* [PATCH 1/38] KVM: Create kvm-intel.ko module
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
@ 2006-11-27 12:11 ` Avi Kivity
2006-11-27 12:36 ` Ingo Molnar
2006-11-27 12:12 ` [PATCH 2/38] KVM: Make /dev/registration happen when the arch specific module is loaded Avi Kivity
` (36 subsequent siblings)
37 siblings, 1 reply; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:11 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
This patch lays the foundation for arch-independent kvm: it creates the
kvm-intel.ko module and prepares an arch function vector for the arch-specific
implementation.
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -235,11 +235,18 @@ struct kvm_stat {
u32 irq_exits;
};
+struct kvm_arch_ops {
+};
+
extern struct kvm_stat kvm_stat;
+extern struct kvm_arch_ops *kvm_arch_ops;
#define kvm_printf(kvm, fmt ...) printk(KERN_DEBUG fmt)
#define vcpu_printf(vcpu, fmt...) kvm_printf(vcpu->kvm, fmt)
+void kvm_init_arch(struct kvm_arch_ops *ops);
+void kvm_exit_arch(void);
+
void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
int kvm_mmu_init(struct kvm_vcpu *vcpu);
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -41,6 +41,7 @@
MODULE_AUTHOR("Qumranet");
MODULE_LICENSE("GPL");
+struct kvm_arch_ops *kvm_arch_ops;
struct kvm_stat kvm_stat;
static struct kvm_stats_debugfs_item {
@@ -3611,6 +3612,15 @@ static void kvm_exit_debug(void)
hpa_t bad_page_address;
+void kvm_init_arch(struct kvm_arch_ops *ops)
+{
+ kvm_arch_ops = ops;
+}
+
+void kvm_exit_arch(void)
+{
+}
+
static __init int kvm_init(void)
{
static struct page *bad_page;
@@ -3670,3 +3680,6 @@ static __exit void kvm_exit(void)
module_init(kvm_init)
module_exit(kvm_exit)
+
+EXPORT_SYMBOL_GPL(kvm_init_arch);
+EXPORT_SYMBOL_GPL(kvm_exit_arch);
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- /dev/null
+++ linux-2.6/drivers/kvm/vmx.c
@@ -0,0 +1,39 @@
+/*
+ * Kernel-based Virtual Machine driver for Linux
+ *
+ * This module enables machines with Intel VT-x extensions to run virtual
+ * machines without emulation or binary translation.
+ *
+ * Copyright (C) 2006 Qumranet, Inc.
+ *
+ * Authors:
+ * Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
+ * Yaniv Kamay <yaniv-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2. See
+ * the COPYING file in the top-level directory.
+ *
+ */
+
+#include "kvm.h"
+#include <linux/module.h>
+
+MODULE_AUTHOR("Qumranet");
+MODULE_LICENSE("GPL");
+
+static struct kvm_arch_ops vmx_arch_ops = {
+};
+
+static int __init vmx_init(void)
+{
+ kvm_init_arch(&vmx_arch_ops);
+ return 0;
+}
+
+static void __exit vmx_exit(void)
+{
+ kvm_exit_arch();
+}
+
+module_init(vmx_init)
+module_exit(vmx_exit)
Index: linux-2.6/drivers/kvm/Kconfig
===================================================================
--- linux-2.6.orig/drivers/kvm/Kconfig
+++ linux-2.6/drivers/kvm/Kconfig
@@ -6,8 +6,9 @@ config KVM
depends on X86 && EXPERIMENTAL
---help---
Support hosting fully virtualized guest machines using hardware
- virtualization extensions. You will need a fairly recent Intel
- processor equipped with VT extensions.
+ virtualization extensions. You will need a fairly recent
+ processor equipped with virtualization extensions. You will also
+ need to select one or more of the processor modules below.
This module provides access to the hardware capabilities through
a character device node named /dev/kvm.
@@ -16,3 +17,10 @@ config KVM
will be called kvm.
If unsure, say N.
+
+config KVM_INTEL
+ tristate "KVM for Intel processors support"
+ depends on KVM
+ ---help---
+ Provides support for KVM on Intel processors equipped with the VT
+ extensions.
Index: linux-2.6/drivers/kvm/Makefile
===================================================================
--- linux-2.6.orig/drivers/kvm/Makefile
+++ linux-2.6/drivers/kvm/Makefile
@@ -4,3 +4,5 @@
kvm-objs := kvm_main.o mmu.o x86_emulate.o
obj-$(CONFIG_KVM) += kvm.o
+kvm-intel-objs = vmx.o
+obj-$(CONFIG_KVM_INTEL) += kvm-intel.o
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 2/38] KVM: Make /dev/registration happen when the arch specific module is loaded
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
2006-11-27 12:11 ` [PATCH 1/38] KVM: Create kvm-intel.ko module Avi Kivity
@ 2006-11-27 12:12 ` Avi Kivity
2006-11-27 12:13 ` [PATCH 0/38] KVM: Decouple Intel VT implementation from base kvm Avi Kivity
` (35 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:12 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
This avoids exposing the driver capabilities before they are loaded.
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -244,7 +244,7 @@ extern struct kvm_arch_ops *kvm_arch_ops
#define kvm_printf(kvm, fmt ...) printk(KERN_DEBUG fmt)
#define vcpu_printf(vcpu, fmt...) kvm_printf(vcpu->kvm, fmt)
-void kvm_init_arch(struct kvm_arch_ops *ops);
+int kvm_init_arch(struct kvm_arch_ops *ops, struct module *module);
void kvm_exit_arch(void);
void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -3558,7 +3558,6 @@ static int kvm_dev_mmap(struct file *fil
}
static struct file_operations kvm_chardev_ops = {
- .owner = THIS_MODULE,
.open = kvm_dev_open,
.release = kvm_dev_release,
.unlocked_ioctl = kvm_dev_ioctl,
@@ -3612,13 +3611,26 @@ static void kvm_exit_debug(void)
hpa_t bad_page_address;
-void kvm_init_arch(struct kvm_arch_ops *ops)
+int kvm_init_arch(struct kvm_arch_ops *ops, struct module *module)
{
+ int r;
+
kvm_arch_ops = ops;
+ kvm_chardev_ops.owner = module;
+
+ r = misc_register(&kvm_dev);
+ if (r) {
+ printk (KERN_ERR "kvm: misc device register failed\n");
+ goto out_free;
+ }
+
+out_free:
+ return r;
}
void kvm_exit_arch(void)
{
+ misc_deregister(&kvm_dev);
}
static __init int kvm_init(void)
@@ -3644,13 +3656,6 @@ static __init int kvm_init(void)
on_each_cpu(kvm_enable, 0, 0, 1);
register_reboot_notifier(&kvm_reboot_notifier);
- r = misc_register(&kvm_dev);
- if (r) {
- printk (KERN_ERR "kvm: misc device register failed\n");
- goto out_free;
- }
-
-
if ((bad_page = alloc_page(GFP_KERNEL)) == NULL) {
r = -ENOMEM;
goto out_free;
@@ -3671,7 +3676,6 @@ out:
static __exit void kvm_exit(void)
{
kvm_exit_debug();
- misc_deregister(&kvm_dev);
unregister_reboot_notifier(&kvm_reboot_notifier);
on_each_cpu(kvm_disable, 0, 0, 1);
free_kvm_area();
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -26,7 +26,7 @@ static struct kvm_arch_ops vmx_arch_ops
static int __init vmx_init(void)
{
- kvm_init_arch(&vmx_arch_ops);
+ kvm_init_arch(&vmx_arch_ops, THIS_MODULE);
return 0;
}
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* Re: [PATCH 0/38] KVM: Decouple Intel VT implementation from base kvm
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
2006-11-27 12:11 ` [PATCH 1/38] KVM: Create kvm-intel.ko module Avi Kivity
2006-11-27 12:12 ` [PATCH 2/38] KVM: Make /dev/registration happen when the arch specific module is loaded Avi Kivity
@ 2006-11-27 12:13 ` Avi Kivity
2006-11-27 12:13 ` [PATCH 3/38] KV: Make hardware detection an arch operation Avi Kivity
` (34 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:13 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f; +Cc: Andrew Morton, linux-kernel
Avi Kivity wrote:
> This patchset prepares kvm for multiple architecture implementations.
> The kvm.ko module is split into kvm.ko and kvm-intel.ko, and a
> function vector is used to dispatch arch specific operations to the
> arch module.
>
It should be noted that the patchset is bisect friendly (and will even
boot a guest for you at any point in the set).
--
error compiling committee.c: too many arguments to function
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 3/38] KV: Make hardware detection an arch operation
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (2 preceding siblings ...)
2006-11-27 12:13 ` [PATCH 0/38] KVM: Decouple Intel VT implementation from base kvm Avi Kivity
@ 2006-11-27 12:13 ` Avi Kivity
2006-11-27 12:14 ` [PATCH 4/38] KVM: Make the per-cpu enable/disable functions arch operations Avi Kivity
` (33 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:13 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -236,6 +236,8 @@ struct kvm_stat {
};
struct kvm_arch_ops {
+ int (*cpu_has_kvm_support)(void); /* __init */
+ int (*disabled_by_bios)(void); /* __init */
};
extern struct kvm_stat kvm_stat;
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -527,12 +527,6 @@ static void free_vmcs(struct vmcs *vmcs)
free_pages((unsigned long)vmcs, vmcs_descriptor.order);
}
-static __init int cpu_has_kvm_support(void)
-{
- unsigned long ecx = cpuid_ecx(1);
- return test_bit(5, &ecx); /* CPUID.1:ECX.VMX[bit 5] -> VT */
-}
-
static __exit void free_kvm_area(void)
{
int cpu;
@@ -559,14 +553,6 @@ static __init int alloc_kvm_area(void)
return 0;
}
-static __init int vmx_disabled_by_bios(void)
-{
- u64 msr;
-
- rdmsrl(MSR_IA32_FEATURE_CONTROL, msr);
- return (msr & 5) == 1; /* locked but not enabled */
-}
-
static __init void kvm_enable(void *garbage)
{
int cpu = raw_smp_processor_id();
@@ -3616,6 +3602,16 @@ int kvm_init_arch(struct kvm_arch_ops *o
int r;
kvm_arch_ops = ops;
+
+ if (!kvm_arch_ops->cpu_has_kvm_support()) {
+ printk(KERN_ERR "kvm: no hardware support\n");
+ return -EOPNOTSUPP;
+ }
+ if (kvm_arch_ops->disabled_by_bios()) {
+ printk(KERN_ERR "kvm: disabled by bios\n");
+ return -EOPNOTSUPP;
+ }
+
kvm_chardev_ops.owner = module;
r = misc_register(&kvm_dev);
@@ -3638,15 +3634,6 @@ static __init int kvm_init(void)
static struct page *bad_page;
int r = 0;
- if (!cpu_has_kvm_support()) {
- printk(KERN_ERR "kvm: no hardware support\n");
- return -EOPNOTSUPP;
- }
- if (vmx_disabled_by_bios()) {
- printk(KERN_ERR "kvm: disabled by bios\n");
- return -EOPNOTSUPP;
- }
-
kvm_init_debug();
setup_vmcs_descriptor();
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -18,10 +18,28 @@
#include "kvm.h"
#include <linux/module.h>
+#define MSR_IA32_FEATURE_CONTROL 0x03a
+
MODULE_AUTHOR("Qumranet");
MODULE_LICENSE("GPL");
+static __init int cpu_has_kvm_support(void)
+{
+ unsigned long ecx = cpuid_ecx(1);
+ return test_bit(5, &ecx); /* CPUID.1:ECX.VMX[bit 5] -> VT */
+}
+
+static __init int vmx_disabled_by_bios(void)
+{
+ u64 msr;
+
+ rdmsrl(MSR_IA32_FEATURE_CONTROL, msr);
+ return (msr & 5) == 1; /* locked but not enabled */
+}
+
static struct kvm_arch_ops vmx_arch_ops = {
+ .cpu_has_kvm_support = cpu_has_kvm_support,
+ .disabled_by_bios = vmx_disabled_by_bios,
};
static int __init vmx_init(void)
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 4/38] KVM: Make the per-cpu enable/disable functions arch operations
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (3 preceding siblings ...)
2006-11-27 12:13 ` [PATCH 3/38] KV: Make hardware detection an arch operation Avi Kivity
@ 2006-11-27 12:14 ` Avi Kivity
2006-11-27 12:15 ` [PATCH 5/38] KVM: Make the hardware setup operations (non-percpu) " Avi Kivity
` (32 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:14 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -238,6 +238,8 @@ struct kvm_stat {
struct kvm_arch_ops {
int (*cpu_has_kvm_support)(void); /* __init */
int (*disabled_by_bios)(void); /* __init */
+ void (*hardware_enable)(void *dummy); /* __init */
+ void (*hardware_disable)(void *dummy);
};
extern struct kvm_stat kvm_stat;
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -135,7 +135,6 @@ static const u32 vmx_msr_index[] = {
#define CR0_RESEVED_BITS 0xffffffff1ffaffc0ULL
#define LMSW_GUEST_MASK 0x0eULL
#define CR4_RESEVED_BITS (~((1ULL << 11) - 1))
-#define CR4_VMXE 0x2000
#define CR8_RESEVED_BITS (~0x0fULL)
#define EFER_RESERVED_BITS 0xfffffffffffff2fe
@@ -300,7 +299,8 @@ static void reload_tss(void)
#endif
}
-static DEFINE_PER_CPU(struct vmcs *, vmxarea);
+DEFINE_PER_CPU(struct vmcs *, vmxarea);
+EXPORT_SYMBOL_GPL(per_cpu__vmxarea); /* temporary hack */
static DEFINE_PER_CPU(struct vmcs *, current_vmcs);
static struct vmcs_descriptor {
@@ -553,26 +553,6 @@ static __init int alloc_kvm_area(void)
return 0;
}
-static __init void kvm_enable(void *garbage)
-{
- int cpu = raw_smp_processor_id();
- u64 phys_addr = __pa(per_cpu(vmxarea, cpu));
- u64 old;
-
- rdmsrl(MSR_IA32_FEATURE_CONTROL, old);
- if ((old & 5) == 0)
- /* enable and lock */
- wrmsrl(MSR_IA32_FEATURE_CONTROL, old | 5);
- write_cr4(read_cr4() | CR4_VMXE); /* FIXME: not cpu hotplug safe */
- asm volatile (ASM_VMX_VMXON_RAX : : "a"(&phys_addr), "m"(phys_addr)
- : "memory", "cc");
-}
-
-static void kvm_disable(void *garbage)
-{
- asm volatile (ASM_VMX_VMXOFF : : : "cc");
-}
-
static int kvm_dev_open(struct inode *inode, struct file *filp)
{
struct kvm *kvm = kzalloc(sizeof(struct kvm), GFP_KERNEL);
@@ -3565,8 +3545,8 @@ static int kvm_reboot(struct notifier_bl
* Some (well, at least mine) BIOSes hang on reboot if
* in vmx root mode.
*/
- printk(KERN_INFO "kvm: exiting vmx mode\n");
- on_each_cpu(kvm_disable, 0, 0, 1);
+ printk(KERN_INFO "kvm: exiting hardware virtualization\n");
+ on_each_cpu(kvm_arch_ops->hardware_disable, 0, 0, 1);
}
return NOTIFY_OK;
}
@@ -3612,6 +3592,9 @@ int kvm_init_arch(struct kvm_arch_ops *o
return -EOPNOTSUPP;
}
+ on_each_cpu(kvm_arch_ops->hardware_enable, 0, 0, 1);
+ register_reboot_notifier(&kvm_reboot_notifier);
+
kvm_chardev_ops.owner = module;
r = misc_register(&kvm_dev);
@@ -3627,6 +3610,9 @@ out_free:
void kvm_exit_arch(void)
{
misc_deregister(&kvm_dev);
+
+ unregister_reboot_notifier(&kvm_reboot_notifier);
+ on_each_cpu(kvm_arch_ops->hardware_disable, 0, 0, 1);
}
static __init int kvm_init(void)
@@ -3640,8 +3626,6 @@ static __init int kvm_init(void)
r = alloc_kvm_area();
if (r)
goto out;
- on_each_cpu(kvm_enable, 0, 0, 1);
- register_reboot_notifier(&kvm_reboot_notifier);
if ((bad_page = alloc_page(GFP_KERNEL)) == NULL) {
r = -ENOMEM;
@@ -3663,8 +3647,6 @@ out:
static __exit void kvm_exit(void)
{
kvm_exit_debug();
- unregister_reboot_notifier(&kvm_reboot_notifier);
- on_each_cpu(kvm_disable, 0, 0, 1);
free_kvm_area();
__free_page(pfn_to_page(bad_page_address >> PAGE_SHIFT));
}
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -23,6 +23,8 @@
MODULE_AUTHOR("Qumranet");
MODULE_LICENSE("GPL");
+DECLARE_PER_CPU(struct vmcs *, vmxarea);
+
static __init int cpu_has_kvm_support(void)
{
unsigned long ecx = cpuid_ecx(1);
@@ -37,9 +39,31 @@ static __init int vmx_disabled_by_bios(v
return (msr & 5) == 1; /* locked but not enabled */
}
+static __init void hardware_enable(void *garbage)
+{
+ int cpu = raw_smp_processor_id();
+ u64 phys_addr = __pa(per_cpu(vmxarea, cpu));
+ u64 old;
+
+ rdmsrl(MSR_IA32_FEATURE_CONTROL, old);
+ if ((old & 5) == 0)
+ /* enable and lock */
+ wrmsrl(MSR_IA32_FEATURE_CONTROL, old | 5);
+ write_cr4(read_cr4() | CR4_VMXE); /* FIXME: not cpu hotplug safe */
+ asm volatile (ASM_VMX_VMXON_RAX : : "a"(&phys_addr), "m"(phys_addr)
+ : "memory", "cc");
+}
+
+static void hardware_disable(void *garbage)
+{
+ asm volatile (ASM_VMX_VMXOFF : : : "cc");
+}
+
static struct kvm_arch_ops vmx_arch_ops = {
.cpu_has_kvm_support = cpu_has_kvm_support,
.disabled_by_bios = vmx_disabled_by_bios,
+ .hardware_enable = hardware_enable,
+ .hardware_disable = hardware_disable,
};
static int __init vmx_init(void)
Index: linux-2.6/drivers/kvm/vmx.h
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.h
+++ linux-2.6/drivers/kvm/vmx.h
@@ -284,4 +284,6 @@ enum vmcs_field {
#define AR_RESERVD_MASK 0xfffe0f00
+#define CR4_VMXE 0x2000
+
#endif
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 5/38] KVM: Make the hardware setup operations (non-percpu) arch operations
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (4 preceding siblings ...)
2006-11-27 12:14 ` [PATCH 4/38] KVM: Make the per-cpu enable/disable functions arch operations Avi Kivity
@ 2006-11-27 12:15 ` Avi Kivity
2006-11-27 12:17 ` [PATCH 7/38] KVM: Make msr accessors " Avi Kivity
` (31 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:15 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -240,6 +240,8 @@ struct kvm_arch_ops {
int (*disabled_by_bios)(void); /* __init */
void (*hardware_enable)(void *dummy); /* __init */
void (*hardware_disable)(void *dummy);
+ int (*hardware_setup)(void); /* __init */
+ void (*hardware_unsetup)(void); /* __exit */
};
extern struct kvm_stat kvm_stat;
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -124,7 +124,6 @@ static const u32 vmx_msr_index[] = {
#define MSR_IA32_TIME_STAMP_COUNTER 0x010
#define MSR_IA32_FEATURE_CONTROL 0x03a
-#define MSR_IA32_VMX_BASIC_MSR 0x480
#define MSR_IA32_VMX_PINBASED_CTLS_MSR 0x481
#define MSR_IA32_VMX_PROCBASED_CTLS_MSR 0x482
#define MSR_IA32_VMX_EXIT_CTLS_MSR 0x483
@@ -303,11 +302,12 @@ DEFINE_PER_CPU(struct vmcs *, vmxarea);
EXPORT_SYMBOL_GPL(per_cpu__vmxarea); /* temporary hack */
static DEFINE_PER_CPU(struct vmcs *, current_vmcs);
-static struct vmcs_descriptor {
+struct vmcs_descriptor {
int size;
int order;
u32 revision_id;
} vmcs_descriptor;
+EXPORT_SYMBOL_GPL(vmcs_descriptor);
#ifdef __x86_64__
static unsigned long read_msr(unsigned long msr)
@@ -394,16 +394,6 @@ int kvm_write_guest(struct kvm_vcpu *vcp
return req_size - size;
}
-static __init void setup_vmcs_descriptor(void)
-{
- u32 vmx_msr_low, vmx_msr_high;
-
- rdmsr(MSR_IA32_VMX_BASIC_MSR, vmx_msr_low, vmx_msr_high);
- vmcs_descriptor.size = vmx_msr_high & 0x1fff;
- vmcs_descriptor.order = get_order(vmcs_descriptor.size);
- vmcs_descriptor.revision_id = vmx_msr_low;
-};
-
static void vmcs_clear(struct vmcs *vmcs)
{
u64 phys_addr = __pa(vmcs);
@@ -501,8 +491,7 @@ static void vcpu_put(struct kvm_vcpu *vc
mutex_unlock(&vcpu->mutex);
}
-
-static struct vmcs *alloc_vmcs_cpu(int cpu)
+struct vmcs *alloc_vmcs_cpu(int cpu)
{
int node = cpu_to_node(cpu);
struct page *pages;
@@ -516,42 +505,18 @@ static struct vmcs *alloc_vmcs_cpu(int c
vmcs->revision_id = vmcs_descriptor.revision_id; /* vmcs revision id */
return vmcs;
}
+EXPORT_SYMBOL_GPL(alloc_vmcs_cpu);
static struct vmcs *alloc_vmcs(void)
{
return alloc_vmcs_cpu(smp_processor_id());
}
-static void free_vmcs(struct vmcs *vmcs)
+void free_vmcs(struct vmcs *vmcs)
{
free_pages((unsigned long)vmcs, vmcs_descriptor.order);
}
-
-static __exit void free_kvm_area(void)
-{
- int cpu;
-
- for_each_online_cpu(cpu)
- free_vmcs(per_cpu(vmxarea, cpu));
-}
-
-static __init int alloc_kvm_area(void)
-{
- int cpu;
-
- for_each_online_cpu(cpu) {
- struct vmcs *vmcs;
-
- vmcs = alloc_vmcs_cpu(cpu);
- if (!vmcs) {
- free_kvm_area();
- return -ENOMEM;
- }
-
- per_cpu(vmxarea, cpu) = vmcs;
- }
- return 0;
-}
+EXPORT_SYMBOL_GPL(free_vmcs);
static int kvm_dev_open(struct inode *inode, struct file *filp)
{
@@ -3592,6 +3557,10 @@ int kvm_init_arch(struct kvm_arch_ops *o
return -EOPNOTSUPP;
}
+ r = kvm_arch_ops->hardware_setup();
+ if (r < 0)
+ return r;
+
on_each_cpu(kvm_arch_ops->hardware_enable, 0, 0, 1);
register_reboot_notifier(&kvm_reboot_notifier);
@@ -3603,7 +3572,12 @@ int kvm_init_arch(struct kvm_arch_ops *o
goto out_free;
}
+ return r;
+
out_free:
+ unregister_reboot_notifier(&kvm_reboot_notifier);
+ on_each_cpu(kvm_arch_ops->hardware_disable, 0, 0, 1);
+ kvm_arch_ops->hardware_unsetup();
return r;
}
@@ -3613,6 +3587,7 @@ void kvm_exit_arch(void)
unregister_reboot_notifier(&kvm_reboot_notifier);
on_each_cpu(kvm_arch_ops->hardware_disable, 0, 0, 1);
+ kvm_arch_ops->hardware_unsetup();
}
static __init int kvm_init(void)
@@ -3622,14 +3597,9 @@ static __init int kvm_init(void)
kvm_init_debug();
- setup_vmcs_descriptor();
- r = alloc_kvm_area();
- if (r)
- goto out;
-
if ((bad_page = alloc_page(GFP_KERNEL)) == NULL) {
r = -ENOMEM;
- goto out_free;
+ goto out;
}
bad_page_address = page_to_pfn(bad_page) << PAGE_SHIFT;
@@ -3637,8 +3607,6 @@ static __init int kvm_init(void)
return r;
-out_free:
- free_kvm_area();
out:
kvm_exit_debug();
return r;
@@ -3647,7 +3615,6 @@ out:
static __exit void kvm_exit(void)
{
kvm_exit_debug();
- free_kvm_area();
__free_page(pfn_to_page(bad_page_address >> PAGE_SHIFT));
}
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -25,6 +25,12 @@ MODULE_LICENSE("GPL");
DECLARE_PER_CPU(struct vmcs *, vmxarea);
+extern struct vmcs_descriptor {
+ int size;
+ int order;
+ u32 revision_id;
+} vmcs_descriptor;
+
static __init int cpu_has_kvm_support(void)
{
unsigned long ecx = cpuid_ecx(1);
@@ -59,9 +65,62 @@ static void hardware_disable(void *garba
asm volatile (ASM_VMX_VMXOFF : : : "cc");
}
+static __init void setup_vmcs_descriptor(void)
+{
+ u32 vmx_msr_low, vmx_msr_high;
+
+ rdmsr(MSR_IA32_VMX_BASIC_MSR, vmx_msr_low, vmx_msr_high);
+ vmcs_descriptor.size = vmx_msr_high & 0x1fff;
+ vmcs_descriptor.order = get_order(vmcs_descriptor.size);
+ vmcs_descriptor.revision_id = vmx_msr_low;
+};
+
+void free_vmcs(struct vmcs *vmcs);
+
+static __exit void free_kvm_area(void)
+{
+ int cpu;
+
+ for_each_online_cpu(cpu)
+ free_vmcs(per_cpu(vmxarea, cpu));
+}
+
+extern struct vmcs *alloc_vmcs_cpu(int cpu);
+
+static __init int alloc_kvm_area(void)
+{
+ int cpu;
+
+ for_each_online_cpu(cpu) {
+ struct vmcs *vmcs;
+
+ vmcs = alloc_vmcs_cpu(cpu);
+ if (!vmcs) {
+ free_kvm_area();
+ return -ENOMEM;
+ }
+
+ per_cpu(vmxarea, cpu) = vmcs;
+ }
+ return 0;
+}
+
+static __init int hardware_setup(void)
+{
+ setup_vmcs_descriptor();
+ return alloc_kvm_area();
+}
+
+static __exit void hardware_unsetup(void)
+{
+ free_kvm_area();
+}
+
static struct kvm_arch_ops vmx_arch_ops = {
.cpu_has_kvm_support = cpu_has_kvm_support,
.disabled_by_bios = vmx_disabled_by_bios,
+ .hardware_setup = hardware_setup,
+ .hardware_unsetup = hardware_unsetup,
.hardware_enable = hardware_enable,
.hardware_disable = hardware_disable,
};
Index: linux-2.6/drivers/kvm/vmx.h
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.h
+++ linux-2.6/drivers/kvm/vmx.h
@@ -286,4 +286,6 @@ enum vmcs_field {
#define CR4_VMXE 0x2000
+#define MSR_IA32_VMX_BASIC_MSR 0x480
+
#endif
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 7/38] KVM: Make msr accessors arch operations
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (5 preceding siblings ...)
2006-11-27 12:15 ` [PATCH 5/38] KVM: Make the hardware setup operations (non-percpu) " Avi Kivity
@ 2006-11-27 12:17 ` Avi Kivity
2006-11-27 12:18 ` [PATCH 8/38] KVM: Make the segment " Avi Kivity
` (30 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:17 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -245,6 +245,8 @@ struct kvm_arch_ops {
int (*set_guest_debug)(struct kvm_vcpu *vcpu,
struct kvm_debug_guest *dbg);
+ int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
+ int (*set_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
};
extern struct kvm_stat kvm_stat;
@@ -401,6 +403,8 @@ static inline struct kvm_mmu_page *page_
#define ASM_VMX_VMXOFF ".byte 0x0f, 0x01, 0xc4"
#define ASM_VMX_VMXON_RAX ".byte 0xf3, 0x0f, 0xc7, 0x30"
+#define MSR_IA32_TIME_STAMP_COUNTER 0x010
+
#ifdef __x86_64__
/*
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -122,7 +122,6 @@ static const u32 vmx_msr_index[] = {
#define TSS_REDIRECTION_SIZE (256 / 8)
#define RMODE_TSS_SIZE (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
-#define MSR_IA32_TIME_STAMP_COUNTER 0x010
#define MSR_IA32_FEATURE_CONTROL 0x03a
#define MSR_IA32_VMX_PINBASED_CTLS_MSR 0x481
#define MSR_IA32_VMX_PROCBASED_CTLS_MSR 0x482
@@ -143,7 +142,7 @@ static const u32 vmx_msr_index[] = {
#define HOST_IS_64 0
#endif
-static struct vmx_msr_entry *find_msr_entry(struct kvm_vcpu *vcpu, u32 msr)
+struct vmx_msr_entry *find_msr_entry(struct kvm_vcpu *vcpu, u32 msr)
{
int i;
@@ -152,6 +151,7 @@ static struct vmx_msr_entry *find_msr_en
return &vcpu->guest_msrs[i];
return 0;
}
+EXPORT_SYMBOL_GPL(find_msr_entry);
struct descriptor_table {
u16 limit;
@@ -657,7 +657,7 @@ static void inject_gp(struct kvm_vcpu *v
* reads and returns guest's timestamp counter "register"
* guest_tsc = host_tsc + tsc_offset -- 21.3
*/
-static u64 guest_read_tsc(void)
+u64 guest_read_tsc(void)
{
u64 host_tsc, tsc_offset;
@@ -665,18 +665,20 @@ static u64 guest_read_tsc(void)
tsc_offset = vmcs_read64(TSC_OFFSET);
return host_tsc + tsc_offset;
}
+EXPORT_SYMBOL_GPL(guest_read_tsc);
/*
* writes 'guest_tsc' into guest's timestamp counter "register"
* guest_tsc = host_tsc + tsc_offset ==> tsc_offset = guest_tsc - host_tsc
*/
-static void guest_write_tsc(u64 guest_tsc)
+void guest_write_tsc(u64 guest_tsc)
{
u64 host_tsc;
rdtscll(host_tsc);
vmcs_write64(TSC_OFFSET, guest_tsc - host_tsc);
}
+EXPORT_SYMBOL_GPL(guest_write_tsc);
static void update_exception_bitmap(struct kvm_vcpu *vcpu)
{
@@ -2242,67 +2244,7 @@ static int handle_cpuid(struct kvm_vcpu
*/
static int get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata)
{
- u64 data;
- struct vmx_msr_entry *msr;
-
- if (!pdata) {
- printk(KERN_ERR "BUG: get_msr called with NULL pdata\n");
- return -EINVAL;
- }
-
- switch (msr_index) {
-#ifdef __x86_64__
- case MSR_FS_BASE:
- data = vmcs_readl(GUEST_FS_BASE);
- break;
- case MSR_GS_BASE:
- data = vmcs_readl(GUEST_GS_BASE);
- break;
- case MSR_EFER:
- data = vcpu->shadow_efer;
- break;
-#endif
- case MSR_IA32_TIME_STAMP_COUNTER:
- data = guest_read_tsc();
- break;
- case MSR_IA32_SYSENTER_CS:
- data = vmcs_read32(GUEST_SYSENTER_CS);
- break;
- case MSR_IA32_SYSENTER_EIP:
- data = vmcs_read32(GUEST_SYSENTER_EIP);
- break;
- case MSR_IA32_SYSENTER_ESP:
- data = vmcs_read32(GUEST_SYSENTER_ESP);
- break;
- case MSR_IA32_MC0_CTL:
- case MSR_IA32_MCG_STATUS:
- case MSR_IA32_MCG_CAP:
- case MSR_IA32_MC0_MISC:
- case MSR_IA32_MC0_MISC+4:
- case MSR_IA32_MC0_MISC+8:
- case MSR_IA32_MC0_MISC+12:
- case MSR_IA32_MC0_MISC+16:
- case MSR_IA32_UCODE_REV:
- /* MTRR registers */
- case 0xfe:
- case 0x200 ... 0x2ff:
- data = 0;
- break;
- case MSR_IA32_APICBASE:
- data = vcpu->apic_base;
- break;
- default:
- msr = find_msr_entry(vcpu, msr_index);
- if (!msr) {
- printk(KERN_ERR "kvm: unhandled rdmsr: %x\n", msr_index);
- return 1;
- }
- data = msr->data;
- break;
- }
-
- *pdata = data;
- return 0;
+ return kvm_arch_ops->get_msr(vcpu, msr_index, pdata);
}
static int handle_rdmsr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
@@ -2324,7 +2266,7 @@ static int handle_rdmsr(struct kvm_vcpu
#ifdef __x86_64__
-static void set_efer(struct kvm_vcpu *vcpu, u64 efer)
+void set_efer(struct kvm_vcpu *vcpu, u64 efer)
{
struct vmx_msr_entry *msr;
@@ -2353,10 +2295,10 @@ static void set_efer(struct kvm_vcpu *vc
msr->data = efer;
skip_emulated_instruction(vcpu);
}
+EXPORT_SYMBOL_GPL(set_efer);
#endif
-
/*
* Writes msr value into into the appropriate "register".
* Returns 0 on success, non-0 otherwise.
@@ -2364,56 +2306,7 @@ static void set_efer(struct kvm_vcpu *vc
*/
static int set_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 data)
{
- struct vmx_msr_entry *msr;
- switch (msr_index) {
-#ifdef __x86_64__
- case MSR_FS_BASE:
- vmcs_writel(GUEST_FS_BASE, data);
- break;
- case MSR_GS_BASE:
- vmcs_writel(GUEST_GS_BASE, data);
- break;
-#endif
- case MSR_IA32_SYSENTER_CS:
- vmcs_write32(GUEST_SYSENTER_CS, data);
- break;
- case MSR_IA32_SYSENTER_EIP:
- vmcs_write32(GUEST_SYSENTER_EIP, data);
- break;
- case MSR_IA32_SYSENTER_ESP:
- vmcs_write32(GUEST_SYSENTER_ESP, data);
- break;
-#ifdef __x86_64
- case MSR_EFER:
- set_efer(vcpu, data);
- break;
- case MSR_IA32_MC0_STATUS:
- printk(KERN_WARNING "%s: MSR_IA32_MC0_STATUS 0x%llx, nop\n"
- , __FUNCTION__, data);
- break;
-#endif
- case MSR_IA32_TIME_STAMP_COUNTER: {
- guest_write_tsc(data);
- break;
- }
- case MSR_IA32_UCODE_REV:
- case MSR_IA32_UCODE_WRITE:
- case 0x200 ... 0x2ff: /* MTRRs */
- break;
- case MSR_IA32_APICBASE:
- vcpu->apic_base = data;
- break;
- default:
- msr = find_msr_entry(vcpu, msr_index);
- if (!msr) {
- printk(KERN_ERR "kvm: unhandled wrmsr: 0x%x\n", msr_index);
- return 1;
- }
- msr->data = data;
- break;
- }
-
- return 0;
+ return kvm_arch_ops->set_msr(vcpu, msr_index, data);
}
static int handle_wrmsr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -31,6 +31,146 @@ extern struct vmcs_descriptor {
u32 revision_id;
} vmcs_descriptor;
+u64 guest_read_tsc(void);
+void guest_write_tsc(u64 guest_tsc);
+struct vmx_msr_entry *find_msr_entry(struct kvm_vcpu *vcpu, u32 msr);
+
+#ifdef __x86_64__
+
+void set_efer(struct kvm_vcpu *vcpu, u64 efer);
+
+#endif
+
+
+/*
+ * Reads an msr value (of 'msr_index') into 'pdata'.
+ * Returns 0 on success, non-0 otherwise.
+ * Assumes vcpu_load() was already called.
+ */
+static int vmx_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata)
+{
+ u64 data;
+ struct vmx_msr_entry *msr;
+
+ if (!pdata) {
+ printk(KERN_ERR "BUG: get_msr called with NULL pdata\n");
+ return -EINVAL;
+ }
+
+ switch (msr_index) {
+#ifdef __x86_64__
+ case MSR_FS_BASE:
+ data = vmcs_readl(GUEST_FS_BASE);
+ break;
+ case MSR_GS_BASE:
+ data = vmcs_readl(GUEST_GS_BASE);
+ break;
+ case MSR_EFER:
+ data = vcpu->shadow_efer;
+ break;
+#endif
+ case MSR_IA32_TIME_STAMP_COUNTER:
+ data = guest_read_tsc();
+ break;
+ case MSR_IA32_SYSENTER_CS:
+ data = vmcs_read32(GUEST_SYSENTER_CS);
+ break;
+ case MSR_IA32_SYSENTER_EIP:
+ data = vmcs_read32(GUEST_SYSENTER_EIP);
+ break;
+ case MSR_IA32_SYSENTER_ESP:
+ data = vmcs_read32(GUEST_SYSENTER_ESP);
+ break;
+ case MSR_IA32_MC0_CTL:
+ case MSR_IA32_MCG_STATUS:
+ case MSR_IA32_MCG_CAP:
+ case MSR_IA32_MC0_MISC:
+ case MSR_IA32_MC0_MISC+4:
+ case MSR_IA32_MC0_MISC+8:
+ case MSR_IA32_MC0_MISC+12:
+ case MSR_IA32_MC0_MISC+16:
+ case MSR_IA32_UCODE_REV:
+ /* MTRR registers */
+ case 0xfe:
+ case 0x200 ... 0x2ff:
+ data = 0;
+ break;
+ case MSR_IA32_APICBASE:
+ data = vcpu->apic_base;
+ break;
+ default:
+ msr = find_msr_entry(vcpu, msr_index);
+ if (!msr) {
+ printk(KERN_ERR "kvm: unhandled rdmsr: %x\n", msr_index);
+ return 1;
+ }
+ data = msr->data;
+ break;
+ }
+
+ *pdata = data;
+ return 0;
+}
+
+/*
+ * Writes msr value into into the appropriate "register".
+ * Returns 0 on success, non-0 otherwise.
+ * Assumes vcpu_load() was already called.
+ */
+static int vmx_set_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 data)
+{
+ struct vmx_msr_entry *msr;
+ switch (msr_index) {
+#ifdef __x86_64__
+ case MSR_FS_BASE:
+ vmcs_writel(GUEST_FS_BASE, data);
+ break;
+ case MSR_GS_BASE:
+ vmcs_writel(GUEST_GS_BASE, data);
+ break;
+#endif
+ case MSR_IA32_SYSENTER_CS:
+ vmcs_write32(GUEST_SYSENTER_CS, data);
+ break;
+ case MSR_IA32_SYSENTER_EIP:
+ vmcs_write32(GUEST_SYSENTER_EIP, data);
+ break;
+ case MSR_IA32_SYSENTER_ESP:
+ vmcs_write32(GUEST_SYSENTER_ESP, data);
+ break;
+#ifdef __x86_64
+ case MSR_EFER:
+ set_efer(vcpu, data);
+ break;
+ case MSR_IA32_MC0_STATUS:
+ printk(KERN_WARNING "%s: MSR_IA32_MC0_STATUS 0x%llx, nop\n"
+ , __FUNCTION__, data);
+ break;
+#endif
+ case MSR_IA32_TIME_STAMP_COUNTER: {
+ guest_write_tsc(data);
+ break;
+ }
+ case MSR_IA32_UCODE_REV:
+ case MSR_IA32_UCODE_WRITE:
+ case 0x200 ... 0x2ff: /* MTRRs */
+ break;
+ case MSR_IA32_APICBASE:
+ vcpu->apic_base = data;
+ break;
+ default:
+ msr = find_msr_entry(vcpu, msr_index);
+ if (!msr) {
+ printk(KERN_ERR "kvm: unhandled wrmsr: 0x%x\n", msr_index);
+ return 1;
+ }
+ msr->data = data;
+ break;
+ }
+
+ return 0;
+}
+
static int set_guest_debug(struct kvm_vcpu *vcpu, struct kvm_debug_guest *dbg)
{
unsigned long dr7 = 0x400;
@@ -169,6 +309,8 @@ static struct kvm_arch_ops vmx_arch_ops
.hardware_disable = hardware_disable,
.set_guest_debug = set_guest_debug,
+ .get_msr = vmx_get_msr,
+ .set_msr = vmx_set_msr,
};
static int __init vmx_init(void)
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 8/38] KVM: Make the segment accessors arch operations
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (6 preceding siblings ...)
2006-11-27 12:17 ` [PATCH 7/38] KVM: Make msr accessors " Avi Kivity
@ 2006-11-27 12:18 ` Avi Kivity
2006-11-27 12:19 ` [PATCH 9/38] KVM: Cache guest cr4 in vcpu structure Avi Kivity
` (29 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:18 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -247,6 +247,10 @@ struct kvm_arch_ops {
struct kvm_debug_guest *dbg);
int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
int (*set_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
+ void (*get_segment)(struct kvm_vcpu *vcpu,
+ struct kvm_segment *var, int seg);
+ void (*set_segment)(struct kvm_vcpu *vcpu,
+ struct kvm_segment *var, int seg);
};
extern struct kvm_stat kvm_stat;
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -36,6 +36,7 @@
#include <asm/desc.h>
#include "vmx.h"
+#include "kvm_vmx.h"
#include "x86_emulate.h"
MODULE_AUTHOR("Qumranet");
@@ -82,7 +83,7 @@ enum {
GUEST_##seg##_AR_BYTES, \
}
-static struct kvm_vmx_segment_field {
+struct kvm_vmx_segment_field {
unsigned selector;
unsigned base;
unsigned limit;
@@ -97,6 +98,7 @@ static struct kvm_vmx_segment_field {
VMX_SEGMENT_FIELD(TR),
VMX_SEGMENT_FIELD(LDTR),
};
+EXPORT_SYMBOL_GPL(kvm_vmx_segment_fields);
static const u32 vmx_msr_index[] = {
#ifdef __x86_64__
@@ -625,22 +627,6 @@ void vmcs_writel(unsigned long field, un
}
EXPORT_SYMBOL_GPL(vmcs_writel);
-static void vmcs_write16(unsigned long field, u16 value)
-{
- vmcs_writel(field, value);
-}
-
-static void vmcs_write64(unsigned long field, u64 value)
-{
-#ifdef __x86_64__
- vmcs_writel(field, value);
-#else
- vmcs_writel(field, value);
- asm volatile ("");
- vmcs_writel(field+1, value >> 32);
-#endif
-}
-
static void inject_gp(struct kvm_vcpu *vcpu)
{
printk(KERN_DEBUG "inject_general_protection: rip 0x%lx\n",
@@ -2821,26 +2807,10 @@ static int kvm_dev_ioctl_set_regs(struct
return 0;
}
-static void get_segment(struct kvm_segment *var, int seg)
+static void get_segment(struct kvm_vcpu *vcpu,
+ struct kvm_segment *var, int seg)
{
- struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
- u32 ar;
-
- var->base = vmcs_readl(sf->base);
- var->limit = vmcs_read32(sf->limit);
- var->selector = vmcs_read16(sf->selector);
- ar = vmcs_read32(sf->ar_bytes);
- if (ar & AR_UNUSABLE_MASK)
- ar = 0;
- var->type = ar & 15;
- var->s = (ar >> 4) & 1;
- var->dpl = (ar >> 5) & 3;
- var->present = (ar >> 7) & 1;
- var->avl = (ar >> 12) & 1;
- var->l = (ar >> 13) & 1;
- var->db = (ar >> 14) & 1;
- var->g = (ar >> 15) & 1;
- var->unusable = (ar >> 16) & 1;
+ return kvm_arch_ops->get_segment(vcpu, var, seg);
}
static int kvm_dev_ioctl_get_sregs(struct kvm *kvm, struct kvm_sregs *sregs)
@@ -2853,15 +2823,15 @@ static int kvm_dev_ioctl_get_sregs(struc
if (!vcpu)
return -ENOENT;
- get_segment(&sregs->cs, VCPU_SREG_CS);
- get_segment(&sregs->ds, VCPU_SREG_DS);
- get_segment(&sregs->es, VCPU_SREG_ES);
- get_segment(&sregs->fs, VCPU_SREG_FS);
- get_segment(&sregs->gs, VCPU_SREG_GS);
- get_segment(&sregs->ss, VCPU_SREG_SS);
+ get_segment(vcpu, &sregs->cs, VCPU_SREG_CS);
+ get_segment(vcpu, &sregs->ds, VCPU_SREG_DS);
+ get_segment(vcpu, &sregs->es, VCPU_SREG_ES);
+ get_segment(vcpu, &sregs->fs, VCPU_SREG_FS);
+ get_segment(vcpu, &sregs->gs, VCPU_SREG_GS);
+ get_segment(vcpu, &sregs->ss, VCPU_SREG_SS);
- get_segment(&sregs->tr, VCPU_SREG_TR);
- get_segment(&sregs->ldt, VCPU_SREG_LDTR);
+ get_segment(vcpu, &sregs->tr, VCPU_SREG_TR);
+ get_segment(vcpu, &sregs->ldt, VCPU_SREG_LDTR);
#define get_dtable(var, table) \
sregs->var.limit = vmcs_read32(GUEST_##table##_LIMIT), \
@@ -2887,27 +2857,10 @@ static int kvm_dev_ioctl_get_sregs(struc
return 0;
}
-static void set_segment(struct kvm_segment *var, int seg)
+static void set_segment(struct kvm_vcpu *vcpu,
+ struct kvm_segment *var, int seg)
{
- struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
- u32 ar;
-
- vmcs_writel(sf->base, var->base);
- vmcs_write32(sf->limit, var->limit);
- vmcs_write16(sf->selector, var->selector);
- if (var->unusable)
- ar = 1 << 16;
- else {
- ar = var->type & 15;
- ar |= (var->s & 1) << 4;
- ar |= (var->dpl & 3) << 5;
- ar |= (var->present & 1) << 7;
- ar |= (var->avl & 1) << 12;
- ar |= (var->l & 1) << 13;
- ar |= (var->db & 1) << 14;
- ar |= (var->g & 1) << 15;
- }
- vmcs_write32(sf->ar_bytes, ar);
+ return kvm_arch_ops->set_segment(vcpu, var, seg);
}
static int kvm_dev_ioctl_set_sregs(struct kvm *kvm, struct kvm_sregs *sregs)
@@ -2922,15 +2875,15 @@ static int kvm_dev_ioctl_set_sregs(struc
if (!vcpu)
return -ENOENT;
- set_segment(&sregs->cs, VCPU_SREG_CS);
- set_segment(&sregs->ds, VCPU_SREG_DS);
- set_segment(&sregs->es, VCPU_SREG_ES);
- set_segment(&sregs->fs, VCPU_SREG_FS);
- set_segment(&sregs->gs, VCPU_SREG_GS);
- set_segment(&sregs->ss, VCPU_SREG_SS);
+ set_segment(vcpu, &sregs->cs, VCPU_SREG_CS);
+ set_segment(vcpu, &sregs->ds, VCPU_SREG_DS);
+ set_segment(vcpu, &sregs->es, VCPU_SREG_ES);
+ set_segment(vcpu, &sregs->fs, VCPU_SREG_FS);
+ set_segment(vcpu, &sregs->gs, VCPU_SREG_GS);
+ set_segment(vcpu, &sregs->ss, VCPU_SREG_SS);
- set_segment(&sregs->tr, VCPU_SREG_TR);
- set_segment(&sregs->ldt, VCPU_SREG_LDTR);
+ set_segment(vcpu, &sregs->tr, VCPU_SREG_TR);
+ set_segment(vcpu, &sregs->ldt, VCPU_SREG_LDTR);
#define set_dtable(var, table) \
vmcs_write32(GUEST_##table##_LIMIT, sregs->var.limit), \
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -17,6 +17,8 @@
#include "kvm.h"
#include <linux/module.h>
+#include "vmx.h"
+#include "kvm_vmx.h"
#define MSR_IA32_FEATURE_CONTROL 0x03a
@@ -31,6 +33,13 @@ extern struct vmcs_descriptor {
u32 revision_id;
} vmcs_descriptor;
+extern struct kvm_vmx_segment_field {
+ unsigned selector;
+ unsigned base;
+ unsigned limit;
+ unsigned ar_bytes;
+} kvm_vmx_segment_fields[];
+
u64 guest_read_tsc(void);
void guest_write_tsc(u64 guest_tsc);
struct vmx_msr_entry *find_msr_entry(struct kvm_vcpu *vcpu, u32 msr);
@@ -300,6 +309,53 @@ static __exit void hardware_unsetup(void
free_kvm_area();
}
+static void vmx_get_segment(struct kvm_vcpu *vcpu,
+ struct kvm_segment *var, int seg)
+{
+ struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
+ u32 ar;
+
+ var->base = vmcs_readl(sf->base);
+ var->limit = vmcs_read32(sf->limit);
+ var->selector = vmcs_read16(sf->selector);
+ ar = vmcs_read32(sf->ar_bytes);
+ if (ar & AR_UNUSABLE_MASK)
+ ar = 0;
+ var->type = ar & 15;
+ var->s = (ar >> 4) & 1;
+ var->dpl = (ar >> 5) & 3;
+ var->present = (ar >> 7) & 1;
+ var->avl = (ar >> 12) & 1;
+ var->l = (ar >> 13) & 1;
+ var->db = (ar >> 14) & 1;
+ var->g = (ar >> 15) & 1;
+ var->unusable = (ar >> 16) & 1;
+}
+
+static void vmx_set_segment(struct kvm_vcpu *vcpu,
+ struct kvm_segment *var, int seg)
+{
+ struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
+ u32 ar;
+
+ vmcs_writel(sf->base, var->base);
+ vmcs_write32(sf->limit, var->limit);
+ vmcs_write16(sf->selector, var->selector);
+ if (var->unusable)
+ ar = 1 << 16;
+ else {
+ ar = var->type & 15;
+ ar |= (var->s & 1) << 4;
+ ar |= (var->dpl & 3) << 5;
+ ar |= (var->present & 1) << 7;
+ ar |= (var->avl & 1) << 12;
+ ar |= (var->l & 1) << 13;
+ ar |= (var->db & 1) << 14;
+ ar |= (var->g & 1) << 15;
+ }
+ vmcs_write32(sf->ar_bytes, ar);
+}
+
static struct kvm_arch_ops vmx_arch_ops = {
.cpu_has_kvm_support = cpu_has_kvm_support,
.disabled_by_bios = vmx_disabled_by_bios,
@@ -311,6 +367,8 @@ static struct kvm_arch_ops vmx_arch_ops
.set_guest_debug = set_guest_debug,
.get_msr = vmx_get_msr,
.set_msr = vmx_set_msr,
+ .get_segment = vmx_get_segment,
+ .set_segment = vmx_set_segment,
};
static int __init vmx_init(void)
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 9/38] KVM: Cache guest cr4 in vcpu structure
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (7 preceding siblings ...)
2006-11-27 12:18 ` [PATCH 8/38] KVM: Make the segment " Avi Kivity
@ 2006-11-27 12:19 ` Avi Kivity
2006-11-27 12:20 ` [PATCH 10/38] KVM: Cache guest cr0 " Avi Kivity
` (28 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:19 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
This eliminates needing to have an arch operation to get cr4.
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -168,6 +168,7 @@ struct kvm_vcpu {
unsigned long cr2;
unsigned long cr3;
+ unsigned long cr4;
unsigned long cr8;
u64 shadow_efer;
u64 apic_base;
@@ -335,20 +336,14 @@ static inline int is_long_mode(void)
return vmcs_read32(VM_ENTRY_CONTROLS) & VM_ENTRY_CONTROLS_IA32E_MASK;
}
-static inline unsigned long guest_cr4(void)
+static inline int is_pae(struct kvm_vcpu *vcpu)
{
- return (vmcs_readl(CR4_READ_SHADOW) & KVM_GUEST_CR4_MASK) |
- (vmcs_readl(GUEST_CR4) & ~KVM_GUEST_CR4_MASK);
+ return vcpu->cr4 & CR4_PAE_MASK;
}
-static inline int is_pae(void)
+static inline int is_pse(struct kvm_vcpu *vcpu)
{
- return guest_cr4() & CR4_PAE_MASK;
-}
-
-static inline int is_pse(void)
-{
- return guest_cr4() & CR4_PSE_MASK;
+ return vcpu->cr4 & CR4_PSE_MASK;
}
static inline unsigned long guest_cr0(void)
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -940,7 +940,7 @@ static void set_cr0(struct kvm_vcpu *vcp
#ifdef __x86_64__
if ((vcpu->shadow_efer & EFER_LME)) {
u32 guest_cs_ar;
- if (!is_pae()) {
+ if (!is_pae(vcpu)) {
printk(KERN_DEBUG "set_cr0: #GP, start paging "
"in long mode while PAE is disabled\n");
inject_gp(vcpu);
@@ -956,7 +956,7 @@ static void set_cr0(struct kvm_vcpu *vcp
}
} else
#endif
- if (is_pae() &&
+ if (is_pae(vcpu) &&
pdptrs_have_reserved_bits_set(vcpu, vcpu->cr3)) {
printk(KERN_DEBUG "set_cr0: #GP, pdptrs "
"reserved bits\n");
@@ -993,6 +993,7 @@ static void __set_cr4(struct kvm_vcpu *v
vmcs_writel(CR4_READ_SHADOW, cr4);
vmcs_writel(GUEST_CR4, cr4 | (vcpu->rmode.active ?
KVM_RMODE_VM_CR4_ALWAYS_ON : KVM_PMODE_VM_CR4_ALWAYS_ON));
+ vcpu->cr4 = cr4;
}
static void set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
@@ -1010,7 +1011,7 @@ static void set_cr4(struct kvm_vcpu *vcp
inject_gp(vcpu);
return;
}
- } else if (is_paging() && !is_pae() && (cr4 & CR4_PAE_MASK)
+ } else if (is_paging() && !is_pae(vcpu) && (cr4 & CR4_PAE_MASK)
&& pdptrs_have_reserved_bits_set(vcpu, vcpu->cr3)) {
printk(KERN_DEBUG "set_cr4: #GP, pdptrs reserved bits\n");
inject_gp(vcpu);
@@ -1041,7 +1042,7 @@ static void set_cr3(struct kvm_vcpu *vcp
inject_gp(vcpu);
return;
}
- if (is_paging() && is_pae() &&
+ if (is_paging() && is_pae(vcpu) &&
pdptrs_have_reserved_bits_set(vcpu, cr3)) {
printk(KERN_DEBUG "set_cr3: #GP, pdptrs "
"reserved bits\n");
@@ -1902,7 +1903,7 @@ unsigned long realmode_get_cr(struct kvm
case 3:
return vcpu->cr3;
case 4:
- return guest_cr4();
+ return vcpu->cr4;
default:
vcpu_printf(vcpu, "%s: unexpected cr %u\n", __FUNCTION__, cr);
return 0;
@@ -1924,7 +1925,7 @@ void realmode_set_cr(struct kvm_vcpu *vc
set_cr3(vcpu, val);
break;
case 4:
- set_cr4(vcpu, mk_cr_64(guest_cr4(), val));
+ set_cr4(vcpu, mk_cr_64(vcpu->cr4, val));
break;
default:
vcpu_printf(vcpu, "%s: unexpected cr %u\n", __FUNCTION__, cr);
@@ -2844,7 +2845,7 @@ static int kvm_dev_ioctl_get_sregs(struc
sregs->cr0 = guest_cr0();
sregs->cr2 = vcpu->cr2;
sregs->cr3 = vcpu->cr3;
- sregs->cr4 = guest_cr4();
+ sregs->cr4 = vcpu->cr4;
sregs->cr8 = vcpu->cr8;
sregs->efer = vcpu->shadow_efer;
sregs->apic_base = vcpu->apic_base;
@@ -2912,7 +2913,7 @@ static int kvm_dev_ioctl_set_sregs(struc
vmcs_writel(GUEST_CR0,
(sregs->cr0 & ~KVM_GUEST_CR0_MASK) | KVM_VM_CR0_ALWAYS_ON);
- mmu_reset_needed |= guest_cr4() != sregs->cr4;
+ mmu_reset_needed |= vcpu->cr4 != sregs->cr4;
__set_cr4(vcpu, sregs->cr4);
if (mmu_reset_needed)
Index: linux-2.6/drivers/kvm/mmu.c
===================================================================
--- linux-2.6.orig/drivers/kvm/mmu.c
+++ linux-2.6/drivers/kvm/mmu.c
@@ -564,7 +564,7 @@ static int paging64_init_context(struct
{
struct kvm_mmu *context = &vcpu->mmu;
- ASSERT(is_pae());
+ ASSERT(is_pae(vcpu));
context->new_cr3 = paging_new_cr3;
context->page_fault = paging64_page_fault;
context->inval_page = paging_inval_page;
@@ -618,7 +618,7 @@ static int init_kvm_mmu(struct kvm_vcpu
return nonpaging_init_context(vcpu);
else if (is_long_mode())
return paging64_init_context(vcpu);
- else if (is_pae())
+ else if (is_pae(vcpu))
return paging32E_init_context(vcpu);
else
return paging32_init_context(vcpu);
Index: linux-2.6/drivers/kvm/paging_tmpl.h
===================================================================
--- linux-2.6.orig/drivers/kvm/paging_tmpl.h
+++ linux-2.6/drivers/kvm/paging_tmpl.h
@@ -70,7 +70,7 @@ static void FNAME(init_walker)(struct gu
hpa = safe_gpa_to_hpa(vcpu, vcpu->cr3 & PT64_BASE_ADDR_MASK);
walker->table = kmap_atomic(pfn_to_page(hpa >> PAGE_SHIFT), KM_USER0);
- ASSERT((!is_long_mode() && is_pae()) ||
+ ASSERT((!is_long_mode() && is_pae(vcpu)) ||
(vcpu->cr3 & ~(PAGE_MASK | CR3_FLAGS_MASK)) == 0);
walker->table = (pt_element_t *)( (unsigned long)walker->table |
@@ -133,7 +133,7 @@ static pt_element_t *FNAME(fetch_guest)(
!is_present_pte(walker->table[index]) ||
(walker->level == PT_DIRECTORY_LEVEL &&
(walker->table[index] & PT_PAGE_SIZE_MASK) &&
- (PTTYPE == 64 || is_pse())))
+ (PTTYPE == 64 || is_pse(vcpu))))
return &walker->table[index];
if (walker->level != 3 || is_long_mode())
walker->inherited_ar &= walker->table[index];
@@ -369,7 +369,7 @@ static gpa_t FNAME(gva_to_gpa)(struct kv
if (walker.level == PT_DIRECTORY_LEVEL) {
ASSERT((guest_pte & PT_PAGE_SIZE_MASK));
- ASSERT(PTTYPE == 64 || is_pse());
+ ASSERT(PTTYPE == 64 || is_pse(vcpu));
gpa = (guest_pte & PT_DIR_BASE_ADDR_MASK) | (vaddr &
(PT_LEVEL_MASK(PT_PAGE_TABLE_LEVEL) | ~PAGE_MASK));
Index: linux-2.6/drivers/kvm/kvm_vmx.h
===================================================================
--- /dev/null
+++ linux-2.6/drivers/kvm/kvm_vmx.h
@@ -0,0 +1,20 @@
+#ifndef __KVM_VMX_H
+#define __KVM_VMX_H
+
+static inline void vmcs_write16(unsigned long field, u16 value)
+{
+ vmcs_writel(field, value);
+}
+
+static inline void vmcs_write64(unsigned long field, u64 value)
+{
+#ifdef __x86_64__
+ vmcs_writel(field, value);
+#else
+ vmcs_writel(field, value);
+ asm volatile ("");
+ vmcs_writel(field+1, value >> 32);
+#endif
+}
+
+#endif
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 10/38] KVM: Cache guest cr0 in vcpu structure
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (8 preceding siblings ...)
2006-11-27 12:19 ` [PATCH 9/38] KVM: Cache guest cr4 in vcpu structure Avi Kivity
@ 2006-11-27 12:20 ` Avi Kivity
2006-11-27 12:21 ` [PATCH 11/38] KVM: Add get_segment_base() arch accessor Avi Kivity
` (27 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:20 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
This eliminates needing to have an arch operation to get cr0.
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -41,7 +41,6 @@
| CR0_NW_MASK | CR0_CD_MASK)
#define KVM_VM_CR0_ALWAYS_ON \
(CR0_PG_MASK | CR0_PE_MASK | CR0_WP_MASK | CR0_NE_MASK)
-
#define KVM_GUEST_CR4_MASK \
(CR4_PSE_MASK | CR4_PAE_MASK | CR4_PGE_MASK | CR4_VMXE_MASK | CR4_VME_MASK)
#define KVM_PMODE_VM_CR4_ALWAYS_ON (CR4_VMXE_MASK | CR4_PAE_MASK)
@@ -166,6 +165,7 @@ struct kvm_vcpu {
unsigned long regs[NR_VCPU_REGS]; /* for rsp: vcpu_load_rsp_rip() */
unsigned long rip; /* needs vcpu_load_rsp_rip() */
+ unsigned long cr0;
unsigned long cr2;
unsigned long cr3;
unsigned long cr4;
@@ -346,20 +346,14 @@ static inline int is_pse(struct kvm_vcpu
return vcpu->cr4 & CR4_PSE_MASK;
}
-static inline unsigned long guest_cr0(void)
-{
- return (vmcs_readl(CR0_READ_SHADOW) & KVM_GUEST_CR0_MASK) |
- (vmcs_readl(GUEST_CR0) & ~KVM_GUEST_CR0_MASK);
-}
-
static inline unsigned guest_cpl(void)
{
return vmcs_read16(GUEST_CS_SELECTOR) & SELECTOR_RPL_MASK;
}
-static inline int is_paging(void)
+static inline int is_paging(struct kvm_vcpu *vcpu)
{
- return guest_cr0() & CR0_PG_MASK;
+ return vcpu->cr0 & CR0_PG_MASK;
}
static inline int is_page_fault(u32 intr_info)
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -875,9 +875,9 @@ static void __set_cr0(struct kvm_vcpu *v
#ifdef __x86_64__
if (vcpu->shadow_efer & EFER_LME) {
- if (!is_paging() && (cr0 & CR0_PG_MASK))
+ if (!is_paging(vcpu) && (cr0 & CR0_PG_MASK))
enter_lmode(vcpu);
- if (is_paging() && !(cr0 & CR0_PG_MASK))
+ if (is_paging(vcpu) && !(cr0 & CR0_PG_MASK))
exit_lmode(vcpu);
}
#endif
@@ -885,6 +885,7 @@ static void __set_cr0(struct kvm_vcpu *v
vmcs_writel(CR0_READ_SHADOW, cr0);
vmcs_writel(GUEST_CR0,
(cr0 & ~KVM_GUEST_CR0_MASK) | KVM_VM_CR0_ALWAYS_ON);
+ vcpu->cr0 = cr0;
}
static int pdptrs_have_reserved_bits_set(struct kvm_vcpu *vcpu,
@@ -918,7 +919,7 @@ static void set_cr0(struct kvm_vcpu *vcp
{
if (cr0 & CR0_RESEVED_BITS) {
printk(KERN_DEBUG "set_cr0: 0x%lx #GP, reserved bits 0x%lx\n",
- cr0, guest_cr0());
+ cr0, vcpu->cr0);
inject_gp(vcpu);
return;
}
@@ -936,7 +937,7 @@ static void set_cr0(struct kvm_vcpu *vcp
return;
}
- if (!is_paging() && (cr0 & CR0_PG_MASK)) {
+ if (!is_paging(vcpu) && (cr0 & CR0_PG_MASK)) {
#ifdef __x86_64__
if ((vcpu->shadow_efer & EFER_LME)) {
u32 guest_cs_ar;
@@ -975,7 +976,7 @@ static void set_cr0(struct kvm_vcpu *vcp
static void lmsw(struct kvm_vcpu *vcpu, unsigned long msw)
{
- unsigned long cr0 = guest_cr0();
+ unsigned long cr0 = vcpu->cr0;
if ((msw & CR0_PE_MASK) && !(cr0 & CR0_PE_MASK)) {
enter_pmode(vcpu);
@@ -986,6 +987,7 @@ static void lmsw(struct kvm_vcpu *vcpu,
vmcs_writel(GUEST_CR0, (vmcs_readl(GUEST_CR0) & ~LMSW_GUEST_MASK)
| (msw & LMSW_GUEST_MASK));
+ vcpu->cr0 = (vcpu->cr0 & ~0xfffful) | msw;
}
static void __set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
@@ -1011,7 +1013,7 @@ static void set_cr4(struct kvm_vcpu *vcp
inject_gp(vcpu);
return;
}
- } else if (is_paging() && !is_pae(vcpu) && (cr4 & CR4_PAE_MASK)
+ } else if (is_paging(vcpu) && !is_pae(vcpu) && (cr4 & CR4_PAE_MASK)
&& pdptrs_have_reserved_bits_set(vcpu, vcpu->cr3)) {
printk(KERN_DEBUG "set_cr4: #GP, pdptrs reserved bits\n");
inject_gp(vcpu);
@@ -1042,7 +1044,7 @@ static void set_cr3(struct kvm_vcpu *vcp
inject_gp(vcpu);
return;
}
- if (is_paging() && is_pae(vcpu) &&
+ if (is_paging(vcpu) && is_pae(vcpu) &&
pdptrs_have_reserved_bits_set(vcpu, cr3)) {
printk(KERN_DEBUG "set_cr3: #GP, pdptrs "
"reserved bits\n");
@@ -1897,7 +1899,7 @@ unsigned long realmode_get_cr(struct kvm
{
switch (cr) {
case 0:
- return guest_cr0();
+ return vcpu->cr0;
case 2:
return vcpu->cr2;
case 3:
@@ -1915,7 +1917,7 @@ void realmode_set_cr(struct kvm_vcpu *vc
{
switch (cr) {
case 0:
- set_cr0(vcpu, mk_cr_64(guest_cr0(), val));
+ set_cr0(vcpu, mk_cr_64(vcpu->cr0, val));
*rflags = vmcs_readl(GUEST_RFLAGS);
break;
case 2:
@@ -2264,7 +2266,8 @@ void set_efer(struct kvm_vcpu *vcpu, u64
return;
}
- if (is_paging() && (vcpu->shadow_efer & EFER_LME) != (efer & EFER_LME)) {
+ if (is_paging(vcpu)
+ && (vcpu->shadow_efer & EFER_LME) != (efer & EFER_LME)) {
printk(KERN_DEBUG "set_efer: #GP, change LME while paging\n");
inject_gp(vcpu);
return;
@@ -2842,7 +2845,7 @@ static int kvm_dev_ioctl_get_sregs(struc
get_dtable(gdt, GDTR);
#undef get_dtable
- sregs->cr0 = guest_cr0();
+ sregs->cr0 = vcpu->cr0;
sregs->cr2 = vcpu->cr2;
sregs->cr3 = vcpu->cr3;
sregs->cr4 = vcpu->cr4;
@@ -2906,12 +2909,13 @@ static int kvm_dev_ioctl_set_sregs(struc
#endif
vcpu->apic_base = sregs->apic_base;
- mmu_reset_needed |= guest_cr0() != sregs->cr0;
+ mmu_reset_needed |= vcpu->cr0 != sregs->cr0;
vcpu->rmode.active = ((sregs->cr0 & CR0_PE_MASK) == 0);
update_exception_bitmap(vcpu);
vmcs_writel(CR0_READ_SHADOW, sregs->cr0);
vmcs_writel(GUEST_CR0,
(sregs->cr0 & ~KVM_GUEST_CR0_MASK) | KVM_VM_CR0_ALWAYS_ON);
+ vcpu->cr0 = sregs->cr0;
mmu_reset_needed |= vcpu->cr4 != sregs->cr4;
__set_cr4(vcpu, sregs->cr4);
Index: linux-2.6/drivers/kvm/mmu.c
===================================================================
--- linux-2.6.orig/drivers/kvm/mmu.c
+++ linux-2.6/drivers/kvm/mmu.c
@@ -138,9 +138,9 @@
#define PT_DIRECTORY_LEVEL 2
#define PT_PAGE_TABLE_LEVEL 1
-static int is_write_protection(void)
+static int is_write_protection(struct kvm_vcpu *vcpu)
{
- return guest_cr0() & CR0_WP_MASK;
+ return vcpu->cr0 & CR0_WP_MASK;
}
static int is_cpuid_PSE36(void)
@@ -321,7 +321,7 @@ static void nonpaging_flush(struct kvm_v
root = kvm_mmu_alloc_page(vcpu, 0);
ASSERT(VALID_PAGE(root));
vcpu->mmu.root_hpa = root;
- if (is_paging())
+ if (is_paging(vcpu))
root |= (vcpu->cr3 & (CR3_PCD_MASK | CR3_WPT_MASK));
vmcs_writel(GUEST_CR3, root);
}
@@ -614,7 +614,7 @@ static int init_kvm_mmu(struct kvm_vcpu
ASSERT(vcpu);
ASSERT(!VALID_PAGE(vcpu->mmu.root_hpa));
- if (!is_paging())
+ if (!is_paging(vcpu))
return nonpaging_init_context(vcpu);
else if (is_long_mode())
return paging64_init_context(vcpu);
Index: linux-2.6/drivers/kvm/paging_tmpl.h
===================================================================
--- linux-2.6.orig/drivers/kvm/paging_tmpl.h
+++ linux-2.6/drivers/kvm/paging_tmpl.h
@@ -251,7 +251,7 @@ static int FNAME(fix_write_pf)(struct kv
* supervisor write protection is enabled.
*/
if (!writable_shadow) {
- if (is_write_protection())
+ if (is_write_protection(vcpu))
return 0;
*shadow_ent &= ~PT_USER_MASK;
}
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 11/38] KVM: Add get_segment_base() arch accessor
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (9 preceding siblings ...)
2006-11-27 12:20 ` [PATCH 10/38] KVM: Cache guest cr0 " Avi Kivity
@ 2006-11-27 12:21 ` Avi Kivity
2006-11-27 12:22 ` [PATCH 12/38] KVM: Add idt and gdt descriptor accessors Avi Kivity
` (26 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:21 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
This could have been implemented in terms of the existing get_segment(), but
that make unnecessary vmx accesses.
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -248,6 +248,7 @@ struct kvm_arch_ops {
struct kvm_debug_guest *dbg);
int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
int (*set_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 data);
+ u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
void (*get_segment)(struct kvm_vcpu *vcpu,
struct kvm_segment *var, int seg);
void (*set_segment)(struct kvm_vcpu *vcpu,
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -1776,12 +1776,19 @@ static int emulator_cmpxchg_emulated(uns
return emulator_write_emulated(addr, new, bytes, ctxt);
}
+static unsigned long get_segment_base(struct kvm_vcpu *vcpu, int seg)
+{
+ return kvm_arch_ops->get_segment_base(vcpu, seg);
+}
+
static void report_emulation_failure(struct x86_emulate_ctxt *ctxt)
{
static int reported;
u8 opcodes[4];
unsigned long rip = vmcs_readl(GUEST_RIP);
- unsigned long rip_linear = rip + vmcs_readl(GUEST_CS_BASE);
+ unsigned long rip_linear;
+
+ rip_linear = rip + get_segment_base(ctxt->vcpu, VCPU_SREG_CS);
if (reported)
return;
@@ -1835,14 +1842,14 @@ static int emulate_instruction(struct kv
emulate_ctxt.es_base = 0;
emulate_ctxt.ss_base = 0;
} else {
- emulate_ctxt.cs_base = vmcs_readl(GUEST_CS_BASE);
- emulate_ctxt.ds_base = vmcs_readl(GUEST_DS_BASE);
- emulate_ctxt.es_base = vmcs_readl(GUEST_ES_BASE);
- emulate_ctxt.ss_base = vmcs_readl(GUEST_SS_BASE);
+ emulate_ctxt.cs_base = get_segment_base(vcpu, VCPU_SREG_CS);
+ emulate_ctxt.ds_base = get_segment_base(vcpu, VCPU_SREG_DS);
+ emulate_ctxt.es_base = get_segment_base(vcpu, VCPU_SREG_ES);
+ emulate_ctxt.ss_base = get_segment_base(vcpu, VCPU_SREG_SS);
}
- emulate_ctxt.gs_base = vmcs_readl(GUEST_GS_BASE);
- emulate_ctxt.fs_base = vmcs_readl(GUEST_FS_BASE);
+ emulate_ctxt.gs_base = get_segment_base(vcpu, VCPU_SREG_GS);
+ emulate_ctxt.fs_base = get_segment_base(vcpu, VCPU_SREG_FS);
vcpu->mmio_is_write = 0;
r = x86_emulate_memop(&emulate_ctxt, &emulate_ops);
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -309,6 +309,13 @@ static __exit void hardware_unsetup(void
free_kvm_area();
}
+static u64 vmx_get_segment_base(struct kvm_vcpu *vcpu, int seg)
+{
+ struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
+
+ return vmcs_readl(sf->base);
+}
+
static void vmx_get_segment(struct kvm_vcpu *vcpu,
struct kvm_segment *var, int seg)
{
@@ -367,6 +374,7 @@ static struct kvm_arch_ops vmx_arch_ops
.set_guest_debug = set_guest_debug,
.get_msr = vmx_get_msr,
.set_msr = vmx_set_msr,
+ .get_segment_base = vmx_get_segment_base,
.get_segment = vmx_get_segment,
.set_segment = vmx_set_segment,
};
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 12/38] KVM: Add idt and gdt descriptor accessors
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (10 preceding siblings ...)
2006-11-27 12:21 ` [PATCH 11/38] KVM: Add get_segment_base() arch accessor Avi Kivity
@ 2006-11-27 12:22 ` Avi Kivity
2006-11-27 12:23 ` [PATCH 13/38] KVM: Make syncing the register file to the vcpu structure an arch operation Avi Kivity
` (25 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:22 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -236,6 +236,11 @@ struct kvm_stat {
u32 irq_exits;
};
+struct descriptor_table {
+ u16 limit;
+ unsigned long base;
+} __attribute__((packed));
+
struct kvm_arch_ops {
int (*cpu_has_kvm_support)(void); /* __init */
int (*disabled_by_bios)(void); /* __init */
@@ -253,6 +258,10 @@ struct kvm_arch_ops {
struct kvm_segment *var, int seg);
void (*set_segment)(struct kvm_vcpu *vcpu,
struct kvm_segment *var, int seg);
+ void (*get_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
+ void (*set_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
+ void (*get_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
+ void (*set_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
};
extern struct kvm_stat kvm_stat;
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -155,11 +155,6 @@ struct vmx_msr_entry *find_msr_entry(str
}
EXPORT_SYMBOL_GPL(find_msr_entry);
-struct descriptor_table {
- u16 limit;
- unsigned long base;
-} __attribute__((packed));
-
static void get_gdt(struct descriptor_table *table)
{
asm ("sgdt %0" : "=m"(*table));
@@ -2827,6 +2822,7 @@ static void get_segment(struct kvm_vcpu
static int kvm_dev_ioctl_get_sregs(struct kvm *kvm, struct kvm_sregs *sregs)
{
struct kvm_vcpu *vcpu;
+ struct descriptor_table dt;
if (sregs->vcpu < 0 || sregs->vcpu >= KVM_MAX_VCPUS)
return -EINVAL;
@@ -2844,13 +2840,12 @@ static int kvm_dev_ioctl_get_sregs(struc
get_segment(vcpu, &sregs->tr, VCPU_SREG_TR);
get_segment(vcpu, &sregs->ldt, VCPU_SREG_LDTR);
-#define get_dtable(var, table) \
- sregs->var.limit = vmcs_read32(GUEST_##table##_LIMIT), \
- sregs->var.base = vmcs_readl(GUEST_##table##_BASE)
-
- get_dtable(idt, IDTR);
- get_dtable(gdt, GDTR);
-#undef get_dtable
+ kvm_arch_ops->get_idt(vcpu, &dt);
+ sregs->idt.limit = dt.limit;
+ sregs->idt.base = dt.base;
+ kvm_arch_ops->get_gdt(vcpu, &dt);
+ sregs->gdt.limit = dt.limit;
+ sregs->gdt.base = dt.base;
sregs->cr0 = vcpu->cr0;
sregs->cr2 = vcpu->cr2;
@@ -2879,6 +2874,7 @@ static int kvm_dev_ioctl_set_sregs(struc
struct kvm_vcpu *vcpu;
int mmu_reset_needed = 0;
int i;
+ struct descriptor_table dt;
if (sregs->vcpu < 0 || sregs->vcpu >= KVM_MAX_VCPUS)
return -EINVAL;
@@ -2896,13 +2892,12 @@ static int kvm_dev_ioctl_set_sregs(struc
set_segment(vcpu, &sregs->tr, VCPU_SREG_TR);
set_segment(vcpu, &sregs->ldt, VCPU_SREG_LDTR);
-#define set_dtable(var, table) \
- vmcs_write32(GUEST_##table##_LIMIT, sregs->var.limit), \
- vmcs_writel(GUEST_##table##_BASE, sregs->var.base)
-
- set_dtable(idt, IDTR);
- set_dtable(gdt, GDTR);
-#undef set_dtable
+ dt.limit = sregs->idt.limit;
+ dt.base = sregs->idt.base;
+ kvm_arch_ops->set_idt(vcpu, &dt);
+ dt.limit = sregs->gdt.limit;
+ dt.base = sregs->gdt.base;
+ kvm_arch_ops->set_gdt(vcpu, &dt);
vcpu->cr2 = sregs->cr2;
mmu_reset_needed |= vcpu->cr3 != sregs->cr3;
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -363,6 +363,30 @@ static void vmx_set_segment(struct kvm_v
vmcs_write32(sf->ar_bytes, ar);
}
+static void vmx_get_idt(struct kvm_vcpu *vcpu, struct descriptor_table *dt)
+{
+ dt->limit = vmcs_read32(GUEST_IDTR_LIMIT);
+ dt->base = vmcs_readl(GUEST_IDTR_BASE);
+}
+
+static void vmx_set_idt(struct kvm_vcpu *vcpu, struct descriptor_table *dt)
+{
+ vmcs_write32(GUEST_IDTR_LIMIT, dt->limit);
+ vmcs_writel(GUEST_IDTR_BASE, dt->base);
+}
+
+static void vmx_get_gdt(struct kvm_vcpu *vcpu, struct descriptor_table *dt)
+{
+ dt->limit = vmcs_read32(GUEST_GDTR_LIMIT);
+ dt->base = vmcs_readl(GUEST_GDTR_BASE);
+}
+
+static void vmx_set_gdt(struct kvm_vcpu *vcpu, struct descriptor_table *dt)
+{
+ vmcs_write32(GUEST_GDTR_LIMIT, dt->limit);
+ vmcs_writel(GUEST_GDTR_BASE, dt->base);
+}
+
static struct kvm_arch_ops vmx_arch_ops = {
.cpu_has_kvm_support = cpu_has_kvm_support,
.disabled_by_bios = vmx_disabled_by_bios,
@@ -377,6 +401,10 @@ static struct kvm_arch_ops vmx_arch_ops
.get_segment_base = vmx_get_segment_base,
.get_segment = vmx_get_segment,
.set_segment = vmx_set_segment,
+ .get_idt = vmx_get_idt,
+ .set_idt = vmx_set_idt,
+ .get_gdt = vmx_get_gdt,
+ .set_gdt = vmx_set_gdt,
};
static int __init vmx_init(void)
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 13/38] KVM: Make syncing the register file to the vcpu structure an arch operation
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (11 preceding siblings ...)
2006-11-27 12:22 ` [PATCH 12/38] KVM: Add idt and gdt descriptor accessors Avi Kivity
@ 2006-11-27 12:23 ` Avi Kivity
2006-11-27 12:24 ` [PATCH 14/38] KVM: Make the vcpu execution loop " Avi Kivity
` (24 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:23 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
This copies any general purpose guest registers maintained by the hardware
to the vcpu structure (and back).
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -262,6 +262,8 @@ struct kvm_arch_ops {
void (*set_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
void (*get_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
void (*set_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
+ void (*cache_regs)(struct kvm_vcpu *vcpu);
+ void (*decache_regs)(struct kvm_vcpu *vcpu);
};
extern struct kvm_stat kvm_stat;
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -1327,26 +1327,6 @@ out:
}
/*
- * Sync the rsp and rip registers into the vcpu structure. This allows
- * registers to be accessed by indexing vcpu->regs.
- */
-static void vcpu_load_rsp_rip(struct kvm_vcpu *vcpu)
-{
- vcpu->regs[VCPU_REGS_RSP] = vmcs_readl(GUEST_RSP);
- vcpu->rip = vmcs_readl(GUEST_RIP);
-}
-
-/*
- * Syncs rsp and rip back into the vmcs. Should be called after possible
- * modification.
- */
-static void vcpu_put_rsp_rip(struct kvm_vcpu *vcpu)
-{
- vmcs_writel(GUEST_RSP, vcpu->regs[VCPU_REGS_RSP]);
- vmcs_writel(GUEST_RIP, vcpu->rip);
-}
-
-/*
* Creates some virtual cpus. Good luck creating more than one.
*/
static int kvm_dev_ioctl_create_vcpu(struct kvm *kvm, int n)
@@ -1819,7 +1799,7 @@ static int emulate_instruction(struct kv
int r;
u32 cs_ar;
- vcpu_load_rsp_rip(vcpu);
+ kvm_arch_ops->cache_regs(vcpu);
cs_ar = vmcs_read32(GUEST_CS_AR_BYTES);
@@ -1864,7 +1844,7 @@ static int emulate_instruction(struct kv
return EMULATE_DO_MMIO;
}
- vcpu_put_rsp_rip(vcpu);
+ kvm_arch_ops->decache_regs(vcpu);
vmcs_writel(GUEST_RFLAGS, emulate_ctxt.eflags);
if (vcpu->mmio_is_write)
@@ -2134,22 +2114,22 @@ static int handle_cr(struct kvm_vcpu *vc
case 0: /* mov to cr */
switch (cr) {
case 0:
- vcpu_load_rsp_rip(vcpu);
+ kvm_arch_ops->cache_regs(vcpu);
set_cr0(vcpu, vcpu->regs[reg]);
skip_emulated_instruction(vcpu);
return 1;
case 3:
- vcpu_load_rsp_rip(vcpu);
+ kvm_arch_ops->cache_regs(vcpu);
set_cr3(vcpu, vcpu->regs[reg]);
skip_emulated_instruction(vcpu);
return 1;
case 4:
- vcpu_load_rsp_rip(vcpu);
+ kvm_arch_ops->cache_regs(vcpu);
set_cr4(vcpu, vcpu->regs[reg]);
skip_emulated_instruction(vcpu);
return 1;
case 8:
- vcpu_load_rsp_rip(vcpu);
+ kvm_arch_ops->cache_regs(vcpu);
set_cr8(vcpu, vcpu->regs[reg]);
skip_emulated_instruction(vcpu);
return 1;
@@ -2158,17 +2138,17 @@ static int handle_cr(struct kvm_vcpu *vc
case 1: /*mov from cr*/
switch (cr) {
case 3:
- vcpu_load_rsp_rip(vcpu);
+ kvm_arch_ops->cache_regs(vcpu);
vcpu->regs[reg] = vcpu->cr3;
- vcpu_put_rsp_rip(vcpu);
+ kvm_arch_ops->decache_regs(vcpu);
skip_emulated_instruction(vcpu);
return 1;
case 8:
printk(KERN_DEBUG "handle_cr: read CR8 "
"cpu erratum AA15\n");
- vcpu_load_rsp_rip(vcpu);
+ kvm_arch_ops->cache_regs(vcpu);
vcpu->regs[reg] = vcpu->cr8;
- vcpu_put_rsp_rip(vcpu);
+ kvm_arch_ops->decache_regs(vcpu);
skip_emulated_instruction(vcpu);
return 1;
}
@@ -2200,7 +2180,7 @@ static int handle_dr(struct kvm_vcpu *vc
exit_qualification = vmcs_read64(EXIT_QUALIFICATION);
dr = exit_qualification & 7;
reg = (exit_qualification >> 8) & 15;
- vcpu_load_rsp_rip(vcpu);
+ kvm_arch_ops->cache_regs(vcpu);
if (exit_qualification & 16) {
/* mov from dr */
switch (dr) {
@@ -2217,7 +2197,7 @@ static int handle_dr(struct kvm_vcpu *vc
} else {
/* mov to dr */
}
- vcpu_put_rsp_rip(vcpu);
+ kvm_arch_ops->decache_regs(vcpu);
skip_emulated_instruction(vcpu);
return 1;
}
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -180,6 +180,26 @@ static int vmx_set_msr(struct kvm_vcpu *
return 0;
}
+/*
+ * Sync the rsp and rip registers into the vcpu structure. This allows
+ * registers to be accessed by indexing vcpu->regs.
+ */
+static void vcpu_load_rsp_rip(struct kvm_vcpu *vcpu)
+{
+ vcpu->regs[VCPU_REGS_RSP] = vmcs_readl(GUEST_RSP);
+ vcpu->rip = vmcs_readl(GUEST_RIP);
+}
+
+/*
+ * Syncs rsp and rip back into the vmcs. Should be called after possible
+ * modification.
+ */
+static void vcpu_put_rsp_rip(struct kvm_vcpu *vcpu)
+{
+ vmcs_writel(GUEST_RSP, vcpu->regs[VCPU_REGS_RSP]);
+ vmcs_writel(GUEST_RIP, vcpu->rip);
+}
+
static int set_guest_debug(struct kvm_vcpu *vcpu, struct kvm_debug_guest *dbg)
{
unsigned long dr7 = 0x400;
@@ -405,6 +425,8 @@ static struct kvm_arch_ops vmx_arch_ops
.set_idt = vmx_set_idt,
.get_gdt = vmx_get_gdt,
.set_gdt = vmx_set_gdt,
+ .cache_regs = vcpu_load_rsp_rip,
+ .decache_regs = vcpu_put_rsp_rip,
};
static int __init vmx_init(void)
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 14/38] KVM: Make the vcpu execution loop an arch operation
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (12 preceding siblings ...)
2006-11-27 12:23 ` [PATCH 13/38] KVM: Make syncing the register file to the vcpu structure an arch operation Avi Kivity
@ 2006-11-27 12:24 ` Avi Kivity
2006-11-27 12:25 ` [PATCH 15/38] KVM: Move the vmx exit handlers to vmx.c Avi Kivity
` (23 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:24 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -264,6 +264,9 @@ struct kvm_arch_ops {
void (*set_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
void (*cache_regs)(struct kvm_vcpu *vcpu);
void (*decache_regs)(struct kvm_vcpu *vcpu);
+
+ int (*run)(struct kvm_vcpu *vcpu, struct kvm_run *run);
+ unsigned long vmx_return; /* temporary hack */
};
extern struct kvm_stat kvm_stat;
@@ -306,6 +309,11 @@ unsigned long realmode_get_cr(struct kvm
void realmode_set_cr(struct kvm_vcpu *vcpu, int cr, unsigned long value,
unsigned long *rflags);
+void load_msrs(struct vmx_msr_entry *e, int n);
+void save_msrs(struct vmx_msr_entry *e, int n);
+void kvm_resched(struct kvm_vcpu *vcpu);
+int kvm_handle_exit(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu);
+
int kvm_read_guest(struct kvm_vcpu *vcpu,
gva_t addr,
unsigned long size,
@@ -398,6 +406,69 @@ static inline struct kvm_mmu_page *page_
return (struct kvm_mmu_page *)page->private;
}
+static inline u16 read_fs(void)
+{
+ u16 seg;
+ asm ("mov %%fs, %0" : "=g"(seg));
+ return seg;
+}
+
+static inline u16 read_gs(void)
+{
+ u16 seg;
+ asm ("mov %%gs, %0" : "=g"(seg));
+ return seg;
+}
+
+static inline u16 read_ldt(void)
+{
+ u16 ldt;
+ asm ("sldt %0" : "=g"(ldt));
+ return ldt;
+}
+
+static inline void load_fs(u16 sel)
+{
+ asm ("mov %0, %%fs" : : "rm"(sel));
+}
+
+static inline void load_gs(u16 sel)
+{
+ asm ("mov %0, %%gs" : : "rm"(sel));
+}
+
+#ifndef load_ldt
+static inline void load_ldt(u16 sel)
+{
+ asm ("lldt %0" : : "g"(sel));
+}
+#endif
+
+#ifdef __x86_64__
+static inline unsigned long read_msr(unsigned long msr)
+{
+ u64 value;
+
+ rdmsrl(msr, value);
+ return value;
+}
+#endif
+
+static inline void fx_save(void *image)
+{
+ asm ("fxsave (%0)":: "r" (image));
+}
+
+static inline void fx_restore(void *image)
+{
+ asm ("fxrstor (%0)":: "r" (image));
+}
+
+static inline void fpu_init(void)
+{
+ asm ("finit");
+}
+
#define ASM_VMX_VMCLEAR_RAX ".byte 0x66, 0x0f, 0xc7, 0x30"
#define ASM_VMX_VMLAUNCH ".byte 0x0f, 0x01, 0xc2"
#define ASM_VMX_VMRESUME ".byte 0x0f, 0x01, 0xc3"
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -44,6 +44,7 @@ MODULE_LICENSE("GPL");
struct kvm_arch_ops *kvm_arch_ops;
struct kvm_stat kvm_stat;
+EXPORT_SYMBOL_GPL(kvm_stat);
static struct kvm_stats_debugfs_item {
const char *name;
@@ -108,16 +109,6 @@ static const u32 vmx_msr_index[] = {
};
#define NR_VMX_MSR (sizeof(vmx_msr_index) / sizeof(*vmx_msr_index))
-#ifdef __x86_64__
-/*
- * avoid save/load MSR_SYSCALL_MASK and MSR_LSTAR by std vt
- * mechanism (cpu bug AA24)
- */
-#define NR_BAD_MSRS 2
-#else
-#define NR_BAD_MSRS 0
-#endif
-
#define TSS_IOPB_BASE_OFFSET 0x66
#define TSS_BASE_SIZE 0x68
#define TSS_IOPB_SIZE (65536 / 8)
@@ -165,59 +156,6 @@ static void get_idt(struct descriptor_ta
asm ("sidt %0" : "=m"(*table));
}
-static u16 read_fs(void)
-{
- u16 seg;
- asm ("mov %%fs, %0" : "=g"(seg));
- return seg;
-}
-
-static u16 read_gs(void)
-{
- u16 seg;
- asm ("mov %%gs, %0" : "=g"(seg));
- return seg;
-}
-
-static u16 read_ldt(void)
-{
- u16 ldt;
- asm ("sldt %0" : "=g"(ldt));
- return ldt;
-}
-
-static void load_fs(u16 sel)
-{
- asm ("mov %0, %%fs" : : "rm"(sel));
-}
-
-static void load_gs(u16 sel)
-{
- asm ("mov %0, %%gs" : : "rm"(sel));
-}
-
-#ifndef load_ldt
-static void load_ldt(u16 sel)
-{
- asm ("lldt %0" : : "g"(sel));
-}
-#endif
-
-static void fx_save(void *image)
-{
- asm ("fxsave (%0)":: "r" (image));
-}
-
-static void fx_restore(void *image)
-{
- asm ("fxrstor (%0)":: "r" (image));
-}
-
-static void fpu_init(void)
-{
- asm ("finit");
-}
-
struct segment_descriptor {
u16 limit_low;
u16 base_low;
@@ -278,23 +216,6 @@ static unsigned long read_tr_base(void)
return segment_base(tr);
}
-static void reload_tss(void)
-{
-#ifndef __x86_64__
-
- /*
- * VT restores TR but not its size. Useless.
- */
- struct descriptor_table gdt;
- struct segment_descriptor *descs;
-
- get_gdt(&gdt);
- descs = (void *)gdt.base;
- descs[GDT_ENTRY_TSS].type = 9; /* available TSS */
- load_TR_desc();
-#endif
-}
-
DEFINE_PER_CPU(struct vmcs *, vmxarea);
EXPORT_SYMBOL_GPL(per_cpu__vmxarea); /* temporary hack */
static DEFINE_PER_CPU(struct vmcs *, current_vmcs);
@@ -306,24 +227,12 @@ struct vmcs_descriptor {
} vmcs_descriptor;
EXPORT_SYMBOL_GPL(vmcs_descriptor);
-#ifdef __x86_64__
-static unsigned long read_msr(unsigned long msr)
-{
- u64 value;
-
- rdmsrl(msr, value);
- return value;
-}
-#endif
-
static inline struct page *_gfn_to_page(struct kvm *kvm, gfn_t gfn)
{
struct kvm_memory_slot *slot = gfn_to_memslot(kvm, gfn);
return (slot) ? slot->phys_mem[gfn - slot->base_gfn] : 0;
}
-
-
int kvm_read_guest(struct kvm_vcpu *vcpu,
gva_t addr,
unsigned long size,
@@ -357,6 +266,7 @@ int kvm_read_guest(struct kvm_vcpu *vcpu
}
return req_size - size;
}
+EXPORT_SYMBOL_GPL(kvm_read_guest);
int kvm_write_guest(struct kvm_vcpu *vcpu,
gva_t addr,
@@ -390,6 +300,7 @@ int kvm_write_guest(struct kvm_vcpu *vcp
}
return req_size - size;
}
+EXPORT_SYMBOL_GPL(kvm_write_guest);
static void vmcs_clear(struct vmcs *vmcs)
{
@@ -1120,7 +1031,6 @@ static void seg_setup(int seg)
*/
static int kvm_vcpu_setup(struct kvm_vcpu *vcpu)
{
- extern asmlinkage void kvm_vmx_return(void);
u32 host_sysenter_cs;
u32 junk;
unsigned long a;
@@ -1251,7 +1161,7 @@ static int kvm_vcpu_setup(struct kvm_vcp
vmcs_writel(HOST_IDTR_BASE, dt.base); /* 22.2.4 */
- vmcs_writel(HOST_RIP, (unsigned long)kvm_vmx_return); /* 22.2.5 */
+ vmcs_writel(HOST_RIP, kvm_arch_ops->vmx_return); /* 22.2.5 */
rdmsr(MSR_IA32_SYSENTER_CS, host_sysenter_cs, junk);
vmcs_write32(HOST_IA32_SYSENTER_CS, host_sysenter_cs);
@@ -2344,7 +2254,7 @@ static const int kvm_vmx_max_exit_handle
* The guest has exited. See if we can fix it or if we need userspace
* assistance.
*/
-static int kvm_handle_exit(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
+int kvm_handle_exit(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
{
u32 vectoring_info = vmcs_read32(IDT_VECTORING_INFO_FIELD);
u32 exit_reason = vmcs_read32(VM_EXIT_REASON);
@@ -2363,127 +2273,39 @@ static int kvm_handle_exit(struct kvm_ru
}
return 0;
}
+EXPORT_SYMBOL_GPL(kvm_handle_exit);
-static void inject_rmode_irq(struct kvm_vcpu *vcpu, int irq)
+void kvm_resched(struct kvm_vcpu *vcpu)
{
- u16 ent[2];
- u16 cs;
- u16 ip;
- unsigned long flags;
- unsigned long ss_base = vmcs_readl(GUEST_SS_BASE);
- u16 sp = vmcs_readl(GUEST_RSP);
- u32 ss_limit = vmcs_read32(GUEST_SS_LIMIT);
-
- if (sp > ss_limit || sp - 6 > sp) {
- vcpu_printf(vcpu, "%s: #SS, rsp 0x%lx ss 0x%lx limit 0x%x\n",
- __FUNCTION__,
- vmcs_readl(GUEST_RSP),
- vmcs_readl(GUEST_SS_BASE),
- vmcs_read32(GUEST_SS_LIMIT));
- return;
- }
-
- if (kvm_read_guest(vcpu, irq * sizeof(ent), sizeof(ent), &ent) !=
- sizeof(ent)) {
- vcpu_printf(vcpu, "%s: read guest err\n", __FUNCTION__);
- return;
- }
-
- flags = vmcs_readl(GUEST_RFLAGS);
- cs = vmcs_readl(GUEST_CS_BASE) >> 4;
- ip = vmcs_readl(GUEST_RIP);
-
-
- if (kvm_write_guest(vcpu, ss_base + sp - 2, 2, &flags) != 2 ||
- kvm_write_guest(vcpu, ss_base + sp - 4, 2, &cs) != 2 ||
- kvm_write_guest(vcpu, ss_base + sp - 6, 2, &ip) != 2) {
- vcpu_printf(vcpu, "%s: write guest err\n", __FUNCTION__);
- return;
- }
-
- vmcs_writel(GUEST_RFLAGS, flags &
- ~( X86_EFLAGS_IF | X86_EFLAGS_AC | X86_EFLAGS_TF));
- vmcs_write16(GUEST_CS_SELECTOR, ent[1]) ;
- vmcs_writel(GUEST_CS_BASE, ent[1] << 4);
- vmcs_writel(GUEST_RIP, ent[0]);
- vmcs_writel(GUEST_RSP, (vmcs_readl(GUEST_RSP) & ~0xffff) | (sp - 6));
-}
-
-static void kvm_do_inject_irq(struct kvm_vcpu *vcpu)
-{
- int word_index = __ffs(vcpu->irq_summary);
- int bit_index = __ffs(vcpu->irq_pending[word_index]);
- int irq = word_index * BITS_PER_LONG + bit_index;
-
- clear_bit(bit_index, &vcpu->irq_pending[word_index]);
- if (!vcpu->irq_pending[word_index])
- clear_bit(word_index, &vcpu->irq_summary);
-
- if (vcpu->rmode.active) {
- inject_rmode_irq(vcpu, irq);
- return;
- }
- vmcs_write32(VM_ENTRY_INTR_INFO_FIELD,
- irq | INTR_TYPE_EXT_INTR | INTR_INFO_VALID_MASK);
-}
-
-static void kvm_try_inject_irq(struct kvm_vcpu *vcpu)
-{
- if ((vmcs_readl(GUEST_RFLAGS) & X86_EFLAGS_IF)
- && (vmcs_read32(GUEST_INTERRUPTIBILITY_INFO) & 3) == 0)
- /*
- * Interrupts enabled, and not blocked by sti or mov ss. Good.
- */
- kvm_do_inject_irq(vcpu);
- else
- /*
- * Interrupts blocked. Wait for unblock.
- */
- vmcs_write32(CPU_BASED_VM_EXEC_CONTROL,
- vmcs_read32(CPU_BASED_VM_EXEC_CONTROL)
- | CPU_BASED_VIRTUAL_INTR_PENDING);
-}
-
-static void kvm_guest_debug_pre(struct kvm_vcpu *vcpu)
-{
- struct kvm_guest_debug *dbg = &vcpu->guest_debug;
-
- set_debugreg(dbg->bp[0], 0);
- set_debugreg(dbg->bp[1], 1);
- set_debugreg(dbg->bp[2], 2);
- set_debugreg(dbg->bp[3], 3);
-
- if (dbg->singlestep) {
- unsigned long flags;
-
- flags = vmcs_readl(GUEST_RFLAGS);
- flags |= X86_EFLAGS_TF | X86_EFLAGS_RF;
- vmcs_writel(GUEST_RFLAGS, flags);
- }
+ vcpu_put(vcpu);
+ cond_resched();
+ /* Cannot fail - no vcpu unplug yet. */
+ vcpu_load(vcpu->kvm, vcpu_slot(vcpu));
}
+EXPORT_SYMBOL_GPL(kvm_resched);
-static void load_msrs(struct vmx_msr_entry *e, int n)
+void load_msrs(struct vmx_msr_entry *e, int n)
{
int i;
for (i = 0; i < n; ++i)
wrmsrl(e[i].index, e[i].data);
}
+EXPORT_SYMBOL_GPL(load_msrs);
-static void save_msrs(struct vmx_msr_entry *e, int n)
+void save_msrs(struct vmx_msr_entry *e, int n)
{
int i;
for (i = 0; i < n; ++i)
rdmsrl(e[i].index, e[i].data);
}
+EXPORT_SYMBOL_GPL(save_msrs);
static int kvm_dev_ioctl_run(struct kvm *kvm, struct kvm_run *kvm_run)
{
struct kvm_vcpu *vcpu;
- u8 fail;
- u16 fs_sel, gs_sel, ldt_sel;
- int fs_gs_ldt_reload_needed;
+ int r;
if (kvm_run->vcpu < 0 || kvm_run->vcpu >= KVM_MAX_VCPUS)
return -EINVAL;
@@ -2504,211 +2326,10 @@ static int kvm_dev_ioctl_run(struct kvm
vcpu->mmio_needed = 0;
-again:
- /*
- * Set host fs and gs selectors. Unfortunately, 22.2.3 does not
- * allow segment selectors with cpl > 0 or ti == 1.
- */
- fs_sel = read_fs();
- gs_sel = read_gs();
- ldt_sel = read_ldt();
- fs_gs_ldt_reload_needed = (fs_sel & 7) | (gs_sel & 7) | ldt_sel;
- if (!fs_gs_ldt_reload_needed) {
- vmcs_write16(HOST_FS_SELECTOR, fs_sel);
- vmcs_write16(HOST_GS_SELECTOR, gs_sel);
- } else {
- vmcs_write16(HOST_FS_SELECTOR, 0);
- vmcs_write16(HOST_GS_SELECTOR, 0);
- }
-
-#ifdef __x86_64__
- vmcs_writel(HOST_FS_BASE, read_msr(MSR_FS_BASE));
- vmcs_writel(HOST_GS_BASE, read_msr(MSR_GS_BASE));
-#endif
-
- if (vcpu->irq_summary &&
- !(vmcs_read32(VM_ENTRY_INTR_INFO_FIELD) & INTR_INFO_VALID_MASK))
- kvm_try_inject_irq(vcpu);
-
- if (vcpu->guest_debug.enabled)
- kvm_guest_debug_pre(vcpu);
-
- fx_save(vcpu->host_fx_image);
- fx_restore(vcpu->guest_fx_image);
-
- save_msrs(vcpu->host_msrs, vcpu->nmsrs);
- load_msrs(vcpu->guest_msrs, NR_BAD_MSRS);
-
- asm (
- /* Store host registers */
- "pushf \n\t"
-#ifdef __x86_64__
- "push %%rax; push %%rbx; push %%rdx;"
- "push %%rsi; push %%rdi; push %%rbp;"
- "push %%r8; push %%r9; push %%r10; push %%r11;"
- "push %%r12; push %%r13; push %%r14; push %%r15;"
- "push %%rcx \n\t"
- ASM_VMX_VMWRITE_RSP_RDX "\n\t"
-#else
- "pusha; push %%ecx \n\t"
- ASM_VMX_VMWRITE_RSP_RDX "\n\t"
-#endif
- /* Check if vmlaunch of vmresume is needed */
- "cmp $0, %1 \n\t"
- /* Load guest registers. Don't clobber flags. */
-#ifdef __x86_64__
- "mov %c[cr2](%3), %%rax \n\t"
- "mov %%rax, %%cr2 \n\t"
- "mov %c[rax](%3), %%rax \n\t"
- "mov %c[rbx](%3), %%rbx \n\t"
- "mov %c[rdx](%3), %%rdx \n\t"
- "mov %c[rsi](%3), %%rsi \n\t"
- "mov %c[rdi](%3), %%rdi \n\t"
- "mov %c[rbp](%3), %%rbp \n\t"
- "mov %c[r8](%3), %%r8 \n\t"
- "mov %c[r9](%3), %%r9 \n\t"
- "mov %c[r10](%3), %%r10 \n\t"
- "mov %c[r11](%3), %%r11 \n\t"
- "mov %c[r12](%3), %%r12 \n\t"
- "mov %c[r13](%3), %%r13 \n\t"
- "mov %c[r14](%3), %%r14 \n\t"
- "mov %c[r15](%3), %%r15 \n\t"
- "mov %c[rcx](%3), %%rcx \n\t" /* kills %3 (rcx) */
-#else
- "mov %c[cr2](%3), %%eax \n\t"
- "mov %%eax, %%cr2 \n\t"
- "mov %c[rax](%3), %%eax \n\t"
- "mov %c[rbx](%3), %%ebx \n\t"
- "mov %c[rdx](%3), %%edx \n\t"
- "mov %c[rsi](%3), %%esi \n\t"
- "mov %c[rdi](%3), %%edi \n\t"
- "mov %c[rbp](%3), %%ebp \n\t"
- "mov %c[rcx](%3), %%ecx \n\t" /* kills %3 (ecx) */
-#endif
- /* Enter guest mode */
- "jne launched \n\t"
- ASM_VMX_VMLAUNCH "\n\t"
- "jmp kvm_vmx_return \n\t"
- "launched: " ASM_VMX_VMRESUME "\n\t"
- ".globl kvm_vmx_return \n\t"
- "kvm_vmx_return: "
- /* Save guest registers, load host registers, keep flags */
-#ifdef __x86_64__
- "xchg %3, 0(%%rsp) \n\t"
- "mov %%rax, %c[rax](%3) \n\t"
- "mov %%rbx, %c[rbx](%3) \n\t"
- "pushq 0(%%rsp); popq %c[rcx](%3) \n\t"
- "mov %%rdx, %c[rdx](%3) \n\t"
- "mov %%rsi, %c[rsi](%3) \n\t"
- "mov %%rdi, %c[rdi](%3) \n\t"
- "mov %%rbp, %c[rbp](%3) \n\t"
- "mov %%r8, %c[r8](%3) \n\t"
- "mov %%r9, %c[r9](%3) \n\t"
- "mov %%r10, %c[r10](%3) \n\t"
- "mov %%r11, %c[r11](%3) \n\t"
- "mov %%r12, %c[r12](%3) \n\t"
- "mov %%r13, %c[r13](%3) \n\t"
- "mov %%r14, %c[r14](%3) \n\t"
- "mov %%r15, %c[r15](%3) \n\t"
- "mov %%cr2, %%rax \n\t"
- "mov %%rax, %c[cr2](%3) \n\t"
- "mov 0(%%rsp), %3 \n\t"
-
- "pop %%rcx; pop %%r15; pop %%r14; pop %%r13; pop %%r12;"
- "pop %%r11; pop %%r10; pop %%r9; pop %%r8;"
- "pop %%rbp; pop %%rdi; pop %%rsi;"
- "pop %%rdx; pop %%rbx; pop %%rax \n\t"
-#else
- "xchg %3, 0(%%esp) \n\t"
- "mov %%eax, %c[rax](%3) \n\t"
- "mov %%ebx, %c[rbx](%3) \n\t"
- "pushl 0(%%esp); popl %c[rcx](%3) \n\t"
- "mov %%edx, %c[rdx](%3) \n\t"
- "mov %%esi, %c[rsi](%3) \n\t"
- "mov %%edi, %c[rdi](%3) \n\t"
- "mov %%ebp, %c[rbp](%3) \n\t"
- "mov %%cr2, %%eax \n\t"
- "mov %%eax, %c[cr2](%3) \n\t"
- "mov 0(%%esp), %3 \n\t"
-
- "pop %%ecx; popa \n\t"
-#endif
- "setbe %0 \n\t"
- "popf \n\t"
- : "=g" (fail)
- : "r"(vcpu->launched), "d"((unsigned long)HOST_RSP),
- "c"(vcpu),
- [rax]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RAX])),
- [rbx]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RBX])),
- [rcx]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RCX])),
- [rdx]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RDX])),
- [rsi]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RSI])),
- [rdi]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RDI])),
- [rbp]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RBP])),
-#ifdef __x86_64__
- [r8 ]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R8 ])),
- [r9 ]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R9 ])),
- [r10]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R10])),
- [r11]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R11])),
- [r12]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R12])),
- [r13]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R13])),
- [r14]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R14])),
- [r15]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R15])),
-#endif
- [cr2]"i"(offsetof(struct kvm_vcpu, cr2))
- : "cc", "memory" );
-
- ++kvm_stat.exits;
-
- save_msrs(vcpu->guest_msrs, NR_BAD_MSRS);
- load_msrs(vcpu->host_msrs, NR_BAD_MSRS);
-
- fx_save(vcpu->guest_fx_image);
- fx_restore(vcpu->host_fx_image);
-
-#ifndef __x86_64__
- asm ("mov %0, %%ds; mov %0, %%es" : : "r"(__USER_DS));
-#endif
-
- kvm_run->exit_type = 0;
- if (fail) {
- kvm_run->exit_type = KVM_EXIT_TYPE_FAIL_ENTRY;
- kvm_run->exit_reason = vmcs_read32(VM_INSTRUCTION_ERROR);
- } else {
- if (fs_gs_ldt_reload_needed) {
- load_ldt(ldt_sel);
- load_fs(fs_sel);
- /*
- * If we have to reload gs, we must take care to
- * preserve our gs base.
- */
- local_irq_disable();
- load_gs(gs_sel);
-#ifdef __x86_64__
- wrmsrl(MSR_GS_BASE, vmcs_readl(HOST_GS_BASE));
-#endif
- local_irq_enable();
-
- reload_tss();
- }
- vcpu->launched = 1;
- kvm_run->exit_type = KVM_EXIT_TYPE_VM_EXIT;
- if (kvm_handle_exit(kvm_run, vcpu)) {
- /* Give scheduler a change to reschedule. */
- vcpu_put(vcpu);
- if (signal_pending(current)) {
- ++kvm_stat.signal_exits;
- return -EINTR;
- }
- cond_resched();
- /* Cannot fail - no vcpu unplug yet. */
- vcpu_load(kvm, vcpu_slot(vcpu));
- goto again;
- }
- }
+ r = kvm_arch_ops->run(vcpu, kvm_run);
vcpu_put(vcpu);
- return 0;
+ return r;
}
static int kvm_dev_ioctl_get_regs(struct kvm *kvm, struct kvm_regs *regs)
Index: linux-2.6/drivers/kvm/kvm_vmx.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_vmx.h
+++ linux-2.6/drivers/kvm/kvm_vmx.h
@@ -17,4 +17,14 @@ static inline void vmcs_write64(unsigned
#endif
}
+#ifdef __x86_64__
+/*
+ * avoid save/load MSR_SYSCALL_MASK and MSR_LSTAR by std vt
+ * mechanism (cpu bug AA24)
+ */
+#define NR_BAD_MSRS 2
+#else
+#define NR_BAD_MSRS 0
+#endif
+
#endif
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -50,6 +50,22 @@ void set_efer(struct kvm_vcpu *vcpu, u64
#endif
+static void reload_tss(void)
+{
+#ifndef __x86_64__
+
+ /*
+ * VT restores TR but not its size. Useless.
+ */
+ struct descriptor_table gdt;
+ struct segment_descriptor *descs;
+
+ get_gdt(&gdt);
+ descs = (void *)gdt.base;
+ descs[GDT_ENTRY_TSS].type = 9; /* available TSS */
+ load_TR_desc();
+#endif
+}
/*
* Reads an msr value (of 'msr_index') into 'pdata'.
@@ -407,6 +423,314 @@ static void vmx_set_gdt(struct kvm_vcpu
vmcs_writel(GUEST_GDTR_BASE, dt->base);
}
+static void inject_rmode_irq(struct kvm_vcpu *vcpu, int irq)
+{
+ u16 ent[2];
+ u16 cs;
+ u16 ip;
+ unsigned long flags;
+ unsigned long ss_base = vmcs_readl(GUEST_SS_BASE);
+ u16 sp = vmcs_readl(GUEST_RSP);
+ u32 ss_limit = vmcs_read32(GUEST_SS_LIMIT);
+
+ if (sp > ss_limit || sp - 6 > sp) {
+ vcpu_printf(vcpu, "%s: #SS, rsp 0x%lx ss 0x%lx limit 0x%x\n",
+ __FUNCTION__,
+ vmcs_readl(GUEST_RSP),
+ vmcs_readl(GUEST_SS_BASE),
+ vmcs_read32(GUEST_SS_LIMIT));
+ return;
+ }
+
+ if (kvm_read_guest(vcpu, irq * sizeof(ent), sizeof(ent), &ent) !=
+ sizeof(ent)) {
+ vcpu_printf(vcpu, "%s: read guest err\n", __FUNCTION__);
+ return;
+ }
+
+ flags = vmcs_readl(GUEST_RFLAGS);
+ cs = vmcs_readl(GUEST_CS_BASE) >> 4;
+ ip = vmcs_readl(GUEST_RIP);
+
+
+ if (kvm_write_guest(vcpu, ss_base + sp - 2, 2, &flags) != 2 ||
+ kvm_write_guest(vcpu, ss_base + sp - 4, 2, &cs) != 2 ||
+ kvm_write_guest(vcpu, ss_base + sp - 6, 2, &ip) != 2) {
+ vcpu_printf(vcpu, "%s: write guest err\n", __FUNCTION__);
+ return;
+ }
+
+ vmcs_writel(GUEST_RFLAGS, flags &
+ ~( X86_EFLAGS_IF | X86_EFLAGS_AC | X86_EFLAGS_TF));
+ vmcs_write16(GUEST_CS_SELECTOR, ent[1]) ;
+ vmcs_writel(GUEST_CS_BASE, ent[1] << 4);
+ vmcs_writel(GUEST_RIP, ent[0]);
+ vmcs_writel(GUEST_RSP, (vmcs_readl(GUEST_RSP) & ~0xffff) | (sp - 6));
+}
+
+static void kvm_do_inject_irq(struct kvm_vcpu *vcpu)
+{
+ int word_index = __ffs(vcpu->irq_summary);
+ int bit_index = __ffs(vcpu->irq_pending[word_index]);
+ int irq = word_index * BITS_PER_LONG + bit_index;
+
+ clear_bit(bit_index, &vcpu->irq_pending[word_index]);
+ if (!vcpu->irq_pending[word_index])
+ clear_bit(word_index, &vcpu->irq_summary);
+
+ if (vcpu->rmode.active) {
+ inject_rmode_irq(vcpu, irq);
+ return;
+ }
+ vmcs_write32(VM_ENTRY_INTR_INFO_FIELD,
+ irq | INTR_TYPE_EXT_INTR | INTR_INFO_VALID_MASK);
+}
+
+static void kvm_try_inject_irq(struct kvm_vcpu *vcpu)
+{
+ if ((vmcs_readl(GUEST_RFLAGS) & X86_EFLAGS_IF)
+ && (vmcs_read32(GUEST_INTERRUPTIBILITY_INFO) & 3) == 0)
+ /*
+ * Interrupts enabled, and not blocked by sti or mov ss. Good.
+ */
+ kvm_do_inject_irq(vcpu);
+ else
+ /*
+ * Interrupts blocked. Wait for unblock.
+ */
+ vmcs_write32(CPU_BASED_VM_EXEC_CONTROL,
+ vmcs_read32(CPU_BASED_VM_EXEC_CONTROL)
+ | CPU_BASED_VIRTUAL_INTR_PENDING);
+}
+
+static void kvm_guest_debug_pre(struct kvm_vcpu *vcpu)
+{
+ struct kvm_guest_debug *dbg = &vcpu->guest_debug;
+
+ set_debugreg(dbg->bp[0], 0);
+ set_debugreg(dbg->bp[1], 1);
+ set_debugreg(dbg->bp[2], 2);
+ set_debugreg(dbg->bp[3], 3);
+
+ if (dbg->singlestep) {
+ unsigned long flags;
+
+ flags = vmcs_readl(GUEST_RFLAGS);
+ flags |= X86_EFLAGS_TF | X86_EFLAGS_RF;
+ vmcs_writel(GUEST_RFLAGS, flags);
+ }
+}
+
+static int vmx_vcpu_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
+{
+ u8 fail;
+ u16 fs_sel, gs_sel, ldt_sel;
+ int fs_gs_ldt_reload_needed;
+
+again:
+ /*
+ * Set host fs and gs selectors. Unfortunately, 22.2.3 does not
+ * allow segment selectors with cpl > 0 or ti == 1.
+ */
+ fs_sel = read_fs();
+ gs_sel = read_gs();
+ ldt_sel = read_ldt();
+ fs_gs_ldt_reload_needed = (fs_sel & 7) | (gs_sel & 7) | ldt_sel;
+ if (!fs_gs_ldt_reload_needed) {
+ vmcs_write16(HOST_FS_SELECTOR, fs_sel);
+ vmcs_write16(HOST_GS_SELECTOR, gs_sel);
+ } else {
+ vmcs_write16(HOST_FS_SELECTOR, 0);
+ vmcs_write16(HOST_GS_SELECTOR, 0);
+ }
+
+#ifdef __x86_64__
+ vmcs_writel(HOST_FS_BASE, read_msr(MSR_FS_BASE));
+ vmcs_writel(HOST_GS_BASE, read_msr(MSR_GS_BASE));
+#endif
+
+ if (vcpu->irq_summary &&
+ !(vmcs_read32(VM_ENTRY_INTR_INFO_FIELD) & INTR_INFO_VALID_MASK))
+ kvm_try_inject_irq(vcpu);
+
+ if (vcpu->guest_debug.enabled)
+ kvm_guest_debug_pre(vcpu);
+
+ fx_save(vcpu->host_fx_image);
+ fx_restore(vcpu->guest_fx_image);
+
+ save_msrs(vcpu->host_msrs, vcpu->nmsrs);
+ load_msrs(vcpu->guest_msrs, NR_BAD_MSRS);
+
+ asm (
+ /* Store host registers */
+ "pushf \n\t"
+#ifdef __x86_64__
+ "push %%rax; push %%rbx; push %%rdx;"
+ "push %%rsi; push %%rdi; push %%rbp;"
+ "push %%r8; push %%r9; push %%r10; push %%r11;"
+ "push %%r12; push %%r13; push %%r14; push %%r15;"
+ "push %%rcx \n\t"
+ ASM_VMX_VMWRITE_RSP_RDX "\n\t"
+#else
+ "pusha; push %%ecx \n\t"
+ ASM_VMX_VMWRITE_RSP_RDX "\n\t"
+#endif
+ /* Check if vmlaunch of vmresume is needed */
+ "cmp $0, %1 \n\t"
+ /* Load guest registers. Don't clobber flags. */
+#ifdef __x86_64__
+ "mov %c[cr2](%3), %%rax \n\t"
+ "mov %%rax, %%cr2 \n\t"
+ "mov %c[rax](%3), %%rax \n\t"
+ "mov %c[rbx](%3), %%rbx \n\t"
+ "mov %c[rdx](%3), %%rdx \n\t"
+ "mov %c[rsi](%3), %%rsi \n\t"
+ "mov %c[rdi](%3), %%rdi \n\t"
+ "mov %c[rbp](%3), %%rbp \n\t"
+ "mov %c[r8](%3), %%r8 \n\t"
+ "mov %c[r9](%3), %%r9 \n\t"
+ "mov %c[r10](%3), %%r10 \n\t"
+ "mov %c[r11](%3), %%r11 \n\t"
+ "mov %c[r12](%3), %%r12 \n\t"
+ "mov %c[r13](%3), %%r13 \n\t"
+ "mov %c[r14](%3), %%r14 \n\t"
+ "mov %c[r15](%3), %%r15 \n\t"
+ "mov %c[rcx](%3), %%rcx \n\t" /* kills %3 (rcx) */
+#else
+ "mov %c[cr2](%3), %%eax \n\t"
+ "mov %%eax, %%cr2 \n\t"
+ "mov %c[rax](%3), %%eax \n\t"
+ "mov %c[rbx](%3), %%ebx \n\t"
+ "mov %c[rdx](%3), %%edx \n\t"
+ "mov %c[rsi](%3), %%esi \n\t"
+ "mov %c[rdi](%3), %%edi \n\t"
+ "mov %c[rbp](%3), %%ebp \n\t"
+ "mov %c[rcx](%3), %%ecx \n\t" /* kills %3 (ecx) */
+#endif
+ /* Enter guest mode */
+ "jne launched \n\t"
+ ASM_VMX_VMLAUNCH "\n\t"
+ "jmp kvm_vmx_return \n\t"
+ "launched: " ASM_VMX_VMRESUME "\n\t"
+ ".globl kvm_vmx_return \n\t"
+ "kvm_vmx_return: "
+ /* Save guest registers, load host registers, keep flags */
+#ifdef __x86_64__
+ "xchg %3, 0(%%rsp) \n\t"
+ "mov %%rax, %c[rax](%3) \n\t"
+ "mov %%rbx, %c[rbx](%3) \n\t"
+ "pushq 0(%%rsp); popq %c[rcx](%3) \n\t"
+ "mov %%rdx, %c[rdx](%3) \n\t"
+ "mov %%rsi, %c[rsi](%3) \n\t"
+ "mov %%rdi, %c[rdi](%3) \n\t"
+ "mov %%rbp, %c[rbp](%3) \n\t"
+ "mov %%r8, %c[r8](%3) \n\t"
+ "mov %%r9, %c[r9](%3) \n\t"
+ "mov %%r10, %c[r10](%3) \n\t"
+ "mov %%r11, %c[r11](%3) \n\t"
+ "mov %%r12, %c[r12](%3) \n\t"
+ "mov %%r13, %c[r13](%3) \n\t"
+ "mov %%r14, %c[r14](%3) \n\t"
+ "mov %%r15, %c[r15](%3) \n\t"
+ "mov %%cr2, %%rax \n\t"
+ "mov %%rax, %c[cr2](%3) \n\t"
+ "mov 0(%%rsp), %3 \n\t"
+
+ "pop %%rcx; pop %%r15; pop %%r14; pop %%r13; pop %%r12;"
+ "pop %%r11; pop %%r10; pop %%r9; pop %%r8;"
+ "pop %%rbp; pop %%rdi; pop %%rsi;"
+ "pop %%rdx; pop %%rbx; pop %%rax \n\t"
+#else
+ "xchg %3, 0(%%esp) \n\t"
+ "mov %%eax, %c[rax](%3) \n\t"
+ "mov %%ebx, %c[rbx](%3) \n\t"
+ "pushl 0(%%esp); popl %c[rcx](%3) \n\t"
+ "mov %%edx, %c[rdx](%3) \n\t"
+ "mov %%esi, %c[rsi](%3) \n\t"
+ "mov %%edi, %c[rdi](%3) \n\t"
+ "mov %%ebp, %c[rbp](%3) \n\t"
+ "mov %%cr2, %%eax \n\t"
+ "mov %%eax, %c[cr2](%3) \n\t"
+ "mov 0(%%esp), %3 \n\t"
+
+ "pop %%ecx; popa \n\t"
+#endif
+ "setbe %0 \n\t"
+ "popf \n\t"
+ : "=g" (fail)
+ : "r"(vcpu->launched), "d"((unsigned long)HOST_RSP),
+ "c"(vcpu),
+ [rax]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RAX])),
+ [rbx]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RBX])),
+ [rcx]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RCX])),
+ [rdx]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RDX])),
+ [rsi]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RSI])),
+ [rdi]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RDI])),
+ [rbp]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_RBP])),
+#ifdef __x86_64__
+ [r8 ]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R8 ])),
+ [r9 ]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R9 ])),
+ [r10]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R10])),
+ [r11]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R11])),
+ [r12]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R12])),
+ [r13]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R13])),
+ [r14]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R14])),
+ [r15]"i"(offsetof(struct kvm_vcpu, regs[VCPU_REGS_R15])),
+#endif
+ [cr2]"i"(offsetof(struct kvm_vcpu, cr2))
+ : "cc", "memory" );
+
+ ++kvm_stat.exits;
+
+ save_msrs(vcpu->guest_msrs, NR_BAD_MSRS);
+ load_msrs(vcpu->host_msrs, NR_BAD_MSRS);
+
+ fx_save(vcpu->guest_fx_image);
+ fx_restore(vcpu->host_fx_image);
+
+#ifndef __x86_64__
+ asm ("mov %0, %%ds; mov %0, %%es" : : "r"(__USER_DS));
+#endif
+
+ kvm_run->exit_type = 0;
+ if (fail) {
+ kvm_run->exit_type = KVM_EXIT_TYPE_FAIL_ENTRY;
+ kvm_run->exit_reason = vmcs_read32(VM_INSTRUCTION_ERROR);
+ } else {
+ if (fs_gs_ldt_reload_needed) {
+ load_ldt(ldt_sel);
+ load_fs(fs_sel);
+ /*
+ * If we have to reload gs, we must take care to
+ * preserve our gs base.
+ */
+ local_irq_disable();
+ load_gs(gs_sel);
+#ifdef __x86_64__
+ wrmsrl(MSR_GS_BASE, vmcs_readl(HOST_GS_BASE));
+#endif
+ local_irq_enable();
+
+ reload_tss();
+ }
+ vcpu->launched = 1;
+ kvm_run->exit_type = KVM_EXIT_TYPE_VM_EXIT;
+ if (kvm_handle_exit(kvm_run, vcpu)) {
+ /* Give scheduler a change to reschedule. */
+ if (signal_pending(current)) {
+ ++kvm_stat.signal_exits;
+ return -EINTR;
+ }
+ kvm_resched(vcpu);
+ goto again;
+ }
+ }
+ return 0;
+}
+
+extern asmlinkage void kvm_vmx_return(void);
+
static struct kvm_arch_ops vmx_arch_ops = {
.cpu_has_kvm_support = cpu_has_kvm_support,
.disabled_by_bios = vmx_disabled_by_bios,
@@ -427,6 +751,9 @@ static struct kvm_arch_ops vmx_arch_ops
.set_gdt = vmx_set_gdt,
.cache_regs = vcpu_load_rsp_rip,
.decache_regs = vcpu_put_rsp_rip,
+
+ .run = vmx_vcpu_run,
+ .vmx_return = (unsigned long)kvm_vmx_return,
};
static int __init vmx_init(void)
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 15/38] KVM: Move the vmx exit handlers to vmx.c
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (13 preceding siblings ...)
2006-11-27 12:24 ` [PATCH 14/38] KVM: Make the vcpu execution loop " Avi Kivity
@ 2006-11-27 12:25 ` Avi Kivity
2006-11-27 12:26 ` [PATCH 16/38] KVM: Make vcpu_setup() an arch operation Avi Kivity
` (22 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:25 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -266,6 +266,7 @@ struct kvm_arch_ops {
void (*decache_regs)(struct kvm_vcpu *vcpu);
int (*run)(struct kvm_vcpu *vcpu, struct kvm_run *run);
+ void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
unsigned long vmx_return; /* temporary hack */
};
@@ -300,6 +301,14 @@ static inline struct page *gfn_to_page(s
struct kvm_memory_slot *gfn_to_memslot(struct kvm *kvm, gfn_t gfn);
void mark_page_dirty(struct kvm *kvm, gfn_t gfn);
+enum emulation_result {
+ EMULATE_DONE, /* no further processing */
+ EMULATE_DO_MMIO, /* kvm_run filled with mmio request */
+ EMULATE_FAIL, /* can't emulate this instruction */
+};
+
+int emulate_instruction(struct kvm_vcpu *vcpu, struct kvm_run *run,
+ unsigned long cr2, u16 error_code);
void realmode_lgdt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
void realmode_lidt(struct kvm_vcpu *vcpu, u16 size, unsigned long address);
void realmode_lmsw(struct kvm_vcpu *vcpu, unsigned long msw,
@@ -309,10 +318,22 @@ unsigned long realmode_get_cr(struct kvm
void realmode_set_cr(struct kvm_vcpu *vcpu, int cr, unsigned long value,
unsigned long *rflags);
+void set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
+void set_cr3(struct kvm_vcpu *vcpu, unsigned long cr0);
+void set_cr4(struct kvm_vcpu *vcpu, unsigned long cr0);
+void set_cr8(struct kvm_vcpu *vcpu, unsigned long cr0);
+void lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
+
+void inject_gp(struct kvm_vcpu *vcpu);
+
+#ifdef __x86_64__
+void set_efer(struct kvm_vcpu *vcpu, u64 efer);
+#endif
+
+
void load_msrs(struct vmx_msr_entry *e, int n);
void save_msrs(struct vmx_msr_entry *e, int n);
void kvm_resched(struct kvm_vcpu *vcpu);
-int kvm_handle_exit(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu);
int kvm_read_guest(struct kvm_vcpu *vcpu,
gva_t addr,
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -533,7 +533,7 @@ void vmcs_writel(unsigned long field, un
}
EXPORT_SYMBOL_GPL(vmcs_writel);
-static void inject_gp(struct kvm_vcpu *vcpu)
+void inject_gp(struct kvm_vcpu *vcpu)
{
printk(KERN_DEBUG "inject_general_protection: rip 0x%lx\n",
vmcs_readl(GUEST_RIP));
@@ -544,6 +544,7 @@ static void inject_gp(struct kvm_vcpu *v
INTR_INFO_DELIEVER_CODE_MASK |
INTR_INFO_VALID_MASK);
}
+EXPORT_SYMBOL_GPL(inject_gp);
/*
* reads and returns guest's timestamp counter "register"
@@ -821,7 +822,7 @@ static int pdptrs_have_reserved_bits_set
return i != 4;
}
-static void set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
+void set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
{
if (cr0 & CR0_RESEVED_BITS) {
printk(KERN_DEBUG "set_cr0: 0x%lx #GP, reserved bits 0x%lx\n",
@@ -879,8 +880,9 @@ static void set_cr0(struct kvm_vcpu *vcp
spin_unlock(&vcpu->kvm->lock);
return;
}
+EXPORT_SYMBOL_GPL(set_cr0);
-static void lmsw(struct kvm_vcpu *vcpu, unsigned long msw)
+void lmsw(struct kvm_vcpu *vcpu, unsigned long msw)
{
unsigned long cr0 = vcpu->cr0;
@@ -895,6 +897,7 @@ static void lmsw(struct kvm_vcpu *vcpu,
| (msw & LMSW_GUEST_MASK));
vcpu->cr0 = (vcpu->cr0 & ~0xfffful) | msw;
}
+EXPORT_SYMBOL_GPL(lmsw);
static void __set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
{
@@ -904,7 +907,7 @@ static void __set_cr4(struct kvm_vcpu *v
vcpu->cr4 = cr4;
}
-static void set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
+void set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
{
if (cr4 & CR4_RESEVED_BITS) {
printk(KERN_DEBUG "set_cr4: #GP, reserved bits\n");
@@ -935,8 +938,9 @@ static void set_cr4(struct kvm_vcpu *vcp
kvm_mmu_reset_context(vcpu);
spin_unlock(&vcpu->kvm->lock);
}
+EXPORT_SYMBOL_GPL(set_cr4);
-static void set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3)
+void set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3)
{
if (is_long_mode()) {
if ( cr3 & CR3_L_MODE_RESEVED_BITS) {
@@ -964,8 +968,9 @@ static void set_cr3(struct kvm_vcpu *vcp
vcpu->mmu.new_cr3(vcpu);
spin_unlock(&vcpu->kvm->lock);
}
+EXPORT_SYMBOL_GPL(set_cr3);
-static void set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8)
+void set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8)
{
if ( cr8 & CR8_RESEVED_BITS) {
printk(KERN_DEBUG "set_cr8: #GP, reserved bits 0x%lx\n", cr8);
@@ -974,6 +979,7 @@ static void set_cr8(struct kvm_vcpu *vcp
}
vcpu->cr8 = cr8;
}
+EXPORT_SYMBOL_GPL(set_cr8);
static u32 get_rdx_init_val(void)
{
@@ -1534,25 +1540,6 @@ void mark_page_dirty(struct kvm *kvm, gf
}
}
-static void skip_emulated_instruction(struct kvm_vcpu *vcpu)
-{
- unsigned long rip;
- u32 interruptibility;
-
- rip = vmcs_readl(GUEST_RIP);
- rip += vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
- vmcs_writel(GUEST_RIP, rip);
-
- /*
- * We emulated an instruction, so temporary interrupt blocking
- * should be removed, if set.
- */
- interruptibility = vmcs_read32(GUEST_INTERRUPTIBILITY_INFO);
- if (interruptibility & 3)
- vmcs_write32(GUEST_INTERRUPTIBILITY_INFO,
- interruptibility & ~3);
-}
-
static int emulator_read_std(unsigned long addr,
unsigned long *val,
unsigned int bytes,
@@ -1694,16 +1681,10 @@ struct x86_emulate_ops emulate_ops = {
.cmpxchg_emulated = emulator_cmpxchg_emulated,
};
-enum emulation_result {
- EMULATE_DONE, /* no further processing */
- EMULATE_DO_MMIO, /* kvm_run filled with mmio request */
- EMULATE_FAIL, /* can't emulate this instruction */
-};
-
-static int emulate_instruction(struct kvm_vcpu *vcpu,
- struct kvm_run *run,
- unsigned long cr2,
- u16 error_code)
+int emulate_instruction(struct kvm_vcpu *vcpu,
+ struct kvm_run *run,
+ unsigned long cr2,
+ u16 error_code)
{
struct x86_emulate_ctxt emulate_ctxt;
int r;
@@ -1762,6 +1743,7 @@ static int emulate_instruction(struct kv
return EMULATE_DONE;
}
+EXPORT_SYMBOL_GPL(emulate_instruction);
static u64 mk_cr_64(u64 curr_cr, u32 new_val)
{
@@ -1826,298 +1808,6 @@ void realmode_set_cr(struct kvm_vcpu *vc
}
}
-static int handle_rmode_exception(struct kvm_vcpu *vcpu,
- int vec, u32 err_code)
-{
- if (!vcpu->rmode.active)
- return 0;
-
- if (vec == GP_VECTOR && err_code == 0)
- if (emulate_instruction(vcpu, 0, 0, 0) == EMULATE_DONE)
- return 1;
- return 0;
-}
-
-static int handle_exception(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
-{
- u32 intr_info, error_code;
- unsigned long cr2, rip;
- u32 vect_info;
- enum emulation_result er;
-
- vect_info = vmcs_read32(IDT_VECTORING_INFO_FIELD);
- intr_info = vmcs_read32(VM_EXIT_INTR_INFO);
-
- if ((vect_info & VECTORING_INFO_VALID_MASK) &&
- !is_page_fault(intr_info)) {
- printk(KERN_ERR "%s: unexpected, vectoring info 0x%x "
- "intr info 0x%x\n", __FUNCTION__, vect_info, intr_info);
- }
-
- if (is_external_interrupt(vect_info)) {
- int irq = vect_info & VECTORING_INFO_VECTOR_MASK;
- set_bit(irq, vcpu->irq_pending);
- set_bit(irq / BITS_PER_LONG, &vcpu->irq_summary);
- }
-
- if ((intr_info & INTR_INFO_INTR_TYPE_MASK) == 0x200) { /* nmi */
- asm ("int $2");
- return 1;
- }
- error_code = 0;
- rip = vmcs_readl(GUEST_RIP);
- if (intr_info & INTR_INFO_DELIEVER_CODE_MASK)
- error_code = vmcs_read32(VM_EXIT_INTR_ERROR_CODE);
- if (is_page_fault(intr_info)) {
- cr2 = vmcs_readl(EXIT_QUALIFICATION);
-
- spin_lock(&vcpu->kvm->lock);
- if (!vcpu->mmu.page_fault(vcpu, cr2, error_code)) {
- spin_unlock(&vcpu->kvm->lock);
- return 1;
- }
-
- er = emulate_instruction(vcpu, kvm_run, cr2, error_code);
- spin_unlock(&vcpu->kvm->lock);
-
- switch (er) {
- case EMULATE_DONE:
- return 1;
- case EMULATE_DO_MMIO:
- ++kvm_stat.mmio_exits;
- kvm_run->exit_reason = KVM_EXIT_MMIO;
- return 0;
- case EMULATE_FAIL:
- vcpu_printf(vcpu, "%s: emulate fail\n", __FUNCTION__);
- break;
- default:
- BUG();
- }
- }
-
- if (vcpu->rmode.active &&
- handle_rmode_exception(vcpu, intr_info & INTR_INFO_VECTOR_MASK,
- error_code))
- return 1;
-
- if ((intr_info & (INTR_INFO_INTR_TYPE_MASK | INTR_INFO_VECTOR_MASK)) == (INTR_TYPE_EXCEPTION | 1)) {
- kvm_run->exit_reason = KVM_EXIT_DEBUG;
- return 0;
- }
- kvm_run->exit_reason = KVM_EXIT_EXCEPTION;
- kvm_run->ex.exception = intr_info & INTR_INFO_VECTOR_MASK;
- kvm_run->ex.error_code = error_code;
- return 0;
-}
-
-static int handle_external_interrupt(struct kvm_vcpu *vcpu,
- struct kvm_run *kvm_run)
-{
- ++kvm_stat.irq_exits;
- return 1;
-}
-
-
-static int get_io_count(struct kvm_vcpu *vcpu, u64 *count)
-{
- u64 inst;
- gva_t rip;
- int countr_size;
- int i, n;
-
- if ((vmcs_readl(GUEST_RFLAGS) & X86_EFLAGS_VM)) {
- countr_size = 2;
- } else {
- u32 cs_ar = vmcs_read32(GUEST_CS_AR_BYTES);
-
- countr_size = (cs_ar & AR_L_MASK) ? 8:
- (cs_ar & AR_DB_MASK) ? 4: 2;
- }
-
- rip = vmcs_readl(GUEST_RIP);
- if (countr_size != 8)
- rip += vmcs_readl(GUEST_CS_BASE);
-
- n = kvm_read_guest(vcpu, rip, sizeof(inst), &inst);
-
- for (i = 0; i < n; i++) {
- switch (((u8*)&inst)[i]) {
- case 0xf0:
- case 0xf2:
- case 0xf3:
- case 0x2e:
- case 0x36:
- case 0x3e:
- case 0x26:
- case 0x64:
- case 0x65:
- case 0x66:
- break;
- case 0x67:
- countr_size = (countr_size == 2) ? 4: (countr_size >> 1);
- default:
- goto done;
- }
- }
- return 0;
-done:
- countr_size *= 8;
- *count = vcpu->regs[VCPU_REGS_RCX] & (~0ULL >> (64 - countr_size));
- return 1;
-}
-
-static int handle_io(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
-{
- u64 exit_qualification;
-
- ++kvm_stat.io_exits;
- exit_qualification = vmcs_read64(EXIT_QUALIFICATION);
- kvm_run->exit_reason = KVM_EXIT_IO;
- if (exit_qualification & 8)
- kvm_run->io.direction = KVM_EXIT_IO_IN;
- else
- kvm_run->io.direction = KVM_EXIT_IO_OUT;
- kvm_run->io.size = (exit_qualification & 7) + 1;
- kvm_run->io.string = (exit_qualification & 16) != 0;
- kvm_run->io.string_down
- = (vmcs_readl(GUEST_RFLAGS) & X86_EFLAGS_DF) != 0;
- kvm_run->io.rep = (exit_qualification & 32) != 0;
- kvm_run->io.port = exit_qualification >> 16;
- if (kvm_run->io.string) {
- if (!get_io_count(vcpu, &kvm_run->io.count))
- return 1;
- kvm_run->io.address = vmcs_readl(GUEST_LINEAR_ADDRESS);
- } else
- kvm_run->io.value = vcpu->regs[VCPU_REGS_RAX]; /* rax */
- return 0;
-}
-
-static int handle_invlpg(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
-{
- u64 address = vmcs_read64(EXIT_QUALIFICATION);
- int instruction_length = vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
- spin_lock(&vcpu->kvm->lock);
- vcpu->mmu.inval_page(vcpu, address);
- spin_unlock(&vcpu->kvm->lock);
- vmcs_writel(GUEST_RIP, vmcs_readl(GUEST_RIP) + instruction_length);
- return 1;
-}
-
-static int handle_cr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
-{
- u64 exit_qualification;
- int cr;
- int reg;
-
-#ifdef KVM_DEBUG
- if (guest_cpl() != 0) {
- vcpu_printf(vcpu, "%s: not supervisor\n", __FUNCTION__);
- inject_gp(vcpu);
- return 1;
- }
-#endif
-
- exit_qualification = vmcs_read64(EXIT_QUALIFICATION);
- cr = exit_qualification & 15;
- reg = (exit_qualification >> 8) & 15;
- switch ((exit_qualification >> 4) & 3) {
- case 0: /* mov to cr */
- switch (cr) {
- case 0:
- kvm_arch_ops->cache_regs(vcpu);
- set_cr0(vcpu, vcpu->regs[reg]);
- skip_emulated_instruction(vcpu);
- return 1;
- case 3:
- kvm_arch_ops->cache_regs(vcpu);
- set_cr3(vcpu, vcpu->regs[reg]);
- skip_emulated_instruction(vcpu);
- return 1;
- case 4:
- kvm_arch_ops->cache_regs(vcpu);
- set_cr4(vcpu, vcpu->regs[reg]);
- skip_emulated_instruction(vcpu);
- return 1;
- case 8:
- kvm_arch_ops->cache_regs(vcpu);
- set_cr8(vcpu, vcpu->regs[reg]);
- skip_emulated_instruction(vcpu);
- return 1;
- };
- break;
- case 1: /*mov from cr*/
- switch (cr) {
- case 3:
- kvm_arch_ops->cache_regs(vcpu);
- vcpu->regs[reg] = vcpu->cr3;
- kvm_arch_ops->decache_regs(vcpu);
- skip_emulated_instruction(vcpu);
- return 1;
- case 8:
- printk(KERN_DEBUG "handle_cr: read CR8 "
- "cpu erratum AA15\n");
- kvm_arch_ops->cache_regs(vcpu);
- vcpu->regs[reg] = vcpu->cr8;
- kvm_arch_ops->decache_regs(vcpu);
- skip_emulated_instruction(vcpu);
- return 1;
- }
- break;
- case 3: /* lmsw */
- lmsw(vcpu, (exit_qualification >> LMSW_SOURCE_DATA_SHIFT) & 0x0f);
-
- skip_emulated_instruction(vcpu);
- return 1;
- default:
- break;
- }
- kvm_run->exit_reason = 0;
- printk(KERN_ERR "kvm: unhandled control register: op %d cr %d\n",
- (int)(exit_qualification >> 4) & 3, cr);
- return 0;
-}
-
-static int handle_dr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
-{
- u64 exit_qualification;
- unsigned long val;
- int dr, reg;
-
- /*
- * FIXME: this code assumes the host is debugging the guest.
- * need to deal with guest debugging itself too.
- */
- exit_qualification = vmcs_read64(EXIT_QUALIFICATION);
- dr = exit_qualification & 7;
- reg = (exit_qualification >> 8) & 15;
- kvm_arch_ops->cache_regs(vcpu);
- if (exit_qualification & 16) {
- /* mov from dr */
- switch (dr) {
- case 6:
- val = 0xffff0ff0;
- break;
- case 7:
- val = 0x400;
- break;
- default:
- val = 0;
- }
- vcpu->regs[reg] = val;
- } else {
- /* mov to dr */
- }
- kvm_arch_ops->decache_regs(vcpu);
- skip_emulated_instruction(vcpu);
- return 1;
-}
-
-static int handle_cpuid(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
-{
- kvm_run->exit_reason = KVM_EXIT_CPUID;
- return 0;
-}
-
/*
* Reads an msr value (of 'msr_index') into 'pdata'.
* Returns 0 on success, non-0 otherwise.
@@ -2128,23 +1818,6 @@ static int get_msr(struct kvm_vcpu *vcpu
return kvm_arch_ops->get_msr(vcpu, msr_index, pdata);
}
-static int handle_rdmsr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
-{
- u32 ecx = vcpu->regs[VCPU_REGS_RCX];
- u64 data;
-
- if (get_msr(vcpu, ecx, &data)) {
- inject_gp(vcpu);
- return 1;
- }
-
- /* FIXME: handling of bits 32:63 of rax, rdx */
- vcpu->regs[VCPU_REGS_RAX] = data & -1u;
- vcpu->regs[VCPU_REGS_RDX] = (data >> 32) & -1u;
- skip_emulated_instruction(vcpu);
- return 1;
-}
-
#ifdef __x86_64__
void set_efer(struct kvm_vcpu *vcpu, u64 efer)
@@ -2175,7 +1848,6 @@ void set_efer(struct kvm_vcpu *vcpu, u64
if (!(efer & EFER_LMA))
efer &= ~EFER_LME;
msr->data = efer;
- skip_emulated_instruction(vcpu);
}
EXPORT_SYMBOL_GPL(set_efer);
@@ -2191,90 +1863,6 @@ static int set_msr(struct kvm_vcpu *vcpu
return kvm_arch_ops->set_msr(vcpu, msr_index, data);
}
-static int handle_wrmsr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
-{
- u32 ecx = vcpu->regs[VCPU_REGS_RCX];
- u64 data = (vcpu->regs[VCPU_REGS_RAX] & -1u)
- | ((u64)(vcpu->regs[VCPU_REGS_RDX] & -1u) << 32);
-
- if (set_msr(vcpu, ecx, data) != 0) {
- inject_gp(vcpu);
- return 1;
- }
-
- if (ecx != MSR_EFER)
- skip_emulated_instruction(vcpu);
- return 1;
-}
-
-static int handle_interrupt_window(struct kvm_vcpu *vcpu,
- struct kvm_run *kvm_run)
-{
- /* Turn off interrupt window reporting. */
- vmcs_write32(CPU_BASED_VM_EXEC_CONTROL,
- vmcs_read32(CPU_BASED_VM_EXEC_CONTROL)
- & ~CPU_BASED_VIRTUAL_INTR_PENDING);
- return 1;
-}
-
-static int handle_halt(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
-{
- skip_emulated_instruction(vcpu);
- if (vcpu->irq_summary && (vmcs_readl(GUEST_RFLAGS) & X86_EFLAGS_IF))
- return 1;
-
- kvm_run->exit_reason = KVM_EXIT_HLT;
- return 0;
-}
-
-/*
- * The exit handlers return 1 if the exit was handled fully and guest execution
- * may resume. Otherwise they set the kvm_run parameter to indicate what needs
- * to be done to userspace and return 0.
- */
-static int (*kvm_vmx_exit_handlers[])(struct kvm_vcpu *vcpu,
- struct kvm_run *kvm_run) = {
- [EXIT_REASON_EXCEPTION_NMI] = handle_exception,
- [EXIT_REASON_EXTERNAL_INTERRUPT] = handle_external_interrupt,
- [EXIT_REASON_IO_INSTRUCTION] = handle_io,
- [EXIT_REASON_INVLPG] = handle_invlpg,
- [EXIT_REASON_CR_ACCESS] = handle_cr,
- [EXIT_REASON_DR_ACCESS] = handle_dr,
- [EXIT_REASON_CPUID] = handle_cpuid,
- [EXIT_REASON_MSR_READ] = handle_rdmsr,
- [EXIT_REASON_MSR_WRITE] = handle_wrmsr,
- [EXIT_REASON_PENDING_INTERRUPT] = handle_interrupt_window,
- [EXIT_REASON_HLT] = handle_halt,
-};
-
-static const int kvm_vmx_max_exit_handlers =
- sizeof(kvm_vmx_exit_handlers) / sizeof(*kvm_vmx_exit_handlers);
-
-/*
- * The guest has exited. See if we can fix it or if we need userspace
- * assistance.
- */
-int kvm_handle_exit(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
-{
- u32 vectoring_info = vmcs_read32(IDT_VECTORING_INFO_FIELD);
- u32 exit_reason = vmcs_read32(VM_EXIT_REASON);
-
- if ( (vectoring_info & VECTORING_INFO_VALID_MASK) &&
- exit_reason != EXIT_REASON_EXCEPTION_NMI )
- printk(KERN_WARNING "%s: unexpected, valid vectoring info and "
- "exit reason is 0x%x\n", __FUNCTION__, exit_reason);
- kvm_run->instruction_length = vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
- if (exit_reason < kvm_vmx_max_exit_handlers
- && kvm_vmx_exit_handlers[exit_reason])
- return kvm_vmx_exit_handlers[exit_reason](vcpu, kvm_run);
- else {
- kvm_run->exit_reason = KVM_EXIT_UNKNOWN;
- kvm_run->hw.hardware_exit_reason = exit_reason;
- }
- return 0;
-}
-EXPORT_SYMBOL_GPL(kvm_handle_exit);
-
void kvm_resched(struct kvm_vcpu *vcpu)
{
vcpu_put(vcpu);
@@ -2315,7 +1903,7 @@ static int kvm_dev_ioctl_run(struct kvm
return -ENOENT;
if (kvm_run->emulated) {
- skip_emulated_instruction(vcpu);
+ kvm_arch_ops->skip_emulated_instruction(vcpu);
kvm_run->emulated = 0;
}
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -44,11 +44,24 @@ u64 guest_read_tsc(void);
void guest_write_tsc(u64 guest_tsc);
struct vmx_msr_entry *find_msr_entry(struct kvm_vcpu *vcpu, u32 msr);
-#ifdef __x86_64__
+static void skip_emulated_instruction(struct kvm_vcpu *vcpu)
+{
+ unsigned long rip;
+ u32 interruptibility;
-void set_efer(struct kvm_vcpu *vcpu, u64 efer);
+ rip = vmcs_readl(GUEST_RIP);
+ rip += vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
+ vmcs_writel(GUEST_RIP, rip);
-#endif
+ /*
+ * We emulated an instruction, so temporary interrupt blocking
+ * should be removed, if set.
+ */
+ interruptibility = vmcs_read32(GUEST_INTERRUPTIBILITY_INFO);
+ if (interruptibility & 3)
+ vmcs_write32(GUEST_INTERRUPTIBILITY_INFO,
+ interruptibility & ~3);
+}
static void reload_tss(void)
{
@@ -521,6 +534,397 @@ static void kvm_guest_debug_pre(struct k
}
}
+static int handle_rmode_exception(struct kvm_vcpu *vcpu,
+ int vec, u32 err_code)
+{
+ if (!vcpu->rmode.active)
+ return 0;
+
+ if (vec == GP_VECTOR && err_code == 0)
+ if (emulate_instruction(vcpu, 0, 0, 0) == EMULATE_DONE)
+ return 1;
+ return 0;
+}
+
+static int handle_exception(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
+{
+ u32 intr_info, error_code;
+ unsigned long cr2, rip;
+ u32 vect_info;
+ enum emulation_result er;
+
+ vect_info = vmcs_read32(IDT_VECTORING_INFO_FIELD);
+ intr_info = vmcs_read32(VM_EXIT_INTR_INFO);
+
+ if ((vect_info & VECTORING_INFO_VALID_MASK) &&
+ !is_page_fault(intr_info)) {
+ printk(KERN_ERR "%s: unexpected, vectoring info 0x%x "
+ "intr info 0x%x\n", __FUNCTION__, vect_info, intr_info);
+ }
+
+ if (is_external_interrupt(vect_info)) {
+ int irq = vect_info & VECTORING_INFO_VECTOR_MASK;
+ set_bit(irq, vcpu->irq_pending);
+ set_bit(irq / BITS_PER_LONG, &vcpu->irq_summary);
+ }
+
+ if ((intr_info & INTR_INFO_INTR_TYPE_MASK) == 0x200) { /* nmi */
+ asm ("int $2");
+ return 1;
+ }
+ error_code = 0;
+ rip = vmcs_readl(GUEST_RIP);
+ if (intr_info & INTR_INFO_DELIEVER_CODE_MASK)
+ error_code = vmcs_read32(VM_EXIT_INTR_ERROR_CODE);
+ if (is_page_fault(intr_info)) {
+ cr2 = vmcs_readl(EXIT_QUALIFICATION);
+
+ spin_lock(&vcpu->kvm->lock);
+ if (!vcpu->mmu.page_fault(vcpu, cr2, error_code)) {
+ spin_unlock(&vcpu->kvm->lock);
+ return 1;
+ }
+
+ er = emulate_instruction(vcpu, kvm_run, cr2, error_code);
+ spin_unlock(&vcpu->kvm->lock);
+
+ switch (er) {
+ case EMULATE_DONE:
+ return 1;
+ case EMULATE_DO_MMIO:
+ ++kvm_stat.mmio_exits;
+ kvm_run->exit_reason = KVM_EXIT_MMIO;
+ return 0;
+ case EMULATE_FAIL:
+ vcpu_printf(vcpu, "%s: emulate fail\n", __FUNCTION__);
+ break;
+ default:
+ BUG();
+ }
+ }
+
+ if (vcpu->rmode.active &&
+ handle_rmode_exception(vcpu, intr_info & INTR_INFO_VECTOR_MASK,
+ error_code))
+ return 1;
+
+ if ((intr_info & (INTR_INFO_INTR_TYPE_MASK | INTR_INFO_VECTOR_MASK)) == (INTR_TYPE_EXCEPTION | 1)) {
+ kvm_run->exit_reason = KVM_EXIT_DEBUG;
+ return 0;
+ }
+ kvm_run->exit_reason = KVM_EXIT_EXCEPTION;
+ kvm_run->ex.exception = intr_info & INTR_INFO_VECTOR_MASK;
+ kvm_run->ex.error_code = error_code;
+ return 0;
+}
+
+static int handle_external_interrupt(struct kvm_vcpu *vcpu,
+ struct kvm_run *kvm_run)
+{
+ ++kvm_stat.irq_exits;
+ return 1;
+}
+
+
+static int get_io_count(struct kvm_vcpu *vcpu, u64 *count)
+{
+ u64 inst;
+ gva_t rip;
+ int countr_size;
+ int i, n;
+
+ if ((vmcs_readl(GUEST_RFLAGS) & X86_EFLAGS_VM)) {
+ countr_size = 2;
+ } else {
+ u32 cs_ar = vmcs_read32(GUEST_CS_AR_BYTES);
+
+ countr_size = (cs_ar & AR_L_MASK) ? 8:
+ (cs_ar & AR_DB_MASK) ? 4: 2;
+ }
+
+ rip = vmcs_readl(GUEST_RIP);
+ if (countr_size != 8)
+ rip += vmcs_readl(GUEST_CS_BASE);
+
+ n = kvm_read_guest(vcpu, rip, sizeof(inst), &inst);
+
+ for (i = 0; i < n; i++) {
+ switch (((u8*)&inst)[i]) {
+ case 0xf0:
+ case 0xf2:
+ case 0xf3:
+ case 0x2e:
+ case 0x36:
+ case 0x3e:
+ case 0x26:
+ case 0x64:
+ case 0x65:
+ case 0x66:
+ break;
+ case 0x67:
+ countr_size = (countr_size == 2) ? 4: (countr_size >> 1);
+ default:
+ goto done;
+ }
+ }
+ return 0;
+done:
+ countr_size *= 8;
+ *count = vcpu->regs[VCPU_REGS_RCX] & (~0ULL >> (64 - countr_size));
+ return 1;
+}
+
+static int handle_io(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
+{
+ u64 exit_qualification;
+
+ ++kvm_stat.io_exits;
+ exit_qualification = vmcs_read64(EXIT_QUALIFICATION);
+ kvm_run->exit_reason = KVM_EXIT_IO;
+ if (exit_qualification & 8)
+ kvm_run->io.direction = KVM_EXIT_IO_IN;
+ else
+ kvm_run->io.direction = KVM_EXIT_IO_OUT;
+ kvm_run->io.size = (exit_qualification & 7) + 1;
+ kvm_run->io.string = (exit_qualification & 16) != 0;
+ kvm_run->io.string_down
+ = (vmcs_readl(GUEST_RFLAGS) & X86_EFLAGS_DF) != 0;
+ kvm_run->io.rep = (exit_qualification & 32) != 0;
+ kvm_run->io.port = exit_qualification >> 16;
+ if (kvm_run->io.string) {
+ if (!get_io_count(vcpu, &kvm_run->io.count))
+ return 1;
+ kvm_run->io.address = vmcs_readl(GUEST_LINEAR_ADDRESS);
+ } else
+ kvm_run->io.value = vcpu->regs[VCPU_REGS_RAX]; /* rax */
+ return 0;
+}
+
+static int handle_invlpg(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
+{
+ u64 address = vmcs_read64(EXIT_QUALIFICATION);
+ int instruction_length = vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
+ spin_lock(&vcpu->kvm->lock);
+ vcpu->mmu.inval_page(vcpu, address);
+ spin_unlock(&vcpu->kvm->lock);
+ vmcs_writel(GUEST_RIP, vmcs_readl(GUEST_RIP) + instruction_length);
+ return 1;
+}
+
+static int handle_cr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
+{
+ u64 exit_qualification;
+ int cr;
+ int reg;
+
+#ifdef KVM_DEBUG
+ if (guest_cpl() != 0) {
+ vcpu_printf(vcpu, "%s: not supervisor\n", __FUNCTION__);
+ inject_gp(vcpu);
+ return 1;
+ }
+#endif
+
+ exit_qualification = vmcs_read64(EXIT_QUALIFICATION);
+ cr = exit_qualification & 15;
+ reg = (exit_qualification >> 8) & 15;
+ switch ((exit_qualification >> 4) & 3) {
+ case 0: /* mov to cr */
+ switch (cr) {
+ case 0:
+ vcpu_load_rsp_rip(vcpu);
+ set_cr0(vcpu, vcpu->regs[reg]);
+ skip_emulated_instruction(vcpu);
+ return 1;
+ case 3:
+ vcpu_load_rsp_rip(vcpu);
+ set_cr3(vcpu, vcpu->regs[reg]);
+ skip_emulated_instruction(vcpu);
+ return 1;
+ case 4:
+ vcpu_load_rsp_rip(vcpu);
+ set_cr4(vcpu, vcpu->regs[reg]);
+ skip_emulated_instruction(vcpu);
+ return 1;
+ case 8:
+ vcpu_load_rsp_rip(vcpu);
+ set_cr8(vcpu, vcpu->regs[reg]);
+ skip_emulated_instruction(vcpu);
+ return 1;
+ };
+ break;
+ case 1: /*mov from cr*/
+ switch (cr) {
+ case 3:
+ vcpu_load_rsp_rip(vcpu);
+ vcpu->regs[reg] = vcpu->cr3;
+ vcpu_put_rsp_rip(vcpu);
+ skip_emulated_instruction(vcpu);
+ return 1;
+ case 8:
+ printk(KERN_DEBUG "handle_cr: read CR8 "
+ "cpu erratum AA15\n");
+ vcpu_load_rsp_rip(vcpu);
+ vcpu->regs[reg] = vcpu->cr8;
+ vcpu_put_rsp_rip(vcpu);
+ skip_emulated_instruction(vcpu);
+ return 1;
+ }
+ break;
+ case 3: /* lmsw */
+ lmsw(vcpu, (exit_qualification >> LMSW_SOURCE_DATA_SHIFT) & 0x0f);
+
+ skip_emulated_instruction(vcpu);
+ return 1;
+ default:
+ break;
+ }
+ kvm_run->exit_reason = 0;
+ printk(KERN_ERR "kvm: unhandled control register: op %d cr %d\n",
+ (int)(exit_qualification >> 4) & 3, cr);
+ return 0;
+}
+
+static int handle_dr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
+{
+ u64 exit_qualification;
+ unsigned long val;
+ int dr, reg;
+
+ /*
+ * FIXME: this code assumes the host is debugging the guest.
+ * need to deal with guest debugging itself too.
+ */
+ exit_qualification = vmcs_read64(EXIT_QUALIFICATION);
+ dr = exit_qualification & 7;
+ reg = (exit_qualification >> 8) & 15;
+ vcpu_load_rsp_rip(vcpu);
+ if (exit_qualification & 16) {
+ /* mov from dr */
+ switch (dr) {
+ case 6:
+ val = 0xffff0ff0;
+ break;
+ case 7:
+ val = 0x400;
+ break;
+ default:
+ val = 0;
+ }
+ vcpu->regs[reg] = val;
+ } else {
+ /* mov to dr */
+ }
+ vcpu_put_rsp_rip(vcpu);
+ skip_emulated_instruction(vcpu);
+ return 1;
+}
+
+static int handle_cpuid(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
+{
+ kvm_run->exit_reason = KVM_EXIT_CPUID;
+ return 0;
+}
+
+static int handle_rdmsr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
+{
+ u32 ecx = vcpu->regs[VCPU_REGS_RCX];
+ u64 data;
+
+ if (vmx_get_msr(vcpu, ecx, &data)) {
+ inject_gp(vcpu);
+ return 1;
+ }
+
+ /* FIXME: handling of bits 32:63 of rax, rdx */
+ vcpu->regs[VCPU_REGS_RAX] = data & -1u;
+ vcpu->regs[VCPU_REGS_RDX] = (data >> 32) & -1u;
+ skip_emulated_instruction(vcpu);
+ return 1;
+}
+
+static int handle_wrmsr(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
+{
+ u32 ecx = vcpu->regs[VCPU_REGS_RCX];
+ u64 data = (vcpu->regs[VCPU_REGS_RAX] & -1u)
+ | ((u64)(vcpu->regs[VCPU_REGS_RDX] & -1u) << 32);
+
+ if (vmx_set_msr(vcpu, ecx, data) != 0) {
+ inject_gp(vcpu);
+ return 1;
+ }
+
+ skip_emulated_instruction(vcpu);
+ return 1;
+}
+
+static int handle_interrupt_window(struct kvm_vcpu *vcpu,
+ struct kvm_run *kvm_run)
+{
+ /* Turn off interrupt window reporting. */
+ vmcs_write32(CPU_BASED_VM_EXEC_CONTROL,
+ vmcs_read32(CPU_BASED_VM_EXEC_CONTROL)
+ & ~CPU_BASED_VIRTUAL_INTR_PENDING);
+ return 1;
+}
+
+static int handle_halt(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
+{
+ skip_emulated_instruction(vcpu);
+ if (vcpu->irq_summary && (vmcs_readl(GUEST_RFLAGS) & X86_EFLAGS_IF))
+ return 1;
+
+ kvm_run->exit_reason = KVM_EXIT_HLT;
+ return 0;
+}
+
+/*
+ * The exit handlers return 1 if the exit was handled fully and guest execution
+ * may resume. Otherwise they set the kvm_run parameter to indicate what needs
+ * to be done to userspace and return 0.
+ */
+static int (*kvm_vmx_exit_handlers[])(struct kvm_vcpu *vcpu,
+ struct kvm_run *kvm_run) = {
+ [EXIT_REASON_EXCEPTION_NMI] = handle_exception,
+ [EXIT_REASON_EXTERNAL_INTERRUPT] = handle_external_interrupt,
+ [EXIT_REASON_IO_INSTRUCTION] = handle_io,
+ [EXIT_REASON_INVLPG] = handle_invlpg,
+ [EXIT_REASON_CR_ACCESS] = handle_cr,
+ [EXIT_REASON_DR_ACCESS] = handle_dr,
+ [EXIT_REASON_CPUID] = handle_cpuid,
+ [EXIT_REASON_MSR_READ] = handle_rdmsr,
+ [EXIT_REASON_MSR_WRITE] = handle_wrmsr,
+ [EXIT_REASON_PENDING_INTERRUPT] = handle_interrupt_window,
+ [EXIT_REASON_HLT] = handle_halt,
+};
+
+static const int kvm_vmx_max_exit_handlers =
+ sizeof(kvm_vmx_exit_handlers) / sizeof(*kvm_vmx_exit_handlers);
+
+/*
+ * The guest has exited. See if we can fix it or if we need userspace
+ * assistance.
+ */
+static int kvm_handle_exit(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
+{
+ u32 vectoring_info = vmcs_read32(IDT_VECTORING_INFO_FIELD);
+ u32 exit_reason = vmcs_read32(VM_EXIT_REASON);
+
+ if ( (vectoring_info & VECTORING_INFO_VALID_MASK) &&
+ exit_reason != EXIT_REASON_EXCEPTION_NMI )
+ printk(KERN_WARNING "%s: unexpected, valid vectoring info and "
+ "exit reason is 0x%x\n", __FUNCTION__, exit_reason);
+ kvm_run->instruction_length = vmcs_read32(VM_EXIT_INSTRUCTION_LEN);
+ if (exit_reason < kvm_vmx_max_exit_handlers
+ && kvm_vmx_exit_handlers[exit_reason])
+ return kvm_vmx_exit_handlers[exit_reason](vcpu, kvm_run);
+ else {
+ kvm_run->exit_reason = KVM_EXIT_UNKNOWN;
+ kvm_run->hw.hardware_exit_reason = exit_reason;
+ }
+ return 0;
+}
+
static int vmx_vcpu_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
u8 fail;
@@ -753,6 +1157,7 @@ static struct kvm_arch_ops vmx_arch_ops
.decache_regs = vcpu_put_rsp_rip,
.run = vmx_vcpu_run,
+ .skip_emulated_instruction = skip_emulated_instruction,
.vmx_return = (unsigned long)kvm_vmx_return,
};
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 16/38] KVM: Make vcpu_setup() an arch operation
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (14 preceding siblings ...)
2006-11-27 12:25 ` [PATCH 15/38] KVM: Move the vmx exit handlers to vmx.c Avi Kivity
@ 2006-11-27 12:26 ` Avi Kivity
2006-11-27 12:27 ` [PATCH 17/38] KVM: Make __set_cr0() (and dependencies) arch operations Avi Kivity
` (21 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:26 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -153,6 +153,17 @@ enum {
NR_VCPU_REGS
};
+enum {
+ VCPU_SREG_CS,
+ VCPU_SREG_DS,
+ VCPU_SREG_ES,
+ VCPU_SREG_FS,
+ VCPU_SREG_GS,
+ VCPU_SREG_SS,
+ VCPU_SREG_TR,
+ VCPU_SREG_LDTR,
+};
+
struct kvm_vcpu {
struct kvm *kvm;
struct vmcs *vmcs;
@@ -266,8 +277,8 @@ struct kvm_arch_ops {
void (*decache_regs)(struct kvm_vcpu *vcpu);
int (*run)(struct kvm_vcpu *vcpu, struct kvm_run *run);
+ int (*vcpu_setup)(struct kvm_vcpu *vcpu);
void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
- unsigned long vmx_return; /* temporary hack */
};
extern struct kvm_stat kvm_stat;
@@ -324,12 +335,18 @@ void set_cr4(struct kvm_vcpu *vcpu, unsi
void set_cr8(struct kvm_vcpu *vcpu, unsigned long cr0);
void lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
+void __set_efer(struct kvm_vcpu *vcpu, u64 efer);
+void __set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
+void __set_cr4(struct kvm_vcpu *vcpu, unsigned long cr0);
+
+int rmode_tss_base(struct kvm* kvm); /* temporary hack */
void inject_gp(struct kvm_vcpu *vcpu);
#ifdef __x86_64__
void set_efer(struct kvm_vcpu *vcpu, u64 efer);
#endif
+void fx_init(struct kvm_vcpu *vcpu);
void load_msrs(struct vmx_msr_entry *e, int n);
void save_msrs(struct vmx_msr_entry *e, int n);
@@ -348,6 +365,12 @@ int kvm_write_guest(struct kvm_vcpu *vcp
void vmcs_writel(unsigned long field, unsigned long value);
unsigned long vmcs_readl(unsigned long field);
+static inline struct page *_gfn_to_page(struct kvm *kvm, gfn_t gfn)
+{
+ struct kvm_memory_slot *slot = gfn_to_memslot(kvm, gfn);
+ return (slot) ? slot->phys_mem[gfn - slot->base_gfn] : 0;
+}
+
static inline u16 vmcs_read16(unsigned long field)
{
return vmcs_readl(field);
@@ -465,6 +488,11 @@ static inline void load_ldt(u16 sel)
}
#endif
+static inline void get_idt(struct descriptor_table *table)
+{
+ asm ("sidt %0" : "=m"(*table));
+}
+
#ifdef __x86_64__
static inline unsigned long read_msr(unsigned long msr)
{
@@ -490,6 +518,11 @@ static inline void fpu_init(void)
asm ("finit");
}
+static inline u32 get_rdx_init_val(void)
+{
+ return 0x600; /* P6 family */
+}
+
#define ASM_VMX_VMCLEAR_RAX ".byte 0x66, 0x0f, 0xc7, 0x30"
#define ASM_VMX_VMLAUNCH ".byte 0x0f, 0x01, 0xc2"
#define ASM_VMX_VMRESUME ".byte 0x0f, 0x01, 0xc3"
@@ -502,6 +535,12 @@ static inline void fpu_init(void)
#define MSR_IA32_TIME_STAMP_COUNTER 0x010
+#define TSS_IOPB_BASE_OFFSET 0x66
+#define TSS_BASE_SIZE 0x68
+#define TSS_IOPB_SIZE (65536 / 8)
+#define TSS_REDIRECTION_SIZE (256 / 8)
+#define RMODE_TSS_SIZE (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
+
#ifdef __x86_64__
/*
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -65,17 +65,6 @@ static struct kvm_stats_debugfs_item {
static struct dentry *debugfs_dir;
-enum {
- VCPU_SREG_CS,
- VCPU_SREG_DS,
- VCPU_SREG_ES,
- VCPU_SREG_FS,
- VCPU_SREG_GS,
- VCPU_SREG_SS,
- VCPU_SREG_TR,
- VCPU_SREG_LDTR,
-};
-
#define VMX_SEGMENT_FIELD(seg) \
[VCPU_SREG_##seg] { \
GUEST_##seg##_SELECTOR, \
@@ -101,26 +90,6 @@ struct kvm_vmx_segment_field {
};
EXPORT_SYMBOL_GPL(kvm_vmx_segment_fields);
-static const u32 vmx_msr_index[] = {
-#ifdef __x86_64__
- MSR_SYSCALL_MASK, MSR_LSTAR, MSR_CSTAR, MSR_KERNEL_GS_BASE,
-#endif
- MSR_EFER, MSR_K6_STAR,
-};
-#define NR_VMX_MSR (sizeof(vmx_msr_index) / sizeof(*vmx_msr_index))
-
-#define TSS_IOPB_BASE_OFFSET 0x66
-#define TSS_BASE_SIZE 0x68
-#define TSS_IOPB_SIZE (65536 / 8)
-#define TSS_REDIRECTION_SIZE (256 / 8)
-#define RMODE_TSS_SIZE (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
-
-#define MSR_IA32_FEATURE_CONTROL 0x03a
-#define MSR_IA32_VMX_PINBASED_CTLS_MSR 0x481
-#define MSR_IA32_VMX_PROCBASED_CTLS_MSR 0x482
-#define MSR_IA32_VMX_EXIT_CTLS_MSR 0x483
-#define MSR_IA32_VMX_ENTRY_CTLS_MSR 0x484
-
#define MAX_IO_MSRS 256
#define CR0_RESEVED_BITS 0xffffffff1ffaffc0ULL
@@ -129,12 +98,6 @@ static const u32 vmx_msr_index[] = {
#define CR8_RESEVED_BITS (~0x0fULL)
#define EFER_RESERVED_BITS 0xfffffffffffff2fe
-#ifdef __x86_64__
-#define HOST_IS_64 1
-#else
-#define HOST_IS_64 0
-#endif
-
struct vmx_msr_entry *find_msr_entry(struct kvm_vcpu *vcpu, u32 msr)
{
int i;
@@ -151,11 +114,6 @@ static void get_gdt(struct descriptor_ta
asm ("sgdt %0" : "=m"(*table));
}
-static void get_idt(struct descriptor_table *table)
-{
- asm ("sidt %0" : "=m"(*table));
-}
-
struct segment_descriptor {
u16 limit_low;
u16 base_low;
@@ -227,12 +185,6 @@ struct vmcs_descriptor {
} vmcs_descriptor;
EXPORT_SYMBOL_GPL(vmcs_descriptor);
-static inline struct page *_gfn_to_page(struct kvm *kvm, gfn_t gfn)
-{
- struct kvm_memory_slot *slot = gfn_to_memslot(kvm, gfn);
- return (slot) ? slot->phys_mem[gfn - slot->base_gfn] : 0;
-}
-
int kvm_read_guest(struct kvm_vcpu *vcpu,
gva_t addr,
unsigned long size,
@@ -630,11 +582,12 @@ static void enter_pmode(struct kvm_vcpu
vmcs_write32(GUEST_CS_AR_BYTES, 0x9b);
}
-static int rmode_tss_base(struct kvm* kvm)
+int rmode_tss_base(struct kvm* kvm)
{
gfn_t base_gfn = kvm->memslots[0].base_gfn + kvm->memslots[0].npages - 3;
return base_gfn << PAGE_SHIFT;
}
+EXPORT_SYMBOL_GPL(rmode_tss_base);
static void fix_rmode_seg(int seg, struct kvm_save_segment *save)
{
@@ -686,41 +639,9 @@ static void enter_rmode(struct kvm_vcpu
fix_rmode_seg(VCPU_SREG_FS, &vcpu->rmode.fs);
}
-static int init_rmode_tss(struct kvm* kvm)
-{
- struct page *p1, *p2, *p3;
- gfn_t fn = rmode_tss_base(kvm) >> PAGE_SHIFT;
- char *page;
-
- p1 = _gfn_to_page(kvm, fn++);
- p2 = _gfn_to_page(kvm, fn++);
- p3 = _gfn_to_page(kvm, fn);
-
- if (!p1 || !p2 || !p3) {
- kvm_printf(kvm,"%s: gfn_to_page failed\n", __FUNCTION__);
- return 0;
- }
-
- page = kmap_atomic(p1, KM_USER0);
- memset(page, 0, PAGE_SIZE);
- *(u16*)(page + 0x66) = TSS_BASE_SIZE + TSS_REDIRECTION_SIZE;
- kunmap_atomic(page, KM_USER0);
-
- page = kmap_atomic(p2, KM_USER0);
- memset(page, 0, PAGE_SIZE);
- kunmap_atomic(page, KM_USER0);
-
- page = kmap_atomic(p3, KM_USER0);
- memset(page, 0, PAGE_SIZE);
- *(page + RMODE_TSS_SIZE - 2 * PAGE_SIZE - 1) = ~0;
- kunmap_atomic(page, KM_USER0);
-
- return 1;
-}
-
#ifdef __x86_64__
-static void __set_efer(struct kvm_vcpu *vcpu, u64 efer)
+void __set_efer(struct kvm_vcpu *vcpu, u64 efer)
{
struct vmx_msr_entry *msr = find_msr_entry(vcpu, MSR_EFER);
@@ -739,6 +660,7 @@ static void __set_efer(struct kvm_vcpu *
msr->data = efer & ~EFER_LME;
}
}
+EXPORT_SYMBOL_GPL(__set_efer);
static void enter_lmode(struct kvm_vcpu *vcpu)
{
@@ -772,7 +694,7 @@ static void exit_lmode(struct kvm_vcpu *
#endif
-static void __set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
+void __set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
{
if (vcpu->rmode.active && (cr0 & CR0_PE_MASK))
enter_pmode(vcpu);
@@ -794,6 +716,7 @@ static void __set_cr0(struct kvm_vcpu *v
(cr0 & ~KVM_GUEST_CR0_MASK) | KVM_VM_CR0_ALWAYS_ON);
vcpu->cr0 = cr0;
}
+EXPORT_SYMBOL_GPL(__set_cr0);
static int pdptrs_have_reserved_bits_set(struct kvm_vcpu *vcpu,
unsigned long cr3)
@@ -899,13 +822,14 @@ void lmsw(struct kvm_vcpu *vcpu, unsigne
}
EXPORT_SYMBOL_GPL(lmsw);
-static void __set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
+void __set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
{
vmcs_writel(CR4_READ_SHADOW, cr4);
vmcs_writel(GUEST_CR4, cr4 | (vcpu->rmode.active ?
KVM_RMODE_VM_CR4_ALWAYS_ON : KVM_PMODE_VM_CR4_ALWAYS_ON));
vcpu->cr4 = cr4;
}
+EXPORT_SYMBOL_GPL(__set_cr4);
void set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
{
@@ -981,12 +905,7 @@ void set_cr8(struct kvm_vcpu *vcpu, unsi
}
EXPORT_SYMBOL_GPL(set_cr8);
-static u32 get_rdx_init_val(void)
-{
- return 0x600; /* P6 family */
-}
-
-static void fx_init(struct kvm_vcpu *vcpu)
+void fx_init(struct kvm_vcpu *vcpu)
{
struct __attribute__ ((__packed__)) fx_image_s {
u16 control; //fcw
@@ -1010,237 +929,7 @@ static void fx_init(struct kvm_vcpu *vcp
memset(vcpu->guest_fx_image + sizeof(struct fx_image_s),
0, FX_IMAGE_SIZE - sizeof(struct fx_image_s));
}
-
-static void vmcs_write32_fixedbits(u32 msr, u32 vmcs_field, u32 val)
-{
- u32 msr_high, msr_low;
-
- rdmsr(msr, msr_low, msr_high);
-
- val &= msr_high;
- val |= msr_low;
- vmcs_write32(vmcs_field, val);
-}
-
-static void seg_setup(int seg)
-{
- struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
-
- vmcs_write16(sf->selector, 0);
- vmcs_writel(sf->base, 0);
- vmcs_write32(sf->limit, 0xffff);
- vmcs_write32(sf->ar_bytes, 0x93);
-}
-
-/*
- * Sets up the vmcs for emulated real mode.
- */
-static int kvm_vcpu_setup(struct kvm_vcpu *vcpu)
-{
- u32 host_sysenter_cs;
- u32 junk;
- unsigned long a;
- struct descriptor_table dt;
- int i;
- int ret;
- int nr_good_msrs;
-
-
- if (!init_rmode_tss(vcpu->kvm)) {
- ret = 0;
- goto out;
- }
-
- memset(vcpu->regs, 0, sizeof(vcpu->regs));
- vcpu->regs[VCPU_REGS_RDX] = get_rdx_init_val();
- vcpu->cr8 = 0;
- vcpu->apic_base = 0xfee00000 |
- /*for vcpu 0*/ MSR_IA32_APICBASE_BSP |
- MSR_IA32_APICBASE_ENABLE;
-
- fx_init(vcpu);
-
- /*
- * GUEST_CS_BASE should really be 0xffff0000, but VT vm86 mode
- * insists on having GUEST_CS_BASE == GUEST_CS_SELECTOR << 4. Sigh.
- */
- vmcs_write16(GUEST_CS_SELECTOR, 0xf000);
- vmcs_writel(GUEST_CS_BASE, 0x000f0000);
- vmcs_write32(GUEST_CS_LIMIT, 0xffff);
- vmcs_write32(GUEST_CS_AR_BYTES, 0x9b);
-
- seg_setup(VCPU_SREG_DS);
- seg_setup(VCPU_SREG_ES);
- seg_setup(VCPU_SREG_FS);
- seg_setup(VCPU_SREG_GS);
- seg_setup(VCPU_SREG_SS);
-
- vmcs_write16(GUEST_TR_SELECTOR, 0);
- vmcs_writel(GUEST_TR_BASE, 0);
- vmcs_write32(GUEST_TR_LIMIT, 0xffff);
- vmcs_write32(GUEST_TR_AR_BYTES, 0x008b);
-
- vmcs_write16(GUEST_LDTR_SELECTOR, 0);
- vmcs_writel(GUEST_LDTR_BASE, 0);
- vmcs_write32(GUEST_LDTR_LIMIT, 0xffff);
- vmcs_write32(GUEST_LDTR_AR_BYTES, 0x00082);
-
- vmcs_write32(GUEST_SYSENTER_CS, 0);
- vmcs_writel(GUEST_SYSENTER_ESP, 0);
- vmcs_writel(GUEST_SYSENTER_EIP, 0);
-
- vmcs_writel(GUEST_RFLAGS, 0x02);
- vmcs_writel(GUEST_RIP, 0xfff0);
- vmcs_writel(GUEST_RSP, 0);
-
- vmcs_writel(GUEST_CR3, 0);
-
- //todo: dr0 = dr1 = dr2 = dr3 = 0; dr6 = 0xffff0ff0
- vmcs_writel(GUEST_DR7, 0x400);
-
- vmcs_writel(GUEST_GDTR_BASE, 0);
- vmcs_write32(GUEST_GDTR_LIMIT, 0xffff);
-
- vmcs_writel(GUEST_IDTR_BASE, 0);
- vmcs_write32(GUEST_IDTR_LIMIT, 0xffff);
-
- vmcs_write32(GUEST_ACTIVITY_STATE, 0);
- vmcs_write32(GUEST_INTERRUPTIBILITY_INFO, 0);
- vmcs_write32(GUEST_PENDING_DBG_EXCEPTIONS, 0);
-
- /* I/O */
- vmcs_write64(IO_BITMAP_A, 0);
- vmcs_write64(IO_BITMAP_B, 0);
-
- guest_write_tsc(0);
-
- vmcs_write64(VMCS_LINK_POINTER, -1ull); /* 22.3.1.5 */
-
- /* Special registers */
- vmcs_write64(GUEST_IA32_DEBUGCTL, 0);
-
- /* Control */
- vmcs_write32_fixedbits(MSR_IA32_VMX_PINBASED_CTLS_MSR,
- PIN_BASED_VM_EXEC_CONTROL,
- PIN_BASED_EXT_INTR_MASK /* 20.6.1 */
- | PIN_BASED_NMI_EXITING /* 20.6.1 */
- );
- vmcs_write32_fixedbits(MSR_IA32_VMX_PROCBASED_CTLS_MSR,
- CPU_BASED_VM_EXEC_CONTROL,
- CPU_BASED_HLT_EXITING /* 20.6.2 */
- | CPU_BASED_CR8_LOAD_EXITING /* 20.6.2 */
- | CPU_BASED_CR8_STORE_EXITING /* 20.6.2 */
- | CPU_BASED_UNCOND_IO_EXITING /* 20.6.2 */
- | CPU_BASED_INVDPG_EXITING
- | CPU_BASED_MOV_DR_EXITING
- | CPU_BASED_USE_TSC_OFFSETING /* 21.3 */
- );
-
- vmcs_write32(EXCEPTION_BITMAP, 1 << PF_VECTOR);
- vmcs_write32(PAGE_FAULT_ERROR_CODE_MASK, 0);
- vmcs_write32(PAGE_FAULT_ERROR_CODE_MATCH, 0);
- vmcs_write32(CR3_TARGET_COUNT, 0); /* 22.2.1 */
-
- vmcs_writel(HOST_CR0, read_cr0()); /* 22.2.3 */
- vmcs_writel(HOST_CR4, read_cr4()); /* 22.2.3, 22.2.5 */
- vmcs_writel(HOST_CR3, read_cr3()); /* 22.2.3 FIXME: shadow tables */
-
- vmcs_write16(HOST_CS_SELECTOR, __KERNEL_CS); /* 22.2.4 */
- vmcs_write16(HOST_DS_SELECTOR, __KERNEL_DS); /* 22.2.4 */
- vmcs_write16(HOST_ES_SELECTOR, __KERNEL_DS); /* 22.2.4 */
- vmcs_write16(HOST_FS_SELECTOR, read_fs()); /* 22.2.4 */
- vmcs_write16(HOST_GS_SELECTOR, read_gs()); /* 22.2.4 */
- vmcs_write16(HOST_SS_SELECTOR, __KERNEL_DS); /* 22.2.4 */
-#ifdef __x86_64__
- rdmsrl(MSR_FS_BASE, a);
- vmcs_writel(HOST_FS_BASE, a); /* 22.2.4 */
- rdmsrl(MSR_GS_BASE, a);
- vmcs_writel(HOST_GS_BASE, a); /* 22.2.4 */
-#else
- vmcs_writel(HOST_FS_BASE, 0); /* 22.2.4 */
- vmcs_writel(HOST_GS_BASE, 0); /* 22.2.4 */
-#endif
-
- vmcs_write16(HOST_TR_SELECTOR, GDT_ENTRY_TSS*8); /* 22.2.4 */
-
- get_idt(&dt);
- vmcs_writel(HOST_IDTR_BASE, dt.base); /* 22.2.4 */
-
-
- vmcs_writel(HOST_RIP, kvm_arch_ops->vmx_return); /* 22.2.5 */
-
- rdmsr(MSR_IA32_SYSENTER_CS, host_sysenter_cs, junk);
- vmcs_write32(HOST_IA32_SYSENTER_CS, host_sysenter_cs);
- rdmsrl(MSR_IA32_SYSENTER_ESP, a);
- vmcs_writel(HOST_IA32_SYSENTER_ESP, a); /* 22.2.3 */
- rdmsrl(MSR_IA32_SYSENTER_EIP, a);
- vmcs_writel(HOST_IA32_SYSENTER_EIP, a); /* 22.2.3 */
-
- ret = -ENOMEM;
- vcpu->guest_msrs = kmalloc(PAGE_SIZE, GFP_KERNEL);
- if (!vcpu->guest_msrs)
- goto out;
- vcpu->host_msrs = kmalloc(PAGE_SIZE, GFP_KERNEL);
- if (!vcpu->host_msrs)
- goto out_free_guest_msrs;
-
- for (i = 0; i < NR_VMX_MSR; ++i) {
- u32 index = vmx_msr_index[i];
- u32 data_low, data_high;
- u64 data;
- int j = vcpu->nmsrs;
-
- if (rdmsr_safe(index, &data_low, &data_high) < 0)
- continue;
- data = data_low | ((u64)data_high << 32);
- vcpu->host_msrs[j].index = index;
- vcpu->host_msrs[j].reserved = 0;
- vcpu->host_msrs[j].data = data;
- vcpu->guest_msrs[j] = vcpu->host_msrs[j];
- ++vcpu->nmsrs;
- }
- printk("msrs: %d\n", vcpu->nmsrs);
-
- nr_good_msrs = vcpu->nmsrs - NR_BAD_MSRS;
- vmcs_writel(VM_ENTRY_MSR_LOAD_ADDR,
- virt_to_phys(vcpu->guest_msrs + NR_BAD_MSRS));
- vmcs_writel(VM_EXIT_MSR_STORE_ADDR,
- virt_to_phys(vcpu->guest_msrs + NR_BAD_MSRS));
- vmcs_writel(VM_EXIT_MSR_LOAD_ADDR,
- virt_to_phys(vcpu->host_msrs + NR_BAD_MSRS));
- vmcs_write32_fixedbits(MSR_IA32_VMX_EXIT_CTLS_MSR, VM_EXIT_CONTROLS,
- (HOST_IS_64 << 9)); /* 22.2,1, 20.7.1 */
- vmcs_write32(VM_EXIT_MSR_STORE_COUNT, nr_good_msrs); /* 22.2.2 */
- vmcs_write32(VM_EXIT_MSR_LOAD_COUNT, nr_good_msrs); /* 22.2.2 */
- vmcs_write32(VM_ENTRY_MSR_LOAD_COUNT, nr_good_msrs); /* 22.2.2 */
-
-
- /* 22.2.1, 20.8.1 */
- vmcs_write32_fixedbits(MSR_IA32_VMX_ENTRY_CTLS_MSR,
- VM_ENTRY_CONTROLS, 0);
- vmcs_write32(VM_ENTRY_INTR_INFO_FIELD, 0); /* 22.2.1 */
-
- vmcs_writel(VIRTUAL_APIC_PAGE_ADDR, 0);
- vmcs_writel(TPR_THRESHOLD, 0);
-
- vmcs_writel(CR0_GUEST_HOST_MASK, KVM_GUEST_CR0_MASK);
- vmcs_writel(CR4_GUEST_HOST_MASK, KVM_GUEST_CR4_MASK);
-
- __set_cr0(vcpu, 0x60000010); // enter rmode
- __set_cr4(vcpu, 0);
-#ifdef __x86_64__
- __set_efer(vcpu, 0);
-#endif
-
- ret = kvm_mmu_init(vcpu);
-
- return ret;
-
-out_free_guest_msrs:
- kfree(vcpu->guest_msrs);
-out:
- return ret;
-}
+EXPORT_SYMBOL_GPL(fx_init);
/*
* Creates some virtual cpus. Good luck creating more than one.
@@ -1281,7 +970,9 @@ static int kvm_dev_ioctl_create_vcpu(str
__vcpu_load(vcpu);
- r = kvm_vcpu_setup(vcpu);
+ r = kvm_arch_ops->vcpu_setup(vcpu);
+ if (r >= 0)
+ r = kvm_mmu_init(vcpu);
vcpu_put(vcpu);
@@ -1514,6 +1205,7 @@ struct kvm_memory_slot *gfn_to_memslot(s
}
return 0;
}
+EXPORT_SYMBOL_GPL(gfn_to_memslot);
void mark_page_dirty(struct kvm *kvm, gfn_t gfn)
{
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -19,6 +19,9 @@
#include <linux/module.h>
#include "vmx.h"
#include "kvm_vmx.h"
+#include <linux/mm.h>
+#include <linux/highmem.h>
+#include <asm/io.h>
#define MSR_IA32_FEATURE_CONTROL 0x03a
@@ -27,6 +30,12 @@ MODULE_LICENSE("GPL");
DECLARE_PER_CPU(struct vmcs *, vmxarea);
+#ifdef __x86_64__
+#define HOST_IS_64 1
+#else
+#define HOST_IS_64 0
+#endif
+
extern struct vmcs_descriptor {
int size;
int order;
@@ -40,6 +49,14 @@ extern struct kvm_vmx_segment_field {
unsigned ar_bytes;
} kvm_vmx_segment_fields[];
+static const u32 vmx_msr_index[] = {
+#ifdef __x86_64__
+ MSR_SYSCALL_MASK, MSR_LSTAR, MSR_CSTAR, MSR_KERNEL_GS_BASE,
+#endif
+ MSR_EFER, MSR_K6_STAR,
+};
+#define NR_VMX_MSR (sizeof(vmx_msr_index) / sizeof(*vmx_msr_index))
+
u64 guest_read_tsc(void);
void guest_write_tsc(u64 guest_tsc);
struct vmx_msr_entry *find_msr_entry(struct kvm_vcpu *vcpu, u32 msr);
@@ -436,6 +453,267 @@ static void vmx_set_gdt(struct kvm_vcpu
vmcs_writel(GUEST_GDTR_BASE, dt->base);
}
+static int init_rmode_tss(struct kvm* kvm)
+{
+ struct page *p1, *p2, *p3;
+ gfn_t fn = rmode_tss_base(kvm) >> PAGE_SHIFT;
+ char *page;
+
+ p1 = _gfn_to_page(kvm, fn++);
+ p2 = _gfn_to_page(kvm, fn++);
+ p3 = _gfn_to_page(kvm, fn);
+
+ if (!p1 || !p2 || !p3) {
+ kvm_printf(kvm,"%s: gfn_to_page failed\n", __FUNCTION__);
+ return 0;
+ }
+
+ page = kmap_atomic(p1, KM_USER0);
+ memset(page, 0, PAGE_SIZE);
+ *(u16*)(page + 0x66) = TSS_BASE_SIZE + TSS_REDIRECTION_SIZE;
+ kunmap_atomic(page, KM_USER0);
+
+ page = kmap_atomic(p2, KM_USER0);
+ memset(page, 0, PAGE_SIZE);
+ kunmap_atomic(page, KM_USER0);
+
+ page = kmap_atomic(p3, KM_USER0);
+ memset(page, 0, PAGE_SIZE);
+ *(page + RMODE_TSS_SIZE - 2 * PAGE_SIZE - 1) = ~0;
+ kunmap_atomic(page, KM_USER0);
+
+ return 1;
+}
+
+static void vmcs_write32_fixedbits(u32 msr, u32 vmcs_field, u32 val)
+{
+ u32 msr_high, msr_low;
+
+ rdmsr(msr, msr_low, msr_high);
+
+ val &= msr_high;
+ val |= msr_low;
+ vmcs_write32(vmcs_field, val);
+}
+
+static void seg_setup(int seg)
+{
+ struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
+
+ vmcs_write16(sf->selector, 0);
+ vmcs_writel(sf->base, 0);
+ vmcs_write32(sf->limit, 0xffff);
+ vmcs_write32(sf->ar_bytes, 0x93);
+}
+
+/*
+ * Sets up the vmcs for emulated real mode.
+ */
+static int vmx_vcpu_setup(struct kvm_vcpu *vcpu)
+{
+ u32 host_sysenter_cs;
+ u32 junk;
+ unsigned long a;
+ struct descriptor_table dt;
+ int i;
+ int ret = 0;
+ int nr_good_msrs;
+ extern asmlinkage void kvm_vmx_return(void);
+
+ if (!init_rmode_tss(vcpu->kvm)) {
+ ret = -ENOMEM;
+ goto out;
+ }
+
+ memset(vcpu->regs, 0, sizeof(vcpu->regs));
+ vcpu->regs[VCPU_REGS_RDX] = get_rdx_init_val();
+ vcpu->cr8 = 0;
+ vcpu->apic_base = 0xfee00000 |
+ /*for vcpu 0*/ MSR_IA32_APICBASE_BSP |
+ MSR_IA32_APICBASE_ENABLE;
+
+ fx_init(vcpu);
+
+ /*
+ * GUEST_CS_BASE should really be 0xffff0000, but VT vm86 mode
+ * insists on having GUEST_CS_BASE == GUEST_CS_SELECTOR << 4. Sigh.
+ */
+ vmcs_write16(GUEST_CS_SELECTOR, 0xf000);
+ vmcs_writel(GUEST_CS_BASE, 0x000f0000);
+ vmcs_write32(GUEST_CS_LIMIT, 0xffff);
+ vmcs_write32(GUEST_CS_AR_BYTES, 0x9b);
+
+ seg_setup(VCPU_SREG_DS);
+ seg_setup(VCPU_SREG_ES);
+ seg_setup(VCPU_SREG_FS);
+ seg_setup(VCPU_SREG_GS);
+ seg_setup(VCPU_SREG_SS);
+
+ vmcs_write16(GUEST_TR_SELECTOR, 0);
+ vmcs_writel(GUEST_TR_BASE, 0);
+ vmcs_write32(GUEST_TR_LIMIT, 0xffff);
+ vmcs_write32(GUEST_TR_AR_BYTES, 0x008b);
+
+ vmcs_write16(GUEST_LDTR_SELECTOR, 0);
+ vmcs_writel(GUEST_LDTR_BASE, 0);
+ vmcs_write32(GUEST_LDTR_LIMIT, 0xffff);
+ vmcs_write32(GUEST_LDTR_AR_BYTES, 0x00082);
+
+ vmcs_write32(GUEST_SYSENTER_CS, 0);
+ vmcs_writel(GUEST_SYSENTER_ESP, 0);
+ vmcs_writel(GUEST_SYSENTER_EIP, 0);
+
+ vmcs_writel(GUEST_RFLAGS, 0x02);
+ vmcs_writel(GUEST_RIP, 0xfff0);
+ vmcs_writel(GUEST_RSP, 0);
+
+ vmcs_writel(GUEST_CR3, 0);
+
+ //todo: dr0 = dr1 = dr2 = dr3 = 0; dr6 = 0xffff0ff0
+ vmcs_writel(GUEST_DR7, 0x400);
+
+ vmcs_writel(GUEST_GDTR_BASE, 0);
+ vmcs_write32(GUEST_GDTR_LIMIT, 0xffff);
+
+ vmcs_writel(GUEST_IDTR_BASE, 0);
+ vmcs_write32(GUEST_IDTR_LIMIT, 0xffff);
+
+ vmcs_write32(GUEST_ACTIVITY_STATE, 0);
+ vmcs_write32(GUEST_INTERRUPTIBILITY_INFO, 0);
+ vmcs_write32(GUEST_PENDING_DBG_EXCEPTIONS, 0);
+
+ /* I/O */
+ vmcs_write64(IO_BITMAP_A, 0);
+ vmcs_write64(IO_BITMAP_B, 0);
+
+ guest_write_tsc(0);
+
+ vmcs_write64(VMCS_LINK_POINTER, -1ull); /* 22.3.1.5 */
+
+ /* Special registers */
+ vmcs_write64(GUEST_IA32_DEBUGCTL, 0);
+
+ /* Control */
+ vmcs_write32_fixedbits(MSR_IA32_VMX_PINBASED_CTLS_MSR,
+ PIN_BASED_VM_EXEC_CONTROL,
+ PIN_BASED_EXT_INTR_MASK /* 20.6.1 */
+ | PIN_BASED_NMI_EXITING /* 20.6.1 */
+ );
+ vmcs_write32_fixedbits(MSR_IA32_VMX_PROCBASED_CTLS_MSR,
+ CPU_BASED_VM_EXEC_CONTROL,
+ CPU_BASED_HLT_EXITING /* 20.6.2 */
+ | CPU_BASED_CR8_LOAD_EXITING /* 20.6.2 */
+ | CPU_BASED_CR8_STORE_EXITING /* 20.6.2 */
+ | CPU_BASED_UNCOND_IO_EXITING /* 20.6.2 */
+ | CPU_BASED_INVDPG_EXITING
+ | CPU_BASED_MOV_DR_EXITING
+ | CPU_BASED_USE_TSC_OFFSETING /* 21.3 */
+ );
+
+ vmcs_write32(EXCEPTION_BITMAP, 1 << PF_VECTOR);
+ vmcs_write32(PAGE_FAULT_ERROR_CODE_MASK, 0);
+ vmcs_write32(PAGE_FAULT_ERROR_CODE_MATCH, 0);
+ vmcs_write32(CR3_TARGET_COUNT, 0); /* 22.2.1 */
+
+ vmcs_writel(HOST_CR0, read_cr0()); /* 22.2.3 */
+ vmcs_writel(HOST_CR4, read_cr4()); /* 22.2.3, 22.2.5 */
+ vmcs_writel(HOST_CR3, read_cr3()); /* 22.2.3 FIXME: shadow tables */
+
+ vmcs_write16(HOST_CS_SELECTOR, __KERNEL_CS); /* 22.2.4 */
+ vmcs_write16(HOST_DS_SELECTOR, __KERNEL_DS); /* 22.2.4 */
+ vmcs_write16(HOST_ES_SELECTOR, __KERNEL_DS); /* 22.2.4 */
+ vmcs_write16(HOST_FS_SELECTOR, read_fs()); /* 22.2.4 */
+ vmcs_write16(HOST_GS_SELECTOR, read_gs()); /* 22.2.4 */
+ vmcs_write16(HOST_SS_SELECTOR, __KERNEL_DS); /* 22.2.4 */
+#ifdef __x86_64__
+ rdmsrl(MSR_FS_BASE, a);
+ vmcs_writel(HOST_FS_BASE, a); /* 22.2.4 */
+ rdmsrl(MSR_GS_BASE, a);
+ vmcs_writel(HOST_GS_BASE, a); /* 22.2.4 */
+#else
+ vmcs_writel(HOST_FS_BASE, 0); /* 22.2.4 */
+ vmcs_writel(HOST_GS_BASE, 0); /* 22.2.4 */
+#endif
+
+ vmcs_write16(HOST_TR_SELECTOR, GDT_ENTRY_TSS*8); /* 22.2.4 */
+
+ get_idt(&dt);
+ vmcs_writel(HOST_IDTR_BASE, dt.base); /* 22.2.4 */
+
+
+ vmcs_writel(HOST_RIP, (unsigned long)kvm_vmx_return); /* 22.2.5 */
+
+ rdmsr(MSR_IA32_SYSENTER_CS, host_sysenter_cs, junk);
+ vmcs_write32(HOST_IA32_SYSENTER_CS, host_sysenter_cs);
+ rdmsrl(MSR_IA32_SYSENTER_ESP, a);
+ vmcs_writel(HOST_IA32_SYSENTER_ESP, a); /* 22.2.3 */
+ rdmsrl(MSR_IA32_SYSENTER_EIP, a);
+ vmcs_writel(HOST_IA32_SYSENTER_EIP, a); /* 22.2.3 */
+
+ ret = -ENOMEM;
+ vcpu->guest_msrs = kmalloc(PAGE_SIZE, GFP_KERNEL);
+ if (!vcpu->guest_msrs)
+ goto out;
+ vcpu->host_msrs = kmalloc(PAGE_SIZE, GFP_KERNEL);
+ if (!vcpu->host_msrs)
+ goto out_free_guest_msrs;
+
+ for (i = 0; i < NR_VMX_MSR; ++i) {
+ u32 index = vmx_msr_index[i];
+ u32 data_low, data_high;
+ u64 data;
+ int j = vcpu->nmsrs;
+
+ if (rdmsr_safe(index, &data_low, &data_high) < 0)
+ continue;
+ data = data_low | ((u64)data_high << 32);
+ vcpu->host_msrs[j].index = index;
+ vcpu->host_msrs[j].reserved = 0;
+ vcpu->host_msrs[j].data = data;
+ vcpu->guest_msrs[j] = vcpu->host_msrs[j];
+ ++vcpu->nmsrs;
+ }
+ printk("msrs: %d\n", vcpu->nmsrs);
+
+ nr_good_msrs = vcpu->nmsrs - NR_BAD_MSRS;
+ vmcs_writel(VM_ENTRY_MSR_LOAD_ADDR,
+ virt_to_phys(vcpu->guest_msrs + NR_BAD_MSRS));
+ vmcs_writel(VM_EXIT_MSR_STORE_ADDR,
+ virt_to_phys(vcpu->guest_msrs + NR_BAD_MSRS));
+ vmcs_writel(VM_EXIT_MSR_LOAD_ADDR,
+ virt_to_phys(vcpu->host_msrs + NR_BAD_MSRS));
+ vmcs_write32_fixedbits(MSR_IA32_VMX_EXIT_CTLS_MSR, VM_EXIT_CONTROLS,
+ (HOST_IS_64 << 9)); /* 22.2,1, 20.7.1 */
+ vmcs_write32(VM_EXIT_MSR_STORE_COUNT, nr_good_msrs); /* 22.2.2 */
+ vmcs_write32(VM_EXIT_MSR_LOAD_COUNT, nr_good_msrs); /* 22.2.2 */
+ vmcs_write32(VM_ENTRY_MSR_LOAD_COUNT, nr_good_msrs); /* 22.2.2 */
+
+
+ /* 22.2.1, 20.8.1 */
+ vmcs_write32_fixedbits(MSR_IA32_VMX_ENTRY_CTLS_MSR,
+ VM_ENTRY_CONTROLS, 0);
+ vmcs_write32(VM_ENTRY_INTR_INFO_FIELD, 0); /* 22.2.1 */
+
+ vmcs_writel(VIRTUAL_APIC_PAGE_ADDR, 0);
+ vmcs_writel(TPR_THRESHOLD, 0);
+
+ vmcs_writel(CR0_GUEST_HOST_MASK, KVM_GUEST_CR0_MASK);
+ vmcs_writel(CR4_GUEST_HOST_MASK, KVM_GUEST_CR4_MASK);
+
+ __set_cr0(vcpu, 0x60000010); // enter rmode
+ __set_cr4(vcpu, 0);
+#ifdef __x86_64__
+ __set_efer(vcpu, 0);
+#endif
+
+ return 0;
+
+out_free_guest_msrs:
+ kfree(vcpu->guest_msrs);
+out:
+ return ret;
+}
+
static void inject_rmode_irq(struct kvm_vcpu *vcpu, int irq)
{
u16 ent[2];
@@ -1133,8 +1411,6 @@ again:
return 0;
}
-extern asmlinkage void kvm_vmx_return(void);
-
static struct kvm_arch_ops vmx_arch_ops = {
.cpu_has_kvm_support = cpu_has_kvm_support,
.disabled_by_bios = vmx_disabled_by_bios,
@@ -1158,7 +1434,7 @@ static struct kvm_arch_ops vmx_arch_ops
.run = vmx_vcpu_run,
.skip_emulated_instruction = skip_emulated_instruction,
- .vmx_return = (unsigned long)kvm_vmx_return,
+ .vcpu_setup = vmx_vcpu_setup,
};
static int __init vmx_init(void)
Index: linux-2.6/drivers/kvm/vmx.h
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.h
+++ linux-2.6/drivers/kvm/vmx.h
@@ -287,5 +287,10 @@ enum vmcs_field {
#define CR4_VMXE 0x2000
#define MSR_IA32_VMX_BASIC_MSR 0x480
+#define MSR_IA32_FEATURE_CONTROL 0x03a
+#define MSR_IA32_VMX_PINBASED_CTLS_MSR 0x481
+#define MSR_IA32_VMX_PROCBASED_CTLS_MSR 0x482
+#define MSR_IA32_VMX_EXIT_CTLS_MSR 0x483
+#define MSR_IA32_VMX_ENTRY_CTLS_MSR 0x484
#endif
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 17/38] KVM: Make __set_cr0() (and dependencies) arch operations
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (15 preceding siblings ...)
2006-11-27 12:26 ` [PATCH 16/38] KVM: Make vcpu_setup() an arch operation Avi Kivity
@ 2006-11-27 12:27 ` Avi Kivity
2006-11-27 12:28 ` [PATCH 18/38] KVM: Make __set_cr4() an arch operation Avi Kivity
` (20 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:27 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -269,6 +269,7 @@ struct kvm_arch_ops {
struct kvm_segment *var, int seg);
void (*set_segment)(struct kvm_vcpu *vcpu,
struct kvm_segment *var, int seg);
+ void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
void (*get_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
void (*set_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
void (*get_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
@@ -279,6 +280,8 @@ struct kvm_arch_ops {
int (*run)(struct kvm_vcpu *vcpu, struct kvm_run *run);
int (*vcpu_setup)(struct kvm_vcpu *vcpu);
void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
+
+ void (*update_exception_bitmap)(struct kvm_vcpu *vcpu); /* hack */
};
extern struct kvm_stat kvm_stat;
@@ -336,10 +339,8 @@ void set_cr8(struct kvm_vcpu *vcpu, unsi
void lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
void __set_efer(struct kvm_vcpu *vcpu, u64 efer);
-void __set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
void __set_cr4(struct kvm_vcpu *vcpu, unsigned long cr0);
-int rmode_tss_base(struct kvm* kvm); /* temporary hack */
void inject_gp(struct kvm_vcpu *vcpu);
#ifdef __x86_64__
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -525,120 +525,6 @@ void guest_write_tsc(u64 guest_tsc)
}
EXPORT_SYMBOL_GPL(guest_write_tsc);
-static void update_exception_bitmap(struct kvm_vcpu *vcpu)
-{
- if (vcpu->rmode.active)
- vmcs_write32(EXCEPTION_BITMAP, ~0);
- else
- vmcs_write32(EXCEPTION_BITMAP, 1 << PF_VECTOR);
-}
-
-static void fix_pmode_dataseg(int seg, struct kvm_save_segment *save)
-{
- struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
-
- if (vmcs_readl(sf->base) == save->base) {
- vmcs_write16(sf->selector, save->selector);
- vmcs_writel(sf->base, save->base);
- vmcs_write32(sf->limit, save->limit);
- vmcs_write32(sf->ar_bytes, save->ar);
- } else {
- u32 dpl = (vmcs_read16(sf->selector) & SELECTOR_RPL_MASK)
- << AR_DPL_SHIFT;
- vmcs_write32(sf->ar_bytes, 0x93 | dpl);
- }
-}
-
-static void enter_pmode(struct kvm_vcpu *vcpu)
-{
- unsigned long flags;
-
- vcpu->rmode.active = 0;
-
- vmcs_writel(GUEST_TR_BASE, vcpu->rmode.tr.base);
- vmcs_write32(GUEST_TR_LIMIT, vcpu->rmode.tr.limit);
- vmcs_write32(GUEST_TR_AR_BYTES, vcpu->rmode.tr.ar);
-
- flags = vmcs_readl(GUEST_RFLAGS);
- flags &= ~(IOPL_MASK | X86_EFLAGS_VM);
- flags |= (vcpu->rmode.save_iopl << IOPL_SHIFT);
- vmcs_writel(GUEST_RFLAGS, flags);
-
- vmcs_writel(GUEST_CR4, (vmcs_readl(GUEST_CR4) & ~CR4_VME_MASK) |
- (vmcs_readl(CR4_READ_SHADOW) & CR4_VME_MASK));
-
- update_exception_bitmap(vcpu);
-
- fix_pmode_dataseg(VCPU_SREG_ES, &vcpu->rmode.es);
- fix_pmode_dataseg(VCPU_SREG_DS, &vcpu->rmode.ds);
- fix_pmode_dataseg(VCPU_SREG_GS, &vcpu->rmode.gs);
- fix_pmode_dataseg(VCPU_SREG_FS, &vcpu->rmode.fs);
-
- vmcs_write16(GUEST_SS_SELECTOR, 0);
- vmcs_write32(GUEST_SS_AR_BYTES, 0x93);
-
- vmcs_write16(GUEST_CS_SELECTOR,
- vmcs_read16(GUEST_CS_SELECTOR) & ~SELECTOR_RPL_MASK);
- vmcs_write32(GUEST_CS_AR_BYTES, 0x9b);
-}
-
-int rmode_tss_base(struct kvm* kvm)
-{
- gfn_t base_gfn = kvm->memslots[0].base_gfn + kvm->memslots[0].npages - 3;
- return base_gfn << PAGE_SHIFT;
-}
-EXPORT_SYMBOL_GPL(rmode_tss_base);
-
-static void fix_rmode_seg(int seg, struct kvm_save_segment *save)
-{
- struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
-
- save->selector = vmcs_read16(sf->selector);
- save->base = vmcs_readl(sf->base);
- save->limit = vmcs_read32(sf->limit);
- save->ar = vmcs_read32(sf->ar_bytes);
- vmcs_write16(sf->selector, vmcs_readl(sf->base) >> 4);
- vmcs_write32(sf->limit, 0xffff);
- vmcs_write32(sf->ar_bytes, 0xf3);
-}
-
-static void enter_rmode(struct kvm_vcpu *vcpu)
-{
- unsigned long flags;
-
- vcpu->rmode.active = 1;
-
- vcpu->rmode.tr.base = vmcs_readl(GUEST_TR_BASE);
- vmcs_writel(GUEST_TR_BASE, rmode_tss_base(vcpu->kvm));
-
- vcpu->rmode.tr.limit = vmcs_read32(GUEST_TR_LIMIT);
- vmcs_write32(GUEST_TR_LIMIT, RMODE_TSS_SIZE - 1);
-
- vcpu->rmode.tr.ar = vmcs_read32(GUEST_TR_AR_BYTES);
- vmcs_write32(GUEST_TR_AR_BYTES, 0x008b);
-
- flags = vmcs_readl(GUEST_RFLAGS);
- vcpu->rmode.save_iopl = (flags & IOPL_MASK) >> IOPL_SHIFT;
-
- flags |= IOPL_MASK | X86_EFLAGS_VM;
-
- vmcs_writel(GUEST_RFLAGS, flags);
- vmcs_writel(GUEST_CR4, vmcs_readl(GUEST_CR4) | CR4_VME_MASK);
- update_exception_bitmap(vcpu);
-
- vmcs_write16(GUEST_SS_SELECTOR, vmcs_readl(GUEST_SS_BASE) >> 4);
- vmcs_write32(GUEST_SS_LIMIT, 0xffff);
- vmcs_write32(GUEST_SS_AR_BYTES, 0xf3);
-
- vmcs_write32(GUEST_CS_AR_BYTES, 0xf3);
- vmcs_write16(GUEST_CS_SELECTOR, vmcs_readl(GUEST_CS_BASE) >> 4);
-
- fix_rmode_seg(VCPU_SREG_ES, &vcpu->rmode.es);
- fix_rmode_seg(VCPU_SREG_DS, &vcpu->rmode.ds);
- fix_rmode_seg(VCPU_SREG_GS, &vcpu->rmode.gs);
- fix_rmode_seg(VCPU_SREG_FS, &vcpu->rmode.fs);
-}
-
#ifdef __x86_64__
void __set_efer(struct kvm_vcpu *vcpu, u64 efer)
@@ -662,62 +548,8 @@ void __set_efer(struct kvm_vcpu *vcpu, u
}
EXPORT_SYMBOL_GPL(__set_efer);
-static void enter_lmode(struct kvm_vcpu *vcpu)
-{
- u32 guest_tr_ar;
-
- guest_tr_ar = vmcs_read32(GUEST_TR_AR_BYTES);
- if ((guest_tr_ar & AR_TYPE_MASK) != AR_TYPE_BUSY_64_TSS) {
- printk(KERN_DEBUG "%s: tss fixup for long mode. \n",
- __FUNCTION__);
- vmcs_write32(GUEST_TR_AR_BYTES,
- (guest_tr_ar & ~AR_TYPE_MASK)
- | AR_TYPE_BUSY_64_TSS);
- }
-
- vcpu->shadow_efer |= EFER_LMA;
-
- find_msr_entry(vcpu, MSR_EFER)->data |= EFER_LMA | EFER_LME;
- vmcs_write32(VM_ENTRY_CONTROLS,
- vmcs_read32(VM_ENTRY_CONTROLS)
- | VM_ENTRY_CONTROLS_IA32E_MASK);
-}
-
-static void exit_lmode(struct kvm_vcpu *vcpu)
-{
- vcpu->shadow_efer &= ~EFER_LMA;
-
- vmcs_write32(VM_ENTRY_CONTROLS,
- vmcs_read32(VM_ENTRY_CONTROLS)
- & ~VM_ENTRY_CONTROLS_IA32E_MASK);
-}
-
-#endif
-
-void __set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
-{
- if (vcpu->rmode.active && (cr0 & CR0_PE_MASK))
- enter_pmode(vcpu);
-
- if (!vcpu->rmode.active && !(cr0 & CR0_PE_MASK))
- enter_rmode(vcpu);
-
-#ifdef __x86_64__
- if (vcpu->shadow_efer & EFER_LME) {
- if (!is_paging(vcpu) && (cr0 & CR0_PG_MASK))
- enter_lmode(vcpu);
- if (is_paging(vcpu) && !(cr0 & CR0_PG_MASK))
- exit_lmode(vcpu);
- }
#endif
- vmcs_writel(CR0_READ_SHADOW, cr0);
- vmcs_writel(GUEST_CR0,
- (cr0 & ~KVM_GUEST_CR0_MASK) | KVM_VM_CR0_ALWAYS_ON);
- vcpu->cr0 = cr0;
-}
-EXPORT_SYMBOL_GPL(__set_cr0);
-
static int pdptrs_have_reserved_bits_set(struct kvm_vcpu *vcpu,
unsigned long cr3)
{
@@ -797,7 +629,9 @@ void set_cr0(struct kvm_vcpu *vcpu, unsi
}
- __set_cr0(vcpu, cr0);
+ kvm_arch_ops->set_cr0(vcpu, cr0);
+ vcpu->cr0 = cr0;
+
spin_lock(&vcpu->kvm->lock);
kvm_mmu_reset_context(vcpu);
spin_unlock(&vcpu->kvm->lock);
@@ -807,18 +641,7 @@ EXPORT_SYMBOL_GPL(set_cr0);
void lmsw(struct kvm_vcpu *vcpu, unsigned long msw)
{
- unsigned long cr0 = vcpu->cr0;
-
- if ((msw & CR0_PE_MASK) && !(cr0 & CR0_PE_MASK)) {
- enter_pmode(vcpu);
- vmcs_writel(CR0_READ_SHADOW, cr0 | CR0_PE_MASK);
-
- } else
- printk(KERN_DEBUG "lmsw: unexpected\n");
-
- vmcs_writel(GUEST_CR0, (vmcs_readl(GUEST_CR0) & ~LMSW_GUEST_MASK)
- | (msw & LMSW_GUEST_MASK));
- vcpu->cr0 = (vcpu->cr0 & ~0xfffful) | msw;
+ set_cr0(vcpu, (vcpu->cr0 & ~0x0ful) | (msw & 0x0f));
}
EXPORT_SYMBOL_GPL(lmsw);
@@ -1794,7 +1617,7 @@ static int kvm_dev_ioctl_set_sregs(struc
mmu_reset_needed |= vcpu->cr0 != sregs->cr0;
vcpu->rmode.active = ((sregs->cr0 & CR0_PE_MASK) == 0);
- update_exception_bitmap(vcpu);
+ kvm_arch_ops->update_exception_bitmap(vcpu);
vmcs_writel(CR0_READ_SHADOW, sregs->cr0);
vmcs_writel(GUEST_CR0,
(sregs->cr0 & ~KVM_GUEST_CR0_MASK) | KVM_VM_CR0_ALWAYS_ON);
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -375,6 +375,176 @@ static __exit void hardware_unsetup(void
free_kvm_area();
}
+static void update_exception_bitmap(struct kvm_vcpu *vcpu)
+{
+ if (vcpu->rmode.active)
+ vmcs_write32(EXCEPTION_BITMAP, ~0);
+ else
+ vmcs_write32(EXCEPTION_BITMAP, 1 << PF_VECTOR);
+}
+
+static void fix_pmode_dataseg(int seg, struct kvm_save_segment *save)
+{
+ struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
+
+ if (vmcs_readl(sf->base) == save->base) {
+ vmcs_write16(sf->selector, save->selector);
+ vmcs_writel(sf->base, save->base);
+ vmcs_write32(sf->limit, save->limit);
+ vmcs_write32(sf->ar_bytes, save->ar);
+ } else {
+ u32 dpl = (vmcs_read16(sf->selector) & SELECTOR_RPL_MASK)
+ << AR_DPL_SHIFT;
+ vmcs_write32(sf->ar_bytes, 0x93 | dpl);
+ }
+}
+
+static void enter_pmode(struct kvm_vcpu *vcpu)
+{
+ unsigned long flags;
+
+ vcpu->rmode.active = 0;
+
+ vmcs_writel(GUEST_TR_BASE, vcpu->rmode.tr.base);
+ vmcs_write32(GUEST_TR_LIMIT, vcpu->rmode.tr.limit);
+ vmcs_write32(GUEST_TR_AR_BYTES, vcpu->rmode.tr.ar);
+
+ flags = vmcs_readl(GUEST_RFLAGS);
+ flags &= ~(IOPL_MASK | X86_EFLAGS_VM);
+ flags |= (vcpu->rmode.save_iopl << IOPL_SHIFT);
+ vmcs_writel(GUEST_RFLAGS, flags);
+
+ vmcs_writel(GUEST_CR4, (vmcs_readl(GUEST_CR4) & ~CR4_VME_MASK) |
+ (vmcs_readl(CR4_READ_SHADOW) & CR4_VME_MASK));
+
+ update_exception_bitmap(vcpu);
+
+ fix_pmode_dataseg(VCPU_SREG_ES, &vcpu->rmode.es);
+ fix_pmode_dataseg(VCPU_SREG_DS, &vcpu->rmode.ds);
+ fix_pmode_dataseg(VCPU_SREG_GS, &vcpu->rmode.gs);
+ fix_pmode_dataseg(VCPU_SREG_FS, &vcpu->rmode.fs);
+
+ vmcs_write16(GUEST_SS_SELECTOR, 0);
+ vmcs_write32(GUEST_SS_AR_BYTES, 0x93);
+
+ vmcs_write16(GUEST_CS_SELECTOR,
+ vmcs_read16(GUEST_CS_SELECTOR) & ~SELECTOR_RPL_MASK);
+ vmcs_write32(GUEST_CS_AR_BYTES, 0x9b);
+}
+
+static int rmode_tss_base(struct kvm* kvm)
+{
+ gfn_t base_gfn = kvm->memslots[0].base_gfn + kvm->memslots[0].npages - 3;
+ return base_gfn << PAGE_SHIFT;
+}
+
+static void fix_rmode_seg(int seg, struct kvm_save_segment *save)
+{
+ struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
+
+ save->selector = vmcs_read16(sf->selector);
+ save->base = vmcs_readl(sf->base);
+ save->limit = vmcs_read32(sf->limit);
+ save->ar = vmcs_read32(sf->ar_bytes);
+ vmcs_write16(sf->selector, vmcs_readl(sf->base) >> 4);
+ vmcs_write32(sf->limit, 0xffff);
+ vmcs_write32(sf->ar_bytes, 0xf3);
+}
+
+static void enter_rmode(struct kvm_vcpu *vcpu)
+{
+ unsigned long flags;
+
+ vcpu->rmode.active = 1;
+
+ vcpu->rmode.tr.base = vmcs_readl(GUEST_TR_BASE);
+ vmcs_writel(GUEST_TR_BASE, rmode_tss_base(vcpu->kvm));
+
+ vcpu->rmode.tr.limit = vmcs_read32(GUEST_TR_LIMIT);
+ vmcs_write32(GUEST_TR_LIMIT, RMODE_TSS_SIZE - 1);
+
+ vcpu->rmode.tr.ar = vmcs_read32(GUEST_TR_AR_BYTES);
+ vmcs_write32(GUEST_TR_AR_BYTES, 0x008b);
+
+ flags = vmcs_readl(GUEST_RFLAGS);
+ vcpu->rmode.save_iopl = (flags & IOPL_MASK) >> IOPL_SHIFT;
+
+ flags |= IOPL_MASK | X86_EFLAGS_VM;
+
+ vmcs_writel(GUEST_RFLAGS, flags);
+ vmcs_writel(GUEST_CR4, vmcs_readl(GUEST_CR4) | CR4_VME_MASK);
+ update_exception_bitmap(vcpu);
+
+ vmcs_write16(GUEST_SS_SELECTOR, vmcs_readl(GUEST_SS_BASE) >> 4);
+ vmcs_write32(GUEST_SS_LIMIT, 0xffff);
+ vmcs_write32(GUEST_SS_AR_BYTES, 0xf3);
+
+ vmcs_write32(GUEST_CS_AR_BYTES, 0xf3);
+ vmcs_write16(GUEST_CS_SELECTOR, vmcs_readl(GUEST_CS_BASE) >> 4);
+
+ fix_rmode_seg(VCPU_SREG_ES, &vcpu->rmode.es);
+ fix_rmode_seg(VCPU_SREG_DS, &vcpu->rmode.ds);
+ fix_rmode_seg(VCPU_SREG_GS, &vcpu->rmode.gs);
+ fix_rmode_seg(VCPU_SREG_FS, &vcpu->rmode.fs);
+}
+
+#ifdef __x86_64__
+
+static void enter_lmode(struct kvm_vcpu *vcpu)
+{
+ u32 guest_tr_ar;
+
+ guest_tr_ar = vmcs_read32(GUEST_TR_AR_BYTES);
+ if ((guest_tr_ar & AR_TYPE_MASK) != AR_TYPE_BUSY_64_TSS) {
+ printk(KERN_DEBUG "%s: tss fixup for long mode. \n",
+ __FUNCTION__);
+ vmcs_write32(GUEST_TR_AR_BYTES,
+ (guest_tr_ar & ~AR_TYPE_MASK)
+ | AR_TYPE_BUSY_64_TSS);
+ }
+
+ vcpu->shadow_efer |= EFER_LMA;
+
+ find_msr_entry(vcpu, MSR_EFER)->data |= EFER_LMA | EFER_LME;
+ vmcs_write32(VM_ENTRY_CONTROLS,
+ vmcs_read32(VM_ENTRY_CONTROLS)
+ | VM_ENTRY_CONTROLS_IA32E_MASK);
+}
+
+static void exit_lmode(struct kvm_vcpu *vcpu)
+{
+ vcpu->shadow_efer &= ~EFER_LMA;
+
+ vmcs_write32(VM_ENTRY_CONTROLS,
+ vmcs_read32(VM_ENTRY_CONTROLS)
+ & ~VM_ENTRY_CONTROLS_IA32E_MASK);
+}
+
+#endif
+
+static void vmx_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
+{
+ if (vcpu->rmode.active && (cr0 & CR0_PE_MASK))
+ enter_pmode(vcpu);
+
+ if (!vcpu->rmode.active && !(cr0 & CR0_PE_MASK))
+ enter_rmode(vcpu);
+
+#ifdef __x86_64__
+ if (vcpu->shadow_efer & EFER_LME) {
+ if (!is_paging(vcpu) && (cr0 & CR0_PG_MASK))
+ enter_lmode(vcpu);
+ if (is_paging(vcpu) && !(cr0 & CR0_PG_MASK))
+ exit_lmode(vcpu);
+ }
+#endif
+
+ vmcs_writel(CR0_READ_SHADOW, cr0);
+ vmcs_writel(GUEST_CR0,
+ (cr0 & ~KVM_GUEST_CR0_MASK) | KVM_VM_CR0_ALWAYS_ON);
+ vcpu->cr0 = cr0;
+}
+
static u64 vmx_get_segment_base(struct kvm_vcpu *vcpu, int seg)
{
struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
@@ -700,7 +870,8 @@ static int vmx_vcpu_setup(struct kvm_vcp
vmcs_writel(CR0_GUEST_HOST_MASK, KVM_GUEST_CR0_MASK);
vmcs_writel(CR4_GUEST_HOST_MASK, KVM_GUEST_CR4_MASK);
- __set_cr0(vcpu, 0x60000010); // enter rmode
+ vcpu->cr0 = 0x60000010;
+ vmx_set_cr0(vcpu, vcpu->cr0); // enter rmode
__set_cr4(vcpu, 0);
#ifdef __x86_64__
__set_efer(vcpu, 0);
@@ -1425,6 +1596,7 @@ static struct kvm_arch_ops vmx_arch_ops
.get_segment_base = vmx_get_segment_base,
.get_segment = vmx_get_segment,
.set_segment = vmx_set_segment,
+ .set_cr0 = vmx_set_cr0,
.get_idt = vmx_get_idt,
.set_idt = vmx_set_idt,
.get_gdt = vmx_get_gdt,
@@ -1435,6 +1607,8 @@ static struct kvm_arch_ops vmx_arch_ops
.run = vmx_vcpu_run,
.skip_emulated_instruction = skip_emulated_instruction,
.vcpu_setup = vmx_vcpu_setup,
+
+ .update_exception_bitmap = update_exception_bitmap,
};
static int __init vmx_init(void)
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 18/38] KVM: Make __set_cr4() an arch operation
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (16 preceding siblings ...)
2006-11-27 12:27 ` [PATCH 17/38] KVM: Make __set_cr0() (and dependencies) arch operations Avi Kivity
@ 2006-11-27 12:28 ` Avi Kivity
2006-11-27 12:29 ` [PATCH 19/38] KVM: Make __set_efer() " Avi Kivity
` (19 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:28 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -270,6 +270,7 @@ struct kvm_arch_ops {
void (*set_segment)(struct kvm_vcpu *vcpu,
struct kvm_segment *var, int seg);
void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
+ void (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
void (*get_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
void (*set_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
void (*get_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
@@ -339,7 +340,6 @@ void set_cr8(struct kvm_vcpu *vcpu, unsi
void lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
void __set_efer(struct kvm_vcpu *vcpu, u64 efer);
-void __set_cr4(struct kvm_vcpu *vcpu, unsigned long cr0);
void inject_gp(struct kvm_vcpu *vcpu);
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -645,15 +645,6 @@ void lmsw(struct kvm_vcpu *vcpu, unsigne
}
EXPORT_SYMBOL_GPL(lmsw);
-void __set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
-{
- vmcs_writel(CR4_READ_SHADOW, cr4);
- vmcs_writel(GUEST_CR4, cr4 | (vcpu->rmode.active ?
- KVM_RMODE_VM_CR4_ALWAYS_ON : KVM_PMODE_VM_CR4_ALWAYS_ON));
- vcpu->cr4 = cr4;
-}
-EXPORT_SYMBOL_GPL(__set_cr4);
-
void set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
{
if (cr4 & CR4_RESEVED_BITS) {
@@ -680,7 +671,7 @@ void set_cr4(struct kvm_vcpu *vcpu, unsi
inject_gp(vcpu);
return;
}
- __set_cr4(vcpu, cr4);
+ kvm_arch_ops->set_cr4(vcpu, cr4);
spin_lock(&vcpu->kvm->lock);
kvm_mmu_reset_context(vcpu);
spin_unlock(&vcpu->kvm->lock);
@@ -1624,7 +1615,7 @@ static int kvm_dev_ioctl_set_sregs(struc
vcpu->cr0 = sregs->cr0;
mmu_reset_needed |= vcpu->cr4 != sregs->cr4;
- __set_cr4(vcpu, sregs->cr4);
+ kvm_arch_ops->set_cr4(vcpu, sregs->cr4);
if (mmu_reset_needed)
kvm_mmu_reset_context(vcpu);
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -545,6 +545,14 @@ static void vmx_set_cr0(struct kvm_vcpu
vcpu->cr0 = cr0;
}
+static void vmx_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
+{
+ vmcs_writel(CR4_READ_SHADOW, cr4);
+ vmcs_writel(GUEST_CR4, cr4 | (vcpu->rmode.active ?
+ KVM_RMODE_VM_CR4_ALWAYS_ON : KVM_PMODE_VM_CR4_ALWAYS_ON));
+ vcpu->cr4 = cr4;
+}
+
static u64 vmx_get_segment_base(struct kvm_vcpu *vcpu, int seg)
{
struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
@@ -872,7 +880,7 @@ static int vmx_vcpu_setup(struct kvm_vcp
vcpu->cr0 = 0x60000010;
vmx_set_cr0(vcpu, vcpu->cr0); // enter rmode
- __set_cr4(vcpu, 0);
+ vmx_set_cr4(vcpu, 0);
#ifdef __x86_64__
__set_efer(vcpu, 0);
#endif
@@ -1597,6 +1605,7 @@ static struct kvm_arch_ops vmx_arch_ops
.get_segment = vmx_get_segment,
.set_segment = vmx_set_segment,
.set_cr0 = vmx_set_cr0,
+ .set_cr4 = vmx_set_cr4,
.get_idt = vmx_get_idt,
.set_idt = vmx_set_idt,
.get_gdt = vmx_get_gdt,
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 19/38] KVM: Make __set_efer() an arch operation
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (17 preceding siblings ...)
2006-11-27 12:28 ` [PATCH 18/38] KVM: Make __set_cr4() an arch operation Avi Kivity
@ 2006-11-27 12:29 ` Avi Kivity
2006-11-27 13:39 ` Christoph Hellwig
2006-11-27 12:30 ` [PATCH 20/38] KVM: Make set_cr3() and tlb flushing arch operations Avi Kivity
` (18 subsequent siblings)
37 siblings, 1 reply; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:29 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -271,6 +271,7 @@ struct kvm_arch_ops {
struct kvm_segment *var, int seg);
void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
void (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
+ void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
void (*get_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
void (*set_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
void (*get_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
@@ -339,8 +340,6 @@ void set_cr4(struct kvm_vcpu *vcpu, unsi
void set_cr8(struct kvm_vcpu *vcpu, unsigned long cr0);
void lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
-void __set_efer(struct kvm_vcpu *vcpu, u64 efer);
-
void inject_gp(struct kvm_vcpu *vcpu);
#ifdef __x86_64__
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -525,31 +525,6 @@ void guest_write_tsc(u64 guest_tsc)
}
EXPORT_SYMBOL_GPL(guest_write_tsc);
-#ifdef __x86_64__
-
-void __set_efer(struct kvm_vcpu *vcpu, u64 efer)
-{
- struct vmx_msr_entry *msr = find_msr_entry(vcpu, MSR_EFER);
-
- vcpu->shadow_efer = efer;
- if (efer & EFER_LMA) {
- vmcs_write32(VM_ENTRY_CONTROLS,
- vmcs_read32(VM_ENTRY_CONTROLS) |
- VM_ENTRY_CONTROLS_IA32E_MASK);
- msr->data = efer;
-
- } else {
- vmcs_write32(VM_ENTRY_CONTROLS,
- vmcs_read32(VM_ENTRY_CONTROLS) &
- ~VM_ENTRY_CONTROLS_IA32E_MASK);
-
- msr->data = efer & ~EFER_LME;
- }
-}
-EXPORT_SYMBOL_GPL(__set_efer);
-
-#endif
-
static int pdptrs_have_reserved_bits_set(struct kvm_vcpu *vcpu,
unsigned long cr3)
{
@@ -1602,7 +1577,7 @@ static int kvm_dev_ioctl_set_sregs(struc
mmu_reset_needed |= vcpu->shadow_efer != sregs->efer;
#ifdef __x86_64__
- __set_efer(vcpu, sregs->efer);
+ kvm_arch_ops->set_efer(vcpu, sregs->efer);
#endif
vcpu->apic_base = sregs->apic_base;
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -553,6 +553,30 @@ static void vmx_set_cr4(struct kvm_vcpu
vcpu->cr4 = cr4;
}
+#ifdef __x86_64__
+
+static void vmx_set_efer(struct kvm_vcpu *vcpu, u64 efer)
+{
+ struct vmx_msr_entry *msr = find_msr_entry(vcpu, MSR_EFER);
+
+ vcpu->shadow_efer = efer;
+ if (efer & EFER_LMA) {
+ vmcs_write32(VM_ENTRY_CONTROLS,
+ vmcs_read32(VM_ENTRY_CONTROLS) |
+ VM_ENTRY_CONTROLS_IA32E_MASK);
+ msr->data = efer;
+
+ } else {
+ vmcs_write32(VM_ENTRY_CONTROLS,
+ vmcs_read32(VM_ENTRY_CONTROLS) &
+ ~VM_ENTRY_CONTROLS_IA32E_MASK);
+
+ msr->data = efer & ~EFER_LME;
+ }
+}
+
+#endif
+
static u64 vmx_get_segment_base(struct kvm_vcpu *vcpu, int seg)
{
struct kvm_vmx_segment_field *sf = &kvm_vmx_segment_fields[seg];
@@ -882,7 +906,7 @@ static int vmx_vcpu_setup(struct kvm_vcp
vmx_set_cr0(vcpu, vcpu->cr0); // enter rmode
vmx_set_cr4(vcpu, 0);
#ifdef __x86_64__
- __set_efer(vcpu, 0);
+ vmx_set_efer(vcpu, 0);
#endif
return 0;
@@ -1606,6 +1630,7 @@ static struct kvm_arch_ops vmx_arch_ops
.set_segment = vmx_set_segment,
.set_cr0 = vmx_set_cr0,
.set_cr4 = vmx_set_cr4,
+ .set_efer = vmx_set_efer,
.get_idt = vmx_get_idt,
.set_idt = vmx_set_idt,
.get_gdt = vmx_get_gdt,
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* Re: [PATCH 19/38] KVM: Make __set_efer() an arch operation
2006-11-27 12:29 ` [PATCH 19/38] KVM: Make __set_efer() " Avi Kivity
@ 2006-11-27 13:39 ` Christoph Hellwig
[not found] ` <20061127133944.GA4155-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
0 siblings, 1 reply; 49+ messages in thread
From: Christoph Hellwig @ 2006-11-27 13:39 UTC (permalink / raw)
To: Avi Kivity; +Cc: kvm-devel, linux-kernel, akpm
On Mon, Nov 27, 2006 at 12:29:38PM -0000, Avi Kivity wrote:
> #ifdef __x86_64__
> - __set_efer(vcpu, sregs->efer);
> + kvm_arch_ops->set_efer(vcpu, sregs->efer);
> #endif
I think it would be much better to make ->set_efer a noop for 32bit,
and have different operation vectors for 32 vs 64 bit.
> #ifdef __x86_64__
> - __set_efer(vcpu, 0);
> + vmx_set_efer(vcpu, 0);
> #endif
Similarly vmx_set_efer should just become a noop on 32bit.
^ permalink raw reply [flat|nested] 49+ messages in thread
* [PATCH 20/38] KVM: Make set_cr3() and tlb flushing arch operations
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (18 preceding siblings ...)
2006-11-27 12:29 ` [PATCH 19/38] KVM: Make __set_efer() " Avi Kivity
@ 2006-11-27 12:30 ` Avi Kivity
2006-11-27 12:31 ` [PATCH 21/38] KVM: Make inject_page_fault() an arch operation Avi Kivity
` (17 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:30 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -270,6 +270,7 @@ struct kvm_arch_ops {
void (*set_segment)(struct kvm_vcpu *vcpu,
struct kvm_segment *var, int seg);
void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
+ void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
void (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
void (*get_idt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
@@ -279,6 +280,8 @@ struct kvm_arch_ops {
void (*cache_regs)(struct kvm_vcpu *vcpu);
void (*decache_regs)(struct kvm_vcpu *vcpu);
+ void (*flush_tlb)(struct kvm_vcpu *vcpu);
+
int (*run)(struct kvm_vcpu *vcpu, struct kvm_run *run);
int (*vcpu_setup)(struct kvm_vcpu *vcpu);
void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
Index: linux-2.6/drivers/kvm/mmu.c
===================================================================
--- linux-2.6.orig/drivers/kvm/mmu.c
+++ linux-2.6/drivers/kvm/mmu.c
@@ -323,7 +323,7 @@ static void nonpaging_flush(struct kvm_v
vcpu->mmu.root_hpa = root;
if (is_paging(vcpu))
root |= (vcpu->cr3 & (CR3_PCD_MASK | CR3_WPT_MASK));
- vmcs_writel(GUEST_CR3, root);
+ kvm_arch_ops->set_cr3(vcpu, root);
}
static gpa_t nonpaging_gva_to_gpa(struct kvm_vcpu *vcpu, gva_t vaddr)
@@ -386,7 +386,7 @@ static int nonpaging_init_context(struct
context->shadow_root_level = PT32E_ROOT_LEVEL;
context->root_hpa = kvm_mmu_alloc_page(vcpu, 0);
ASSERT(VALID_PAGE(context->root_hpa));
- vmcs_writel(GUEST_CR3, context->root_hpa);
+ kvm_arch_ops->set_cr3(vcpu, context->root_hpa);
return 0;
}
@@ -539,9 +539,7 @@ static void paging_inval_page(struct kvm
table[index] = 0;
release_pt_page_64(vcpu, page_addr, PT_PAGE_TABLE_LEVEL);
- //flush tlb
- vmcs_writel(GUEST_CR3, vcpu->mmu.root_hpa |
- (vcpu->cr3 & (CR3_PCD_MASK | CR3_WPT_MASK)));
+ kvm_arch_ops->flush_tlb(vcpu);
return;
}
}
@@ -574,7 +572,7 @@ static int paging64_init_context(struct
context->shadow_root_level = PT64_ROOT_LEVEL;
context->root_hpa = kvm_mmu_alloc_page(vcpu, 0);
ASSERT(VALID_PAGE(context->root_hpa));
- vmcs_writel(GUEST_CR3, context->root_hpa |
+ kvm_arch_ops->set_cr3(vcpu, context->root_hpa |
(vcpu->cr3 & (CR3_PCD_MASK | CR3_WPT_MASK)));
return 0;
}
@@ -592,7 +590,7 @@ static int paging32_init_context(struct
context->shadow_root_level = PT32E_ROOT_LEVEL;
context->root_hpa = kvm_mmu_alloc_page(vcpu, 0);
ASSERT(VALID_PAGE(context->root_hpa));
- vmcs_writel(GUEST_CR3, context->root_hpa |
+ kvm_arch_ops->set_cr3(vcpu, context->root_hpa |
(vcpu->cr3 & (CR3_PCD_MASK | CR3_WPT_MASK)));
return 0;
}
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -545,6 +545,11 @@ static void vmx_set_cr0(struct kvm_vcpu
vcpu->cr0 = cr0;
}
+static void vmx_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3)
+{
+ vmcs_writel(GUEST_CR3, cr3);
+}
+
static void vmx_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
{
vmcs_writel(CR4_READ_SHADOW, cr4);
@@ -1614,6 +1619,11 @@ again:
return 0;
}
+static void vmx_flush_tlb(struct kvm_vcpu *vcpu)
+{
+ vmcs_writel(GUEST_CR3, vmcs_readl(GUEST_CR3));
+}
+
static struct kvm_arch_ops vmx_arch_ops = {
.cpu_has_kvm_support = cpu_has_kvm_support,
.disabled_by_bios = vmx_disabled_by_bios,
@@ -1629,6 +1639,7 @@ static struct kvm_arch_ops vmx_arch_ops
.get_segment = vmx_get_segment,
.set_segment = vmx_set_segment,
.set_cr0 = vmx_set_cr0,
+ .set_cr3 = vmx_set_cr3,
.set_cr4 = vmx_set_cr4,
.set_efer = vmx_set_efer,
.get_idt = vmx_get_idt,
@@ -1638,6 +1649,8 @@ static struct kvm_arch_ops vmx_arch_ops
.cache_regs = vcpu_load_rsp_rip,
.decache_regs = vcpu_put_rsp_rip,
+ .flush_tlb = vmx_flush_tlb,
+
.run = vmx_vcpu_run,
.skip_emulated_instruction = skip_emulated_instruction,
.vcpu_setup = vmx_vcpu_setup,
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 21/38] KVM: Make inject_page_fault() an arch operation
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (19 preceding siblings ...)
2006-11-27 12:30 ` [PATCH 20/38] KVM: Make set_cr3() and tlb flushing arch operations Avi Kivity
@ 2006-11-27 12:31 ` Avi Kivity
2006-11-27 12:32 ` [PATCH 22/38] KVM: Make inject_gp() " Avi Kivity
` (16 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:31 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -281,6 +281,8 @@ struct kvm_arch_ops {
void (*decache_regs)(struct kvm_vcpu *vcpu);
void (*flush_tlb)(struct kvm_vcpu *vcpu);
+ void (*inject_page_fault)(struct kvm_vcpu *vcpu,
+ unsigned long addr, u32 err_code);
int (*run)(struct kvm_vcpu *vcpu, struct kvm_run *run);
int (*vcpu_setup)(struct kvm_vcpu *vcpu);
Index: linux-2.6/drivers/kvm/mmu.c
===================================================================
--- linux-2.6.orig/drivers/kvm/mmu.c
+++ linux-2.6/drivers/kvm/mmu.c
@@ -455,32 +455,7 @@ static void inject_page_fault(struct kvm
u64 addr,
u32 err_code)
{
- u32 vect_info = vmcs_read32(IDT_VECTORING_INFO_FIELD);
-
- pgprintk("inject_page_fault: 0x%llx err 0x%x\n", addr, err_code);
-
- ++kvm_stat.pf_guest;
-
- if (is_page_fault(vect_info)) {
- printk(KERN_DEBUG "inject_page_fault: "
- "double fault 0x%llx @ 0x%lx\n",
- addr, vmcs_readl(GUEST_RIP));
- vmcs_write32(VM_ENTRY_EXCEPTION_ERROR_CODE, 0);
- vmcs_write32(VM_ENTRY_INTR_INFO_FIELD,
- DF_VECTOR |
- INTR_TYPE_EXCEPTION |
- INTR_INFO_DELIEVER_CODE_MASK |
- INTR_INFO_VALID_MASK);
- return;
- }
- vcpu->cr2 = addr;
- vmcs_write32(VM_ENTRY_EXCEPTION_ERROR_CODE, err_code);
- vmcs_write32(VM_ENTRY_INTR_INFO_FIELD,
- PF_VECTOR |
- INTR_TYPE_EXCEPTION |
- INTR_INFO_DELIEVER_CODE_MASK |
- INTR_INFO_VALID_MASK);
-
+ kvm_arch_ops->inject_page_fault(vcpu, addr, err_code);
}
static inline int fix_read_pf(u64 *shadow_ent)
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -1624,6 +1624,36 @@ static void vmx_flush_tlb(struct kvm_vcp
vmcs_writel(GUEST_CR3, vmcs_readl(GUEST_CR3));
}
+static void vmx_inject_page_fault(struct kvm_vcpu *vcpu,
+ unsigned long addr,
+ u32 err_code)
+{
+ u32 vect_info = vmcs_read32(IDT_VECTORING_INFO_FIELD);
+
+ ++kvm_stat.pf_guest;
+
+ if (is_page_fault(vect_info)) {
+ printk(KERN_DEBUG "inject_page_fault: "
+ "double fault 0x%lx @ 0x%lx\n",
+ addr, vmcs_readl(GUEST_RIP));
+ vmcs_write32(VM_ENTRY_EXCEPTION_ERROR_CODE, 0);
+ vmcs_write32(VM_ENTRY_INTR_INFO_FIELD,
+ DF_VECTOR |
+ INTR_TYPE_EXCEPTION |
+ INTR_INFO_DELIEVER_CODE_MASK |
+ INTR_INFO_VALID_MASK);
+ return;
+ }
+ vcpu->cr2 = addr;
+ vmcs_write32(VM_ENTRY_EXCEPTION_ERROR_CODE, err_code);
+ vmcs_write32(VM_ENTRY_INTR_INFO_FIELD,
+ PF_VECTOR |
+ INTR_TYPE_EXCEPTION |
+ INTR_INFO_DELIEVER_CODE_MASK |
+ INTR_INFO_VALID_MASK);
+
+}
+
static struct kvm_arch_ops vmx_arch_ops = {
.cpu_has_kvm_support = cpu_has_kvm_support,
.disabled_by_bios = vmx_disabled_by_bios,
@@ -1650,6 +1680,7 @@ static struct kvm_arch_ops vmx_arch_ops
.decache_regs = vcpu_put_rsp_rip,
.flush_tlb = vmx_flush_tlb,
+ .inject_page_fault = vmx_inject_page_fault,
.run = vmx_vcpu_run,
.skip_emulated_instruction = skip_emulated_instruction,
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 22/38] KVM: Make inject_gp() an arch operation
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (20 preceding siblings ...)
2006-11-27 12:31 ` [PATCH 21/38] KVM: Make inject_page_fault() an arch operation Avi Kivity
@ 2006-11-27 12:32 ` Avi Kivity
2006-11-27 12:33 ` [PATCH 23/38] KVM: Use the idt and gdt accessors in realmode emulation Avi Kivity
` (15 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:32 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -284,6 +284,8 @@ struct kvm_arch_ops {
void (*inject_page_fault)(struct kvm_vcpu *vcpu,
unsigned long addr, u32 err_code);
+ void (*inject_gp)(struct kvm_vcpu *vcpu, unsigned err_code);
+
int (*run)(struct kvm_vcpu *vcpu, struct kvm_run *run);
int (*vcpu_setup)(struct kvm_vcpu *vcpu);
void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
@@ -345,8 +347,6 @@ void set_cr4(struct kvm_vcpu *vcpu, unsi
void set_cr8(struct kvm_vcpu *vcpu, unsigned long cr0);
void lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
-void inject_gp(struct kvm_vcpu *vcpu);
-
#ifdef __x86_64__
void set_efer(struct kvm_vcpu *vcpu, u64 efer);
#endif
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -485,18 +485,10 @@ void vmcs_writel(unsigned long field, un
}
EXPORT_SYMBOL_GPL(vmcs_writel);
-void inject_gp(struct kvm_vcpu *vcpu)
+static void inject_gp(struct kvm_vcpu *vcpu)
{
- printk(KERN_DEBUG "inject_general_protection: rip 0x%lx\n",
- vmcs_readl(GUEST_RIP));
- vmcs_write32(VM_ENTRY_EXCEPTION_ERROR_CODE, 0);
- vmcs_write32(VM_ENTRY_INTR_INFO_FIELD,
- GP_VECTOR |
- INTR_TYPE_EXCEPTION |
- INTR_INFO_DELIEVER_CODE_MASK |
- INTR_INFO_VALID_MASK);
+ kvm_arch_ops->inject_gp(vcpu, 0);
}
-EXPORT_SYMBOL_GPL(inject_gp);
/*
* reads and returns guest's timestamp counter "register"
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -80,6 +80,18 @@ static void skip_emulated_instruction(st
interruptibility & ~3);
}
+static void vmx_inject_gp(struct kvm_vcpu *vcpu, unsigned error_code)
+{
+ printk(KERN_DEBUG "inject_general_protection: rip 0x%lx\n",
+ vmcs_readl(GUEST_RIP));
+ vmcs_write32(VM_ENTRY_EXCEPTION_ERROR_CODE, error_code);
+ vmcs_write32(VM_ENTRY_INTR_INFO_FIELD,
+ GP_VECTOR |
+ INTR_TYPE_EXCEPTION |
+ INTR_INFO_DELIEVER_CODE_MASK |
+ INTR_INFO_VALID_MASK);
+}
+
static void reload_tss(void)
{
#ifndef __x86_64__
@@ -1318,7 +1330,7 @@ static int handle_rdmsr(struct kvm_vcpu
u64 data;
if (vmx_get_msr(vcpu, ecx, &data)) {
- inject_gp(vcpu);
+ vmx_inject_gp(vcpu, 0);
return 1;
}
@@ -1336,7 +1348,7 @@ static int handle_wrmsr(struct kvm_vcpu
| ((u64)(vcpu->regs[VCPU_REGS_RDX] & -1u) << 32);
if (vmx_set_msr(vcpu, ecx, data) != 0) {
- inject_gp(vcpu);
+ vmx_inject_gp(vcpu, 0);
return 1;
}
@@ -1682,6 +1694,8 @@ static struct kvm_arch_ops vmx_arch_ops
.flush_tlb = vmx_flush_tlb,
.inject_page_fault = vmx_inject_page_fault,
+ .inject_gp = vmx_inject_gp,
+
.run = vmx_vcpu_run,
.skip_emulated_instruction = skip_emulated_instruction,
.vcpu_setup = vmx_vcpu_setup,
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 23/38] KVM: Use the idt and gdt accessors in realmode emulation
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (21 preceding siblings ...)
2006-11-27 12:32 ` [PATCH 22/38] KVM: Make inject_gp() " Avi Kivity
@ 2006-11-27 12:33 ` Avi Kivity
2006-11-27 12:34 ` [PATCH 24/38] KVM: Use the general purpose register accessors rather than direct access Avi Kivity
` (14 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:33 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -1225,14 +1225,16 @@ static u64 mk_cr_64(u64 curr_cr, u32 new
void realmode_lgdt(struct kvm_vcpu *vcpu, u16 limit, unsigned long base)
{
- vmcs_writel(GUEST_GDTR_BASE, base);
- vmcs_write32(GUEST_GDTR_LIMIT, limit);
+ struct descriptor_table dt = { limit, base };
+
+ kvm_arch_ops->set_gdt(vcpu, &dt);
}
void realmode_lidt(struct kvm_vcpu *vcpu, u16 limit, unsigned long base)
{
- vmcs_writel(GUEST_IDTR_BASE, base);
- vmcs_write32(GUEST_IDTR_LIMIT, limit);
+ struct descriptor_table dt = { limit, base };
+
+ kvm_arch_ops->set_idt(vcpu, &dt);
}
void realmode_lmsw(struct kvm_vcpu *vcpu, unsigned long msw,
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 24/38] KVM: Use the general purpose register accessors rather than direct access
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (22 preceding siblings ...)
2006-11-27 12:33 ` [PATCH 23/38] KVM: Use the idt and gdt accessors in realmode emulation Avi Kivity
@ 2006-11-27 12:34 ` Avi Kivity
2006-11-27 12:35 ` [PATCH 25/38] KVM: Move the vmx tsc accessors to vmx.c Avi Kivity
` (13 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:34 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -1130,7 +1130,7 @@ static void report_emulation_failure(str
{
static int reported;
u8 opcodes[4];
- unsigned long rip = vmcs_readl(GUEST_RIP);
+ unsigned long rip = ctxt->vcpu->rip;
unsigned long rip_linear;
rip_linear = rip + get_segment_base(ctxt->vcpu, VCPU_SREG_CS);
@@ -1406,13 +1406,15 @@ static int kvm_dev_ioctl_get_regs(struct
if (!vcpu)
return -ENOENT;
+ kvm_arch_ops->cache_regs(vcpu);
+
regs->rax = vcpu->regs[VCPU_REGS_RAX];
regs->rbx = vcpu->regs[VCPU_REGS_RBX];
regs->rcx = vcpu->regs[VCPU_REGS_RCX];
regs->rdx = vcpu->regs[VCPU_REGS_RDX];
regs->rsi = vcpu->regs[VCPU_REGS_RSI];
regs->rdi = vcpu->regs[VCPU_REGS_RDI];
- regs->rsp = vmcs_readl(GUEST_RSP);
+ regs->rsp = vcpu->regs[VCPU_REGS_RSP];
regs->rbp = vcpu->regs[VCPU_REGS_RBP];
#ifdef __x86_64__
regs->r8 = vcpu->regs[VCPU_REGS_R8];
@@ -1425,7 +1427,7 @@ static int kvm_dev_ioctl_get_regs(struct
regs->r15 = vcpu->regs[VCPU_REGS_R15];
#endif
- regs->rip = vmcs_readl(GUEST_RIP);
+ regs->rip = vcpu->rip;
regs->rflags = vmcs_readl(GUEST_RFLAGS);
/*
@@ -1456,7 +1458,7 @@ static int kvm_dev_ioctl_set_regs(struct
vcpu->regs[VCPU_REGS_RDX] = regs->rdx;
vcpu->regs[VCPU_REGS_RSI] = regs->rsi;
vcpu->regs[VCPU_REGS_RDI] = regs->rdi;
- vmcs_writel(GUEST_RSP, regs->rsp);
+ vcpu->regs[VCPU_REGS_RSP] = regs->rsp;
vcpu->regs[VCPU_REGS_RBP] = regs->rbp;
#ifdef __x86_64__
vcpu->regs[VCPU_REGS_R8] = regs->r8;
@@ -1469,9 +1471,11 @@ static int kvm_dev_ioctl_set_regs(struct
vcpu->regs[VCPU_REGS_R15] = regs->r15;
#endif
- vmcs_writel(GUEST_RIP, regs->rip);
+ vcpu->rip = regs->rip;
vmcs_writel(GUEST_RFLAGS, regs->rflags);
+ kvm_arch_ops->decache_regs(vcpu);
+
vcpu_put(vcpu);
return 0;
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 25/38] KVM: Move the vmx tsc accessors to vmx.c
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (23 preceding siblings ...)
2006-11-27 12:34 ` [PATCH 24/38] KVM: Use the general purpose register accessors rather than direct access Avi Kivity
@ 2006-11-27 12:35 ` Avi Kivity
2006-11-27 12:36 ` [PATCH 26/38] KVM: Access rflags through an arch operation Avi Kivity
` (12 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:35 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -490,33 +490,6 @@ static void inject_gp(struct kvm_vcpu *v
kvm_arch_ops->inject_gp(vcpu, 0);
}
-/*
- * reads and returns guest's timestamp counter "register"
- * guest_tsc = host_tsc + tsc_offset -- 21.3
- */
-u64 guest_read_tsc(void)
-{
- u64 host_tsc, tsc_offset;
-
- rdtscll(host_tsc);
- tsc_offset = vmcs_read64(TSC_OFFSET);
- return host_tsc + tsc_offset;
-}
-EXPORT_SYMBOL_GPL(guest_read_tsc);
-
-/*
- * writes 'guest_tsc' into guest's timestamp counter "register"
- * guest_tsc = host_tsc + tsc_offset ==> tsc_offset = guest_tsc - host_tsc
- */
-void guest_write_tsc(u64 guest_tsc)
-{
- u64 host_tsc;
-
- rdtscll(host_tsc);
- vmcs_write64(TSC_OFFSET, guest_tsc - host_tsc);
-}
-EXPORT_SYMBOL_GPL(guest_write_tsc);
-
static int pdptrs_have_reserved_bits_set(struct kvm_vcpu *vcpu,
unsigned long cr3)
{
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -57,8 +57,6 @@ static const u32 vmx_msr_index[] = {
};
#define NR_VMX_MSR (sizeof(vmx_msr_index) / sizeof(*vmx_msr_index))
-u64 guest_read_tsc(void);
-void guest_write_tsc(u64 guest_tsc);
struct vmx_msr_entry *find_msr_entry(struct kvm_vcpu *vcpu, u32 msr);
static void skip_emulated_instruction(struct kvm_vcpu *vcpu)
@@ -92,6 +90,31 @@ static void vmx_inject_gp(struct kvm_vcp
INTR_INFO_VALID_MASK);
}
+/*
+ * reads and returns guest's timestamp counter "register"
+ * guest_tsc = host_tsc + tsc_offset -- 21.3
+ */
+static u64 guest_read_tsc(void)
+{
+ u64 host_tsc, tsc_offset;
+
+ rdtscll(host_tsc);
+ tsc_offset = vmcs_read64(TSC_OFFSET);
+ return host_tsc + tsc_offset;
+}
+
+/*
+ * writes 'guest_tsc' into guest's timestamp counter "register"
+ * guest_tsc = host_tsc + tsc_offset ==> tsc_offset = guest_tsc - host_tsc
+ */
+static void guest_write_tsc(u64 guest_tsc)
+{
+ u64 host_tsc;
+
+ rdtscll(host_tsc);
+ vmcs_write64(TSC_OFFSET, guest_tsc - host_tsc);
+}
+
static void reload_tss(void)
{
#ifndef __x86_64__
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 26/38] KVM: Access rflags through an arch operation
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (24 preceding siblings ...)
2006-11-27 12:35 ` [PATCH 25/38] KVM: Move the vmx tsc accessors to vmx.c Avi Kivity
@ 2006-11-27 12:36 ` Avi Kivity
2006-11-27 12:37 ` [PATCH 27/38] KVM: Move the vmx segment field definitions to vmx.c Avi Kivity
` (11 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:36 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -279,6 +279,8 @@ struct kvm_arch_ops {
void (*set_gdt)(struct kvm_vcpu *vcpu, struct descriptor_table *dt);
void (*cache_regs)(struct kvm_vcpu *vcpu);
void (*decache_regs)(struct kvm_vcpu *vcpu);
+ unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
+ void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
void (*flush_tlb)(struct kvm_vcpu *vcpu);
void (*inject_page_fault)(struct kvm_vcpu *vcpu,
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -1141,7 +1141,7 @@ int emulate_instruction(struct kvm_vcpu
cs_ar = vmcs_read32(GUEST_CS_AR_BYTES);
emulate_ctxt.vcpu = vcpu;
- emulate_ctxt.eflags = vmcs_readl(GUEST_RFLAGS);
+ emulate_ctxt.eflags = kvm_arch_ops->get_rflags(vcpu);
emulate_ctxt.cr2 = cr2;
emulate_ctxt.mode = (emulate_ctxt.eflags & X86_EFLAGS_VM)
? X86EMUL_MODE_REAL : (cs_ar & AR_L_MASK)
@@ -1182,7 +1182,7 @@ int emulate_instruction(struct kvm_vcpu
}
kvm_arch_ops->decache_regs(vcpu);
- vmcs_writel(GUEST_RFLAGS, emulate_ctxt.eflags);
+ kvm_arch_ops->set_rflags(vcpu, emulate_ctxt.eflags);
if (vcpu->mmio_is_write)
return EMULATE_DO_MMIO;
@@ -1214,7 +1214,7 @@ void realmode_lmsw(struct kvm_vcpu *vcpu
unsigned long *rflags)
{
lmsw(vcpu, msw);
- *rflags = vmcs_readl(GUEST_RFLAGS);
+ *rflags = kvm_arch_ops->get_rflags(vcpu);
}
unsigned long realmode_get_cr(struct kvm_vcpu *vcpu, int cr)
@@ -1240,7 +1240,7 @@ void realmode_set_cr(struct kvm_vcpu *vc
switch (cr) {
case 0:
set_cr0(vcpu, mk_cr_64(vcpu->cr0, val));
- *rflags = vmcs_readl(GUEST_RFLAGS);
+ *rflags = kvm_arch_ops->get_rflags(vcpu);
break;
case 2:
vcpu->cr2 = val;
@@ -1401,7 +1401,7 @@ static int kvm_dev_ioctl_get_regs(struct
#endif
regs->rip = vcpu->rip;
- regs->rflags = vmcs_readl(GUEST_RFLAGS);
+ regs->rflags = kvm_arch_ops->get_rflags(vcpu);
/*
* Don't leak debug flags in case they were set for guest debugging
@@ -1445,7 +1445,7 @@ static int kvm_dev_ioctl_set_regs(struct
#endif
vcpu->rip = regs->rip;
- vmcs_writel(GUEST_RFLAGS, regs->rflags);
+ kvm_arch_ops->set_rflags(vcpu, regs->rflags);
kvm_arch_ops->decache_regs(vcpu);
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -59,6 +59,16 @@ static const u32 vmx_msr_index[] = {
struct vmx_msr_entry *find_msr_entry(struct kvm_vcpu *vcpu, u32 msr);
+static unsigned long vmx_get_rflags(struct kvm_vcpu *vcpu)
+{
+ return vmcs_readl(GUEST_RFLAGS);
+}
+
+static void vmx_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags)
+{
+ vmcs_writel(GUEST_RFLAGS, rflags);
+}
+
static void skip_emulated_instruction(struct kvm_vcpu *vcpu)
{
unsigned long rip;
@@ -1713,6 +1723,8 @@ static struct kvm_arch_ops vmx_arch_ops
.set_gdt = vmx_set_gdt,
.cache_regs = vcpu_load_rsp_rip,
.decache_regs = vcpu_put_rsp_rip,
+ .get_rflags = vmx_get_rflags,
+ .set_rflags = vmx_set_rflags,
.flush_tlb = vmx_flush_tlb,
.inject_page_fault = vmx_inject_page_fault,
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 27/38] KVM: Move the vmx segment field definitions to vmx.c
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (25 preceding siblings ...)
2006-11-27 12:36 ` [PATCH 26/38] KVM: Access rflags through an arch operation Avi Kivity
@ 2006-11-27 12:37 ` Avi Kivity
2006-11-27 12:38 ` [PATCH 28/38] KVM: Add an arch accessor for cs d/b and l bits Avi Kivity
` (10 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:37 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -65,31 +65,6 @@ static struct kvm_stats_debugfs_item {
static struct dentry *debugfs_dir;
-#define VMX_SEGMENT_FIELD(seg) \
- [VCPU_SREG_##seg] { \
- GUEST_##seg##_SELECTOR, \
- GUEST_##seg##_BASE, \
- GUEST_##seg##_LIMIT, \
- GUEST_##seg##_AR_BYTES, \
- }
-
-struct kvm_vmx_segment_field {
- unsigned selector;
- unsigned base;
- unsigned limit;
- unsigned ar_bytes;
-} kvm_vmx_segment_fields[] = {
- VMX_SEGMENT_FIELD(CS),
- VMX_SEGMENT_FIELD(DS),
- VMX_SEGMENT_FIELD(ES),
- VMX_SEGMENT_FIELD(FS),
- VMX_SEGMENT_FIELD(GS),
- VMX_SEGMENT_FIELD(SS),
- VMX_SEGMENT_FIELD(TR),
- VMX_SEGMENT_FIELD(LDTR),
-};
-EXPORT_SYMBOL_GPL(kvm_vmx_segment_fields);
-
#define MAX_IO_MSRS 256
#define CR0_RESEVED_BITS 0xffffffff1ffaffc0ULL
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -42,12 +42,29 @@ extern struct vmcs_descriptor {
u32 revision_id;
} vmcs_descriptor;
-extern struct kvm_vmx_segment_field {
+#define VMX_SEGMENT_FIELD(seg) \
+ [VCPU_SREG_##seg] { \
+ GUEST_##seg##_SELECTOR, \
+ GUEST_##seg##_BASE, \
+ GUEST_##seg##_LIMIT, \
+ GUEST_##seg##_AR_BYTES, \
+ }
+
+static struct kvm_vmx_segment_field {
unsigned selector;
unsigned base;
unsigned limit;
unsigned ar_bytes;
-} kvm_vmx_segment_fields[];
+} kvm_vmx_segment_fields[] = {
+ VMX_SEGMENT_FIELD(CS),
+ VMX_SEGMENT_FIELD(DS),
+ VMX_SEGMENT_FIELD(ES),
+ VMX_SEGMENT_FIELD(FS),
+ VMX_SEGMENT_FIELD(GS),
+ VMX_SEGMENT_FIELD(SS),
+ VMX_SEGMENT_FIELD(TR),
+ VMX_SEGMENT_FIELD(LDTR),
+};
static const u32 vmx_msr_index[] = {
#ifdef __x86_64__
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 28/38] KVM: Add an arch accessor for cs d/b and l bits
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (26 preceding siblings ...)
2006-11-27 12:37 ` [PATCH 27/38] KVM: Move the vmx segment field definitions to vmx.c Avi Kivity
@ 2006-11-27 12:38 ` Avi Kivity
2006-11-27 12:39 ` [PATCH 29/38] KVM: Add a set_cr0_no_modeswitch() arch accessor Avi Kivity
` (9 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:38 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
These are used for detecting the current processor mode.
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -269,6 +269,7 @@ struct kvm_arch_ops {
struct kvm_segment *var, int seg);
void (*set_segment)(struct kvm_vcpu *vcpu,
struct kvm_segment *var, int seg);
+ void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
void (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -517,15 +517,16 @@ void set_cr0(struct kvm_vcpu *vcpu, unsi
if (!is_paging(vcpu) && (cr0 & CR0_PG_MASK)) {
#ifdef __x86_64__
if ((vcpu->shadow_efer & EFER_LME)) {
- u32 guest_cs_ar;
+ int cs_db, cs_l;
+
if (!is_pae(vcpu)) {
printk(KERN_DEBUG "set_cr0: #GP, start paging "
"in long mode while PAE is disabled\n");
inject_gp(vcpu);
return;
}
- guest_cs_ar = vmcs_read32(GUEST_CS_AR_BYTES);
- if (guest_cs_ar & SEGMENT_AR_L_MASK) {
+ kvm_arch_ops->get_cs_db_l_bits(vcpu, &cs_db, &cs_l);
+ if (cs_l) {
printk(KERN_DEBUG "set_cr0: #GP, start paging "
"in long mode while CS.L == 1\n");
inject_gp(vcpu);
@@ -1109,18 +1110,18 @@ int emulate_instruction(struct kvm_vcpu
{
struct x86_emulate_ctxt emulate_ctxt;
int r;
- u32 cs_ar;
+ int cs_db, cs_l;
kvm_arch_ops->cache_regs(vcpu);
- cs_ar = vmcs_read32(GUEST_CS_AR_BYTES);
+ kvm_arch_ops->get_cs_db_l_bits(vcpu, &cs_db, &cs_l);
emulate_ctxt.vcpu = vcpu;
emulate_ctxt.eflags = kvm_arch_ops->get_rflags(vcpu);
emulate_ctxt.cr2 = cr2;
emulate_ctxt.mode = (emulate_ctxt.eflags & X86_EFLAGS_VM)
- ? X86EMUL_MODE_REAL : (cs_ar & AR_L_MASK)
- ? X86EMUL_MODE_PROT64 : (cs_ar & AR_DB_MASK)
+ ? X86EMUL_MODE_REAL : cs_l
+ ? X86EMUL_MODE_PROT64 : cs_db
? X86EMUL_MODE_PROT32 : X86EMUL_MODE_PROT16;
if (emulate_ctxt.mode == X86EMUL_MODE_PROT64) {
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -698,6 +698,14 @@ static void vmx_set_segment(struct kvm_v
vmcs_write32(sf->ar_bytes, ar);
}
+static void vmx_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l)
+{
+ u32 ar = vmcs_read32(GUEST_CS_AR_BYTES);
+
+ *db = (ar >> 14) & 1;
+ *l = (ar >> 13) & 1;
+}
+
static void vmx_get_idt(struct kvm_vcpu *vcpu, struct descriptor_table *dt)
{
dt->limit = vmcs_read32(GUEST_IDTR_LIMIT);
@@ -1730,6 +1738,7 @@ static struct kvm_arch_ops vmx_arch_ops
.get_segment_base = vmx_get_segment_base,
.get_segment = vmx_get_segment,
.set_segment = vmx_set_segment,
+ .get_cs_db_l_bits = vmx_get_cs_db_l_bits,
.set_cr0 = vmx_set_cr0,
.set_cr3 = vmx_set_cr3,
.set_cr4 = vmx_set_cr4,
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 29/38] KVM: Add a set_cr0_no_modeswitch() arch accessor
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (27 preceding siblings ...)
2006-11-27 12:38 ` [PATCH 28/38] KVM: Add an arch accessor for cs d/b and l bits Avi Kivity
@ 2006-11-27 12:39 ` Avi Kivity
2006-11-27 12:40 ` [PATCH 30/38] KVM: Make vcpu_load() and vcpu_put() arch operations Avi Kivity
` (8 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:39 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
A side effect of using vm86 mode to simulate real mode is that we mangle
the segment registers on mode switching. The new set_cr0_no_modeswitch()
accessor avoids the mangling when the restore is due to loading the virtual
machine from a file.
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -271,6 +271,8 @@ struct kvm_arch_ops {
struct kvm_segment *var, int seg);
void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
+ void (*set_cr0_no_modeswitch)(struct kvm_vcpu *vcpu,
+ unsigned long cr0);
void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
void (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
@@ -292,8 +294,6 @@ struct kvm_arch_ops {
int (*run)(struct kvm_vcpu *vcpu, struct kvm_run *run);
int (*vcpu_setup)(struct kvm_vcpu *vcpu);
void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
-
- void (*update_exception_bitmap)(struct kvm_vcpu *vcpu); /* hack */
};
extern struct kvm_stat kvm_stat;
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -1529,12 +1529,7 @@ static int kvm_dev_ioctl_set_sregs(struc
vcpu->apic_base = sregs->apic_base;
mmu_reset_needed |= vcpu->cr0 != sregs->cr0;
- vcpu->rmode.active = ((sregs->cr0 & CR0_PE_MASK) == 0);
- kvm_arch_ops->update_exception_bitmap(vcpu);
- vmcs_writel(CR0_READ_SHADOW, sregs->cr0);
- vmcs_writel(GUEST_CR0,
- (sregs->cr0 & ~KVM_GUEST_CR0_MASK) | KVM_VM_CR0_ALWAYS_ON);
- vcpu->cr0 = sregs->cr0;
+ kvm_arch_ops->set_cr0_no_modeswitch(vcpu, sregs->cr0);
mmu_reset_needed |= vcpu->cr4 != sregs->cr4;
kvm_arch_ops->set_cr4(vcpu, sregs->cr4);
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -607,6 +607,19 @@ static void vmx_set_cr0(struct kvm_vcpu
vcpu->cr0 = cr0;
}
+/*
+ * Used when restoring the VM to avoid corrupting segment registers
+ */
+static void vmx_set_cr0_no_modeswitch(struct kvm_vcpu *vcpu, unsigned long cr0)
+{
+ vcpu->rmode.active = ((cr0 & CR0_PE_MASK) == 0);
+ update_exception_bitmap(vcpu);
+ vmcs_writel(CR0_READ_SHADOW, cr0);
+ vmcs_writel(GUEST_CR0,
+ (cr0 & ~KVM_GUEST_CR0_MASK) | KVM_VM_CR0_ALWAYS_ON);
+ vcpu->cr0 = cr0;
+}
+
static void vmx_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3)
{
vmcs_writel(GUEST_CR3, cr3);
@@ -1740,6 +1753,7 @@ static struct kvm_arch_ops vmx_arch_ops
.set_segment = vmx_set_segment,
.get_cs_db_l_bits = vmx_get_cs_db_l_bits,
.set_cr0 = vmx_set_cr0,
+ .set_cr0_no_modeswitch = vmx_set_cr0_no_modeswitch,
.set_cr3 = vmx_set_cr3,
.set_cr4 = vmx_set_cr4,
.set_efer = vmx_set_efer,
@@ -1760,8 +1774,6 @@ static struct kvm_arch_ops vmx_arch_ops
.run = vmx_vcpu_run,
.skip_emulated_instruction = skip_emulated_instruction,
.vcpu_setup = vmx_vcpu_setup,
-
- .update_exception_bitmap = update_exception_bitmap,
};
static int __init vmx_init(void)
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 30/38] KVM: Make vcpu_load() and vcpu_put() arch operations
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (28 preceding siblings ...)
2006-11-27 12:39 ` [PATCH 29/38] KVM: Add a set_cr0_no_modeswitch() arch accessor Avi Kivity
@ 2006-11-27 12:40 ` Avi Kivity
2006-11-27 12:41 ` [PATCH 31/38] KVM: Make vcpu creation and destruction " Avi Kivity
` (7 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:40 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -260,6 +260,9 @@ struct kvm_arch_ops {
int (*hardware_setup)(void); /* __init */
void (*hardware_unsetup)(void); /* __exit */
+ struct kvm_vcpu *(*vcpu_load)(struct kvm_vcpu *vcpu);
+ void (*vcpu_put)(struct kvm_vcpu *vcpu);
+
int (*set_guest_debug)(struct kvm_vcpu *vcpu,
struct kvm_debug_guest *dbg);
int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
@@ -296,6 +299,8 @@ struct kvm_arch_ops {
void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
};
+void __vcpu_clear(void *arg); /* temporary hack */
+
extern struct kvm_stat kvm_stat;
extern struct kvm_arch_ops *kvm_arch_ops;
@@ -373,6 +378,8 @@ int kvm_write_guest(struct kvm_vcpu *vcp
void vmcs_writel(unsigned long field, unsigned long value);
unsigned long vmcs_readl(unsigned long field);
+unsigned long segment_base(u16 selector);
+
static inline struct page *_gfn_to_page(struct kvm *kvm, gfn_t gfn)
{
struct kvm_memory_slot *slot = gfn_to_memslot(kvm, gfn);
@@ -501,6 +508,18 @@ static inline void get_idt(struct descri
asm ("sidt %0" : "=m"(*table));
}
+static inline void get_gdt(struct descriptor_table *table)
+{
+ asm ("sgdt %0" : "=m"(*table));
+}
+
+static inline unsigned long read_tr_base(void)
+{
+ u16 tr;
+ asm ("str %0" : "=g"(tr));
+ return segment_base(tr);
+}
+
#ifdef __x86_64__
static inline unsigned long read_msr(unsigned long msr)
{
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -84,11 +84,6 @@ struct vmx_msr_entry *find_msr_entry(str
}
EXPORT_SYMBOL_GPL(find_msr_entry);
-static void get_gdt(struct descriptor_table *table)
-{
- asm ("sgdt %0" : "=m"(*table));
-}
-
struct segment_descriptor {
u16 limit_low;
u16 base_low;
@@ -115,7 +110,7 @@ struct segment_descriptor_64 {
#endif
-static unsigned long segment_base(u16 selector)
+unsigned long segment_base(u16 selector)
{
struct descriptor_table gdt;
struct segment_descriptor *d;
@@ -141,17 +136,12 @@ static unsigned long segment_base(u16 se
#endif
return v;
}
-
-static unsigned long read_tr_base(void)
-{
- u16 tr;
- asm ("str %0" : "=g"(tr));
- return segment_base(tr);
-}
+EXPORT_SYMBOL_GPL(segment_base);
DEFINE_PER_CPU(struct vmcs *, vmxarea);
EXPORT_SYMBOL_GPL(per_cpu__vmxarea); /* temporary hack */
-static DEFINE_PER_CPU(struct vmcs *, current_vmcs);
+DEFINE_PER_CPU(struct vmcs *, current_vmcs);
+EXPORT_SYMBOL_GPL(per_cpu__current_vmcs); /* temporary hack */
struct vmcs_descriptor {
int size;
@@ -242,7 +232,7 @@ static void vmcs_clear(struct vmcs *vmcs
vmcs, phys_addr);
}
-static void __vcpu_clear(void *arg)
+void __vcpu_clear(void *arg)
{
struct kvm_vcpu *vcpu = arg;
int cpu = smp_processor_id();
@@ -252,6 +242,7 @@ static void __vcpu_clear(void *arg)
if (per_cpu(current_vmcs, cpu) == vcpu->vmcs)
per_cpu(current_vmcs, cpu) = 0;
}
+EXPORT_SYMBOL_GPL(__vcpu_clear);
static int vcpu_slot(struct kvm_vcpu *vcpu)
{
@@ -259,53 +250,6 @@ static int vcpu_slot(struct kvm_vcpu *vc
}
/*
- * Switches to specified vcpu, until a matching vcpu_put(), but assumes
- * vcpu mutex is already taken.
- */
-static struct kvm_vcpu *__vcpu_load(struct kvm_vcpu *vcpu)
-{
- u64 phys_addr = __pa(vcpu->vmcs);
- int cpu;
-
- cpu = get_cpu();
-
- if (vcpu->cpu != cpu) {
- smp_call_function(__vcpu_clear, vcpu, 0, 1);
- vcpu->launched = 0;
- }
-
- if (per_cpu(current_vmcs, cpu) != vcpu->vmcs) {
- u8 error;
-
- per_cpu(current_vmcs, cpu) = vcpu->vmcs;
- asm volatile (ASM_VMX_VMPTRLD_RAX "; setna %0"
- : "=g"(error) : "a"(&phys_addr), "m"(phys_addr)
- : "cc");
- if (error)
- printk(KERN_ERR "kvm: vmptrld %p/%llx fail\n",
- vcpu->vmcs, phys_addr);
- }
-
- if (vcpu->cpu != cpu) {
- struct descriptor_table dt;
- unsigned long sysenter_esp;
-
- vcpu->cpu = cpu;
- /*
- * Linux uses per-cpu TSS and GDT, so set these when switching
- * processors.
- */
- vmcs_writel(HOST_TR_BASE, read_tr_base()); /* 22.2.4 */
- get_gdt(&dt);
- vmcs_writel(HOST_GDTR_BASE, dt.base); /* 22.2.4 */
-
- rdmsrl(MSR_IA32_SYSENTER_ESP, sysenter_esp);
- vmcs_writel(HOST_IA32_SYSENTER_ESP, sysenter_esp); /* 22.2.3 */
- }
- return vcpu;
-}
-
-/*
* Switches to specified vcpu, until a matching vcpu_put()
*/
static struct kvm_vcpu *vcpu_load(struct kvm *kvm, int vcpu_slot)
@@ -317,11 +261,12 @@ static struct kvm_vcpu *vcpu_load(struct
mutex_unlock(&vcpu->mutex);
return 0;
}
- return __vcpu_load(vcpu);
+ return kvm_arch_ops->vcpu_load(vcpu);
}
static void vcpu_put(struct kvm_vcpu *vcpu)
{
+ kvm_arch_ops->vcpu_put(vcpu);
put_cpu();
mutex_unlock(&vcpu->mutex);
}
@@ -698,7 +643,7 @@ static int kvm_dev_ioctl_create_vcpu(str
vcpu->vmcs = vmcs;
vcpu->launched = 0;
- __vcpu_load(vcpu);
+ kvm_arch_ops->vcpu_load(vcpu);
r = kvm_arch_ops->vcpu_setup(vcpu);
if (r >= 0)
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -29,6 +29,7 @@ MODULE_AUTHOR("Qumranet");
MODULE_LICENSE("GPL");
DECLARE_PER_CPU(struct vmcs *, vmxarea);
+DECLARE_PER_CPU(struct vmcs *, current_vmcs);
#ifdef __x86_64__
#define HOST_IS_64 1
@@ -76,6 +77,58 @@ static const u32 vmx_msr_index[] = {
struct vmx_msr_entry *find_msr_entry(struct kvm_vcpu *vcpu, u32 msr);
+/*
+ * Switches to specified vcpu, until a matching vcpu_put(), but assumes
+ * vcpu mutex is already taken.
+ */
+static struct kvm_vcpu *vmx_vcpu_load(struct kvm_vcpu *vcpu)
+{
+ u64 phys_addr = __pa(vcpu->vmcs);
+ int cpu;
+
+ cpu = get_cpu();
+
+ if (vcpu->cpu != cpu) {
+ smp_call_function(__vcpu_clear, vcpu, 0, 1);
+ vcpu->launched = 0;
+ }
+
+ if (per_cpu(current_vmcs, cpu) != vcpu->vmcs) {
+ u8 error;
+
+ per_cpu(current_vmcs, cpu) = vcpu->vmcs;
+ asm volatile (ASM_VMX_VMPTRLD_RAX "; setna %0"
+ : "=g"(error) : "a"(&phys_addr), "m"(phys_addr)
+ : "cc");
+ if (error)
+ printk(KERN_ERR "kvm: vmptrld %p/%llx fail\n",
+ vcpu->vmcs, phys_addr);
+ }
+
+ if (vcpu->cpu != cpu) {
+ struct descriptor_table dt;
+ unsigned long sysenter_esp;
+
+ vcpu->cpu = cpu;
+ /*
+ * Linux uses per-cpu TSS and GDT, so set these when switching
+ * processors.
+ */
+ vmcs_writel(HOST_TR_BASE, read_tr_base()); /* 22.2.4 */
+ get_gdt(&dt);
+ vmcs_writel(HOST_GDTR_BASE, dt.base); /* 22.2.4 */
+
+ rdmsrl(MSR_IA32_SYSENTER_ESP, sysenter_esp);
+ vmcs_writel(HOST_IA32_SYSENTER_ESP, sysenter_esp); /* 22.2.3 */
+ }
+ return vcpu;
+}
+
+static void vmx_vcpu_put(struct kvm_vcpu *vcpu)
+{
+ put_cpu();
+}
+
static unsigned long vmx_get_rflags(struct kvm_vcpu *vcpu)
{
return vmcs_readl(GUEST_RFLAGS);
@@ -1745,6 +1798,9 @@ static struct kvm_arch_ops vmx_arch_ops
.hardware_enable = hardware_enable,
.hardware_disable = hardware_disable,
+ .vcpu_load = vmx_vcpu_load,
+ .vcpu_put = vmx_vcpu_put,
+
.set_guest_debug = set_guest_debug,
.get_msr = vmx_get_msr,
.set_msr = vmx_set_msr,
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 31/38] KVM: Make vcpu creation and destruction arch operations
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (29 preceding siblings ...)
2006-11-27 12:40 ` [PATCH 30/38] KVM: Make vcpu_load() and vcpu_put() arch operations Avi Kivity
@ 2006-11-27 12:41 ` Avi Kivity
2006-11-27 12:42 ` [PATCH 32/38] KVM: Move vmcs static variables to vmx.c Avi Kivity
` (6 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:41 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -260,6 +260,9 @@ struct kvm_arch_ops {
int (*hardware_setup)(void); /* __init */
void (*hardware_unsetup)(void); /* __exit */
+ int (*vcpu_create)(struct kvm_vcpu *vcpu);
+ void (*vcpu_free)(struct kvm_vcpu *vcpu);
+
struct kvm_vcpu *(*vcpu_load)(struct kvm_vcpu *vcpu);
void (*vcpu_put)(struct kvm_vcpu *vcpu);
@@ -299,8 +302,6 @@ struct kvm_arch_ops {
void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
};
-void __vcpu_clear(void *arg); /* temporary hack */
-
extern struct kvm_stat kvm_stat;
extern struct kvm_arch_ops *kvm_arch_ops;
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -219,31 +219,6 @@ int kvm_write_guest(struct kvm_vcpu *vcp
}
EXPORT_SYMBOL_GPL(kvm_write_guest);
-static void vmcs_clear(struct vmcs *vmcs)
-{
- u64 phys_addr = __pa(vmcs);
- u8 error;
-
- asm volatile (ASM_VMX_VMCLEAR_RAX "; setna %0"
- : "=g"(error) : "a"(&phys_addr), "m"(phys_addr)
- : "cc", "memory");
- if (error)
- printk(KERN_ERR "kvm: vmclear fail: %p/%llx\n",
- vmcs, phys_addr);
-}
-
-void __vcpu_clear(void *arg)
-{
- struct kvm_vcpu *vcpu = arg;
- int cpu = smp_processor_id();
-
- if (vcpu->cpu == cpu)
- vmcs_clear(vcpu->vmcs);
- if (per_cpu(current_vmcs, cpu) == vcpu->vmcs)
- per_cpu(current_vmcs, cpu) = 0;
-}
-EXPORT_SYMBOL_GPL(__vcpu_clear);
-
static int vcpu_slot(struct kvm_vcpu *vcpu)
{
return vcpu - vcpu->kvm->vcpus;
@@ -271,33 +246,6 @@ static void vcpu_put(struct kvm_vcpu *vc
mutex_unlock(&vcpu->mutex);
}
-struct vmcs *alloc_vmcs_cpu(int cpu)
-{
- int node = cpu_to_node(cpu);
- struct page *pages;
- struct vmcs *vmcs;
-
- pages = alloc_pages_node(node, GFP_KERNEL, vmcs_descriptor.order);
- if (!pages)
- return 0;
- vmcs = page_address(pages);
- memset(vmcs, 0, vmcs_descriptor.size);
- vmcs->revision_id = vmcs_descriptor.revision_id; /* vmcs revision id */
- return vmcs;
-}
-EXPORT_SYMBOL_GPL(alloc_vmcs_cpu);
-
-static struct vmcs *alloc_vmcs(void)
-{
- return alloc_vmcs_cpu(smp_processor_id());
-}
-
-void free_vmcs(struct vmcs *vmcs)
-{
- free_pages((unsigned long)vmcs, vmcs_descriptor.order);
-}
-EXPORT_SYMBOL_GPL(free_vmcs);
-
static int kvm_dev_open(struct inode *inode, struct file *filp)
{
struct kvm *kvm = kzalloc(sizeof(struct kvm), GFP_KERNEL);
@@ -350,18 +298,9 @@ static void kvm_free_physmem(struct kvm
kvm_free_physmem_slot(&kvm->memslots[i], 0);
}
-static void kvm_free_vmcs(struct kvm_vcpu *vcpu)
-{
- if (vcpu->vmcs) {
- on_each_cpu(__vcpu_clear, vcpu, 0, 1);
- free_vmcs(vcpu->vmcs);
- vcpu->vmcs = 0;
- }
-}
-
static void kvm_free_vcpu(struct kvm_vcpu *vcpu)
{
- kvm_free_vmcs(vcpu);
+ kvm_arch_ops->vcpu_free(vcpu);
kvm_mmu_destroy(vcpu);
}
@@ -613,7 +552,6 @@ static int kvm_dev_ioctl_create_vcpu(str
{
int r;
struct kvm_vcpu *vcpu;
- struct vmcs *vmcs;
r = -EINVAL;
if (n < 0 || n >= KVM_MAX_VCPUS)
@@ -634,14 +572,9 @@ static int kvm_dev_ioctl_create_vcpu(str
vcpu->cpu = -1; /* First load will set up TR */
vcpu->kvm = kvm;
- vmcs = alloc_vmcs();
- if (!vmcs) {
- mutex_unlock(&vcpu->mutex);
+ r = kvm_arch_ops->vcpu_create(vcpu);
+ if (r < 0)
goto out_free_vcpus;
- }
- vmcs_clear(vmcs);
- vcpu->vmcs = vmcs;
- vcpu->launched = 0;
kvm_arch_ops->vcpu_load(vcpu);
@@ -658,6 +591,7 @@ static int kvm_dev_ioctl_create_vcpu(str
out_free_vcpus:
kvm_free_vcpu(vcpu);
+ mutex_unlock(&vcpu->mutex);
out:
return r;
}
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -77,6 +77,30 @@ static const u32 vmx_msr_index[] = {
struct vmx_msr_entry *find_msr_entry(struct kvm_vcpu *vcpu, u32 msr);
+static void vmcs_clear(struct vmcs *vmcs)
+{
+ u64 phys_addr = __pa(vmcs);
+ u8 error;
+
+ asm volatile (ASM_VMX_VMCLEAR_RAX "; setna %0"
+ : "=g"(error) : "a"(&phys_addr), "m"(phys_addr)
+ : "cc", "memory");
+ if (error)
+ printk(KERN_ERR "kvm: vmclear fail: %p/%llx\n",
+ vmcs, phys_addr);
+}
+
+static void __vcpu_clear(void *arg)
+{
+ struct kvm_vcpu *vcpu = arg;
+ int cpu = smp_processor_id();
+
+ if (vcpu->cpu == cpu)
+ vmcs_clear(vcpu->vmcs);
+ if (per_cpu(current_vmcs, cpu) == vcpu->vmcs)
+ per_cpu(current_vmcs, cpu) = 0;
+}
+
/*
* Switches to specified vcpu, until a matching vcpu_put(), but assumes
* vcpu mutex is already taken.
@@ -449,7 +473,30 @@ static __init void setup_vmcs_descriptor
vmcs_descriptor.revision_id = vmx_msr_low;
};
-void free_vmcs(struct vmcs *vmcs);
+static struct vmcs *alloc_vmcs_cpu(int cpu)
+{
+ int node = cpu_to_node(cpu);
+ struct page *pages;
+ struct vmcs *vmcs;
+
+ pages = alloc_pages_node(node, GFP_KERNEL, vmcs_descriptor.order);
+ if (!pages)
+ return 0;
+ vmcs = page_address(pages);
+ memset(vmcs, 0, vmcs_descriptor.size);
+ vmcs->revision_id = vmcs_descriptor.revision_id; /* vmcs revision id */
+ return vmcs;
+}
+
+static struct vmcs *alloc_vmcs(void)
+{
+ return alloc_vmcs_cpu(smp_processor_id());
+}
+
+static void free_vmcs(struct vmcs *vmcs)
+{
+ free_pages((unsigned long)vmcs, vmcs_descriptor.order);
+}
static __exit void free_kvm_area(void)
{
@@ -1790,6 +1837,33 @@ static void vmx_inject_page_fault(struct
}
+static void vmx_free_vmcs(struct kvm_vcpu *vcpu)
+{
+ if (vcpu->vmcs) {
+ on_each_cpu(__vcpu_clear, vcpu, 0, 1);
+ free_vmcs(vcpu->vmcs);
+ vcpu->vmcs = 0;
+ }
+}
+
+static void vmx_free_vcpu(struct kvm_vcpu *vcpu)
+{
+ vmx_free_vmcs(vcpu);
+}
+
+static int vmx_create_vcpu(struct kvm_vcpu *vcpu)
+{
+ struct vmcs *vmcs;
+
+ vmcs = alloc_vmcs();
+ if (!vmcs)
+ return -ENOMEM;
+ vmcs_clear(vmcs);
+ vcpu->vmcs = vmcs;
+ vcpu->launched = 0;
+ return 0;
+}
+
static struct kvm_arch_ops vmx_arch_ops = {
.cpu_has_kvm_support = cpu_has_kvm_support,
.disabled_by_bios = vmx_disabled_by_bios,
@@ -1798,6 +1872,9 @@ static struct kvm_arch_ops vmx_arch_ops
.hardware_enable = hardware_enable,
.hardware_disable = hardware_disable,
+ .vcpu_create = vmx_create_vcpu,
+ .vcpu_free = vmx_free_vcpu,
+
.vcpu_load = vmx_vcpu_load,
.vcpu_put = vmx_vcpu_put,
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 32/38] KVM: Move vmcs static variables to vmx.c
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (30 preceding siblings ...)
2006-11-27 12:41 ` [PATCH 31/38] KVM: Make vcpu creation and destruction " Avi Kivity
@ 2006-11-27 12:42 ` Avi Kivity
2006-11-27 12:43 ` [PATCH 33/38] KVM: Make is_long_mode() an arch operation Avi Kivity
` (5 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:42 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -138,18 +138,6 @@ unsigned long segment_base(u16 selector)
}
EXPORT_SYMBOL_GPL(segment_base);
-DEFINE_PER_CPU(struct vmcs *, vmxarea);
-EXPORT_SYMBOL_GPL(per_cpu__vmxarea); /* temporary hack */
-DEFINE_PER_CPU(struct vmcs *, current_vmcs);
-EXPORT_SYMBOL_GPL(per_cpu__current_vmcs); /* temporary hack */
-
-struct vmcs_descriptor {
- int size;
- int order;
- u32 revision_id;
-} vmcs_descriptor;
-EXPORT_SYMBOL_GPL(vmcs_descriptor);
-
int kvm_read_guest(struct kvm_vcpu *vcpu,
gva_t addr,
unsigned long size,
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -28,8 +28,8 @@
MODULE_AUTHOR("Qumranet");
MODULE_LICENSE("GPL");
-DECLARE_PER_CPU(struct vmcs *, vmxarea);
-DECLARE_PER_CPU(struct vmcs *, current_vmcs);
+static DEFINE_PER_CPU(struct vmcs *, vmxarea);
+static DEFINE_PER_CPU(struct vmcs *, current_vmcs);
#ifdef __x86_64__
#define HOST_IS_64 1
@@ -37,7 +37,7 @@ DECLARE_PER_CPU(struct vmcs *, current_v
#define HOST_IS_64 0
#endif
-extern struct vmcs_descriptor {
+static struct vmcs_descriptor {
int size;
int order;
u32 revision_id;
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 33/38] KVM: Make is_long_mode() an arch operation
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (31 preceding siblings ...)
2006-11-27 12:42 ` [PATCH 32/38] KVM: Move vmcs static variables to vmx.c Avi Kivity
@ 2006-11-27 12:43 ` Avi Kivity
2006-11-27 12:44 ` [PATCH 34/38] KVM: Use the tlb flush arch operation instead of an inline Avi Kivity
` (4 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:43 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -275,6 +275,7 @@ struct kvm_arch_ops {
struct kvm_segment *var, int seg);
void (*set_segment)(struct kvm_vcpu *vcpu,
struct kvm_segment *var, int seg);
+ int (*is_long_mode)(struct kvm_vcpu *vcpu);
void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
void (*set_cr0_no_modeswitch)(struct kvm_vcpu *vcpu,
@@ -411,11 +412,6 @@ static inline void vmcs_write32(unsigned
vmcs_writel(field, value);
}
-static inline int is_long_mode(void)
-{
- return vmcs_read32(VM_ENTRY_CONTROLS) & VM_ENTRY_CONTROLS_IA32E_MASK;
-}
-
static inline int is_pae(struct kvm_vcpu *vcpu)
{
return vcpu->cr4 & CR4_PAE_MASK;
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -441,7 +441,7 @@ void set_cr4(struct kvm_vcpu *vcpu, unsi
return;
}
- if (is_long_mode()) {
+ if (kvm_arch_ops->is_long_mode(vcpu)) {
if (!(cr4 & CR4_PAE_MASK)) {
printk(KERN_DEBUG "set_cr4: #GP, clearing PAE while "
"in long mode\n");
@@ -468,7 +468,7 @@ EXPORT_SYMBOL_GPL(set_cr4);
void set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3)
{
- if (is_long_mode()) {
+ if (kvm_arch_ops->is_long_mode(vcpu)) {
if ( cr3 & CR3_L_MODE_RESEVED_BITS) {
printk(KERN_DEBUG "set_cr3: #GP, reserved bits\n");
inject_gp(vcpu);
Index: linux-2.6/drivers/kvm/mmu.c
===================================================================
--- linux-2.6.orig/drivers/kvm/mmu.c
+++ linux-2.6/drivers/kvm/mmu.c
@@ -589,7 +589,7 @@ static int init_kvm_mmu(struct kvm_vcpu
if (!is_paging(vcpu))
return nonpaging_init_context(vcpu);
- else if (is_long_mode())
+ else if (kvm_arch_ops->is_long_mode(vcpu))
return paging64_init_context(vcpu);
else if (is_pae(vcpu))
return paging32E_init_context(vcpu);
Index: linux-2.6/drivers/kvm/paging_tmpl.h
===================================================================
--- linux-2.6.orig/drivers/kvm/paging_tmpl.h
+++ linux-2.6/drivers/kvm/paging_tmpl.h
@@ -70,7 +70,7 @@ static void FNAME(init_walker)(struct gu
hpa = safe_gpa_to_hpa(vcpu, vcpu->cr3 & PT64_BASE_ADDR_MASK);
walker->table = kmap_atomic(pfn_to_page(hpa >> PAGE_SHIFT), KM_USER0);
- ASSERT((!is_long_mode() && is_pae(vcpu)) ||
+ ASSERT((!kvm_arch_ops->is_long_mode(vcpu) && is_pae(vcpu)) ||
(vcpu->cr3 & ~(PAGE_MASK | CR3_FLAGS_MASK)) == 0);
walker->table = (pt_element_t *)( (unsigned long)walker->table |
@@ -135,7 +135,7 @@ static pt_element_t *FNAME(fetch_guest)(
(walker->table[index] & PT_PAGE_SIZE_MASK) &&
(PTTYPE == 64 || is_pse(vcpu))))
return &walker->table[index];
- if (walker->level != 3 || is_long_mode())
+ if (walker->level != 3 || kvm_arch_ops->is_long_mode(vcpu))
walker->inherited_ar &= walker->table[index];
paddr = safe_gpa_to_hpa(vcpu, walker->table[index] & PT_BASE_ADDR_MASK);
kunmap_atomic(walker->table, KM_USER0);
@@ -204,7 +204,7 @@ static u64 *FNAME(fetch)(struct kvm_vcpu
shadow_addr = kvm_mmu_alloc_page(vcpu, shadow_ent);
if (!VALID_PAGE(shadow_addr))
return ERR_PTR(-ENOMEM);
- if (!is_long_mode() && level == 3)
+ if (!kvm_arch_ops->is_long_mode(vcpu) && level == 3)
*shadow_ent = shadow_addr |
(*guest_ent & (PT_PRESENT_MASK | PT_PWT_MASK | PT_PCD_MASK));
else {
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -811,6 +811,11 @@ static void vmx_set_segment(struct kvm_v
vmcs_write32(sf->ar_bytes, ar);
}
+static int vmx_is_long_mode(struct kvm_vcpu *vcpu)
+{
+ return vmcs_read32(VM_ENTRY_CONTROLS) & VM_ENTRY_CONTROLS_IA32E_MASK;
+}
+
static void vmx_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l)
{
u32 ar = vmcs_read32(GUEST_CS_AR_BYTES);
@@ -1884,6 +1889,7 @@ static struct kvm_arch_ops vmx_arch_ops
.get_segment_base = vmx_get_segment_base,
.get_segment = vmx_get_segment,
.set_segment = vmx_set_segment,
+ .is_long_mode = vmx_is_long_mode,
.get_cs_db_l_bits = vmx_get_cs_db_l_bits,
.set_cr0 = vmx_set_cr0,
.set_cr0_no_modeswitch = vmx_set_cr0_no_modeswitch,
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 34/38] KVM: Use the tlb flush arch operation instead of an inline
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (32 preceding siblings ...)
2006-11-27 12:43 ` [PATCH 33/38] KVM: Make is_long_mode() an arch operation Avi Kivity
@ 2006-11-27 12:44 ` Avi Kivity
2006-11-27 12:45 ` [PATCH 35/38] KVM: Remove guest_cpl() Avi Kivity
` (3 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:44 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -445,11 +445,6 @@ static inline int is_external_interrupt(
== (INTR_TYPE_EXT_INTR | INTR_INFO_VALID_MASK);
}
-static inline void flush_guest_tlb(struct kvm_vcpu *vcpu)
-{
- vmcs_writel(GUEST_CR3, vmcs_readl(GUEST_CR3));
-}
-
static inline int memslot_id(struct kvm *kvm, struct kvm_memory_slot *slot)
{
return slot - kvm->memslots;
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -775,7 +775,7 @@ static int kvm_dev_ioctl_get_dirty_log(s
if (!vcpu)
continue;
- flush_guest_tlb(vcpu);
+ kvm_arch_ops->flush_tlb(vcpu);
vcpu_put(vcpu);
}
}
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 35/38] KVM: Remove guest_cpl()
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (33 preceding siblings ...)
2006-11-27 12:44 ` [PATCH 34/38] KVM: Use the tlb flush arch operation instead of an inline Avi Kivity
@ 2006-11-27 12:45 ` Avi Kivity
2006-11-27 12:46 ` [PATCH 36/38] KVM: Move vmcs accessors to vmx.c Avi Kivity
` (2 subsequent siblings)
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:45 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
This was used only in debugging.
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -422,11 +422,6 @@ static inline int is_pse(struct kvm_vcpu
return vcpu->cr4 & CR4_PSE_MASK;
}
-static inline unsigned guest_cpl(void)
-{
- return vmcs_read16(GUEST_CS_SELECTOR) & SELECTOR_RPL_MASK;
-}
-
static inline int is_paging(struct kvm_vcpu *vcpu)
{
return vcpu->cr0 & CR0_PG_MASK;
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -1391,14 +1391,6 @@ static int handle_cr(struct kvm_vcpu *vc
int cr;
int reg;
-#ifdef KVM_DEBUG
- if (guest_cpl() != 0) {
- vcpu_printf(vcpu, "%s: not supervisor\n", __FUNCTION__);
- inject_gp(vcpu);
- return 1;
- }
-#endif
-
exit_qualification = vmcs_read64(EXIT_QUALIFICATION);
cr = exit_qualification & 15;
reg = (exit_qualification >> 8) & 15;
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 36/38] KVM: Move vmcs accessors to vmx.c
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (34 preceding siblings ...)
2006-11-27 12:45 ` [PATCH 35/38] KVM: Remove guest_cpl() Avi Kivity
@ 2006-11-27 12:46 ` Avi Kivity
2006-11-27 12:47 ` [PATCH 37/38] KVM: Move vmx helper inlines " Avi Kivity
2006-11-27 12:48 ` [PATCH 38/38] KVM: Remove vmx includes from arch independent code Avi Kivity
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:46 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -377,9 +377,6 @@ int kvm_write_guest(struct kvm_vcpu *vcp
unsigned long size,
void *data);
-void vmcs_writel(unsigned long field, unsigned long value);
-unsigned long vmcs_readl(unsigned long field);
-
unsigned long segment_base(u16 selector);
static inline struct page *_gfn_to_page(struct kvm *kvm, gfn_t gfn)
@@ -388,30 +385,6 @@ static inline struct page *_gfn_to_page(
return (slot) ? slot->phys_mem[gfn - slot->base_gfn] : 0;
}
-static inline u16 vmcs_read16(unsigned long field)
-{
- return vmcs_readl(field);
-}
-
-static inline u32 vmcs_read32(unsigned long field)
-{
- return vmcs_readl(field);
-}
-
-static inline u64 vmcs_read64(unsigned long field)
-{
-#ifdef __x86_64__
- return vmcs_readl(field);
-#else
- return vmcs_readl(field) | ((u64)vmcs_readl(field+1) << 32);
-#endif
-}
-
-static inline void vmcs_write32(unsigned long field, u32 value)
-{
- vmcs_writel(field, value);
-}
-
static inline int is_pae(struct kvm_vcpu *vcpu)
{
return vcpu->cr4 & CR4_PAE_MASK;
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -310,28 +310,6 @@ static int kvm_dev_release(struct inode
return 0;
}
-unsigned long vmcs_readl(unsigned long field)
-{
- unsigned long value;
-
- asm volatile (ASM_VMX_VMREAD_RDX_RAX
- : "=a"(value) : "d"(field) : "cc");
- return value;
-}
-EXPORT_SYMBOL_GPL(vmcs_readl);
-
-void vmcs_writel(unsigned long field, unsigned long value)
-{
- u8 error;
-
- asm volatile (ASM_VMX_VMWRITE_RAX_RDX "; setna %0"
- : "=q"(error) : "a"(value), "d"(field) : "cc" );
- if (error)
- printk(KERN_ERR "vmwrite error: reg %lx value %lx (err %d)\n",
- field, value, vmcs_read32(VM_INSTRUCTION_ERROR));
-}
-EXPORT_SYMBOL_GPL(vmcs_writel);
-
static void inject_gp(struct kvm_vcpu *vcpu)
{
kvm_arch_ops->inject_gp(vcpu, 0);
Index: linux-2.6/drivers/kvm/kvm_vmx.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_vmx.h
+++ linux-2.6/drivers/kvm/kvm_vmx.h
@@ -1,22 +1,6 @@
#ifndef __KVM_VMX_H
#define __KVM_VMX_H
-static inline void vmcs_write16(unsigned long field, u16 value)
-{
- vmcs_writel(field, value);
-}
-
-static inline void vmcs_write64(unsigned long field, u64 value)
-{
-#ifdef __x86_64__
- vmcs_writel(field, value);
-#else
- vmcs_writel(field, value);
- asm volatile ("");
- vmcs_writel(field+1, value >> 32);
-#endif
-}
-
#ifdef __x86_64__
/*
* avoid save/load MSR_SYSCALL_MASK and MSR_LSTAR by std vt
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -101,6 +101,66 @@ static void __vcpu_clear(void *arg)
per_cpu(current_vmcs, cpu) = 0;
}
+static unsigned long vmcs_readl(unsigned long field)
+{
+ unsigned long value;
+
+ asm volatile (ASM_VMX_VMREAD_RDX_RAX
+ : "=a"(value) : "d"(field) : "cc");
+ return value;
+}
+
+static u16 vmcs_read16(unsigned long field)
+{
+ return vmcs_readl(field);
+}
+
+static u32 vmcs_read32(unsigned long field)
+{
+ return vmcs_readl(field);
+}
+
+static u64 vmcs_read64(unsigned long field)
+{
+#ifdef __x86_64__
+ return vmcs_readl(field);
+#else
+ return vmcs_readl(field) | ((u64)vmcs_readl(field+1) << 32);
+#endif
+}
+
+static void vmcs_writel(unsigned long field, unsigned long value)
+{
+ u8 error;
+
+ asm volatile (ASM_VMX_VMWRITE_RAX_RDX "; setna %0"
+ : "=q"(error) : "a"(value), "d"(field) : "cc" );
+ if (error)
+ printk(KERN_ERR "vmwrite error: reg %lx value %lx (err %d)\n",
+ field, value, vmcs_read32(VM_INSTRUCTION_ERROR));
+}
+
+static void vmcs_write16(unsigned long field, u16 value)
+{
+ vmcs_writel(field, value);
+}
+
+static void vmcs_write32(unsigned long field, u32 value)
+{
+ vmcs_writel(field, value);
+}
+
+static void vmcs_write64(unsigned long field, u64 value)
+{
+#ifdef __x86_64__
+ vmcs_writel(field, value);
+#else
+ vmcs_writel(field, value);
+ asm volatile ("");
+ vmcs_writel(field+1, value >> 32);
+#endif
+}
+
/*
* Switches to specified vcpu, until a matching vcpu_put(), but assumes
* vcpu mutex is already taken.
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 37/38] KVM: Move vmx helper inlines to vmx.c
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (35 preceding siblings ...)
2006-11-27 12:46 ` [PATCH 36/38] KVM: Move vmcs accessors to vmx.c Avi Kivity
@ 2006-11-27 12:47 ` Avi Kivity
2006-11-27 12:48 ` [PATCH 38/38] KVM: Remove vmx includes from arch independent code Avi Kivity
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:47 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm.h
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm.h
+++ linux-2.6/drivers/kvm/kvm.h
@@ -400,19 +400,6 @@ static inline int is_paging(struct kvm_v
return vcpu->cr0 & CR0_PG_MASK;
}
-static inline int is_page_fault(u32 intr_info)
-{
- return (intr_info & (INTR_INFO_INTR_TYPE_MASK | INTR_INFO_VECTOR_MASK |
- INTR_INFO_VALID_MASK)) ==
- (INTR_TYPE_EXCEPTION | PF_VECTOR | INTR_INFO_VALID_MASK);
-}
-
-static inline int is_external_interrupt(u32 intr_info)
-{
- return (intr_info & (INTR_INFO_INTR_TYPE_MASK | INTR_INFO_VALID_MASK))
- == (INTR_TYPE_EXT_INTR | INTR_INFO_VALID_MASK);
-}
-
static inline int memslot_id(struct kvm *kvm, struct kvm_memory_slot *slot)
{
return slot - kvm->memslots;
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -77,6 +77,19 @@ static const u32 vmx_msr_index[] = {
struct vmx_msr_entry *find_msr_entry(struct kvm_vcpu *vcpu, u32 msr);
+static inline int is_page_fault(u32 intr_info)
+{
+ return (intr_info & (INTR_INFO_INTR_TYPE_MASK | INTR_INFO_VECTOR_MASK |
+ INTR_INFO_VALID_MASK)) ==
+ (INTR_TYPE_EXCEPTION | PF_VECTOR | INTR_INFO_VALID_MASK);
+}
+
+static inline int is_external_interrupt(u32 intr_info)
+{
+ return (intr_info & (INTR_INFO_INTR_TYPE_MASK | INTR_INFO_VALID_MASK))
+ == (INTR_TYPE_EXT_INTR | INTR_INFO_VALID_MASK);
+}
+
static void vmcs_clear(struct vmcs *vmcs)
{
u64 phys_addr = __pa(vmcs);
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread* [PATCH 38/38] KVM: Remove vmx includes from arch independent code
[not found] ` <456AD5C6.1090406-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
` (36 preceding siblings ...)
2006-11-27 12:47 ` [PATCH 37/38] KVM: Move vmx helper inlines " Avi Kivity
@ 2006-11-27 12:48 ` Avi Kivity
37 siblings, 0 replies; 49+ messages in thread
From: Avi Kivity @ 2006-11-27 12:48 UTC (permalink / raw)
To: kvm-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f
Cc: akpm-3NddpPZAyC0, linux-kernel-u79uwXL29TY76Z2rM5mHXA
Signed-off-by: Avi Kivity <avi-atKUWr5tajBWk0Htik3J/w@public.gmane.org>
Index: linux-2.6/drivers/kvm/kvm_main.c
===================================================================
--- linux-2.6.orig/drivers/kvm/kvm_main.c
+++ linux-2.6/drivers/kvm/kvm_main.c
@@ -35,8 +35,6 @@
#include <linux/file.h>
#include <asm/desc.h>
-#include "vmx.h"
-#include "kvm_vmx.h"
#include "x86_emulate.h"
MODULE_AUTHOR("Qumranet");
Index: linux-2.6/drivers/kvm/vmx.c
===================================================================
--- linux-2.6.orig/drivers/kvm/vmx.c
+++ linux-2.6/drivers/kvm/vmx.c
@@ -16,9 +16,9 @@
*/
#include "kvm.h"
-#include <linux/module.h>
#include "vmx.h"
#include "kvm_vmx.h"
+#include <linux/module.h>
#include <linux/mm.h>
#include <linux/highmem.h>
#include <asm/io.h>
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
^ permalink raw reply [flat|nested] 49+ messages in thread