KVM: ia64: Re-organize data sturure of guests' data area
1. Increase the size of data area to 64M
2. Support more vcpus and memory, 128 vcpus and 256G memory are supported
for guests.
3. Add the boundary check for memory and vcpu allocation.
With this patch, kvm guest's data area looks as follow:
*
* +----------------------+ ------- KVM_VM_DATA_SIZE
* | vcpu[n]'s data | | ___________________KVM_STK_OFFSET
* | | | / |
* | .......... | | /vcpu's struct&stack |
* | .......... | | /---------------------|---- 0
* | vcpu[5]'s data | | / vpd |
* | vcpu[4]'s data | |/-----------------------|
* | vcpu[3]'s data | / vtlb |
* | vcpu[2]'s data | /|------------------------|
* | vcpu[1]'s data |/ | vhpt |
* | vcpu[0]'s data |____________________________|
* +----------------------+ |
* | memory dirty log | |
* +----------------------+ |
* | vm's data struct | |
* +----------------------+ |
* | | |
* | | |
* | | |
* | | |
* | | |
* | | |
* | | |
* | vm's p2m table | |
* | | |
* | | |
* | | | |
* vm's data->| | | |
* +----------------------+ ------- 0
* To support large memory, needs to increase the size of p2m.
* To support more vcpus, needs to ensure it has enough space to
* hold vcpus' data.
*/
Signed-off-by: Xiantao Zhang <xiantao.zhang@intel.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
diff --git a/arch/ia64/include/asm/kvm_host.h b/arch/ia64/include/asm/kvm_host.h
index c60d324..678e264 100644
--- a/arch/ia64/include/asm/kvm_host.h
+++ b/arch/ia64/include/asm/kvm_host.h
@@ -23,17 +23,6 @@
#ifndef __ASM_KVM_HOST_H
#define __ASM_KVM_HOST_H
-
-#include <linux/types.h>
-#include <linux/mm.h>
-#include <linux/kvm.h>
-#include <linux/kvm_para.h>
-#include <linux/kvm_types.h>
-
-#include <asm/pal.h>
-#include <asm/sal.h>
-
-#define KVM_MAX_VCPUS 4
#define KVM_MEMORY_SLOTS 32
/* memory slots that does not exposed to userspace */
#define KVM_PRIVATE_MEM_SLOTS 4
@@ -52,68 +41,127 @@
#define EXIT_REASON_PTC_G 8
/*Define vmm address space and vm data space.*/
-#define KVM_VMM_SIZE (16UL<<20)
+#define KVM_VMM_SIZE (__IA64_UL_CONST(16)<<20)
#define KVM_VMM_SHIFT 24
-#define KVM_VMM_BASE 0xD000000000000000UL
-#define VMM_SIZE (8UL<<20)
+#define KVM_VMM_BASE 0xD000000000000000
+#define VMM_SIZE (__IA64_UL_CONST(8)<<20)
/*
* Define vm_buffer, used by PAL Services, base address.
- * Note: vmbuffer is in the VMM-BLOCK, the size must be < 8M
+ * Note: vm_buffer is in the VMM-BLOCK, the size must be < 8M
*/
#define KVM_VM_BUFFER_BASE (KVM_VMM_BASE + VMM_SIZE)
-#define KVM_VM_BUFFER_SIZE (8UL<<20)
+#define KVM_VM_BUFFER_SIZE (__IA64_UL_CONST(8)<<20)
-/*Define Virtual machine data layout.*/
-#define KVM_VM_DATA_SHIFT 24
-#define KVM_VM_DATA_SIZE (1UL << KVM_VM_DATA_SHIFT)
-#define KVM_VM_DATA_BASE (KVM_VMM_BASE + KVM_VMM_SIZE)
+/*
+ * kvm guest's data area looks as follow:
+ *
+ * +----------------------+ ------- KVM_VM_DATA_SIZE
+ * | vcpu[n]'s data | | ___________________KVM_STK_OFFSET
+ * | | | / |
+ * | .......... | | /vcpu's struct&stack |
+ * | .......... | | /---------------------|---- 0
+ * | vcpu[5]'s data | | / vpd |
+ * | vcpu[4]'s data | |/-----------------------|
+ * | vcpu[3]'s data | / vtlb |
+ * | vcpu[2]'s data | /|------------------------|
+ * | vcpu[1]'s data |/ | vhpt |
+ * | vcpu[0]'s data |____________________________|
+ * +----------------------+ |
+ * | memory dirty log | |
+ * +----------------------+ |
+ * | vm's data struct | |
+ * +----------------------+ |
+ * | | |
+ * | | |
+ * | | |
+ * | | |
+ * | | |
+ * | | |
+ * | | |
+ * | vm's p2m table | |
+ * | | |
+ * | | |
+ * | | | |
+ * vm's data->| | | |
+ * +----------------------+ ------- 0
+ * To support large memory, needs to increase the size of p2m.
+ * To support more vcpus, needs to ensure it has enough space to
+ * hold vcpus' data.
+ */
+#define KVM_VM_DATA_SHIFT 26
+#define KVM_VM_DATA_SIZE (__IA64_UL_CONST(1) << KVM_VM_DATA_SHIFT)
+#define KVM_VM_DATA_BASE (KVM_VMM_BASE + KVM_VM_DATA_SIZE)
-#define KVM_P2M_BASE KVM_VM_DATA_BASE
-#define KVM_P2M_OFS 0
-#define KVM_P2M_SIZE (8UL << 20)
+#define KVM_P2M_BASE KVM_VM_DATA_BASE
+#define KVM_P2M_SIZE (__IA64_UL_CONST(24) << 20)
-#define KVM_VHPT_BASE (KVM_P2M_BASE + KVM_P2M_SIZE)
-#define KVM_VHPT_OFS KVM_P2M_SIZE
-#define KVM_VHPT_BLOCK_SIZE (2UL << 20)
-#define VHPT_SHIFT 18
-#define VHPT_SIZE (1UL << VHPT_SHIFT)
-#define VHPT_NUM_ENTRIES (1<<(VHPT_SHIFT-5))
+#define VHPT_SHIFT 16
+#define VHPT_SIZE (__IA64_UL_CONST(1) << VHPT_SHIFT)
+#define VHPT_NUM_ENTRIES (__IA64_UL_CONST(1) << (VHPT_SHIFT-5))
-#define KVM_VTLB_BASE (KVM_VHPT_BASE+KVM_VHPT_BLOCK_SIZE)
-#define KVM_VTLB_OFS (KVM_VHPT_OFS+KVM_VHPT_BLOCK_SIZE)
-#define KVM_VTLB_BLOCK_SIZE (1UL<<20)
-#define VTLB_SHIFT 17
-#define VTLB_SIZE (1UL<<VTLB_SHIFT)
-#define VTLB_NUM_ENTRIES (1<<(VTLB_SHIFT-5))
+#define VTLB_SHIFT 16
+#define VTLB_SIZE (__IA64_UL_CONST(1) << VTLB_SHIFT)
+#define VTLB_NUM_ENTRIES (1UL << (VHPT_SHIFT-5))
-#define KVM_VPD_BASE (KVM_VTLB_BASE+KVM_VTLB_BLOCK_SIZE)
-#define KVM_VPD_OFS (KVM_VTLB_OFS+KVM_VTLB_BLOCK_SIZE)
-#define KVM_VPD_BLOCK_SIZE (2UL<<20)
-#define VPD_SHIFT 16
-#define VPD_SIZE (1UL<<VPD_SHIFT)
+#define VPD_SHIFT 16
+#define VPD_SIZE (__IA64_UL_CONST(1) << VPD_SHIFT)
-#define KVM_VCPU_BASE (KVM_VPD_BASE+KVM_VPD_BLOCK_SIZE)
-#define KVM_VCPU_OFS (KVM_VPD_OFS+KVM_VPD_BLOCK_SIZE)
-#define KVM_VCPU_BLOCK_SIZE (2UL<<20)
-#define VCPU_SHIFT 18
-#define VCPU_SIZE (1UL<<VCPU_SHIFT)
-#define MAX_VCPU_NUM KVM_VCPU_BLOCK_SIZE/VCPU_SIZE
+#define VCPU_STRUCT_SHIFT 16
+#define VCPU_STRUCT_SIZE (__IA64_UL_CONST(1) << VCPU_STRUCT_SHIFT)
-#define KVM_VM_BASE (KVM_VCPU_BASE+KVM_VCPU_BLOCK_SIZE)
-#define KVM_VM_OFS (KVM_VCPU_OFS+KVM_VCPU_BLOCK_SIZE)
-#define KVM_VM_BLOCK_SIZE (1UL<<19)
+#define KVM_STK_OFFSET VCPU_STRUCT_SIZE
-#define KVM_MEM_DIRTY_LOG_BASE (KVM_VM_BASE+KVM_VM_BLOCK_SIZE)
-#define KVM_MEM_DIRTY_LOG_OFS (KVM_VM_OFS+KVM_VM_BLOCK_SIZE)
-#define KVM_MEM_DIRTY_LOG_SIZE (1UL<<19)
+#define KVM_VM_STRUCT_SHIFT 19
+#define KVM_VM_STRUCT_SIZE (__IA64_UL_CONST(1) << KVM_VM_STRUCT_SHIFT)
-/* Get vpd, vhpt, tlb, vcpu, base*/
-#define VPD_ADDR(n) (KVM_VPD_BASE+n*VPD_SIZE)
-#define VHPT_ADDR(n) (KVM_VHPT_BASE+n*VHPT_SIZE)
-#define VTLB_ADDR(n) (KVM_VTLB_BASE+n*VTLB_SIZE)
-#define VCPU_ADDR(n) (KVM_VCPU_BASE+n*VCPU_SIZE)
+#define KVM_MEM_DIRY_LOG_SHIFT 19
+#define KVM_MEM_DIRTY_LOG_SIZE (__IA64_UL_CONST(1) << KVM_MEM_DIRY_LOG_SHIFT)
+
+#ifndef __ASSEMBLY__
+
+/*Define the max vcpus and memory for Guests.*/
+#define KVM_MAX_VCPUS (KVM_VM_DATA_SIZE - KVM_P2M_SIZE - KVM_VM_STRUCT_SIZE -\
+ KVM_MEM_DIRTY_LOG_SIZE) / sizeof(struct kvm_vcpu_data)
+#define KVM_MAX_MEM_SIZE (KVM_P2M_SIZE >> 3 << PAGE_SHIFT)
+
+#include <linux/types.h>
+#include <linux/mm.h>
+#include <linux/kvm.h>
+#include <linux/kvm_para.h>
+#include <linux/kvm_types.h>
+
+#include <asm/pal.h>
+#include <asm/sal.h>
+#include <asm/page.h>
+
+struct kvm_vcpu_data {
+ char vcpu_vhpt[VHPT_SIZE];
+ char vcpu_vtlb[VTLB_SIZE];
+ char vcpu_vpd[VPD_SIZE];
+ char vcpu_struct[VCPU_STRUCT_SIZE];
+};
+
+struct kvm_vm_data {
+ char kvm_p2m[KVM_P2M_SIZE];
+ char kvm_vm_struct[KVM_VM_STRUCT_SIZE];
+ char kvm_mem_dirty_log[KVM_MEM_DIRTY_LOG_SIZE];
+ struct kvm_vcpu_data vcpu_data[KVM_MAX_VCPUS];
+};
+
+#define VCPU_BASE(n) KVM_VM_DATA_BASE + \
+ offsetof(struct kvm_vm_data, vcpu_data[n])
+#define VM_BASE KVM_VM_DATA_BASE + \
+ offsetof(struct kvm_vm_data, kvm_vm_struct)
+#define KVM_MEM_DIRTY_LOG_BASE KVM_VM_DATA_BASE + \
+ offsetof(struct kvm_vm_data, kvm_mem_dirty_log)
+
+#define VHPT_BASE(n) (VCPU_BASE(n) + offsetof(struct kvm_vcpu_data, vcpu_vhpt))
+#define VTLB_BASE(n) (VCPU_BASE(n) + offsetof(struct kvm_vcpu_data, vcpu_vtlb))
+#define VPD_BASE(n) (VCPU_BASE(n) + offsetof(struct kvm_vcpu_data, vcpu_vpd))
+#define VCPU_STRUCT_BASE(n) (VCPU_BASE(n) + \
+ offsetof(struct kvm_vcpu_data, vcpu_struct))
/*IO section definitions*/
#define IOREQ_READ 1
@@ -403,14 +451,13 @@
};
struct kvm_arch {
+ spinlock_t dirty_log_lock;
+
unsigned long vm_base;
unsigned long metaphysical_rr0;
unsigned long metaphysical_rr4;
unsigned long vmm_init_rr;
- unsigned long vhpt_base;
- unsigned long vtlb_base;
- unsigned long vpd_base;
- spinlock_t dirty_log_lock;
+
struct kvm_ioapic *vioapic;
struct kvm_vm_stat stat;
struct kvm_sal_data rdv_sal_data;
@@ -512,7 +559,7 @@
static inline struct kvm_pt_regs *vcpu_regs(struct kvm_vcpu *v)
{
- return (struct kvm_pt_regs *) ((unsigned long) v + IA64_STK_OFFSET) - 1;
+ return (struct kvm_pt_regs *) ((unsigned long) v + KVM_STK_OFFSET) - 1;
}
typedef int kvm_vmm_entry(void);
@@ -531,5 +578,6 @@
void kvm_sal_emul(struct kvm_vcpu *vcpu);
static inline void kvm_inject_nmi(struct kvm_vcpu *vcpu) {}
+#endif /* __ASSEMBLY__*/
#endif
diff --git a/arch/ia64/kvm/kvm-ia64.c b/arch/ia64/kvm/kvm-ia64.c
index af1464f..43e45f6 100644
--- a/arch/ia64/kvm/kvm-ia64.c
+++ b/arch/ia64/kvm/kvm-ia64.c
@@ -698,27 +698,24 @@
return r;
}
-/*
- * Allocate 16M memory for every vm to hold its specific data.
- * Its memory map is defined in kvm_host.h.
- */
static struct kvm *kvm_alloc_kvm(void)
{
struct kvm *kvm;
uint64_t vm_base;
+ BUG_ON(sizeof(struct kvm) > KVM_VM_STRUCT_SIZE);
+
vm_base = __get_free_pages(GFP_KERNEL, get_order(KVM_VM_DATA_SIZE));
if (!vm_base)
return ERR_PTR(-ENOMEM);
- printk(KERN_DEBUG"kvm: VM data's base Address:0x%lx\n", vm_base);
- /* Zero all pages before use! */
memset((void *)vm_base, 0, KVM_VM_DATA_SIZE);
-
- kvm = (struct kvm *)(vm_base + KVM_VM_OFS);
+ kvm = (struct kvm *)(vm_base +
+ offsetof(struct kvm_vm_data, kvm_vm_struct));
kvm->arch.vm_base = vm_base;
+ printk(KERN_DEBUG"kvm: vm's data area:0x%lx\n", vm_base);
return kvm;
}
@@ -760,21 +757,12 @@
static void kvm_init_vm(struct kvm *kvm)
{
- long vm_base;
-
BUG_ON(!kvm);
kvm->arch.metaphysical_rr0 = GUEST_PHYSICAL_RR0;
kvm->arch.metaphysical_rr4 = GUEST_PHYSICAL_RR4;
kvm->arch.vmm_init_rr = VMM_INIT_RR;
- vm_base = kvm->arch.vm_base;
- if (vm_base) {
- kvm->arch.vhpt_base = vm_base + KVM_VHPT_OFS;
- kvm->arch.vtlb_base = vm_base + KVM_VTLB_OFS;
- kvm->arch.vpd_base = vm_base + KVM_VPD_OFS;
- }
-
/*
*Fill P2M entries for MMIO/IO ranges
*/
@@ -864,7 +852,7 @@
goto out;
r = copy_from_user(vcpu + 1, regs->saved_stack +
sizeof(struct kvm_vcpu),
- IA64_STK_OFFSET - sizeof(struct kvm_vcpu));
+ KVM_STK_OFFSET - sizeof(struct kvm_vcpu));
if (r)
goto out;
vcpu->arch.exit_data =
@@ -1166,10 +1154,11 @@
/*Set entry address for first run.*/
regs->cr_iip = PALE_RESET_ENTRY;
- /*Initilize itc offset for vcpus*/
+ /*Initialize itc offset for vcpus*/
itc_offset = 0UL - ia64_getreg(_IA64_REG_AR_ITC);
- for (i = 0; i < MAX_VCPU_NUM; i++) {
- v = (struct kvm_vcpu *)((char *)vcpu + VCPU_SIZE * i);
+ for (i = 0; i < KVM_MAX_VCPUS; i++) {
+ v = (struct kvm_vcpu *)((char *)vcpu +
+ sizeof(struct kvm_vcpu_data) * i);
v->arch.itc_offset = itc_offset;
v->arch.last_itc = 0;
}
@@ -1183,7 +1172,7 @@
vcpu->arch.apic->vcpu = vcpu;
p_ctx->gr[1] = 0;
- p_ctx->gr[12] = (unsigned long)((char *)vmm_vcpu + IA64_STK_OFFSET);
+ p_ctx->gr[12] = (unsigned long)((char *)vmm_vcpu + KVM_STK_OFFSET);
p_ctx->gr[13] = (unsigned long)vmm_vcpu;
p_ctx->psr = 0x1008522000UL;
p_ctx->ar[40] = FPSR_DEFAULT; /*fpsr*/
@@ -1218,12 +1207,12 @@
vcpu->arch.hlt_timer.function = hlt_timer_fn;
vcpu->arch.last_run_cpu = -1;
- vcpu->arch.vpd = (struct vpd *)VPD_ADDR(vcpu->vcpu_id);
+ vcpu->arch.vpd = (struct vpd *)VPD_BASE(vcpu->vcpu_id);
vcpu->arch.vsa_base = kvm_vsa_base;
vcpu->arch.__gp = kvm_vmm_gp;
vcpu->arch.dirty_log_lock_pa = __pa(&kvm->arch.dirty_log_lock);
- vcpu->arch.vhpt.hash = (struct thash_data *)VHPT_ADDR(vcpu->vcpu_id);
- vcpu->arch.vtlb.hash = (struct thash_data *)VTLB_ADDR(vcpu->vcpu_id);
+ vcpu->arch.vhpt.hash = (struct thash_data *)VHPT_BASE(vcpu->vcpu_id);
+ vcpu->arch.vtlb.hash = (struct thash_data *)VTLB_BASE(vcpu->vcpu_id);
init_ptce_info(vcpu);
r = 0;
@@ -1273,12 +1262,22 @@
int r;
int cpu;
+ BUG_ON(sizeof(struct kvm_vcpu) > VCPU_STRUCT_SIZE/2);
+
+ r = -EINVAL;
+ if (id >= KVM_MAX_VCPUS) {
+ printk(KERN_ERR"kvm: Can't configure vcpus > %ld",
+ KVM_MAX_VCPUS);
+ goto fail;
+ }
+
r = -ENOMEM;
if (!vm_base) {
printk(KERN_ERR"kvm: Create vcpu[%d] error!\n", id);
goto fail;
}
- vcpu = (struct kvm_vcpu *)(vm_base + KVM_VCPU_OFS + VCPU_SIZE * id);
+ vcpu = (struct kvm_vcpu *)(vm_base + offsetof(struct kvm_vm_data,
+ vcpu_data[id].vcpu_struct));
vcpu->kvm = kvm;
cpu = get_cpu();
@@ -1396,7 +1395,7 @@
sizeof(union context));
if (r)
goto out;
- r = copy_to_user(regs->saved_stack, (void *)vcpu, IA64_STK_OFFSET);
+ r = copy_to_user(regs->saved_stack, (void *)vcpu, KVM_STK_OFFSET);
if (r)
goto out;
SAVE_REGS(mp_state);
@@ -1457,6 +1456,9 @@
struct kvm_memory_slot *memslot = &kvm->memslots[mem->slot];
unsigned long base_gfn = memslot->base_gfn;
+ if (base_gfn + npages > (KVM_MAX_MEM_SIZE >> PAGE_SHIFT))
+ return -ENOMEM;
+
for (i = 0; i < npages; i++) {
pfn = gfn_to_pfn(kvm, base_gfn + i);
if (!kvm_is_mmio_pfn(pfn)) {
@@ -1631,8 +1633,8 @@
struct kvm_memory_slot *memslot;
int r, i;
long n, base;
- unsigned long *dirty_bitmap = (unsigned long *)((void *)kvm - KVM_VM_OFS
- + KVM_MEM_DIRTY_LOG_OFS);
+ unsigned long *dirty_bitmap = (unsigned long *)(kvm->arch.vm_base +
+ offsetof(struct kvm_vm_data, kvm_mem_dirty_log));
r = -EINVAL;
if (log->slot >= KVM_MEMORY_SLOTS)
diff --git a/arch/ia64/kvm/kvm_minstate.h b/arch/ia64/kvm/kvm_minstate.h
index 2cc41d1..b2bcaa2 100644
--- a/arch/ia64/kvm/kvm_minstate.h
+++ b/arch/ia64/kvm/kvm_minstate.h
@@ -24,6 +24,8 @@
#include <asm/asmmacro.h>
#include <asm/types.h>
#include <asm/kregs.h>
+#include <asm/kvm_host.h>
+
#include "asm-offsets.h"
#define KVM_MINSTATE_START_SAVE_MIN \
@@ -33,7 +35,7 @@
addl r22 = VMM_RBS_OFFSET,r1; /* compute base of RBS */ \
;; \
lfetch.fault.excl.nt1 [r22]; \
- addl r1 = IA64_STK_OFFSET-VMM_PT_REGS_SIZE,r1; /* compute base of memory stack */ \
+ addl r1 = KVM_STK_OFFSET-VMM_PT_REGS_SIZE, r1; \
mov r23 = ar.bspstore; /* save ar.bspstore */ \
;; \
mov ar.bspstore = r22; /* switch to kernel RBS */\
diff --git a/arch/ia64/kvm/misc.h b/arch/ia64/kvm/misc.h
index e585c46..dd979e0 100644
--- a/arch/ia64/kvm/misc.h
+++ b/arch/ia64/kvm/misc.h
@@ -27,7 +27,8 @@
*/
static inline uint64_t *kvm_host_get_pmt(struct kvm *kvm)
{
- return (uint64_t *)(kvm->arch.vm_base + KVM_P2M_OFS);
+ return (uint64_t *)(kvm->arch.vm_base +
+ offsetof(struct kvm_vm_data, kvm_p2m));
}
static inline void kvm_set_pmt_entry(struct kvm *kvm, gfn_t gfn,
diff --git a/arch/ia64/kvm/vcpu.c b/arch/ia64/kvm/vcpu.c
index e44027c..a528d70 100644
--- a/arch/ia64/kvm/vcpu.c
+++ b/arch/ia64/kvm/vcpu.c
@@ -816,8 +816,9 @@
unsigned long vitv = VCPU(vcpu, itv);
if (vcpu->vcpu_id == 0) {
- for (i = 0; i < MAX_VCPU_NUM; i++) {
- v = (struct kvm_vcpu *)((char *)vcpu + VCPU_SIZE * i);
+ for (i = 0; i < KVM_MAX_VCPUS; i++) {
+ v = (struct kvm_vcpu *)((char *)vcpu +
+ sizeof(struct kvm_vcpu_data) * i);
VMX(v, itc_offset) = itc_offset;
VMX(v, last_itc) = 0;
}
diff --git a/arch/ia64/kvm/vtlb.c b/arch/ia64/kvm/vtlb.c
index e22b933..6b6307a 100644
--- a/arch/ia64/kvm/vtlb.c
+++ b/arch/ia64/kvm/vtlb.c
@@ -183,8 +183,8 @@
u64 i, dirty_pages = 1;
u64 base_gfn = (pte&_PAGE_PPN_MASK) >> PAGE_SHIFT;
spinlock_t *lock = __kvm_va(v->arch.dirty_log_lock_pa);
- void *dirty_bitmap = (void *)v - (KVM_VCPU_OFS + v->vcpu_id * VCPU_SIZE)
- + KVM_MEM_DIRTY_LOG_OFS;
+ void *dirty_bitmap = (void *)KVM_MEM_DIRTY_LOG_BASE;
+
dirty_pages <<= ps <= PAGE_SHIFT ? 0 : ps - PAGE_SHIFT;
vmm_spin_lock(lock);
