Merge branch 'kvm-ppc-next' of...

Merge branch 'kvm-ppc-next' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc into HEAD

The big feature this time is support for POWER9 using the radix-tree
MMU for host and guest.  This required some changes to arch/powerpc
code, so I talked with Michael Ellerman and he created a topic branch
with this patchset, which I merged into kvm-ppc-next and which Michael
will pull into his tree.  Michael also put in some patches from Nick
Piggin which fix bugs in the interrupt vector code in relocatable
kernels when coming from a KVM guest.

Other notable changes include:

* Add the ability to change the size of the hashed page table,
  from David Gibson.

* XICS (interrupt controller) emulation fixes and improvements,
  from Li Zhong.

* Bug fixes from myself and Thomas Huth.

These patches define some new KVM capabilities and ioctls, but there
should be no conflicts with anything else currently upstream, as far
as I am aware.

Documentation/virtual/kvm/api.txt

@@ -2443,18 +2443,20 @@ are, it will do nothing and return an EBUSY error.
 The parameter is a pointer to a 32-bit unsigned integer variable
 containing the order (log base 2) of the desired size of the hash
 table, which must be between 18 and 46.  On successful return from the
-ioctl, it will have been updated with the order of the hash table that
-was allocated.
+ioctl, the value will not be changed by the kernel.
 
 If no hash table has been allocated when any vcpu is asked to run
 (with the KVM_RUN ioctl), the host kernel will allocate a
 default-sized hash table (16 MB).
 
 If this ioctl is called when a hash table has already been allocated,
-the kernel will clear out the existing hash table (zero all HPTEs) and
-return the hash table order in the parameter.  (If the guest is using
-the virtualized real-mode area (VRMA) facility, the kernel will
-re-create the VMRA HPTEs on the next KVM_RUN of any vcpu.)
+with a different order from the existing hash table, the existing hash
+table will be freed and a new one allocated.  If this is ioctl is
+called when a hash table has already been allocated of the same order
+as specified, the kernel will clear out the existing hash table (zero
+all HPTEs).  In either case, if the guest is using the virtualized
+real-mode area (VRMA) facility, the kernel will re-create the VMRA
+HPTEs on the next KVM_RUN of any vcpu.
 
 4.77 KVM_S390_INTERRUPT
 
@@ -3177,7 +3179,7 @@ of IOMMU pages.
 
 The rest of functionality is identical to KVM_CREATE_SPAPR_TCE.
 
-4.98 KVM_REINJECT_CONTROL
+4.99 KVM_REINJECT_CONTROL
 
 Capability: KVM_CAP_REINJECT_CONTROL
 Architectures: x86
@@ -3201,6 +3203,166 @@ struct kvm_reinject_control {
 pit_reinject = 0 (!reinject mode) is recommended, unless running an old
 operating system that uses the PIT for timing (e.g. Linux 2.4.x).
 
+4.100 KVM_PPC_CONFIGURE_V3_MMU
+
+Capability: KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3
+Architectures: ppc
+Type: vm ioctl
+Parameters: struct kvm_ppc_mmuv3_cfg (in)
+Returns: 0 on success,
+         -EFAULT if struct kvm_ppc_mmuv3_cfg cannot be read,
+         -EINVAL if the configuration is invalid
+
+This ioctl controls whether the guest will use radix or HPT (hashed
+page table) translation, and sets the pointer to the process table for
+the guest.
+
+struct kvm_ppc_mmuv3_cfg {
+	__u64	flags;
+	__u64	process_table;
+};
+
+There are two bits that can be set in flags; KVM_PPC_MMUV3_RADIX and
+KVM_PPC_MMUV3_GTSE.  KVM_PPC_MMUV3_RADIX, if set, configures the guest
+to use radix tree translation, and if clear, to use HPT translation.
+KVM_PPC_MMUV3_GTSE, if set and if KVM permits it, configures the guest
+to be able to use the global TLB and SLB invalidation instructions;
+if clear, the guest may not use these instructions.
+
+The process_table field specifies the address and size of the guest
+process table, which is in the guest's space.  This field is formatted
+as the second doubleword of the partition table entry, as defined in
+the Power ISA V3.00, Book III section 5.7.6.1.
+
+4.101 KVM_PPC_GET_RMMU_INFO
+
+Capability: KVM_CAP_PPC_RADIX_MMU
+Architectures: ppc
+Type: vm ioctl
+Parameters: struct kvm_ppc_rmmu_info (out)
+Returns: 0 on success,
+	 -EFAULT if struct kvm_ppc_rmmu_info cannot be written,
+	 -EINVAL if no useful information can be returned
+
+This ioctl returns a structure containing two things: (a) a list
+containing supported radix tree geometries, and (b) a list that maps
+page sizes to put in the "AP" (actual page size) field for the tlbie
+(TLB invalidate entry) instruction.
+
+struct kvm_ppc_rmmu_info {
+	struct kvm_ppc_radix_geom {
+		__u8	page_shift;
+		__u8	level_bits[4];
+		__u8	pad[3];
+	}	geometries[8];
+	__u32	ap_encodings[8];
+};
+
+The geometries[] field gives up to 8 supported geometries for the
+radix page table, in terms of the log base 2 of the smallest page
+size, and the number of bits indexed at each level of the tree, from
+the PTE level up to the PGD level in that order.  Any unused entries
+will have 0 in the page_shift field.
+
+The ap_encodings gives the supported page sizes and their AP field
+encodings, encoded with the AP value in the top 3 bits and the log
+base 2 of the page size in the bottom 6 bits.
+
+4.102 KVM_PPC_RESIZE_HPT_PREPARE
+
+Capability: KVM_CAP_SPAPR_RESIZE_HPT
+Architectures: powerpc
+Type: vm ioctl
+Parameters: struct kvm_ppc_resize_hpt (in)
+Returns: 0 on successful completion,
+	 >0 if a new HPT is being prepared, the value is an estimated
+             number of milliseconds until preparation is complete
+         -EFAULT if struct kvm_reinject_control cannot be read,
+	 -EINVAL if the supplied shift or flags are invalid
+	 -ENOMEM if unable to allocate the new HPT
+	 -ENOSPC if there was a hash collision when moving existing
+                  HPT entries to the new HPT
+	 -EIO on other error conditions
+
+Used to implement the PAPR extension for runtime resizing of a guest's
+Hashed Page Table (HPT).  Specifically this starts, stops or monitors
+the preparation of a new potential HPT for the guest, essentially
+implementing the H_RESIZE_HPT_PREPARE hypercall.
+
+If called with shift > 0 when there is no pending HPT for the guest,
+this begins preparation of a new pending HPT of size 2^(shift) bytes.
+It then returns a positive integer with the estimated number of
+milliseconds until preparation is complete.
+
+If called when there is a pending HPT whose size does not match that
+requested in the parameters, discards the existing pending HPT and
+creates a new one as above.
+
+If called when there is a pending HPT of the size requested, will:
+  * If preparation of the pending HPT is already complete, return 0
+  * If preparation of the pending HPT has failed, return an error
+    code, then discard the pending HPT.
+  * If preparation of the pending HPT is still in progress, return an
+    estimated number of milliseconds until preparation is complete.
+
+If called with shift == 0, discards any currently pending HPT and
+returns 0 (i.e. cancels any in-progress preparation).
+
+flags is reserved for future expansion, currently setting any bits in
+flags will result in an -EINVAL.
+
+Normally this will be called repeatedly with the same parameters until
+it returns <= 0.  The first call will initiate preparation, subsequent
+ones will monitor preparation until it completes or fails.
+
+struct kvm_ppc_resize_hpt {
+	__u64 flags;
+	__u32 shift;
+	__u32 pad;
+};
+
+4.103 KVM_PPC_RESIZE_HPT_COMMIT
+
+Capability: KVM_CAP_SPAPR_RESIZE_HPT
+Architectures: powerpc
+Type: vm ioctl
+Parameters: struct kvm_ppc_resize_hpt (in)
+Returns: 0 on successful completion,
+         -EFAULT if struct kvm_reinject_control cannot be read,
+	 -EINVAL if the supplied shift or flags are invalid
+	 -ENXIO is there is no pending HPT, or the pending HPT doesn't
+                 have the requested size
+	 -EBUSY if the pending HPT is not fully prepared
+	 -ENOSPC if there was a hash collision when moving existing
+                  HPT entries to the new HPT
+	 -EIO on other error conditions
+
+Used to implement the PAPR extension for runtime resizing of a guest's
+Hashed Page Table (HPT).  Specifically this requests that the guest be
+transferred to working with the new HPT, essentially implementing the
+H_RESIZE_HPT_COMMIT hypercall.
+
+This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has
+returned 0 with the same parameters.  In other cases
+KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or
+-EBUSY, though others may be possible if the preparation was started,
+but failed).
+
+This will have undefined effects on the guest if it has not already
+placed itself in a quiescent state where no vcpu will make MMU enabled
+memory accesses.
+
+On succsful completion, the pending HPT will become the guest's active
+HPT and the previous HPT will be discarded.
+
+On failure, the guest will still be operating on its previous HPT.
+
+struct kvm_ppc_resize_hpt {
+	__u64 flags;
+	__u32 shift;
+	__u32 pad;
+};
+
 5. The kvm_run structure
 ------------------------
 
@@ -3942,3 +4104,21 @@ In order to use SynIC, it has to be activated by setting this
 capability via KVM_ENABLE_CAP ioctl on the vcpu fd. Note that this
 will disable the use of APIC hardware virtualization even if supported
 by the CPU, as it's incompatible with SynIC auto-EOI behavior.
+
+8.3 KVM_CAP_PPC_RADIX_MMU
+
+Architectures: ppc
+
+This capability, if KVM_CHECK_EXTENSION indicates that it is
+available, means that that the kernel can support guests using the
+radix MMU defined in Power ISA V3.00 (as implemented in the POWER9
+processor).
+
+8.4 KVM_CAP_PPC_HASH_MMU_V3
+
+Architectures: ppc
+
+This capability, if KVM_CHECK_EXTENSION indicates that it is
+available, means that that the kernel can support guests using the
+hashed page table MMU defined in Power ISA V3.00 (as implemented in
+the POWER9 processor), including in-memory segment tables.

arch/powerpc/include/asm/book3s/64/mmu.h

@@ -44,10 +44,20 @@ struct patb_entry {
 };
 extern struct patb_entry *partition_tb;
 
+/* Bits in patb0 field */
 #define PATB_HR		(1UL << 63)
-#define PATB_GR		(1UL << 63)
 #define RPDB_MASK	0x0ffffffffffff00fUL
 #define RPDB_SHIFT	(1UL << 8)
+#define RTS1_SHIFT	61		/* top 2 bits of radix tree size */
+#define RTS1_MASK	(3UL << RTS1_SHIFT)
+#define RTS2_SHIFT	5		/* bottom 3 bits of radix tree size */
+#define RTS2_MASK	(7UL << RTS2_SHIFT)
+#define RPDS_MASK	0x1f		/* root page dir. size field */
+
+/* Bits in patb1 field */
+#define PATB_GR		(1UL << 63)	/* guest uses radix; must match HR */
+#define PRTS_MASK	0x1f		/* process table size field */
+
 /*
  * Limit process table to PAGE_SIZE table. This
  * also limit the max pid we can support.
@@ -138,5 +148,11 @@ static inline void setup_initial_memory_limit(phys_addr_t first_memblock_base,
 extern int (*register_process_table)(unsigned long base, unsigned long page_size,
 				     unsigned long tbl_size);
 
+#ifdef CONFIG_PPC_PSERIES
+extern void radix_init_pseries(void);
+#else
+static inline void radix_init_pseries(void) { };
+#endif
+
 #endif /* __ASSEMBLY__ */
 #endif /* _ASM_POWERPC_BOOK3S_64_MMU_H_ */

arch/powerpc/include/asm/exception-64s.h

@@ -97,6 +97,15 @@
 	ld	reg,PACAKBASE(r13);					\
 	ori	reg,reg,(ABS_ADDR(label))@l;
 
+/*
+ * Branches from unrelocated code (e.g., interrupts) to labels outside
+ * head-y require >64K offsets.
+ */
+#define __LOAD_FAR_HANDLER(reg, label)					\
+	ld	reg,PACAKBASE(r13);					\
+	ori	reg,reg,(ABS_ADDR(label))@l;				\
+	addis	reg,reg,(ABS_ADDR(label))@h;
+
 /* Exception register prefixes */
 #define EXC_HV	H
 #define EXC_STD
@@ -227,13 +236,41 @@ END_FTR_SECTION_NESTED(ftr,ftr,943)
 	mtctr	reg;							\
 	bctr
 
+/*
+ * KVM requires __LOAD_FAR_HANDLER.
+ *
+ * __BRANCH_TO_KVM_EXIT branches are also a special case because they
+ * explicitly use r9 then reload it from PACA before branching. Hence
+ * the double-underscore.
+ */
+#define __BRANCH_TO_KVM_EXIT(area, label)				\
+	mfctr	r9;							\
+	std	r9,HSTATE_SCRATCH1(r13);				\
+	__LOAD_FAR_HANDLER(r9, label);					\
+	mtctr	r9;							\
+	ld	r9,area+EX_R9(r13);					\
+	bctr
+
+#define BRANCH_TO_KVM(reg, label)					\
+	__LOAD_FAR_HANDLER(reg, label);					\
+	mtctr	reg;							\
+	bctr
+
 #else
 #define BRANCH_TO_COMMON(reg, label)					\
 	b	label
 
+#define BRANCH_TO_KVM(reg, label)					\
+	b	label
+
+#define __BRANCH_TO_KVM_EXIT(area, label)				\
+	ld	r9,area+EX_R9(r13);					\
+	b	label
+
 #endif
 
-#define __KVM_HANDLER_PROLOG(area, n)					\
+
+#define __KVM_HANDLER(area, h, n)					\
 	BEGIN_FTR_SECTION_NESTED(947)					\
 	ld	r10,area+EX_CFAR(r13);					\
 	std	r10,HSTATE_CFAR(r13);					\
@@ -243,30 +280,28 @@ END_FTR_SECTION_NESTED(ftr,ftr,943)
 	std	r10,HSTATE_PPR(r13);					\
 	END_FTR_SECTION_NESTED(CPU_FTR_HAS_PPR,CPU_FTR_HAS_PPR,948);	\
 	ld	r10,area+EX_R10(r13);					\
-	stw	r9,HSTATE_SCRATCH1(r13);				\
-	ld	r9,area+EX_R9(r13);					\
 	std	r12,HSTATE_SCRATCH0(r13);				\
-
-#define __KVM_HANDLER(area, h, n)					\
-	__KVM_HANDLER_PROLOG(area, n)					\
-	li	r12,n;							\
-	b	kvmppc_interrupt
+	sldi	r12,r9,32;						\
+	ori	r12,r12,(n);						\
+	/* This reloads r9 before branching to kvmppc_interrupt */	\
+	__BRANCH_TO_KVM_EXIT(area, kvmppc_interrupt)
 
 #define __KVM_HANDLER_SKIP(area, h, n)					\
 	cmpwi	r10,KVM_GUEST_MODE_SKIP;				\
-	ld	r10,area+EX_R10(r13);					\
 	beq	89f;							\
-	stw	r9,HSTATE_SCRATCH1(r13);				\
 	BEGIN_FTR_SECTION_NESTED(948)					\
-	ld	r9,area+EX_PPR(r13);					\
-	std	r9,HSTATE_PPR(r13);					\
+	ld	r10,area+EX_PPR(r13);					\
+	std	r10,HSTATE_PPR(r13);					\
 	END_FTR_SECTION_NESTED(CPU_FTR_HAS_PPR,CPU_FTR_HAS_PPR,948);	\
-	ld	r9,area+EX_R9(r13);					\
+	ld	r10,area+EX_R10(r13);					\
 	std	r12,HSTATE_SCRATCH0(r13);				\
-	li	r12,n;							\
-	b	kvmppc_interrupt;					\
+	sldi	r12,r9,32;						\
+	ori	r12,r12,(n);						\
+	/* This reloads r9 before branching to kvmppc_interrupt */	\
+	__BRANCH_TO_KVM_EXIT(area, kvmppc_interrupt);			\
 89:	mtocrf	0x80,r9;						\
 	ld	r9,area+EX_R9(r13);					\
+	ld	r10,area+EX_R10(r13);					\
 	b	kvmppc_skip_##h##interrupt
 
 #ifdef CONFIG_KVM_BOOK3S_64_HANDLER
@@ -393,12 +428,12 @@ END_FTR_SECTION_NESTED(ftr,ftr,943)
 	EXCEPTION_RELON_PROLOG_PSERIES_1(label, EXC_STD)
 
 #define STD_RELON_EXCEPTION_HV(loc, vec, label)		\
-	/* No guest interrupts come through here */	\
 	SET_SCRATCH0(r13);	/* save r13 */		\
-	EXCEPTION_RELON_PROLOG_PSERIES(PACA_EXGEN, label, EXC_HV, NOTEST, vec);
+	EXCEPTION_RELON_PROLOG_PSERIES(PACA_EXGEN, label,	\
+				       EXC_HV, KVMTEST_HV, vec);
 
 #define STD_RELON_EXCEPTION_HV_OOL(vec, label)			\
-	EXCEPTION_PROLOG_1(PACA_EXGEN, NOTEST, vec);		\
+	EXCEPTION_PROLOG_1(PACA_EXGEN, KVMTEST_HV, vec);	\
 	EXCEPTION_RELON_PROLOG_PSERIES_1(label, EXC_HV)
 
 /* This associate vector numbers with bits in paca->irq_happened */
@@ -475,10 +510,10 @@ END_FTR_SECTION_NESTED(ftr,ftr,943)
 
 #define MASKABLE_RELON_EXCEPTION_HV(loc, vec, label)			\
 	_MASKABLE_RELON_EXCEPTION_PSERIES(vec, label,			\
-					  EXC_HV, SOFTEN_NOTEST_HV)
+					  EXC_HV, SOFTEN_TEST_HV)
 
 #define MASKABLE_RELON_EXCEPTION_HV_OOL(vec, label)			\
-	EXCEPTION_PROLOG_1(PACA_EXGEN, SOFTEN_NOTEST_HV, vec);		\
+	EXCEPTION_PROLOG_1(PACA_EXGEN, SOFTEN_TEST_HV, vec);		\
 	EXCEPTION_PROLOG_PSERIES_1(label, EXC_HV)
 
 /*

arch/powerpc/include/asm/head-64.h

@@ -218,7 +218,7 @@ end_##sname:
 
 #ifdef CONFIG_KVM_BOOK3S_64_HANDLER
 #define TRAMP_KVM_BEGIN(name)						\
-	TRAMP_REAL_BEGIN(name)
+	TRAMP_VIRT_BEGIN(name)
 #else
 #define TRAMP_KVM_BEGIN(name)
 #endif

arch/powerpc/include/asm/hvcall.h

@@ -276,6 +276,7 @@
 #define H_GET_MPP_X		0x314
 #define H_SET_MODE		0x31C
 #define H_CLEAR_HPT		0x358
+#define H_REGISTER_PROC_TBL	0x37C
 #define H_SIGNAL_SYS_RESET	0x380
 #define MAX_HCALL_OPCODE	H_SIGNAL_SYS_RESET
 
@@ -313,6 +314,16 @@
 #define H_SIGNAL_SYS_RESET_ALL_OTHERS		-2
 /* >= 0 values are CPU number */
 
+/* Flag values used in H_REGISTER_PROC_TBL hcall */
+#define PROC_TABLE_OP_MASK	0x18
+#define PROC_TABLE_DEREG	0x10
+#define PROC_TABLE_NEW		0x18
+#define PROC_TABLE_TYPE_MASK	0x06
+#define PROC_TABLE_HPT_SLB	0x00
+#define PROC_TABLE_HPT_PT	0x02
+#define PROC_TABLE_RADIX	0x04
+#define PROC_TABLE_GTSE		0x01
+
 #ifndef __ASSEMBLY__
 
 /**

arch/powerpc/include/asm/kvm_book3s.h

@@ -170,6 +170,8 @@ extern int kvmppc_book3s_hv_page_fault(struct kvm_run *run,
 			unsigned long status);
 extern long kvmppc_hv_find_lock_hpte(struct kvm *kvm, gva_t eaddr,
 			unsigned long slb_v, unsigned long valid);
+extern int kvmppc_hv_emulate_mmio(struct kvm_run *run, struct kvm_vcpu *vcpu,
+			unsigned long gpa, gva_t ea, int is_store);
 
 extern void kvmppc_mmu_hpte_cache_map(struct kvm_vcpu *vcpu, struct hpte_cache *pte);
 extern struct hpte_cache *kvmppc_mmu_hpte_cache_next(struct kvm_vcpu *vcpu);
@@ -182,6 +184,25 @@ extern void kvmppc_mmu_hpte_sysexit(void);
 extern int kvmppc_mmu_hv_init(void);
 extern int kvmppc_book3s_hcall_implemented(struct kvm *kvm, unsigned long hc);
 
+extern int kvmppc_book3s_radix_page_fault(struct kvm_run *run,
+			struct kvm_vcpu *vcpu,
+			unsigned long ea, unsigned long dsisr);
+extern int kvmppc_mmu_radix_xlate(struct kvm_vcpu *vcpu, gva_t eaddr,
+			struct kvmppc_pte *gpte, bool data, bool iswrite);
+extern int kvmppc_init_vm_radix(struct kvm *kvm);
+extern void kvmppc_free_radix(struct kvm *kvm);
+extern int kvmppc_radix_init(void);
+extern void kvmppc_radix_exit(void);
+extern int kvm_unmap_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
+			unsigned long gfn);
+extern int kvm_age_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
+			unsigned long gfn);
+extern int kvm_test_age_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
+			unsigned long gfn);
+extern long kvmppc_hv_get_dirty_log_radix(struct kvm *kvm,
+			struct kvm_memory_slot *memslot, unsigned long *map);
+extern int kvmhv_get_rmmu_info(struct kvm *kvm, struct kvm_ppc_rmmu_info *info);
+
 /* XXX remove this export when load_last_inst() is generic */
 extern int kvmppc_ld(struct kvm_vcpu *vcpu, ulong *eaddr, int size, void *ptr, bool data);
 extern void kvmppc_book3s_queue_irqprio(struct kvm_vcpu *vcpu, unsigned int vec);
@@ -211,8 +232,11 @@ extern long kvmppc_do_h_enter(struct kvm *kvm, unsigned long flags,
 extern long kvmppc_do_h_remove(struct kvm *kvm, unsigned long flags,
 			unsigned long pte_index, unsigned long avpn,
 			unsigned long *hpret);
-extern long kvmppc_hv_get_dirty_log(struct kvm *kvm,
+extern long kvmppc_hv_get_dirty_log_hpt(struct kvm *kvm,
 			struct kvm_memory_slot *memslot, unsigned long *map);
+extern void kvmppc_harvest_vpa_dirty(struct kvmppc_vpa *vpa,
+			struct kvm_memory_slot *memslot,
+			unsigned long *map);
 extern void kvmppc_update_lpcr(struct kvm *kvm, unsigned long lpcr,
 			unsigned long mask);
 extern void kvmppc_set_fscr(struct kvm_vcpu *vcpu, u64 fscr);

arch/powerpc/include/asm/kvm_book3s_64.h

@@ -22,6 +22,10 @@
 
 #include <asm/book3s/64/mmu-hash.h>
 
+/* Power architecture requires HPT is at least 256kiB, at most 64TiB */
+#define PPC_MIN_HPT_ORDER	18
+#define PPC_MAX_HPT_ORDER	46
+
 #ifdef CONFIG_KVM_BOOK3S_PR_POSSIBLE
 static inline struct kvmppc_book3s_shadow_vcpu *svcpu_get(struct kvm_vcpu *vcpu)
 {
@@ -36,6 +40,12 @@ static inline void svcpu_put(struct kvmppc_book3s_shadow_vcpu *svcpu)
 #endif
 
 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
+
+static inline bool kvm_is_radix(struct kvm *kvm)
+{
+	return kvm->arch.radix;
+}
+
 #define KVM_DEFAULT_HPT_ORDER	24	/* 16MB HPT by default */
 #endif
 
@@ -350,6 +360,18 @@ extern void kvmppc_mmu_debugfs_init(struct kvm *kvm);
 
 extern void kvmhv_rm_send_ipi(int cpu);
 
+static inline unsigned long kvmppc_hpt_npte(struct kvm_hpt_info *hpt)
+{
+	/* HPTEs are 2**4 bytes long */
+	return 1UL << (hpt->order - 4);
+}
+
+static inline unsigned long kvmppc_hpt_mask(struct kvm_hpt_info *hpt)
+{
+	/* 128 (2**7) bytes in each HPTEG */
+	return (1UL << (hpt->order - 7)) - 1;
+}
+
 #endif /* CONFIG_KVM_BOOK3S_HV_POSSIBLE */
 
 #endif /* __ASM_KVM_BOOK3S_64_H__ */

arch/powerpc/include/asm/kvm_host.h

@@ -241,12 +241,24 @@ struct kvm_arch_memory_slot {
 #endif /* CONFIG_KVM_BOOK3S_HV_POSSIBLE */
 };
 
+struct kvm_hpt_info {
+	/* Host virtual (linear mapping) address of guest HPT */
+	unsigned long virt;
+	/* Array of reverse mapping entries for each guest HPTE */
+	struct revmap_entry *rev;
+	/* Guest HPT size is 2**(order) bytes */
+	u32 order;
+	/* 1 if HPT allocated with CMA, 0 otherwise */
+	int cma;
+};
+
+struct kvm_resize_hpt;
+
 struct kvm_arch {
 	unsigned int lpid;
 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
 	unsigned int tlb_sets;
-	unsigned long hpt_virt;
-	struct revmap_entry *revmap;
+	struct kvm_hpt_info hpt;
 	atomic64_t mmio_update;
 	unsigned int host_lpid;
 	unsigned long host_lpcr;
@@ -256,16 +268,17 @@ struct kvm_arch {
 	unsigned long lpcr;
 	unsigned long vrma_slb_v;
 	int hpte_setup_done;
-	u32 hpt_order;
 	atomic_t vcpus_running;
 	u32 online_vcores;
-	unsigned long hpt_npte;
-	unsigned long hpt_mask;
 	atomic_t hpte_mod_interest;
 	cpumask_t need_tlb_flush;
-	int hpt_cma_alloc;
+	cpumask_t cpu_in_guest;
+	u8 radix;
+	pgd_t *pgtable;
+	u64 process_table;
 	struct dentry *debugfs_dir;
 	struct dentry *htab_dentry;
+	struct kvm_resize_hpt *resize_hpt; /* protected by kvm->lock */
 #endif /* CONFIG_KVM_BOOK3S_HV_POSSIBLE */
 #ifdef CONFIG_KVM_BOOK3S_PR_POSSIBLE
 	struct mutex hpt_mutex;
@@ -603,6 +616,7 @@ struct kvm_vcpu_arch {
 	ulong fault_dar;
 	u32 fault_dsisr;
 	unsigned long intr_msr;
+	ulong fault_gpa;	/* guest real address of page fault (POWER9) */
 #endif
 
 #ifdef CONFIG_BOOKE
@@ -657,6 +671,7 @@ struct kvm_vcpu_arch {
 	int state;
 	int ptid;
 	int thread_cpu;
+	int prev_cpu;
 	bool timer_running;
 	wait_queue_head_t cpu_run;
 

arch/powerpc/include/asm/kvm_ppc.h

@@ -155,9 +155,10 @@ extern void kvmppc_core_destroy_mmu(struct kvm_vcpu *vcpu);
 extern int kvmppc_kvm_pv(struct kvm_vcpu *vcpu);
 extern void kvmppc_map_magic(struct kvm_vcpu *vcpu);
 
-extern long kvmppc_alloc_hpt(struct kvm *kvm, u32 *htab_orderp);
-extern long kvmppc_alloc_reset_hpt(struct kvm *kvm, u32 *htab_orderp);
-extern void kvmppc_free_hpt(struct kvm *kvm);
+extern int kvmppc_allocate_hpt(struct kvm_hpt_info *info, u32 order);
+extern void kvmppc_set_hpt(struct kvm *kvm, struct kvm_hpt_info *info);
+extern long kvmppc_alloc_reset_hpt(struct kvm *kvm, int order);
+extern void kvmppc_free_hpt(struct kvm_hpt_info *info);
 extern long kvmppc_prepare_vrma(struct kvm *kvm,
 				struct kvm_userspace_memory_region *mem);
 extern void kvmppc_map_vrma(struct kvm_vcpu *vcpu,
@@ -186,8 +187,8 @@ extern long kvmppc_h_stuff_tce(struct kvm_vcpu *vcpu,
 		unsigned long tce_value, unsigned long npages);
 extern long kvmppc_h_get_tce(struct kvm_vcpu *vcpu, unsigned long liobn,
 			     unsigned long ioba);
-extern struct page *kvm_alloc_hpt(unsigned long nr_pages);
-extern void kvm_release_hpt(struct page *page, unsigned long nr_pages);
+extern struct page *kvm_alloc_hpt_cma(unsigned long nr_pages);
+extern void kvm_free_hpt_cma(struct page *page, unsigned long nr_pages);
 extern int kvmppc_core_init_vm(struct kvm *kvm);
 extern void kvmppc_core_destroy_vm(struct kvm *kvm);
 extern void kvmppc_core_free_memslot(struct kvm *kvm,
@@ -214,6 +215,10 @@ extern void kvmppc_bookehv_exit(void);
 extern int kvmppc_prepare_to_enter(struct kvm_vcpu *vcpu);
 
 extern int kvm_vm_ioctl_get_htab_fd(struct kvm *kvm, struct kvm_get_htab_fd *);
+extern long kvm_vm_ioctl_resize_hpt_prepare(struct kvm *kvm,
+					    struct kvm_ppc_resize_hpt *rhpt);
+extern long kvm_vm_ioctl_resize_hpt_commit(struct kvm *kvm,
+					   struct kvm_ppc_resize_hpt *rhpt);
 
 int kvm_vcpu_ioctl_interrupt(struct kvm_vcpu *vcpu, struct kvm_interrupt *irq);
 
@@ -291,6 +296,8 @@ struct kvmppc_ops {
 				       struct irq_bypass_producer *);
 	void (*irq_bypass_del_producer)(struct irq_bypass_consumer *,
 					struct irq_bypass_producer *);
+	int (*configure_mmu)(struct kvm *kvm, struct kvm_ppc_mmuv3_cfg *cfg);
+	int (*get_rmmu_info)(struct kvm *kvm, struct kvm_ppc_rmmu_info *info);
 };
 
 extern struct kvmppc_ops *kvmppc_hv_ops;

arch/powerpc/include/asm/prom.h

@@ -121,6 +121,8 @@ struct of_drconf_cell {
 #define OV1_PPC_2_06		0x02	/* set if we support PowerPC 2.06 */
 #define OV1_PPC_2_07		0x01	/* set if we support PowerPC 2.07 */
 
+#define OV1_PPC_3_00		0x80	/* set if we support PowerPC 3.00 */
+
 /* Option vector 2: Open Firmware options supported */
 #define OV2_REAL_MODE		0x20	/* set if we want OF in real mode */
 
@@ -151,10 +153,17 @@ struct of_drconf_cell {
 #define OV5_XCMO		0x0440	/* Page Coalescing */
 #define OV5_TYPE1_AFFINITY	0x0580	/* Type 1 NUMA affinity */
 #define OV5_PRRN		0x0540	/* Platform Resource Reassignment */
-#define OV5_PFO_HW_RNG		0x0E80	/* PFO Random Number Generator */
-#define OV5_PFO_HW_842		0x0E40	/* PFO Compression Accelerator */
-#define OV5_PFO_HW_ENCR		0x0E20	/* PFO Encryption Accelerator */
-#define OV5_SUB_PROCESSORS	0x0F01	/* 1,2,or 4 Sub-Processors supported */
+#define OV5_PFO_HW_RNG		0x1180	/* PFO Random Number Generator */
+#define OV5_PFO_HW_842		0x1140	/* PFO Compression Accelerator */
+#define OV5_PFO_HW_ENCR		0x1120	/* PFO Encryption Accelerator */
+#define OV5_SUB_PROCESSORS	0x1501	/* 1,2,or 4 Sub-Processors supported */
+#define OV5_XIVE_EXPLOIT	0x1701	/* XIVE exploitation supported */
+#define OV5_MMU_RADIX_300	0x1880	/* ISA v3.00 radix MMU supported */
+#define OV5_MMU_HASH_300	0x1840	/* ISA v3.00 hash MMU supported */
+#define OV5_MMU_SEGM_RADIX	0x1820	/* radix mode (no segmentation) */
+#define OV5_MMU_PROC_TBL	0x1810	/* hcall selects SLB or proc table */
+#define OV5_MMU_SLB		0x1800	/* always use SLB */
+#define OV5_MMU_GTSE		0x1808	/* Guest translation shootdown */
 
 /* Option Vector 6: IBM PAPR hints */
 #define OV6_LINUX		0x02	/* Linux is our OS */

arch/powerpc/include/asm/reg.h

@@ -274,10 +274,14 @@
 #define SPRN_DSISR	0x012	/* Data Storage Interrupt Status Register */
 #define   DSISR_NOHPTE		0x40000000	/* no translation found */
 #define   DSISR_PROTFAULT	0x08000000	/* protection fault */
+#define   DSISR_BADACCESS	0x04000000	/* bad access to CI or G */
 #define   DSISR_ISSTORE		0x02000000	/* access was a store */
 #define   DSISR_DABRMATCH	0x00400000	/* hit data breakpoint */
 #define   DSISR_NOSEGMENT	0x00200000	/* SLB miss */
 #define   DSISR_KEYFAULT	0x00200000	/* Key fault */
+#define   DSISR_UNSUPP_MMU	0x00080000	/* Unsupported MMU config */
+#define   DSISR_SET_RC		0x00040000	/* Failed setting of R/C bits */
+#define   DSISR_PGDIRFAULT      0x00020000      /* Fault on page directory */
 #define SPRN_TBRL	0x10C	/* Time Base Read Lower Register (user, R/O) */
 #define SPRN_TBRU	0x10D	/* Time Base Read Upper Register (user, R/O) */
 #define SPRN_CIR	0x11B	/* Chip Information Register (hyper, R/0) */

arch/powerpc/include/uapi/asm/kvm.h

@@ -413,6 +413,26 @@ struct kvm_get_htab_header {
 	__u16	n_invalid;
 };
 
+/* For KVM_PPC_CONFIGURE_V3_MMU */
+struct kvm_ppc_mmuv3_cfg {
+	__u64	flags;
+	__u64	process_table;	/* second doubleword of partition table entry */
+};
+
+/* Flag values for KVM_PPC_CONFIGURE_V3_MMU */
+#define KVM_PPC_MMUV3_RADIX	1	/* 1 = radix mode, 0 = HPT */
+#define KVM_PPC_MMUV3_GTSE	2	/* global translation shootdown enb. */
+
+/* For KVM_PPC_GET_RMMU_INFO */
+struct kvm_ppc_rmmu_info {
+	struct kvm_ppc_radix_geom {
+		__u8	page_shift;
+		__u8	level_bits[4];
+		__u8	pad[3];
+	}	geometries[8];
+	__u32	ap_encodings[8];
+};
+
 /* Per-vcpu XICS interrupt controller state */
 #define KVM_REG_PPC_ICP_STATE	(KVM_REG_PPC | KVM_REG_SIZE_U64 | 0x8c)
 
@@ -613,5 +633,7 @@ struct kvm_get_htab_header {
 #define  KVM_XICS_LEVEL_SENSITIVE	(1ULL << 40)
 #define  KVM_XICS_MASKED		(1ULL << 41)
 #define  KVM_XICS_PENDING		(1ULL << 42)
+#define  KVM_XICS_PRESENTED		(1ULL << 43)
+#define  KVM_XICS_QUEUED		(1ULL << 44)
 
 #endif /* __LINUX_KVM_POWERPC_H */

arch/powerpc/kernel/asm-offsets.c

@@ -498,6 +498,7 @@ int main(void)
 	DEFINE(KVM_NEED_FLUSH, offsetof(struct kvm, arch.need_tlb_flush.bits));
 	DEFINE(KVM_ENABLED_HCALLS, offsetof(struct kvm, arch.enabled_hcalls));
 	DEFINE(KVM_VRMA_SLB_V, offsetof(struct kvm, arch.vrma_slb_v));
+	DEFINE(KVM_RADIX, offsetof(struct kvm, arch.radix));
 	DEFINE(VCPU_DSISR, offsetof(struct kvm_vcpu, arch.shregs.dsisr));
 	DEFINE(VCPU_DAR, offsetof(struct kvm_vcpu, arch.shregs.dar));
 	DEFINE(VCPU_VPA, offsetof(struct kvm_vcpu, arch.vpa.pinned_addr));
@@ -537,6 +538,7 @@ int main(void)
 	DEFINE(VCPU_SLB_NR, offsetof(struct kvm_vcpu, arch.slb_nr));
 	DEFINE(VCPU_FAULT_DSISR, offsetof(struct kvm_vcpu, arch.fault_dsisr));
 	DEFINE(VCPU_FAULT_DAR, offsetof(struct kvm_vcpu, arch.fault_dar));
+	DEFINE(VCPU_FAULT_GPA, offsetof(struct kvm_vcpu, arch.fault_gpa));
 	DEFINE(VCPU_INTR_MSR, offsetof(struct kvm_vcpu, arch.intr_msr));
 	DEFINE(VCPU_LAST_INST, offsetof(struct kvm_vcpu, arch.last_inst));
 	DEFINE(VCPU_TRAP, offsetof(struct kvm_vcpu, arch.trap));

arch/powerpc/kernel/exceptions-64s.S

@@ -142,7 +142,7 @@ END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_300)
 	lbz	r0,HSTATE_HWTHREAD_REQ(r13)
 	cmpwi	r0,0
 	beq	1f
-	b	kvm_start_guest
+	BRANCH_TO_KVM(r10, kvm_start_guest)
 1:
 #endif
 
@@ -717,13 +717,9 @@ hardware_interrupt_hv:
 	BEGIN_FTR_SECTION
 		_MASKABLE_EXCEPTION_PSERIES(0x500, hardware_interrupt_common,
 					    EXC_HV, SOFTEN_TEST_HV)
-do_kvm_H0x500:
-		KVM_HANDLER(PACA_EXGEN, EXC_HV, 0x502)
 	FTR_SECTION_ELSE
 		_MASKABLE_EXCEPTION_PSERIES(0x500, hardware_interrupt_common,
 					    EXC_STD, SOFTEN_TEST_PR)
-do_kvm_0x500:
-		KVM_HANDLER(PACA_EXGEN, EXC_STD, 0x500)
 	ALT_FTR_SECTION_END_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_206)
 EXC_REAL_END(hardware_interrupt, 0x500, 0x600)
 
@@ -737,6 +733,8 @@ hardware_interrupt_relon_hv:
 	ALT_FTR_SECTION_END_IFSET(CPU_FTR_HVMODE)
 EXC_VIRT_END(hardware_interrupt, 0x4500, 0x4600)
 
+TRAMP_KVM(PACA_EXGEN, 0x500)
+TRAMP_KVM_HV(PACA_EXGEN, 0x500)
 EXC_COMMON_ASYNC(hardware_interrupt_common, 0x500, do_IRQ)
 
 
@@ -832,6 +830,31 @@ EXC_VIRT(trap_0b, 0x4b00, 0x4c00, 0xb00)
 TRAMP_KVM(PACA_EXGEN, 0xb00)
 EXC_COMMON(trap_0b_common, 0xb00, unknown_exception)
 
+#ifdef CONFIG_KVM_BOOK3S_64_HANDLER
+	 /*
+	  * If CONFIG_KVM_BOOK3S_64_HANDLER is set, save the PPR (on systems
+	  * that support it) before changing to HMT_MEDIUM. That allows the KVM
+	  * code to save that value into the guest state (it is the guest's PPR
+	  * value). Otherwise just change to HMT_MEDIUM as userspace has
+	  * already saved the PPR.
+	  */
+#define SYSCALL_KVMTEST							\
+	SET_SCRATCH0(r13);						\
+	GET_PACA(r13);							\
+	std	r9,PACA_EXGEN+EX_R9(r13);				\
+	OPT_GET_SPR(r9, SPRN_PPR, CPU_FTR_HAS_PPR);			\
+	HMT_MEDIUM;							\
+	std	r10,PACA_EXGEN+EX_R10(r13);				\
+	OPT_SAVE_REG_TO_PACA(PACA_EXGEN+EX_PPR, r9, CPU_FTR_HAS_PPR);	\
+	mfcr	r9;							\
+	KVMTEST_PR(0xc00);						\
+	GET_SCRATCH0(r13)
+
+#else
+#define SYSCALL_KVMTEST							\
+	HMT_MEDIUM
+#endif
+	
 #define LOAD_SYSCALL_HANDLER(reg)					\
 	__LOAD_HANDLER(reg, system_call_common)
 
@@ -885,34 +908,14 @@ END_FTR_SECTION_IFSET(CPU_FTR_REAL_LE)				\
 #endif
 
 EXC_REAL_BEGIN(system_call, 0xc00, 0xd00)
-	 /*
-	  * If CONFIG_KVM_BOOK3S_64_HANDLER is set, save the PPR (on systems
-	  * that support it) before changing to HMT_MEDIUM. That allows the KVM
-	  * code to save that value into the guest state (it is the guest's PPR
-	  * value). Otherwise just change to HMT_MEDIUM as userspace has
-	  * already saved the PPR.
-	  */
-#ifdef CONFIG_KVM_BOOK3S_64_HANDLER
-	SET_SCRATCH0(r13)
-	GET_PACA(r13)
-	std	r9,PACA_EXGEN+EX_R9(r13)
-	OPT_GET_SPR(r9, SPRN_PPR, CPU_FTR_HAS_PPR);
-	HMT_MEDIUM;
-	std	r10,PACA_EXGEN+EX_R10(r13)
-	OPT_SAVE_REG_TO_PACA(PACA_EXGEN+EX_PPR, r9, CPU_FTR_HAS_PPR);
-	mfcr	r9
-	KVMTEST_PR(0xc00)
-	GET_SCRATCH0(r13)
-#else
-	HMT_MEDIUM;
-#endif
+	SYSCALL_KVMTEST
 	SYSCALL_PSERIES_1
 	SYSCALL_PSERIES_2_RFID
 	SYSCALL_PSERIES_3
 EXC_REAL_END(system_call, 0xc00, 0xd00)
 
 EXC_VIRT_BEGIN(system_call, 0x4c00, 0x4d00)
-	HMT_MEDIUM
+	SYSCALL_KVMTEST
 	SYSCALL_PSERIES_1
 	SYSCALL_PSERIES_2_DIRECT
 	SYSCALL_PSERIES_3
@@ -927,7 +930,7 @@ TRAMP_KVM(PACA_EXGEN, 0xd00)
 EXC_COMMON(single_step_common, 0xd00, single_step_exception)
 
 EXC_REAL_OOL_HV(h_data_storage, 0xe00, 0xe20)
-EXC_VIRT_NONE(0x4e00, 0x4e20)
+EXC_VIRT_OOL_HV(h_data_storage, 0x4e00, 0x4e20, 0xe00)
 TRAMP_KVM_HV_SKIP(PACA_EXGEN, 0xe00)
 EXC_COMMON_BEGIN(h_data_storage_common)
 	mfspr   r10,SPRN_HDAR
@@ -943,7 +946,7 @@ EXC_COMMON_BEGIN(h_data_storage_common)
 
 
 EXC_REAL_OOL_HV(h_instr_storage, 0xe20, 0xe40)
-EXC_VIRT_NONE(0x4e20, 0x4e40)
+EXC_VIRT_OOL_HV(h_instr_storage, 0x4e20, 0x4e40, 0xe20)
 TRAMP_KVM_HV(PACA_EXGEN, 0xe20)
 EXC_COMMON(h_instr_storage_common, 0xe20, unknown_exception)
 

arch/powerpc/kernel/prom_init.c

@@ -649,6 +649,7 @@ static void __init early_cmdline_parse(void)
 struct option_vector1 {
 	u8 byte1;
 	u8 arch_versions;
+	u8 arch_versions3;
 } __packed;
 
 struct option_vector2 {
@@ -691,6 +692,9 @@ struct option_vector5 {
 	u8 reserved2;
 	__be16 reserved3;
 	u8 subprocessors;
+	u8 byte22;
+	u8 intarch;
+	u8 mmu;
 } __packed;
 
 struct option_vector6 {
@@ -700,7 +704,7 @@ struct option_vector6 {
 } __packed;
 
 struct ibm_arch_vec {
-	struct { u32 mask, val; } pvrs[10];
+	struct { u32 mask, val; } pvrs[12];
 
 	u8 num_vectors;
 
@@ -749,6 +753,14 @@ struct ibm_arch_vec __cacheline_aligned ibm_architecture_vec = {
 			.mask = cpu_to_be32(0xffff0000), /* POWER8 */
 			.val  = cpu_to_be32(0x004d0000),
 		},
+		{
+			.mask = cpu_to_be32(0xffff0000), /* POWER9 */
+			.val  = cpu_to_be32(0x004e0000),
+		},
+		{
+			.mask = cpu_to_be32(0xffffffff), /* all 3.00-compliant */
+			.val  = cpu_to_be32(0x0f000005),
+		},
 		{
 			.mask = cpu_to_be32(0xffffffff), /* all 2.07-compliant */
 			.val  = cpu_to_be32(0x0f000004),
@@ -774,6 +786,7 @@ struct ibm_arch_vec __cacheline_aligned ibm_architecture_vec = {
 		.byte1 = 0,
 		.arch_versions = OV1_PPC_2_00 | OV1_PPC_2_01 | OV1_PPC_2_02 | OV1_PPC_2_03 |
 				 OV1_PPC_2_04 | OV1_PPC_2_05 | OV1_PPC_2_06 | OV1_PPC_2_07,
+		.arch_versions3 = OV1_PPC_3_00,
 	},
 
 	.vec2_len = VECTOR_LENGTH(sizeof(struct option_vector2)),
@@ -836,6 +849,9 @@ struct ibm_arch_vec __cacheline_aligned ibm_architecture_vec = {
 		.reserved2 = 0,
 		.reserved3 = 0,
 		.subprocessors = 1,
+		.intarch = 0,
+		.mmu = OV5_FEAT(OV5_MMU_RADIX_300) | OV5_FEAT(OV5_MMU_HASH_300) |
+			OV5_FEAT(OV5_MMU_PROC_TBL) | OV5_FEAT(OV5_MMU_GTSE),
 	},
 
 	/* option vector 6: IBM PAPR hints */

arch/powerpc/kvm/Makefile

@@ -70,7 +70,8 @@ endif
 kvm-hv-y += \
 	book3s_hv.o \
 	book3s_hv_interrupts.o \
-	book3s_64_mmu_hv.o
+	book3s_64_mmu_hv.o \
+	book3s_64_mmu_radix.o
 
 kvm-book3s_64-builtin-xics-objs-$(CONFIG_KVM_XICS) := \
 	book3s_hv_rm_xics.o

arch/powerpc/kvm/book3s.c

@@ -239,6 +239,7 @@ void kvmppc_core_queue_data_storage(struct kvm_vcpu *vcpu, ulong dar,
 	kvmppc_set_dsisr(vcpu, flags);
 	kvmppc_book3s_queue_irqprio(vcpu, BOOK3S_INTERRUPT_DATA_STORAGE);
 }
+EXPORT_SYMBOL_GPL(kvmppc_core_queue_data_storage);	/* used by kvm_hv */
 
 void kvmppc_core_queue_inst_storage(struct kvm_vcpu *vcpu, ulong flags)
 {

arch/powerpc/kvm/book3s_64_mmu_hv.c

@@ -40,84 +40,104 @@
 
 #include "trace_hv.h"
 
-/* Power architecture requires HPT is at least 256kB */
-#define PPC_MIN_HPT_ORDER	18
+//#define DEBUG_RESIZE_HPT	1
+
+#ifdef DEBUG_RESIZE_HPT
+#define resize_hpt_debug(resize, ...)				\
+	do {							\
+		printk(KERN_DEBUG "RESIZE HPT %p: ", resize);	\
+		printk(__VA_ARGS__);				\
+	} while (0)
+#else
+#define resize_hpt_debug(resize, ...)				\
+	do { } while (0)
+#endif
 
 static long kvmppc_virtmode_do_h_enter(struct kvm *kvm, unsigned long flags,
 				long pte_index, unsigned long pteh,
 				unsigned long ptel, unsigned long *pte_idx_ret);
+
+struct kvm_resize_hpt {
+	/* These fields read-only after init */
+	struct kvm *kvm;
+	struct work_struct work;
+	u32 order;
+
+	/* These fields protected by kvm->lock */
+	int error;
+	bool prepare_done;
+
+	/* Private to the work thread, until prepare_done is true,
+	 * then protected by kvm->resize_hpt_sem */
+	struct kvm_hpt_info hpt;
+};
+
 static void kvmppc_rmap_reset(struct kvm *kvm);
 
-long kvmppc_alloc_hpt(struct kvm *kvm, u32 *htab_orderp)
+int kvmppc_allocate_hpt(struct kvm_hpt_info *info, u32 order)
 {
 	unsigned long hpt = 0;
-	struct revmap_entry *rev;
+	int cma = 0;
 	struct page *page = NULL;
-	long order = KVM_DEFAULT_HPT_ORDER;
+	struct revmap_entry *rev;
+	unsigned long npte;
 
-	if (htab_orderp) {
-		order = *htab_orderp;
-		if (order < PPC_MIN_HPT_ORDER)
-			order = PPC_MIN_HPT_ORDER;
-	}
+	if ((order < PPC_MIN_HPT_ORDER) || (order > PPC_MAX_HPT_ORDER))
+		return -EINVAL;
 
-	kvm->arch.hpt_cma_alloc = 0;
-	page = kvm_alloc_hpt(1ul << (order - PAGE_SHIFT));
+	page = kvm_alloc_hpt_cma(1ul << (order - PAGE_SHIFT));
 	if (page) {
 		hpt = (unsigned long)pfn_to_kaddr(page_to_pfn(page));
 		memset((void *)hpt, 0, (1ul << order));
-		kvm->arch.hpt_cma_alloc = 1;
+		cma = 1;
 	}
 
-	/* Lastly try successively smaller sizes from the page allocator */
-	/* Only do this if userspace didn't specify a size via ioctl */
-	while (!hpt && order > PPC_MIN_HPT_ORDER && !htab_orderp) {
-		hpt = __get_free_pages(GFP_KERNEL|__GFP_ZERO|__GFP_REPEAT|
-				       __GFP_NOWARN, order - PAGE_SHIFT);
-		if (!hpt)
-			--order;
-	}
+	if (!hpt)
+		hpt = __get_free_pages(GFP_KERNEL|__GFP_ZERO|__GFP_REPEAT
+				       |__GFP_NOWARN, order - PAGE_SHIFT);
 
 	if (!hpt)
 		return -ENOMEM;
 
-	kvm->arch.hpt_virt = hpt;
-	kvm->arch.hpt_order = order;
 	/* HPTEs are 2**4 bytes long */
-	kvm->arch.hpt_npte = 1ul << (order - 4);
-	/* 128 (2**7) bytes in each HPTEG */
-	kvm->arch.hpt_mask = (1ul << (order - 7)) - 1;
-
-	atomic64_set(&kvm->arch.mmio_update, 0);
+	npte = 1ul << (order - 4);
 
 	/* Allocate reverse map array */
-	rev = vmalloc(sizeof(struct revmap_entry) * kvm->arch.hpt_npte);
+	rev = vmalloc(sizeof(struct revmap_entry) * npte);
 	if (!rev) {
-		pr_err("kvmppc_alloc_hpt: Couldn't alloc reverse map array\n");
-		goto out_freehpt;
+		pr_err("kvmppc_allocate_hpt: Couldn't alloc reverse map array\n");
+		if (cma)
+			kvm_free_hpt_cma(page, 1 << (order - PAGE_SHIFT));
+		else
+			free_pages(hpt, order - PAGE_SHIFT);
+		return -ENOMEM;
 	}
-	kvm->arch.revmap = rev;
-	kvm->arch.sdr1 = __pa(hpt) | (order - 18);
 
-	pr_info("KVM guest htab at %lx (order %ld), LPID %x\n",
-		hpt, order, kvm->arch.lpid);
+	info->order = order;
+	info->virt = hpt;
+	info->cma = cma;
+	info->rev = rev;
 
-	if (htab_orderp)
-		*htab_orderp = order;
 	return 0;
+}
 
- out_freehpt:
-	if (kvm->arch.hpt_cma_alloc)
-		kvm_release_hpt(page, 1 << (order - PAGE_SHIFT));
-	else
-		free_pages(hpt, order - PAGE_SHIFT);
-	return -ENOMEM;
+void kvmppc_set_hpt(struct kvm *kvm, struct kvm_hpt_info *info)
+{
+	atomic64_set(&kvm->arch.mmio_update, 0);
+	kvm->arch.hpt = *info;
+	kvm->arch.sdr1 = __pa(info->virt) | (info->order - 18);
+
+	pr_info("KVM guest htab at %lx (order %ld), LPID %x\n",
+		info->virt, (long)info->order, kvm->arch.lpid);
 }
 
-long kvmppc_alloc_reset_hpt(struct kvm *kvm, u32 *htab_orderp)
+long kvmppc_alloc_reset_hpt(struct kvm *kvm, int order)
 {
 	long err = -EBUSY;
-	long order;
+	struct kvm_hpt_info info;
+
+	if (kvm_is_radix(kvm))
+		return -EINVAL;
 
 	mutex_lock(&kvm->lock);
 	if (kvm->arch.hpte_setup_done) {
@@ -129,37 +149,44 @@ long kvmppc_alloc_reset_hpt(struct kvm *kvm, u32 *htab_orderp)
 			goto out;
 		}
 	}
-	if (kvm->arch.hpt_virt) {
-		order = kvm->arch.hpt_order;
+	if (kvm->arch.hpt.order == order) {
+		/* We already have a suitable HPT */
+
 		/* Set the entire HPT to 0, i.e. invalid HPTEs */
-		memset((void *)kvm->arch.hpt_virt, 0, 1ul << order);
+		memset((void *)kvm->arch.hpt.virt, 0, 1ul << order);
 		/*
 		 * Reset all the reverse-mapping chains for all memslots
 		 */
 		kvmppc_rmap_reset(kvm);
 		/* Ensure that each vcpu will flush its TLB on next entry. */
 		cpumask_setall(&kvm->arch.need_tlb_flush);
-		*htab_orderp = order;
 		err = 0;
-	} else {
-		err = kvmppc_alloc_hpt(kvm, htab_orderp);
-		order = *htab_orderp;
+		goto out;
 	}
- out:
+
+	if (kvm->arch.hpt.virt)
+		kvmppc_free_hpt(&kvm->arch.hpt);
+
+	err = kvmppc_allocate_hpt(&info, order);
+	if (err < 0)
+		goto out;
+	kvmppc_set_hpt(kvm, &info);
+
+out:
 	mutex_unlock(&kvm->lock);
 	return err;
 }
 
-void kvmppc_free_hpt(struct kvm *kvm)
+void kvmppc_free_hpt(struct kvm_hpt_info *info)
 {
-	kvmppc_free_lpid(kvm->arch.lpid);
-	vfree(kvm->arch.revmap);
-	if (kvm->arch.hpt_cma_alloc)
-		kvm_release_hpt(virt_to_page(kvm->arch.hpt_virt),
-				1 << (kvm->arch.hpt_order - PAGE_SHIFT));
-	else
-		free_pages(kvm->arch.hpt_virt,
-			   kvm->arch.hpt_order - PAGE_SHIFT);
+	vfree(info->rev);
+	if (info->cma)
+		kvm_free_hpt_cma(virt_to_page(info->virt),
+				 1 << (info->order - PAGE_SHIFT));
+	else if (info->virt)
+		free_pages(info->virt, info->order - PAGE_SHIFT);
+	info->virt = 0;
+	info->order = 0;
 }
 
 /* Bits in first HPTE dword for pagesize 4k, 64k or 16M */
@@ -194,8 +221,8 @@ void kvmppc_map_vrma(struct kvm_vcpu *vcpu, struct kvm_memory_slot *memslot,
 	if (npages > 1ul << (40 - porder))
 		npages = 1ul << (40 - porder);
 	/* Can't use more than 1 HPTE per HPTEG */
-	if (npages > kvm->arch.hpt_mask + 1)
-		npages = kvm->arch.hpt_mask + 1;
+	if (npages > kvmppc_hpt_mask(&kvm->arch.hpt) + 1)
+		npages = kvmppc_hpt_mask(&kvm->arch.hpt) + 1;
 
 	hp0 = HPTE_V_1TB_SEG | (VRMA_VSID << (40 - 16)) |
 		HPTE_V_BOLTED | hpte0_pgsize_encoding(psize);
@@ -205,7 +232,8 @@ void kvmppc_map_vrma(struct kvm_vcpu *vcpu, struct kvm_memory_slot *memslot,
 	for (i = 0; i < npages; ++i) {
 		addr = i << porder;
 		/* can't use hpt_hash since va > 64 bits */
-		hash = (i ^ (VRMA_VSID ^ (VRMA_VSID << 25))) & kvm->arch.hpt_mask;
+		hash = (i ^ (VRMA_VSID ^ (VRMA_VSID << 25)))
+			& kvmppc_hpt_mask(&kvm->arch.hpt);
 		/*
 		 * We assume that the hash table is empty and no
 		 * vcpus are using it at this stage.  Since we create
@@ -338,11 +366,11 @@ static int kvmppc_mmu_book3s_64_hv_xlate(struct kvm_vcpu *vcpu, gva_t eaddr,
 		preempt_enable();
 		return -ENOENT;
 	}
-	hptep = (__be64 *)(kvm->arch.hpt_virt + (index << 4));
+	hptep = (__be64 *)(kvm->arch.hpt.virt + (index << 4));
 	v = orig_v = be64_to_cpu(hptep[0]) & ~HPTE_V_HVLOCK;
 	if (cpu_has_feature(CPU_FTR_ARCH_300))
 		v = hpte_new_to_old_v(v, be64_to_cpu(hptep[1]));
-	gr = kvm->arch.revmap[index].guest_rpte;
+	gr = kvm->arch.hpt.rev[index].guest_rpte;
 
 	unlock_hpte(hptep, orig_v);
 	preempt_enable();
@@ -392,8 +420,8 @@ static int instruction_is_store(unsigned int instr)
 	return (instr & mask) != 0;
 }
 
-static int kvmppc_hv_emulate_mmio(struct kvm_run *run, struct kvm_vcpu *vcpu,
-				  unsigned long gpa, gva_t ea, int is_store)
+int kvmppc_hv_emulate_mmio(struct kvm_run *run, struct kvm_vcpu *vcpu,
+			   unsigned long gpa, gva_t ea, int is_store)
 {
 	u32 last_inst;
 
@@ -458,6 +486,9 @@ int kvmppc_book3s_hv_page_fault(struct kvm_run *run, struct kvm_vcpu *vcpu,
 	unsigned long rcbits;
 	long mmio_update;
 
+	if (kvm_is_radix(kvm))
+		return kvmppc_book3s_radix_page_fault(run, vcpu, ea, dsisr);
+
 	/*
 	 * Real-mode code has already searched the HPT and found the
 	 * entry we're interested in.  Lock the entry and check that
@@ -480,8 +511,8 @@ int kvmppc_book3s_hv_page_fault(struct kvm_run *run, struct kvm_vcpu *vcpu,
 		}
 	}
 	index = vcpu->arch.pgfault_index;
-	hptep = (__be64 *)(kvm->arch.hpt_virt + (index << 4));
-	rev = &kvm->arch.revmap[index];
+	hptep = (__be64 *)(kvm->arch.hpt.virt + (index << 4));
+	rev = &kvm->arch.hpt.rev[index];
 	preempt_disable();
 	while (!try_lock_hpte(hptep, HPTE_V_HVLOCK))
 		cpu_relax();
@@ -695,12 +726,13 @@ static void kvmppc_rmap_reset(struct kvm *kvm)
 	srcu_read_unlock(&kvm->srcu, srcu_idx);
 }
 
+typedef int (*hva_handler_fn)(struct kvm *kvm, struct kvm_memory_slot *memslot,
+			      unsigned long gfn);
+
 static int kvm_handle_hva_range(struct kvm *kvm,
 				unsigned long start,
 				unsigned long end,
-				int (*handler)(struct kvm *kvm,
-					       unsigned long *rmapp,
-					       unsigned long gfn))
+				hva_handler_fn handler)
 {
 	int ret;
 	int retval = 0;
@@ -725,9 +757,7 @@ static int kvm_handle_hva_range(struct kvm *kvm,
 		gfn_end = hva_to_gfn_memslot(hva_end + PAGE_SIZE - 1, memslot);
 
 		for (; gfn < gfn_end; ++gfn) {
-			gfn_t gfn_offset = gfn - memslot->base_gfn;
-
-			ret = handler(kvm, &memslot->arch.rmap[gfn_offset], gfn);
+			ret = handler(kvm, memslot, gfn);
 			retval |= ret;
 		}
 	}
@@ -736,20 +766,61 @@ static int kvm_handle_hva_range(struct kvm *kvm,
 }
 
 static int kvm_handle_hva(struct kvm *kvm, unsigned long hva,
-			  int (*handler)(struct kvm *kvm, unsigned long *rmapp,
-					 unsigned long gfn))
+			  hva_handler_fn handler)
 {
 	return kvm_handle_hva_range(kvm, hva, hva + 1, handler);
 }
 
-static int kvm_unmap_rmapp(struct kvm *kvm, unsigned long *rmapp,
+/* Must be called with both HPTE and rmap locked */
+static void kvmppc_unmap_hpte(struct kvm *kvm, unsigned long i,
+			      unsigned long *rmapp, unsigned long gfn)
+{
+	__be64 *hptep = (__be64 *) (kvm->arch.hpt.virt + (i << 4));
+	struct revmap_entry *rev = kvm->arch.hpt.rev;
+	unsigned long j, h;
+	unsigned long ptel, psize, rcbits;
+
+	j = rev[i].forw;
+	if (j == i) {
+		/* chain is now empty */
+		*rmapp &= ~(KVMPPC_RMAP_PRESENT | KVMPPC_RMAP_INDEX);
+	} else {
+		/* remove i from chain */
+		h = rev[i].back;
+		rev[h].forw = j;
+		rev[j].back = h;
+		rev[i].forw = rev[i].back = i;
+		*rmapp = (*rmapp & ~KVMPPC_RMAP_INDEX) | j;
+	}
+
+	/* Now check and modify the HPTE */
+	ptel = rev[i].guest_rpte;
+	psize = hpte_page_size(be64_to_cpu(hptep[0]), ptel);
+	if ((be64_to_cpu(hptep[0]) & HPTE_V_VALID) &&
+	    hpte_rpn(ptel, psize) == gfn) {
+		hptep[0] |= cpu_to_be64(HPTE_V_ABSENT);
+		kvmppc_invalidate_hpte(kvm, hptep, i);
+		hptep[1] &= ~cpu_to_be64(HPTE_R_KEY_HI | HPTE_R_KEY_LO);
+		/* Harvest R and C */
+		rcbits = be64_to_cpu(hptep[1]) & (HPTE_R_R | HPTE_R_C);
+		*rmapp |= rcbits << KVMPPC_RMAP_RC_SHIFT;
+		if (rcbits & HPTE_R_C)
+			kvmppc_update_rmap_change(rmapp, psize);
+		if (rcbits & ~rev[i].guest_rpte) {
+			rev[i].guest_rpte = ptel | rcbits;
+			note_hpte_modification(kvm, &rev[i]);
+		}
+	}
+}
+
+static int kvm_unmap_rmapp(struct kvm *kvm, struct kvm_memory_slot *memslot,
 			   unsigned long gfn)
 {
-	struct revmap_entry *rev = kvm->arch.revmap;
-	unsigned long h, i, j;
+	unsigned long i;
 	__be64 *hptep;
-	unsigned long ptel, psize, rcbits;
+	unsigned long *rmapp;
 
+	rmapp = &memslot->arch.rmap[gfn - memslot->base_gfn];
 	for (;;) {
 		lock_rmap(rmapp);
 		if (!(*rmapp & KVMPPC_RMAP_PRESENT)) {
@@ -763,7 +834,7 @@ static int kvm_unmap_rmapp(struct kvm *kvm, unsigned long *rmapp,
 		 * rmap chain lock.
 		 */
 		i = *rmapp & KVMPPC_RMAP_INDEX;
-		hptep = (__be64 *) (kvm->arch.hpt_virt + (i << 4));
+		hptep = (__be64 *) (kvm->arch.hpt.virt + (i << 4));
 		if (!try_lock_hpte(hptep, HPTE_V_HVLOCK)) {
 			/* unlock rmap before spinning on the HPTE lock */
 			unlock_rmap(rmapp);
@@ -771,37 +842,8 @@ static int kvm_unmap_rmapp(struct kvm *kvm, unsigned long *rmapp,
 				cpu_relax();
 			continue;
 		}
-		j = rev[i].forw;
-		if (j == i) {
-			/* chain is now empty */
-			*rmapp &= ~(KVMPPC_RMAP_PRESENT | KVMPPC_RMAP_INDEX);
-		} else {
-			/* remove i from chain */
-			h = rev[i].back;
-			rev[h].forw = j;
-			rev[j].back = h;
-			rev[i].forw = rev[i].back = i;
-			*rmapp = (*rmapp & ~KVMPPC_RMAP_INDEX) | j;
-		}
 
-		/* Now check and modify the HPTE */
-		ptel = rev[i].guest_rpte;
-		psize = hpte_page_size(be64_to_cpu(hptep[0]), ptel);
-		if ((be64_to_cpu(hptep[0]) & HPTE_V_VALID) &&
-		    hpte_rpn(ptel, psize) == gfn) {
-			hptep[0] |= cpu_to_be64(HPTE_V_ABSENT);
-			kvmppc_invalidate_hpte(kvm, hptep, i);
-			hptep[1] &= ~cpu_to_be64(HPTE_R_KEY_HI | HPTE_R_KEY_LO);
-			/* Harvest R and C */
-			rcbits = be64_to_cpu(hptep[1]) & (HPTE_R_R | HPTE_R_C);
-			*rmapp |= rcbits << KVMPPC_RMAP_RC_SHIFT;
-			if (rcbits & HPTE_R_C)
-				kvmppc_update_rmap_change(rmapp, psize);
-			if (rcbits & ~rev[i].guest_rpte) {
-				rev[i].guest_rpte = ptel | rcbits;
-				note_hpte_modification(kvm, &rev[i]);
-			}
-		}
+		kvmppc_unmap_hpte(kvm, i, rmapp, gfn);
 		unlock_rmap(rmapp);
 		__unlock_hpte(hptep, be64_to_cpu(hptep[0]));
 	}
@@ -810,26 +852,36 @@ static int kvm_unmap_rmapp(struct kvm *kvm, unsigned long *rmapp,
 
 int kvm_unmap_hva_hv(struct kvm *kvm, unsigned long hva)
 {
-	kvm_handle_hva(kvm, hva, kvm_unmap_rmapp);
+	hva_handler_fn handler;
+
+	handler = kvm_is_radix(kvm) ? kvm_unmap_radix : kvm_unmap_rmapp;
+	kvm_handle_hva(kvm, hva, handler);
 	return 0;
 }
 
 int kvm_unmap_hva_range_hv(struct kvm *kvm, unsigned long start, unsigned long end)
 {
-	kvm_handle_hva_range(kvm, start, end, kvm_unmap_rmapp);
+	hva_handler_fn handler;
+
+	handler = kvm_is_radix(kvm) ? kvm_unmap_radix : kvm_unmap_rmapp;
+	kvm_handle_hva_range(kvm, start, end, handler);
 	return 0;
 }
 
 void kvmppc_core_flush_memslot_hv(struct kvm *kvm,
 				  struct kvm_memory_slot *memslot)
 {
-	unsigned long *rmapp;
 	unsigned long gfn;
 	unsigned long n;
+	unsigned long *rmapp;
 
-	rmapp = memslot->arch.rmap;
 	gfn = memslot->base_gfn;
-	for (n = memslot->npages; n; --n) {
+	rmapp = memslot->arch.rmap;
+	for (n = memslot->npages; n; --n, ++gfn) {
+		if (kvm_is_radix(kvm)) {
+			kvm_unmap_radix(kvm, memslot, gfn);
+			continue;
+		}
 		/*
 		 * Testing the present bit without locking is OK because
 		 * the memslot has been marked invalid already, and hence
@@ -837,20 +889,21 @@ void kvmppc_core_flush_memslot_hv(struct kvm *kvm,
 		 * thus the present bit can't go from 0 to 1.
 		 */
 		if (*rmapp & KVMPPC_RMAP_PRESENT)
-			kvm_unmap_rmapp(kvm, rmapp, gfn);
+			kvm_unmap_rmapp(kvm, memslot, gfn);
 		++rmapp;
-		++gfn;
 	}
 }
 
-static int kvm_age_rmapp(struct kvm *kvm, unsigned long *rmapp,
+static int kvm_age_rmapp(struct kvm *kvm, struct kvm_memory_slot *memslot,
 			 unsigned long gfn)
 {
-	struct revmap_entry *rev = kvm->arch.revmap;
+	struct revmap_entry *rev = kvm->arch.hpt.rev;
 	unsigned long head, i, j;
 	__be64 *hptep;
 	int ret = 0;
+	unsigned long *rmapp;
 
+	rmapp = &memslot->arch.rmap[gfn - memslot->base_gfn];
  retry:
 	lock_rmap(rmapp);
 	if (*rmapp & KVMPPC_RMAP_REFERENCED) {
@@ -864,7 +917,7 @@ static int kvm_age_rmapp(struct kvm *kvm, unsigned long *rmapp,
 
 	i = head = *rmapp & KVMPPC_RMAP_INDEX;
 	do {
-		hptep = (__be64 *) (kvm->arch.hpt_virt + (i << 4));
+		hptep = (__be64 *) (kvm->arch.hpt.virt + (i << 4));
 		j = rev[i].forw;
 
 		/* If this HPTE isn't referenced, ignore it */
@@ -898,17 +951,22 @@ static int kvm_age_rmapp(struct kvm *kvm, unsigned long *rmapp,
 
 int kvm_age_hva_hv(struct kvm *kvm, unsigned long start, unsigned long end)
 {
-	return kvm_handle_hva_range(kvm, start, end, kvm_age_rmapp);
+	hva_handler_fn handler;
+
+	handler = kvm_is_radix(kvm) ? kvm_age_radix : kvm_age_rmapp;
+	return kvm_handle_hva_range(kvm, start, end, handler);
 }
 
-static int kvm_test_age_rmapp(struct kvm *kvm, unsigned long *rmapp,
+static int kvm_test_age_rmapp(struct kvm *kvm, struct kvm_memory_slot *memslot,
 			      unsigned long gfn)
 {
-	struct revmap_entry *rev = kvm->arch.revmap;
+	struct revmap_entry *rev = kvm->arch.hpt.rev;
 	unsigned long head, i, j;
 	unsigned long *hp;
 	int ret = 1;
+	unsigned long *rmapp;
 
+	rmapp = &memslot->arch.rmap[gfn - memslot->base_gfn];
 	if (*rmapp & KVMPPC_RMAP_REFERENCED)
 		return 1;
 
@@ -919,7 +977,7 @@ static int kvm_test_age_rmapp(struct kvm *kvm, unsigned long *rmapp,
 	if (*rmapp & KVMPPC_RMAP_PRESENT) {
 		i = head = *rmapp & KVMPPC_RMAP_INDEX;
 		do {
-			hp = (unsigned long *)(kvm->arch.hpt_virt + (i << 4));
+			hp = (unsigned long *)(kvm->arch.hpt.virt + (i << 4));
 			j = rev[i].forw;
 			if (be64_to_cpu(hp[1]) & HPTE_R_R)
 				goto out;
@@ -934,12 +992,18 @@ static int kvm_test_age_rmapp(struct kvm *kvm, unsigned long *rmapp,
 
 int kvm_test_age_hva_hv(struct kvm *kvm, unsigned long hva)
 {
-	return kvm_handle_hva(kvm, hva, kvm_test_age_rmapp);
+	hva_handler_fn handler;
+
+	handler = kvm_is_radix(kvm) ? kvm_test_age_radix : kvm_test_age_rmapp;
+	return kvm_handle_hva(kvm, hva, handler);
 }
 
 void kvm_set_spte_hva_hv(struct kvm *kvm, unsigned long hva, pte_t pte)
 {
-	kvm_handle_hva(kvm, hva, kvm_unmap_rmapp);
+	hva_handler_fn handler;
+
+	handler = kvm_is_radix(kvm) ? kvm_unmap_radix : kvm_unmap_rmapp;
+	kvm_handle_hva(kvm, hva, handler);
 }
 
 static int vcpus_running(struct kvm *kvm)
@@ -953,7 +1017,7 @@ static int vcpus_running(struct kvm *kvm)
  */
 static int kvm_test_clear_dirty_npages(struct kvm *kvm, unsigned long *rmapp)
 {
-	struct revmap_entry *rev = kvm->arch.revmap;
+	struct revmap_entry *rev = kvm->arch.hpt.rev;
 	unsigned long head, i, j;
 	unsigned long n;
 	unsigned long v, r;
@@ -978,7 +1042,7 @@ static int kvm_test_clear_dirty_npages(struct kvm *kvm, unsigned long *rmapp)
 	i = head = *rmapp & KVMPPC_RMAP_INDEX;
 	do {
 		unsigned long hptep1;
-		hptep = (__be64 *) (kvm->arch.hpt_virt + (i << 4));
+		hptep = (__be64 *) (kvm->arch.hpt.virt + (i << 4));
 		j = rev[i].forw;
 
 		/*
@@ -1040,7 +1104,7 @@ static int kvm_test_clear_dirty_npages(struct kvm *kvm, unsigned long *rmapp)
 	return npages_dirty;
 }
 
-static void harvest_vpa_dirty(struct kvmppc_vpa *vpa,
+void kvmppc_harvest_vpa_dirty(struct kvmppc_vpa *vpa,
 			      struct kvm_memory_slot *memslot,
 			      unsigned long *map)
 {
@@ -1058,12 +1122,11 @@ static void harvest_vpa_dirty(struct kvmppc_vpa *vpa,
 		__set_bit_le(gfn - memslot->base_gfn, map);
 }
 
-long kvmppc_hv_get_dirty_log(struct kvm *kvm, struct kvm_memory_slot *memslot,
-			     unsigned long *map)
+long kvmppc_hv_get_dirty_log_hpt(struct kvm *kvm,
+			struct kvm_memory_slot *memslot, unsigned long *map)
 {
 	unsigned long i, j;
 	unsigned long *rmapp;
-	struct kvm_vcpu *vcpu;
 
 	preempt_disable();
 	rmapp = memslot->arch.rmap;
@@ -1079,15 +1142,6 @@ long kvmppc_hv_get_dirty_log(struct kvm *kvm, struct kvm_memory_slot *memslot,
 				__set_bit_le(j, map);
 		++rmapp;
 	}
-
-	/* Harvest dirty bits from VPA and DTL updates */
-	/* Note: we never modify the SLB shadow buffer areas */
-	kvm_for_each_vcpu(i, vcpu, kvm) {
-		spin_lock(&vcpu->arch.vpa_update_lock);
-		harvest_vpa_dirty(&vcpu->arch.vpa, memslot, map);
-		harvest_vpa_dirty(&vcpu->arch.dtl, memslot, map);
-		spin_unlock(&vcpu->arch.vpa_update_lock);
-	}
 	preempt_enable();
 	return 0;
 }
@@ -1142,14 +1196,366 @@ void kvmppc_unpin_guest_page(struct kvm *kvm, void *va, unsigned long gpa,
 	srcu_idx = srcu_read_lock(&kvm->srcu);
 	memslot = gfn_to_memslot(kvm, gfn);
 	if (memslot) {
-		rmap = &memslot->arch.rmap[gfn - memslot->base_gfn];
-		lock_rmap(rmap);
-		*rmap |= KVMPPC_RMAP_CHANGED;
-		unlock_rmap(rmap);
+		if (!kvm_is_radix(kvm)) {
+			rmap = &memslot->arch.rmap[gfn - memslot->base_gfn];
+			lock_rmap(rmap);
+			*rmap |= KVMPPC_RMAP_CHANGED;
+			unlock_rmap(rmap);
+		} else if (memslot->dirty_bitmap) {
+			mark_page_dirty(kvm, gfn);
+		}
 	}
 	srcu_read_unlock(&kvm->srcu, srcu_idx);
 }
 
+/*
+ * HPT resizing
+ */
+static int resize_hpt_allocate(struct kvm_resize_hpt *resize)
+{
+	int rc;
+
+	rc = kvmppc_allocate_hpt(&resize->hpt, resize->order);
+	if (rc < 0)
+		return rc;
+
+	resize_hpt_debug(resize, "resize_hpt_allocate(): HPT @ 0x%lx\n",
+			 resize->hpt.virt);
+
+	return 0;
+}
+
+static unsigned long resize_hpt_rehash_hpte(struct kvm_resize_hpt *resize,
+					    unsigned long idx)
+{
+	struct kvm *kvm = resize->kvm;
+	struct kvm_hpt_info *old = &kvm->arch.hpt;
+	struct kvm_hpt_info *new = &resize->hpt;
+	unsigned long old_hash_mask = (1ULL << (old->order - 7)) - 1;
+	unsigned long new_hash_mask = (1ULL << (new->order - 7)) - 1;
+	__be64 *hptep, *new_hptep;
+	unsigned long vpte, rpte, guest_rpte;
+	int ret;
+	struct revmap_entry *rev;
+	unsigned long apsize, psize, avpn, pteg, hash;
+	unsigned long new_idx, new_pteg, replace_vpte;
+
+	hptep = (__be64 *)(old->virt + (idx << 4));
+
+	/* Guest is stopped, so new HPTEs can't be added or faulted
+	 * in, only unmapped or altered by host actions.  So, it's
+	 * safe to check this before we take the HPTE lock */
+	vpte = be64_to_cpu(hptep[0]);
+	if (!(vpte & HPTE_V_VALID) && !(vpte & HPTE_V_ABSENT))
+		return 0; /* nothing to do */
+
+	while (!try_lock_hpte(hptep, HPTE_V_HVLOCK))
+		cpu_relax();
+
+	vpte = be64_to_cpu(hptep[0]);
+
+	ret = 0;
+	if (!(vpte & HPTE_V_VALID) && !(vpte & HPTE_V_ABSENT))
+		/* Nothing to do */
+		goto out;
+
+	/* Unmap */
+	rev = &old->rev[idx];
+	guest_rpte = rev->guest_rpte;
+
+	ret = -EIO;
+	apsize = hpte_page_size(vpte, guest_rpte);
+	if (!apsize)
+		goto out;
+
+	if (vpte & HPTE_V_VALID) {
+		unsigned long gfn = hpte_rpn(guest_rpte, apsize);
+		int srcu_idx = srcu_read_lock(&kvm->srcu);
+		struct kvm_memory_slot *memslot =
+			__gfn_to_memslot(kvm_memslots(kvm), gfn);
+
+		if (memslot) {
+			unsigned long *rmapp;
+			rmapp = &memslot->arch.rmap[gfn - memslot->base_gfn];
+
+			lock_rmap(rmapp);
+			kvmppc_unmap_hpte(kvm, idx, rmapp, gfn);
+			unlock_rmap(rmapp);
+		}
+
+		srcu_read_unlock(&kvm->srcu, srcu_idx);
+	}
+
+	/* Reload PTE after unmap */
+	vpte = be64_to_cpu(hptep[0]);
+
+	BUG_ON(vpte & HPTE_V_VALID);
+	BUG_ON(!(vpte & HPTE_V_ABSENT));
+
+	ret = 0;
+	if (!(vpte & HPTE_V_BOLTED))
+		goto out;
+
+	rpte = be64_to_cpu(hptep[1]);
+	psize = hpte_base_page_size(vpte, rpte);
+	avpn = HPTE_V_AVPN_VAL(vpte) & ~((psize - 1) >> 23);
+	pteg = idx / HPTES_PER_GROUP;
+	if (vpte & HPTE_V_SECONDARY)
+		pteg = ~pteg;
+
+	if (!(vpte & HPTE_V_1TB_SEG)) {
+		unsigned long offset, vsid;
+
+		/* We only have 28 - 23 bits of offset in avpn */
+		offset = (avpn & 0x1f) << 23;
+		vsid = avpn >> 5;
+		/* We can find more bits from the pteg value */
+		if (psize < (1ULL << 23))
+			offset |= ((vsid ^ pteg) & old_hash_mask) * psize;
+
+		hash = vsid ^ (offset / psize);
+	} else {
+		unsigned long offset, vsid;
+
+		/* We only have 40 - 23 bits of seg_off in avpn */
+		offset = (avpn & 0x1ffff) << 23;
+		vsid = avpn >> 17;
+		if (psize < (1ULL << 23))
+			offset |= ((vsid ^ (vsid << 25) ^ pteg) & old_hash_mask) * psize;
+
+		hash = vsid ^ (vsid << 25) ^ (offset / psize);
+	}
+
+	new_pteg = hash & new_hash_mask;
+	if (vpte & HPTE_V_SECONDARY) {
+		BUG_ON(~pteg != (hash & old_hash_mask));
+		new_pteg = ~new_pteg;
+	} else {
+		BUG_ON(pteg != (hash & old_hash_mask));
+	}
+
+	new_idx = new_pteg * HPTES_PER_GROUP + (idx % HPTES_PER_GROUP);
+	new_hptep = (__be64 *)(new->virt + (new_idx << 4));
+
+	replace_vpte = be64_to_cpu(new_hptep[0]);
+
+	if (replace_vpte & (HPTE_V_VALID | HPTE_V_ABSENT)) {
+		BUG_ON(new->order >= old->order);
+
+		if (replace_vpte & HPTE_V_BOLTED) {
+			if (vpte & HPTE_V_BOLTED)
+				/* Bolted collision, nothing we can do */
+				ret = -ENOSPC;
+			/* Discard the new HPTE */
+			goto out;
+		}
+
+		/* Discard the previous HPTE */
+	}
+
+	new_hptep[1] = cpu_to_be64(rpte);
+	new->rev[new_idx].guest_rpte = guest_rpte;
+	/* No need for a barrier, since new HPT isn't active */
+	new_hptep[0] = cpu_to_be64(vpte);
+	unlock_hpte(new_hptep, vpte);
+
+out:
+	unlock_hpte(hptep, vpte);
+	return ret;
+}
+
+static int resize_hpt_rehash(struct kvm_resize_hpt *resize)
+{
+	struct kvm *kvm = resize->kvm;
+	unsigned  long i;
+	int rc;
+
+	for (i = 0; i < kvmppc_hpt_npte(&kvm->arch.hpt); i++) {
+		rc = resize_hpt_rehash_hpte(resize, i);
+		if (rc != 0)
+			return rc;
+	}
+
+	return 0;
+}
+
+static void resize_hpt_pivot(struct kvm_resize_hpt *resize)
+{
+	struct kvm *kvm = resize->kvm;
+	struct kvm_hpt_info hpt_tmp;
+
+	/* Exchange the pending tables in the resize structure with
+	 * the active tables */
+
+	resize_hpt_debug(resize, "resize_hpt_pivot()\n");
+
+	spin_lock(&kvm->mmu_lock);
+	asm volatile("ptesync" : : : "memory");
+
+	hpt_tmp = kvm->arch.hpt;
+	kvmppc_set_hpt(kvm, &resize->hpt);
+	resize->hpt = hpt_tmp;
+
+	spin_unlock(&kvm->mmu_lock);
+
+	synchronize_srcu_expedited(&kvm->srcu);
+
+	resize_hpt_debug(resize, "resize_hpt_pivot() done\n");
+}
+
+static void resize_hpt_release(struct kvm *kvm, struct kvm_resize_hpt *resize)
+{
+	BUG_ON(kvm->arch.resize_hpt != resize);
+
+	if (resize->hpt.virt)
+		kvmppc_free_hpt(&resize->hpt);
+
+	kvm->arch.resize_hpt = NULL;
+	kfree(resize);
+}
+
+static void resize_hpt_prepare_work(struct work_struct *work)
+{
+	struct kvm_resize_hpt *resize = container_of(work,
+						     struct kvm_resize_hpt,
+						     work);
+	struct kvm *kvm = resize->kvm;
+	int err;
+
+	resize_hpt_debug(resize, "resize_hpt_prepare_work(): order = %d\n",
+			 resize->order);
+
+	err = resize_hpt_allocate(resize);
+
+	mutex_lock(&kvm->lock);
+
+	resize->error = err;
+	resize->prepare_done = true;
+
+	mutex_unlock(&kvm->lock);
+}
+
+long kvm_vm_ioctl_resize_hpt_prepare(struct kvm *kvm,
+				     struct kvm_ppc_resize_hpt *rhpt)
+{
+	unsigned long flags = rhpt->flags;
+	unsigned long shift = rhpt->shift;
+	struct kvm_resize_hpt *resize;
+	int ret;
+
+	if (flags != 0)
+		return -EINVAL;
+
+	if (shift && ((shift < 18) || (shift > 46)))
+		return -EINVAL;
+
+	mutex_lock(&kvm->lock);
+
+	resize = kvm->arch.resize_hpt;
+
+	if (resize) {
+		if (resize->order == shift) {
+			/* Suitable resize in progress */
+			if (resize->prepare_done) {
+				ret = resize->error;
+				if (ret != 0)
+					resize_hpt_release(kvm, resize);
+			} else {
+				ret = 100; /* estimated time in ms */
+			}
+
+			goto out;
+		}
+
+		/* not suitable, cancel it */
+		resize_hpt_release(kvm, resize);
+	}
+
+	ret = 0;
+	if (!shift)
+		goto out; /* nothing to do */
+
+	/* start new resize */
+
+	resize = kzalloc(sizeof(*resize), GFP_KERNEL);
+	resize->order = shift;
+	resize->kvm = kvm;
+	INIT_WORK(&resize->work, resize_hpt_prepare_work);
+	kvm->arch.resize_hpt = resize;
+
+	schedule_work(&resize->work);
+
+	ret = 100; /* estimated time in ms */
+
+out:
+	mutex_unlock(&kvm->lock);
+	return ret;
+}
+
+static void resize_hpt_boot_vcpu(void *opaque)
+{
+	/* Nothing to do, just force a KVM exit */
+}
+
+long kvm_vm_ioctl_resize_hpt_commit(struct kvm *kvm,
+				    struct kvm_ppc_resize_hpt *rhpt)
+{
+	unsigned long flags = rhpt->flags;
+	unsigned long shift = rhpt->shift;
+	struct kvm_resize_hpt *resize;
+	long ret;
+
+	if (flags != 0)
+		return -EINVAL;
+
+	if (shift && ((shift < 18) || (shift > 46)))
+		return -EINVAL;
+
+	mutex_lock(&kvm->lock);
+
+	resize = kvm->arch.resize_hpt;
+
+	/* This shouldn't be possible */
+	ret = -EIO;
+	if (WARN_ON(!kvm->arch.hpte_setup_done))
+		goto out_no_hpt;
+
+	/* Stop VCPUs from running while we mess with the HPT */
+	kvm->arch.hpte_setup_done = 0;
+	smp_mb();
+
+	/* Boot all CPUs out of the guest so they re-read
+	 * hpte_setup_done */
+	on_each_cpu(resize_hpt_boot_vcpu, NULL, 1);
+
+	ret = -ENXIO;
+	if (!resize || (resize->order != shift))
+		goto out;
+
+	ret = -EBUSY;
+	if (!resize->prepare_done)
+		goto out;
+
+	ret = resize->error;
+	if (ret != 0)
+		goto out;
+
+	ret = resize_hpt_rehash(resize);
+	if (ret != 0)
+		goto out;
+
+	resize_hpt_pivot(resize);
+
+out:
+	/* Let VCPUs run again */
+	kvm->arch.hpte_setup_done = 1;
+	smp_mb();
+out_no_hpt:
+	resize_hpt_release(kvm, resize);
+	mutex_unlock(&kvm->lock);
+	return ret;
+}
+
 /*
  * Functions for reading and writing the hash table via reads and
  * writes on a file descriptor.
@@ -1290,8 +1696,8 @@ static ssize_t kvm_htab_read(struct file *file, char __user *buf,
 	flags = ctx->flags;
 
 	i = ctx->index;
-	hptp = (__be64 *)(kvm->arch.hpt_virt + (i * HPTE_SIZE));
-	revp = kvm->arch.revmap + i;
+	hptp = (__be64 *)(kvm->arch.hpt.virt + (i * HPTE_SIZE));
+	revp = kvm->arch.hpt.rev + i;
 	lbuf = (unsigned long __user *)buf;
 
 	nb = 0;
@@ -1306,7 +1712,7 @@ static ssize_t kvm_htab_read(struct file *file, char __user *buf,
 
 		/* Skip uninteresting entries, i.e. clean on not-first pass */
 		if (!first_pass) {
-			while (i < kvm->arch.hpt_npte &&
+			while (i < kvmppc_hpt_npte(&kvm->arch.hpt) &&
 			       !hpte_dirty(revp, hptp)) {
 				++i;
 				hptp += 2;
@@ -1316,7 +1722,7 @@ static ssize_t kvm_htab_read(struct file *file, char __user *buf,
 		hdr.index = i;
 
 		/* Grab a series of valid entries */
-		while (i < kvm->arch.hpt_npte &&
+		while (i < kvmppc_hpt_npte(&kvm->arch.hpt) &&
 		       hdr.n_valid < 0xffff &&
 		       nb + HPTE_SIZE < count &&
 		       record_hpte(flags, hptp, hpte, revp, 1, first_pass)) {
@@ -1332,7 +1738,7 @@ static ssize_t kvm_htab_read(struct file *file, char __user *buf,
 			++revp;
 		}
 		/* Now skip invalid entries while we can */
-		while (i < kvm->arch.hpt_npte &&
+		while (i < kvmppc_hpt_npte(&kvm->arch.hpt) &&
 		       hdr.n_invalid < 0xffff &&
 		       record_hpte(flags, hptp, hpte, revp, 0, first_pass)) {
 			/* found an invalid entry */
@@ -1353,7 +1759,7 @@ static ssize_t kvm_htab_read(struct file *file, char __user *buf,
 		}
 
 		/* Check if we've wrapped around the hash table */
-		if (i >= kvm->arch.hpt_npte) {
+		if (i >= kvmppc_hpt_npte(&kvm->arch.hpt)) {
 			i = 0;
 			ctx->first_pass = 0;
 			break;
@@ -1412,11 +1818,11 @@ static ssize_t kvm_htab_write(struct file *file, const char __user *buf,
 
 		err = -EINVAL;
 		i = hdr.index;
-		if (i >= kvm->arch.hpt_npte ||
-		    i + hdr.n_valid + hdr.n_invalid > kvm->arch.hpt_npte)
+		if (i >= kvmppc_hpt_npte(&kvm->arch.hpt) ||
+		    i + hdr.n_valid + hdr.n_invalid > kvmppc_hpt_npte(&kvm->arch.hpt))
 			break;
 
-		hptp = (__be64 *)(kvm->arch.hpt_virt + (i * HPTE_SIZE));
+		hptp = (__be64 *)(kvm->arch.hpt.virt + (i * HPTE_SIZE));
 		lbuf = (unsigned long __user *)buf;
 		for (j = 0; j < hdr.n_valid; ++j) {
 			__be64 hpte_v;
@@ -1603,8 +2009,9 @@ static ssize_t debugfs_htab_read(struct file *file, char __user *buf,
 
 	kvm = p->kvm;
 	i = p->hpt_index;
-	hptp = (__be64 *)(kvm->arch.hpt_virt + (i * HPTE_SIZE));
-	for (; len != 0 && i < kvm->arch.hpt_npte; ++i, hptp += 2) {
+	hptp = (__be64 *)(kvm->arch.hpt.virt + (i * HPTE_SIZE));
+	for (; len != 0 && i < kvmppc_hpt_npte(&kvm->arch.hpt);
+	     ++i, hptp += 2) {
 		if (!(be64_to_cpu(hptp[0]) & (HPTE_V_VALID | HPTE_V_ABSENT)))
 			continue;
 
@@ -1614,7 +2021,7 @@ static ssize_t debugfs_htab_read(struct file *file, char __user *buf,
 			cpu_relax();
 		v = be64_to_cpu(hptp[0]) & ~HPTE_V_HVLOCK;
 		hr = be64_to_cpu(hptp[1]);
-		gr = kvm->arch.revmap[i].guest_rpte;
+		gr = kvm->arch.hpt.rev[i].guest_rpte;
 		unlock_hpte(hptp, v);
 		preempt_enable();
 
@@ -1675,7 +2082,10 @@ void kvmppc_mmu_book3s_hv_init(struct kvm_vcpu *vcpu)
 
 	vcpu->arch.slb_nr = 32;		/* POWER7/POWER8 */
 
-	mmu->xlate = kvmppc_mmu_book3s_64_hv_xlate;
+	if (kvm_is_radix(vcpu->kvm))
+		mmu->xlate = kvmppc_mmu_radix_xlate;
+	else
+		mmu->xlate = kvmppc_mmu_book3s_64_hv_xlate;
 	mmu->reset_msr = kvmppc_mmu_book3s_64_hv_reset_msr;
 
 	vcpu->arch.hflags |= BOOK3S_HFLAG_SLB;

arch/powerpc/kvm/book3s_64_mmu_radix.c

+/*
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, version 2, as
+ * published by the Free Software Foundation.
+ *
+ * Copyright 2016 Paul Mackerras, IBM Corp. <paulus@au1.ibm.com>
+ */
+
+#include <linux/types.h>
+#include <linux/string.h>
+#include <linux/kvm.h>
+#include <linux/kvm_host.h>
+
+#include <asm/kvm_ppc.h>
+#include <asm/kvm_book3s.h>
+#include <asm/page.h>
+#include <asm/mmu.h>
+#include <asm/pgtable.h>
+#include <asm/pgalloc.h>
+
+/*
+ * Supported radix tree geometry.
+ * Like p9, we support either 5 or 9 bits at the first (lowest) level,
+ * for a page size of 64k or 4k.
+ */
+static int p9_supported_radix_bits[4] = { 5, 9, 9, 13 };
+
+int kvmppc_mmu_radix_xlate(struct kvm_vcpu *vcpu, gva_t eaddr,
+			   struct kvmppc_pte *gpte, bool data, bool iswrite)
+{
+	struct kvm *kvm = vcpu->kvm;
+	u32 pid;
+	int ret, level, ps;
+	__be64 prte, rpte;
+	unsigned long root, pte, index;
+	unsigned long rts, bits, offset;
+	unsigned long gpa;
+	unsigned long proc_tbl_size;
+
+	/* Work out effective PID */
+	switch (eaddr >> 62) {
+	case 0:
+		pid = vcpu->arch.pid;
+		break;
+	case 3:
+		pid = 0;
+		break;
+	default:
+		return -EINVAL;
+	}
+	proc_tbl_size = 1 << ((kvm->arch.process_table & PRTS_MASK) + 12);
+	if (pid * 16 >= proc_tbl_size)
+		return -EINVAL;
+
+	/* Read partition table to find root of tree for effective PID */
+	ret = kvm_read_guest(kvm, kvm->arch.process_table + pid * 16,
+			     &prte, sizeof(prte));
+	if (ret)
+		return ret;
+
+	root = be64_to_cpu(prte);
+	rts = ((root & RTS1_MASK) >> (RTS1_SHIFT - 3)) |
+		((root & RTS2_MASK) >> RTS2_SHIFT);
+	bits = root & RPDS_MASK;
+	root = root & RPDB_MASK;
+
+	/* P9 DD1 interprets RTS (radix tree size) differently */
+	offset = rts + 31;
+	if (cpu_has_feature(CPU_FTR_POWER9_DD1))
+		offset -= 3;
+
+	/* current implementations only support 52-bit space */
+	if (offset != 52)
+		return -EINVAL;
+
+	for (level = 3; level >= 0; --level) {
+		if (level && bits != p9_supported_radix_bits[level])
+			return -EINVAL;
+		if (level == 0 && !(bits == 5 || bits == 9))
+			return -EINVAL;
+		offset -= bits;
+		index = (eaddr >> offset) & ((1UL << bits) - 1);
+		/* check that low bits of page table base are zero */
+		if (root & ((1UL << (bits + 3)) - 1))
+			return -EINVAL;
+		ret = kvm_read_guest(kvm, root + index * 8,
+				     &rpte, sizeof(rpte));
+		if (ret)
+			return ret;
+		pte = __be64_to_cpu(rpte);
+		if (!(pte & _PAGE_PRESENT))
+			return -ENOENT;
+		if (pte & _PAGE_PTE)
+			break;
+		bits = pte & 0x1f;
+		root = pte & 0x0fffffffffffff00ul;
+	}
+	/* need a leaf at lowest level; 512GB pages not supported */
+	if (level < 0 || level == 3)
+		return -EINVAL;
+
+	/* offset is now log base 2 of the page size */
+	gpa = pte & 0x01fffffffffff000ul;
+	if (gpa & ((1ul << offset) - 1))
+		return -EINVAL;
+	gpa += eaddr & ((1ul << offset) - 1);
+	for (ps = MMU_PAGE_4K; ps < MMU_PAGE_COUNT; ++ps)
+		if (offset == mmu_psize_defs[ps].shift)
+			break;
+	gpte->page_size = ps;
+
+	gpte->eaddr = eaddr;
+	gpte->raddr = gpa;
+
+	/* Work out permissions */
+	gpte->may_read = !!(pte & _PAGE_READ);
+	gpte->may_write = !!(pte & _PAGE_WRITE);
+	gpte->may_execute = !!(pte & _PAGE_EXEC);
+	if (kvmppc_get_msr(vcpu) & MSR_PR) {
+		if (pte & _PAGE_PRIVILEGED) {
+			gpte->may_read = 0;
+			gpte->may_write = 0;
+			gpte->may_execute = 0;
+		}
+	} else {
+		if (!(pte & _PAGE_PRIVILEGED)) {
+			/* Check AMR/IAMR to see if strict mode is in force */
+			if (vcpu->arch.amr & (1ul << 62))
+				gpte->may_read = 0;
+			if (vcpu->arch.amr & (1ul << 63))
+				gpte->may_write = 0;
+			if (vcpu->arch.iamr & (1ul << 62))
+				gpte->may_execute = 0;
+		}
+	}
+
+	return 0;
+}
+
+#ifdef CONFIG_PPC_64K_PAGES
+#define MMU_BASE_PSIZE	MMU_PAGE_64K
+#else
+#define MMU_BASE_PSIZE	MMU_PAGE_4K
+#endif
+
+static void kvmppc_radix_tlbie_page(struct kvm *kvm, unsigned long addr,
+				    unsigned int pshift)
+{
+	int psize = MMU_BASE_PSIZE;
+
+	if (pshift >= PMD_SHIFT)
+		psize = MMU_PAGE_2M;
+	addr &= ~0xfffUL;
+	addr |= mmu_psize_defs[psize].ap << 5;
+	asm volatile("ptesync": : :"memory");
+	asm volatile(PPC_TLBIE_5(%0, %1, 0, 0, 1)
+		     : : "r" (addr), "r" (kvm->arch.lpid) : "memory");
+	asm volatile("ptesync": : :"memory");
+}
+
+unsigned long kvmppc_radix_update_pte(struct kvm *kvm, pte_t *ptep,
+				      unsigned long clr, unsigned long set,
+				      unsigned long addr, unsigned int shift)
+{
+	unsigned long old = 0;
+
+	if (!(clr & _PAGE_PRESENT) && cpu_has_feature(CPU_FTR_POWER9_DD1) &&
+	    pte_present(*ptep)) {
+		/* have to invalidate it first */
+		old = __radix_pte_update(ptep, _PAGE_PRESENT, 0);
+		kvmppc_radix_tlbie_page(kvm, addr, shift);
+		set |= _PAGE_PRESENT;
+		old &= _PAGE_PRESENT;
+	}
+	return __radix_pte_update(ptep, clr, set) | old;
+}
+
+void kvmppc_radix_set_pte_at(struct kvm *kvm, unsigned long addr,
+			     pte_t *ptep, pte_t pte)
+{
+	radix__set_pte_at(kvm->mm, addr, ptep, pte, 0);
+}
+
+static struct kmem_cache *kvm_pte_cache;
+
+static pte_t *kvmppc_pte_alloc(void)
+{
+	return kmem_cache_alloc(kvm_pte_cache, GFP_KERNEL);
+}
+
+static void kvmppc_pte_free(pte_t *ptep)
+{
+	kmem_cache_free(kvm_pte_cache, ptep);
+}
+
+static int kvmppc_create_pte(struct kvm *kvm, pte_t pte, unsigned long gpa,
+			     unsigned int level, unsigned long mmu_seq)
+{
+	pgd_t *pgd;
+	pud_t *pud, *new_pud = NULL;
+	pmd_t *pmd, *new_pmd = NULL;
+	pte_t *ptep, *new_ptep = NULL;
+	unsigned long old;
+	int ret;
+
+	/* Traverse the guest's 2nd-level tree, allocate new levels needed */
+	pgd = kvm->arch.pgtable + pgd_index(gpa);
+	pud = NULL;
+	if (pgd_present(*pgd))
+		pud = pud_offset(pgd, gpa);
+	else
+		new_pud = pud_alloc_one(kvm->mm, gpa);
+
+	pmd = NULL;
+	if (pud && pud_present(*pud))
+		pmd = pmd_offset(pud, gpa);
+	else
+		new_pmd = pmd_alloc_one(kvm->mm, gpa);
+
+	if (level == 0 && !(pmd && pmd_present(*pmd)))
+		new_ptep = kvmppc_pte_alloc();
+
+	/* Check if we might have been invalidated; let the guest retry if so */
+	spin_lock(&kvm->mmu_lock);
+	ret = -EAGAIN;
+	if (mmu_notifier_retry(kvm, mmu_seq))
+		goto out_unlock;
+
+	/* Now traverse again under the lock and change the tree */
+	ret = -ENOMEM;
+	if (pgd_none(*pgd)) {
+		if (!new_pud)
+			goto out_unlock;
+		pgd_populate(kvm->mm, pgd, new_pud);
+		new_pud = NULL;
+	}
+	pud = pud_offset(pgd, gpa);
+	if (pud_none(*pud)) {
+		if (!new_pmd)
+			goto out_unlock;
+		pud_populate(kvm->mm, pud, new_pmd);
+		new_pmd = NULL;
+	}
+	pmd = pmd_offset(pud, gpa);
+	if (pmd_large(*pmd)) {
+		/* Someone else has instantiated a large page here; retry */
+		ret = -EAGAIN;
+		goto out_unlock;
+	}
+	if (level == 1 && !pmd_none(*pmd)) {
+		/*
+		 * There's a page table page here, but we wanted
+		 * to install a large page.  Tell the caller and let
+		 * it try installing a normal page if it wants.
+		 */
+		ret = -EBUSY;
+		goto out_unlock;
+	}
+	if (level == 0) {
+		if (pmd_none(*pmd)) {
+			if (!new_ptep)
+				goto out_unlock;
+			pmd_populate(kvm->mm, pmd, new_ptep);
+			new_ptep = NULL;
+		}
+		ptep = pte_offset_kernel(pmd, gpa);
+		if (pte_present(*ptep)) {
+			/* PTE was previously valid, so invalidate it */
+			old = kvmppc_radix_update_pte(kvm, ptep, _PAGE_PRESENT,
+						      0, gpa, 0);
+			kvmppc_radix_tlbie_page(kvm, gpa, 0);
+			if (old & _PAGE_DIRTY)
+				mark_page_dirty(kvm, gpa >> PAGE_SHIFT);
+		}
+		kvmppc_radix_set_pte_at(kvm, gpa, ptep, pte);
+	} else {
+		kvmppc_radix_set_pte_at(kvm, gpa, pmdp_ptep(pmd), pte);
+	}
+	ret = 0;
+
+ out_unlock:
+	spin_unlock(&kvm->mmu_lock);
+	if (new_pud)
+		pud_free(kvm->mm, new_pud);
+	if (new_pmd)
+		pmd_free(kvm->mm, new_pmd);
+	if (new_ptep)
+		kvmppc_pte_free(new_ptep);
+	return ret;
+}
+
+int kvmppc_book3s_radix_page_fault(struct kvm_run *run, struct kvm_vcpu *vcpu,
+				   unsigned long ea, unsigned long dsisr)
+{
+	struct kvm *kvm = vcpu->kvm;
+	unsigned long mmu_seq, pte_size;
+	unsigned long gpa, gfn, hva, pfn;
+	struct kvm_memory_slot *memslot;
+	struct page *page = NULL, *pages[1];
+	long ret, npages, ok;
+	unsigned int writing;
+	struct vm_area_struct *vma;
+	unsigned long flags;
+	pte_t pte, *ptep;
+	unsigned long pgflags;
+	unsigned int shift, level;
+
+	/* Check for unusual errors */
+	if (dsisr & DSISR_UNSUPP_MMU) {
+		pr_err("KVM: Got unsupported MMU fault\n");
+		return -EFAULT;
+	}
+	if (dsisr & DSISR_BADACCESS) {
+		/* Reflect to the guest as DSI */
+		pr_err("KVM: Got radix HV page fault with DSISR=%lx\n", dsisr);
+		kvmppc_core_queue_data_storage(vcpu, ea, dsisr);
+		return RESUME_GUEST;
+	}
+
+	/* Translate the logical address and get the page */
+	gpa = vcpu->arch.fault_gpa & ~0xfffUL;
+	gpa &= ~0xF000000000000000ul;
+	gfn = gpa >> PAGE_SHIFT;
+	if (!(dsisr & DSISR_PGDIRFAULT))
+		gpa |= ea & 0xfff;
+	memslot = gfn_to_memslot(kvm, gfn);
+
+	/* No memslot means it's an emulated MMIO region */
+	if (!memslot || (memslot->flags & KVM_MEMSLOT_INVALID)) {
+		if (dsisr & (DSISR_PGDIRFAULT | DSISR_BADACCESS |
+			     DSISR_SET_RC)) {
+			/*
+			 * Bad address in guest page table tree, or other
+			 * unusual error - reflect it to the guest as DSI.
+			 */
+			kvmppc_core_queue_data_storage(vcpu, ea, dsisr);
+			return RESUME_GUEST;
+		}
+		return kvmppc_hv_emulate_mmio(run, vcpu, gpa, ea,
+					      dsisr & DSISR_ISSTORE);
+	}
+
+	/* used to check for invalidations in progress */
+	mmu_seq = kvm->mmu_notifier_seq;
+	smp_rmb();
+
+	writing = (dsisr & DSISR_ISSTORE) != 0;
+	hva = gfn_to_hva_memslot(memslot, gfn);
+	if (dsisr & DSISR_SET_RC) {
+		/*
+		 * Need to set an R or C bit in the 2nd-level tables;
+		 * if the relevant bits aren't already set in the linux
+		 * page tables, fall through to do the gup_fast to
+		 * set them in the linux page tables too.
+		 */
+		ok = 0;
+		pgflags = _PAGE_ACCESSED;
+		if (writing)
+			pgflags |= _PAGE_DIRTY;
+		local_irq_save(flags);
+		ptep = __find_linux_pte_or_hugepte(current->mm->pgd, hva,
+						   NULL, NULL);
+		if (ptep) {
+			pte = READ_ONCE(*ptep);
+			if (pte_present(pte) &&
+			    (pte_val(pte) & pgflags) == pgflags)
+				ok = 1;
+		}
+		local_irq_restore(flags);
+		if (ok) {
+			spin_lock(&kvm->mmu_lock);
+			if (mmu_notifier_retry(vcpu->kvm, mmu_seq)) {
+				spin_unlock(&kvm->mmu_lock);
+				return RESUME_GUEST;
+			}
+			ptep = __find_linux_pte_or_hugepte(kvm->arch.pgtable,
+							gpa, NULL, &shift);
+			if (ptep && pte_present(*ptep)) {
+				kvmppc_radix_update_pte(kvm, ptep, 0, pgflags,
+							gpa, shift);
+				spin_unlock(&kvm->mmu_lock);
+				return RESUME_GUEST;
+			}
+			spin_unlock(&kvm->mmu_lock);
+		}
+	}
+
+	ret = -EFAULT;
+	pfn = 0;
+	pte_size = PAGE_SIZE;
+	pgflags = _PAGE_READ | _PAGE_EXEC;
+	level = 0;
+	npages = get_user_pages_fast(hva, 1, writing, pages);
+	if (npages < 1) {
+		/* Check if it's an I/O mapping */
+		down_read(&current->mm->mmap_sem);
+		vma = find_vma(current->mm, hva);
+		if (vma && vma->vm_start <= hva && hva < vma->vm_end &&
+		    (vma->vm_flags & VM_PFNMAP)) {
+			pfn = vma->vm_pgoff +
+				((hva - vma->vm_start) >> PAGE_SHIFT);
+			pgflags = pgprot_val(vma->vm_page_prot);
+		}
+		up_read(&current->mm->mmap_sem);
+		if (!pfn)
+			return -EFAULT;
+	} else {
+		page = pages[0];
+		pfn = page_to_pfn(page);
+		if (PageHuge(page)) {
+			page = compound_head(page);
+			pte_size <<= compound_order(page);
+			/* See if we can insert a 2MB large-page PTE here */
+			if (pte_size >= PMD_SIZE &&
+			    (gpa & PMD_MASK & PAGE_MASK) ==
+			    (hva & PMD_MASK & PAGE_MASK)) {
+				level = 1;
+				pfn &= ~((PMD_SIZE >> PAGE_SHIFT) - 1);
+			}
+		}
+		/* See if we can provide write access */
+		if (writing) {
+			/*
+			 * We assume gup_fast has set dirty on the host PTE.
+			 */
+			pgflags |= _PAGE_WRITE;
+		} else {
+			local_irq_save(flags);
+			ptep = __find_linux_pte_or_hugepte(current->mm->pgd,
+							hva, NULL, NULL);
+			if (ptep && pte_write(*ptep) && pte_dirty(*ptep))
+				pgflags |= _PAGE_WRITE;
+			local_irq_restore(flags);
+		}
+	}
+
+	/*
+	 * Compute the PTE value that we need to insert.
+	 */
+	pgflags |= _PAGE_PRESENT | _PAGE_PTE | _PAGE_ACCESSED;
+	if (pgflags & _PAGE_WRITE)
+		pgflags |= _PAGE_DIRTY;
+	pte = pfn_pte(pfn, __pgprot(pgflags));
+
+	/* Allocate space in the tree and write the PTE */
+	ret = kvmppc_create_pte(kvm, pte, gpa, level, mmu_seq);
+	if (ret == -EBUSY) {
+		/*
+		 * There's already a PMD where wanted to install a large page;
+		 * for now, fall back to installing a small page.
+		 */
+		level = 0;
+		pfn |= gfn & ((PMD_SIZE >> PAGE_SHIFT) - 1);
+		pte = pfn_pte(pfn, __pgprot(pgflags));
+		ret = kvmppc_create_pte(kvm, pte, gpa, level, mmu_seq);
+	}
+	if (ret == 0 || ret == -EAGAIN)
+		ret = RESUME_GUEST;
+
+	if (page) {
+		/*
+		 * We drop pages[0] here, not page because page might
+		 * have been set to the head page of a compound, but
+		 * we have to drop the reference on the correct tail
+		 * page to match the get inside gup()
+		 */
+		put_page(pages[0]);
+	}
+	return ret;
+}
+
+static void mark_pages_dirty(struct kvm *kvm, struct kvm_memory_slot *memslot,
+			     unsigned long gfn, unsigned int order)
+{
+	unsigned long i, limit;
+	unsigned long *dp;
+
+	if (!memslot->dirty_bitmap)
+		return;
+	limit = 1ul << order;
+	if (limit < BITS_PER_LONG) {
+		for (i = 0; i < limit; ++i)
+			mark_page_dirty(kvm, gfn + i);
+		return;
+	}
+	dp = memslot->dirty_bitmap + (gfn - memslot->base_gfn);
+	limit /= BITS_PER_LONG;
+	for (i = 0; i < limit; ++i)
+		*dp++ = ~0ul;
+}
+
+/* Called with kvm->lock held */
+int kvm_unmap_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
+		    unsigned long gfn)
+{
+	pte_t *ptep;
+	unsigned long gpa = gfn << PAGE_SHIFT;
+	unsigned int shift;
+	unsigned long old;
+
+	ptep = __find_linux_pte_or_hugepte(kvm->arch.pgtable, gpa,
+					   NULL, &shift);
+	if (ptep && pte_present(*ptep)) {
+		old = kvmppc_radix_update_pte(kvm, ptep, _PAGE_PRESENT, 0,
+					      gpa, shift);
+		kvmppc_radix_tlbie_page(kvm, gpa, shift);
+		if (old & _PAGE_DIRTY) {
+			if (!shift)
+				mark_page_dirty(kvm, gfn);
+			else
+				mark_pages_dirty(kvm, memslot,
+						 gfn, shift - PAGE_SHIFT);
+		}
+	}
+	return 0;				
+}
+
+/* Called with kvm->lock held */
+int kvm_age_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
+		  unsigned long gfn)
+{
+	pte_t *ptep;
+	unsigned long gpa = gfn << PAGE_SHIFT;
+	unsigned int shift;
+	int ref = 0;
+
+	ptep = __find_linux_pte_or_hugepte(kvm->arch.pgtable, gpa,
+					   NULL, &shift);
+	if (ptep && pte_present(*ptep) && pte_young(*ptep)) {
+		kvmppc_radix_update_pte(kvm, ptep, _PAGE_ACCESSED, 0,
+					gpa, shift);
+		/* XXX need to flush tlb here? */
+		ref = 1;
+	}
+	return ref;
+}
+
+/* Called with kvm->lock held */
+int kvm_test_age_radix(struct kvm *kvm, struct kvm_memory_slot *memslot,
+		       unsigned long gfn)
+{
+	pte_t *ptep;
+	unsigned long gpa = gfn << PAGE_SHIFT;
+	unsigned int shift;
+	int ref = 0;
+
+	ptep = __find_linux_pte_or_hugepte(kvm->arch.pgtable, gpa,
+					   NULL, &shift);
+	if (ptep && pte_present(*ptep) && pte_young(*ptep))
+		ref = 1;
+	return ref;
+}
+
+/* Returns the number of PAGE_SIZE pages that are dirty */
+static int kvm_radix_test_clear_dirty(struct kvm *kvm,
+				struct kvm_memory_slot *memslot, int pagenum)
+{
+	unsigned long gfn = memslot->base_gfn + pagenum;
+	unsigned long gpa = gfn << PAGE_SHIFT;
+	pte_t *ptep;
+	unsigned int shift;
+	int ret = 0;
+
+	ptep = __find_linux_pte_or_hugepte(kvm->arch.pgtable, gpa,
+					   NULL, &shift);
+	if (ptep && pte_present(*ptep) && pte_dirty(*ptep)) {
+		ret = 1;
+		if (shift)
+			ret = 1 << (shift - PAGE_SHIFT);
+		kvmppc_radix_update_pte(kvm, ptep, _PAGE_DIRTY, 0,
+					gpa, shift);
+		kvmppc_radix_tlbie_page(kvm, gpa, shift);
+	}
+	return ret;
+}
+
+long kvmppc_hv_get_dirty_log_radix(struct kvm *kvm,
+			struct kvm_memory_slot *memslot, unsigned long *map)
+{
+	unsigned long i, j;
+	unsigned long n, *p;
+	int npages;
+
+	/*
+	 * Radix accumulates dirty bits in the first half of the
+	 * memslot's dirty_bitmap area, for when pages are paged
+	 * out or modified by the host directly.  Pick up these
+	 * bits and add them to the map.
+	 */
+	n = kvm_dirty_bitmap_bytes(memslot) / sizeof(long);
+	p = memslot->dirty_bitmap;
+	for (i = 0; i < n; ++i)
+		map[i] |= xchg(&p[i], 0);
+
+	for (i = 0; i < memslot->npages; i = j) {
+		npages = kvm_radix_test_clear_dirty(kvm, memslot, i);
+
+		/*
+		 * Note that if npages > 0 then i must be a multiple of npages,
+		 * since huge pages are only used to back the guest at guest
+		 * real addresses that are a multiple of their size.
+		 * Since we have at most one PTE covering any given guest
+		 * real address, if npages > 1 we can skip to i + npages.
+		 */
+		j = i + 1;
+		if (npages)
+			for (j = i; npages; ++j, --npages)
+				__set_bit_le(j, map);
+	}
+	return 0;
+}
+
+static void add_rmmu_ap_encoding(struct kvm_ppc_rmmu_info *info,
+				 int psize, int *indexp)
+{
+	if (!mmu_psize_defs[psize].shift)
+		return;
+	info->ap_encodings[*indexp] = mmu_psize_defs[psize].shift |
+		(mmu_psize_defs[psize].ap << 29);
+	++(*indexp);
+}
+
+int kvmhv_get_rmmu_info(struct kvm *kvm, struct kvm_ppc_rmmu_info *info)
+{
+	int i;
+
+	if (!radix_enabled())
+		return -EINVAL;
+	memset(info, 0, sizeof(*info));
+
+	/* 4k page size */
+	info->geometries[0].page_shift = 12;
+	info->geometries[0].level_bits[0] = 9;
+	for (i = 1; i < 4; ++i)
+		info->geometries[0].level_bits[i] = p9_supported_radix_bits[i];
+	/* 64k page size */
+	info->geometries[1].page_shift = 16;
+	for (i = 0; i < 4; ++i)
+		info->geometries[1].level_bits[i] = p9_supported_radix_bits[i];
+
+	i = 0;
+	add_rmmu_ap_encoding(info, MMU_PAGE_4K, &i);
+	add_rmmu_ap_encoding(info, MMU_PAGE_64K, &i);
+	add_rmmu_ap_encoding(info, MMU_PAGE_2M, &i);
+	add_rmmu_ap_encoding(info, MMU_PAGE_1G, &i);
+
+	return 0;
+}
+
+int kvmppc_init_vm_radix(struct kvm *kvm)
+{
+	kvm->arch.pgtable = pgd_alloc(kvm->mm);
+	if (!kvm->arch.pgtable)
+		return -ENOMEM;
+	return 0;
+}
+
+void kvmppc_free_radix(struct kvm *kvm)
+{
+	unsigned long ig, iu, im;
+	pte_t *pte;
+	pmd_t *pmd;
+	pud_t *pud;
+	pgd_t *pgd;
+
+	if (!kvm->arch.pgtable)
+		return;
+	pgd = kvm->arch.pgtable;
+	for (ig = 0; ig < PTRS_PER_PGD; ++ig, ++pgd) {
+		if (!pgd_present(*pgd))
+			continue;
+		pud = pud_offset(pgd, 0);
+		for (iu = 0; iu < PTRS_PER_PUD; ++iu, ++pud) {
+			if (!pud_present(*pud))
+				continue;
+			pmd = pmd_offset(pud, 0);
+			for (im = 0; im < PTRS_PER_PMD; ++im, ++pmd) {
+				if (pmd_huge(*pmd)) {
+					pmd_clear(pmd);
+					continue;
+				}
+				if (!pmd_present(*pmd))
+					continue;
+				pte = pte_offset_map(pmd, 0);
+				memset(pte, 0, sizeof(long) << PTE_INDEX_SIZE);
+				kvmppc_pte_free(pte);
+				pmd_clear(pmd);
+			}
+			pmd_free(kvm->mm, pmd_offset(pud, 0));
+			pud_clear(pud);
+		}
+		pud_free(kvm->mm, pud_offset(pgd, 0));
+		pgd_clear(pgd);
+	}
+	pgd_free(kvm->mm, kvm->arch.pgtable);
+}
+
+static void pte_ctor(void *addr)
+{
+	memset(addr, 0, PTE_TABLE_SIZE);
+}
+
+int kvmppc_radix_init(void)
+{
+	unsigned long size = sizeof(void *) << PTE_INDEX_SIZE;
+
+	kvm_pte_cache = kmem_cache_create("kvm-pte", size, size, 0, pte_ctor);
+	if (!kvm_pte_cache)
+		return -ENOMEM;
+	return 0;
+}
+
+void kvmppc_radix_exit(void)
+{
+	kmem_cache_destroy(kvm_pte_cache);
+}

arch/powerpc/kvm/book3s_hv.c

@@ -182,7 +182,8 @@ static void kvmppc_fast_vcpu_kick_hv(struct kvm_vcpu *vcpu)
 		++vcpu->stat.halt_wakeup;
 	}
 
-	if (kvmppc_ipi_thread(vcpu->arch.thread_cpu))
+	cpu = READ_ONCE(vcpu->arch.thread_cpu);
+	if (cpu >= 0 && kvmppc_ipi_thread(cpu))
 		return;
 
 	/* CPU points to the first thread of the core */
@@ -773,12 +774,8 @@ int kvmppc_pseries_do_hcall(struct kvm_vcpu *vcpu)
 		}
 		tvcpu->arch.prodded = 1;
 		smp_mb();
-		if (vcpu->arch.ceded) {
-			if (swait_active(&vcpu->wq)) {
-				swake_up(&vcpu->wq);
-				vcpu->stat.halt_wakeup++;
-			}
-		}
+		if (tvcpu->arch.ceded)
+			kvmppc_fast_vcpu_kick_hv(tvcpu);
 		break;
 	case H_CONFER:
 		target = kvmppc_get_gpr(vcpu, 4);
@@ -1135,7 +1132,7 @@ static void kvmppc_set_lpcr(struct kvm_vcpu *vcpu, u64 new_lpcr,
 	/*
 	 * Userspace can only modify DPFD (default prefetch depth),
 	 * ILE (interrupt little-endian) and TC (translation control).
-	 * On POWER8 userspace can also modify AIL (alt. interrupt loc.)
+	 * On POWER8 and POWER9 userspace can also modify AIL (alt. interrupt loc.).
 	 */
 	mask = LPCR_DPFD | LPCR_ILE | LPCR_TC;
 	if (cpu_has_feature(CPU_FTR_ARCH_207S))
@@ -1821,6 +1818,7 @@ static struct kvm_vcpu *kvmppc_core_vcpu_create_hv(struct kvm *kvm,
 	vcpu->arch.vcore = vcore;
 	vcpu->arch.ptid = vcpu->vcpu_id - vcore->first_vcpuid;
 	vcpu->arch.thread_cpu = -1;
+	vcpu->arch.prev_cpu = -1;
 
 	vcpu->arch.cpu_type = KVM_CPU_3S_64;
 	kvmppc_sanity_check(vcpu);
@@ -1950,11 +1948,33 @@ static void kvmppc_release_hwthread(int cpu)
 	tpaca->kvm_hstate.kvm_split_mode = NULL;
 }
 
+static void do_nothing(void *x)
+{
+}
+
+static void radix_flush_cpu(struct kvm *kvm, int cpu, struct kvm_vcpu *vcpu)
+{
+	int i;
+
+	cpu = cpu_first_thread_sibling(cpu);
+	cpumask_set_cpu(cpu, &kvm->arch.need_tlb_flush);
+	/*
+	 * Make sure setting of bit in need_tlb_flush precedes
+	 * testing of cpu_in_guest bits.  The matching barrier on
+	 * the other side is the first smp_mb() in kvmppc_run_core().
+	 */
+	smp_mb();
+	for (i = 0; i < threads_per_core; ++i)
+		if (cpumask_test_cpu(cpu + i, &kvm->arch.cpu_in_guest))
+			smp_call_function_single(cpu + i, do_nothing, NULL, 1);
+}
+
 static void kvmppc_start_thread(struct kvm_vcpu *vcpu, struct kvmppc_vcore *vc)
 {
 	int cpu;
 	struct paca_struct *tpaca;
 	struct kvmppc_vcore *mvc = vc->master_vcore;
+	struct kvm *kvm = vc->kvm;
 
 	cpu = vc->pcpu;
 	if (vcpu) {
@@ -1965,6 +1985,27 @@ static void kvmppc_start_thread(struct kvm_vcpu *vcpu, struct kvmppc_vcore *vc)
 		cpu += vcpu->arch.ptid;
 		vcpu->cpu = mvc->pcpu;
 		vcpu->arch.thread_cpu = cpu;
+
+		/*
+		 * With radix, the guest can do TLB invalidations itself,
+		 * and it could choose to use the local form (tlbiel) if
+		 * it is invalidating a translation that has only ever been
+		 * used on one vcpu.  However, that doesn't mean it has
+		 * only ever been used on one physical cpu, since vcpus
+		 * can move around between pcpus.  To cope with this, when
+		 * a vcpu moves from one pcpu to another, we need to tell
+		 * any vcpus running on the same core as this vcpu previously
+		 * ran to flush the TLB.  The TLB is shared between threads,
+		 * so we use a single bit in .need_tlb_flush for all 4 threads.
+		 */
+		if (kvm_is_radix(kvm) && vcpu->arch.prev_cpu != cpu) {
+			if (vcpu->arch.prev_cpu >= 0 &&
+			    cpu_first_thread_sibling(vcpu->arch.prev_cpu) !=
+			    cpu_first_thread_sibling(cpu))
+				radix_flush_cpu(kvm, vcpu->arch.prev_cpu, vcpu);
+			vcpu->arch.prev_cpu = cpu;
+		}
+		cpumask_set_cpu(cpu, &kvm->arch.cpu_in_guest);
 	}
 	tpaca = &paca[cpu];
 	tpaca->kvm_hstate.kvm_vcpu = vcpu;
@@ -2552,6 +2593,7 @@ static noinline void kvmppc_run_core(struct kvmppc_vcore *vc)
 		kvmppc_release_hwthread(pcpu + i);
 		if (sip && sip->napped[i])
 			kvmppc_ipi_thread(pcpu + i);
+		cpumask_clear_cpu(pcpu + i, &vc->kvm->arch.cpu_in_guest);
 	}
 
 	kvmppc_set_host_core(pcpu);
@@ -2620,7 +2662,8 @@ static int kvmppc_vcore_check_block(struct kvmppc_vcore *vc)
 	int i;
 
 	for_each_runnable_thread(i, vcpu, vc) {
-		if (vcpu->arch.pending_exceptions || !vcpu->arch.ceded)
+		if (vcpu->arch.pending_exceptions || !vcpu->arch.ceded ||
+		    vcpu->arch.prodded)
 			return 1;
 	}
 
@@ -2806,7 +2849,7 @@ static int kvmppc_run_vcpu(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
 			break;
 		n_ceded = 0;
 		for_each_runnable_thread(i, v, vc) {
-			if (!v->arch.pending_exceptions)
+			if (!v->arch.pending_exceptions && !v->arch.prodded)
 				n_ceded += v->arch.ceded;
 			else
 				v->arch.ceded = 0;
@@ -2877,7 +2920,7 @@ static int kvmppc_vcpu_run_hv(struct kvm_run *run, struct kvm_vcpu *vcpu)
 	smp_mb();
 
 	/* On the first time here, set up HTAB and VRMA */
-	if (!vcpu->kvm->arch.hpte_setup_done) {
+	if (!kvm_is_radix(vcpu->kvm) && !vcpu->kvm->arch.hpte_setup_done) {
 		r = kvmppc_hv_setup_htab_rma(vcpu);
 		if (r)
 			goto out;
@@ -2939,6 +2982,13 @@ static int kvm_vm_ioctl_get_smmu_info_hv(struct kvm *kvm,
 {
 	struct kvm_ppc_one_seg_page_size *sps;
 
+	/*
+	 * Since we don't yet support HPT guests on a radix host,
+	 * return an error if the host uses radix.
+	 */
+	if (radix_enabled())
+		return -EINVAL;
+
 	info->flags = KVM_PPC_PAGE_SIZES_REAL;
 	if (mmu_has_feature(MMU_FTR_1T_SEGMENT))
 		info->flags |= KVM_PPC_1T_SEGMENTS;
@@ -2961,8 +3011,10 @@ static int kvm_vm_ioctl_get_dirty_log_hv(struct kvm *kvm,
 {
 	struct kvm_memslots *slots;
 	struct kvm_memory_slot *memslot;
-	int r;
+	int i, r;
 	unsigned long n;
+	unsigned long *buf;
+	struct kvm_vcpu *vcpu;
 
 	mutex_lock(&kvm->slots_lock);
 
@@ -2976,15 +3028,32 @@ static int kvm_vm_ioctl_get_dirty_log_hv(struct kvm *kvm,
 	if (!memslot->dirty_bitmap)
 		goto out;
 
+	/*
+	 * Use second half of bitmap area because radix accumulates
+	 * bits in the first half.
+	 */
 	n = kvm_dirty_bitmap_bytes(memslot);
-	memset(memslot->dirty_bitmap, 0, n);
+	buf = memslot->dirty_bitmap + n / sizeof(long);
+	memset(buf, 0, n);
 
-	r = kvmppc_hv_get_dirty_log(kvm, memslot, memslot->dirty_bitmap);
+	if (kvm_is_radix(kvm))
+		r = kvmppc_hv_get_dirty_log_radix(kvm, memslot, buf);
+	else
+		r = kvmppc_hv_get_dirty_log_hpt(kvm, memslot, buf);
 	if (r)
 		goto out;
 
+	/* Harvest dirty bits from VPA and DTL updates */
+	/* Note: we never modify the SLB shadow buffer areas */
+	kvm_for_each_vcpu(i, vcpu, kvm) {
+		spin_lock(&vcpu->arch.vpa_update_lock);
+		kvmppc_harvest_vpa_dirty(&vcpu->arch.vpa, memslot, buf);
+		kvmppc_harvest_vpa_dirty(&vcpu->arch.dtl, memslot, buf);
+		spin_unlock(&vcpu->arch.vpa_update_lock);
+	}
+
 	r = -EFAULT;
-	if (copy_to_user(log->dirty_bitmap, memslot->dirty_bitmap, n))
+	if (copy_to_user(log->dirty_bitmap, buf, n))
 		goto out;
 
 	r = 0;
@@ -3005,6 +3074,15 @@ static void kvmppc_core_free_memslot_hv(struct kvm_memory_slot *free,
 static int kvmppc_core_create_memslot_hv(struct kvm_memory_slot *slot,
 					 unsigned long npages)
 {
+	/*
+	 * For now, if radix_enabled() then we only support radix guests,
+	 * and in that case we don't need the rmap array.
+	 */
+	if (radix_enabled()) {
+		slot->arch.rmap = NULL;
+		return 0;
+	}
+
 	slot->arch.rmap = vzalloc(npages * sizeof(*slot->arch.rmap));
 	if (!slot->arch.rmap)
 		return -ENOMEM;
@@ -3037,7 +3115,7 @@ static void kvmppc_core_commit_memory_region_hv(struct kvm *kvm,
 	if (npages)
 		atomic64_inc(&kvm->arch.mmio_update);
 
-	if (npages && old->npages) {
+	if (npages && old->npages && !kvm_is_radix(kvm)) {
 		/*
 		 * If modifying a memslot, reset all the rmap dirty bits.
 		 * If this is a new memslot, we don't need to do anything
@@ -3046,7 +3124,7 @@ static void kvmppc_core_commit_memory_region_hv(struct kvm *kvm,
 		 */
 		slots = kvm_memslots(kvm);
 		memslot = id_to_memslot(slots, mem->slot);
-		kvmppc_hv_get_dirty_log(kvm, memslot, NULL);
+		kvmppc_hv_get_dirty_log_hpt(kvm, memslot, NULL);
 	}
 }
 
@@ -3085,14 +3163,20 @@ static void kvmppc_setup_partition_table(struct kvm *kvm)
 {
 	unsigned long dw0, dw1;
 
-	/* PS field - page size for VRMA */
-	dw0 = ((kvm->arch.vrma_slb_v & SLB_VSID_L) >> 1) |
-		((kvm->arch.vrma_slb_v & SLB_VSID_LP) << 1);
-	/* HTABSIZE and HTABORG fields */
-	dw0 |= kvm->arch.sdr1;
+	if (!kvm_is_radix(kvm)) {
+		/* PS field - page size for VRMA */
+		dw0 = ((kvm->arch.vrma_slb_v & SLB_VSID_L) >> 1) |
+			((kvm->arch.vrma_slb_v & SLB_VSID_LP) << 1);
+		/* HTABSIZE and HTABORG fields */
+		dw0 |= kvm->arch.sdr1;
 
-	/* Second dword has GR=0; other fields are unused since UPRT=0 */
-	dw1 = 0;
+		/* Second dword as set by userspace */
+		dw1 = kvm->arch.process_table;
+	} else {
+		dw0 = PATB_HR | radix__get_tree_size() |
+			__pa(kvm->arch.pgtable) | RADIX_PGD_INDEX_SIZE;
+		dw1 = PATB_GR | kvm->arch.process_table;
+	}
 
 	mmu_partition_table_set_entry(kvm->arch.lpid, dw0, dw1);
 }
@@ -3113,12 +3197,23 @@ static int kvmppc_hv_setup_htab_rma(struct kvm_vcpu *vcpu)
 		goto out;	/* another vcpu beat us to it */
 
 	/* Allocate hashed page table (if not done already) and reset it */
-	if (!kvm->arch.hpt_virt) {
-		err = kvmppc_alloc_hpt(kvm, NULL);
-		if (err) {
+	if (!kvm->arch.hpt.virt) {
+		int order = KVM_DEFAULT_HPT_ORDER;
+		struct kvm_hpt_info info;
+
+		err = kvmppc_allocate_hpt(&info, order);
+		/* If we get here, it means userspace didn't specify a
+		 * size explicitly.  So, try successively smaller
+		 * sizes if the default failed. */
+		while ((err == -ENOMEM) && --order >= PPC_MIN_HPT_ORDER)
+			err  = kvmppc_allocate_hpt(&info, order);
+
+		if (err < 0) {
 			pr_err("KVM: Couldn't alloc HPT\n");
 			goto out;
 		}
+
+		kvmppc_set_hpt(kvm, &info);
 	}
 
 	/* Look up the memslot for guest physical address 0 */
@@ -3262,6 +3357,7 @@ static int kvmppc_core_init_vm_hv(struct kvm *kvm)
 {
 	unsigned long lpcr, lpid;
 	char buf[32];
+	int ret;
 
 	/* Allocate the guest's logical partition ID */
 
@@ -3309,13 +3405,33 @@ static int kvmppc_core_init_vm_hv(struct kvm *kvm)
 		lpcr |= LPCR_HVICE;
 	}
 
+	/*
+	 * For now, if the host uses radix, the guest must be radix.
+	 */
+	if (radix_enabled()) {
+		kvm->arch.radix = 1;
+		lpcr &= ~LPCR_VPM1;
+		lpcr |= LPCR_UPRT | LPCR_GTSE | LPCR_HR;
+		ret = kvmppc_init_vm_radix(kvm);
+		if (ret) {
+			kvmppc_free_lpid(kvm->arch.lpid);
+			return ret;
+		}
+		kvmppc_setup_partition_table(kvm);
+	}
+
 	kvm->arch.lpcr = lpcr;
 
+	/* Initialization for future HPT resizes */
+	kvm->arch.resize_hpt = NULL;
+
 	/*
 	 * Work out how many sets the TLB has, for the use of
 	 * the TLB invalidation loop in book3s_hv_rmhandlers.S.
 	 */
-	if (cpu_has_feature(CPU_FTR_ARCH_300))
+	if (kvm_is_radix(kvm))
+		kvm->arch.tlb_sets = POWER9_TLB_SETS_RADIX;	/* 128 */
+	else if (cpu_has_feature(CPU_FTR_ARCH_300))
 		kvm->arch.tlb_sets = POWER9_TLB_SETS_HASH;	/* 256 */
 	else if (cpu_has_feature(CPU_FTR_ARCH_207S))
 		kvm->arch.tlb_sets = POWER8_TLB_SETS;		/* 512 */
@@ -3325,8 +3441,11 @@ static int kvmppc_core_init_vm_hv(struct kvm *kvm)
 	/*
 	 * Track that we now have a HV mode VM active. This blocks secondary
 	 * CPU threads from coming online.
+	 * On POWER9, we only need to do this for HPT guests on a radix
+	 * host, which is not yet supported.
 	 */
-	kvm_hv_vm_activated();
+	if (!cpu_has_feature(CPU_FTR_ARCH_300))
+		kvm_hv_vm_activated();
 
 	/*
 	 * Create a debugfs directory for the VM
@@ -3352,11 +3471,17 @@ static void kvmppc_core_destroy_vm_hv(struct kvm *kvm)
 {
 	debugfs_remove_recursive(kvm->arch.debugfs_dir);
 
-	kvm_hv_vm_deactivated();
+	if (!cpu_has_feature(CPU_FTR_ARCH_300))
+		kvm_hv_vm_deactivated();
 
 	kvmppc_free_vcores(kvm);
 
-	kvmppc_free_hpt(kvm);
+	kvmppc_free_lpid(kvm->arch.lpid);
+
+	if (kvm_is_radix(kvm))
+		kvmppc_free_radix(kvm);
+	else
+		kvmppc_free_hpt(&kvm->arch.hpt);
 
 	kvmppc_free_pimap(kvm);
 }
@@ -3385,11 +3510,6 @@ static int kvmppc_core_check_processor_compat_hv(void)
 	if (!cpu_has_feature(CPU_FTR_HVMODE) ||
 	    !cpu_has_feature(CPU_FTR_ARCH_206))
 		return -EIO;
-	/*
-	 * Disable KVM for Power9 in radix mode.
-	 */
-	if (cpu_has_feature(CPU_FTR_ARCH_300) && radix_enabled())
-		return -EIO;
 
 	return 0;
 }
@@ -3587,12 +3707,9 @@ static long kvm_arch_vm_ioctl_hv(struct file *filp,
 		r = -EFAULT;
 		if (get_user(htab_order, (u32 __user *)argp))
 			break;
-		r = kvmppc_alloc_reset_hpt(kvm, &htab_order);
+		r = kvmppc_alloc_reset_hpt(kvm, htab_order);
 		if (r)
 			break;
-		r = -EFAULT;
-		if (put_user(htab_order, (u32 __user *)argp))
-			break;
 		r = 0;
 		break;
 	}
@@ -3607,6 +3724,28 @@ static long kvm_arch_vm_ioctl_hv(struct file *filp,
 		break;
 	}
 
+	case KVM_PPC_RESIZE_HPT_PREPARE: {
+		struct kvm_ppc_resize_hpt rhpt;
+
+		r = -EFAULT;
+		if (copy_from_user(&rhpt, argp, sizeof(rhpt)))
+			break;
+
+		r = kvm_vm_ioctl_resize_hpt_prepare(kvm, &rhpt);
+		break;
+	}
+
+	case KVM_PPC_RESIZE_HPT_COMMIT: {
+		struct kvm_ppc_resize_hpt rhpt;
+
+		r = -EFAULT;
+		if (copy_from_user(&rhpt, argp, sizeof(rhpt)))
+			break;
+
+		r = kvm_vm_ioctl_resize_hpt_commit(kvm, &rhpt);
+		break;
+	}
+
 	default:
 		r = -ENOTTY;
 	}
@@ -3657,6 +3796,41 @@ static void init_default_hcalls(void)
 	}
 }
 
+static int kvmhv_configure_mmu(struct kvm *kvm, struct kvm_ppc_mmuv3_cfg *cfg)
+{
+	unsigned long lpcr;
+	int radix;
+
+	/* If not on a POWER9, reject it */
+	if (!cpu_has_feature(CPU_FTR_ARCH_300))
+		return -ENODEV;
+
+	/* If any unknown flags set, reject it */
+	if (cfg->flags & ~(KVM_PPC_MMUV3_RADIX | KVM_PPC_MMUV3_GTSE))
+		return -EINVAL;
+
+	/* We can't change a guest to/from radix yet */
+	radix = !!(cfg->flags & KVM_PPC_MMUV3_RADIX);
+	if (radix != kvm_is_radix(kvm))
+		return -EINVAL;
+
+	/* GR (guest radix) bit in process_table field must match */
+	if (!!(cfg->process_table & PATB_GR) != radix)
+		return -EINVAL;
+
+	/* Process table size field must be reasonable, i.e. <= 24 */
+	if ((cfg->process_table & PRTS_MASK) > 24)
+		return -EINVAL;
+
+	kvm->arch.process_table = cfg->process_table;
+	kvmppc_setup_partition_table(kvm);
+
+	lpcr = (cfg->flags & KVM_PPC_MMUV3_GTSE) ? LPCR_GTSE : 0;
+	kvmppc_update_lpcr(kvm, lpcr, LPCR_GTSE);
+
+	return 0;
+}
+
 static struct kvmppc_ops kvm_ops_hv = {
 	.get_sregs = kvm_arch_vcpu_ioctl_get_sregs_hv,
 	.set_sregs = kvm_arch_vcpu_ioctl_set_sregs_hv,
@@ -3694,6 +3868,8 @@ static struct kvmppc_ops kvm_ops_hv = {
 	.irq_bypass_add_producer = kvmppc_irq_bypass_add_producer_hv,
 	.irq_bypass_del_producer = kvmppc_irq_bypass_del_producer_hv,
 #endif
+	.configure_mmu = kvmhv_configure_mmu,
+	.get_rmmu_info = kvmhv_get_rmmu_info,
 };
 
 static int kvm_init_subcore_bitmap(void)
@@ -3728,6 +3904,11 @@ static int kvm_init_subcore_bitmap(void)
 	return 0;
 }
 
+static int kvmppc_radix_possible(void)
+{
+	return cpu_has_feature(CPU_FTR_ARCH_300) && radix_enabled();
+}
+
 static int kvmppc_book3s_init_hv(void)
 {
 	int r;
@@ -3767,12 +3948,19 @@ static int kvmppc_book3s_init_hv(void)
 	init_vcore_lists();
 
 	r = kvmppc_mmu_hv_init();
+	if (r)
+		return r;
+
+	if (kvmppc_radix_possible())
+		r = kvmppc_radix_init();
 	return r;
 }
 
 static void kvmppc_book3s_exit_hv(void)
 {
 	kvmppc_free_host_rm_ops();
+	if (kvmppc_radix_possible())
+		kvmppc_radix_exit();
 	kvmppc_hv_ops = NULL;
 }
 

arch/powerpc/kvm/book3s_hv_builtin.c

@@ -29,6 +29,11 @@
 #include <asm/opal.h>
 #include <asm/smp.h>
 
+static bool in_realmode(void)
+{
+	return !(mfmsr() & MSR_IR);
+}
+
 #define KVM_CMA_CHUNK_ORDER	18
 
 /*
@@ -52,19 +57,19 @@ static int __init early_parse_kvm_cma_resv(char *p)
 }
 early_param("kvm_cma_resv_ratio", early_parse_kvm_cma_resv);
 
-struct page *kvm_alloc_hpt(unsigned long nr_pages)
+struct page *kvm_alloc_hpt_cma(unsigned long nr_pages)
 {
 	VM_BUG_ON(order_base_2(nr_pages) < KVM_CMA_CHUNK_ORDER - PAGE_SHIFT);
 
 	return cma_alloc(kvm_cma, nr_pages, order_base_2(HPT_ALIGN_PAGES));
 }
-EXPORT_SYMBOL_GPL(kvm_alloc_hpt);
+EXPORT_SYMBOL_GPL(kvm_alloc_hpt_cma);
 
-void kvm_release_hpt(struct page *page, unsigned long nr_pages)
+void kvm_free_hpt_cma(struct page *page, unsigned long nr_pages)
 {
 	cma_release(kvm_cma, page, nr_pages);
 }
-EXPORT_SYMBOL_GPL(kvm_release_hpt);
+EXPORT_SYMBOL_GPL(kvm_free_hpt_cma);
 
 /**
  * kvm_cma_reserve() - reserve area for kvm hash pagetable
@@ -200,7 +205,6 @@ static inline void rm_writeb(unsigned long paddr, u8 val)
 
 /*
  * Send an interrupt or message to another CPU.
- * This can only be called in real mode.
  * The caller needs to include any barrier needed to order writes
  * to memory vs. the IPI/message.
  */
@@ -226,7 +230,9 @@ void kvmhv_rm_send_ipi(int cpu)
 
 	/* Else poke the target with an IPI */
 	xics_phys = paca[cpu].kvm_hstate.xics_phys;
-	if (xics_phys)
+	if (!in_realmode())
+		opal_int_set_mfrr(get_hard_smp_processor_id(cpu), IPI_PRIORITY);
+	else if (xics_phys)
 		rm_writeb(xics_phys + XICS_MFRR, IPI_PRIORITY);
 	else
 		opal_rm_int_set_mfrr(get_hard_smp_processor_id(cpu),
@@ -412,14 +418,15 @@ static long kvmppc_read_one_intr(bool *again)
 
 	/* Now read the interrupt from the ICP */
 	xics_phys = local_paca->kvm_hstate.xics_phys;
-	if (!xics_phys) {
-		/* Use OPAL to read the XIRR */
+	rc = 0;
+	if (!in_realmode())
+		rc = opal_int_get_xirr(&xirr, false);
+	else if (!xics_phys)
 		rc = opal_rm_int_get_xirr(&xirr, false);
-		if (rc < 0)
-			return 1;
-	} else {
+	else
 		xirr = _lwzcix(xics_phys + XICS_XIRR);
-	}
+	if (rc < 0)
+		return 1;
 
 	/*
 	 * Save XIRR for later. Since we get control in reverse endian
@@ -445,15 +452,19 @@ static long kvmppc_read_one_intr(bool *again)
 	 * If it is an IPI, clear the MFRR and EOI it.
 	 */
 	if (xisr == XICS_IPI) {
-		if (xics_phys) {
+		rc = 0;
+		if (!in_realmode()) {
+			opal_int_set_mfrr(hard_smp_processor_id(), 0xff);
+			rc = opal_int_eoi(h_xirr);
+		} else if (xics_phys) {
 			_stbcix(xics_phys + XICS_MFRR, 0xff);
 			_stwcix(xics_phys + XICS_XIRR, xirr);
 		} else {
 			opal_rm_int_set_mfrr(hard_smp_processor_id(), 0xff);
 			rc = opal_rm_int_eoi(h_xirr);
-			/* If rc > 0, there is another interrupt pending */
-			*again = rc > 0;
 		}
+		/* If rc > 0, there is another interrupt pending */
+		*again = rc > 0;
 
 		/*
 		 * Need to ensure side effects of above stores
@@ -471,7 +482,10 @@ static long kvmppc_read_one_intr(bool *again)
 			/* We raced with the host,
 			 * we need to resend that IPI, bummer
 			 */
-			if (xics_phys)
+			if (!in_realmode())
+				opal_int_set_mfrr(hard_smp_processor_id(),
+						  IPI_PRIORITY);
+			else if (xics_phys)
 				_stbcix(xics_phys + XICS_MFRR, IPI_PRIORITY);
 			else
 				opal_rm_int_set_mfrr(hard_smp_processor_id(),

arch/powerpc/kvm/book3s_hv_rm_mmu.c

@@ -43,6 +43,7 @@ static void *real_vmalloc_addr(void *x)
 static int global_invalidates(struct kvm *kvm, unsigned long flags)
 {
 	int global;
+	int cpu;
 
 	/*
 	 * If there is only one vcore, and it's currently running,
@@ -60,8 +61,14 @@ static int global_invalidates(struct kvm *kvm, unsigned long flags)
 		/* any other core might now have stale TLB entries... */
 		smp_wmb();
 		cpumask_setall(&kvm->arch.need_tlb_flush);
-		cpumask_clear_cpu(local_paca->kvm_hstate.kvm_vcore->pcpu,
-				  &kvm->arch.need_tlb_flush);
+		cpu = local_paca->kvm_hstate.kvm_vcore->pcpu;
+		/*
+		 * On POWER9, threads are independent but the TLB is shared,
+		 * so use the bit for the first thread to represent the core.
+		 */
+		if (cpu_has_feature(CPU_FTR_ARCH_300))
+			cpu = cpu_first_thread_sibling(cpu);
+		cpumask_clear_cpu(cpu, &kvm->arch.need_tlb_flush);
 	}
 
 	return global;
@@ -79,10 +86,10 @@ void kvmppc_add_revmap_chain(struct kvm *kvm, struct revmap_entry *rev,
 
 	if (*rmap & KVMPPC_RMAP_PRESENT) {
 		i = *rmap & KVMPPC_RMAP_INDEX;
-		head = &kvm->arch.revmap[i];
+		head = &kvm->arch.hpt.rev[i];
 		if (realmode)
 			head = real_vmalloc_addr(head);
-		tail = &kvm->arch.revmap[head->back];
+		tail = &kvm->arch.hpt.rev[head->back];
 		if (realmode)
 			tail = real_vmalloc_addr(tail);
 		rev->forw = i;
@@ -147,8 +154,8 @@ static void remove_revmap_chain(struct kvm *kvm, long pte_index,
 	lock_rmap(rmap);
 
 	head = *rmap & KVMPPC_RMAP_INDEX;
-	next = real_vmalloc_addr(&kvm->arch.revmap[rev->forw]);
-	prev = real_vmalloc_addr(&kvm->arch.revmap[rev->back]);
+	next = real_vmalloc_addr(&kvm->arch.hpt.rev[rev->forw]);
+	prev = real_vmalloc_addr(&kvm->arch.hpt.rev[rev->back]);
 	next->back = rev->back;
 	prev->forw = rev->forw;
 	if (head == pte_index) {
@@ -182,6 +189,8 @@ long kvmppc_do_h_enter(struct kvm *kvm, unsigned long flags,
 	unsigned long mmu_seq;
 	unsigned long rcbits, irq_flags = 0;
 
+	if (kvm_is_radix(kvm))
+		return H_FUNCTION;
 	psize = hpte_page_size(pteh, ptel);
 	if (!psize)
 		return H_PARAMETER;
@@ -283,11 +292,11 @@ long kvmppc_do_h_enter(struct kvm *kvm, unsigned long flags,
 
 	/* Find and lock the HPTEG slot to use */
  do_insert:
-	if (pte_index >= kvm->arch.hpt_npte)
+	if (pte_index >= kvmppc_hpt_npte(&kvm->arch.hpt))
 		return H_PARAMETER;
 	if (likely((flags & H_EXACT) == 0)) {
 		pte_index &= ~7UL;
-		hpte = (__be64 *)(kvm->arch.hpt_virt + (pte_index << 4));
+		hpte = (__be64 *)(kvm->arch.hpt.virt + (pte_index << 4));
 		for (i = 0; i < 8; ++i) {
 			if ((be64_to_cpu(*hpte) & HPTE_V_VALID) == 0 &&
 			    try_lock_hpte(hpte, HPTE_V_HVLOCK | HPTE_V_VALID |
@@ -318,7 +327,7 @@ long kvmppc_do_h_enter(struct kvm *kvm, unsigned long flags,
 		}
 		pte_index += i;
 	} else {
-		hpte = (__be64 *)(kvm->arch.hpt_virt + (pte_index << 4));
+		hpte = (__be64 *)(kvm->arch.hpt.virt + (pte_index << 4));
 		if (!try_lock_hpte(hpte, HPTE_V_HVLOCK | HPTE_V_VALID |
 				   HPTE_V_ABSENT)) {
 			/* Lock the slot and check again */
@@ -335,7 +344,7 @@ long kvmppc_do_h_enter(struct kvm *kvm, unsigned long flags,
 	}
 
 	/* Save away the guest's idea of the second HPTE dword */
-	rev = &kvm->arch.revmap[pte_index];
+	rev = &kvm->arch.hpt.rev[pte_index];
 	if (realmode)
 		rev = real_vmalloc_addr(rev);
 	if (rev) {
@@ -458,9 +467,11 @@ long kvmppc_do_h_remove(struct kvm *kvm, unsigned long flags,
 	struct revmap_entry *rev;
 	u64 pte, orig_pte, pte_r;
 
-	if (pte_index >= kvm->arch.hpt_npte)
+	if (kvm_is_radix(kvm))
+		return H_FUNCTION;
+	if (pte_index >= kvmppc_hpt_npte(&kvm->arch.hpt))
 		return H_PARAMETER;
-	hpte = (__be64 *)(kvm->arch.hpt_virt + (pte_index << 4));
+	hpte = (__be64 *)(kvm->arch.hpt.virt + (pte_index << 4));
 	while (!try_lock_hpte(hpte, HPTE_V_HVLOCK))
 		cpu_relax();
 	pte = orig_pte = be64_to_cpu(hpte[0]);
@@ -476,7 +487,7 @@ long kvmppc_do_h_remove(struct kvm *kvm, unsigned long flags,
 		return H_NOT_FOUND;
 	}
 
-	rev = real_vmalloc_addr(&kvm->arch.revmap[pte_index]);
+	rev = real_vmalloc_addr(&kvm->arch.hpt.rev[pte_index]);
 	v = pte & ~HPTE_V_HVLOCK;
 	if (v & HPTE_V_VALID) {
 		hpte[0] &= ~cpu_to_be64(HPTE_V_VALID);
@@ -529,6 +540,8 @@ long kvmppc_h_bulk_remove(struct kvm_vcpu *vcpu)
 	struct revmap_entry *rev, *revs[4];
 	u64 hp0, hp1;
 
+	if (kvm_is_radix(kvm))
+		return H_FUNCTION;
 	global = global_invalidates(kvm, 0);
 	for (i = 0; i < 4 && ret == H_SUCCESS; ) {
 		n = 0;
@@ -544,13 +557,13 @@ long kvmppc_h_bulk_remove(struct kvm_vcpu *vcpu)
 				break;
 			}
 			if (req != 1 || flags == 3 ||
-			    pte_index >= kvm->arch.hpt_npte) {
+			    pte_index >= kvmppc_hpt_npte(&kvm->arch.hpt)) {
 				/* parameter error */
 				args[j] = ((0xa0 | flags) << 56) + pte_index;
 				ret = H_PARAMETER;
 				break;
 			}
-			hp = (__be64 *) (kvm->arch.hpt_virt + (pte_index << 4));
+			hp = (__be64 *) (kvm->arch.hpt.virt + (pte_index << 4));
 			/* to avoid deadlock, don't spin except for first */
 			if (!try_lock_hpte(hp, HPTE_V_HVLOCK)) {
 				if (n)
@@ -587,7 +600,7 @@ long kvmppc_h_bulk_remove(struct kvm_vcpu *vcpu)
 			}
 
 			args[j] = ((0x80 | flags) << 56) + pte_index;
-			rev = real_vmalloc_addr(&kvm->arch.revmap[pte_index]);
+			rev = real_vmalloc_addr(&kvm->arch.hpt.rev[pte_index]);
 			note_hpte_modification(kvm, rev);
 
 			if (!(hp0 & HPTE_V_VALID)) {
@@ -642,10 +655,12 @@ long kvmppc_h_protect(struct kvm_vcpu *vcpu, unsigned long flags,
 	unsigned long v, r, rb, mask, bits;
 	u64 pte_v, pte_r;
 
-	if (pte_index >= kvm->arch.hpt_npte)
+	if (kvm_is_radix(kvm))
+		return H_FUNCTION;
+	if (pte_index >= kvmppc_hpt_npte(&kvm->arch.hpt))
 		return H_PARAMETER;
 
-	hpte = (__be64 *)(kvm->arch.hpt_virt + (pte_index << 4));
+	hpte = (__be64 *)(kvm->arch.hpt.virt + (pte_index << 4));
 	while (!try_lock_hpte(hpte, HPTE_V_HVLOCK))
 		cpu_relax();
 	v = pte_v = be64_to_cpu(hpte[0]);
@@ -665,7 +680,7 @@ long kvmppc_h_protect(struct kvm_vcpu *vcpu, unsigned long flags,
 	/* Update guest view of 2nd HPTE dword */
 	mask = HPTE_R_PP0 | HPTE_R_PP | HPTE_R_N |
 		HPTE_R_KEY_HI | HPTE_R_KEY_LO;
-	rev = real_vmalloc_addr(&kvm->arch.revmap[pte_index]);
+	rev = real_vmalloc_addr(&kvm->arch.hpt.rev[pte_index]);
 	if (rev) {
 		r = (rev->guest_rpte & ~mask) | bits;
 		rev->guest_rpte = r;
@@ -711,15 +726,17 @@ long kvmppc_h_read(struct kvm_vcpu *vcpu, unsigned long flags,
 	int i, n = 1;
 	struct revmap_entry *rev = NULL;
 
-	if (pte_index >= kvm->arch.hpt_npte)
+	if (kvm_is_radix(kvm))
+		return H_FUNCTION;
+	if (pte_index >= kvmppc_hpt_npte(&kvm->arch.hpt))
 		return H_PARAMETER;
 	if (flags & H_READ_4) {
 		pte_index &= ~3;
 		n = 4;
 	}
-	rev = real_vmalloc_addr(&kvm->arch.revmap[pte_index]);
+	rev = real_vmalloc_addr(&kvm->arch.hpt.rev[pte_index]);
 	for (i = 0; i < n; ++i, ++pte_index) {
-		hpte = (__be64 *)(kvm->arch.hpt_virt + (pte_index << 4));
+		hpte = (__be64 *)(kvm->arch.hpt.virt + (pte_index << 4));
 		v = be64_to_cpu(hpte[0]) & ~HPTE_V_HVLOCK;
 		r = be64_to_cpu(hpte[1]);
 		if (cpu_has_feature(CPU_FTR_ARCH_300)) {
@@ -750,11 +767,13 @@ long kvmppc_h_clear_ref(struct kvm_vcpu *vcpu, unsigned long flags,
 	unsigned long *rmap;
 	long ret = H_NOT_FOUND;
 
-	if (pte_index >= kvm->arch.hpt_npte)
+	if (kvm_is_radix(kvm))
+		return H_FUNCTION;
+	if (pte_index >= kvmppc_hpt_npte(&kvm->arch.hpt))
 		return H_PARAMETER;
 
-	rev = real_vmalloc_addr(&kvm->arch.revmap[pte_index]);
-	hpte = (__be64 *)(kvm->arch.hpt_virt + (pte_index << 4));
+	rev = real_vmalloc_addr(&kvm->arch.hpt.rev[pte_index]);
+	hpte = (__be64 *)(kvm->arch.hpt.virt + (pte_index << 4));
 	while (!try_lock_hpte(hpte, HPTE_V_HVLOCK))
 		cpu_relax();
 	v = be64_to_cpu(hpte[0]);
@@ -796,11 +815,13 @@ long kvmppc_h_clear_mod(struct kvm_vcpu *vcpu, unsigned long flags,
 	unsigned long *rmap;
 	long ret = H_NOT_FOUND;
 
-	if (pte_index >= kvm->arch.hpt_npte)
+	if (kvm_is_radix(kvm))
+		return H_FUNCTION;
+	if (pte_index >= kvmppc_hpt_npte(&kvm->arch.hpt))
 		return H_PARAMETER;
 
-	rev = real_vmalloc_addr(&kvm->arch.revmap[pte_index]);
-	hpte = (__be64 *)(kvm->arch.hpt_virt + (pte_index << 4));
+	rev = real_vmalloc_addr(&kvm->arch.hpt.rev[pte_index]);
+	hpte = (__be64 *)(kvm->arch.hpt.virt + (pte_index << 4));
 	while (!try_lock_hpte(hpte, HPTE_V_HVLOCK))
 		cpu_relax();
 	v = be64_to_cpu(hpte[0]);
@@ -949,7 +970,7 @@ long kvmppc_hv_find_lock_hpte(struct kvm *kvm, gva_t eaddr, unsigned long slb_v,
 		somask = (1UL << 28) - 1;
 		vsid = (slb_v & ~SLB_VSID_B) >> SLB_VSID_SHIFT;
 	}
-	hash = (vsid ^ ((eaddr & somask) >> pshift)) & kvm->arch.hpt_mask;
+	hash = (vsid ^ ((eaddr & somask) >> pshift)) & kvmppc_hpt_mask(&kvm->arch.hpt);
 	avpn = slb_v & ~(somask >> 16);	/* also includes B */
 	avpn |= (eaddr & somask) >> 16;
 
@@ -960,7 +981,7 @@ long kvmppc_hv_find_lock_hpte(struct kvm *kvm, gva_t eaddr, unsigned long slb_v,
 	val |= avpn;
 
 	for (;;) {
-		hpte = (__be64 *)(kvm->arch.hpt_virt + (hash << 7));
+		hpte = (__be64 *)(kvm->arch.hpt.virt + (hash << 7));
 
 		for (i = 0; i < 16; i += 2) {
 			/* Read the PTE racily */
@@ -996,7 +1017,7 @@ long kvmppc_hv_find_lock_hpte(struct kvm *kvm, gva_t eaddr, unsigned long slb_v,
 		if (val & HPTE_V_SECONDARY)
 			break;
 		val |= HPTE_V_SECONDARY;
-		hash = hash ^ kvm->arch.hpt_mask;
+		hash = hash ^ kvmppc_hpt_mask(&kvm->arch.hpt);
 	}
 	return -1;
 }
@@ -1045,14 +1066,14 @@ long kvmppc_hpte_hv_fault(struct kvm_vcpu *vcpu, unsigned long addr,
 				return status;	/* there really was no HPTE */
 			return 0;	/* for prot fault, HPTE disappeared */
 		}
-		hpte = (__be64 *)(kvm->arch.hpt_virt + (index << 4));
+		hpte = (__be64 *)(kvm->arch.hpt.virt + (index << 4));
 		v = orig_v = be64_to_cpu(hpte[0]) & ~HPTE_V_HVLOCK;
 		r = be64_to_cpu(hpte[1]);
 		if (cpu_has_feature(CPU_FTR_ARCH_300)) {
 			v = hpte_new_to_old_v(v, r);
 			r = hpte_new_to_old_r(r);
 		}
-		rev = real_vmalloc_addr(&kvm->arch.revmap[index]);
+		rev = real_vmalloc_addr(&kvm->arch.hpt.rev[index]);
 		gr = rev->guest_rpte;
 
 		unlock_hpte(hpte, orig_v);

arch/powerpc/kvm/book3s_hv_rm_xics.c

@@ -35,7 +35,7 @@ int kvm_irq_bypass = 1;
 EXPORT_SYMBOL(kvm_irq_bypass);
 
 static void icp_rm_deliver_irq(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
-			    u32 new_irq);
+			    u32 new_irq, bool check_resend);
 static int xics_opal_rm_set_server(unsigned int hw_irq, int server_cpu);
 
 /* -- ICS routines -- */
@@ -44,20 +44,12 @@ static void ics_rm_check_resend(struct kvmppc_xics *xics,
 {
 	int i;
 
-	arch_spin_lock(&ics->lock);
-
 	for (i = 0; i < KVMPPC_XICS_IRQ_PER_ICS; i++) {
 		struct ics_irq_state *state = &ics->irq_state[i];
-
-		if (!state->resend)
-			continue;
-
-		arch_spin_unlock(&ics->lock);
-		icp_rm_deliver_irq(xics, icp, state->number);
-		arch_spin_lock(&ics->lock);
+		if (state->resend)
+			icp_rm_deliver_irq(xics, icp, state->number, true);
 	}
 
-	arch_spin_unlock(&ics->lock);
 }
 
 /* -- ICP routines -- */
@@ -70,11 +62,9 @@ static inline void icp_send_hcore_msg(int hcore, struct kvm_vcpu *vcpu)
 	hcpu = hcore << threads_shift;
 	kvmppc_host_rm_ops_hv->rm_core[hcore].rm_data = vcpu;
 	smp_muxed_ipi_set_message(hcpu, PPC_MSG_RM_HOST_ACTION);
-	if (paca[hcpu].kvm_hstate.xics_phys)
-		icp_native_cause_ipi_rm(hcpu);
-	else
-		opal_rm_int_set_mfrr(get_hard_smp_processor_id(hcpu),
-				     IPI_PRIORITY);
+	kvmppc_set_host_ipi(hcpu, 1);
+	smp_mb();
+	kvmhv_rm_send_ipi(hcpu);
 }
 #else
 static inline void icp_send_hcore_msg(int hcore, struct kvm_vcpu *vcpu) { }
@@ -290,7 +280,7 @@ static bool icp_rm_try_to_deliver(struct kvmppc_icp *icp, u32 irq, u8 priority,
 }
 
 static void icp_rm_deliver_irq(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
-			    u32 new_irq)
+			    u32 new_irq, bool check_resend)
 {
 	struct ics_irq_state *state;
 	struct kvmppc_ics *ics;
@@ -335,6 +325,10 @@ static void icp_rm_deliver_irq(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
 		}
 	}
 
+	if (check_resend)
+		if (!state->resend)
+			goto out;
+
 	/* Clear the resend bit of that interrupt */
 	state->resend = 0;
 
@@ -380,7 +374,9 @@ static void icp_rm_deliver_irq(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
 		 */
 		if (reject && reject != XICS_IPI) {
 			arch_spin_unlock(&ics->lock);
+			icp->n_reject++;
 			new_irq = reject;
+			check_resend = 0;
 			goto again;
 		}
 	} else {
@@ -388,9 +384,15 @@ static void icp_rm_deliver_irq(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
 		 * We failed to deliver the interrupt we need to set the
 		 * resend map bit and mark the ICS state as needing a resend
 		 */
-		set_bit(ics->icsid, icp->resend_map);
 		state->resend = 1;
 
+		/*
+		 * Make sure when checking resend, we don't miss the resend
+		 * if resend_map bit is seen and cleared.
+		 */
+		smp_wmb();
+		set_bit(ics->icsid, icp->resend_map);
+
 		/*
 		 * If the need_resend flag got cleared in the ICP some time
 		 * between icp_rm_try_to_deliver() atomic update and now, then
@@ -399,7 +401,9 @@ static void icp_rm_deliver_irq(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
 		 */
 		smp_mb();
 		if (!icp->state.need_resend) {
+			state->resend = 0;
 			arch_spin_unlock(&ics->lock);
+			check_resend = 0;
 			goto again;
 		}
 	}
@@ -594,7 +598,7 @@ int kvmppc_rm_h_ipi(struct kvm_vcpu *vcpu, unsigned long server,
 	/* Handle reject in real mode */
 	if (reject && reject != XICS_IPI) {
 		this_icp->n_reject++;
-		icp_rm_deliver_irq(xics, icp, reject);
+		icp_rm_deliver_irq(xics, icp, reject, false);
 	}
 
 	/* Handle resends in real mode */
@@ -662,59 +666,45 @@ int kvmppc_rm_h_cppr(struct kvm_vcpu *vcpu, unsigned long cppr)
 	 */
 	if (reject && reject != XICS_IPI) {
 		icp->n_reject++;
-		icp_rm_deliver_irq(xics, icp, reject);
+		icp_rm_deliver_irq(xics, icp, reject, false);
 	}
  bail:
 	return check_too_hard(xics, icp);
 }
 
-int kvmppc_rm_h_eoi(struct kvm_vcpu *vcpu, unsigned long xirr)
+static int ics_rm_eoi(struct kvm_vcpu *vcpu, u32 irq)
 {
 	struct kvmppc_xics *xics = vcpu->kvm->arch.xics;
 	struct kvmppc_icp *icp = vcpu->arch.icp;
 	struct kvmppc_ics *ics;
 	struct ics_irq_state *state;
-	u32 irq = xirr & 0x00ffffff;
 	u16 src;
-
-	if (!xics || !xics->real_mode)
-		return H_TOO_HARD;
+	u32 pq_old, pq_new;
 
 	/*
-	 * ICP State: EOI
-	 *
-	 * Note: If EOI is incorrectly used by SW to lower the CPPR
-	 * value (ie more favored), we do not check for rejection of
-	 * a pending interrupt, this is a SW error and PAPR sepcifies
-	 * that we don't have to deal with it.
+	 * ICS EOI handling: For LSI, if P bit is still set, we need to
+	 * resend it.
 	 *
-	 * The sending of an EOI to the ICS is handled after the
-	 * CPPR update
-	 *
-	 * ICP State: Down_CPPR which we handle
-	 * in a separate function as it's shared with H_CPPR.
+	 * For MSI, we move Q bit into P (and clear Q). If it is set,
+	 * resend it.
 	 */
-	icp_rm_down_cppr(xics, icp, xirr >> 24);
 
-	/* IPIs have no EOI */
-	if (irq == XICS_IPI)
-		goto bail;
-	/*
-	 * EOI handling: If the interrupt is still asserted, we need to
-	 * resend it. We can take a lockless "peek" at the ICS state here.
-	 *
-	 * "Message" interrupts will never have "asserted" set
-	 */
 	ics = kvmppc_xics_find_ics(xics, irq, &src);
 	if (!ics)
 		goto bail;
+
 	state = &ics->irq_state[src];
 
-	/* Still asserted, resend it */
-	if (state->asserted) {
-		icp->n_reject++;
-		icp_rm_deliver_irq(xics, icp, irq);
-	}
+	if (state->lsi)
+		pq_new = state->pq_state;
+	else
+		do {
+			pq_old = state->pq_state;
+			pq_new = pq_old >> 1;
+		} while (cmpxchg(&state->pq_state, pq_old, pq_new) != pq_old);
+
+	if (pq_new & PQ_PRESENTED)
+		icp_rm_deliver_irq(xics, NULL, irq, false);
 
 	if (!hlist_empty(&vcpu->kvm->irq_ack_notifier_list)) {
 		icp->rm_action |= XICS_RM_NOTIFY_EOI;
@@ -735,10 +725,43 @@ int kvmppc_rm_h_eoi(struct kvm_vcpu *vcpu, unsigned long xirr)
 			state->intr_cpu = -1;
 		}
 	}
+
  bail:
 	return check_too_hard(xics, icp);
 }
 
+int kvmppc_rm_h_eoi(struct kvm_vcpu *vcpu, unsigned long xirr)
+{
+	struct kvmppc_xics *xics = vcpu->kvm->arch.xics;
+	struct kvmppc_icp *icp = vcpu->arch.icp;
+	u32 irq = xirr & 0x00ffffff;
+
+	if (!xics || !xics->real_mode)
+		return H_TOO_HARD;
+
+	/*
+	 * ICP State: EOI
+	 *
+	 * Note: If EOI is incorrectly used by SW to lower the CPPR
+	 * value (ie more favored), we do not check for rejection of
+	 * a pending interrupt, this is a SW error and PAPR specifies
+	 * that we don't have to deal with it.
+	 *
+	 * The sending of an EOI to the ICS is handled after the
+	 * CPPR update
+	 *
+	 * ICP State: Down_CPPR which we handle
+	 * in a separate function as it's shared with H_CPPR.
+	 */
+	icp_rm_down_cppr(xics, icp, xirr >> 24);
+
+	/* IPIs have no EOI */
+	if (irq == XICS_IPI)
+		return check_too_hard(xics, icp);
+
+	return ics_rm_eoi(vcpu, irq);
+}
+
 unsigned long eoi_rc;
 
 static void icp_eoi(struct irq_chip *c, u32 hwirq, __be32 xirr, bool *again)
@@ -825,14 +848,33 @@ long kvmppc_deliver_irq_passthru(struct kvm_vcpu *vcpu,
 {
 	struct kvmppc_xics *xics;
 	struct kvmppc_icp *icp;
+	struct kvmppc_ics *ics;
+	struct ics_irq_state *state;
 	u32 irq;
+	u16 src;
+	u32 pq_old, pq_new;
 
 	irq = irq_map->v_hwirq;
 	xics = vcpu->kvm->arch.xics;
 	icp = vcpu->arch.icp;
 
 	kvmppc_rm_handle_irq_desc(irq_map->desc);
-	icp_rm_deliver_irq(xics, icp, irq);
+
+	ics = kvmppc_xics_find_ics(xics, irq, &src);
+	if (!ics)
+		return 2;
+
+	state = &ics->irq_state[src];
+
+	/* only MSIs register bypass producers, so it must be MSI here */
+	do {
+		pq_old = state->pq_state;
+		pq_new = ((pq_old << 1) & 3) | PQ_PRESENTED;
+	} while (cmpxchg(&state->pq_state, pq_old, pq_new) != pq_old);
+
+	/* Test P=1, Q=0, this is the only case where we present */
+	if (pq_new == PQ_PRESENTED)
+		icp_rm_deliver_irq(xics, icp, irq, false);
 
 	/* EOI the interrupt */
 	icp_eoi(irq_desc_get_chip(irq_map->desc), irq_map->r_hwirq, xirr,

arch/powerpc/kvm/book3s_hv_rmhandlers.S

@@ -148,6 +148,15 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S)
 	addi	r1, r1, 112
 	ld	r7, HSTATE_HOST_MSR(r13)
 
+	/*
+	 * If we came back from the guest via a relocation-on interrupt,
+	 * we will be in virtual mode at this point, which makes it a
+	 * little easier to get back to the caller.
+	 */
+	mfmsr	r0
+	andi.	r0, r0, MSR_IR		/* in real mode? */
+	bne	.Lvirt_return
+
 	cmpwi	cr1, r12, BOOK3S_INTERRUPT_MACHINE_CHECK
 	cmpwi	r12, BOOK3S_INTERRUPT_EXTERNAL
 	beq	11f
@@ -181,6 +190,26 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S)
 	mtspr SPRN_HSRR1, r7
 	ba    0xe80
 
+	/* Virtual-mode return - can't get here for HMI or machine check */
+.Lvirt_return:
+	cmpwi	r12, BOOK3S_INTERRUPT_EXTERNAL
+	beq	16f
+	cmpwi	r12, BOOK3S_INTERRUPT_H_DOORBELL
+	beq	17f
+	andi.	r0, r7, MSR_EE		/* were interrupts hard-enabled? */
+	beq	18f
+	mtmsrd	r7, 1			/* if so then re-enable them */
+18:	mtlr	r8
+	blr
+
+16:	mtspr	SPRN_HSRR0, r8		/* jump to reloc-on external vector */
+	mtspr	SPRN_HSRR1, r7
+	b	exc_virt_0x4500_hardware_interrupt
+
+17:	mtspr	SPRN_HSRR0, r8
+	mtspr	SPRN_HSRR1, r7
+	b	exc_virt_0x4e80_h_doorbell
+
 kvmppc_primary_no_guest:
 	/* We handle this much like a ceded vcpu */
 	/* put the HDEC into the DEC, since HDEC interrupts don't wake us */
@@ -518,6 +547,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S)
 /* Stack frame offsets */
 #define STACK_SLOT_TID		(112-16)
 #define STACK_SLOT_PSSCR	(112-24)
+#define STACK_SLOT_PID		(112-32)
 
 .global kvmppc_hv_entry
 kvmppc_hv_entry:
@@ -530,6 +560,7 @@ kvmppc_hv_entry:
 	 * R1 = host R1
 	 * R2 = TOC
 	 * all other volatile GPRS = free
+	 * Does not preserve non-volatile GPRs or CR fields
 	 */
 	mflr	r0
 	std	r0, PPC_LR_STKOFF(r1)
@@ -549,32 +580,38 @@ kvmppc_hv_entry:
 	bl	kvmhv_start_timing
 1:
 #endif
-	/* Clear out SLB */
+
+	/* Use cr7 as an indication of radix mode */
+	ld	r5, HSTATE_KVM_VCORE(r13)
+	ld	r9, VCORE_KVM(r5)	/* pointer to struct kvm */
+	lbz	r0, KVM_RADIX(r9)
+	cmpwi	cr7, r0, 0
+
+	/* Clear out SLB if hash */
+	bne	cr7, 2f
 	li	r6,0
 	slbmte	r6,r6
 	slbia
 	ptesync
-
+2:
 	/*
 	 * POWER7/POWER8 host -> guest partition switch code.
 	 * We don't have to lock against concurrent tlbies,
 	 * but we do have to coordinate across hardware threads.
 	 */
 	/* Set bit in entry map iff exit map is zero. */
-	ld	r5, HSTATE_KVM_VCORE(r13)
 	li	r7, 1
 	lbz	r6, HSTATE_PTID(r13)
 	sld	r7, r7, r6
-	addi	r9, r5, VCORE_ENTRY_EXIT
-21:	lwarx	r3, 0, r9
+	addi	r8, r5, VCORE_ENTRY_EXIT
+21:	lwarx	r3, 0, r8
 	cmpwi	r3, 0x100		/* any threads starting to exit? */
 	bge	secondary_too_late	/* if so we're too late to the party */
 	or	r3, r3, r7
-	stwcx.	r3, 0, r9
+	stwcx.	r3, 0, r8
 	bne	21b
 
 	/* Primary thread switches to guest partition. */
-	ld	r9,VCORE_KVM(r5)	/* pointer to struct kvm */
 	cmpwi	r6,0
 	bne	10f
 	lwz	r7,KVM_LPID(r9)
@@ -590,30 +627,44 @@ END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_300)
 
 	/* See if we need to flush the TLB */
 	lhz	r6,PACAPACAINDEX(r13)	/* test_bit(cpu, need_tlb_flush) */
+BEGIN_FTR_SECTION
+	/*
+	 * On POWER9, individual threads can come in here, but the
+	 * TLB is shared between the 4 threads in a core, hence
+	 * invalidating on one thread invalidates for all.
+	 * Thus we make all 4 threads use the same bit here.
+	 */
+	clrrdi	r6,r6,2
+END_FTR_SECTION_IFSET(CPU_FTR_ARCH_300)
 	clrldi	r7,r6,64-6		/* extract bit number (6 bits) */
 	srdi	r6,r6,6			/* doubleword number */
 	sldi	r6,r6,3			/* address offset */
 	add	r6,r6,r9
 	addi	r6,r6,KVM_NEED_FLUSH	/* dword in kvm->arch.need_tlb_flush */
-	li	r0,1
-	sld	r0,r0,r7
+	li	r8,1
+	sld	r8,r8,r7
 	ld	r7,0(r6)
-	and.	r7,r7,r0
+	and.	r7,r7,r8
 	beq	22f
-23:	ldarx	r7,0,r6			/* if set, clear the bit */
-	andc	r7,r7,r0
-	stdcx.	r7,0,r6
-	bne	23b
 	/* Flush the TLB of any entries for this LPID */
-	lwz	r6,KVM_TLB_SETS(r9)
-	li	r0,0			/* RS for P9 version of tlbiel */
-	mtctr	r6
+	lwz	r0,KVM_TLB_SETS(r9)
+	mtctr	r0
 	li	r7,0x800		/* IS field = 0b10 */
 	ptesync
-28:	tlbiel	r7
+	li	r0,0			/* RS for P9 version of tlbiel */
+	bne	cr7, 29f
+28:	tlbiel	r7			/* On P9, rs=0, RIC=0, PRS=0, R=0 */
 	addi	r7,r7,0x1000
 	bdnz	28b
-	ptesync
+	b	30f
+29:	PPC_TLBIEL(7,0,2,1,1)		/* for radix, RIC=2, PRS=1, R=1 */
+	addi	r7,r7,0x1000
+	bdnz	29b
+30:	ptesync
+23:	ldarx	r7,0,r6			/* clear the bit after TLB flushed */
+	andc	r7,r7,r8
+	stdcx.	r7,0,r6
+	bne	23b
 
 	/* Add timebase offset onto timebase */
 22:	ld	r8,VCORE_TB_OFFSET(r5)
@@ -658,7 +709,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S)
 	beq	kvmppc_primary_no_guest
 kvmppc_got_guest:
 
-	/* Load up guest SLB entries */
+	/* Load up guest SLB entries (N.B. slb_max will be 0 for radix) */
 	lwz	r5,VCPU_SLB_MAX(r4)
 	cmpwi	r5,0
 	beq	9f
@@ -696,8 +747,10 @@ kvmppc_got_guest:
 BEGIN_FTR_SECTION
 	mfspr	r5, SPRN_TIDR
 	mfspr	r6, SPRN_PSSCR
+	mfspr	r7, SPRN_PID
 	std	r5, STACK_SLOT_TID(r1)
 	std	r6, STACK_SLOT_PSSCR(r1)
+	std	r7, STACK_SLOT_PID(r1)
 END_FTR_SECTION_IFSET(CPU_FTR_ARCH_300)
 
 BEGIN_FTR_SECTION
@@ -823,6 +876,9 @@ END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_207S)
 	mtspr	SPRN_BESCR, r6
 	mtspr	SPRN_PID, r7
 	mtspr	SPRN_WORT, r8
+BEGIN_FTR_SECTION
+	PPC_INVALIDATE_ERAT
+END_FTR_SECTION_IFSET(CPU_FTR_POWER9_DD1)
 BEGIN_FTR_SECTION
 	/* POWER8-only registers */
 	ld	r5, VCPU_TCSCR(r4)
@@ -1057,13 +1113,13 @@ hdec_soon:
 kvmppc_interrupt_hv:
 	/*
 	 * Register contents:
-	 * R12		= interrupt vector
+	 * R12		= (guest CR << 32) | interrupt vector
 	 * R13		= PACA
-	 * guest CR, R12 saved in shadow VCPU SCRATCH1/0
+	 * guest R12 saved in shadow VCPU SCRATCH0
+	 * guest CTR saved in shadow VCPU SCRATCH1 if RELOCATABLE
 	 * guest R13 saved in SPRN_SCRATCH0
 	 */
 	std	r9, HSTATE_SCRATCH2(r13)
-
 	lbz	r9, HSTATE_IN_GUEST(r13)
 	cmpwi	r9, KVM_GUEST_MODE_HOST_HV
 	beq	kvmppc_bad_host_intr
@@ -1094,8 +1150,9 @@ kvmppc_interrupt_hv:
 	std	r10, VCPU_GPR(R10)(r9)
 	std	r11, VCPU_GPR(R11)(r9)
 	ld	r3, HSTATE_SCRATCH0(r13)
-	lwz	r4, HSTATE_SCRATCH1(r13)
 	std	r3, VCPU_GPR(R12)(r9)
+	/* CR is in the high half of r12 */
+	srdi	r4, r12, 32
 	stw	r4, VCPU_CR(r9)
 BEGIN_FTR_SECTION
 	ld	r3, HSTATE_CFAR(r13)
@@ -1114,6 +1171,8 @@ END_FTR_SECTION_IFSET(CPU_FTR_HAS_PPR)
 	mfspr	r11, SPRN_SRR1
 	std	r10, VCPU_SRR0(r9)
 	std	r11, VCPU_SRR1(r9)
+	/* trap is in the low half of r12, clear CR from the high half */
+	clrldi	r12, r12, 32
 	andi.	r0, r12, 2		/* need to read HSRR0/1? */
 	beq	1f
 	mfspr	r10, SPRN_HSRR0
@@ -1149,7 +1208,12 @@ END_FTR_SECTION_IFSET(CPU_FTR_HAS_PPR)
 11:	stw	r3,VCPU_HEIR(r9)
 
 	/* these are volatile across C function calls */
+#ifdef CONFIG_RELOCATABLE
+	ld	r3, HSTATE_SCRATCH1(r13)
+	mtctr	r3
+#else
 	mfctr	r3
+#endif
 	mfxer	r4
 	std	r3, VCPU_CTR(r9)
 	std	r4, VCPU_XER(r9)
@@ -1285,11 +1349,15 @@ mc_cont:
 	mtspr	SPRN_CTRLT,r6
 4:
 	/* Read the guest SLB and save it away */
+	ld	r5, VCPU_KVM(r9)
+	lbz	r0, KVM_RADIX(r5)
+	cmpwi	r0, 0
+	li	r5, 0
+	bne	3f			/* for radix, save 0 entries */
 	lwz	r0,VCPU_SLB_NR(r9)	/* number of entries in SLB */
 	mtctr	r0
 	li	r6,0
 	addi	r7,r9,VCPU_SLB
-	li	r5,0
 1:	slbmfee	r8,r6
 	andis.	r0,r8,SLB_ESID_V@h
 	beq	2f
@@ -1301,7 +1369,7 @@ mc_cont:
 	addi	r5,r5,1
 2:	addi	r6,r6,1
 	bdnz	1b
-	stw	r5,VCPU_SLB_MAX(r9)
+3:	stw	r5,VCPU_SLB_MAX(r9)
 
 	/*
 	 * Save the guest PURR/SPURR
@@ -1550,9 +1618,14 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S)
 BEGIN_FTR_SECTION
 	ld	r5, STACK_SLOT_TID(r1)
 	ld	r6, STACK_SLOT_PSSCR(r1)
+	ld	r7, STACK_SLOT_PID(r1)
 	mtspr	SPRN_TIDR, r5
 	mtspr	SPRN_PSSCR, r6
+	mtspr	SPRN_PID, r7
 END_FTR_SECTION_IFSET(CPU_FTR_ARCH_300)
+BEGIN_FTR_SECTION
+	PPC_INVALIDATE_ERAT
+END_FTR_SECTION_IFSET(CPU_FTR_POWER9_DD1)
 
 	/*
 	 * POWER7/POWER8 guest -> host partition switch code.
@@ -1663,6 +1736,9 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S)
 	isync
 
 	/* load host SLB entries */
+BEGIN_MMU_FTR_SECTION
+	b	0f
+END_MMU_FTR_SECTION_IFSET(MMU_FTR_TYPE_RADIX)
 	ld	r8,PACA_SLBSHADOWPTR(r13)
 
 	.rept	SLB_NUM_BOLTED
@@ -1675,7 +1751,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S)
 	slbmte	r6,r5
 1:	addi	r8,r8,16
 	.endr
-
+0:
 #ifdef CONFIG_KVM_BOOK3S_HV_EXIT_TIMING
 	/* Finish timing, if we have a vcpu */
 	ld	r4, HSTATE_KVM_VCPU(r13)
@@ -1702,11 +1778,19 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S)
  * reflect the HDSI to the guest as a DSI.
  */
 kvmppc_hdsi:
+	ld	r3, VCPU_KVM(r9)
+	lbz	r0, KVM_RADIX(r3)
+	cmpwi	r0, 0
 	mfspr	r4, SPRN_HDAR
 	mfspr	r6, SPRN_HDSISR
+	bne	.Lradix_hdsi		/* on radix, just save DAR/DSISR/ASDR */
 	/* HPTE not found fault or protection fault? */
 	andis.	r0, r6, (DSISR_NOHPTE | DSISR_PROTFAULT)@h
 	beq	1f			/* if not, send it to the guest */
+BEGIN_FTR_SECTION
+	mfspr	r5, SPRN_ASDR		/* on POWER9, use ASDR to get VSID */
+	b	4f
+END_FTR_SECTION_IFSET(CPU_FTR_ARCH_300)
 	andi.	r0, r11, MSR_DR		/* data relocation enabled? */
 	beq	3f
 	clrrdi	r0, r4, 28
@@ -1776,13 +1860,29 @@ fast_interrupt_c_return:
 	stb	r0, HSTATE_IN_GUEST(r13)
 	b	guest_exit_cont
 
+.Lradix_hdsi:
+	std	r4, VCPU_FAULT_DAR(r9)
+	stw	r6, VCPU_FAULT_DSISR(r9)
+.Lradix_hisi:
+	mfspr	r5, SPRN_ASDR
+	std	r5, VCPU_FAULT_GPA(r9)
+	b	guest_exit_cont
+
 /*
  * Similarly for an HISI, reflect it to the guest as an ISI unless
  * it is an HPTE not found fault for a page that we have paged out.
  */
 kvmppc_hisi:
+	ld	r3, VCPU_KVM(r9)
+	lbz	r0, KVM_RADIX(r3)
+	cmpwi	r0, 0
+	bne	.Lradix_hisi		/* for radix, just save ASDR */
 	andis.	r0, r11, SRR1_ISI_NOPT@h
 	beq	1f
+BEGIN_FTR_SECTION
+	mfspr	r5, SPRN_ASDR		/* on POWER9, use ASDR to get VSID */
+	b	4f
+END_FTR_SECTION_IFSET(CPU_FTR_ARCH_300)
 	andi.	r0, r11, MSR_IR		/* instruction relocation enabled? */
 	beq	3f
 	clrrdi	r0, r10, 28

arch/powerpc/kvm/book3s_pr.c

@@ -902,6 +902,69 @@ static void kvmppc_clear_debug(struct kvm_vcpu *vcpu)
 	}
 }
 
+static int kvmppc_exit_pr_progint(struct kvm_run *run, struct kvm_vcpu *vcpu,
+				  unsigned int exit_nr)
+{
+	enum emulation_result er;
+	ulong flags;
+	u32 last_inst;
+	int emul, r;
+
+	/*
+	 * shadow_srr1 only contains valid flags if we came here via a program
+	 * exception. The other exceptions (emulation assist, FP unavailable,
+	 * etc.) do not provide flags in SRR1, so use an illegal-instruction
+	 * exception when injecting a program interrupt into the guest.
+	 */
+	if (exit_nr == BOOK3S_INTERRUPT_PROGRAM)
+		flags = vcpu->arch.shadow_srr1 & 0x1f0000ull;
+	else
+		flags = SRR1_PROGILL;
+
+	emul = kvmppc_get_last_inst(vcpu, INST_GENERIC, &last_inst);
+	if (emul != EMULATE_DONE)
+		return RESUME_GUEST;
+
+	if (kvmppc_get_msr(vcpu) & MSR_PR) {
+#ifdef EXIT_DEBUG
+		pr_info("Userspace triggered 0x700 exception at\n 0x%lx (0x%x)\n",
+			kvmppc_get_pc(vcpu), last_inst);
+#endif
+		if ((last_inst & 0xff0007ff) != (INS_DCBZ & 0xfffffff7)) {
+			kvmppc_core_queue_program(vcpu, flags);
+			return RESUME_GUEST;
+		}
+	}
+
+	vcpu->stat.emulated_inst_exits++;
+	er = kvmppc_emulate_instruction(run, vcpu);
+	switch (er) {
+	case EMULATE_DONE:
+		r = RESUME_GUEST_NV;
+		break;
+	case EMULATE_AGAIN:
+		r = RESUME_GUEST;
+		break;
+	case EMULATE_FAIL:
+		pr_crit("%s: emulation at %lx failed (%08x)\n",
+			__func__, kvmppc_get_pc(vcpu), last_inst);
+		kvmppc_core_queue_program(vcpu, flags);
+		r = RESUME_GUEST;
+		break;
+	case EMULATE_DO_MMIO:
+		run->exit_reason = KVM_EXIT_MMIO;
+		r = RESUME_HOST_NV;
+		break;
+	case EMULATE_EXIT_USER:
+		r = RESUME_HOST_NV;
+		break;
+	default:
+		BUG();
+	}
+
+	return r;
+}
+
 int kvmppc_handle_exit_pr(struct kvm_run *run, struct kvm_vcpu *vcpu,
 			  unsigned int exit_nr)
 {
@@ -1044,71 +1107,8 @@ int kvmppc_handle_exit_pr(struct kvm_run *run, struct kvm_vcpu *vcpu,
 		break;
 	case BOOK3S_INTERRUPT_PROGRAM:
 	case BOOK3S_INTERRUPT_H_EMUL_ASSIST:
-	{
-		enum emulation_result er;
-		ulong flags;
-		u32 last_inst;
-		int emul;
-
-program_interrupt:
-		/*
-		 * shadow_srr1 only contains valid flags if we came here via
-		 * a program exception. The other exceptions (emulation assist,
-		 * FP unavailable, etc.) do not provide flags in SRR1, so use
-		 * an illegal-instruction exception when injecting a program
-		 * interrupt into the guest.
-		 */
-		if (exit_nr == BOOK3S_INTERRUPT_PROGRAM)
-			flags = vcpu->arch.shadow_srr1 & 0x1f0000ull;
-		else
-			flags = SRR1_PROGILL;
-
-		emul = kvmppc_get_last_inst(vcpu, INST_GENERIC, &last_inst);
-		if (emul != EMULATE_DONE) {
-			r = RESUME_GUEST;
-			break;
-		}
-
-		if (kvmppc_get_msr(vcpu) & MSR_PR) {
-#ifdef EXIT_DEBUG
-			pr_info("Userspace triggered 0x700 exception at\n 0x%lx (0x%x)\n",
-				kvmppc_get_pc(vcpu), last_inst);
-#endif
-			if ((last_inst & 0xff0007ff) !=
-			    (INS_DCBZ & 0xfffffff7)) {
-				kvmppc_core_queue_program(vcpu, flags);
-				r = RESUME_GUEST;
-				break;
-			}
-		}
-
-		vcpu->stat.emulated_inst_exits++;
-		er = kvmppc_emulate_instruction(run, vcpu);
-		switch (er) {
-		case EMULATE_DONE:
-			r = RESUME_GUEST_NV;
-			break;
-		case EMULATE_AGAIN:
-			r = RESUME_GUEST;
-			break;
-		case EMULATE_FAIL:
-			printk(KERN_CRIT "%s: emulation at %lx failed (%08x)\n",
-			       __func__, kvmppc_get_pc(vcpu), last_inst);
-			kvmppc_core_queue_program(vcpu, flags);
-			r = RESUME_GUEST;
-			break;
-		case EMULATE_DO_MMIO:
-			run->exit_reason = KVM_EXIT_MMIO;
-			r = RESUME_HOST_NV;
-			break;
-		case EMULATE_EXIT_USER:
-			r = RESUME_HOST_NV;
-			break;
-		default:
-			BUG();
-		}
+		r = kvmppc_exit_pr_progint(run, vcpu, exit_nr);
 		break;
-	}
 	case BOOK3S_INTERRUPT_SYSCALL:
 	{
 		u32 last_sc;
@@ -1185,7 +1185,7 @@ int kvmppc_handle_exit_pr(struct kvm_run *run, struct kvm_vcpu *vcpu,
 			emul = kvmppc_get_last_inst(vcpu, INST_GENERIC,
 						    &last_inst);
 			if (emul == EMULATE_DONE)
-				goto program_interrupt;
+				r = kvmppc_exit_pr_progint(run, vcpu, exit_nr);
 			else
 				r = RESUME_GUEST;
 

arch/powerpc/kvm/book3s_segment.S

@@ -167,20 +167,38 @@ kvmppc_handler_trampoline_enter_end:
  *                                                                            *
  *****************************************************************************/
 
-.global kvmppc_handler_trampoline_exit
-kvmppc_handler_trampoline_exit:
-
 .global kvmppc_interrupt_pr
 kvmppc_interrupt_pr:
+	/* 64-bit entry. Register usage at this point:
+	 *
+	 * SPRG_SCRATCH0   = guest R13
+	 * R12             = (guest CR << 32) | exit handler id
+	 * R13             = PACA
+	 * HSTATE.SCRATCH0 = guest R12
+	 * HSTATE.SCRATCH1 = guest CTR if RELOCATABLE
+	 */
+#ifdef CONFIG_PPC64
+	/* Match 32-bit entry */
+#ifdef CONFIG_RELOCATABLE
+	std	r9, HSTATE_SCRATCH2(r13)
+	ld	r9, HSTATE_SCRATCH1(r13)
+	mtctr	r9
+	ld	r9, HSTATE_SCRATCH2(r13)
+#endif
+	rotldi	r12, r12, 32		  /* Flip R12 halves for stw */
+	stw	r12, HSTATE_SCRATCH1(r13) /* CR is now in the low half */
+	srdi	r12, r12, 32		  /* shift trap into low half */
+#endif
 
+.global kvmppc_handler_trampoline_exit
+kvmppc_handler_trampoline_exit:
 	/* Register usage at this point:
 	 *
-	 * SPRG_SCRATCH0  = guest R13
-	 * R12            = exit handler id
-	 * R13            = shadow vcpu (32-bit) or PACA (64-bit)
+	 * SPRG_SCRATCH0   = guest R13
+	 * R12             = exit handler id
+	 * R13             = shadow vcpu (32-bit) or PACA (64-bit)
 	 * HSTATE.SCRATCH0 = guest R12
 	 * HSTATE.SCRATCH1 = guest CR
-	 *
 	 */
 
 	/* Save registers */

arch/powerpc/kvm/book3s_xics.c

@@ -63,7 +63,7 @@
 /* -- ICS routines -- */
 
 static void icp_deliver_irq(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
-			    u32 new_irq);
+			    u32 new_irq, bool check_resend);
 
 /*
  * Return value ideally indicates how the interrupt was handled, but no
@@ -75,6 +75,7 @@ static int ics_deliver_irq(struct kvmppc_xics *xics, u32 irq, u32 level)
 	struct ics_irq_state *state;
 	struct kvmppc_ics *ics;
 	u16 src;
+	u32 pq_old, pq_new;
 
 	XICS_DBG("ics deliver %#x (level: %d)\n", irq, level);
 
@@ -87,25 +88,41 @@ static int ics_deliver_irq(struct kvmppc_xics *xics, u32 irq, u32 level)
 	if (!state->exists)
 		return -EINVAL;
 
+	if (level == KVM_INTERRUPT_SET_LEVEL || level == KVM_INTERRUPT_SET)
+		level = 1;
+	else if (level == KVM_INTERRUPT_UNSET)
+		level = 0;
 	/*
-	 * We set state->asserted locklessly. This should be fine as
-	 * we are the only setter, thus concurrent access is undefined
-	 * to begin with.
+	 * Take other values the same as 1, consistent with original code.
+	 * maybe WARN here?
 	 */
-	if ((level == 1 && state->lsi) || level == KVM_INTERRUPT_SET_LEVEL)
-		state->asserted = 1;
-	else if (level == 0 || level == KVM_INTERRUPT_UNSET) {
-		state->asserted = 0;
+
+	if (!state->lsi && level == 0) /* noop for MSI */
 		return 0;
-	}
+
+	do {
+		pq_old = state->pq_state;
+		if (state->lsi) {
+			if (level) {
+				if (pq_old & PQ_PRESENTED)
+					/* Setting already set LSI ... */
+					return 0;
+
+				pq_new = PQ_PRESENTED;
+			} else
+				pq_new = 0;
+		} else
+			pq_new = ((pq_old << 1) & 3) | PQ_PRESENTED;
+	} while (cmpxchg(&state->pq_state, pq_old, pq_new) != pq_old);
+
+	/* Test P=1, Q=0, this is the only case where we present */
+	if (pq_new == PQ_PRESENTED)
+		icp_deliver_irq(xics, NULL, irq, false);
 
 	/* Record which CPU this arrived on for passed-through interrupts */
 	if (state->host_irq)
 		state->intr_cpu = raw_smp_processor_id();
 
-	/* Attempt delivery */
-	icp_deliver_irq(xics, NULL, irq);
-
 	return 0;
 }
 
@@ -114,29 +131,14 @@ static void ics_check_resend(struct kvmppc_xics *xics, struct kvmppc_ics *ics,
 {
 	int i;
 
-	unsigned long flags;
-
-	local_irq_save(flags);
-	arch_spin_lock(&ics->lock);
-
 	for (i = 0; i < KVMPPC_XICS_IRQ_PER_ICS; i++) {
 		struct ics_irq_state *state = &ics->irq_state[i];
-
-		if (!state->resend)
-			continue;
-
-		XICS_DBG("resend %#x prio %#x\n", state->number,
-			      state->priority);
-
-		arch_spin_unlock(&ics->lock);
-		local_irq_restore(flags);
-		icp_deliver_irq(xics, icp, state->number);
-		local_irq_save(flags);
-		arch_spin_lock(&ics->lock);
+		if (state->resend) {
+			XICS_DBG("resend %#x prio %#x\n", state->number,
+				      state->priority);
+			icp_deliver_irq(xics, icp, state->number, true);
+		}
 	}
-
-	arch_spin_unlock(&ics->lock);
-	local_irq_restore(flags);
 }
 
 static bool write_xive(struct kvmppc_xics *xics, struct kvmppc_ics *ics,
@@ -155,6 +157,7 @@ static bool write_xive(struct kvmppc_xics *xics, struct kvmppc_ics *ics,
 	deliver = false;
 	if ((state->masked_pending || state->resend) && priority != MASKED) {
 		state->masked_pending = 0;
+		state->resend = 0;
 		deliver = true;
 	}
 
@@ -189,7 +192,7 @@ int kvmppc_xics_set_xive(struct kvm *kvm, u32 irq, u32 server, u32 priority)
 		 state->masked_pending, state->resend);
 
 	if (write_xive(xics, ics, state, server, priority, priority))
-		icp_deliver_irq(xics, icp, irq);
+		icp_deliver_irq(xics, icp, irq, false);
 
 	return 0;
 }
@@ -242,7 +245,7 @@ int kvmppc_xics_int_on(struct kvm *kvm, u32 irq)
 
 	if (write_xive(xics, ics, state, state->server, state->saved_priority,
 		       state->saved_priority))
-		icp_deliver_irq(xics, icp, irq);
+		icp_deliver_irq(xics, icp, irq, false);
 
 	return 0;
 }
@@ -376,7 +379,7 @@ static bool icp_try_to_deliver(struct kvmppc_icp *icp, u32 irq, u8 priority,
 }
 
 static void icp_deliver_irq(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
-			    u32 new_irq)
+			    u32 new_irq, bool check_resend)
 {
 	struct ics_irq_state *state;
 	struct kvmppc_ics *ics;
@@ -422,6 +425,10 @@ static void icp_deliver_irq(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
 		}
 	}
 
+	if (check_resend)
+		if (!state->resend)
+			goto out;
+
 	/* Clear the resend bit of that interrupt */
 	state->resend = 0;
 
@@ -470,6 +477,7 @@ static void icp_deliver_irq(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
 			arch_spin_unlock(&ics->lock);
 			local_irq_restore(flags);
 			new_irq = reject;
+			check_resend = 0;
 			goto again;
 		}
 	} else {
@@ -477,9 +485,15 @@ static void icp_deliver_irq(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
 		 * We failed to deliver the interrupt we need to set the
 		 * resend map bit and mark the ICS state as needing a resend
 		 */
-		set_bit(ics->icsid, icp->resend_map);
 		state->resend = 1;
 
+		/*
+		 * Make sure when checking resend, we don't miss the resend
+		 * if resend_map bit is seen and cleared.
+		 */
+		smp_wmb();
+		set_bit(ics->icsid, icp->resend_map);
+
 		/*
 		 * If the need_resend flag got cleared in the ICP some time
 		 * between icp_try_to_deliver() atomic update and now, then
@@ -488,8 +502,10 @@ static void icp_deliver_irq(struct kvmppc_xics *xics, struct kvmppc_icp *icp,
 		 */
 		smp_mb();
 		if (!icp->state.need_resend) {
+			state->resend = 0;
 			arch_spin_unlock(&ics->lock);
 			local_irq_restore(flags);
+			check_resend = 0;
 			goto again;
 		}
 	}
@@ -681,7 +697,7 @@ static noinline int kvmppc_h_ipi(struct kvm_vcpu *vcpu, unsigned long server,
 
 	/* Handle reject */
 	if (reject && reject != XICS_IPI)
-		icp_deliver_irq(xics, icp, reject);
+		icp_deliver_irq(xics, icp, reject, false);
 
 	/* Handle resend */
 	if (resend)
@@ -761,17 +777,54 @@ static noinline void kvmppc_h_cppr(struct kvm_vcpu *vcpu, unsigned long cppr)
 	 * attempt (see comments in icp_deliver_irq).
 	 */
 	if (reject && reject != XICS_IPI)
-		icp_deliver_irq(xics, icp, reject);
+		icp_deliver_irq(xics, icp, reject, false);
 }
 
-static noinline int kvmppc_h_eoi(struct kvm_vcpu *vcpu, unsigned long xirr)
+static int ics_eoi(struct kvm_vcpu *vcpu, u32 irq)
 {
 	struct kvmppc_xics *xics = vcpu->kvm->arch.xics;
 	struct kvmppc_icp *icp = vcpu->arch.icp;
 	struct kvmppc_ics *ics;
 	struct ics_irq_state *state;
-	u32 irq = xirr & 0x00ffffff;
 	u16 src;
+	u32 pq_old, pq_new;
+
+	/*
+	 * ICS EOI handling: For LSI, if P bit is still set, we need to
+	 * resend it.
+	 *
+	 * For MSI, we move Q bit into P (and clear Q). If it is set,
+	 * resend it.
+	 */
+
+	ics = kvmppc_xics_find_ics(xics, irq, &src);
+	if (!ics) {
+		XICS_DBG("ios_eoi: IRQ 0x%06x not found !\n", irq);
+		return H_PARAMETER;
+	}
+	state = &ics->irq_state[src];
+
+	if (state->lsi)
+		pq_new = state->pq_state;
+	else
+		do {
+			pq_old = state->pq_state;
+			pq_new = pq_old >> 1;
+		} while (cmpxchg(&state->pq_state, pq_old, pq_new) != pq_old);
+
+	if (pq_new & PQ_PRESENTED)
+		icp_deliver_irq(xics, icp, irq, false);
+
+	kvm_notify_acked_irq(vcpu->kvm, 0, irq);
+
+	return H_SUCCESS;
+}
+
+static noinline int kvmppc_h_eoi(struct kvm_vcpu *vcpu, unsigned long xirr)
+{
+	struct kvmppc_xics *xics = vcpu->kvm->arch.xics;
+	struct kvmppc_icp *icp = vcpu->arch.icp;
+	u32 irq = xirr & 0x00ffffff;
 
 	XICS_DBG("h_eoi vcpu %d eoi %#lx\n", vcpu->vcpu_id, xirr);
 
@@ -794,26 +847,8 @@ static noinline int kvmppc_h_eoi(struct kvm_vcpu *vcpu, unsigned long xirr)
 	/* IPIs have no EOI */
 	if (irq == XICS_IPI)
 		return H_SUCCESS;
-	/*
-	 * EOI handling: If the interrupt is still asserted, we need to
-	 * resend it. We can take a lockless "peek" at the ICS state here.
-	 *
-	 * "Message" interrupts will never have "asserted" set
-	 */
-	ics = kvmppc_xics_find_ics(xics, irq, &src);
-	if (!ics) {
-		XICS_DBG("h_eoi: IRQ 0x%06x not found !\n", irq);
-		return H_PARAMETER;
-	}
-	state = &ics->irq_state[src];
 
-	/* Still asserted, resend it */
-	if (state->asserted)
-		icp_deliver_irq(xics, icp, irq);
-
-	kvm_notify_acked_irq(vcpu->kvm, 0, irq);
-
-	return H_SUCCESS;
+	return ics_eoi(vcpu, irq);
 }
 
 int kvmppc_xics_rm_complete(struct kvm_vcpu *vcpu, u32 hcall)
@@ -832,10 +867,6 @@ int kvmppc_xics_rm_complete(struct kvm_vcpu *vcpu, u32 hcall)
 		icp->n_rm_check_resend++;
 		icp_check_resend(xics, icp->rm_resend_icp);
 	}
-	if (icp->rm_action & XICS_RM_REJECT) {
-		icp->n_rm_reject++;
-		icp_deliver_irq(xics, icp, icp->rm_reject);
-	}
 	if (icp->rm_action & XICS_RM_NOTIFY_EOI) {
 		icp->n_rm_notify_eoi++;
 		kvm_notify_acked_irq(vcpu->kvm, 0, icp->rm_eoied_irq);
@@ -920,7 +951,7 @@ static int xics_debug_show(struct seq_file *m, void *private)
 	int icsid, i;
 	unsigned long flags;
 	unsigned long t_rm_kick_vcpu, t_rm_check_resend;
-	unsigned long t_rm_reject, t_rm_notify_eoi;
+	unsigned long t_rm_notify_eoi;
 	unsigned long t_reject, t_check_resend;
 
 	if (!kvm)
@@ -929,7 +960,6 @@ static int xics_debug_show(struct seq_file *m, void *private)
 	t_rm_kick_vcpu = 0;
 	t_rm_notify_eoi = 0;
 	t_rm_check_resend = 0;
-	t_rm_reject = 0;
 	t_check_resend = 0;
 	t_reject = 0;
 
@@ -952,14 +982,13 @@ static int xics_debug_show(struct seq_file *m, void *private)
 		t_rm_kick_vcpu += icp->n_rm_kick_vcpu;
 		t_rm_notify_eoi += icp->n_rm_notify_eoi;
 		t_rm_check_resend += icp->n_rm_check_resend;
-		t_rm_reject += icp->n_rm_reject;
 		t_check_resend += icp->n_check_resend;
 		t_reject += icp->n_reject;
 	}
 
-	seq_printf(m, "ICP Guest->Host totals: kick_vcpu=%lu check_resend=%lu reject=%lu notify_eoi=%lu\n",
+	seq_printf(m, "ICP Guest->Host totals: kick_vcpu=%lu check_resend=%lu notify_eoi=%lu\n",
 			t_rm_kick_vcpu, t_rm_check_resend,
-			t_rm_reject, t_rm_notify_eoi);
+			t_rm_notify_eoi);
 	seq_printf(m, "ICP Real Mode totals: check_resend=%lu resend=%lu\n",
 			t_check_resend, t_reject);
 	for (icsid = 0; icsid <= KVMPPC_XICS_MAX_ICS_ID; icsid++) {
@@ -977,9 +1006,9 @@ static int xics_debug_show(struct seq_file *m, void *private)
 		for (i = 0; i < KVMPPC_XICS_IRQ_PER_ICS; i++) {
 			struct ics_irq_state *irq = &ics->irq_state[i];
 
-			seq_printf(m, "irq 0x%06x: server %#x prio %#x save prio %#x asserted %d resend %d masked pending %d\n",
+			seq_printf(m, "irq 0x%06x: server %#x prio %#x save prio %#x pq_state %d resend %d masked pending %d\n",
 				   irq->number, irq->server, irq->priority,
-				   irq->saved_priority, irq->asserted,
+				   irq->saved_priority, irq->pq_state,
 				   irq->resend, irq->masked_pending);
 
 		}
@@ -1198,10 +1227,17 @@ static int xics_get_source(struct kvmppc_xics *xics, long irq, u64 addr)
 		val |= prio << KVM_XICS_PRIORITY_SHIFT;
 		if (irqp->lsi) {
 			val |= KVM_XICS_LEVEL_SENSITIVE;
-			if (irqp->asserted)
+			if (irqp->pq_state & PQ_PRESENTED)
 				val |= KVM_XICS_PENDING;
 		} else if (irqp->masked_pending || irqp->resend)
 			val |= KVM_XICS_PENDING;
+
+		if (irqp->pq_state & PQ_PRESENTED)
+			val |= KVM_XICS_PRESENTED;
+
+		if (irqp->pq_state & PQ_QUEUED)
+			val |= KVM_XICS_QUEUED;
+
 		ret = 0;
 	}
 	arch_spin_unlock(&ics->lock);
@@ -1253,18 +1289,20 @@ static int xics_set_source(struct kvmppc_xics *xics, long irq, u64 addr)
 	irqp->resend = 0;
 	irqp->masked_pending = 0;
 	irqp->lsi = 0;
-	irqp->asserted = 0;
-	if (val & KVM_XICS_LEVEL_SENSITIVE) {
+	irqp->pq_state = 0;
+	if (val & KVM_XICS_LEVEL_SENSITIVE)
 		irqp->lsi = 1;
-		if (val & KVM_XICS_PENDING)
-			irqp->asserted = 1;
-	}
+	/* If PENDING, set P in case P is not saved because of old code */
+	if (val & KVM_XICS_PRESENTED || val & KVM_XICS_PENDING)
+		irqp->pq_state |= PQ_PRESENTED;
+	if (val & KVM_XICS_QUEUED)
+		irqp->pq_state |= PQ_QUEUED;
 	irqp->exists = 1;
 	arch_spin_unlock(&ics->lock);
 	local_irq_restore(flags);
 
 	if (val & KVM_XICS_PENDING)
-		icp_deliver_irq(xics, NULL, irqp->number);
+		icp_deliver_irq(xics, NULL, irqp->number, false);
 
 	return 0;
 }

arch/powerpc/kvm/book3s_xics.h

@@ -31,16 +31,19 @@
 /* Priority value to use for disabling an interrupt */
 #define MASKED	0xff
 
+#define PQ_PRESENTED	1
+#define PQ_QUEUED	2
+
 /* State for one irq source */
 struct ics_irq_state {
 	u32 number;
 	u32 server;
+	u32 pq_state;
 	u8  priority;
 	u8  saved_priority;
 	u8  resend;
 	u8  masked_pending;
 	u8  lsi;		/* level-sensitive interrupt */
-	u8  asserted; /* Only for LSI */
 	u8  exists;
 	int intr_cpu;
 	u32 host_irq;
@@ -73,7 +76,6 @@ struct kvmppc_icp {
 	 */
 #define XICS_RM_KICK_VCPU	0x1
 #define XICS_RM_CHECK_RESEND	0x2
-#define XICS_RM_REJECT		0x4
 #define XICS_RM_NOTIFY_EOI	0x8
 	u32 rm_action;
 	struct kvm_vcpu *rm_kick_target;
@@ -84,7 +86,6 @@ struct kvmppc_icp {
 	/* Counters for each reason we exited real mode */
 	unsigned long n_rm_kick_vcpu;
 	unsigned long n_rm_check_resend;
-	unsigned long n_rm_reject;
 	unsigned long n_rm_notify_eoi;
 	/* Counters for handling ICP processing in real mode */
 	unsigned long n_check_resend;

arch/powerpc/kvm/powerpc.c

@@ -565,6 +565,13 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 	case KVM_CAP_PPC_HWRNG:
 		r = kvmppc_hwrng_present();
 		break;
+	case KVM_CAP_PPC_MMU_RADIX:
+		r = !!(hv_enabled && radix_enabled());
+		break;
+	case KVM_CAP_PPC_MMU_HASH_V3:
+		r = !!(hv_enabled && !radix_enabled() &&
+		       cpu_has_feature(CPU_FTR_ARCH_300));
+		break;
 #endif
 	case KVM_CAP_SYNC_MMU:
 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
@@ -605,6 +612,9 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 	case KVM_CAP_SPAPR_MULTITCE:
 		r = 1;
 		break;
+	case KVM_CAP_SPAPR_RESIZE_HPT:
+		r = !!hv_enabled;
+		break;
 #endif
 	case KVM_CAP_PPC_HTM:
 		r = cpu_has_feature(CPU_FTR_TM_COMP) &&
@@ -1468,6 +1478,31 @@ long kvm_arch_vm_ioctl(struct file *filp,
 		r = kvm_vm_ioctl_rtas_define_token(kvm, argp);
 		break;
 	}
+	case KVM_PPC_CONFIGURE_V3_MMU: {
+		struct kvm *kvm = filp->private_data;
+		struct kvm_ppc_mmuv3_cfg cfg;
+
+		r = -EINVAL;
+		if (!kvm->arch.kvm_ops->configure_mmu)
+			goto out;
+		r = -EFAULT;
+		if (copy_from_user(&cfg, argp, sizeof(cfg)))
+			goto out;
+		r = kvm->arch.kvm_ops->configure_mmu(kvm, &cfg);
+		break;
+	}
+	case KVM_PPC_GET_RMMU_INFO: {
+		struct kvm *kvm = filp->private_data;
+		struct kvm_ppc_rmmu_info info;
+
+		r = -EINVAL;
+		if (!kvm->arch.kvm_ops->get_rmmu_info)
+			goto out;
+		r = kvm->arch.kvm_ops->get_rmmu_info(kvm, &info);
+		if (r >= 0 && copy_to_user(argp, &info, sizeof(info)))
+			r = -EFAULT;
+		break;
+	}
 	default: {
 		struct kvm *kvm = filp->private_data;
 		r = kvm->arch.kvm_ops->arch_vm_ioctl(filp, ioctl, arg);

arch/powerpc/mm/init-common.c

@@ -41,6 +41,7 @@ static void pmd_ctor(void *addr)
 }
 
 struct kmem_cache *pgtable_cache[MAX_PGTABLE_INDEX_SIZE];
+EXPORT_SYMBOL_GPL(pgtable_cache);	/* used by kvm_hv module */
 
 /*
  * Create a kmem_cache() for pagetables.  This is not used for PTE
@@ -82,7 +83,7 @@ void pgtable_cache_add(unsigned shift, void (*ctor)(void *))
 	pgtable_cache[shift - 1] = new;
 	pr_debug("Allocated pgtable cache for order %d\n", shift);
 }
-
+EXPORT_SYMBOL_GPL(pgtable_cache_add);	/* used by kvm_hv module */
 
 void pgtable_cache_init(void)
 {

arch/powerpc/mm/init_64.c

@@ -42,6 +42,8 @@
 #include <linux/memblock.h>
 #include <linux/hugetlb.h>
 #include <linux/slab.h>
+#include <linux/of_fdt.h>
+#include <linux/libfdt.h>
 
 #include <asm/pgalloc.h>
 #include <asm/page.h>
@@ -344,12 +346,45 @@ static int __init parse_disable_radix(char *p)
 }
 early_param("disable_radix", parse_disable_radix);
 
+/*
+ * If we're running under a hypervisor, we need to check the contents of
+ * /chosen/ibm,architecture-vec-5 to see if the hypervisor is willing to do
+ * radix.  If not, we clear the radix feature bit so we fall back to hash.
+ */
+static void early_check_vec5(void)
+{
+	unsigned long root, chosen;
+	int size;
+	const u8 *vec5;
+
+	root = of_get_flat_dt_root();
+	chosen = of_get_flat_dt_subnode_by_name(root, "chosen");
+	if (chosen == -FDT_ERR_NOTFOUND)
+		return;
+	vec5 = of_get_flat_dt_prop(chosen, "ibm,architecture-vec-5", &size);
+	if (!vec5)
+		return;
+	if (size <= OV5_INDX(OV5_MMU_RADIX_300) ||
+	    !(vec5[OV5_INDX(OV5_MMU_RADIX_300)] & OV5_FEAT(OV5_MMU_RADIX_300)))
+		/* Hypervisor doesn't support radix */
+		cur_cpu_spec->mmu_features &= ~MMU_FTR_TYPE_RADIX;
+}
+
 void __init mmu_early_init_devtree(void)
 {
 	/* Disable radix mode based on kernel command line. */
 	if (disable_radix)
 		cur_cpu_spec->mmu_features &= ~MMU_FTR_TYPE_RADIX;
 
+	/*
+	 * Check /chosen/ibm,architecture-vec-5 if running as a guest.
+	 * When running bare-metal, we can use radix if we like
+	 * even though the ibm,architecture-vec-5 property created by
+	 * skiboot doesn't have the necessary bits set.
+	 */
+	if (early_radix_enabled() && !(mfmsr() & MSR_HV))
+		early_check_vec5();
+
 	if (early_radix_enabled())
 		radix__early_init_devtree();
 	else

arch/powerpc/mm/pgtable-radix.c

@@ -401,6 +401,8 @@ void __init radix__early_init_mmu(void)
 		mtspr(SPRN_LPCR, lpcr | LPCR_UPRT | LPCR_HR);
 		radix_init_partition_table();
 		radix_init_amor();
+	} else {
+		radix_init_pseries();
 	}
 
 	memblock_set_current_limit(MEMBLOCK_ALLOC_ANYWHERE);

arch/powerpc/mm/pgtable_64.c

@@ -454,13 +454,23 @@ void __init mmu_partition_table_init(void)
 void mmu_partition_table_set_entry(unsigned int lpid, unsigned long dw0,
 				   unsigned long dw1)
 {
+	unsigned long old = be64_to_cpu(partition_tb[lpid].patb0);
+
 	partition_tb[lpid].patb0 = cpu_to_be64(dw0);
 	partition_tb[lpid].patb1 = cpu_to_be64(dw1);
 
-	/* Global flush of TLBs and partition table caches for this lpid */
+	/*
+	 * Global flush of TLBs and partition table caches for this lpid.
+	 * The type of flush (hash or radix) depends on what the previous
+	 * use of this partition ID was, not the new use.
+	 */
 	asm volatile("ptesync" : : : "memory");
-	asm volatile(PPC_TLBIE_5(%0,%1,2,0,0) : :
-		     "r" (TLBIEL_INVAL_SET_LPID), "r" (lpid));
+	if (old & PATB_HR)
+		asm volatile(PPC_TLBIE_5(%0,%1,2,0,1) : :
+			     "r" (TLBIEL_INVAL_SET_LPID), "r" (lpid));
+	else
+		asm volatile(PPC_TLBIE_5(%0,%1,2,0,0) : :
+			     "r" (TLBIEL_INVAL_SET_LPID), "r" (lpid));
 	asm volatile("eieio; tlbsync; ptesync" : : : "memory");
 }
 EXPORT_SYMBOL_GPL(mmu_partition_table_set_entry);

arch/powerpc/platforms/pseries/firmware.c

@@ -126,7 +126,7 @@ static void __init fw_vec5_feature_init(const char *vec5, unsigned long len)
 		index = OV5_INDX(vec5_fw_features_table[i].feature);
 		feat = OV5_FEAT(vec5_fw_features_table[i].feature);
 
-		if (vec5[index] & feat)
+		if (index < len && (vec5[index] & feat))
 			powerpc_firmware_features |=
 				vec5_fw_features_table[i].val;
 	}

arch/powerpc/platforms/pseries/lpar.c

@@ -609,6 +609,29 @@ static int __init disable_bulk_remove(char *str)
 
 __setup("bulk_remove=", disable_bulk_remove);
 
+/* Actually only used for radix, so far */
+static int pseries_lpar_register_process_table(unsigned long base,
+			unsigned long page_size, unsigned long table_size)
+{
+	long rc;
+	unsigned long flags = PROC_TABLE_NEW;
+
+	if (radix_enabled())
+		flags |= PROC_TABLE_RADIX | PROC_TABLE_GTSE;
+	for (;;) {
+		rc = plpar_hcall_norets(H_REGISTER_PROC_TBL, flags, base,
+					page_size, table_size);
+		if (!H_IS_LONG_BUSY(rc))
+			break;
+		mdelay(get_longbusy_msecs(rc));
+	}
+	if (rc != H_SUCCESS) {
+		pr_err("Failed to register process table (rc=%ld)\n", rc);
+		BUG();
+	}
+	return rc;
+}
+
 void __init hpte_init_pseries(void)
 {
 	mmu_hash_ops.hpte_invalidate	 = pSeries_lpar_hpte_invalidate;
@@ -622,6 +645,12 @@ void __init hpte_init_pseries(void)
 	mmu_hash_ops.hugepage_invalidate = pSeries_lpar_hugepage_invalidate;
 }
 
+void radix_init_pseries(void)
+{
+	pr_info("Using radix MMU under hypervisor\n");
+	register_process_table = pseries_lpar_register_process_table;
+}
+
 #ifdef CONFIG_PPC_SMLPAR
 #define CMO_FREE_HINT_DEFAULT 1
 static int cmo_free_hint_flag = CMO_FREE_HINT_DEFAULT;

include/uapi/linux/kvm.h

@@ -685,6 +685,13 @@ struct kvm_ppc_smmu_info {
 	struct kvm_ppc_one_seg_page_size sps[KVM_PPC_PAGE_SIZES_MAX_SZ];
 };
 
+/* for KVM_PPC_RESIZE_HPT_{PREPARE,COMMIT} */
+struct kvm_ppc_resize_hpt {
+	__u64 flags;
+	__u32 shift;
+	__u32 pad;
+};
+
 #define KVMIO 0xAE
 
 /* machine type bits, to be used as argument to KVM_CREATE_VM */
@@ -871,6 +878,9 @@ struct kvm_ppc_smmu_info {
 #define KVM_CAP_S390_USER_INSTR0 130
 #define KVM_CAP_MSI_DEVID 131
 #define KVM_CAP_PPC_HTM 132
+#define KVM_CAP_SPAPR_RESIZE_HPT 133
+#define KVM_CAP_PPC_MMU_RADIX 134
+#define KVM_CAP_PPC_MMU_HASH_V3 135
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
@@ -1187,6 +1197,13 @@ struct kvm_s390_ucas_mapping {
 #define KVM_ARM_SET_DEVICE_ADDR	  _IOW(KVMIO,  0xab, struct kvm_arm_device_addr)
 /* Available with KVM_CAP_PPC_RTAS */
 #define KVM_PPC_RTAS_DEFINE_TOKEN _IOW(KVMIO,  0xac, struct kvm_rtas_token_args)
+/* Available with KVM_CAP_SPAPR_RESIZE_HPT */
+#define KVM_PPC_RESIZE_HPT_PREPARE _IOR(KVMIO, 0xad, struct kvm_ppc_resize_hpt)
+#define KVM_PPC_RESIZE_HPT_COMMIT  _IOR(KVMIO, 0xae, struct kvm_ppc_resize_hpt)
+/* Available with KVM_CAP_PPC_RADIX_MMU or KVM_CAP_PPC_HASH_MMU_V3 */
+#define KVM_PPC_CONFIGURE_V3_MMU  _IOW(KVMIO,  0xaf, struct kvm_ppc_mmuv3_cfg)
+/* Available with KVM_CAP_PPC_RADIX_MMU */
+#define KVM_PPC_GET_RMMU_INFO	  _IOW(KVMIO,  0xb0, struct kvm_ppc_rmmu_info)
 
 /* ioctl for vm fd */
 #define KVM_CREATE_DEVICE	  _IOWR(KVMIO,  0xe0, struct kvm_create_device)