KVM: PPC: Use get/set for to_svcpu to help preemption

When running the 64-bit Book3s PR code without CONFIG_PREEMPT_NONE, we were
doing a few things wrong, most notably access to PACA fields without making
sure that the pointers stay stable accross the access (preempt_disable()).

This patch moves to_svcpu towards a get/put model which allows us to disable
preemption while accessing the shadow vcpu fields in the PACA. That way we
can run preemptible and everyone's happy!

Reported-by: Jörg Sommer <joerg@alea.gnuu.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>

5 files changed, 110 insertions, 52 deletions

diff --git a/arch/powerpc/kvm/book3s_32_mmu_host.c b/arch/powerpc/kvm/book3s_32_mmu_host.c
index 9fecbfbce77..f922c29bb23 100644
--- a/arch/powerpc/kvm/book3s_32_mmu_host.c
+++ b/arch/powerpc/kvm/book3s_32_mmu_host.c
@@ -151,13 +151,15 @@ int kvmppc_mmu_map_page(struct kvm_vcpu *vcpu, struct kvmppc_pte *orig_pte)
 	bool primary = false;
 	bool evict = false;
 	struct hpte_cache *pte;
+	int r = 0;
 
 	/* Get host physical address for gpa */
 	hpaddr = kvmppc_gfn_to_pfn(vcpu, orig_pte->raddr >> PAGE_SHIFT);
 	if (is_error_pfn(hpaddr)) {
 		printk(KERN_INFO "Couldn't get guest page for gfn %lx!\n",
 				 orig_pte->eaddr);
-		return -EINVAL;
+		r = -EINVAL;
+		goto out;
 	}
 	hpaddr <<= PAGE_SHIFT;
 
@@ -249,7 +251,8 @@ next_pteg:
 
 	kvmppc_mmu_hpte_cache_map(vcpu, pte);
 
-	return 0;
+out:
+	return r;
 }
 
 static struct kvmppc_sid_map *create_sid_map(struct kvm_vcpu *vcpu, u64 gvsid)
@@ -297,12 +300,14 @@ int kvmppc_mmu_map_segment(struct kvm_vcpu *vcpu, ulong eaddr)
 	u64 gvsid;
 	u32 sr;
 	struct kvmppc_sid_map *map;
-	struct kvmppc_book3s_shadow_vcpu *svcpu = to_svcpu(vcpu);
+	struct kvmppc_book3s_shadow_vcpu *svcpu = svcpu_get(vcpu);
+	int r = 0;
 
 	if (vcpu->arch.mmu.esid_to_vsid(vcpu, esid, &gvsid)) {
 		/* Invalidate an entry */
 		svcpu->sr[esid] = SR_INVALID;
-		return -ENOENT;
+		r = -ENOENT;
+		goto out;
 	}
 
 	map = find_sid_vsid(vcpu, gvsid);
@@ -315,17 +320,21 @@ int kvmppc_mmu_map_segment(struct kvm_vcpu *vcpu, ulong eaddr)
 
 	dprintk_sr("MMU: mtsr %d, 0x%x\n", esid, sr);
 
-	return 0;
+out:
+	svcpu_put(svcpu);
+	return r;
 }
 
 void kvmppc_mmu_flush_segments(struct kvm_vcpu *vcpu)
 {
 	int i;
-	struct kvmppc_book3s_shadow_vcpu *svcpu = to_svcpu(vcpu);
+	struct kvmppc_book3s_shadow_vcpu *svcpu = svcpu_get(vcpu);
 
 	dprintk_sr("MMU: flushing all segments (%d)\n", ARRAY_SIZE(svcpu->sr));
 	for (i = 0; i < ARRAY_SIZE(svcpu->sr); i++)
 		svcpu->sr[i] = SR_INVALID;
+
+	svcpu_put(svcpu);
 }
 
 void kvmppc_mmu_destroy(struct kvm_vcpu *vcpu)
diff --git a/arch/powerpc/kvm/book3s_64_mmu_host.c b/arch/powerpc/kvm/book3s_64_mmu_host.c
index fa2f08434ba..6f87f39a1ac 100644
--- a/arch/powerpc/kvm/book3s_64_mmu_host.c
+++ b/arch/powerpc/kvm/book3s_64_mmu_host.c
@@ -88,12 +88,14 @@ int kvmppc_mmu_map_page(struct kvm_vcpu *vcpu, struct kvmppc_pte *orig_pte)
 	int vflags = 0;
 	int attempt = 0;
 	struct kvmppc_sid_map *map;
+	int r = 0;
 
 	/* Get host physical address for gpa */
 	hpaddr = kvmppc_gfn_to_pfn(vcpu, orig_pte->raddr >> PAGE_SHIFT);
 	if (is_error_pfn(hpaddr)) {
 		printk(KERN_INFO "Couldn't get guest page for gfn %lx!\n", orig_pte->eaddr);
-		return -EINVAL;
+		r = -EINVAL;
+		goto out;
 	}
 	hpaddr <<= PAGE_SHIFT;
 	hpaddr |= orig_pte->raddr & (~0xfffULL & ~PAGE_MASK);
@@ -110,7 +112,8 @@ int kvmppc_mmu_map_page(struct kvm_vcpu *vcpu, struct kvmppc_pte *orig_pte)
 		printk(KERN_ERR "KVM: Segment map for 0x%llx (0x%lx) failed\n",
 				vsid, orig_pte->eaddr);
 		WARN_ON(true);
-		return -EINVAL;
+		r = -EINVAL;
+		goto out;
 	}
 
 	vsid = map->host_vsid;
@@ -131,8 +134,10 @@ map_again:
 
 	/* In case we tried normal mapping already, let's nuke old entries */
 	if (attempt > 1)
-		if (ppc_md.hpte_remove(hpteg) < 0)
-			return -1;
+		if (ppc_md.hpte_remove(hpteg) < 0) {
+			r = -1;
+			goto out;
+		}
 
 	ret = ppc_md.hpte_insert(hpteg, va, hpaddr, rflags, vflags, MMU_PAGE_4K, MMU_SEGSIZE_256M);
 
@@ -162,7 +167,8 @@ map_again:
 		kvmppc_mmu_hpte_cache_map(vcpu, pte);
 	}
 
-	return 0;
+out:
+	return r;
 }
 
 static struct kvmppc_sid_map *create_sid_map(struct kvm_vcpu *vcpu, u64 gvsid)
@@ -207,25 +213,30 @@ static struct kvmppc_sid_map *create_sid_map(struct kvm_vcpu *vcpu, u64 gvsid)
 
 static int kvmppc_mmu_next_segment(struct kvm_vcpu *vcpu, ulong esid)
 {
+	struct kvmppc_book3s_shadow_vcpu *svcpu = svcpu_get(vcpu);
 	int i;
 	int max_slb_size = 64;
 	int found_inval = -1;
 	int r;
 
-	if (!to_svcpu(vcpu)->slb_max)
-		to_svcpu(vcpu)->slb_max = 1;
+	if (!svcpu->slb_max)
+		svcpu->slb_max = 1;
 
 	/* Are we overwriting? */
-	for (i = 1; i < to_svcpu(vcpu)->slb_max; i++) {
-		if (!(to_svcpu(vcpu)->slb[i].esid & SLB_ESID_V))
+	for (i = 1; i < svcpu->slb_max; i++) {
+		if (!(svcpu->slb[i].esid & SLB_ESID_V))
 			found_inval = i;
-		else if ((to_svcpu(vcpu)->slb[i].esid & ESID_MASK) == esid)
-			return i;
+		else if ((svcpu->slb[i].esid & ESID_MASK) == esid) {
+			r = i;
+			goto out;
+		}
 	}
 
 	/* Found a spare entry that was invalidated before */
-	if (found_inval > 0)
-		return found_inval;
+	if (found_inval > 0) {
+		r = found_inval;
+		goto out;
+	}
 
 	/* No spare invalid entry, so create one */
 
@@ -233,30 +244,35 @@ static int kvmppc_mmu_next_segment(struct kvm_vcpu *vcpu, ulong esid)
 		max_slb_size = mmu_slb_size;
 
 	/* Overflowing -> purge */
-	if ((to_svcpu(vcpu)->slb_max) == max_slb_size)
+	if ((svcpu->slb_max) == max_slb_size)
 		kvmppc_mmu_flush_segments(vcpu);
 
-	r = to_svcpu(vcpu)->slb_max;
-	to_svcpu(vcpu)->slb_max++;
+	r = svcpu->slb_max;
+	svcpu->slb_max++;
 
+out:
+	svcpu_put(svcpu);
 	return r;
 }
 
 int kvmppc_mmu_map_segment(struct kvm_vcpu *vcpu, ulong eaddr)
 {
+	struct kvmppc_book3s_shadow_vcpu *svcpu = svcpu_get(vcpu);
 	u64 esid = eaddr >> SID_SHIFT;
 	u64 slb_esid = (eaddr & ESID_MASK) | SLB_ESID_V;
 	u64 slb_vsid = SLB_VSID_USER;
 	u64 gvsid;
 	int slb_index;
 	struct kvmppc_sid_map *map;
+	int r = 0;
 
 	slb_index = kvmppc_mmu_next_segment(vcpu, eaddr & ESID_MASK);
 
 	if (vcpu->arch.mmu.esid_to_vsid(vcpu, esid, &gvsid)) {
 		/* Invalidate an entry */
-		to_svcpu(vcpu)->slb[slb_index].esid = 0;
-		return -ENOENT;
+		svcpu->slb[slb_index].esid = 0;
+		r = -ENOENT;
+		goto out;
 	}
 
 	map = find_sid_vsid(vcpu, gvsid);
@@ -269,18 +285,22 @@ int kvmppc_mmu_map_segment(struct kvm_vcpu *vcpu, ulong eaddr)
 	slb_vsid &= ~SLB_VSID_KP;
 	slb_esid |= slb_index;
 
-	to_svcpu(vcpu)->slb[slb_index].esid = slb_esid;
-	to_svcpu(vcpu)->slb[slb_index].vsid = slb_vsid;
+	svcpu->slb[slb_index].esid = slb_esid;
+	svcpu->slb[slb_index].vsid = slb_vsid;
 
 	trace_kvm_book3s_slbmte(slb_vsid, slb_esid);
 
-	return 0;
+out:
+	svcpu_put(svcpu);
+	return r;
 }
 
 void kvmppc_mmu_flush_segments(struct kvm_vcpu *vcpu)
 {
-	to_svcpu(vcpu)->slb_max = 1;
-	to_svcpu(vcpu)->slb[0].esid = 0;
+	struct kvmppc_book3s_shadow_vcpu *svcpu = svcpu_get(vcpu);
+	svcpu->slb_max = 1;
+	svcpu->slb[0].esid = 0;
+	svcpu_put(svcpu);
 }
 
 void kvmppc_mmu_destroy(struct kvm_vcpu *vcpu)
diff --git a/arch/powerpc/kvm/book3s_emulate.c b/arch/powerpc/kvm/book3s_emulate.c
index 0c9dc62532d..f1950d13182 100644
--- a/arch/powerpc/kvm/book3s_emulate.c
+++ b/arch/powerpc/kvm/book3s_emulate.c
@@ -230,9 +230,12 @@ int kvmppc_core_emulate_op(struct kvm_run *run, struct kvm_vcpu *vcpu,
 
 			r = kvmppc_st(vcpu, &addr, 32, zeros, true);
 			if ((r == -ENOENT) || (r == -EPERM)) {
+				struct kvmppc_book3s_shadow_vcpu *svcpu;
+
+				svcpu = svcpu_get(vcpu);
 				*advance = 0;
 				vcpu->arch.shared->dar = vaddr;
-				to_svcpu(vcpu)->fault_dar = vaddr;
+				svcpu->fault_dar = vaddr;
 
 				dsisr = DSISR_ISSTORE;
 				if (r == -ENOENT)
@@ -241,7 +244,8 @@ int kvmppc_core_emulate_op(struct kvm_run *run, struct kvm_vcpu *vcpu,
 					dsisr |= DSISR_PROTFAULT;
 
 				vcpu->arch.shared->dsisr = dsisr;
-				to_svcpu(vcpu)->fault_dsisr = dsisr;
+				svcpu->fault_dsisr = dsisr;
+				svcpu_put(svcpu);
 
 				kvmppc_book3s_queue_irqprio(vcpu,
 					BOOK3S_INTERRUPT_DATA_STORAGE);
diff --git a/arch/powerpc/kvm/book3s_pr.c b/arch/powerpc/kvm/book3s_pr.c
index 857ecde0cfd..0c31507be90 100644
--- a/arch/powerpc/kvm/book3s_pr.c
+++ b/arch/powerpc/kvm/book3s_pr.c
@@ -56,10 +56,12 @@ static int kvmppc_handle_ext(struct kvm_vcpu *vcpu, unsigned int exit_nr,
 void kvmppc_core_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 {
 #ifdef CONFIG_PPC_BOOK3S_64
-	memcpy(to_svcpu(vcpu)->slb, to_book3s(vcpu)->slb_shadow, sizeof(to_svcpu(vcpu)->slb));
+	struct kvmppc_book3s_shadow_vcpu *svcpu = svcpu_get(vcpu);
+	memcpy(svcpu->slb, to_book3s(vcpu)->slb_shadow, sizeof(svcpu->slb));
 	memcpy(&get_paca()->shadow_vcpu, to_book3s(vcpu)->shadow_vcpu,
 	       sizeof(get_paca()->shadow_vcpu));
-	to_svcpu(vcpu)->slb_max = to_book3s(vcpu)->slb_shadow_max;
+	svcpu->slb_max = to_book3s(vcpu)->slb_shadow_max;
+	svcpu_put(svcpu);
 #endif
 
 #ifdef CONFIG_PPC_BOOK3S_32
@@ -70,10 +72,12 @@ void kvmppc_core_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 void kvmppc_core_vcpu_put(struct kvm_vcpu *vcpu)
 {
 #ifdef CONFIG_PPC_BOOK3S_64
-	memcpy(to_book3s(vcpu)->slb_shadow, to_svcpu(vcpu)->slb, sizeof(to_svcpu(vcpu)->slb));
+	struct kvmppc_book3s_shadow_vcpu *svcpu = svcpu_get(vcpu);
+	memcpy(to_book3s(vcpu)->slb_shadow, svcpu->slb, sizeof(svcpu->slb));
 	memcpy(to_book3s(vcpu)->shadow_vcpu, &get_paca()->shadow_vcpu,
 	       sizeof(get_paca()->shadow_vcpu));
-	to_book3s(vcpu)->slb_shadow_max = to_svcpu(vcpu)->slb_max;
+	to_book3s(vcpu)->slb_shadow_max = svcpu->slb_max;
+	svcpu_put(svcpu);
 #endif
 
 	kvmppc_giveup_ext(vcpu, MSR_FP);
@@ -308,19 +312,22 @@ int kvmppc_handle_pagefault(struct kvm_run *run, struct kvm_vcpu *vcpu,
 
 	if (page_found == -ENOENT) {
 		/* Page not found in guest PTE entries */
+		struct kvmppc_book3s_shadow_vcpu *svcpu = svcpu_get(vcpu);
 		vcpu->arch.shared->dar = kvmppc_get_fault_dar(vcpu);
-		vcpu->arch.shared->dsisr = to_svcpu(vcpu)->fault_dsisr;
+		vcpu->arch.shared->dsisr = svcpu->fault_dsisr;
 		vcpu->arch.shared->msr |=
-			(to_svcpu(vcpu)->shadow_srr1 & 0x00000000f8000000ULL);
+			(svcpu->shadow_srr1 & 0x00000000f8000000ULL);
+		svcpu_put(svcpu);
 		kvmppc_book3s_queue_irqprio(vcpu, vec);
 	} else if (page_found == -EPERM) {
 		/* Storage protection */
+		struct kvmppc_book3s_shadow_vcpu *svcpu = svcpu_get(vcpu);
 		vcpu->arch.shared->dar = kvmppc_get_fault_dar(vcpu);
-		vcpu->arch.shared->dsisr =
-			to_svcpu(vcpu)->fault_dsisr & ~DSISR_NOHPTE;
+		vcpu->arch.shared->dsisr = svcpu->fault_dsisr & ~DSISR_NOHPTE;
 		vcpu->arch.shared->dsisr |= DSISR_PROTFAULT;
 		vcpu->arch.shared->msr |=
-			(to_svcpu(vcpu)->shadow_srr1 & 0x00000000f8000000ULL);
+			svcpu->shadow_srr1 & 0x00000000f8000000ULL;
+		svcpu_put(svcpu);
 		kvmppc_book3s_queue_irqprio(vcpu, vec);
 	} else if (page_found == -EINVAL) {
 		/* Page not found in guest SLB */
@@ -521,21 +528,25 @@ int kvmppc_handle_exit(struct kvm_run *run, struct kvm_vcpu *vcpu,
 	kvm_resched(vcpu);
 	switch (exit_nr) {
 	case BOOK3S_INTERRUPT_INST_STORAGE:
+	{
+		struct kvmppc_book3s_shadow_vcpu *svcpu = svcpu_get(vcpu);
+		ulong shadow_srr1 = svcpu->shadow_srr1;
 		vcpu->stat.pf_instruc++;
 
 #ifdef CONFIG_PPC_BOOK3S_32
 		/* We set segments as unused segments when invalidating them. So
 		 * treat the respective fault as segment fault. */
-		if (to_svcpu(vcpu)->sr[kvmppc_get_pc(vcpu) >> SID_SHIFT]
-		    == SR_INVALID) {
+		if (svcpu->sr[kvmppc_get_pc(vcpu) >> SID_SHIFT] == SR_INVALID) {
 			kvmppc_mmu_map_segment(vcpu, kvmppc_get_pc(vcpu));
 			r = RESUME_GUEST;
+			svcpu_put(svcpu);
 			break;
 		}
 #endif
+		svcpu_put(svcpu);
 
 		/* only care about PTEG not found errors, but leave NX alone */
-		if (to_svcpu(vcpu)->shadow_srr1 & 0x40000000) {
+		if (shadow_srr1 & 0x40000000) {
 			r = kvmppc_handle_pagefault(run, vcpu, kvmppc_get_pc(vcpu), exit_nr);
 			vcpu->stat.sp_instruc++;
 		} else if (vcpu->arch.mmu.is_dcbz32(vcpu) &&
@@ -548,33 +559,37 @@ int kvmppc_handle_exit(struct kvm_run *run, struct kvm_vcpu *vcpu,
 			kvmppc_mmu_pte_flush(vcpu, kvmppc_get_pc(vcpu), ~0xFFFUL);
 			r = RESUME_GUEST;
 		} else {
-			vcpu->arch.shared->msr |=
-				to_svcpu(vcpu)->shadow_srr1 & 0x58000000;
+			vcpu->arch.shared->msr |= shadow_srr1 & 0x58000000;
 			kvmppc_book3s_queue_irqprio(vcpu, exit_nr);
 			r = RESUME_GUEST;
 		}
 		break;
+	}
 	case BOOK3S_INTERRUPT_DATA_STORAGE:
 	{
 		ulong dar = kvmppc_get_fault_dar(vcpu);
+		struct kvmppc_book3s_shadow_vcpu *svcpu = svcpu_get(vcpu);
+		u32 fault_dsisr = svcpu->fault_dsisr;
 		vcpu->stat.pf_storage++;
 
 #ifdef CONFIG_PPC_BOOK3S_32
 		/* We set segments as unused segments when invalidating them. So
 		 * treat the respective fault as segment fault. */
-		if ((to_svcpu(vcpu)->sr[dar >> SID_SHIFT]) == SR_INVALID) {
+		if ((svcpu->sr[dar >> SID_SHIFT]) == SR_INVALID) {
 			kvmppc_mmu_map_segment(vcpu, dar);
 			r = RESUME_GUEST;
+			svcpu_put(svcpu);
 			break;
 		}
 #endif
+		svcpu_put(svcpu);
 
 		/* The only case we need to handle is missing shadow PTEs */
-		if (to_svcpu(vcpu)->fault_dsisr & DSISR_NOHPTE) {
+		if (fault_dsisr & DSISR_NOHPTE) {
 			r = kvmppc_handle_pagefault(run, vcpu, dar, exit_nr);
 		} else {
 			vcpu->arch.shared->dar = dar;
-			vcpu->arch.shared->dsisr = to_svcpu(vcpu)->fault_dsisr;
+			vcpu->arch.shared->dsisr = fault_dsisr;
 			kvmppc_book3s_queue_irqprio(vcpu, exit_nr);
 			r = RESUME_GUEST;
 		}
@@ -610,10 +625,13 @@ int kvmppc_handle_exit(struct kvm_run *run, struct kvm_vcpu *vcpu,
 	case BOOK3S_INTERRUPT_PROGRAM:
 	{
 		enum emulation_result er;
+		struct kvmppc_book3s_shadow_vcpu *svcpu;
 		ulong flags;
 
 program_interrupt:
-		flags = to_svcpu(vcpu)->shadow_srr1 & 0x1f0000ull;
+		svcpu = svcpu_get(vcpu);
+		flags = svcpu->shadow_srr1 & 0x1f0000ull;
+		svcpu_put(svcpu);
 
 		if (vcpu->arch.shared->msr & MSR_PR) {
 #ifdef EXIT_DEBUG
@@ -741,14 +759,18 @@ program_interrupt:
 		r = RESUME_GUEST;
 		break;
 	default:
+	{
+		struct kvmppc_book3s_shadow_vcpu *svcpu = svcpu_get(vcpu);
+		ulong shadow_srr1 = svcpu->shadow_srr1;
+		svcpu_put(svcpu);
 		/* Ugh - bork here! What did we get? */
 		printk(KERN_EMERG "exit_nr=0x%x | pc=0x%lx | msr=0x%lx\n",
-			exit_nr, kvmppc_get_pc(vcpu), to_svcpu(vcpu)->shadow_srr1);
+			exit_nr, kvmppc_get_pc(vcpu), shadow_srr1);
 		r = RESUME_HOST;
 		BUG();
 		break;
 	}
-
+	}
 
 	if (!(r & RESUME_HOST)) {
 		/* To avoid clobbering exit_reason, only check for signals if
diff --git a/arch/powerpc/kvm/trace.h b/arch/powerpc/kvm/trace.h
index b135d3d397d..609d8bfb54e 100644
--- a/arch/powerpc/kvm/trace.h
+++ b/arch/powerpc/kvm/trace.h
@@ -118,11 +118,14 @@ TRACE_EVENT(kvm_book3s_exit,
 	),
 
 	TP_fast_assign(
+		struct kvmppc_book3s_shadow_vcpu *svcpu;
 		__entry->exit_nr	= exit_nr;
 		__entry->pc		= kvmppc_get_pc(vcpu);
 		__entry->dar		= kvmppc_get_fault_dar(vcpu);
 		__entry->msr		= vcpu->arch.shared->msr;
-		__entry->srr1		= to_svcpu(vcpu)->shadow_srr1;
+		svcpu = svcpu_get(vcpu);
+		__entry->srr1		= svcpu->shadow_srr1;
+		svcpu_put(svcpu);
 	),
 
 	TP_printk("exit=0x%x | pc=0x%lx | msr=0x%lx | dar=0x%lx | srr1=0x%lx",