[PATCH] x86-64: Make lockless machine check record passing a bit more robust.

[pandora-kernel.git] / arch / x86_64 / kernel / mce.c

diff --git a/arch/x86_64/kernel/mce.c b/arch/x86_64/kernel/mce.c
index 3b267c9..87ea8fd 100644 (file)
--- a/arch/x86_64/kernel/mce.c
+++ b/arch/x86_64/kernel/mce.c
@@ -36,6 +36,7 @@ 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.
@@ -55,15 +56,19 @@ 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)
@@ -197,10 +202,11 @@ 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,
@@ -315,7 +321,7 @@ 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);
 
@@ -402,9 +408,15 @@ 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); 
@@ -476,11 +488,17 @@ static int __init mcheck_disable(char *str)
 }
 
 /* mce=off disables machine check. Note you can reenable it later
-   using sysfs */
+   using sysfs.
+   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
                printk("mce= argument %s ignored. Please use /sys", str); 
        return 0;