perf, x86: Detect broken BIOSes that corrupt the PMU

Some BIOSes use PMU resources, which can cause various bugs:

 - Non-working or erratic PMU based statistics - the PMU can end up
   counting the wrong thing, resulting in misleading statistics

 - Profiling can stop working or it can profile the wrong thing

 - A non-working or erratic NMI watchdog that cannot be relied on

 - The kernel may disturb whatever thing the BIOS tries to use the
   PMU for - possibly causing hardware malfunction in extreme cases.

 - ... and other forms of potential misbehavior

Various forms of such misbehavior has been observed in practice - there are
BIOSes that just corrupt the PMU state, consequences be damned.

The PMU is a CPU resource that is handled by the kernel and the BIOS
stealing+corrupting it is not acceptable nor robust, so we detect it,
warn about it and further refuse to touch the PMU ourselves.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Jason Wessel <jason.wessel@windriver.com>
Cc: Don Zickus <dzickus@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This commit is contained in:
Peter Zijlstra 2010-12-08 15:56:23 +01:00 committed by Ingo Molnar
parent 006b20fe4c
commit 4407204c5c

View file

@ -375,15 +375,53 @@ static void release_pmc_hardware(void) {}
static bool check_hw_exists(void)
{
u64 val, val_new = 0;
int ret = 0;
int i, reg, ret = 0;
/*
* Check to see if the BIOS enabled any of the counters, if so
* complain and bail.
*/
for (i = 0; i < x86_pmu.num_counters; i++) {
reg = x86_pmu.eventsel + i;
ret = rdmsrl_safe(reg, &val);
if (ret)
goto msr_fail;
if (val & ARCH_PERFMON_EVENTSEL_ENABLE)
goto bios_fail;
}
if (x86_pmu.num_counters_fixed) {
reg = MSR_ARCH_PERFMON_FIXED_CTR_CTRL;
ret = rdmsrl_safe(reg, &val);
if (ret)
goto msr_fail;
for (i = 0; i < x86_pmu.num_counters_fixed; i++) {
if (val & (0x03 << i*4))
goto bios_fail;
}
}
/*
* Now write a value and read it back to see if it matches,
* this is needed to detect certain hardware emulators (qemu/kvm)
* that don't trap on the MSR access and always return 0s.
*/
val = 0xabcdUL;
ret |= checking_wrmsrl(x86_pmu.perfctr, val);
ret = checking_wrmsrl(x86_pmu.perfctr, val);
ret |= rdmsrl_safe(x86_pmu.perfctr, &val_new);
if (ret || val != val_new)
return false;
goto msr_fail;
return true;
bios_fail:
printk(KERN_CONT "Broken BIOS detected, using software events only.\n");
printk(KERN_ERR FW_BUG "the BIOS has corrupted hw-PMU resources (MSR %x is %Lx)\n", reg, val);
return false;
msr_fail:
printk(KERN_CONT "Broken PMU hardware detected, using software events only.\n");
return false;
}
static void reserve_ds_buffers(void);
@ -1378,10 +1416,8 @@ int __init init_hw_perf_events(void)
pmu_check_apic();
/* sanity check that the hardware exists or is emulated */
if (!check_hw_exists()) {
pr_cont("Broken PMU hardware detected, software events only.\n");
if (!check_hw_exists())
return 0;
}
pr_cont("%s PMU driver.\n", x86_pmu.name);