KVM: PPC: Book3S HV: Don't rely on host's page size information

This removes the dependence of KVM on the mmu_psize_defs array (which
stores information about hardware support for various page sizes) and
the things derived from it, chiefly hpte_page_sizes[], hpte_page_size(),
hpte_actual_page_size() and get_sllp_encoding().  We also no longer
rely on the mmu_slb_size variable or the MMU_FTR_1T_SEGMENTS feature
bit.

The reason for doing this is so we can support a HPT guest on a radix
host.  In a radix host, the mmu_psize_defs array contains information
about page sizes supported by the MMU in radix mode rather than the
page sizes supported by the MMU in HPT mode.  Similarly, mmu_slb_size
and the MMU_FTR_1T_SEGMENTS bit are not set.

Instead we hard-code knowledge of the behaviour of the HPT MMU in the
POWER7, POWER8 and POWER9 processors (which are the only processors
supported by HV KVM) - specifically the encoding of the LP fields in
the HPT and SLB entries, and the fact that they have 32 SLB entries
and support 1TB segments.
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>

arch/powerpc/include/asm/kvm_book3s_64.h

@@ -107,18 +107,96 @@ static inline void __unlock_hpte(__be64 *hpte, unsigned long hpte_v)
 	hpte[0] = cpu_to_be64(hpte_v);
 }
 
+/*
+ * These functions encode knowledge of the POWER7/8/9 hardware
+ * interpretations of the HPTE LP (large page size) field.
+ */
+static inline int kvmppc_hpte_page_shifts(unsigned long h, unsigned long l)
+{
+	unsigned int lphi;
+
+	if (!(h & HPTE_V_LARGE))
+		return 12;	/* 4kB */
+	lphi = (l >> 16) & 0xf;
+	switch ((l >> 12) & 0xf) {
+	case 0:
+		return !lphi ? 24 : -1;		/* 16MB */
+		break;
+	case 1:
+		return 16;			/* 64kB */
+		break;
+	case 3:
+		return !lphi ? 34 : -1;		/* 16GB */
+		break;
+	case 7:
+		return (16 << 8) + 12;		/* 64kB in 4kB */
+		break;
+	case 8:
+		if (!lphi)
+			return (24 << 8) + 16;	/* 16MB in 64kkB */
+		if (lphi == 3)
+			return (24 << 8) + 12;	/* 16MB in 4kB */
+		break;
+	}
+	return -1;
+}
+
+static inline int kvmppc_hpte_base_page_shift(unsigned long h, unsigned long l)
+{
+	return kvmppc_hpte_page_shifts(h, l) & 0xff;
+}
+
+static inline int kvmppc_hpte_actual_page_shift(unsigned long h, unsigned long l)
+{
+	int tmp = kvmppc_hpte_page_shifts(h, l);
+
+	if (tmp >= 0x100)
+		tmp >>= 8;
+	return tmp;
+}
+
+static inline unsigned long kvmppc_actual_pgsz(unsigned long v, unsigned long r)
+{
+	return 1ul << kvmppc_hpte_actual_page_shift(v, r);
+}
+
+static inline int kvmppc_pgsize_lp_encoding(int base_shift, int actual_shift)
+{
+	switch (base_shift) {
+	case 12:
+		switch (actual_shift) {
+		case 12:
+			return 0;
+		case 16:
+			return 7;
+		case 24:
+			return 0x38;
+		}
+		break;
+	case 16:
+		switch (actual_shift) {
+		case 16:
+			return 1;
+		case 24:
+			return 8;
+		}
+		break;
+	case 24:
+		return 0;
+	}
+	return -1;
+}
+
 static inline unsigned long compute_tlbie_rb(unsigned long v, unsigned long r,
 					     unsigned long pte_index)
 {
-	int i, b_psize = MMU_PAGE_4K, a_psize = MMU_PAGE_4K;
-	unsigned int penc;
+	int a_pgshift, b_pgshift;
 	unsigned long rb = 0, va_low, sllp;
-	unsigned int lp = (r >> LP_SHIFT) & ((1 << LP_BITS) - 1);
 
-	if (v & HPTE_V_LARGE) {
-		i = hpte_page_sizes[lp];
-		b_psize = i & 0xf;
-		a_psize = i >> 4;
+	b_pgshift = a_pgshift = kvmppc_hpte_page_shifts(v, r);
+	if (a_pgshift >= 0x100) {
+		b_pgshift &= 0xff;
+		a_pgshift >>= 8;
 	}
 
 	/*
@@ -152,37 +230,33 @@ static inline unsigned long compute_tlbie_rb(unsigned long v, unsigned long r,
 		va_low ^= v >> (SID_SHIFT_1T - 16);
 	va_low &= 0x7ff;
 
-	switch (b_psize) {
-	case MMU_PAGE_4K:
-		sllp = get_sllp_encoding(a_psize);
+	if (b_pgshift == 12) {
+		if (a_pgshift > 12) {
+			sllp = (a_pgshift == 16) ? 5 : 4;
 			rb |= sllp << 5;	/*  AP field */
+		}
 		rb |= (va_low & 0x7ff) << 12;	/* remaining 11 bits of AVA */
-		break;
-	default:
-	{
+	} else {
 		int aval_shift;
 		/*
 		 * remaining bits of AVA/LP fields
 		 * Also contain the rr bits of LP
 		 */
-		rb |= (va_low << mmu_psize_defs[b_psize].shift) & 0x7ff000;
+		rb |= (va_low << b_pgshift) & 0x7ff000;
 		/*
 		 * Now clear not needed LP bits based on actual psize
 		 */
-		rb &= ~((1ul << mmu_psize_defs[a_psize].shift) - 1);
+		rb &= ~((1ul << a_pgshift) - 1);
 		/*
 		 * AVAL field 58..77 - base_page_shift bits of va
 		 * we have space for 58..64 bits, Missing bits should
 		 * be zero filled. +1 is to take care of L bit shift
 		 */
-		aval_shift = 64 - (77 - mmu_psize_defs[b_psize].shift) + 1;
+		aval_shift = 64 - (77 - b_pgshift) + 1;
 		rb |= ((va_low << aval_shift) & 0xfe);
 
 		rb |= 1;		/* L field */
-		penc = mmu_psize_defs[b_psize].penc[a_psize];
-		rb |= penc << 12;	/* LP field */
-		break;
-	}
+		rb |= r & 0xff000 & ((1ul << a_pgshift) - 1); /* LP field */
 	}
 	rb |= (v >> HPTE_V_SSIZE_SHIFT) << 8;	/* B field */
 	return rb;

arch/powerpc/kvm/book3s_64_mmu_hv.c

@@ -333,7 +333,7 @@ static unsigned long kvmppc_mmu_get_real_addr(unsigned long v, unsigned long r,
 {
 	unsigned long ra_mask;
 
-	ra_mask = hpte_page_size(v, r) - 1;
+	ra_mask = kvmppc_actual_pgsz(v, r) - 1;
 	return (r & HPTE_R_RPN & ~ra_mask) | (ea & ra_mask);
 }
 
@@ -504,7 +504,8 @@ int kvmppc_book3s_hv_page_fault(struct kvm_run *run, struct kvm_vcpu *vcpu,
 		mmio_update = atomic64_read(&kvm->arch.mmio_update);
 		if (mmio_update == vcpu->arch.pgfault_cache->mmio_update) {
 			r = vcpu->arch.pgfault_cache->rpte;
-			psize = hpte_page_size(vcpu->arch.pgfault_hpte[0], r);
+			psize = kvmppc_actual_pgsz(vcpu->arch.pgfault_hpte[0],
+						   r);
 			gpa_base = r & HPTE_R_RPN & ~(psize - 1);
 			gfn_base = gpa_base >> PAGE_SHIFT;
 			gpa = gpa_base | (ea & (psize - 1));
@@ -533,7 +534,7 @@ int kvmppc_book3s_hv_page_fault(struct kvm_run *run, struct kvm_vcpu *vcpu,
 		return RESUME_GUEST;
 
 	/* Translate the logical address and get the page */
-	psize = hpte_page_size(hpte[0], r);
+	psize = kvmppc_actual_pgsz(hpte[0], r);
 	gpa_base = r & HPTE_R_RPN & ~(psize - 1);
 	gfn_base = gpa_base >> PAGE_SHIFT;
 	gpa = gpa_base | (ea & (psize - 1));
@@ -797,7 +798,7 @@ static void kvmppc_unmap_hpte(struct kvm *kvm, unsigned long i,
 
 	/* Now check and modify the HPTE */
 	ptel = rev[i].guest_rpte;
-	psize = hpte_page_size(be64_to_cpu(hptep[0]), ptel);
+	psize = kvmppc_actual_pgsz(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);
@@ -1091,7 +1092,7 @@ static int kvm_test_clear_dirty_npages(struct kvm *kvm, unsigned long *rmapp)
 				rev[i].guest_rpte |= HPTE_R_C;
 				note_hpte_modification(kvm, &rev[i]);
 			}
-			n = hpte_page_size(v, r);
+			n = kvmppc_actual_pgsz(v, r);
 			n = (n + PAGE_SIZE - 1) >> PAGE_SHIFT;
 			if (n > npages_dirty)
 				npages_dirty = n;
@@ -1266,7 +1267,7 @@ static unsigned long resize_hpt_rehash_hpte(struct kvm_resize_hpt *resize,
 	guest_rpte = rev->guest_rpte;
 
 	ret = -EIO;
-	apsize = hpte_page_size(vpte, guest_rpte);
+	apsize = kvmppc_actual_pgsz(vpte, guest_rpte);
 	if (!apsize)
 		goto out;
 

arch/powerpc/kvm/book3s_hv.c

@@ -3300,22 +3300,21 @@ static int kvmppc_vcpu_run_hv(struct kvm_run *run, struct kvm_vcpu *vcpu)
 }
 
 static void kvmppc_add_seg_page_size(struct kvm_ppc_one_seg_page_size **sps,
-				     int linux_psize)
+				     int shift, int sllp)
 {
-	struct mmu_psize_def *def = &mmu_psize_defs[linux_psize];
-
-	if (!def->shift)
-		return;
-	(*sps)->page_shift = def->shift;
-	(*sps)->slb_enc = def->sllp;
-	(*sps)->enc[0].page_shift = def->shift;
-	(*sps)->enc[0].pte_enc = def->penc[linux_psize];
+	(*sps)->page_shift = shift;
+	(*sps)->slb_enc = sllp;
+	(*sps)->enc[0].page_shift = shift;
+	(*sps)->enc[0].pte_enc = kvmppc_pgsize_lp_encoding(shift, shift);
 	/*
-	 * Add 16MB MPSS support if host supports it
+	 * Add 16MB MPSS support (may get filtered out by userspace)
 	 */
-	if (linux_psize != MMU_PAGE_16M && def->penc[MMU_PAGE_16M] != -1) {
+	if (shift != 24) {
+		int penc = kvmppc_pgsize_lp_encoding(shift, 24);
+		if (penc != -1) {
 			(*sps)->enc[1].page_shift = 24;
-		(*sps)->enc[1].pte_enc = def->penc[MMU_PAGE_16M];
+			(*sps)->enc[1].pte_enc = penc;
+		}
 	}
 	(*sps)++;
 }
@@ -3340,16 +3339,15 @@ static int kvm_vm_ioctl_get_smmu_info_hv(struct kvm *kvm,
 	info->data_keys = 32;
 	info->instr_keys = cpu_has_feature(CPU_FTR_ARCH_207S) ? 32 : 0;
 
-	info->flags = KVM_PPC_PAGE_SIZES_REAL;
-	if (mmu_has_feature(MMU_FTR_1T_SEGMENT))
-		info->flags |= KVM_PPC_1T_SEGMENTS;
-	info->slb_size = mmu_slb_size;
+	/* POWER7, 8 and 9 all have 1T segments and 32-entry SLB */
+	info->flags = KVM_PPC_PAGE_SIZES_REAL | KVM_PPC_1T_SEGMENTS;
+	info->slb_size = 32;
 
 	/* We only support these sizes for now, and no muti-size segments */
 	sps = &info->sps[0];
-	kvmppc_add_seg_page_size(&sps, MMU_PAGE_4K);
-	kvmppc_add_seg_page_size(&sps, MMU_PAGE_64K);
-	kvmppc_add_seg_page_size(&sps, MMU_PAGE_16M);
+	kvmppc_add_seg_page_size(&sps, 12, 0);
+	kvmppc_add_seg_page_size(&sps, 16, SLB_VSID_L | SLB_VSID_LP_01);
+	kvmppc_add_seg_page_size(&sps, 24, SLB_VSID_L);
 
 	return 0;
 }
@@ -4352,4 +4350,3 @@ module_exit(kvmppc_book3s_exit_hv);
 MODULE_LICENSE("GPL");
 MODULE_ALIAS_MISCDEV(KVM_MINOR);
 MODULE_ALIAS("devname:kvm");
-

arch/powerpc/kvm/book3s_hv_rm_mmu.c

@@ -129,7 +129,7 @@ static unsigned long *revmap_for_hpte(struct kvm *kvm, unsigned long hpte_v,
 	unsigned long *rmap;
 	unsigned long gfn;
 
-	gfn = hpte_rpn(hpte_gr, hpte_page_size(hpte_v, hpte_gr));
+	gfn = hpte_rpn(hpte_gr, kvmppc_actual_pgsz(hpte_v, hpte_gr));
 	memslot = __gfn_to_memslot(kvm_memslots_raw(kvm), gfn);
 	if (!memslot)
 		return NULL;
@@ -169,7 +169,8 @@ static void remove_revmap_chain(struct kvm *kvm, long pte_index,
 	}
 	*rmap |= rcbits << KVMPPC_RMAP_RC_SHIFT;
 	if (rcbits & HPTE_R_C)
-		kvmppc_update_rmap_change(rmap, hpte_page_size(hpte_v, hpte_r));
+		kvmppc_update_rmap_change(rmap,
+					  kvmppc_actual_pgsz(hpte_v, hpte_r));
 	unlock_rmap(rmap);
 }
 
@@ -193,7 +194,7 @@ long kvmppc_do_h_enter(struct kvm *kvm, unsigned long flags,
 
 	if (kvm_is_radix(kvm))
 		return H_FUNCTION;
-	psize = hpte_page_size(pteh, ptel);
+	psize = kvmppc_actual_pgsz(pteh, ptel);
 	if (!psize)
 		return H_PARAMETER;
 	writing = hpte_is_writable(ptel);
@@ -848,7 +849,7 @@ long kvmppc_h_clear_mod(struct kvm_vcpu *vcpu, unsigned long flags,
 		r = be64_to_cpu(hpte[1]);
 		gr |= r & (HPTE_R_R | HPTE_R_C);
 		if (r & HPTE_R_C) {
-			unsigned long psize = hpte_page_size(v, r);
+			unsigned long psize = kvmppc_actual_pgsz(v, r);
 			hpte[1] = cpu_to_be64(r & ~HPTE_R_C);
 			eieio();
 			rmap = revmap_for_hpte(kvm, v, gr);
@@ -1014,7 +1015,7 @@ long kvmppc_hv_find_lock_hpte(struct kvm *kvm, gva_t eaddr, unsigned long slb_v,
 			 * Check the HPTE again, including base page size
 			 */
 			if ((v & valid) && (v & mask) == val &&
-			    hpte_base_page_size(v, r) == (1ul << pshift))
+			    kvmppc_hpte_base_page_shift(v, r) == pshift)
 				/* Return with the HPTE still locked */
 				return (hash << 3) + (i >> 1);