Awesome
Obsolete
This Readme is obsolete due to a new kernel version.
Installation
Copy this files into your linux kernel.
For new kernel:
- vmx.c is going to /arch/x86/kvm/vmx
- svm.c is going to /arch/x86/kvm/svm
Don't forget to disable rdtscp in your qemu xml config:
Add rdtscp=off to your qemu:arg, so it will look similar to this:
<qemu:arg value="-cpu"/>
<qemu:arg value="host,rdtscp=off,hv_time,kvm=off,hv_vendor_id=null,-hypervisor"/>
Changing timer
You can play with ticks if you want to.
For Intel users:
- Open vmx.c in text editor
- Find handle_rdtsc function
- Change u64 fake_diff = diff / 16;
- 16 is a divider of actual difference in timestamp, you can increase and decrease it
For AMD users:
- Open svm.c in text editor
- Find handle_rdtsc_interception function
- Change u64 fake_diff = diff / 20;
- 20 is a divider of actual difference in timestamp, you can increase and decrease it
Modifying kernel by hand
If you have trouble to build your kernel with files from this repository, you can try to modify vmx and svm files in your kernel by yourself.
Changes for Intel users
Open vmx.c in text editor.
First you need to find setup_vmcs_config function (about 2500 lines) and add this line:
CPU_BASED_CR3_LOAD_EXITING |
CPU_BASED_CR3_STORE_EXITING |
CPU_BASED_UNCOND_IO_EXITING |
CPU_BASED_MOV_DR_EXITING |
CPU_BASED_USE_TSC_OFFSETTING |
CPU_BASED_MWAIT_EXITING |
CPU_BASED_MONITOR_EXITING |
CPU_BASED_INVLPG_EXITING |
CPU_BASED_RDPMC_EXITING |
CPU_BASED_RDTSC_EXITING; //added line
Next you need to find an array of pointers to handlers initialization called static int (*kvm_vmx_exit_handlers[])(struct kvm_vcpu *vcpu) and add the line to the end:
[EXIT_REASON_PREEMPTION_TIMER] = handle_preemption_timer,
[EXIT_REASON_ENCLS] = handle_encls,
[EXIT_REASON_BUS_LOCK] = handle_bus_lock_vmexit,
[EXIT_REASON_RDTSC] = handle_rdtsc, //added line
};
Then add this code before static int (*kvm_vmx_exit_handlers[])(struct kvm_vcpu *vcpu):
static u32 print_once = 1;
static int handle_rdtsc(struct kvm_vcpu *vcpu)
{
static u64 rdtsc_fake = 0;
static u64 rdtsc_prev = 0;
u64 rdtsc_real = rdtsc();
if(print_once)
{
printk("[handle_rdtsc] fake rdtsc vmx function is working\n");
print_once = 0;
rdtsc_fake = rdtsc_real;
}
if(rdtsc_prev != 0)
{
if(rdtsc_real > rdtsc_prev)
{
u64 diff = rdtsc_real - rdtsc_prev;
u64 fake_diff = diff / 16; // if you have 4.2Ghz on your vm, change 16 to 20
rdtsc_fake += fake_diff;
}
}
if(rdtsc_fake > rdtsc_real)
{
rdtsc_fake = rdtsc_real;
}
rdtsc_prev = rdtsc_real;
vcpu->arch.regs[VCPU_REGS_RAX] = rdtsc_fake & -1u;
vcpu->arch.regs[VCPU_REGS_RDX] = (rdtsc_fake >> 32) & -1u;
return skip_emulated_instruction(vcpu);
}
Changes for AMD users
Open svm.c in text editor.
Find static int (*const svm_exit_handlers[])(struct kvm_vcpu *vcpu) (~2700 lines) and add this line:
[SVM_EXIT_AVIC_INCOMPLETE_IPI] = avic_incomplete_ipi_interception,
[SVM_EXIT_AVIC_UNACCELERATED_ACCESS] = avic_unaccelerated_access_interception,
[SVM_EXIT_VMGEXIT] = sev_handle_vmgexit,
[SVM_EXIT_RDTSC] = handle_rdtsc_interception, //added line
};
Next find static void init_vmcb(struct kvm_vcpu *vcpu) (~1000 lines) and add this line:
svm_set_intercept(svm, INTERCEPT_RDPRU);
svm_set_intercept(svm, INTERCEPT_RSM);
svm_set_intercept(svm, INTERCEPT_RDTSC); //added line
After that add this code on top of static int (*const svm_exit_handlers[])(struct kvm_vcpu *vcpu):
static u32 print_once = 1;
static int handle_rdtsc_interception(struct kvm_vcpu *vcpu)
{
static u64 rdtsc_fake = 0;
static u64 rdtsc_prev = 0;
u64 rdtsc_real = rdtsc();
if(print_once)
{
printk("[handle_rdtsc] fake rdtsc svm function is working\n");
print_once = 0;
rdtsc_fake = rdtsc_real;
}
if(rdtsc_prev != 0)
{
if(rdtsc_real > rdtsc_prev)
{
u64 diff = rdtsc_real - rdtsc_prev;
u64 fake_diff = diff / 20; // if you have 3.2Ghz on your vm, change 20 to 16
rdtsc_fake += fake_diff;
}
}
if(rdtsc_fake > rdtsc_real)
{
rdtsc_fake = rdtsc_real;
}
rdtsc_prev = rdtsc_real;
vcpu->arch.regs[VCPU_REGS_RAX] = rdtsc_fake & -1u;
vcpu->arch.regs[VCPU_REGS_RDX] = (rdtsc_fake >> 32) & -1u;
return skip_emulated_instruction(vcpu);
}
Kernel building
- Clone the kernel from official repository or from repository of your os.
- Make changes for your cpu
- Install the requirements for kernel building
sudo apt-get install build-essential libncurses-dev bison flex libssl-dev libelf-dev
- Copy your current config into your kernel folder
cp -v /boot/config-$(uname -r) .config
- Make kernel
make -j $(nproc)
- Install kernel
sudo make INSTALL_MOD_STRIP=1 modules_install && sudo make install
- Reboot your system
- Check if a new kernel was installed
uname -a