2008-02-16 02:52:48 +07:00
|
|
|
/* KVM paravirtual clock driver. A clocksource implementation
|
|
|
|
Copyright (C) 2008 Glauber de Oliveira Costa, Red Hat Inc.
|
|
|
|
|
|
|
|
This program is free software; you can redistribute it and/or modify
|
|
|
|
it under the terms of the GNU General Public License as published by
|
|
|
|
the Free Software Foundation; either version 2 of the License, or
|
|
|
|
(at your option) any later version.
|
|
|
|
|
|
|
|
This program is distributed in the hope that it will be useful,
|
|
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
GNU General Public License for more details.
|
|
|
|
|
|
|
|
You should have received a copy of the GNU General Public License
|
|
|
|
along with this program; if not, write to the Free Software
|
|
|
|
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/clocksource.h>
|
|
|
|
#include <linux/kvm_para.h>
|
2008-06-03 21:17:32 +07:00
|
|
|
#include <asm/pvclock.h>
|
2008-02-16 02:52:48 +07:00
|
|
|
#include <asm/msr.h>
|
|
|
|
#include <asm/apic.h>
|
|
|
|
#include <linux/percpu.h>
|
2012-03-11 02:37:26 +07:00
|
|
|
#include <linux/hardirq.h>
|
2012-11-28 08:28:48 +07:00
|
|
|
#include <linux/memblock.h>
|
2009-08-19 17:35:53 +07:00
|
|
|
|
|
|
|
#include <asm/x86_init.h>
|
2008-03-18 02:08:40 +07:00
|
|
|
#include <asm/reboot.h>
|
2008-02-16 02:52:48 +07:00
|
|
|
|
|
|
|
static int kvmclock = 1;
|
2010-05-11 23:17:44 +07:00
|
|
|
static int msr_kvm_system_time = MSR_KVM_SYSTEM_TIME;
|
|
|
|
static int msr_kvm_wall_clock = MSR_KVM_WALL_CLOCK;
|
2008-02-16 02:52:48 +07:00
|
|
|
|
|
|
|
static int parse_no_kvmclock(char *arg)
|
|
|
|
{
|
|
|
|
kvmclock = 0;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
early_param("no-kvmclock", parse_no_kvmclock);
|
|
|
|
|
|
|
|
/* The hypervisor will put information about time periodically here */
|
2012-11-28 08:28:56 +07:00
|
|
|
static struct pvclock_vsyscall_time_info *hv_clock;
|
2008-06-03 21:17:32 +07:00
|
|
|
static struct pvclock_wall_clock wall_clock;
|
2008-02-16 02:52:48 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The wallclock is the time of day when we booted. Since then, some time may
|
|
|
|
* have elapsed since the hypervisor wrote the data. So we try to account for
|
|
|
|
* that with system time
|
|
|
|
*/
|
2008-05-22 15:37:48 +07:00
|
|
|
static unsigned long kvm_get_wallclock(void)
|
2008-02-16 02:52:48 +07:00
|
|
|
{
|
2008-06-03 21:17:32 +07:00
|
|
|
struct pvclock_vcpu_time_info *vcpu_time;
|
2008-02-16 02:52:48 +07:00
|
|
|
struct timespec ts;
|
|
|
|
int low, high;
|
2012-11-28 08:28:48 +07:00
|
|
|
int cpu;
|
2008-02-16 02:52:48 +07:00
|
|
|
|
2009-08-31 14:04:31 +07:00
|
|
|
low = (int)__pa_symbol(&wall_clock);
|
|
|
|
high = ((u64)__pa_symbol(&wall_clock) >> 32);
|
2010-05-11 23:17:44 +07:00
|
|
|
|
|
|
|
native_write_msr(msr_kvm_wall_clock, low, high);
|
2008-02-16 02:52:48 +07:00
|
|
|
|
2012-11-28 08:28:48 +07:00
|
|
|
preempt_disable();
|
|
|
|
cpu = smp_processor_id();
|
|
|
|
|
2012-11-28 08:28:56 +07:00
|
|
|
vcpu_time = &hv_clock[cpu].pvti;
|
2008-06-03 21:17:32 +07:00
|
|
|
pvclock_read_wallclock(&wall_clock, vcpu_time, &ts);
|
2012-11-28 08:28:48 +07:00
|
|
|
|
|
|
|
preempt_enable();
|
2008-02-16 02:52:48 +07:00
|
|
|
|
2008-06-03 21:17:32 +07:00
|
|
|
return ts.tv_sec;
|
2008-02-16 02:52:48 +07:00
|
|
|
}
|
|
|
|
|
2008-05-22 15:37:48 +07:00
|
|
|
static int kvm_set_wallclock(unsigned long now)
|
2008-02-16 02:52:48 +07:00
|
|
|
{
|
2008-06-03 21:17:32 +07:00
|
|
|
return -1;
|
2008-02-16 02:52:48 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static cycle_t kvm_clock_read(void)
|
|
|
|
{
|
2008-06-03 21:17:32 +07:00
|
|
|
struct pvclock_vcpu_time_info *src;
|
|
|
|
cycle_t ret;
|
2012-11-28 08:28:48 +07:00
|
|
|
int cpu;
|
2008-02-16 02:52:48 +07:00
|
|
|
|
2011-11-15 19:59:07 +07:00
|
|
|
preempt_disable_notrace();
|
2012-11-28 08:28:48 +07:00
|
|
|
cpu = smp_processor_id();
|
2012-11-28 08:28:56 +07:00
|
|
|
src = &hv_clock[cpu].pvti;
|
2008-06-03 21:17:32 +07:00
|
|
|
ret = pvclock_clocksource_read(src);
|
2011-11-15 19:59:07 +07:00
|
|
|
preempt_enable_notrace();
|
2008-06-03 21:17:32 +07:00
|
|
|
return ret;
|
2008-02-16 02:52:48 +07:00
|
|
|
}
|
2008-06-03 21:17:32 +07:00
|
|
|
|
2009-04-22 02:24:00 +07:00
|
|
|
static cycle_t kvm_clock_get_cycles(struct clocksource *cs)
|
|
|
|
{
|
|
|
|
return kvm_clock_read();
|
|
|
|
}
|
|
|
|
|
2008-07-28 21:47:53 +07:00
|
|
|
/*
|
|
|
|
* If we don't do that, there is the possibility that the guest
|
|
|
|
* will calibrate under heavy load - thus, getting a lower lpj -
|
|
|
|
* and execute the delays themselves without load. This is wrong,
|
|
|
|
* because no delay loop can finish beforehand.
|
|
|
|
* Any heuristics is subject to fail, because ultimately, a large
|
|
|
|
* poll of guests can be running and trouble each other. So we preset
|
|
|
|
* lpj here
|
|
|
|
*/
|
|
|
|
static unsigned long kvm_get_tsc_khz(void)
|
|
|
|
{
|
2008-12-06 03:36:45 +07:00
|
|
|
struct pvclock_vcpu_time_info *src;
|
2012-11-28 08:28:48 +07:00
|
|
|
int cpu;
|
|
|
|
unsigned long tsc_khz;
|
|
|
|
|
|
|
|
preempt_disable();
|
|
|
|
cpu = smp_processor_id();
|
2012-11-28 08:28:56 +07:00
|
|
|
src = &hv_clock[cpu].pvti;
|
2012-11-28 08:28:48 +07:00
|
|
|
tsc_khz = pvclock_tsc_khz(src);
|
|
|
|
preempt_enable();
|
|
|
|
return tsc_khz;
|
2008-07-28 21:47:53 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static void kvm_get_preset_lpj(void)
|
|
|
|
{
|
|
|
|
unsigned long khz;
|
|
|
|
u64 lpj;
|
|
|
|
|
2008-12-06 03:36:45 +07:00
|
|
|
khz = kvm_get_tsc_khz();
|
2008-07-28 21:47:53 +07:00
|
|
|
|
|
|
|
lpj = ((u64)khz * 1000);
|
|
|
|
do_div(lpj, HZ);
|
|
|
|
preset_lpj = lpj;
|
|
|
|
}
|
|
|
|
|
2012-03-11 02:37:26 +07:00
|
|
|
bool kvm_check_and_clear_guest_paused(void)
|
|
|
|
{
|
|
|
|
bool ret = false;
|
|
|
|
struct pvclock_vcpu_time_info *src;
|
2012-11-28 08:28:48 +07:00
|
|
|
int cpu = smp_processor_id();
|
|
|
|
|
|
|
|
if (!hv_clock)
|
|
|
|
return ret;
|
2012-03-11 02:37:26 +07:00
|
|
|
|
2012-11-28 08:28:56 +07:00
|
|
|
src = &hv_clock[cpu].pvti;
|
2012-03-11 02:37:26 +07:00
|
|
|
if ((src->flags & PVCLOCK_GUEST_STOPPED) != 0) {
|
2012-11-28 08:28:48 +07:00
|
|
|
src->flags &= ~PVCLOCK_GUEST_STOPPED;
|
2012-03-11 02:37:26 +07:00
|
|
|
ret = true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2008-02-16 02:52:48 +07:00
|
|
|
static struct clocksource kvm_clock = {
|
|
|
|
.name = "kvm-clock",
|
2009-04-22 02:24:00 +07:00
|
|
|
.read = kvm_clock_get_cycles,
|
2008-02-16 02:52:48 +07:00
|
|
|
.rating = 400,
|
|
|
|
.mask = CLOCKSOURCE_MASK(64),
|
|
|
|
.flags = CLOCK_SOURCE_IS_CONTINUOUS,
|
|
|
|
};
|
|
|
|
|
2010-10-14 16:22:49 +07:00
|
|
|
int kvm_register_clock(char *txt)
|
2008-02-16 02:52:48 +07:00
|
|
|
{
|
|
|
|
int cpu = smp_processor_id();
|
2010-08-03 04:35:28 +07:00
|
|
|
int low, high, ret;
|
2013-02-23 23:05:29 +07:00
|
|
|
struct pvclock_vcpu_time_info *src;
|
|
|
|
|
|
|
|
if (!hv_clock)
|
|
|
|
return 0;
|
2010-08-03 04:35:28 +07:00
|
|
|
|
2013-02-23 23:05:29 +07:00
|
|
|
src = &hv_clock[cpu].pvti;
|
x86, kvm: Fix kvm's use of __pa() on percpu areas
In short, it is illegal to call __pa() on an address holding
a percpu variable. This replaces those __pa() calls with
slow_virt_to_phys(). All of the cases in this patch are
in boot time (or CPU hotplug time at worst) code, so the
slow pagetable walking in slow_virt_to_phys() is not expected
to have a performance impact.
The times when this actually matters are pretty obscure
(certain 32-bit NUMA systems), but it _does_ happen. It is
important to keep KVM guests working on these systems because
the real hardware is getting harder and harder to find.
This bug manifested first by me seeing a plain hang at boot
after this message:
CPU 0 irqstacks, hard=f3018000 soft=f301a000
or, sometimes, it would actually make it out to the console:
[ 0.000000] BUG: unable to handle kernel paging request at ffffffff
I eventually traced it down to the KVM async pagefault code.
This can be worked around by disabling that code either at
compile-time, or on the kernel command-line.
The kvm async pagefault code was injecting page faults in
to the guest which the guest misinterpreted because its
"reason" was not being properly sent from the host.
The guest passes a physical address of an per-cpu async page
fault structure via an MSR to the host. Since __pa() is
broken on percpu data, the physical address it sent was
bascially bogus and the host went scribbling on random data.
The guest never saw the real reason for the page fault (it
was injected by the host), assumed that the kernel had taken
a _real_ page fault, and panic()'d. The behavior varied,
though, depending on what got corrupted by the bad write.
Signed-off-by: Dave Hansen <dave@linux.vnet.ibm.com>
Link: http://lkml.kernel.org/r/20130122212435.4905663F@kernel.stglabs.ibm.com
Acked-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
2013-01-23 04:24:35 +07:00
|
|
|
low = (int)slow_virt_to_phys(src) | 1;
|
|
|
|
high = ((u64)slow_virt_to_phys(src) >> 32);
|
2010-08-03 04:35:28 +07:00
|
|
|
ret = native_write_msr_safe(msr_kvm_system_time, low, high);
|
2008-06-03 21:17:32 +07:00
|
|
|
printk(KERN_INFO "kvm-clock: cpu %d, msr %x:%x, %s\n",
|
|
|
|
cpu, high, low, txt);
|
2010-05-11 23:17:44 +07:00
|
|
|
|
2010-08-03 04:35:28 +07:00
|
|
|
return ret;
|
2008-02-16 02:52:48 +07:00
|
|
|
}
|
|
|
|
|
2012-02-13 20:07:27 +07:00
|
|
|
static void kvm_save_sched_clock_state(void)
|
|
|
|
{
|
|
|
|
}
|
|
|
|
|
|
|
|
static void kvm_restore_sched_clock_state(void)
|
|
|
|
{
|
|
|
|
kvm_register_clock("primary cpu clock, resume");
|
|
|
|
}
|
|
|
|
|
2008-04-30 22:39:05 +07:00
|
|
|
#ifdef CONFIG_X86_LOCAL_APIC
|
2008-11-23 00:37:44 +07:00
|
|
|
static void __cpuinit kvm_setup_secondary_clock(void)
|
2008-02-16 02:52:48 +07:00
|
|
|
{
|
|
|
|
/*
|
|
|
|
* Now that the first cpu already had this clocksource initialized,
|
|
|
|
* we shouldn't fail.
|
|
|
|
*/
|
2008-06-03 21:17:32 +07:00
|
|
|
WARN_ON(kvm_register_clock("secondary cpu clock"));
|
2008-02-16 02:52:48 +07:00
|
|
|
}
|
2008-04-30 22:39:05 +07:00
|
|
|
#endif
|
2008-02-16 02:52:48 +07:00
|
|
|
|
2008-03-18 02:08:40 +07:00
|
|
|
/*
|
|
|
|
* After the clock is registered, the host will keep writing to the
|
|
|
|
* registered memory location. If the guest happens to shutdown, this memory
|
|
|
|
* won't be valid. In cases like kexec, in which you install a new kernel, this
|
|
|
|
* means a random memory location will be kept being written. So before any
|
|
|
|
* kind of shutdown from our side, we unregister the clock by writting anything
|
|
|
|
* that does not have the 'enable' bit set in the msr
|
|
|
|
*/
|
|
|
|
#ifdef CONFIG_KEXEC
|
|
|
|
static void kvm_crash_shutdown(struct pt_regs *regs)
|
|
|
|
{
|
2010-05-11 23:17:44 +07:00
|
|
|
native_write_msr(msr_kvm_system_time, 0, 0);
|
2011-07-12 02:28:19 +07:00
|
|
|
kvm_disable_steal_time();
|
2008-03-18 02:08:40 +07:00
|
|
|
native_machine_crash_shutdown(regs);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
static void kvm_shutdown(void)
|
|
|
|
{
|
2010-05-11 23:17:44 +07:00
|
|
|
native_write_msr(msr_kvm_system_time, 0, 0);
|
2011-07-12 02:28:19 +07:00
|
|
|
kvm_disable_steal_time();
|
2008-03-18 02:08:40 +07:00
|
|
|
native_machine_shutdown();
|
|
|
|
}
|
|
|
|
|
2008-02-16 02:52:48 +07:00
|
|
|
void __init kvmclock_init(void)
|
|
|
|
{
|
2012-11-28 08:28:48 +07:00
|
|
|
unsigned long mem;
|
2013-02-19 08:58:14 +07:00
|
|
|
int size;
|
|
|
|
|
|
|
|
size = PAGE_ALIGN(sizeof(struct pvclock_vsyscall_time_info)*NR_CPUS);
|
2012-11-28 08:28:48 +07:00
|
|
|
|
2008-02-16 02:52:48 +07:00
|
|
|
if (!kvm_para_available())
|
|
|
|
return;
|
|
|
|
|
2010-05-11 23:17:44 +07:00
|
|
|
if (kvmclock && kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE2)) {
|
|
|
|
msr_kvm_system_time = MSR_KVM_SYSTEM_TIME_NEW;
|
|
|
|
msr_kvm_wall_clock = MSR_KVM_WALL_CLOCK_NEW;
|
|
|
|
} else if (!(kvmclock && kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE)))
|
|
|
|
return;
|
|
|
|
|
|
|
|
printk(KERN_INFO "kvm-clock: Using msrs %x and %x",
|
|
|
|
msr_kvm_system_time, msr_kvm_wall_clock);
|
|
|
|
|
2013-02-19 08:58:14 +07:00
|
|
|
mem = memblock_alloc(size, PAGE_SIZE);
|
2012-11-28 08:28:48 +07:00
|
|
|
if (!mem)
|
|
|
|
return;
|
|
|
|
hv_clock = __va(mem);
|
|
|
|
|
|
|
|
if (kvm_register_clock("boot clock")) {
|
|
|
|
hv_clock = NULL;
|
2013-02-19 08:58:14 +07:00
|
|
|
memblock_free(mem, size);
|
2010-05-11 23:17:44 +07:00
|
|
|
return;
|
2012-11-28 08:28:48 +07:00
|
|
|
}
|
2010-05-11 23:17:44 +07:00
|
|
|
pv_time_ops.sched_clock = kvm_clock_read;
|
|
|
|
x86_platform.calibrate_tsc = kvm_get_tsc_khz;
|
|
|
|
x86_platform.get_wallclock = kvm_get_wallclock;
|
|
|
|
x86_platform.set_wallclock = kvm_set_wallclock;
|
2008-04-30 22:39:05 +07:00
|
|
|
#ifdef CONFIG_X86_LOCAL_APIC
|
2012-02-07 21:52:44 +07:00
|
|
|
x86_cpuinit.early_percpu_clock_init =
|
2010-05-11 23:17:44 +07:00
|
|
|
kvm_setup_secondary_clock;
|
2008-04-30 22:39:05 +07:00
|
|
|
#endif
|
2012-02-13 20:07:27 +07:00
|
|
|
x86_platform.save_sched_clock_state = kvm_save_sched_clock_state;
|
|
|
|
x86_platform.restore_sched_clock_state = kvm_restore_sched_clock_state;
|
2010-05-11 23:17:44 +07:00
|
|
|
machine_ops.shutdown = kvm_shutdown;
|
2008-03-18 02:08:40 +07:00
|
|
|
#ifdef CONFIG_KEXEC
|
2010-05-11 23:17:44 +07:00
|
|
|
machine_ops.crash_shutdown = kvm_crash_shutdown;
|
2008-03-18 02:08:40 +07:00
|
|
|
#endif
|
2010-05-11 23:17:44 +07:00
|
|
|
kvm_get_preset_lpj();
|
2010-04-27 09:03:05 +07:00
|
|
|
clocksource_register_hz(&kvm_clock, NSEC_PER_SEC);
|
2010-05-11 23:17:44 +07:00
|
|
|
pv_info.paravirt_enabled = 1;
|
|
|
|
pv_info.name = "KVM";
|
2010-05-11 23:17:45 +07:00
|
|
|
|
|
|
|
if (kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE_STABLE_BIT))
|
|
|
|
pvclock_set_flags(PVCLOCK_TSC_STABLE_BIT);
|
2008-02-16 02:52:48 +07:00
|
|
|
}
|
2012-11-28 08:28:56 +07:00
|
|
|
|
|
|
|
int __init kvm_setup_vsyscall_timeinfo(void)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_X86_64
|
|
|
|
int cpu;
|
|
|
|
int ret;
|
|
|
|
u8 flags;
|
|
|
|
struct pvclock_vcpu_time_info *vcpu_time;
|
|
|
|
unsigned int size;
|
|
|
|
|
2013-02-23 23:05:29 +07:00
|
|
|
if (!hv_clock)
|
|
|
|
return 0;
|
|
|
|
|
2013-02-19 08:58:14 +07:00
|
|
|
size = PAGE_ALIGN(sizeof(struct pvclock_vsyscall_time_info)*NR_CPUS);
|
2012-11-28 08:28:56 +07:00
|
|
|
|
|
|
|
preempt_disable();
|
|
|
|
cpu = smp_processor_id();
|
|
|
|
|
|
|
|
vcpu_time = &hv_clock[cpu].pvti;
|
|
|
|
flags = pvclock_read_flags(vcpu_time);
|
|
|
|
|
|
|
|
if (!(flags & PVCLOCK_TSC_STABLE_BIT)) {
|
|
|
|
preempt_enable();
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if ((ret = pvclock_init_vsyscall(hv_clock, size))) {
|
|
|
|
preempt_enable();
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
preempt_enable();
|
|
|
|
|
|
|
|
kvm_clock.archdata.vclock_mode = VCLOCK_PVCLOCK;
|
|
|
|
#endif
|
|
|
|
return 0;
|
|
|
|
}
|