X-Git-Url: https://git.openpandora.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=arch%2Fx86_64%2Fkernel%2Fmce.c;h=969365c0771b242f5b64071e294bba4cd114fca9;hb=413588c7cb8113c03d0044f1d41b832ad7201c29;hp=21e70625a4953e26361c060c432f1803393952c5;hpb=b2b18660066997420b716c1881a6be8b82700d97;p=pandora-kernel.git diff --git a/arch/x86_64/kernel/mce.c b/arch/x86_64/kernel/mce.c index 21e70625a495..969365c0771b 100644 --- a/arch/x86_64/kernel/mce.c +++ b/arch/x86_64/kernel/mce.c @@ -15,6 +15,9 @@ #include #include #include +#include +#include +#include #include #include #include @@ -34,6 +37,7 @@ static unsigned long bank[NR_BANKS] = { [0 ... NR_BANKS-1] = ~0UL }; static unsigned long console_logged; static int notify_user; static int rip_msr; +static int mce_bootlog; /* * Lockless MCE logging infrastructure. @@ -53,15 +57,19 @@ void mce_log(struct mce *mce) smp_wmb(); for (;;) { entry = rcu_dereference(mcelog.next); - /* When the buffer fills up discard new entries. Assume - that the earlier errors are the more interesting. */ - if (entry >= MCE_LOG_LEN) { - set_bit(MCE_OVERFLOW, &mcelog.flags); - return; + for (;;) { + /* When the buffer fills up discard new entries. Assume + that the earlier errors are the more interesting. */ + if (entry >= MCE_LOG_LEN) { + set_bit(MCE_OVERFLOW, &mcelog.flags); + return; + } + /* Old left over entry. Skip. */ + if (mcelog.entry[entry].finished) { + entry++; + continue; + } } - /* Old left over entry. Skip. */ - if (mcelog.entry[entry].finished) - continue; smp_rmb(); next = entry + 1; if (cmpxchg(&mcelog.next, entry, next) == entry) @@ -195,10 +203,11 @@ void do_machine_check(struct pt_regs * regs, long error_code) rdmsrl(MSR_IA32_MC0_ADDR + i*4, m.addr); mce_get_rip(&m, regs); - if (error_code != -1) + if (error_code >= 0) rdtscll(m.tsc); wrmsrl(MSR_IA32_MC0_STATUS + i*4, 0); - mce_log(&m); + if (error_code != -2) + mce_log(&m); /* Did this bank cause the exception? */ /* Assume that the bank with uncorrectable errors did it, @@ -313,7 +322,7 @@ static void mce_init(void *dummy) /* Log the machine checks left over from the previous reset. This also clears all registers */ - do_machine_check(NULL, -1); + do_machine_check(NULL, mce_bootlog ? -1 : -2); set_in_cr4(X86_CR4_MCE); @@ -400,9 +409,15 @@ static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize, loff } err = 0; - for (i = 0; i < next; i++) { - if (!mcelog.entry[i].finished) - continue; + for (i = 0; i < next; i++) { + unsigned long start = jiffies; + while (!mcelog.entry[i].finished) { + if (!time_before(jiffies, start + 2)) { + memset(mcelog.entry + i,0, sizeof(struct mce)); + continue; + } + cpu_relax(); + } smp_rmb(); err |= copy_to_user(buf, mcelog.entry + i, sizeof(struct mce)); buf += sizeof(struct mce); @@ -474,11 +489,20 @@ static int __init mcheck_disable(char *str) } /* mce=off disables machine check. Note you can reenable it later - using sysfs */ + using sysfs. + mce=TOLERANCELEVEL (number, see above) + mce=bootlog Log MCEs from before booting. Disabled by default to work + around buggy BIOS that leave bogus MCEs. */ static int __init mcheck_enable(char *str) { + if (*str == '=') + str++; if (!strcmp(str, "off")) mce_dont_init = 1; + else if (!strcmp(str, "bootlog")) + mce_bootlog = 1; + else if (isdigit(str[0])) + get_option(&str, &tolerant); else printk("mce= argument %s ignored. Please use /sys", str); return 0; @@ -491,10 +515,12 @@ __setup("mce", mcheck_enable); * Sysfs support */ -/* On resume clear all MCE state. Don't want to see leftovers from the BIOS. */ +/* On resume clear all MCE state. Don't want to see leftovers from the BIOS. + Only one CPU is active at this time, the others get readded later using + CPU hotplug. */ static int mce_resume(struct sys_device *dev) { - on_each_cpu(mce_init, NULL, 1, 1); + mce_init(NULL); return 0; } @@ -514,10 +540,7 @@ static struct sysdev_class mce_sysclass = { set_kset_name("machinecheck"), }; -static struct sys_device device_mce = { - .id = 0, - .cls = &mce_sysclass, -}; +static DEFINE_PER_CPU(struct sys_device, device_mce); /* Why are there no generic functions for this? */ #define ACCESSOR(name, var, start) \ @@ -542,27 +565,83 @@ ACCESSOR(bank4ctl,bank[4],mce_restart()) ACCESSOR(tolerant,tolerant,) ACCESSOR(check_interval,check_interval,mce_restart()) -static __cpuinit int mce_init_device(void) +/* Per cpu sysdev init. All of the cpus still share the same ctl bank */ +static __cpuinit int mce_create_device(unsigned int cpu) { int err; + if (!mce_available(&cpu_data[cpu])) + return -EIO; + + per_cpu(device_mce,cpu).id = cpu; + per_cpu(device_mce,cpu).cls = &mce_sysclass; + + err = sysdev_register(&per_cpu(device_mce,cpu)); + + if (!err) { + sysdev_create_file(&per_cpu(device_mce,cpu), &attr_bank0ctl); + sysdev_create_file(&per_cpu(device_mce,cpu), &attr_bank1ctl); + sysdev_create_file(&per_cpu(device_mce,cpu), &attr_bank2ctl); + sysdev_create_file(&per_cpu(device_mce,cpu), &attr_bank3ctl); + sysdev_create_file(&per_cpu(device_mce,cpu), &attr_bank4ctl); + sysdev_create_file(&per_cpu(device_mce,cpu), &attr_tolerant); + sysdev_create_file(&per_cpu(device_mce,cpu), &attr_check_interval); + } + return err; +} + +#ifdef CONFIG_HOTPLUG_CPU +static __cpuinit void mce_remove_device(unsigned int cpu) +{ + sysdev_remove_file(&per_cpu(device_mce,cpu), &attr_bank0ctl); + sysdev_remove_file(&per_cpu(device_mce,cpu), &attr_bank1ctl); + sysdev_remove_file(&per_cpu(device_mce,cpu), &attr_bank2ctl); + sysdev_remove_file(&per_cpu(device_mce,cpu), &attr_bank3ctl); + sysdev_remove_file(&per_cpu(device_mce,cpu), &attr_bank4ctl); + sysdev_remove_file(&per_cpu(device_mce,cpu), &attr_tolerant); + sysdev_remove_file(&per_cpu(device_mce,cpu), &attr_check_interval); + sysdev_unregister(&per_cpu(device_mce,cpu)); +} +#endif + +/* Get notified when a cpu comes on/off. Be hotplug friendly. */ +static __cpuinit int +mce_cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) +{ + unsigned int cpu = (unsigned long)hcpu; + + switch (action) { + case CPU_ONLINE: + mce_create_device(cpu); + break; +#ifdef CONFIG_HOTPLUG_CPU + case CPU_DEAD: + mce_remove_device(cpu); + break; +#endif + } + return NOTIFY_OK; +} + +static struct notifier_block mce_cpu_notifier = { + .notifier_call = mce_cpu_callback, +}; + +static __init int mce_init_device(void) +{ + int err; + int i = 0; + if (!mce_available(&boot_cpu_data)) return -EIO; err = sysdev_class_register(&mce_sysclass); - if (!err) - err = sysdev_register(&device_mce); - if (!err) { - /* could create per CPU objects, but it is not worth it. */ - sysdev_create_file(&device_mce, &attr_bank0ctl); - sysdev_create_file(&device_mce, &attr_bank1ctl); - sysdev_create_file(&device_mce, &attr_bank2ctl); - sysdev_create_file(&device_mce, &attr_bank3ctl); - sysdev_create_file(&device_mce, &attr_bank4ctl); - sysdev_create_file(&device_mce, &attr_tolerant); - sysdev_create_file(&device_mce, &attr_check_interval); - } - + + for_each_online_cpu(i) { + mce_create_device(i); + } + + register_cpu_notifier(&mce_cpu_notifier); misc_register(&mce_log_device); return err; - } + device_initcall(mce_init_device);