ - if (size > sizeof(long)) - align = sizeof(long long); - else if (size > sizeof(int)) - align = sizeof(long); - else if (size > sizeof(short)) - align = sizeof(int); - else if (size > sizeof(char)) - align = sizeof(short); - else - return (char *) ptr; - - r = size % align; + /* if this instance is not in the POLL state, then simply return */ + if (mci->op_state != OP_RUNNING_POLL) + return; - if (r == 0) - return (char *) ptr; + if (init) + INIT_DELAYED_WORK(&mci->work, edac_mc_workq_function); - return (char *) (((unsigned long) ptr) + align - r); + mod_delayed_work(edac_workqueue, &mci->work, msecs_to_jiffies(msec)); } -/** - * edac_mc_alloc: Allocate a struct mem_ctl_info structure - * @size_pvt: size of private storage needed - * @nr_csrows: Number of CWROWS needed for this MC - * @nr_chans: Number of channels for the MC - * - * Everything is kmalloc'ed as one big chunk - more efficient. - * Only can be used if all structures have the same lifetime - otherwise - * you have to allocate and initialize your own structures. +/* + * edac_mc_workq_teardown + * stop the workq processing on this mci * - * Use edac_mc_free() to free mc structures allocated by this function. + * locking model: * - * Returns: - * NULL allocation failed - * struct mem_ctl_info pointer + * called WITHOUT lock held */ -struct mem_ctl_info *edac_mc_alloc(unsigned sz_pvt, unsigned nr_csrows, - unsigned nr_chans) +static void edac_mc_workq_teardown(struct mem_ctl_info *mci) { - struct mem_ctl_info *mci; - struct csrow_info *csi, *csrow; - struct channel_info *chi, *chp, *chan; - void *pvt; - unsigned size; - int row, chn; + int status; - /* Figure out the offsets of the various items from the start of an mc - * structure. We want the alignment of each item to be at least as - * stringent as what the compiler would provide if we could simply - * hardcode everything into a single struct. - */ - mci = (struct mem_ctl_info *) 0; - csi = (struct csrow_info *)align_ptr(&mci[1], sizeof(*csi)); - chi = (struct channel_info *) - align_ptr(&csi[nr_csrows], sizeof(*chi)); - pvt = align_ptr(&chi[nr_chans * nr_csrows], sz_pvt); - size = ((unsigned long) pvt) + sz_pvt; - - if ((mci = kmalloc(size, GFP_KERNEL)) == NULL) - return NULL; + if (mci->op_state != OP_RUNNING_POLL) + return; - /* Adjust pointers so they point within the memory we just allocated - * rather than an imaginary chunk of memory located at address 0. - */ - csi = (struct csrow_info *) (((char *) mci) + ((unsigned long) csi)); - chi = (struct channel_info *) (((char *) mci) + ((unsigned long) chi)); - pvt = sz_pvt ? (((char *) mci) + ((unsigned long) pvt)) : NULL; + status = cancel_delayed_work(&mci->work); + if (status == 0) { + edac_dbg(0, "not canceled, flush the queue\n"); - memset(mci, 0, size); /* clear all fields */ - mci->csrows = csi; - mci->pvt_info = pvt; - mci->nr_csrows = nr_csrows; - - for (row = 0; row < nr_csrows; row++) { - csrow = &csi[row]; - csrow->csrow_idx = row; - csrow->mci = mci; - csrow->nr_channels = nr_chans; - chp = &chi[row * nr_chans]; - csrow->channels = chp; - - for (chn = 0; chn < nr_chans; chn++) { - chan = &chp[chn]; - chan->chan_idx = chn; - chan->csrow = csrow; - } + /* workq instance might be running, wait for it */ + flush_workqueue(edac_workqueue); } - - return mci; } -EXPORT_SYMBOL_GPL(edac_mc_alloc); -/** - * edac_mc_free: Free a previously allocated 'mci' structure - * @mci: pointer to a struct mem_ctl_info structure +/* + * edac_mc_reset_delay_period(unsigned long value) + * + * user space has updated our poll period value, need to + * reset our workq delays */ -void edac_mc_free(struct mem_ctl_info *mci) -{ - kfree(mci); -} -EXPORT_SYMBOL_GPL(edac_mc_free); - -static struct mem_ctl_info *find_mci_by_dev(struct device *dev) +void edac_mc_reset_delay_period(unsigned long value) { struct mem_ctl_info *mci; struct list_head *item; - debugf3("%s()\n", __func__); + mutex_lock(&mem_ctls_mutex); list_for_each(item, &mc_devices) { mci = list_entry(item, struct mem_ctl_info, link); - if (mci->dev == dev) - return mci; + edac_mc_workq_setup(mci, value, false); } - return NULL; + mutex_unlock(&mem_ctls_mutex); } + + /* Return 0 on success, 1 on failure. * Before calling this function, caller must * assign a unique value to mci->mc_idx. + * + * locking model: + * + * called with the mem_ctls_mutex lock held */ -static int add_mc_to_global_list (struct mem_ctl_info *mci) +static int add_mc_to_global_list(struct mem_ctl_info *mci) { struct list_head *item, *insert_before; struct mem_ctl_info *p; insert_before = &mc_devices; - if (unlikely((p = find_mci_by_dev(mci->dev)) != NULL)) + p = find_mci_by_dev(mci->pdev); + if (unlikely(p != NULL)) goto fail0; list_for_each(item, &mc_devices) { @@ -1359,43 +654,68 @@ static int add_mc_to_global_list (struct mem_ctl_info *mci) } list_add_tail_rcu(&mci->link, insert_before); + atomic_inc(&edac_handlers); return 0; fail0: edac_printk(KERN_WARNING, EDAC_MC, - "%s (%s) %s %s already assigned %d\n", p->dev->bus_id, - dev_name(p->dev), p->mod_name, p->ctl_name, p->mc_idx); + "%s (%s) %s %s already assigned %d\n", dev_name(p->pdev), + edac_dev_name(mci), p->mod_name, p->ctl_name, p->mc_idx); return 1; fail1: edac_printk(KERN_WARNING, EDAC_MC, - "bug in low-level driver: attempt to assign\n" - " duplicate mc_idx %d in %s()\n", p->mc_idx, __func__); + "bug in low-level driver: attempt to assign\n" + " duplicate mc_idx %d in %s()\n", p->mc_idx, __func__); return 1; } -static void complete_mc_list_del(struct rcu_head *head) +static int del_mc_from_global_list(struct mem_ctl_info *mci) { - struct mem_ctl_info *mci; + int handlers = atomic_dec_return(&edac_handlers); + list_del_rcu(&mci->link); - mci = container_of(head, struct mem_ctl_info, rcu); + /* these are for safe removal of devices from global list while + * NMI handlers may be traversing list + */ + synchronize_rcu(); INIT_LIST_HEAD(&mci->link); - complete(&mci->complete); + + return handlers; } -static void del_mc_from_global_list(struct mem_ctl_info *mci) +/** + * edac_mc_find: Search for a mem_ctl_info structure whose index is 'idx'. + * + * If found, return a pointer to the structure. + * Else return NULL. + * + * Caller must hold mem_ctls_mutex. + */ +struct mem_ctl_info *edac_mc_find(int idx) { - list_del_rcu(&mci->link); - init_completion(&mci->complete); - call_rcu(&mci->rcu, complete_mc_list_del); - wait_for_completion(&mci->complete); + struct list_head *item; + struct mem_ctl_info *mci; + + list_for_each(item, &mc_devices) { + mci = list_entry(item, struct mem_ctl_info, link); + + if (mci->mc_idx >= idx) { + if (mci->mc_idx == idx) + return mci; + + break; + } + } + + return NULL; } +EXPORT_SYMBOL(edac_mc_find); /** * edac_mc_add_mc: Insert the 'mci' structure into the mci global list and * create sysfs entries associated with mci structure * @mci: pointer to the mci structure to be added to the list - * @mc_idx: A unique numeric identifier to be assigned to the 'mci' structure. * * Return: * 0 Success @@ -1403,10 +723,16 @@ static void del_mc_from_global_list(struct mem_ctl_info *mci) */ /* FIXME - should a warning be printed if no error detection? correction? */ -int edac_mc_add_mc(struct mem_ctl_info *mci, int mc_idx) +int edac_mc_add_mc(struct mem_ctl_info *mci) { - debugf0("%s()\n", __func__); - mci->mc_idx = mc_idx; + int ret = -EINVAL; + edac_dbg(0, "\n"); + + if (mci->mc_idx >= EDAC_MAX_MCS) { + pr_warn_once("Too many memory controllers: %d\n", mci->mc_idx); + return -ENODEV; + } + #ifdef CONFIG_EDAC_DEBUG if (edac_debug_level >= 3) edac_mc_dump_mci(mci); @@ -1415,16 +741,30 @@ int edac_mc_add_mc(struct mem_ctl_info *mci, int mc_idx) int i; for (i = 0; i < mci->nr_csrows; i++) { + struct csrow_info *csrow = mci->csrows[i]; + u32 nr_pages = 0; int j; - edac_mc_dump_csrow(&mci->csrows[i]); - for (j = 0; j < mci->csrows[i].nr_channels; j++) - edac_mc_dump_channel( - &mci->csrows[i].channels[j]); + for (j = 0; j < csrow->nr_channels; j++) + nr_pages += csrow->channels[j]->dimm->nr_pages; + if (!nr_pages) + continue; + edac_mc_dump_csrow(csrow); + for (j = 0; j < csrow->nr_channels; j++) + if (csrow->channels[j]->dimm->nr_pages) + edac_mc_dump_channel(csrow->channels[j]); } + for (i = 0; i < mci->tot_dimms; i++) + if (mci->dimms[i]->nr_pages) + edac_mc_dump_dimm(mci->dimms[i], i); } #endif - down(&mem_ctls_mutex); + mutex_lock(&mem_ctls_mutex); + + if (edac_mc_owner && edac_mc_owner != mci->mod_name) { + ret = -EPERM; + goto fail0; + } if (add_mc_to_global_list(mci)) goto fail0; @@ -1432,25 +772,41 @@ int edac_mc_add_mc(struct mem_ctl_info *mci, int mc_idx) /* set load time so that error rate can be tracked */ mci->start_time = jiffies; - if (edac_create_sysfs_mci_device(mci)) { - edac_mc_printk(mci, KERN_WARNING, + mci->bus = &mc_bus[mci->mc_idx]; + + if (edac_create_sysfs_mci_device(mci)) { + edac_mc_printk(mci, KERN_WARNING, "failed to create sysfs device\n"); - goto fail1; - } + goto fail1; + } + + /* If there IS a check routine, then we are running POLLED */ + if (mci->edac_check != NULL) { + /* This instance is NOW RUNNING */ + mci->op_state = OP_RUNNING_POLL; + + edac_mc_workq_setup(mci, edac_mc_get_poll_msec(), true); + } else { + mci->op_state = OP_RUNNING_INTERRUPT; + } /* Report action taken */ - edac_mc_printk(mci, KERN_INFO, "Giving out device to %s %s: DEV %s\n", - mci->mod_name, mci->ctl_name, dev_name(mci->dev)); + edac_mc_printk(mci, KERN_INFO, + "Giving out device to module %s controller %s: DEV %s (%s)\n", + mci->mod_name, mci->ctl_name, mci->dev_name, + edac_op_state_to_string(mci->op_state)); + + edac_mc_owner = mci->mod_name; - up(&mem_ctls_mutex); + mutex_unlock(&mem_ctls_mutex); return 0; fail1: del_mc_from_global_list(mci); fail0: - up(&mem_ctls_mutex); - return 1; + mutex_unlock(&mem_ctls_mutex); + return ret; } EXPORT_SYMBOL_GPL(edac_mc_add_mc); @@ -1461,38 +817,53 @@ EXPORT_SYMBOL_GPL(edac_mc_add_mc); * * Return pointer to removed mci structure, or NULL if device not found. */ -struct mem_ctl_info * edac_mc_del_mc(struct device *dev) +struct mem_ctl_info *edac_mc_del_mc(struct device *dev) { struct mem_ctl_info *mci; - debugf0("MC: %s()\n", __func__); - down(&mem_ctls_mutex); + edac_dbg(0, "\n"); + + mutex_lock(&mem_ctls_mutex); - if ((mci = find_mci_by_dev(dev)) == NULL) { - up(&mem_ctls_mutex); + /* find the requested mci struct in the global list */ + mci = find_mci_by_dev(dev); + if (mci == NULL) { + mutex_unlock(&mem_ctls_mutex); return NULL; } + if (!del_mc_from_global_list(mci)) + edac_mc_owner = NULL; + mutex_unlock(&mem_ctls_mutex); + + /* flush workq processes */ + edac_mc_workq_teardown(mci); + + /* marking MCI offline */ + mci->op_state = OP_OFFLINE; + + /* remove from sysfs */ edac_remove_sysfs_mci_device(mci); - del_mc_from_global_list(mci); - up(&mem_ctls_mutex); + edac_printk(KERN_INFO, EDAC_MC, "Removed device %d for %s %s: DEV %s\n", mci->mc_idx, - mci->mod_name, mci->ctl_name, dev_name(mci->dev)); + mci->mod_name, mci->ctl_name, edac_dev_name(mci)); + return mci; } EXPORT_SYMBOL_GPL(edac_mc_del_mc); -void edac_mc_scrub_block(unsigned long page, unsigned long offset, u32 size) +static void edac_mc_scrub_block(unsigned long page, unsigned long offset, + u32 size) { struct page *pg; void *virt_addr; unsigned long flags = 0; - debugf3("%s()\n", __func__); + edac_dbg(3, "\n"); /* ECC error page was not in our memory. Ignore it. */ - if(!pfn_valid(page)) + if (!pfn_valid(page)) return; /* Find the actual page structure then map it and fix */ @@ -1501,38 +872,41 @@ void edac_mc_scrub_block(unsigned long page, unsigned long offset, u32 size) if (PageHighMem(pg)) local_irq_save(flags); - virt_addr = kmap_atomic(pg, KM_BOUNCE_READ); + virt_addr = kmap_atomic(pg); /* Perform architecture specific atomic scrub operation */ atomic_scrub(virt_addr + offset, size); /* Unmap and complete */ - kunmap_atomic(virt_addr, KM_BOUNCE_READ); + kunmap_atomic(virt_addr); if (PageHighMem(pg)) local_irq_restore(flags); } -EXPORT_SYMBOL_GPL(edac_mc_scrub_block); /* FIXME - should return -1 */ int edac_mc_find_csrow_by_page(struct mem_ctl_info *mci, unsigned long page) { - struct csrow_info *csrows = mci->csrows; - int row, i; + struct csrow_info **csrows = mci->csrows; + int row, i, j, n; - debugf1("MC%d: %s(): 0x%lx\n", mci->mc_idx, __func__, page); + edac_dbg(1, "MC%d: 0x%lx\n", mci->mc_idx, page); row = -1; for (i = 0; i < mci->nr_csrows; i++) { - struct csrow_info *csrow = &csrows[i]; - - if (csrow->nr_pages == 0) + struct csrow_info *csrow = csrows[i]; + n = 0; + for (j = 0; j < csrow->nr_channels; j++) { + struct dimm_info *dimm = csrow->channels[j]->dimm; + n += dimm->nr_pages; + } + if (n == 0) continue; - debugf3("MC%d: %s(): first(0x%lx) page(0x%lx) last(0x%lx) " - "mask(0x%lx)\n", mci->mc_idx, __func__, - csrow->first_page, page, csrow->last_page, - csrow->page_mask); + edac_dbg(3, "MC%d: first(0x%lx) page(0x%lx) last(0x%lx) mask(0x%lx)\n", + mci->mc_idx, + csrow->first_page, page, csrow->last_page, + csrow->page_mask); if ((page >= csrow->first_page) && (page <= csrow->last_page) && @@ -1546,284 +920,387 @@ int edac_mc_find_csrow_by_page(struct mem_ctl_info *mci, unsigned long page) if (row == -1) edac_mc_printk(mci, KERN_ERR, "could not look up page error address %lx\n", - (unsigned long) page); + (unsigned long)page); return row; } EXPORT_SYMBOL_GPL(edac_mc_find_csrow_by_page); -/* FIXME - setable log (warning/emerg) levels */ -/* FIXME - integrate with evlog: http://evlog.sourceforge.net/ */ -void edac_mc_handle_ce(struct mem_ctl_info *mci, - unsigned long page_frame_number, unsigned long offset_in_page, - unsigned long syndrome, int row, int channel, const char *msg) -{ - unsigned long remapped_page; +const char *edac_layer_name[] = { + [EDAC_MC_LAYER_BRANCH] = "branch", + [EDAC_MC_LAYER_CHANNEL] = "channel", + [EDAC_MC_LAYER_SLOT] = "slot", + [EDAC_MC_LAYER_CHIP_SELECT] = "csrow", + [EDAC_MC_LAYER_ALL_MEM] = "memory", +}; +EXPORT_SYMBOL_GPL(edac_layer_name); - debugf3("MC%d: %s()\n", mci->mc_idx, __func__); +static void edac_inc_ce_error(struct mem_ctl_info *mci, + bool enable_per_layer_report, + const int pos[EDAC_MAX_LAYERS], + const u16 count) +{ + int i, index = 0; - /* FIXME - maybe make panic on INTERNAL ERROR an option */ - if (row >= mci->nr_csrows || row < 0) { - /* something is wrong */ - edac_mc_printk(mci, KERN_ERR, - "INTERNAL ERROR: row out of range " - "(%d >= %d)\n", row, mci->nr_csrows); - edac_mc_handle_ce_no_info(mci, "INTERNAL ERROR"); - return; - } + mci->ce_mc += count; - if (channel >= mci->csrows[row].nr_channels || channel < 0) { - /* something is wrong */ - edac_mc_printk(mci, KERN_ERR, - "INTERNAL ERROR: channel out of range " - "(%d >= %d)\n", channel, - mci->csrows[row].nr_channels); - edac_mc_handle_ce_no_info(mci, "INTERNAL ERROR"); + if (!enable_per_layer_report) { + mci->ce_noinfo_count += count; return; } - if (log_ce) - /* FIXME - put in DIMM location */ - edac_mc_printk(mci, KERN_WARNING, - "CE page 0x%lx, offset 0x%lx, grain %d, syndrome " - "0x%lx, row %d, channel %d, label \"%s\": %s\n", - page_frame_number, offset_in_page, - mci->csrows[row].grain, syndrome, row, channel, - mci->csrows[row].channels[channel].label, msg); - - mci->ce_count++; - mci->csrows[row].ce_count++; - mci->csrows[row].channels[channel].ce_count++; - - if (mci->scrub_mode & SCRUB_SW_SRC) { - /* - * Some MC's can remap memory so that it is still available - * at a different address when PCI devices map into memory. - * MC's that can't do this lose the memory where PCI devices - * are mapped. This mapping is MC dependant and so we call - * back into the MC driver for it to map the MC page to - * a physical (CPU) page which can then be mapped to a virtual - * page - which can then be scrubbed. - */ - remapped_page = mci->ctl_page_to_phys ? - mci->ctl_page_to_phys(mci, page_frame_number) : - page_frame_number; + for (i = 0; i < mci->n_layers; i++) { + if (pos[i] < 0) + break; + index += pos[i]; + mci->ce_per_layer[i][index] += count; - edac_mc_scrub_block(remapped_page, offset_in_page, - mci->csrows[row].grain); + if (i < mci->n_layers - 1) + index *= mci->layers[i + 1].size; } } -EXPORT_SYMBOL_GPL(edac_mc_handle_ce); - -void edac_mc_handle_ce_no_info(struct mem_ctl_info *mci, const char *msg) -{ - if (log_ce) - edac_mc_printk(mci, KERN_WARNING, - "CE - no information available: %s\n", msg); - - mci->ce_noinfo_count++; - mci->ce_count++; -} -EXPORT_SYMBOL_GPL(edac_mc_handle_ce_no_info); -void edac_mc_handle_ue(struct mem_ctl_info *mci, - unsigned long page_frame_number, unsigned long offset_in_page, - int row, const char *msg) +static void edac_inc_ue_error(struct mem_ctl_info *mci, + bool enable_per_layer_report, + const int pos[EDAC_MAX_LAYERS], + const u16 count) { - int len = EDAC_MC_LABEL_LEN * 4; - char labels[len + 1]; - char *pos = labels; - int chan; - int chars; + int i, index = 0; - debugf3("MC%d: %s()\n", mci->mc_idx, __func__); + mci->ue_mc += count; - /* FIXME - maybe make panic on INTERNAL ERROR an option */ - if (row >= mci->nr_csrows || row < 0) { - /* something is wrong */ - edac_mc_printk(mci, KERN_ERR, - "INTERNAL ERROR: row out of range " - "(%d >= %d)\n", row, mci->nr_csrows); - edac_mc_handle_ue_no_info(mci, "INTERNAL ERROR"); + if (!enable_per_layer_report) { + mci->ce_noinfo_count += count; return; } - chars = snprintf(pos, len + 1, "%s", - mci->csrows[row].channels[0].label); - len -= chars; - pos += chars; - - for (chan = 1; (chan < mci->csrows[row].nr_channels) && (len > 0); - chan++) { - chars = snprintf(pos, len + 1, ":%s", - mci->csrows[row].channels[chan].label); - len -= chars; - pos += chars; - } + for (i = 0; i < mci->n_layers; i++) { + if (pos[i] < 0) + break; + index += pos[i]; + mci->ue_per_layer[i][index] += count; - if (log_ue) - edac_mc_printk(mci, KERN_EMERG, - "UE page 0x%lx, offset 0x%lx, grain %d, row %d, " - "labels \"%s\": %s\n", page_frame_number, - offset_in_page, mci->csrows[row].grain, row, labels, - msg); - - if (panic_on_ue) - panic("EDAC MC%d: UE page 0x%lx, offset 0x%lx, grain %d, " - "row %d, labels \"%s\": %s\n", mci->mc_idx, - page_frame_number, offset_in_page, - mci->csrows[row].grain, row, labels, msg); - - mci->ue_count++; - mci->csrows[row].ue_count++; + if (i < mci->n_layers - 1) + index *= mci->layers[i + 1].size; + } } -EXPORT_SYMBOL_GPL(edac_mc_handle_ue); -void edac_mc_handle_ue_no_info(struct mem_ctl_info *mci, const char *msg) +static void edac_ce_error(struct mem_ctl_info *mci, + const u16 error_count, + const int pos[EDAC_MAX_LAYERS], + const char *msg, + const char *location, + const char *label, + const char *detail, + const char *other_detail, + const bool enable_per_layer_report, + const unsigned long page_frame_number, + const unsigned long offset_in_page, + long grain) { - if (panic_on_ue) - panic("EDAC MC%d: Uncorrected Error", mci->mc_idx); + unsigned long remapped_page; + char *msg_aux = ""; - if (log_ue) - edac_mc_printk(mci, KERN_WARNING, - "UE - no information available: %s\n", msg); - mci->ue_noinfo_count++; - mci->ue_count++; -} -EXPORT_SYMBOL_GPL(edac_mc_handle_ue_no_info); + if (*msg) + msg_aux = " "; + + if (edac_mc_get_log_ce()) { + if (other_detail && *other_detail) + edac_mc_printk(mci, KERN_WARNING, + "%d CE %s%son %s (%s %s - %s)\n", + error_count, msg, msg_aux, label, + location, detail, other_detail); + else + edac_mc_printk(mci, KERN_WARNING, + "%d CE %s%son %s (%s %s)\n", + error_count, msg, msg_aux, label, + location, detail); + } + edac_inc_ce_error(mci, enable_per_layer_report, pos, error_count); + if (mci->scrub_mode == SCRUB_SW_SRC) { + /* + * Some memory controllers (called MCs below) can remap + * memory so that it is still available at a different + * address when PCI devices map into memory. + * MC's that can't do this, lose the memory where PCI + * devices are mapped. This mapping is MC-dependent + * and so we call back into the MC driver for it to + * map the MC page to a physical (CPU) page which can + * then be mapped to a virtual page - which can then + * be scrubbed. + */ + remapped_page = mci->ctl_page_to_phys ? + mci->ctl_page_to_phys(mci, page_frame_number) : + page_frame_number; -/* - * Iterate over all MC instances and check for ECC, et al, errors - */ -static inline void check_mc_devices(void) + edac_mc_scrub_block(remapped_page, + offset_in_page, grain); + } +} + +static void edac_ue_error(struct mem_ctl_info *mci, + const u16 error_count, + const int pos[EDAC_MAX_LAYERS], + const char *msg, + const char *location, + const char *label, + const char *detail, + const char *other_detail, + const bool enable_per_layer_report) { - struct list_head *item; - struct mem_ctl_info *mci; + char *msg_aux = ""; - debugf3("%s()\n", __func__); - down(&mem_ctls_mutex); + if (*msg) + msg_aux = " "; - list_for_each(item, &mc_devices) { - mci = list_entry(item, struct mem_ctl_info, link); + if (edac_mc_get_log_ue()) { + if (other_detail && *other_detail) + edac_mc_printk(mci, KERN_WARNING, + "%d UE %s%son %s (%s %s - %s)\n", + error_count, msg, msg_aux, label, + location, detail, other_detail); + else + edac_mc_printk(mci, KERN_WARNING, + "%d UE %s%son %s (%s %s)\n", + error_count, msg, msg_aux, label, + location, detail); + } - if (mci->edac_check != NULL) - mci->edac_check(mci); + if (edac_mc_get_panic_on_ue()) { + if (other_detail && *other_detail) + panic("UE %s%son %s (%s%s - %s)\n", + msg, msg_aux, label, location, detail, other_detail); + else + panic("UE %s%son %s (%s%s)\n", + msg, msg_aux, label, location, detail); } - up(&mem_ctls_mutex); + edac_inc_ue_error(mci, enable_per_layer_report, pos, error_count); } -/* - * Check MC status every poll_msec. - * Check PCI status every poll_msec as well. +/** + * edac_raw_mc_handle_error - reports a memory event to userspace without doing + * anything to discover the error location * - * This where the work gets done for edac. + * @type: severity of the error (CE/UE/Fatal) + * @mci: a struct mem_ctl_info pointer + * @e: error description * - * SMP safe, doesn't use NMI, and auto-rate-limits. + * This raw function is used internally by edac_mc_handle_error(). It should + * only be called directly when the hardware error come directly from BIOS, + * like in the case of APEI GHES driver. */ -static void do_edac_check(void) -{ - debugf3("%s()\n", __func__); - check_mc_devices(); - do_pci_parity_check(); -} - -static int edac_kernel_thread(void *arg) -{ - while (!kthread_should_stop()) { - do_edac_check(); +void edac_raw_mc_handle_error(const enum hw_event_mc_err_type type, + struct mem_ctl_info *mci, + struct edac_raw_error_desc *e) +{ + char detail[80]; + int pos[EDAC_MAX_LAYERS] = { e->top_layer, e->mid_layer, e->low_layer }; + + /* Memory type dependent details about the error */ + if (type == HW_EVENT_ERR_CORRECTED) { + snprintf(detail, sizeof(detail), + "page:0x%lx offset:0x%lx grain:%ld syndrome:0x%lx", + e->page_frame_number, e->offset_in_page, + e->grain, e->syndrome); + edac_ce_error(mci, e->error_count, pos, e->msg, e->location, e->label, + detail, e->other_detail, e->enable_per_layer_report, + e->page_frame_number, e->offset_in_page, e->grain); + } else { + snprintf(detail, sizeof(detail), + "page:0x%lx offset:0x%lx grain:%ld", + e->page_frame_number, e->offset_in_page, e->grain); - /* goto sleep for the interval */ - schedule_timeout_interruptible((HZ * poll_msec) / 1000); - try_to_freeze(); + edac_ue_error(mci, e->error_count, pos, e->msg, e->location, e->label, + detail, e->other_detail, e->enable_per_layer_report); } - return 0; + } +EXPORT_SYMBOL_GPL(edac_raw_mc_handle_error); -/* - * edac_mc_init - * module initialization entry point +/** + * edac_mc_handle_error - reports a memory event to userspace + * + * @type: severity of the error (CE/UE/Fatal) + * @mci: a struct mem_ctl_info pointer + * @error_count: Number of errors of the same type + * @page_frame_number: mem page where the error occurred + * @offset_in_page: offset of the error inside the page + * @syndrome: ECC syndrome + * @top_layer: Memory layer[0] position + * @mid_layer: Memory layer[1] position + * @low_layer: Memory layer[2] position + * @msg: Message meaningful to the end users that + * explains the event + * @other_detail: Technical details about the event that + * may help hardware manufacturers and + * EDAC developers to analyse the event */ -static int __init edac_mc_init(void) -{ - edac_printk(KERN_INFO, EDAC_MC, EDAC_MC_VERSION "\n"); +void edac_mc_handle_error(const enum hw_event_mc_err_type type, + struct mem_ctl_info *mci, + const u16 error_count, + const unsigned long page_frame_number, + const unsigned long offset_in_page, + const unsigned long syndrome, + const int top_layer, + const int mid_layer, + const int low_layer, + const char *msg, + const char *other_detail) +{ + char *p; + int row = -1, chan = -1; + int pos[EDAC_MAX_LAYERS] = { top_layer, mid_layer, low_layer }; + int i, n_labels = 0; + u8 grain_bits; + struct edac_raw_error_desc *e = &mci->error_desc; + + edac_dbg(3, "MC%d\n", mci->mc_idx); + + /* Fills the error report buffer */ + memset(e, 0, sizeof (*e)); + e->error_count = error_count; + e->top_layer = top_layer; + e->mid_layer = mid_layer; + e->low_layer = low_layer; + e->page_frame_number = page_frame_number; + e->offset_in_page = offset_in_page; + e->syndrome = syndrome; + e->msg = msg; + e->other_detail = other_detail; /* - * Harvest and clear any boot/initialization PCI parity errors - * - * FIXME: This only clears errors logged by devices present at time of - * module initialization. We should also do an initial clear - * of each newly hotplugged device. + * Check if the event report is consistent and if the memory + * location is known. If it is known, enable_per_layer_report will be + * true, the DIMM(s) label info will be filled and the per-layer + * error counters will be incremented. */ - clear_pci_parity_errors(); + for (i = 0; i < mci->n_layers; i++) { + if (pos[i] >= (int)mci->layers[i].size) { + + edac_mc_printk(mci, KERN_ERR, + "INTERNAL ERROR: %s value is out of range (%d >= %d)\n", + edac_layer_name[mci->layers[i].type], + pos[i], mci->layers[i].size); + /* + * Instead of just returning it, let's use what's + * known about the error. The increment routines and + * the DIMM filter logic will do the right thing by + * pointing the likely damaged DIMMs. + */ + pos[i] = -1; + } + if (pos[i] >= 0) + e->enable_per_layer_report = true; + } - /* Create the MC sysfs entries */ - if (edac_sysfs_memctrl_setup()) { - edac_printk(KERN_ERR, EDAC_MC, - "Error initializing sysfs code\n"); - return -ENODEV; + /* + * Get the dimm label/grain that applies to the match criteria. + * As the error algorithm may not be able to point to just one memory + * stick, the logic here will get all possible labels that could + * pottentially be affected by the error. + * On FB-DIMM memory controllers, for uncorrected errors, it is common + * to have only the MC channel and the MC dimm (also called "branch") + * but the channel is not known, as the memory is arranged in pairs, + * where each memory belongs to a separate channel within the same + * branch. + */ + p = e->label; + *p = '\0'; + + for (i = 0; i < mci->tot_dimms; i++) { + struct dimm_info *dimm = mci->dimms[i]; + + if (top_layer >= 0 && top_layer != dimm->location[0]) + continue; + if (mid_layer >= 0 && mid_layer != dimm->location[1]) + continue; + if (low_layer >= 0 && low_layer != dimm->location[2]) + continue; + + /* get the max grain, over the error match range */ + if (dimm->grain > e->grain) + e->grain = dimm->grain; + + /* + * If the error is memory-controller wide, there's no need to + * seek for the affected DIMMs because the whole + * channel/memory controller/... may be affected. + * Also, don't show errors for empty DIMM slots. + */ + if (e->enable_per_layer_report && dimm->nr_pages) { + if (n_labels >= EDAC_MAX_LABELS) { + e->enable_per_layer_report = false; + break; + } + n_labels++; + if (p != e->label) { + strcpy(p, OTHER_LABEL); + p += strlen(OTHER_LABEL); + } + strcpy(p, dimm->label); + p += strlen(p); + *p = '\0'; + + /* + * get csrow/channel of the DIMM, in order to allow + * incrementing the compat API counters + */ + edac_dbg(4, "%s csrows map: (%d,%d)\n", + mci->csbased ? "rank" : "dimm", + dimm->csrow, dimm->cschannel); + if (row == -1) + row = dimm->csrow; + else if (row >= 0 && row != dimm->csrow) + row = -2; + + if (chan == -1) + chan = dimm->cschannel; + else if (chan >= 0 && chan != dimm->cschannel) + chan = -2; + } } - /* Create the PCI parity sysfs entries */ - if (edac_sysfs_pci_setup()) { - edac_sysfs_memctrl_teardown(); - edac_printk(KERN_ERR, EDAC_MC, - "EDAC PCI: Error initializing sysfs code\n"); - return -ENODEV; + if (!e->enable_per_layer_report) { + strcpy(e->label, "any memory"); + } else { + edac_dbg(4, "csrow/channel to increment: (%d,%d)\n", row, chan); + if (p == e->label) + strcpy(e->label, "unknown memory"); + if (type == HW_EVENT_ERR_CORRECTED) { + if (row >= 0) { + mci->csrows[row]->ce_count += error_count; + if (chan >= 0) + mci->csrows[row]->channels[chan]->ce_count += error_count; + } + } else + if (row >= 0) + mci->csrows[row]->ue_count += error_count; } - /* create our kernel thread */ - edac_thread = kthread_run(edac_kernel_thread, NULL, "kedac"); + /* Fill the RAM location data */ + p = e->location; - if (IS_ERR(edac_thread)) { - /* remove the sysfs entries */ - edac_sysfs_memctrl_teardown(); - edac_sysfs_pci_teardown(); - return PTR_ERR(edac_thread); - } + for (i = 0; i < mci->n_layers; i++) { + if (pos[i] < 0) + continue; - return 0; -} + p += sprintf(p, "%s:%d ", + edac_layer_name[mci->layers[i].type], + pos[i]); + } + if (p > e->location) + *(p - 1) = '\0'; -/* - * edac_mc_exit() - * module exit/termination functioni - */ -static void __exit edac_mc_exit(void) -{ - debugf0("%s()\n", __func__); - kthread_stop(edac_thread); + /* Report the error via the trace interface */ + grain_bits = fls_long(e->grain) + 1; + trace_mc_event(type, e->msg, e->label, e->error_count, + mci->mc_idx, e->top_layer, e->mid_layer, e->low_layer, + PAGES_TO_MiB(e->page_frame_number) | e->offset_in_page, + grain_bits, e->syndrome, e->other_detail); - /* tear down the sysfs device */ - edac_sysfs_memctrl_teardown(); - edac_sysfs_pci_teardown(); + edac_raw_mc_handle_error(type, mci, e); } - -module_init(edac_mc_init); -module_exit(edac_mc_exit); - -MODULE_LICENSE("GPL"); -MODULE_AUTHOR("Linux Networx (http://lnxi.com) Thayne Harbaugh et al\n" - "Based on work by Dan Hollis et al"); -MODULE_DESCRIPTION("Core library routines for MC reporting"); - -module_param(panic_on_ue, int, 0644); -MODULE_PARM_DESC(panic_on_ue, "Panic on uncorrected error: 0=off 1=on"); -#ifdef CONFIG_PCI -module_param(check_pci_parity, int, 0644); -MODULE_PARM_DESC(check_pci_parity, "Check for PCI bus parity errors: 0=off 1=on"); -module_param(panic_on_pci_parity, int, 0644); -MODULE_PARM_DESC(panic_on_pci_parity, "Panic on PCI Bus Parity error: 0=off 1=on"); -#endif -module_param(log_ue, int, 0644); -MODULE_PARM_DESC(log_ue, "Log uncorrectable error to console: 0=off 1=on"); -module_param(log_ce, int, 0644); -MODULE_PARM_DESC(log_ce, "Log correctable error to console: 0=off 1=on"); -module_param(poll_msec, int, 0644); -MODULE_PARM_DESC(poll_msec, "Polling period in milliseconds"); -#ifdef CONFIG_EDAC_DEBUG -module_param(edac_debug_level, int, 0644); -MODULE_PARM_DESC(edac_debug_level, "Debug level"); -#endif +EXPORT_SYMBOL_GPL(edac_mc_handle_error); |