Skip to contentSkip to conversation

KVM/arm updates for 5.4

PATCH In review Marc Zyngier 2 participants 19 messages Sep 9, 2019 – Sep 10, 2019

Subsystems (5)
Documentation Arm (32-bit) KVM arm64 +1 more subsystems
lore
Share & tools
lore

Marc Zyngier

author maintainer
#0 lore raw

cover KVM/arm updates for 5.4

Paolo, Radim,

Here is the KVM/arm updates for 5.4: a new ITS translation cache improving performance for interrupt injection of of assigned devices, a couple of fixes to allow up to 512 vcpus, and a number of fixes and other cleanups.

Please pull,

	M.

The following changes since commit d45331b00ddb179e291766617259261c112db872:

Linux 5.3-rc4 (2019-08-11 13:26:41 -0700)

are available in the Git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/kvmarm/kvmarm.git tags/kvmarm-5.4

for you to fetch changes up to 92f35b751c71d14250a401246f2c792e3aa5b386:

KVM: arm/arm64: vgic: Allow more than 256 vcpus for KVM_IRQ_LINE (2019-09-09 12:29:09 +0100)

----------------------------------------------------------------
KVM/arm updates for 5.4

- New ITS translation cache
- Allow up to 512 CPUs to be supported with GICv3 (for real this time)
- Now call kvm_arch_vcpu_blocking early in the blocking sequence
- Tidy-up device mappings in S2 when DIC is available
- Clean icache invalidation on VMID rollover
- General cleanup

----------------------------------------------------------------
Alexandru Elisei (1):
KVM: arm/arm64: vgic: Make function comments match function declarations

Eric Auger (1):
KVM: arm/arm64: vgic: Use a single IO device per redistributor

James Morse (1):
arm64: KVM: Device mappings should be execute-never

Marc Zyngier (13):
KVM: arm/arm64: vgic: Add LPI translation cache definition
KVM: arm/arm64: vgic: Add __vgic_put_lpi_locked primitive
KVM: arm/arm64: vgic-its: Add MSI-LPI translation cache invalidation
KVM: arm/arm64: vgic-its: Invalidate MSI-LPI translation cache on specific commands
KVM: arm/arm64: vgic-its: Invalidate MSI-LPI translation cache on disabling LPIs
KVM: arm/arm64: vgic-its: Invalidate MSI-LPI translation cache on ITS disable
KVM: arm/arm64: vgic-its: Invalidate MSI-LPI translation cache on vgic teardown
KVM: arm/arm64: vgic-its: Cache successful MSI->LPI translation
KVM: arm/arm64: vgic-its: Check the LPI translation cache on MSI injection
KVM: arm/arm64: vgic-irqfd: Implement kvm_arch_set_irq_inatomic
KVM: Call kvm_arch_vcpu_blocking early into the blocking sequence
KVM: arm/arm64: vgic: Remove spurious semicolons
KVM: arm/arm64: vgic: Allow more than 256 vcpus for KVM_IRQ_LINE

Mark Rutland (1):
arm64/kvm: Remove VMID rollover I-cache maintenance

 Documentation/virt/kvm/api.txt        |  12 +-
 arch/arm/include/uapi/asm/kvm.h       |   4 +-
 arch/arm64/include/asm/pgtable-prot.h |   2 +-
 arch/arm64/include/uapi/asm/kvm.h     |   4 +-
 arch/arm64/kvm/hyp/tlb.c              |  14 ++-
 include/kvm/arm_vgic.h                |   4 +-
 include/uapi/linux/kvm.h              |   1 +
 virt/kvm/arm/arm.c                    |   2 +
 virt/kvm/arm/vgic/vgic-init.c         |   8 +-
 virt/kvm/arm/vgic/vgic-irqfd.c        |  36 +++++-
 virt/kvm/arm/vgic/vgic-its.c          | 207 ++++++++++++++++++++++++++++++++++
 virt/kvm/arm/vgic/vgic-mmio-v3.c      |  85 +++++---------
 virt/kvm/arm/vgic/vgic-v2.c           |   7 +-
 virt/kvm/arm/vgic/vgic-v3.c           |   7 +-
 virt/kvm/arm/vgic/vgic.c              |  26 +++--
 virt/kvm/arm/vgic/vgic.h              |   5 +
 virt/kvm/kvm_main.c                   |   7 +-
 17 files changed, 339 insertions(+), 92 deletions(-)
author maintainer
#1 lore raw

PATCH 1/17 KVM: arm/arm64: vgic: Add LPI translation cache definition

Add the basic data structure that expresses an MSI to LPI translation as well as the allocation/release hooks.

The size of the cache is arbitrarily defined as 16*nr_vcpus.

---
 include/kvm/arm_vgic.h        |  3 +++
 virt/kvm/arm/vgic/vgic-init.c |  5 ++++
 virt/kvm/arm/vgic/vgic-its.c  | 49 +++++++++++++++++++++++++++++++++++
 virt/kvm/arm/vgic/vgic.h      |  2 ++
 4 files changed, 59 insertions(+)
include/kvm/arm_vgic.h +3 -0
index 7a30524a80ee..ded50a30e2d5 100644
@@ -249,6 +249,9 @@ struct vgic_dist {
 	struct list_head	lpi_list_head;
 	int			lpi_list_count;
 
+	/* LPI translation cache */
+	struct list_head	lpi_translation_cache;
+
 	/* used by vgic-debug */
 	struct vgic_state_iter *iter;
 
virt/kvm/arm/vgic/vgic-init.c +5 -0
index bdbc297d06fb..80127ca9269f 100644
@@ -53,6 +53,7 @@ void kvm_vgic_early_init(struct kvm *kvm)
 	struct vgic_dist *dist = &kvm->arch.vgic;
 
 	INIT_LIST_HEAD(&dist->lpi_list_head);
+	INIT_LIST_HEAD(&dist->lpi_translation_cache);
 	raw_spin_lock_init(&dist->lpi_list_lock);
 }
 
@@ -294,6 +295,7 @@ int vgic_init(struct kvm *kvm)
 	}
 
 	if (vgic_has_its(kvm)) {
+		vgic_lpi_translation_cache_init(kvm);
 		ret = vgic_v4_init(kvm);
 		if (ret)
 			goto out;
@@ -335,6 +337,9 @@ static void kvm_vgic_dist_destroy(struct kvm *kvm)
 		INIT_LIST_HEAD(&dist->rd_regions);
 	}
 
+	if (vgic_has_its(kvm))
+		vgic_lpi_translation_cache_destroy(kvm);
+
 	if (vgic_supports_direct_msis(kvm))
 		vgic_v4_teardown(kvm);
 }
virt/kvm/arm/vgic/vgic-its.c +49 -0

67 lines not on this page · raw message

virt/kvm/arm/vgic/vgic.h +2 -0
index 797e05004d80..a25b790ffa4e 100644
@@ -307,6 +307,8 @@ int vgic_copy_lpi_list(struct kvm *kvm, struct kvm_vcpu *vcpu, u32 **intid_ptr);
 int vgic_its_resolve_lpi(struct kvm *kvm, struct vgic_its *its,
 			 u32 devid, u32 eventid, struct vgic_irq **irq);
 struct vgic_its *vgic_msi_to_its(struct kvm *kvm, struct kvm_msi *msi);
+void vgic_lpi_translation_cache_init(struct kvm *kvm);
+void vgic_lpi_translation_cache_destroy(struct kvm *kvm);
 
 bool vgic_supports_direct_msis(struct kvm *kvm);
 int vgic_v4_init(struct kvm *kvm);
Signature · 1 line
2.20.1
author maintainer
#2 lore raw

PATCH 2/17 KVM: arm/arm64: vgic: Add __vgic_put_lpi_locked primitive

In reply to Marc Zyngier

Our LPI translation cache needs to be able to drop the refcount on an LPI whilst already holding the lpi_list_lock.

Let's add a new primitive for this.

---
 virt/kvm/arm/vgic/vgic.c | 26 +++++++++++++++++---------
 virt/kvm/arm/vgic/vgic.h |  1 +
 2 files changed, 18 insertions(+), 9 deletions(-)
virt/kvm/arm/vgic/vgic.c +17 -9
index 13d4b38a94ec..846bb680fd45 100644
@@ -119,6 +119,22 @@ static void vgic_irq_release(struct kref *ref)
 {
 }
 
+/*
+ * Drop the refcount on the LPI. Must be called with lpi_list_lock held.
+ */
+void __vgic_put_lpi_locked(struct kvm *kvm, struct vgic_irq *irq)
+{
+	struct vgic_dist *dist = &kvm->arch.vgic;
+
+	if (!kref_put(&irq->refcount, vgic_irq_release))
+		return;
+
+	list_del(&irq->lpi_list);
+	dist->lpi_list_count--;
+
+	kfree(irq);
+}
+
 void vgic_put_irq(struct kvm *kvm, struct vgic_irq *irq)
 {
 	struct vgic_dist *dist = &kvm->arch.vgic;
@@ -128,16 +144,8 @@ void vgic_put_irq(struct kvm *kvm, struct vgic_irq *irq)
 		return;
 
 	raw_spin_lock_irqsave(&dist->lpi_list_lock, flags);
-	if (!kref_put(&irq->refcount, vgic_irq_release)) {
-		raw_spin_unlock_irqrestore(&dist->lpi_list_lock, flags);
-		return;
-	};
-
-	list_del(&irq->lpi_list);
-	dist->lpi_list_count--;
+	__vgic_put_lpi_locked(kvm, irq);
 	raw_spin_unlock_irqrestore(&dist->lpi_list_lock, flags);
-
-	kfree(irq);
 }
 
 void vgic_flush_pending_lpis(struct kvm_vcpu *vcpu)
virt/kvm/arm/vgic/vgic.h +1 -0
index a25b790ffa4e..8e9413e317b8 100644
@@ -161,6 +161,7 @@ vgic_get_mmio_region(struct kvm_vcpu *vcpu, struct vgic_io_device *iodev,
 		     gpa_t addr, int len);
 struct vgic_irq *vgic_get_irq(struct kvm *kvm, struct kvm_vcpu *vcpu,
 			      u32 intid);
+void __vgic_put_lpi_locked(struct kvm *kvm, struct vgic_irq *irq);
 void vgic_put_irq(struct kvm *kvm, struct vgic_irq *irq);
 bool vgic_get_phys_line_level(struct vgic_irq *irq);
 void vgic_irq_set_phys_pending(struct vgic_irq *irq, bool pending);
Signature · 1 line
2.20.1
author maintainer
#3 lore raw

PATCH 3/17 KVM: arm/arm64: vgic-its: Add MSI-LPI translation cache invalidation

In reply to Marc Zyngier

There's a number of cases where we need to invalidate the caching of translations, so let's add basic support for that.

---
 virt/kvm/arm/vgic/vgic-its.c | 23 +++++++++++++++++++++++
 virt/kvm/arm/vgic/vgic.h     |  1 +
 2 files changed, 24 insertions(+)
virt/kvm/arm/vgic/vgic-its.c +23 -0
index 0e5c1519bbe2..cc6b5e49a312 100644
@@ -535,6 +535,29 @@ static unsigned long vgic_mmio_read_its_idregs(struct kvm *kvm,
 	return 0;
 }
 
+void vgic_its_invalidate_cache(struct kvm *kvm)
+{
+	struct vgic_dist *dist = &kvm->arch.vgic;
+	struct vgic_translation_cache_entry *cte;
+	unsigned long flags;
+
+	raw_spin_lock_irqsave(&dist->lpi_list_lock, flags);
+
+	list_for_each_entry(cte, &dist->lpi_translation_cache, entry) {
+		/*
+		 * If we hit a NULL entry, there is nothing after this
+		 * point.
+		 */
+		if (!cte->irq)
+			break;
+
+		__vgic_put_lpi_locked(kvm, cte->irq);
+		cte->irq = NULL;
+	}
+
+	raw_spin_unlock_irqrestore(&dist->lpi_list_lock, flags);
+}
+
 int vgic_its_resolve_lpi(struct kvm *kvm, struct vgic_its *its,
 			 u32 devid, u32 eventid, struct vgic_irq **irq)
 {
virt/kvm/arm/vgic/vgic.h +1 -0
index 8e9413e317b8..c7fb4da2ab3b 100644
@@ -310,6 +310,7 @@ int vgic_its_resolve_lpi(struct kvm *kvm, struct vgic_its *its,
 struct vgic_its *vgic_msi_to_its(struct kvm *kvm, struct kvm_msi *msi);
 void vgic_lpi_translation_cache_init(struct kvm *kvm);
 void vgic_lpi_translation_cache_destroy(struct kvm *kvm);
+void vgic_its_invalidate_cache(struct kvm *kvm);
 
 bool vgic_supports_direct_msis(struct kvm *kvm);
 int vgic_v4_init(struct kvm *kvm);
Signature · 1 line
2.20.1
author maintainer
#4 lore raw

PATCH 4/17 KVM: arm/arm64: vgic-its: Invalidate MSI-LPI translation cache on specific commands

In reply to Marc Zyngier

The LPI translation cache needs to be discarded when an ITS command may affect the translation of an LPI (DISCARD, MAPC and MAPD with V=0) or the routing of an LPI to a redistributor with disabled LPIs (MOVI, MOVALL).

We decide to perform a full invalidation of the cache, irrespective of the LPI that is affected. Commands are supposed to be rare enough that it doesn't matter.

---
 virt/kvm/arm/vgic/vgic-its.c | 9 +++++++++
 1 file changed, 9 insertions(+)
virt/kvm/arm/vgic/vgic-its.c +9 -0
index cc6b5e49a312..09a179820816 100644
@@ -722,6 +722,8 @@ static int vgic_its_cmd_handle_discard(struct kvm *kvm, struct vgic_its *its,
 		 * don't bother here since we clear the ITTE anyway and the
 		 * pending state is a property of the ITTE struct.
 		 */
+		vgic_its_invalidate_cache(kvm);
+
 		its_free_ite(kvm, ite);
 		return 0;
 	}
@@ -757,6 +759,8 @@ static int vgic_its_cmd_handle_movi(struct kvm *kvm, struct vgic_its *its,
 	ite->collection = collection;
 	vcpu = kvm_get_vcpu(kvm, collection->target_addr);
 
+	vgic_its_invalidate_cache(kvm);
+
 	return update_affinity(ite->irq, vcpu);
 }
 
@@ -985,6 +989,8 @@ static void vgic_its_free_device(struct kvm *kvm, struct its_device *device)
 	list_for_each_entry_safe(ite, temp, &device->itt_head, ite_list)
 		its_free_ite(kvm, ite);
 
+	vgic_its_invalidate_cache(kvm);
+
 	list_del(&device->dev_list);
 	kfree(device);
 }
@@ -1090,6 +1096,7 @@ static int vgic_its_cmd_handle_mapc(struct kvm *kvm, struct vgic_its *its,
 
 	if (!valid) {
 		vgic_its_free_collection(its, coll_id);
+		vgic_its_invalidate_cache(kvm);
 	} else {
 		collection = find_collection(its, coll_id);
 
@@ -1238,6 +1245,8 @@ static int vgic_its_cmd_handle_movall(struct kvm *kvm, struct vgic_its *its,
 		vgic_put_irq(kvm, irq);
 	}
 
+	vgic_its_invalidate_cache(kvm);
+
 	kfree(intids);
 	return 0;
 }
Signature · 1 line
2.20.1
author maintainer
#5 lore raw

PATCH 5/17 KVM: arm/arm64: vgic-its: Invalidate MSI-LPI translation cache on disabling LPIs

In reply to Marc Zyngier

If a vcpu disables LPIs at its redistributor level, we need to make sure we won't pend more interrupts. For this, we need to invalidate the LPI translation cache.

---
 virt/kvm/arm/vgic/vgic-mmio-v3.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
virt/kvm/arm/vgic/vgic-mmio-v3.c +3 -1
index c45e2d7e942f..fdcfb7ae4491 100644
@@ -192,8 +192,10 @@ static void vgic_mmio_write_v3r_ctlr(struct kvm_vcpu *vcpu,
 
 	vgic_cpu->lpis_enabled = val & GICR_CTLR_ENABLE_LPIS;
 
-	if (was_enabled && !vgic_cpu->lpis_enabled)
+	if (was_enabled && !vgic_cpu->lpis_enabled) {
 		vgic_flush_pending_lpis(vcpu);
+		vgic_its_invalidate_cache(vcpu->kvm);
+	}
 
 	if (!was_enabled && vgic_cpu->lpis_enabled)
 		vgic_enable_lpis(vcpu);
Signature · 1 line
2.20.1
author maintainer
#6 lore raw

PATCH 6/17 KVM: arm/arm64: vgic-its: Invalidate MSI-LPI translation cache on ITS disable

In reply to Marc Zyngier

If an ITS gets disabled, we need to make sure that further interrupts won't hit in the cache. For that, we invalidate the translation cache when the ITS is disabled.

---
 virt/kvm/arm/vgic/vgic-its.c | 2 ++
 1 file changed, 2 insertions(+)
virt/kvm/arm/vgic/vgic-its.c +2 -0
index 09a179820816..05406bd92ce9 100644
@@ -1597,6 +1597,8 @@ static void vgic_mmio_write_its_ctlr(struct kvm *kvm, struct vgic_its *its,
 		goto out;
 
 	its->enabled = !!(val & GITS_CTLR_ENABLE);
+	if (!its->enabled)
+		vgic_its_invalidate_cache(kvm);
 
 	/*
 	 * Try to process any pending commands. This function bails out early
Signature · 1 line
2.20.1
author maintainer
#7 lore raw

PATCH 7/17 KVM: arm/arm64: vgic-its: Invalidate MSI-LPI translation cache on vgic teardown

In reply to Marc Zyngier

In order to avoid leaking vgic_irq structures on teardown, we need to drop all references to LPIs before deallocating the cache itself.

This is done by invalidating the cache on vgic teardown.

---
 virt/kvm/arm/vgic/vgic-its.c | 2 ++
 1 file changed, 2 insertions(+)
virt/kvm/arm/vgic/vgic-its.c +2 -0
index 05406bd92ce9..d3e90a9d0a7a 100644
@@ -1731,6 +1731,8 @@ void vgic_lpi_translation_cache_destroy(struct kvm *kvm)
 	struct vgic_dist *dist = &kvm->arch.vgic;
 	struct vgic_translation_cache_entry *cte, *tmp;
 
+	vgic_its_invalidate_cache(kvm);
+
 	list_for_each_entry_safe(cte, tmp,
 				 &dist->lpi_translation_cache, entry) {
 		list_del(&cte->entry);
Signature · 1 line
2.20.1
author maintainer
#8 lore raw

PATCH 8/17 KVM: arm/arm64: vgic-its: Cache successful MSI->LPI translation

In reply to Marc Zyngier

On a successful translation, preserve the parameters in the LPI translation cache. Each translation is reusing the last slot in the list, naturally evicting the least recently used entry.

---
 virt/kvm/arm/vgic/vgic-its.c | 86 ++++++++++++++++++++++++++++++++++++
 1 file changed, 86 insertions(+)
virt/kvm/arm/vgic/vgic-its.c +86 -0
index d3e90a9d0a7a..e61d3ea0ab40 100644
@@ -535,6 +535,90 @@ static unsigned long vgic_mmio_read_its_idregs(struct kvm *kvm,
 	return 0;
 }
 
+static struct vgic_irq *__vgic_its_check_cache(struct vgic_dist *dist,
+					       phys_addr_t db,
+					       u32 devid, u32 eventid)
+{
+	struct vgic_translation_cache_entry *cte;
+
+	list_for_each_entry(cte, &dist->lpi_translation_cache, entry) {
+		/*
+		 * If we hit a NULL entry, there is nothing after this
+		 * point.
+		 */
+		if (!cte->irq)
+			break;
+
+		if (cte->db != db || cte->devid != devid ||
+		    cte->eventid != eventid)
+			continue;
+
+		/*
+		 * Move this entry to the head, as it is the most
+		 * recently used.
+		 */
+		if (!list_is_first(&cte->entry, &dist->lpi_translation_cache))
+			list_move(&cte->entry, &dist->lpi_translation_cache);
+
+		return cte->irq;
+	}
+
+	return NULL;
+}
+
+static void vgic_its_cache_translation(struct kvm *kvm, struct vgic_its *its,
+				       u32 devid, u32 eventid,
+				       struct vgic_irq *irq)
+{
+	struct vgic_dist *dist = &kvm->arch.vgic;
+	struct vgic_translation_cache_entry *cte;
+	unsigned long flags;
+	phys_addr_t db;
+
+	/* Do not cache a directly injected interrupt */
+	if (irq->hw)
+		return;
+
+	raw_spin_lock_irqsave(&dist->lpi_list_lock, flags);
+
+	if (unlikely(list_empty(&dist->lpi_translation_cache)))
+		goto out;
+
+	/*
+	 * We could have raced with another CPU caching the same
+	 * translation behind our back, so let's check it is not in
+	 * already
+	 */
+	db = its->vgic_its_base + GITS_TRANSLATER;
+	if (__vgic_its_check_cache(dist, db, devid, eventid))
+		goto out;
+
+	/* Always reuse the last entry (LRU policy) */
+	cte = list_last_entry(&dist->lpi_translation_cache,
+			      typeof(*cte), entry);
+
+	/*
+	 * Caching the translation implies having an extra reference
+	 * to the interrupt, so drop the potential reference on what
+	 * was in the cache, and increment it on the new interrupt.
+	 */
+	if (cte->irq)
+		__vgic_put_lpi_locked(kvm, cte->irq);
+
+	vgic_get_irq_kref(irq);
+
+	cte->db		= db;
+	cte->devid	= devid;
+	cte->eventid	= eventid;
+	cte->irq	= irq;
+
+	/* Move the new translation to the head of the list */
+	list_move(&cte->entry, &dist->lpi_translation_cache);
+
+out:
+	raw_spin_unlock_irqrestore(&dist->lpi_list_lock, flags);
+}
+
 void vgic_its_invalidate_cache(struct kvm *kvm)
 {
 	struct vgic_dist *dist = &kvm->arch.vgic;
@@ -578,6 +662,8 @@ int vgic_its_resolve_lpi(struct kvm *kvm, struct vgic_its *its,
 	if (!vcpu->arch.vgic_cpu.lpis_enabled)
 		return -EBUSY;
 
+	vgic_its_cache_translation(kvm, its, devid, eventid, ite->irq);
+
 	*irq = ite->irq;
 	return 0;
 }
Signature · 1 line
2.20.1
author maintainer
#9 lore raw

PATCH 9/17 KVM: arm/arm64: vgic-its: Check the LPI translation cache on MSI injection

In reply to Marc Zyngier

When performing an MSI injection, let's first check if the translation is already in the cache. If so, let's inject it quickly without going through the whole translation process.

---
 virt/kvm/arm/vgic/vgic-its.c | 36 ++++++++++++++++++++++++++++++++++++
 virt/kvm/arm/vgic/vgic.h     |  1 +
 2 files changed, 37 insertions(+)
virt/kvm/arm/vgic/vgic-its.c +36 -0
index e61d3ea0ab40..2be6b66b3856 100644
@@ -566,6 +566,20 @@ static struct vgic_irq *__vgic_its_check_cache(struct vgic_dist *dist,
 	return NULL;
 }
 
+static struct vgic_irq *vgic_its_check_cache(struct kvm *kvm, phys_addr_t db,
+					     u32 devid, u32 eventid)
+{
+	struct vgic_dist *dist = &kvm->arch.vgic;
+	struct vgic_irq *irq;
+	unsigned long flags;
+
+	raw_spin_lock_irqsave(&dist->lpi_list_lock, flags);
+	irq = __vgic_its_check_cache(dist, db, devid, eventid);
+	raw_spin_unlock_irqrestore(&dist->lpi_list_lock, flags);
+
+	return irq;
+}
+
 static void vgic_its_cache_translation(struct kvm *kvm, struct vgic_its *its,
 				       u32 devid, u32 eventid,
 				       struct vgic_irq *irq)
@@ -725,6 +739,25 @@ static int vgic_its_trigger_msi(struct kvm *kvm, struct vgic_its *its,
 	return 0;
 }
 
+int vgic_its_inject_cached_translation(struct kvm *kvm, struct kvm_msi *msi)
+{
+	struct vgic_irq *irq;
+	unsigned long flags;
+	phys_addr_t db;
+
+	db = (u64)msi->address_hi << 32 | msi->address_lo;
+	irq = vgic_its_check_cache(kvm, db, msi->devid, msi->data);
+
+	if (!irq)
+		return -1;
+
+	raw_spin_lock_irqsave(&irq->irq_lock, flags);
+	irq->pending_latch = true;
+	vgic_queue_irq_unlock(kvm, irq, flags);
+
+	return 0;
+}
+
 /*
  * Queries the KVM IO bus framework to get the ITS pointer from the given
  * doorbell address.
@@ -736,6 +769,9 @@ int vgic_its_inject_msi(struct kvm *kvm, struct kvm_msi *msi)
 	struct vgic_its *its;
 	int ret;
 
+	if (!vgic_its_inject_cached_translation(kvm, msi))
+		return 1;
+
 	its = vgic_msi_to_its(kvm, msi);
 	if (IS_ERR(its))
 		return PTR_ERR(its);
virt/kvm/arm/vgic/vgic.h +1 -0
index c7fb4da2ab3b..83066a81b16a 100644
@@ -308,6 +308,7 @@ int vgic_copy_lpi_list(struct kvm *kvm, struct kvm_vcpu *vcpu, u32 **intid_ptr);
 int vgic_its_resolve_lpi(struct kvm *kvm, struct vgic_its *its,
 			 u32 devid, u32 eventid, struct vgic_irq **irq);
 struct vgic_its *vgic_msi_to_its(struct kvm *kvm, struct kvm_msi *msi);
+int vgic_its_inject_cached_translation(struct kvm *kvm, struct kvm_msi *msi);
 void vgic_lpi_translation_cache_init(struct kvm *kvm);
 void vgic_lpi_translation_cache_destroy(struct kvm *kvm);
 void vgic_its_invalidate_cache(struct kvm *kvm);
Signature · 1 line
2.20.1
author maintainer
#10 lore raw

PATCH 10/17 KVM: arm/arm64: vgic-irqfd: Implement kvm_arch_set_irq_inatomic

In reply to Marc Zyngier

Now that we have a cache of MSI->LPI translations, it is pretty easy to implement kvm_arch_set_irq_inatomic (this cache can be parsed without sleeping).

Hopefully, this will improve some LPI-heavy workloads.

---
 virt/kvm/arm/vgic/vgic-irqfd.c | 36 ++++++++++++++++++++++++++++------
 1 file changed, 30 insertions(+), 6 deletions(-)
virt/kvm/arm/vgic/vgic-irqfd.c +30 -6
index c9304b88e720..d8cdfea5cc96 100644
@@ -66,6 +66,15 @@ int kvm_set_routing_entry(struct kvm *kvm,
 	return r;
 }
 
+static void kvm_populate_msi(struct kvm_kernel_irq_routing_entry *e,
+			     struct kvm_msi *msi)
+{
+	msi->address_lo = e->msi.address_lo;
+	msi->address_hi = e->msi.address_hi;
+	msi->data = e->msi.data;
+	msi->flags = e->msi.flags;
+	msi->devid = e->msi.devid;
+}
 /**
  * kvm_set_msi: inject the MSI corresponding to the
  * MSI routing entry
@@ -79,21 +88,36 @@ int kvm_set_msi(struct kvm_kernel_irq_routing_entry *e,
 {
 	struct kvm_msi msi;
 
-	msi.address_lo = e->msi.address_lo;
-	msi.address_hi = e->msi.address_hi;
-	msi.data = e->msi.data;
-	msi.flags = e->msi.flags;
-	msi.devid = e->msi.devid;
-
 	if (!vgic_has_its(kvm))
 		return -ENODEV;
 
 	if (!level)
 		return -1;
 
+	kvm_populate_msi(e, &msi);
 	return vgic_its_inject_msi(kvm, &msi);
 }
 
+/**
+ * kvm_arch_set_irq_inatomic: fast-path for irqfd injection
+ *
+ * Currently only direct MSI injection is supported.
+ */
+int kvm_arch_set_irq_inatomic(struct kvm_kernel_irq_routing_entry *e,
+			      struct kvm *kvm, int irq_source_id, int level,
+			      bool line_status)
+{
+	if (e->type == KVM_IRQ_ROUTING_MSI && vgic_has_its(kvm) && level) {
+		struct kvm_msi msi;
+
+		kvm_populate_msi(e, &msi);
+		if (!vgic_its_inject_cached_translation(kvm, &msi))
+			return 0;
+	}
+
+	return -EWOULDBLOCK;
+}
+
 int kvm_vgic_setup_default_irq_routing(struct kvm *kvm)
 {
 	struct kvm_irq_routing_entry *entries;
Signature · 1 line
2.20.1
author maintainer
#11 lore raw

PATCH 11/17 arm64/kvm: Remove VMID rollover I-cache maintenance

In reply to Marc Zyngier

From: Mark Rutland <[email protected]>

For VPIPT I-caches, we need I-cache maintenance on VMID rollover to avoid an ABA problem. Consider a single vCPU VM, with a pinned stage-2, running with an idmap VA->IPA and idmap IPA->PA. If we don't do maintenance on rollover:

        // VMID A
        Writes insn X to PA 0xF
        Invalidates PA 0xF (for VMID A)

        I$ contains [{A,F}->X]

        [VMID ROLLOVER]

        // VMID B
        Writes insn Y to PA 0xF
        Invalidates PA 0xF (for VMID B)

        I$ contains [{A,F}->X, {B,F}->Y]

        [VMID ROLLOVER]

        // VMID A
        I$ contains [{A,F}->X, {B,F}->Y]

        Unexpectedly hits stale I$ line {A,F}->X.

However, for PIPT and VIPT I-caches, the VMID doesn't affect lookup or constrain maintenance. Given the VMID doesn't affect PIPT and VIPT I-caches, and given VMID rollover is independent of changes to stage-2 mappings, I-cache maintenance cannot be necessary on VMID rollover for PIPT or VIPT I-caches.

This patch removes the maintenance on rollover for VIPT and PIPT I-caches. At the same time, the unnecessary colons are removed from the asm statement to make it more legible.

---
 arch/arm64/kvm/hyp/tlb.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)
arch/arm64/kvm/hyp/tlb.c +13 -1
index d49a14497715..c466060b76d6 100644
@@ -193,6 +193,18 @@ void __hyp_text __kvm_flush_vm_context(void)
 {
 	dsb(ishst);
 	__tlbi(alle1is);
-	asm volatile("ic ialluis" : : );
+
+	/*
+	 * VIPT and PIPT caches are not affected by VMID, so no maintenance
+	 * is necessary across a VMID rollover.
+	 *
+	 * VPIPT caches constrain lookup and maintenance to the active VMID,
+	 * so we need to invalidate lines with a stale VMID to avoid an ABA
+	 * race after multiple rollovers.
+	 *
+	 */
+	if (icache_is_vpipt())
+		asm volatile("ic ialluis");
+
 	dsb(ish);
 }
Signature · 1 line
2.20.1
author maintainer
#12 lore raw

PATCH 12/17 KVM: arm/arm64: vgic: Make function comments match function declarations

In reply to Marc Zyngier

From: Alexandru Elisei <[email protected]>

Since commit 503a62862e8f ("KVM: arm/arm64: vgic: Rely on the GIC driver to parse the firmware tables"), the vgic_v{2,3}_probe functions stopped using a DT node. Commit 909777324588 ("KVM: arm/arm64: vgic-new: vgic_init:
implement kvm_vgic_hyp_init") changed the functions again, and now they require exactly one argument, a struct gic_kvm_info populated by the GIC driver. Unfortunately the comments regressed and state that a DT node is used instead. Change the function comments to reflect the current prototypes.

---
 virt/kvm/arm/vgic/vgic-v2.c | 7 ++++---
 virt/kvm/arm/vgic/vgic-v3.c | 7 ++++---
 2 files changed, 8 insertions(+), 6 deletions(-)
virt/kvm/arm/vgic/vgic-v2.c +4 -3
index 96aab77d0471..e67945020b45 100644
@@ -354,10 +354,11 @@ int vgic_v2_map_resources(struct kvm *kvm)
 DEFINE_STATIC_KEY_FALSE(vgic_v2_cpuif_trap);
 
 /**
- * vgic_v2_probe - probe for a GICv2 compatible interrupt controller in DT
- * @node:	pointer to the DT node
+ * vgic_v2_probe - probe for a VGICv2 compatible interrupt controller
+ * @info:	pointer to the GIC description
  *
- * Returns 0 if a GICv2 has been found, returns an error code otherwise
+ * Returns 0 if the VGICv2 has been probed successfully, returns an error code
+ * otherwise
  */
 int vgic_v2_probe(const struct gic_kvm_info *info)
 {
virt/kvm/arm/vgic/vgic-v3.c +4 -3
index 0c653a1e5215..30955d162a01 100644
@@ -570,10 +570,11 @@ static int __init early_gicv4_enable(char *buf)
 early_param("kvm-arm.vgic_v4_enable", early_gicv4_enable);
 
 /**
- * vgic_v3_probe - probe for a GICv3 compatible interrupt controller in DT
- * @node:	pointer to the DT node
+ * vgic_v3_probe - probe for a VGICv3 compatible interrupt controller
+ * @info:	pointer to the GIC description
  *
- * Returns 0 if a GICv3 has been found, returns an error code otherwise
+ * Returns 0 if the VGICv3 has been probed successfully, returns an error code
+ * otherwise
  */
 int vgic_v3_probe(const struct gic_kvm_info *info)
 {
Signature · 1 line
2.20.1
author maintainer
#13 lore raw

PATCH 13/17 KVM: Call kvm_arch_vcpu_blocking early into the blocking sequence

In reply to Marc Zyngier

When a vpcu is about to block by calling kvm_vcpu_block, we call back into the arch code to allow any form of synchronization that may be required at this point (SVN stops the AVIC, ARM synchronises the VMCR and enables GICv4 doorbells). But this synchronization comes in quite late, as we've potentially waited for halt_poll_ns to expire.

Instead, let's move kvm_arch_vcpu_blocking() to the beginning of kvm_vcpu_block(), which on ARM has several benefits:

- VMCR gets synchronised early, meaning that any interrupt delivered during the polling window will be evaluated with the correct guest PMR
- GICv4 doorbells are enabled, which means that any guest interrupt directly injected during that window will be immediately recognised

Tang Nianyao ran some tests on a GICv4 machine to evaluate such change, and reported up to a 10% improvement for netperf:

<quote>
netperf result:
D06 as server, intel 8180 server as client with change:
package 512 bytes - 5500 Mbits/s package 64 bytes - 760 Mbits/s without change:
package 512 bytes - 5000 Mbits/s package 64 bytes - 710 Mbits/s
</quote>

---
 virt/kvm/kvm_main.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)
virt/kvm/kvm_main.c +3 -4
index c6a91b044d8d..e6de3159e682 100644
@@ -2321,6 +2321,8 @@ void kvm_vcpu_block(struct kvm_vcpu *vcpu)
 	bool waited = false;
 	u64 block_ns;
 
+	kvm_arch_vcpu_blocking(vcpu);
+
 	start = cur = ktime_get();
 	if (vcpu->halt_poll_ns && !kvm_arch_no_poll(vcpu)) {
 		ktime_t stop = ktime_add_ns(ktime_get(), vcpu->halt_poll_ns);
@@ -2341,8 +2343,6 @@ void kvm_vcpu_block(struct kvm_vcpu *vcpu)
 		} while (single_task_running() && ktime_before(cur, stop));
 	}
 
-	kvm_arch_vcpu_blocking(vcpu);
-
 	for (;;) {
 		prepare_to_swait_exclusive(&vcpu->wq, &wait, TASK_INTERRUPTIBLE);
 
@@ -2355,9 +2355,8 @@ void kvm_vcpu_block(struct kvm_vcpu *vcpu)
 
 	finish_swait(&vcpu->wq, &wait);
 	cur = ktime_get();
-
-	kvm_arch_vcpu_unblocking(vcpu);
 out:
+	kvm_arch_vcpu_unblocking(vcpu);
 	block_ns = ktime_to_ns(cur) - ktime_to_ns(start);
 
 	if (!vcpu_valid_wakeup(vcpu))
Signature · 1 line
2.20.1
author maintainer
#14 lore raw

PATCH 14/17 KVM: arm/arm64: vgic: Remove spurious semicolons

In reply to Marc Zyngier

Detected by Coccinelle (and Will Deacon) using
scripts/coccinelle/misc/semicolon.cocci.

---
 virt/kvm/arm/vgic/vgic-init.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
virt/kvm/arm/vgic/vgic-init.c +1 -1
index 80127ca9269f..9175bfd83263 100644
@@ -510,7 +510,7 @@ int kvm_vgic_hyp_init(void)
 		break;
 	default:
 		ret = -ENODEV;
-	};
+	}
 
 	if (ret)
 		return ret;
Signature · 1 line
2.20.1
author maintainer
#15 lore raw

PATCH 15/17 KVM: arm/arm64: vgic: Use a single IO device per redistributor

In reply to Marc Zyngier

From: Eric Auger <[email protected]>

At the moment we use 2 IO devices per GICv3 redistributor: one one for the RD_base frame and one for the SGI_base frame.

Instead we can use a single IO device per redistributor (the 2 frames are contiguous). This saves slots on the KVM_MMIO_BUS which is currently limited to NR_IOBUS_DEVS (1000).

This change allows to instantiate up to 512 redistributors and may speed the guest boot with a large number of VCPUs.

---
 include/kvm/arm_vgic.h           |  1 -
 virt/kvm/arm/vgic/vgic-init.c    |  1 -
 virt/kvm/arm/vgic/vgic-mmio-v3.c | 81 ++++++++++----------------------
 3 files changed, 24 insertions(+), 59 deletions(-)
include/kvm/arm_vgic.h +0 -1
index ded50a30e2d5..af4f09c02bf1 100644
@@ -314,7 +314,6 @@ struct vgic_cpu {
 	 * parts of the redistributor.
 	 */
 	struct vgic_io_device	rd_iodev;
-	struct vgic_io_device	sgi_iodev;
 	struct vgic_redist_region *rdreg;
 
 	/* Contains the attributes and gpa of the LPI pending tables. */
virt/kvm/arm/vgic/vgic-init.c +0 -1
index 9175bfd83263..958e2f0d2207 100644
@@ -193,7 +193,6 @@ int kvm_vgic_vcpu_init(struct kvm_vcpu *vcpu)
 	int i;
 
 	vgic_cpu->rd_iodev.base_addr = VGIC_ADDR_UNDEF;
-	vgic_cpu->sgi_iodev.base_addr = VGIC_ADDR_UNDEF;
 
 	INIT_LIST_HEAD(&vgic_cpu->ap_list_head);
 	raw_spin_lock_init(&vgic_cpu->ap_list_lock);
virt/kvm/arm/vgic/vgic-mmio-v3.c +24 -57

165 lines not on this page · raw message

Signature · 1 line
2.20.1
author maintainer
#16 lore raw

PATCH 16/17 arm64: KVM: Device mappings should be execute-never

In reply to Marc Zyngier

From: James Morse <[email protected]>

Since commit 2f6ea23f63cca ("arm64: KVM: Avoid marking pages as XN in Stage-2 if CTR_EL0.DIC is set"), KVM has stopped marking normal memory as execute-never at stage2 when the system supports D->I Coherency at the PoU. This avoids KVM taking a trap when the page is first executed, in order to clean it to PoU.

The patch that added this change also wrapped PAGE_S2_DEVICE mappings up in this too. The upshot is, if your CPU caches support DIC ... you can execute devices.

Revert the PAGE_S2_DEVICE change so PTE_S2_XN is always used directly.

Fixes 2f6ea23f63cca ("arm64: KVM: Avoid marking pages as XN in Stage-2 if CTR_EL0.DIC is set")
---
 arch/arm64/include/asm/pgtable-prot.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/arm64/include/asm/pgtable-prot.h +1 -1
index 92d2e9f28f28..9a21b84536f2 100644
@@ -77,7 +77,7 @@
 	})
 
 #define PAGE_S2			__pgprot(_PROT_DEFAULT | PAGE_S2_MEMATTR(NORMAL) | PTE_S2_RDONLY | PAGE_S2_XN)
-#define PAGE_S2_DEVICE		__pgprot(_PROT_DEFAULT | PAGE_S2_MEMATTR(DEVICE_nGnRE) | PTE_S2_RDONLY | PAGE_S2_XN)
+#define PAGE_S2_DEVICE		__pgprot(_PROT_DEFAULT | PAGE_S2_MEMATTR(DEVICE_nGnRE) | PTE_S2_RDONLY | PTE_S2_XN)
 
 #define PAGE_NONE		__pgprot(((_PAGE_DEFAULT) & ~PTE_VALID) | PTE_PROT_NONE | PTE_RDONLY | PTE_NG | PTE_PXN | PTE_UXN)
 #define PAGE_SHARED		__pgprot(_PAGE_DEFAULT | PTE_USER | PTE_NG | PTE_PXN | PTE_UXN | PTE_WRITE)
Signature · 1 line
2.20.1
author maintainer
#17 lore raw

PATCH 17/17 KVM: arm/arm64: vgic: Allow more than 256 vcpus for KVM_IRQ_LINE

In reply to Marc Zyngier

While parts of the VGIC support a large number of vcpus (we bravely allow up to 512), other parts are more limited.

One of these limits is visible in the KVM_IRQ_LINE ioctl, which only allows 256 vcpus to be signalled when using the CPU or PPI types. Unfortunately, we've cornered ourselves badly by allocating all the bits in the irq field.

Since the irq_type subfield (8 bit wide) is currently only taking the values 0, 1 and 2 (and we have been careful not to allow anything else), let's reduce this field to only 4 bits, and allocate the remaining 4 bits to a vcpu2_index, which acts as a multiplier:

vcpu_id = 256 * vcpu2_index + vcpu_index

With that, and a new capability (KVM_CAP_ARM_IRQ_LINE_LAYOUT_2) allowing this to be discovered, it becomes possible to inject PPIs to up to 4096 vcpus. But please just don't.

Whilst we're there, add a clarification about the use of KVM_IRQ_LINE on arm, which is not completely conditionned by KVM_CAP_IRQCHIP.

---
 Documentation/virt/kvm/api.txt    | 12 ++++++++++--
 arch/arm/include/uapi/asm/kvm.h   |  4 +++-
 arch/arm64/include/uapi/asm/kvm.h |  4 +++-
 include/uapi/linux/kvm.h          |  1 +
 virt/kvm/arm/arm.c                |  2 ++
 5 files changed, 19 insertions(+), 4 deletions(-)
Documentation/virt/kvm/api.txt +10 -2
index 2d067767b617..25931ca1cb38 100644
@@ -753,8 +753,8 @@ in-kernel irqchip (GIC), and for in-kernel irqchip can tell the GIC to
 use PPIs designated for specific cpus.  The irq field is interpreted
 like this:
 
-  bits:  | 31 ... 24 | 23  ... 16 | 15    ...    0 |
-  field: | irq_type  | vcpu_index |     irq_id     |
+  bits:  |  31 ... 28  | 27 ... 24 | 23  ... 16 | 15 ... 0 |
+  field: | vcpu2_index | irq_type  | vcpu_index |  irq_id  |
 
 The irq_type field has the following values:
 - irq_type[0]: out-of-kernel GIC: irq_id 0 is IRQ, irq_id 1 is FIQ
@@ -766,6 +766,14 @@ The irq_type field has the following values:
 
 In both cases, level is used to assert/deassert the line.
 
+When KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 is supported, the target vcpu is
+identified as (256 * vcpu2_index + vcpu_index). Otherwise, vcpu2_index
+must be zero.
+
+Note that on arm/arm64, the KVM_CAP_IRQCHIP capability only conditions
+injection of interrupts for the in-kernel irqchip. KVM_IRQ_LINE can always
+be used for a userspace interrupt controller.
+
 struct kvm_irq_level {
 	union {
 		__u32 irq;     /* GSI */
arch/arm/include/uapi/asm/kvm.h +3 -1
index a4217c1a5d01..2769360f195c 100644
@@ -266,8 +266,10 @@ struct kvm_vcpu_events {
 #define   KVM_DEV_ARM_ITS_CTRL_RESET		4
 
 /* KVM_IRQ_LINE irq field index values */
+#define KVM_ARM_IRQ_VCPU2_SHIFT		28
+#define KVM_ARM_IRQ_VCPU2_MASK		0xf
 #define KVM_ARM_IRQ_TYPE_SHIFT		24
-#define KVM_ARM_IRQ_TYPE_MASK		0xff
+#define KVM_ARM_IRQ_TYPE_MASK		0xf
 #define KVM_ARM_IRQ_VCPU_SHIFT		16
 #define KVM_ARM_IRQ_VCPU_MASK		0xff
 #define KVM_ARM_IRQ_NUM_SHIFT		0
arch/arm64/include/uapi/asm/kvm.h +3 -1
index 9a507716ae2f..67c21f9bdbad 100644
@@ -325,8 +325,10 @@ struct kvm_vcpu_events {
 #define   KVM_ARM_VCPU_TIMER_IRQ_PTIMER		1
 
 /* KVM_IRQ_LINE irq field index values */
+#define KVM_ARM_IRQ_VCPU2_SHIFT		28
+#define KVM_ARM_IRQ_VCPU2_MASK		0xf
 #define KVM_ARM_IRQ_TYPE_SHIFT		24
-#define KVM_ARM_IRQ_TYPE_MASK		0xff
+#define KVM_ARM_IRQ_TYPE_MASK		0xf
 #define KVM_ARM_IRQ_VCPU_SHIFT		16
 #define KVM_ARM_IRQ_VCPU_MASK		0xff
 #define KVM_ARM_IRQ_NUM_SHIFT		0
include/uapi/linux/kvm.h +1 -0
index 5e3f12d5359e..5414b6588fbb 100644
@@ -996,6 +996,7 @@ struct kvm_ppc_resize_hpt {
 #define KVM_CAP_ARM_PTRAUTH_ADDRESS 171
 #define KVM_CAP_ARM_PTRAUTH_GENERIC 172
 #define KVM_CAP_PMU_EVENT_FILTER 173
+#define KVM_CAP_ARM_IRQ_LINE_LAYOUT_2 174
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
virt/kvm/arm/arm.c +2 -0
index 35a069815baf..86c6aa1cb58e 100644
@@ -196,6 +196,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 	case KVM_CAP_MP_STATE:
 	case KVM_CAP_IMMEDIATE_EXIT:
 	case KVM_CAP_VCPU_EVENTS:
+	case KVM_CAP_ARM_IRQ_LINE_LAYOUT_2:
 		r = 1;
 		break;
 	case KVM_CAP_ARM_SET_DEVICE_ADDR:
@@ -888,6 +889,7 @@ int kvm_vm_ioctl_irq_line(struct kvm *kvm, struct kvm_irq_level *irq_level,
 
 	irq_type = (irq >> KVM_ARM_IRQ_TYPE_SHIFT) & KVM_ARM_IRQ_TYPE_MASK;
 	vcpu_idx = (irq >> KVM_ARM_IRQ_VCPU_SHIFT) & KVM_ARM_IRQ_VCPU_MASK;
+	vcpu_idx += ((irq >> KVM_ARM_IRQ_VCPU2_SHIFT) & KVM_ARM_IRQ_VCPU2_MASK) * (KVM_ARM_IRQ_VCPU_MASK + 1);
 	irq_num = (irq >> KVM_ARM_IRQ_NUM_SHIFT) & KVM_ARM_IRQ_NUM_MASK;
 
 	trace_kvm_irq_line(irq_type, vcpu_idx, irq_num, irq_level->level);
Signature · 1 line
2.20.1