Commit 50d0a0f9 authored by Gerd Hoffmann's avatar Gerd Hoffmann Committed by Avi Kivity

KVM: Make kvm host use the paravirt clocksource structs

This patch updates the kvm host code to use the pvclock structs.
It also makes the paravirt clock compatible with Xen.
Signed-off-by: default avatarGerd Hoffmann <kraxel@redhat.com>
Signed-off-by: default avatarAvi Kivity <avi@qumranet.com>
parent 1c7b67f7
...@@ -492,8 +492,8 @@ static int do_set_msr(struct kvm_vcpu *vcpu, unsigned index, u64 *data) ...@@ -492,8 +492,8 @@ static int do_set_msr(struct kvm_vcpu *vcpu, unsigned index, u64 *data)
static void kvm_write_wall_clock(struct kvm *kvm, gpa_t wall_clock) static void kvm_write_wall_clock(struct kvm *kvm, gpa_t wall_clock)
{ {
static int version; static int version;
struct kvm_wall_clock wc; struct pvclock_wall_clock wc;
struct timespec wc_ts; struct timespec now, sys, boot;
if (!wall_clock) if (!wall_clock)
return; return;
...@@ -502,10 +502,19 @@ static void kvm_write_wall_clock(struct kvm *kvm, gpa_t wall_clock) ...@@ -502,10 +502,19 @@ static void kvm_write_wall_clock(struct kvm *kvm, gpa_t wall_clock)
kvm_write_guest(kvm, wall_clock, &version, sizeof(version)); kvm_write_guest(kvm, wall_clock, &version, sizeof(version));
wc_ts = current_kernel_time(); /*
wc.wc_sec = wc_ts.tv_sec; * The guest calculates current wall clock time by adding
wc.wc_nsec = wc_ts.tv_nsec; * system time (updated by kvm_write_guest_time below) to the
wc.wc_version = version; * wall clock specified here. guest system time equals host
* system time for us, thus we must fill in host boot time here.
*/
now = current_kernel_time();
ktime_get_ts(&sys);
boot = ns_to_timespec(timespec_to_ns(&now) - timespec_to_ns(&sys));
wc.sec = boot.tv_sec;
wc.nsec = boot.tv_nsec;
wc.version = version;
kvm_write_guest(kvm, wall_clock, &wc, sizeof(wc)); kvm_write_guest(kvm, wall_clock, &wc, sizeof(wc));
...@@ -513,6 +522,45 @@ static void kvm_write_wall_clock(struct kvm *kvm, gpa_t wall_clock) ...@@ -513,6 +522,45 @@ static void kvm_write_wall_clock(struct kvm *kvm, gpa_t wall_clock)
kvm_write_guest(kvm, wall_clock, &version, sizeof(version)); kvm_write_guest(kvm, wall_clock, &version, sizeof(version));
} }
static uint32_t div_frac(uint32_t dividend, uint32_t divisor)
{
uint32_t quotient, remainder;
/* Don't try to replace with do_div(), this one calculates
* "(dividend << 32) / divisor" */
__asm__ ( "divl %4"
: "=a" (quotient), "=d" (remainder)
: "0" (0), "1" (dividend), "r" (divisor) );
return quotient;
}
static void kvm_set_time_scale(uint32_t tsc_khz, struct pvclock_vcpu_time_info *hv_clock)
{
uint64_t nsecs = 1000000000LL;
int32_t shift = 0;
uint64_t tps64;
uint32_t tps32;
tps64 = tsc_khz * 1000LL;
while (tps64 > nsecs*2) {
tps64 >>= 1;
shift--;
}
tps32 = (uint32_t)tps64;
while (tps32 <= (uint32_t)nsecs) {
tps32 <<= 1;
shift++;
}
hv_clock->tsc_shift = shift;
hv_clock->tsc_to_system_mul = div_frac(nsecs, tps32);
pr_debug("%s: tsc_khz %u, tsc_shift %d, tsc_mul %u\n",
__FUNCTION__, tsc_khz, hv_clock->tsc_shift,
hv_clock->tsc_to_system_mul);
}
static void kvm_write_guest_time(struct kvm_vcpu *v) static void kvm_write_guest_time(struct kvm_vcpu *v)
{ {
struct timespec ts; struct timespec ts;
...@@ -523,6 +571,11 @@ static void kvm_write_guest_time(struct kvm_vcpu *v) ...@@ -523,6 +571,11 @@ static void kvm_write_guest_time(struct kvm_vcpu *v)
if ((!vcpu->time_page)) if ((!vcpu->time_page))
return; return;
if (unlikely(vcpu->hv_clock_tsc_khz != tsc_khz)) {
kvm_set_time_scale(tsc_khz, &vcpu->hv_clock);
vcpu->hv_clock_tsc_khz = tsc_khz;
}
/* Keep irq disabled to prevent changes to the clock */ /* Keep irq disabled to prevent changes to the clock */
local_irq_save(flags); local_irq_save(flags);
kvm_get_msr(v, MSR_IA32_TIME_STAMP_COUNTER, kvm_get_msr(v, MSR_IA32_TIME_STAMP_COUNTER,
...@@ -537,14 +590,14 @@ static void kvm_write_guest_time(struct kvm_vcpu *v) ...@@ -537,14 +590,14 @@ static void kvm_write_guest_time(struct kvm_vcpu *v)
/* /*
* The interface expects us to write an even number signaling that the * The interface expects us to write an even number signaling that the
* update is finished. Since the guest won't see the intermediate * update is finished. Since the guest won't see the intermediate
* state, we just write "2" at the end * state, we just increase by 2 at the end.
*/ */
vcpu->hv_clock.version = 2; vcpu->hv_clock.version += 2;
shared_kaddr = kmap_atomic(vcpu->time_page, KM_USER0); shared_kaddr = kmap_atomic(vcpu->time_page, KM_USER0);
memcpy(shared_kaddr + vcpu->time_offset, &vcpu->hv_clock, memcpy(shared_kaddr + vcpu->time_offset, &vcpu->hv_clock,
sizeof(vcpu->hv_clock)); sizeof(vcpu->hv_clock));
kunmap_atomic(shared_kaddr, KM_USER0); kunmap_atomic(shared_kaddr, KM_USER0);
...@@ -599,10 +652,6 @@ int kvm_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data) ...@@ -599,10 +652,6 @@ int kvm_set_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 data)
/* ...but clean it before doing the actual write */ /* ...but clean it before doing the actual write */
vcpu->arch.time_offset = data & ~(PAGE_MASK | 1); vcpu->arch.time_offset = data & ~(PAGE_MASK | 1);
vcpu->arch.hv_clock.tsc_to_system_mul =
clocksource_khz2mult(tsc_khz, 22);
vcpu->arch.hv_clock.tsc_shift = 22;
down_read(&current->mm->mmap_sem); down_read(&current->mm->mmap_sem);
vcpu->arch.time_page = vcpu->arch.time_page =
gfn_to_page(vcpu->kvm, data >> PAGE_SHIFT); gfn_to_page(vcpu->kvm, data >> PAGE_SHIFT);
......
...@@ -18,6 +18,7 @@ ...@@ -18,6 +18,7 @@
#include <linux/kvm_para.h> #include <linux/kvm_para.h>
#include <linux/kvm_types.h> #include <linux/kvm_types.h>
#include <asm/pvclock-abi.h>
#include <asm/desc.h> #include <asm/desc.h>
#define KVM_MAX_VCPUS 16 #define KVM_MAX_VCPUS 16
...@@ -282,7 +283,8 @@ struct kvm_vcpu_arch { ...@@ -282,7 +283,8 @@ struct kvm_vcpu_arch {
struct x86_emulate_ctxt emulate_ctxt; struct x86_emulate_ctxt emulate_ctxt;
gpa_t time; gpa_t time;
struct kvm_vcpu_time_info hv_clock; struct pvclock_vcpu_time_info hv_clock;
unsigned int hv_clock_tsc_khz;
unsigned int time_offset; unsigned int time_offset;
struct page *time_page; struct page *time_page;
}; };
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment