Skip to contentSkip to conversation

KVM: PPC: Book3S HV: add XIVE native exploitation mode

PATCH Superseded Cédric Le Goater 5 participants 71 messages Feb 22, 2019 – Mar 15, 2019

Subsystems (4)

v2 of 6 Superseded Latest is v6 →

lore
Share & tools
lore
#0 lore raw

v2 cover KVM: PPC: Book3S HV: add XIVE native exploitation mode

Hello,

On the POWER9 processor, the XIVE interrupt controller can control interrupt sources using MMIOs to trigger events, to EOI or to turn off the sources. Priority management and interrupt acknowledgment is also controlled by MMIO in the CPU presenter subengine.

PowerNV/baremetal Linux runs natively under XIVE but sPAPR guests need special support from the hypervisor to do the same. This is called the XIVE native exploitation mode and today, it can be activated under the PowerPC Hypervisor, pHyp. However, Linux/KVM lacks XIVE native support and still offers the old interrupt mode interface using a KVM device implementing the XICS hcalls over XIVE.

The following series is proposal to add the same support under KVM.

A new KVM device is introduced for the XIVE native exploitation mode. It reuses most of the XICS-over-XIVE glue implementation structures which are internal to KVM but has a completely different interface. A set of KVM device ioctls provide support for the hypervisor calls, all handled in QEMU, to configure the sources and the event queues. From there, all interrupt control is transferred to the guest which can use MMIOs.

These MMIO regions (ESB and TIMA) are exposed to guests in QEMU, similarly to VFIO, and the associated VMAs are populated dynamically with the appropriate pages using a fault handler. These are now implemented using mmap()s of the KVM device fd.

Migration has its own specific needs regarding memory. The patchset provides a specific control to quiesce XIVE before capturing the memory. The save and restore of the internal state is based on the same ioctls used for the hcalls.

On a POWER9 sPAPR machine, the Client Architecture Support (CAS) negotiation process determines whether the guest operates with a interrupt controller using the XICS legacy model, as found on POWER8, or in XIVE exploitation mode. Which means that the KVM interrupt device should be created at runtime, after the machine has started. This requires extra support from KVM to destroy KVM devices. It is introduced at the end of the patcshet as it still requires some attention and a XIVE-only VM would not need.

This is 5.2 material. I expect a couple of respin for fixes, and the OPAL patches have not yet been merged.

GitHub trees available here :

QEMU sPAPR:

https://github.com/legoater/qemu/commits/xive-next

Linux/KVM:

https://github.com/legoater/linux/commits/xive-5.0

OPAL:

https://github.com/legoater/skiboot/commits/xive

Thanks,

C.

Changes since v1:

- Better documentation (was missing)
- Nested support. XIVE not advertised on non PowerNV platforms. This is a good way to test the fallback on QEMU emulated devices.
- ESB and TIMA special mapping done using the KVM device fd
- All hcalls moved to QEMU. Dropped the patch moving the hcall flags.
- Reworked of the KVM device ioctl controls to support hcalls and migration needs to capture/save states
- Merged the control syncing XIVE and marking the EQ page dirty
- Fixed passthrough support using the KVM device file address_space to clear the ESB pages from the mapping
- Misc enhancements and fixes

Cédric Le Goater (16):
powerpc/xive: add OPAL extensions for the XIVE native exploitation
support
KVM: PPC: Book3S HV: add a new KVM device for the XIVE native
exploitation mode
KVM: PPC: Book3S HV: XIVE: introduce a new capability
KVM_CAP_PPC_IRQ_XIVE
KVM: PPC: Book3S HV: XIVE: add a control to initialize a source
KVM: PPC: Book3S HV: XIVE: add a control to configure a source
KVM: PPC: Book3S HV: XIVE: add controls for the EQ configuration
KVM: PPC: Book3S HV: XIVE: add a global reset control
KVM: PPC: Book3S HV: XIVE: add a control to sync the sources
KVM: PPC: Book3S HV: XIVE: add a control to dirty the XIVE EQ pages
KVM: PPC: Book3S HV: XIVE: add get/set accessors for the VP XIVE state
KVM: introduce a 'mmap' method for KVM devices
KVM: PPC: Book3S HV: XIVE: add a TIMA mapping
KVM: PPC: Book3S HV: XIVE: add a mapping for the source ESB pages
KVM: PPC: Book3S HV: XIVE: add passthrough support
KVM: introduce a KVM_DESTROY_DEVICE ioctl
KVM: PPC: Book3S HV: XIVE: clear the vCPU interrupt presenters

 arch/powerpc/include/asm/kvm_host.h           |    2 +
 arch/powerpc/include/asm/kvm_ppc.h            |   32 +
 arch/powerpc/include/asm/opal-api.h           |   11 +-
 arch/powerpc/include/asm/opal.h               |    7 +
 arch/powerpc/include/asm/xive.h               |   17 +
 arch/powerpc/include/uapi/asm/kvm.h           |   48 +
 arch/powerpc/kvm/book3s_xive.h                |   31 +
 include/linux/kvm_host.h                      |    1 +
 include/uapi/linux/kvm.h                      |   10 +
 arch/powerpc/kvm/book3s.c                     |   31 +-
 arch/powerpc/kvm/book3s_xics.c                |   19 +
 arch/powerpc/kvm/book3s_xive.c                |  149 ++-
 arch/powerpc/kvm/book3s_xive_native.c         | 1171 +++++++++++++++++
 arch/powerpc/kvm/powerpc.c                    |   33 +
 arch/powerpc/sysdev/xive/native.c             |  110 ++
 virt/kvm/kvm_main.c                           |   49 +
 Documentation/virtual/kvm/api.txt             |   28 +
 Documentation/virtual/kvm/devices/xive.txt    |  190 +++
 arch/powerpc/kvm/Makefile                     |    2 +-
 .../powerpc/platforms/powernv/opal-wrappers.S |    3 +
 20 files changed, 1896 insertions(+), 48 deletions(-)
 create mode 100644 arch/powerpc/kvm/book3s_xive_native.c
 create mode 100644 Documentation/virtual/kvm/devices/xive.txt
Signature · 1 line
2.20.1
#1 lore raw

PATCH v2 1/16 powerpc/xive: add OPAL extensions for the XIVE native exploitation support

The support for XIVE native exploitation mode in Linux/KVM needs a couple more OPAL calls to configure the sPAPR guest and to get/set the state of the XIVE internal structures.

---
 arch/powerpc/include/asm/opal-api.h           | 11 ++-
 arch/powerpc/include/asm/opal.h               |  7 ++
 arch/powerpc/include/asm/xive.h               | 14 +++
 arch/powerpc/sysdev/xive/native.c             | 99 +++++++++++++++++++
 .../powerpc/platforms/powernv/opal-wrappers.S |  3 +
 5 files changed, 130 insertions(+), 4 deletions(-)
arch/powerpc/include/asm/opal-api.h +7 -4
index 870fb7b239ea..cdfc54f78101 100644
@@ -186,8 +186,8 @@
 #define OPAL_XIVE_FREE_IRQ			140
 #define OPAL_XIVE_SYNC				141
 #define OPAL_XIVE_DUMP				142
-#define OPAL_XIVE_RESERVED3			143
-#define OPAL_XIVE_RESERVED4			144
+#define OPAL_XIVE_GET_QUEUE_STATE		143
+#define OPAL_XIVE_SET_QUEUE_STATE		144
 #define OPAL_SIGNAL_SYSTEM_RESET		145
 #define OPAL_NPU_INIT_CONTEXT			146
 #define OPAL_NPU_DESTROY_CONTEXT		147
@@ -209,8 +209,11 @@
 #define OPAL_SENSOR_GROUP_ENABLE		163
 #define OPAL_PCI_GET_PBCQ_TUNNEL_BAR		164
 #define OPAL_PCI_SET_PBCQ_TUNNEL_BAR		165
-#define	OPAL_NX_COPROC_INIT			167
-#define OPAL_LAST				167
+#define OPAL_HANDLE_HMI2			166
+#define OPAL_NX_COPROC_INIT			167
+#define OPAL_NPU_SET_RELAXED_ORDER		168
+#define OPAL_NPU_GET_RELAXED_ORDER		169
+#define OPAL_XIVE_GET_VP_STATE			170
 
 #define QUIESCE_HOLD			1 /* Spin all calls at entry */
 #define QUIESCE_REJECT			2 /* Fail all calls with OPAL_BUSY */
arch/powerpc/include/asm/opal.h +7 -0
index a55b01c90bb1..4e978d4dea5c 100644
@@ -279,6 +279,13 @@ int64_t opal_xive_allocate_irq(uint32_t chip_id);
 int64_t opal_xive_free_irq(uint32_t girq);
 int64_t opal_xive_sync(uint32_t type, uint32_t id);
 int64_t opal_xive_dump(uint32_t type, uint32_t id);
+int64_t opal_xive_get_queue_state(uint64_t vp, uint32_t prio,
+				  __be32 *out_qtoggle,
+				  __be32 *out_qindex);
+int64_t opal_xive_set_queue_state(uint64_t vp, uint32_t prio,
+				  uint32_t qtoggle,
+				  uint32_t qindex);
+int64_t opal_xive_get_vp_state(uint64_t vp, __be64 *out_w01);
 int64_t opal_pci_set_p2p(uint64_t phb_init, uint64_t phb_target,
 			uint64_t desc, uint16_t pe_number);
 
arch/powerpc/include/asm/xive.h +14 -0
index 3c704f5dd3ae..b579a943407b 100644
@@ -109,12 +109,26 @@ extern int xive_native_configure_queue(u32 vp_id, struct xive_q *q, u8 prio,
 extern void xive_native_disable_queue(u32 vp_id, struct xive_q *q, u8 prio);
 
 extern void xive_native_sync_source(u32 hw_irq);
+extern void xive_native_sync_queue(u32 hw_irq);
 extern bool is_xive_irq(struct irq_chip *chip);
 extern int xive_native_enable_vp(u32 vp_id, bool single_escalation);
 extern int xive_native_disable_vp(u32 vp_id);
 extern int xive_native_get_vp_info(u32 vp_id, u32 *out_cam_id, u32 *out_chip_id);
 extern bool xive_native_has_single_escalation(void);
 
+extern int xive_native_get_queue_info(u32 vp_id, uint32_t prio,
+				      u64 *out_qpage,
+				      u64 *out_qsize,
+				      u64 *out_qeoi_page,
+				      u32 *out_escalate_irq,
+				      u64 *out_qflags);
+
+extern int xive_native_get_queue_state(u32 vp_id, uint32_t prio, u32 *qtoggle,
+				       u32 *qindex);
+extern int xive_native_set_queue_state(u32 vp_id, uint32_t prio, u32 qtoggle,
+				       u32 qindex);
+extern int xive_native_get_vp_state(u32 vp_id, u64 *out_state);
+
 #else
 
 static inline bool xive_enabled(void) { return false; }
arch/powerpc/sysdev/xive/native.c +99 -0

108 lines not on this page · raw message

arch/powerpc/platforms/powernv/opal-wrappers.S +3 -0
index f4875fe3f8ff..3179953d6b56 100644
@@ -309,6 +309,9 @@ OPAL_CALL(opal_xive_get_vp_info,		OPAL_XIVE_GET_VP_INFO);
 OPAL_CALL(opal_xive_set_vp_info,		OPAL_XIVE_SET_VP_INFO);
 OPAL_CALL(opal_xive_sync,			OPAL_XIVE_SYNC);
 OPAL_CALL(opal_xive_dump,			OPAL_XIVE_DUMP);
+OPAL_CALL(opal_xive_get_queue_state,		OPAL_XIVE_GET_QUEUE_STATE);
+OPAL_CALL(opal_xive_set_queue_state,		OPAL_XIVE_SET_QUEUE_STATE);
+OPAL_CALL(opal_xive_get_vp_state,		OPAL_XIVE_GET_VP_STATE);
 OPAL_CALL(opal_signal_system_reset,		OPAL_SIGNAL_SYSTEM_RESET);
 OPAL_CALL(opal_npu_init_context,		OPAL_NPU_INIT_CONTEXT);
 OPAL_CALL(opal_npu_destroy_context,		OPAL_NPU_DESTROY_CONTEXT);
Signature · 1 line
2.20.1
Reviewed-by on v2 1/16
#2 lore raw

Re: [PATCH v2 01/16] powerpc/xive: add OPAL extensions for the XIVE native exploitation support

Quoted 5 lines · Cédric Le Goater

The support for XIVE native exploitation mode in Linux/KVM needs a couple more OPAL calls to configure the sPAPR guest and to get/set the state of the XIVE internal structures.

Quoted 214 lines
---
 arch/powerpc/include/asm/opal-api.h           | 11 ++-
 arch/powerpc/include/asm/opal.h               |  7 ++
 arch/powerpc/include/asm/xive.h               | 14 +++
 arch/powerpc/sysdev/xive/native.c             | 99 +++++++++++++++++++
 .../powerpc/platforms/powernv/opal-wrappers.S |  3 +
 5 files changed, 130 insertions(+), 4 deletions(-)
arch/powerpc/include/asm/opal-api.h +7 -4
index 870fb7b239ea..cdfc54f78101 100644
@@ -186,8 +186,8 @@
 #define OPAL_XIVE_FREE_IRQ			140
 #define OPAL_XIVE_SYNC				141
 #define OPAL_XIVE_DUMP				142
-#define OPAL_XIVE_RESERVED3			143
-#define OPAL_XIVE_RESERVED4			144
+#define OPAL_XIVE_GET_QUEUE_STATE		143
+#define OPAL_XIVE_SET_QUEUE_STATE		144
 #define OPAL_SIGNAL_SYSTEM_RESET		145
 #define OPAL_NPU_INIT_CONTEXT			146
 #define OPAL_NPU_DESTROY_CONTEXT		147
@@ -209,8 +209,11 @@
 #define OPAL_SENSOR_GROUP_ENABLE		163
 #define OPAL_PCI_GET_PBCQ_TUNNEL_BAR		164
 #define OPAL_PCI_SET_PBCQ_TUNNEL_BAR		165
-#define	OPAL_NX_COPROC_INIT			167
-#define OPAL_LAST				167
+#define OPAL_HANDLE_HMI2			166
+#define OPAL_NX_COPROC_INIT			167
+#define OPAL_NPU_SET_RELAXED_ORDER		168
+#define OPAL_NPU_GET_RELAXED_ORDER		169
+#define OPAL_XIVE_GET_VP_STATE			170
 
 #define QUIESCE_HOLD			1 /* Spin all calls at entry */
 #define QUIESCE_REJECT			2 /* Fail all calls with OPAL_BUSY */
arch/powerpc/include/asm/opal.h +7 -0
index a55b01c90bb1..4e978d4dea5c 100644
@@ -279,6 +279,13 @@ int64_t opal_xive_allocate_irq(uint32_t chip_id);
 int64_t opal_xive_free_irq(uint32_t girq);
 int64_t opal_xive_sync(uint32_t type, uint32_t id);
 int64_t opal_xive_dump(uint32_t type, uint32_t id);
+int64_t opal_xive_get_queue_state(uint64_t vp, uint32_t prio,
+				  __be32 *out_qtoggle,
+				  __be32 *out_qindex);
+int64_t opal_xive_set_queue_state(uint64_t vp, uint32_t prio,
+				  uint32_t qtoggle,
+				  uint32_t qindex);
+int64_t opal_xive_get_vp_state(uint64_t vp, __be64 *out_w01);
 int64_t opal_pci_set_p2p(uint64_t phb_init, uint64_t phb_target,
 			uint64_t desc, uint16_t pe_number);
 
arch/powerpc/include/asm/xive.h +14 -0
index 3c704f5dd3ae..b579a943407b 100644
@@ -109,12 +109,26 @@ extern int xive_native_configure_queue(u32 vp_id, struct xive_q *q, u8 prio,
 extern void xive_native_disable_queue(u32 vp_id, struct xive_q *q, u8 prio);
 
 extern void xive_native_sync_source(u32 hw_irq);
+extern void xive_native_sync_queue(u32 hw_irq);
 extern bool is_xive_irq(struct irq_chip *chip);
 extern int xive_native_enable_vp(u32 vp_id, bool single_escalation);
 extern int xive_native_disable_vp(u32 vp_id);
 extern int xive_native_get_vp_info(u32 vp_id, u32 *out_cam_id, u32 *out_chip_id);
 extern bool xive_native_has_single_escalation(void);
 
+extern int xive_native_get_queue_info(u32 vp_id, uint32_t prio,
+				      u64 *out_qpage,
+				      u64 *out_qsize,
+				      u64 *out_qeoi_page,
+				      u32 *out_escalate_irq,
+				      u64 *out_qflags);
+
+extern int xive_native_get_queue_state(u32 vp_id, uint32_t prio, u32 *qtoggle,
+				       u32 *qindex);
+extern int xive_native_set_queue_state(u32 vp_id, uint32_t prio, u32 qtoggle,
+				       u32 qindex);
+extern int xive_native_get_vp_state(u32 vp_id, u64 *out_state);
+
 #else
 
 static inline bool xive_enabled(void) { return false; }
arch/powerpc/sysdev/xive/native.c +99 -0
index 1ca127d052a6..0c037e933e55 100644
@@ -437,6 +437,12 @@ void xive_native_sync_source(u32 hw_irq)
 }
 EXPORT_SYMBOL_GPL(xive_native_sync_source);
 
+void xive_native_sync_queue(u32 hw_irq)
+{
+	opal_xive_sync(XIVE_SYNC_QUEUE, hw_irq);
+}
+EXPORT_SYMBOL_GPL(xive_native_sync_queue);
+
 static const struct xive_ops xive_native_ops = {
 	.populate_irq_data	= xive_native_populate_irq_data,
 	.configure_irq		= xive_native_configure_irq,
@@ -711,3 +717,96 @@ bool xive_native_has_single_escalation(void)
 	return xive_has_single_esc;
 }
 EXPORT_SYMBOL_GPL(xive_native_has_single_escalation);
+
+int xive_native_get_queue_info(u32 vp_id, u32 prio,
+			       u64 *out_qpage,
+			       u64 *out_qsize,
+			       u64 *out_qeoi_page,
+			       u32 *out_escalate_irq,
+			       u64 *out_qflags)
+{
+	__be64 qpage;
+	__be64 qsize;
+	__be64 qeoi_page;
+	__be32 escalate_irq;
+	__be64 qflags;
+	s64 rc;
+
+	rc = opal_xive_get_queue_info(vp_id, prio, &qpage, &qsize,
+				      &qeoi_page, &escalate_irq, &qflags);
+	if (rc) {
+		pr_err("OPAL failed to get queue info for VCPU %d/%d : %lld\n",
+		       vp_id, prio, rc);
+		return -EIO;
+	}
+
+	if (out_qpage)
+		*out_qpage = be64_to_cpu(qpage);
+	if (out_qsize)
+		*out_qsize = be32_to_cpu(qsize);
+	if (out_qeoi_page)
+		*out_qeoi_page = be64_to_cpu(qeoi_page);
+	if (out_escalate_irq)
+		*out_escalate_irq = be32_to_cpu(escalate_irq);
+	if (out_qflags)
+		*out_qflags = be64_to_cpu(qflags);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(xive_native_get_queue_info);
+
+int xive_native_get_queue_state(u32 vp_id, u32 prio, u32 *qtoggle, u32 *qindex)
+{
+	__be32 opal_qtoggle;
+	__be32 opal_qindex;
+	s64 rc;
+
+	rc = opal_xive_get_queue_state(vp_id, prio, &opal_qtoggle,
+				       &opal_qindex);
+	if (rc) {
+		pr_err("OPAL failed to get queue state for VCPU %d/%d : %lld\n",
+		       vp_id, prio, rc);
+		return -EIO;
+	}
+
+	if (qtoggle)
+		*qtoggle = be32_to_cpu(opal_qtoggle);
+	if (qindex)
+		*qindex = be32_to_cpu(opal_qindex);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(xive_native_get_queue_state);
+
+int xive_native_set_queue_state(u32 vp_id, u32 prio, u32 qtoggle, u32 qindex)
+{
+	s64 rc;
+
+	rc = opal_xive_set_queue_state(vp_id, prio, qtoggle, qindex);
+	if (rc) {
+		pr_err("OPAL failed to set queue state for VCPU %d/%d : %lld\n",
+		       vp_id, prio, rc);
+		return -EIO;
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(xive_native_set_queue_state);
+
+int xive_native_get_vp_state(u32 vp_id, u64 *out_state)
+{
+	__be64 state;
+	s64 rc;
+
+	rc = opal_xive_get_vp_state(vp_id, &state);
+	if (rc) {
+		pr_err("OPAL failed to get vp state for VCPU %d : %lld\n",
+		       vp_id, rc);
+		return -EIO;
+	}
+
+	if (out_state)
+		*out_state = be64_to_cpu(state);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(xive_native_get_vp_state);
arch/powerpc/platforms/powernv/opal-wrappers.S +3 -0
index f4875fe3f8ff..3179953d6b56 100644
@@ -309,6 +309,9 @@ OPAL_CALL(opal_xive_get_vp_info,		OPAL_XIVE_GET_VP_INFO);
 OPAL_CALL(opal_xive_set_vp_info,		OPAL_XIVE_SET_VP_INFO);
 OPAL_CALL(opal_xive_sync,			OPAL_XIVE_SYNC);
 OPAL_CALL(opal_xive_dump,			OPAL_XIVE_DUMP);
+OPAL_CALL(opal_xive_get_queue_state,		OPAL_XIVE_GET_QUEUE_STATE);
+OPAL_CALL(opal_xive_set_queue_state,		OPAL_XIVE_SET_QUEUE_STATE);
+OPAL_CALL(opal_xive_get_vp_state,		OPAL_XIVE_GET_VP_STATE);
 OPAL_CALL(opal_signal_system_reset,		OPAL_SIGNAL_SYSTEM_RESET);
 OPAL_CALL(opal_npu_init_context,		OPAL_NPU_INIT_CONTEXT);
 OPAL_CALL(opal_npu_destroy_context,		OPAL_NPU_DESTROY_CONTEXT);
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#3 lore raw

Re: [PATCH v2 01/16] powerpc/xive: add OPAL extensions for the XIVE native exploitation support

In reply to Cédric Le Goater
Quoted 29 lines · Cédric Le Goater

The support for XIVE native exploitation mode in Linux/KVM needs a couple more OPAL calls to configure the sPAPR guest and to get/set the state of the XIVE internal structures.

---
 arch/powerpc/include/asm/opal-api.h           | 11 ++-
 arch/powerpc/include/asm/opal.h               |  7 ++
 arch/powerpc/include/asm/xive.h               | 14 +++
 arch/powerpc/sysdev/xive/native.c             | 99 +++++++++++++++++++
 .../powerpc/platforms/powernv/opal-wrappers.S |  3 +
 5 files changed, 130 insertions(+), 4 deletions(-)
arch/powerpc/include/asm/opal-api.h +2 -2
index 870fb7b239ea..cdfc54f78101 100644
@@ -186,8 +186,8 @@
 #define OPAL_XIVE_FREE_IRQ			140
 #define OPAL_XIVE_SYNC				141
 #define OPAL_XIVE_DUMP				142
-#define OPAL_XIVE_RESERVED3			143
-#define OPAL_XIVE_RESERVED4			144
+#define OPAL_XIVE_GET_QUEUE_STATE		143
+#define OPAL_XIVE_SET_QUEUE_STATE		144
 #define OPAL_SIGNAL_SYSTEM_RESET		145
 #define OPAL_NPU_INIT_CONTEXT			146
 #define OPAL_NPU_DESTROY_CONTEXT		147

arch/powerpc/include/asm/opal-api.h line 216

Quoted 10 lines
arch/powerpc/include/asm/opal-api.h +5 -2
@@ -209,8 +209,11 @@
 #define OPAL_SENSOR_GROUP_ENABLE		163
 #define OPAL_PCI_GET_PBCQ_TUNNEL_BAR		164
 #define OPAL_PCI_SET_PBCQ_TUNNEL_BAR		165
-#define	OPAL_NX_COPROC_INIT			167
-#define OPAL_LAST				167
+#define OPAL_HANDLE_HMI2			166
+#define OPAL_NX_COPROC_INIT			167
+#define OPAL_NPU_SET_RELAXED_ORDER		168
+#define OPAL_NPU_GET_RELAXED_ORDER		169
+#define OPAL_XIVE_GET_VP_STATE			170

You should only be defining the calls you need, leaving gaps for other things, and you need to retain OPAL_LAST. So it should look more like:

-#define OPAL_LAST 167
+#define OPAL_XIVE_GET_VP_STATE 170
+#define OPAL_LAST 170

Also I can't merge this until it's merged into skiboot.

cheers

#4 lore raw

Re: [PATCH v2 01/16] powerpc/xive: add OPAL extensions for the XIVE native exploitation support

Quoted 45 lines · Michael Ellerman
Quoted 29 lines · Cédric Le Goater

The support for XIVE native exploitation mode in Linux/KVM needs a couple more OPAL calls to configure the sPAPR guest and to get/set the state of the XIVE internal structures.

---
 arch/powerpc/include/asm/opal-api.h           | 11 ++-
 arch/powerpc/include/asm/opal.h               |  7 ++
 arch/powerpc/include/asm/xive.h               | 14 +++
 arch/powerpc/sysdev/xive/native.c             | 99 +++++++++++++++++++
 .../powerpc/platforms/powernv/opal-wrappers.S |  3 +
 5 files changed, 130 insertions(+), 4 deletions(-)
arch/powerpc/include/asm/opal-api.h +2 -2
index 870fb7b239ea..cdfc54f78101 100644
@@ -186,8 +186,8 @@
 #define OPAL_XIVE_FREE_IRQ			140
 #define OPAL_XIVE_SYNC				141
 #define OPAL_XIVE_DUMP				142
-#define OPAL_XIVE_RESERVED3			143
-#define OPAL_XIVE_RESERVED4			144
+#define OPAL_XIVE_GET_QUEUE_STATE		143
+#define OPAL_XIVE_SET_QUEUE_STATE		144
 #define OPAL_SIGNAL_SYSTEM_RESET		145
 #define OPAL_NPU_INIT_CONTEXT			146
 #define OPAL_NPU_DESTROY_CONTEXT		147

arch/powerpc/include/asm/opal-api.h line 216

Quoted 10 lines
arch/powerpc/include/asm/opal-api.h +5 -2
@@ -209,8 +209,11 @@
 #define OPAL_SENSOR_GROUP_ENABLE		163
 #define OPAL_PCI_GET_PBCQ_TUNNEL_BAR		164
 #define OPAL_PCI_SET_PBCQ_TUNNEL_BAR		165
-#define	OPAL_NX_COPROC_INIT			167
-#define OPAL_LAST				167
+#define OPAL_HANDLE_HMI2			166
+#define OPAL_NX_COPROC_INIT			167
+#define OPAL_NPU_SET_RELAXED_ORDER		168
+#define OPAL_NPU_GET_RELAXED_ORDER		169
+#define OPAL_XIVE_GET_VP_STATE			170

You should only be defining the calls you need, leaving gaps for other things, and you need to retain OPAL_LAST. So it should look more like:

Quoted 6 lines

-#define OPAL_LAST 167
+#define OPAL_XIVE_GET_VP_STATE 170
+#define OPAL_LAST 170

Also I can't merge this until it's merged into skiboot.

OK. Let's start with skiboot.

C.

#5 lore raw

Re: [PATCH v2 01/16] powerpc/xive: add OPAL extensions for the XIVE native exploitation support

Quoted 46 lines · Cédric Le Goater
Quoted 45 lines · Michael Ellerman
Quoted 29 lines · Cédric Le Goater

The support for XIVE native exploitation mode in Linux/KVM needs a couple more OPAL calls to configure the sPAPR guest and to get/set the state of the XIVE internal structures.

---
 arch/powerpc/include/asm/opal-api.h           | 11 ++-
 arch/powerpc/include/asm/opal.h               |  7 ++
 arch/powerpc/include/asm/xive.h               | 14 +++
 arch/powerpc/sysdev/xive/native.c             | 99 +++++++++++++++++++
 .../powerpc/platforms/powernv/opal-wrappers.S |  3 +
 5 files changed, 130 insertions(+), 4 deletions(-)
arch/powerpc/include/asm/opal-api.h +2 -2
index 870fb7b239ea..cdfc54f78101 100644
@@ -186,8 +186,8 @@
 #define OPAL_XIVE_FREE_IRQ			140
 #define OPAL_XIVE_SYNC				141
 #define OPAL_XIVE_DUMP				142
-#define OPAL_XIVE_RESERVED3			143
-#define OPAL_XIVE_RESERVED4			144
+#define OPAL_XIVE_GET_QUEUE_STATE		143
+#define OPAL_XIVE_SET_QUEUE_STATE		144
 #define OPAL_SIGNAL_SYSTEM_RESET		145
 #define OPAL_NPU_INIT_CONTEXT			146
 #define OPAL_NPU_DESTROY_CONTEXT		147

arch/powerpc/include/asm/opal-api.h line 216

Quoted 10 lines
arch/powerpc/include/asm/opal-api.h +5 -2
@@ -209,8 +209,11 @@
 #define OPAL_SENSOR_GROUP_ENABLE		163
 #define OPAL_PCI_GET_PBCQ_TUNNEL_BAR		164
 #define OPAL_PCI_SET_PBCQ_TUNNEL_BAR		165
-#define	OPAL_NX_COPROC_INIT			167
-#define OPAL_LAST				167
+#define OPAL_HANDLE_HMI2			166
+#define OPAL_NX_COPROC_INIT			167
+#define OPAL_NPU_SET_RELAXED_ORDER		168
+#define OPAL_NPU_GET_RELAXED_ORDER		169
+#define OPAL_XIVE_GET_VP_STATE			170

You should only be defining the calls you need, leaving gaps for other things, and you need to retain OPAL_LAST. So it should look more like:

Quoted 8 lines
Quoted 6 lines

-#define OPAL_LAST 167
+#define OPAL_XIVE_GET_VP_STATE 170
+#define OPAL_LAST 170

Also I can't merge this until it's merged into skiboot.

OK. Let's start with skiboot.

Yeah.. where are we at with skiboot in general. We can't test this downstream until we have a released skiboot with the necessary support.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#6 lore raw

Re: [PATCH v2 01/16] powerpc/xive: add OPAL extensions for the XIVE native exploitation support

Quoted 52 lines · David Gibson
Quoted 51 lines · Cédric Le Goater
Quoted 50 lines · Michael Ellerman
Quoted 29 lines · Cédric Le Goater

The support for XIVE native exploitation mode in Linux/KVM needs a couple more OPAL calls to configure the sPAPR guest and to get/set the state of the XIVE internal structures.

---
 arch/powerpc/include/asm/opal-api.h           | 11 ++-
 arch/powerpc/include/asm/opal.h               |  7 ++
 arch/powerpc/include/asm/xive.h               | 14 +++
 arch/powerpc/sysdev/xive/native.c             | 99 +++++++++++++++++++
 .../powerpc/platforms/powernv/opal-wrappers.S |  3 +
 5 files changed, 130 insertions(+), 4 deletions(-)
arch/powerpc/include/asm/opal-api.h +2 -2
index 870fb7b239ea..cdfc54f78101 100644
@@ -186,8 +186,8 @@
 #define OPAL_XIVE_FREE_IRQ			140
 #define OPAL_XIVE_SYNC				141
 #define OPAL_XIVE_DUMP				142
-#define OPAL_XIVE_RESERVED3			143
-#define OPAL_XIVE_RESERVED4			144
+#define OPAL_XIVE_GET_QUEUE_STATE		143
+#define OPAL_XIVE_SET_QUEUE_STATE		144
 #define OPAL_SIGNAL_SYSTEM_RESET		145
 #define OPAL_NPU_INIT_CONTEXT			146
 #define OPAL_NPU_DESTROY_CONTEXT		147

arch/powerpc/include/asm/opal-api.h line 216

Quoted 10 lines
arch/powerpc/include/asm/opal-api.h +5 -2
@@ -209,8 +209,11 @@
 #define OPAL_SENSOR_GROUP_ENABLE		163
 #define OPAL_PCI_GET_PBCQ_TUNNEL_BAR		164
 #define OPAL_PCI_SET_PBCQ_TUNNEL_BAR		165
-#define	OPAL_NX_COPROC_INIT			167
-#define OPAL_LAST				167
+#define OPAL_HANDLE_HMI2			166
+#define OPAL_NX_COPROC_INIT			167
+#define OPAL_NPU_SET_RELAXED_ORDER		168
+#define OPAL_NPU_GET_RELAXED_ORDER		169
+#define OPAL_XIVE_GET_VP_STATE			170

You should only be defining the calls you need, leaving gaps for other things, and you need to retain OPAL_LAST. So it should look more like:

-#define OPAL_LAST 167
+#define OPAL_XIVE_GET_VP_STATE 170
+#define OPAL_LAST 170

Quoted 7 lines
Quoted 3 lines
Quoted 1 line

Also I can't merge this until it's merged into skiboot.

OK. Let's start with skiboot.

Yeah.. where are we at with skiboot in general. We can't test this downstream until we have a released skiboot with the necessary support.

If we add a flag to remove the OPAL call when setting the EQ, you could test, without migration though.

C.

#7 lore raw

PATCH v2 2/16 KVM: PPC: Book3S HV: add a new KVM device for the XIVE native exploitation mode

In reply to Cédric Le Goater

This is the basic framework for the new KVM device supporting the XIVE native exploitation mode. The user interface exposes a new KVM device to be created by QEMU when running on a L0 hypervisor only. Support for nested guests is not available yet.

---
 arch/powerpc/include/asm/kvm_host.h        |   1 +
 arch/powerpc/include/asm/kvm_ppc.h         |   8 +
 arch/powerpc/include/uapi/asm/kvm.h        |   3 +
 include/uapi/linux/kvm.h                   |   2 +
 arch/powerpc/kvm/book3s.c                  |   7 +-
 arch/powerpc/kvm/book3s_xive_native.c      | 191 +++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  19 ++
 arch/powerpc/kvm/Makefile                  |   2 +-
 8 files changed, 231 insertions(+), 2 deletions(-)
 create mode 100644 arch/powerpc/kvm/book3s_xive_native.c
 create mode 100644 Documentation/virtual/kvm/devices/xive.txt
arch/powerpc/include/asm/kvm_host.h +1 -0
index 091430339db1..9f75a75a07f2 100644
@@ -220,6 +220,7 @@ extern struct kvm_device_ops kvm_xics_ops;
 struct kvmppc_xive;
 struct kvmppc_xive_vcpu;
 extern struct kvm_device_ops kvm_xive_ops;
+extern struct kvm_device_ops kvm_xive_native_ops;
 
 struct kvmppc_passthru_irqmap;
 
arch/powerpc/include/asm/kvm_ppc.h +8 -0
index b3bf4f61b30c..4b72ddde7dc1 100644
@@ -593,6 +593,10 @@ extern int kvmppc_xive_set_icp(struct kvm_vcpu *vcpu, u64 icpval);
 extern int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 irq,
 			       int level, bool line_status);
 extern void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu);
+
+extern void kvmppc_xive_native_init_module(void);
+extern void kvmppc_xive_native_exit_module(void);
+
 #else
 static inline int kvmppc_xive_set_xive(struct kvm *kvm, u32 irq, u32 server,
 				       u32 priority) { return -1; }
@@ -616,6 +620,10 @@ static inline int kvmppc_xive_set_icp(struct kvm_vcpu *vcpu, u64 icpval) { retur
 static inline int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 irq,
 				      int level, bool line_status) { return -ENODEV; }
 static inline void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu) { }
+
+static inline void kvmppc_xive_native_init_module(void) { }
+static inline void kvmppc_xive_native_exit_module(void) { }
+
 #endif /* CONFIG_KVM_XIVE */
 
 #ifdef CONFIG_PPC_POWERNV
arch/powerpc/include/uapi/asm/kvm.h +3 -0
index 8c876c166ef2..b002c0c67787 100644
@@ -675,4 +675,7 @@ struct kvm_ppc_cpu_char {
 #define  KVM_XICS_PRESENTED		(1ULL << 43)
 #define  KVM_XICS_QUEUED		(1ULL << 44)
 
+/* POWER9 XIVE Native Interrupt Controller */
+#define KVM_DEV_XIVE_GRP_CTRL		1
+
 #endif /* __LINUX_KVM_POWERPC_H */
include/uapi/linux/kvm.h +2 -0
index 6d4ea4b6c922..e6368163d3a0 100644
@@ -1211,6 +1211,8 @@ enum kvm_device_type {
 #define KVM_DEV_TYPE_ARM_VGIC_V3	KVM_DEV_TYPE_ARM_VGIC_V3
 	KVM_DEV_TYPE_ARM_VGIC_ITS,
 #define KVM_DEV_TYPE_ARM_VGIC_ITS	KVM_DEV_TYPE_ARM_VGIC_ITS
+	KVM_DEV_TYPE_XIVE,
+#define KVM_DEV_TYPE_XIVE		KVM_DEV_TYPE_XIVE
 	KVM_DEV_TYPE_MAX,
 };
 
arch/powerpc/kvm/book3s.c +6 -1
index 601c094f15ab..96d43f091255 100644
@@ -1040,6 +1040,9 @@ static int kvmppc_book3s_init(void)
 	if (xics_on_xive()) {
 		kvmppc_xive_init_module();
 		kvm_register_device_ops(&kvm_xive_ops, KVM_DEV_TYPE_XICS);
+		kvmppc_xive_native_init_module();
+		kvm_register_device_ops(&kvm_xive_native_ops,
+					KVM_DEV_TYPE_XIVE);
 	} else
 #endif
 		kvm_register_device_ops(&kvm_xics_ops, KVM_DEV_TYPE_XICS);
@@ -1050,8 +1053,10 @@ static int kvmppc_book3s_init(void)
 static void kvmppc_book3s_exit(void)
 {
 #ifdef CONFIG_KVM_XICS
-	if (xics_on_xive())
+	if (xics_on_xive()) {
 		kvmppc_xive_exit_module();
+		kvmppc_xive_native_exit_module();
+	}
 #endif
 #ifdef CONFIG_KVM_BOOK3S_32_HANDLER
 	kvmppc_book3s_exit_pr();
arch/powerpc/kvm/book3s_xive_native.c new file +191 -0

191 lines not on this page · raw message

Documentation/virtual/kvm/devices/xive.txt new file +19 -0
new file mode 100644
index 000000000000..fdbd2ff92a88
@@ -0,0 +1,19 @@
+POWER9 eXternal Interrupt Virtualization Engine (XIVE Gen1)
+==========================================================
+
+Device types supported:
+  KVM_DEV_TYPE_XIVE     POWER9 XIVE Interrupt Controller generation 1
+
+This device acts as a VM interrupt controller. It provides the KVM
+interface to configure the interrupt sources of a VM in the underlying
+POWER9 XIVE interrupt controller.
+
+Only one XIVE instance may be instantiated. A guest XIVE device
+requires a POWER9 host and the guest OS should have support for the
+XIVE native exploitation interrupt mode. If not, it should run using
+the legacy interrupt mode, referred as XICS (POWER7/8).
+
+* Groups:
+
+  1. KVM_DEV_XIVE_GRP_CTRL
+  Provides global controls on the device
arch/powerpc/kvm/Makefile +1 -1
index 64f1135e7732..806cbe488410 100644
@@ -99,7 +99,7 @@ endif
 kvm-book3s_64-objs-$(CONFIG_KVM_XICS) += \
 	book3s_xics.o
 
-kvm-book3s_64-objs-$(CONFIG_KVM_XIVE) += book3s_xive.o
+kvm-book3s_64-objs-$(CONFIG_KVM_XIVE) += book3s_xive.o book3s_xive_native.o
 kvm-book3s_64-objs-$(CONFIG_SPAPR_TCE_IOMMU) += book3s_64_vio.o
 
 kvm-book3s_64-module-objs := \
Signature · 1 line
2.20.1
#8 lore raw

Re: [PATCH v2 02/16] KVM: PPC: Book3S HV: add a new KVM device for the XIVE native exploitation mode

Quoted 203 lines · Cédric Le Goater

This is the basic framework for the new KVM device supporting the XIVE native exploitation mode. The user interface exposes a new KVM device to be created by QEMU when running on a L0 hypervisor only. Support for nested guests is not available yet.

---
 arch/powerpc/include/asm/kvm_host.h        |   1 +
 arch/powerpc/include/asm/kvm_ppc.h         |   8 +
 arch/powerpc/include/uapi/asm/kvm.h        |   3 +
 include/uapi/linux/kvm.h                   |   2 +
 arch/powerpc/kvm/book3s.c                  |   7 +-
 arch/powerpc/kvm/book3s_xive_native.c      | 191 +++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  19 ++
 arch/powerpc/kvm/Makefile                  |   2 +-
 8 files changed, 231 insertions(+), 2 deletions(-)
 create mode 100644 arch/powerpc/kvm/book3s_xive_native.c
 create mode 100644 Documentation/virtual/kvm/devices/xive.txt
arch/powerpc/include/asm/kvm_host.h +1 -0
index 091430339db1..9f75a75a07f2 100644
@@ -220,6 +220,7 @@ extern struct kvm_device_ops kvm_xics_ops;
 struct kvmppc_xive;
 struct kvmppc_xive_vcpu;
 extern struct kvm_device_ops kvm_xive_ops;
+extern struct kvm_device_ops kvm_xive_native_ops;
 
 struct kvmppc_passthru_irqmap;
 
arch/powerpc/include/asm/kvm_ppc.h +8 -0
index b3bf4f61b30c..4b72ddde7dc1 100644
@@ -593,6 +593,10 @@ extern int kvmppc_xive_set_icp(struct kvm_vcpu *vcpu, u64 icpval);
 extern int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 irq,
 			       int level, bool line_status);
 extern void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu);
+
+extern void kvmppc_xive_native_init_module(void);
+extern void kvmppc_xive_native_exit_module(void);
+
 #else
 static inline int kvmppc_xive_set_xive(struct kvm *kvm, u32 irq, u32 server,
 				       u32 priority) { return -1; }
@@ -616,6 +620,10 @@ static inline int kvmppc_xive_set_icp(struct kvm_vcpu *vcpu, u64 icpval) { retur
 static inline int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 irq,
 				      int level, bool line_status) { return -ENODEV; }
 static inline void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu) { }
+
+static inline void kvmppc_xive_native_init_module(void) { }
+static inline void kvmppc_xive_native_exit_module(void) { }
+
 #endif /* CONFIG_KVM_XIVE */
 
 #ifdef CONFIG_PPC_POWERNV
arch/powerpc/include/uapi/asm/kvm.h +3 -0
index 8c876c166ef2..b002c0c67787 100644
@@ -675,4 +675,7 @@ struct kvm_ppc_cpu_char {
 #define  KVM_XICS_PRESENTED		(1ULL << 43)
 #define  KVM_XICS_QUEUED		(1ULL << 44)
 
+/* POWER9 XIVE Native Interrupt Controller */
+#define KVM_DEV_XIVE_GRP_CTRL		1
+
 #endif /* __LINUX_KVM_POWERPC_H */
include/uapi/linux/kvm.h +2 -0
index 6d4ea4b6c922..e6368163d3a0 100644
@@ -1211,6 +1211,8 @@ enum kvm_device_type {
 #define KVM_DEV_TYPE_ARM_VGIC_V3	KVM_DEV_TYPE_ARM_VGIC_V3
 	KVM_DEV_TYPE_ARM_VGIC_ITS,
 #define KVM_DEV_TYPE_ARM_VGIC_ITS	KVM_DEV_TYPE_ARM_VGIC_ITS
+	KVM_DEV_TYPE_XIVE,
+#define KVM_DEV_TYPE_XIVE		KVM_DEV_TYPE_XIVE
 	KVM_DEV_TYPE_MAX,
 };
 
arch/powerpc/kvm/book3s.c +6 -1
index 601c094f15ab..96d43f091255 100644
@@ -1040,6 +1040,9 @@ static int kvmppc_book3s_init(void)
 	if (xics_on_xive()) {
 		kvmppc_xive_init_module();
 		kvm_register_device_ops(&kvm_xive_ops, KVM_DEV_TYPE_XICS);
+		kvmppc_xive_native_init_module();
+		kvm_register_device_ops(&kvm_xive_native_ops,
+					KVM_DEV_TYPE_XIVE);
 	} else
 #endif
 		kvm_register_device_ops(&kvm_xics_ops, KVM_DEV_TYPE_XICS);
@@ -1050,8 +1053,10 @@ static int kvmppc_book3s_init(void)
 static void kvmppc_book3s_exit(void)
 {
 #ifdef CONFIG_KVM_XICS
-	if (xics_on_xive())
+	if (xics_on_xive()) {
 		kvmppc_xive_exit_module();
+		kvmppc_xive_native_exit_module();
+	}
 #endif
 #ifdef CONFIG_KVM_BOOK3S_32_HANDLER
 	kvmppc_book3s_exit_pr();
arch/powerpc/kvm/book3s_xive_native.c new file +89 -0
new file mode 100644
index 000000000000..e475ce83ad14
@@ -0,0 +1,191 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2017-2019, IBM Corporation.
+ */
+
+#define pr_fmt(fmt) "xive-kvm: " fmt
+
+#include <linux/anon_inodes.h>
+#include <linux/kernel.h>
+#include <linux/kvm_host.h>
+#include <linux/err.h>
+#include <linux/gfp.h>
+#include <linux/spinlock.h>
+#include <linux/delay.h>
+#include <linux/percpu.h>
+#include <linux/cpumask.h>
+#include <asm/uaccess.h>
+#include <asm/kvm_book3s.h>
+#include <asm/kvm_ppc.h>
+#include <asm/hvcall.h>
+#include <asm/xics.h>
+#include <asm/xive.h>
+#include <asm/xive-regs.h>
+#include <asm/debug.h>
+#include <asm/debugfs.h>
+#include <asm/time.h>
+#include <asm/opal.h>
+
+#include <linux/debugfs.h>
+#include <linux/seq_file.h>
+
+#include "book3s_xive.h"
+
+static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
+				       struct kvm_device_attr *attr)
+{
+	switch (attr->group) {
+	case KVM_DEV_XIVE_GRP_CTRL:
+		break;
+	}
+	return -ENXIO;
+}
+
+static int kvmppc_xive_native_get_attr(struct kvm_device *dev,
+				       struct kvm_device_attr *attr)
+{
+	return -ENXIO;
+}
+
+static int kvmppc_xive_native_has_attr(struct kvm_device *dev,
+				       struct kvm_device_attr *attr)
+{
+	switch (attr->group) {
+	case KVM_DEV_XIVE_GRP_CTRL:
+		break;
+	}
+	return -ENXIO;
+}
+
+static void kvmppc_xive_native_free(struct kvm_device *dev)
+{
+	struct kvmppc_xive *xive = dev->private;
+	struct kvm *kvm = xive->kvm;
+
+	debugfs_remove(xive->dentry);
+
+	pr_devel("Destroying xive native device\n");
+
+	if (kvm)
+		kvm->arch.xive = NULL;
+
+	if (xive->vp_base != XIVE_INVALID_VP)
+		xive_native_free_vp_block(xive->vp_base);
+
+	kfree(xive);
+	kfree(dev);
+}
+
+static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type)
+{
+	struct kvmppc_xive *xive;
+	struct kvm *kvm = dev->kvm;
+	int ret = 0;
+
+	pr_devel("Creating xive native device\n");
+
+	if (kvm->arch.xive)
+		return -EEXIST;
+

arch/powerpc/kvm/book3s_xive_native.c line 99

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c new file +10 -0
+	xive = kzalloc(sizeof(*xive), GFP_KERNEL);
+	if (!xive)
+		return -ENOMEM;
+
+	dev->private = xive;
+	xive->dev = dev;
+	xive->kvm = kvm;
+	kvm->arch.xive = xive;
+
+	/* We use the default queue size set by the host */

IIUC the queue is examined directly by the guest, so the guest must know its size. In which case letting the host decide the size would be a problem for migration.

Quoted 130 lines
arch/powerpc/kvm/book3s_xive_native.c new file +92 -0
+	xive->q_order = xive_native_default_eq_shift();
+	if (xive->q_order < PAGE_SHIFT)
+		xive->q_page_order = 0;
+	else
+		xive->q_page_order = xive->q_order - PAGE_SHIFT;
+
+	/*
+	 * Allocate a bunch of VPs. KVM_MAX_VCPUS is a large value for
+	 * a default. Getting the max number of CPUs the VM was
+	 * configured with would improve our usage of the XIVE VP space.
+	 */
+	xive->vp_base = xive_native_alloc_vp_block(KVM_MAX_VCPUS);
+	pr_devel("VP_Base=%x\n", xive->vp_base);
+
+	if (xive->vp_base == XIVE_INVALID_VP)
+		ret = -ENOMEM;
+
+	xive->single_escalation = xive_native_has_single_escalation();
+
+	if (ret)
+		kfree(xive);
+
+	return ret;
+}
+
+static int xive_native_debug_show(struct seq_file *m, void *private)
+{
+	struct kvmppc_xive *xive = m->private;
+	struct kvm *kvm = xive->kvm;
+
+	if (!kvm)
+		return 0;
+
+	return 0;
+}
+
+static int xive_native_debug_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, xive_native_debug_show, inode->i_private);
+}
+
+static const struct file_operations xive_native_debug_fops = {
+	.open = xive_native_debug_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = single_release,
+};
+
+static void xive_native_debugfs_init(struct kvmppc_xive *xive)
+{
+	char *name;
+
+	name = kasprintf(GFP_KERNEL, "kvm-xive-%p", xive);
+	if (!name) {
+		pr_err("%s: no memory for name\n", __func__);
+		return;
+	}
+
+	xive->dentry = debugfs_create_file(name, 0444, powerpc_debugfs_root,
+					   xive, &xive_native_debug_fops);
+
+	pr_debug("%s: created %s\n", __func__, name);
+	kfree(name);
+}
+
+static void kvmppc_xive_native_init(struct kvm_device *dev)
+{
+	struct kvmppc_xive *xive = (struct kvmppc_xive *)dev->private;
+
+	/* Register some debug interfaces */
+	xive_native_debugfs_init(xive);
+}
+
+struct kvm_device_ops kvm_xive_native_ops = {
+	.name = "kvm-xive-native",
+	.create = kvmppc_xive_native_create,
+	.init = kvmppc_xive_native_init,
+	.destroy = kvmppc_xive_native_free,
+	.set_attr = kvmppc_xive_native_set_attr,
+	.get_attr = kvmppc_xive_native_get_attr,
+	.has_attr = kvmppc_xive_native_has_attr,
+};
+
+void kvmppc_xive_native_init_module(void)
+{
+	;
+}
+
+void kvmppc_xive_native_exit_module(void)
+{
+	;
+}
Documentation/virtual/kvm/devices/xive.txt new file +19 -0
new file mode 100644
index 000000000000..fdbd2ff92a88
@@ -0,0 +1,19 @@
+POWER9 eXternal Interrupt Virtualization Engine (XIVE Gen1)
+==========================================================
+
+Device types supported:
+  KVM_DEV_TYPE_XIVE     POWER9 XIVE Interrupt Controller generation 1
+
+This device acts as a VM interrupt controller. It provides the KVM
+interface to configure the interrupt sources of a VM in the underlying
+POWER9 XIVE interrupt controller.
+
+Only one XIVE instance may be instantiated. A guest XIVE device
+requires a POWER9 host and the guest OS should have support for the
+XIVE native exploitation interrupt mode. If not, it should run using
+the legacy interrupt mode, referred as XICS (POWER7/8).
+
+* Groups:
+
+  1. KVM_DEV_XIVE_GRP_CTRL
+  Provides global controls on the device
arch/powerpc/kvm/Makefile +1 -1
index 64f1135e7732..806cbe488410 100644
@@ -99,7 +99,7 @@ endif
 kvm-book3s_64-objs-$(CONFIG_KVM_XICS) += \
 	book3s_xics.o
 
-kvm-book3s_64-objs-$(CONFIG_KVM_XIVE) += book3s_xive.o
+kvm-book3s_64-objs-$(CONFIG_KVM_XIVE) += book3s_xive.o book3s_xive_native.o
 kvm-book3s_64-objs-$(CONFIG_SPAPR_TCE_IOMMU) += book3s_64_vio.o
 
 kvm-book3s_64-module-objs := \
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#9 lore raw

Re: [PATCH v2 02/16] KVM: PPC: Book3S HV: add a new KVM device for the XIVE native exploitation mode

Quoted 210 lines · David Gibson
Quoted 209 lines · Cédric Le Goater

This is the basic framework for the new KVM device supporting the XIVE native exploitation mode. The user interface exposes a new KVM device to be created by QEMU when running on a L0 hypervisor only. Support for nested guests is not available yet.

---
 arch/powerpc/include/asm/kvm_host.h        |   1 +
 arch/powerpc/include/asm/kvm_ppc.h         |   8 +
 arch/powerpc/include/uapi/asm/kvm.h        |   3 +
 include/uapi/linux/kvm.h                   |   2 +
 arch/powerpc/kvm/book3s.c                  |   7 +-
 arch/powerpc/kvm/book3s_xive_native.c      | 191 +++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  19 ++
 arch/powerpc/kvm/Makefile                  |   2 +-
 8 files changed, 231 insertions(+), 2 deletions(-)
 create mode 100644 arch/powerpc/kvm/book3s_xive_native.c
 create mode 100644 Documentation/virtual/kvm/devices/xive.txt
arch/powerpc/include/asm/kvm_host.h +1 -0
index 091430339db1..9f75a75a07f2 100644
@@ -220,6 +220,7 @@ extern struct kvm_device_ops kvm_xics_ops;
 struct kvmppc_xive;
 struct kvmppc_xive_vcpu;
 extern struct kvm_device_ops kvm_xive_ops;
+extern struct kvm_device_ops kvm_xive_native_ops;
 
 struct kvmppc_passthru_irqmap;
 
arch/powerpc/include/asm/kvm_ppc.h +8 -0
index b3bf4f61b30c..4b72ddde7dc1 100644
@@ -593,6 +593,10 @@ extern int kvmppc_xive_set_icp(struct kvm_vcpu *vcpu, u64 icpval);
 extern int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 irq,
 			       int level, bool line_status);
 extern void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu);
+
+extern void kvmppc_xive_native_init_module(void);
+extern void kvmppc_xive_native_exit_module(void);
+
 #else
 static inline int kvmppc_xive_set_xive(struct kvm *kvm, u32 irq, u32 server,
 				       u32 priority) { return -1; }
@@ -616,6 +620,10 @@ static inline int kvmppc_xive_set_icp(struct kvm_vcpu *vcpu, u64 icpval) { retur
 static inline int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 irq,
 				      int level, bool line_status) { return -ENODEV; }
 static inline void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu) { }
+
+static inline void kvmppc_xive_native_init_module(void) { }
+static inline void kvmppc_xive_native_exit_module(void) { }
+
 #endif /* CONFIG_KVM_XIVE */
 
 #ifdef CONFIG_PPC_POWERNV
arch/powerpc/include/uapi/asm/kvm.h +3 -0
index 8c876c166ef2..b002c0c67787 100644
@@ -675,4 +675,7 @@ struct kvm_ppc_cpu_char {
 #define  KVM_XICS_PRESENTED		(1ULL << 43)
 #define  KVM_XICS_QUEUED		(1ULL << 44)
 
+/* POWER9 XIVE Native Interrupt Controller */
+#define KVM_DEV_XIVE_GRP_CTRL		1
+
 #endif /* __LINUX_KVM_POWERPC_H */
include/uapi/linux/kvm.h +2 -0
index 6d4ea4b6c922..e6368163d3a0 100644
@@ -1211,6 +1211,8 @@ enum kvm_device_type {
 #define KVM_DEV_TYPE_ARM_VGIC_V3	KVM_DEV_TYPE_ARM_VGIC_V3
 	KVM_DEV_TYPE_ARM_VGIC_ITS,
 #define KVM_DEV_TYPE_ARM_VGIC_ITS	KVM_DEV_TYPE_ARM_VGIC_ITS
+	KVM_DEV_TYPE_XIVE,
+#define KVM_DEV_TYPE_XIVE		KVM_DEV_TYPE_XIVE
 	KVM_DEV_TYPE_MAX,
 };
 
arch/powerpc/kvm/book3s.c +6 -1
index 601c094f15ab..96d43f091255 100644
@@ -1040,6 +1040,9 @@ static int kvmppc_book3s_init(void)
 	if (xics_on_xive()) {
 		kvmppc_xive_init_module();
 		kvm_register_device_ops(&kvm_xive_ops, KVM_DEV_TYPE_XICS);
+		kvmppc_xive_native_init_module();
+		kvm_register_device_ops(&kvm_xive_native_ops,
+					KVM_DEV_TYPE_XIVE);
 	} else
 #endif
 		kvm_register_device_ops(&kvm_xics_ops, KVM_DEV_TYPE_XICS);
@@ -1050,8 +1053,10 @@ static int kvmppc_book3s_init(void)
 static void kvmppc_book3s_exit(void)
 {
 #ifdef CONFIG_KVM_XICS
-	if (xics_on_xive())
+	if (xics_on_xive()) {
 		kvmppc_xive_exit_module();
+		kvmppc_xive_native_exit_module();
+	}
 #endif
 #ifdef CONFIG_KVM_BOOK3S_32_HANDLER
 	kvmppc_book3s_exit_pr();
arch/powerpc/kvm/book3s_xive_native.c new file +95 -0
new file mode 100644
index 000000000000..e475ce83ad14
@@ -0,0 +1,191 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2017-2019, IBM Corporation.
+ */
+
+#define pr_fmt(fmt) "xive-kvm: " fmt
+
+#include <linux/anon_inodes.h>
+#include <linux/kernel.h>
+#include <linux/kvm_host.h>
+#include <linux/err.h>
+#include <linux/gfp.h>
+#include <linux/spinlock.h>
+#include <linux/delay.h>
+#include <linux/percpu.h>
+#include <linux/cpumask.h>
+#include <asm/uaccess.h>
+#include <asm/kvm_book3s.h>
+#include <asm/kvm_ppc.h>
+#include <asm/hvcall.h>
+#include <asm/xics.h>
+#include <asm/xive.h>
+#include <asm/xive-regs.h>
+#include <asm/debug.h>
+#include <asm/debugfs.h>
+#include <asm/time.h>
+#include <asm/opal.h>
+
+#include <linux/debugfs.h>
+#include <linux/seq_file.h>
+
+#include "book3s_xive.h"
+
+static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
+				       struct kvm_device_attr *attr)
+{
+	switch (attr->group) {
+	case KVM_DEV_XIVE_GRP_CTRL:
+		break;
+	}
+	return -ENXIO;
+}
+
+static int kvmppc_xive_native_get_attr(struct kvm_device *dev,
+				       struct kvm_device_attr *attr)
+{
+	return -ENXIO;
+}
+
+static int kvmppc_xive_native_has_attr(struct kvm_device *dev,
+				       struct kvm_device_attr *attr)
+{
+	switch (attr->group) {
+	case KVM_DEV_XIVE_GRP_CTRL:
+		break;
+	}
+	return -ENXIO;
+}
+
+static void kvmppc_xive_native_free(struct kvm_device *dev)
+{
+	struct kvmppc_xive *xive = dev->private;
+	struct kvm *kvm = xive->kvm;
+
+	debugfs_remove(xive->dentry);
+
+	pr_devel("Destroying xive native device\n");
+
+	if (kvm)
+		kvm->arch.xive = NULL;
+
+	if (xive->vp_base != XIVE_INVALID_VP)
+		xive_native_free_vp_block(xive->vp_base);
+
+	kfree(xive);
+	kfree(dev);
+}
+
+static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type)
+{
+	struct kvmppc_xive *xive;
+	struct kvm *kvm = dev->kvm;
+	int ret = 0;
+
+	pr_devel("Creating xive native device\n");
+
+	if (kvm->arch.xive)
+		return -EEXIST;
+
+	xive = kzalloc(sizeof(*xive), GFP_KERNEL);
+	if (!xive)
+		return -ENOMEM;
+
+	dev->private = xive;
+	xive->dev = dev;
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c line 99

Quoted 4 lines
arch/powerpc/kvm/book3s_xive_native.c new file +4 -0
+	xive->kvm = kvm;
+	kvm->arch.xive = xive;
+
+	/* We use the default queue size set by the host */

IIUC the queue is examined directly by the guest, so the guest must know its size. In which case letting the host decide the size would be a problem for migration.

yes. This is a left over from the XICS-over-XIVE KVM device. I will remove the code, we don't use it.

Thanks,

C.

Quoted 131 lines
Quoted 130 lines
arch/powerpc/kvm/book3s_xive_native.c new file +92 -0
+	xive->q_order = xive_native_default_eq_shift();
+	if (xive->q_order < PAGE_SHIFT)
+		xive->q_page_order = 0;
+	else
+		xive->q_page_order = xive->q_order - PAGE_SHIFT;
+
+	/*
+	 * Allocate a bunch of VPs. KVM_MAX_VCPUS is a large value for
+	 * a default. Getting the max number of CPUs the VM was
+	 * configured with would improve our usage of the XIVE VP space.
+	 */
+	xive->vp_base = xive_native_alloc_vp_block(KVM_MAX_VCPUS);
+	pr_devel("VP_Base=%x\n", xive->vp_base);
+
+	if (xive->vp_base == XIVE_INVALID_VP)
+		ret = -ENOMEM;
+
+	xive->single_escalation = xive_native_has_single_escalation();
+
+	if (ret)
+		kfree(xive);
+
+	return ret;
+}
+
+static int xive_native_debug_show(struct seq_file *m, void *private)
+{
+	struct kvmppc_xive *xive = m->private;
+	struct kvm *kvm = xive->kvm;
+
+	if (!kvm)
+		return 0;
+
+	return 0;
+}
+
+static int xive_native_debug_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, xive_native_debug_show, inode->i_private);
+}
+
+static const struct file_operations xive_native_debug_fops = {
+	.open = xive_native_debug_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = single_release,
+};
+
+static void xive_native_debugfs_init(struct kvmppc_xive *xive)
+{
+	char *name;
+
+	name = kasprintf(GFP_KERNEL, "kvm-xive-%p", xive);
+	if (!name) {
+		pr_err("%s: no memory for name\n", __func__);
+		return;
+	}
+
+	xive->dentry = debugfs_create_file(name, 0444, powerpc_debugfs_root,
+					   xive, &xive_native_debug_fops);
+
+	pr_debug("%s: created %s\n", __func__, name);
+	kfree(name);
+}
+
+static void kvmppc_xive_native_init(struct kvm_device *dev)
+{
+	struct kvmppc_xive *xive = (struct kvmppc_xive *)dev->private;
+
+	/* Register some debug interfaces */
+	xive_native_debugfs_init(xive);
+}
+
+struct kvm_device_ops kvm_xive_native_ops = {
+	.name = "kvm-xive-native",
+	.create = kvmppc_xive_native_create,
+	.init = kvmppc_xive_native_init,
+	.destroy = kvmppc_xive_native_free,
+	.set_attr = kvmppc_xive_native_set_attr,
+	.get_attr = kvmppc_xive_native_get_attr,
+	.has_attr = kvmppc_xive_native_has_attr,
+};
+
+void kvmppc_xive_native_init_module(void)
+{
+	;
+}
+
+void kvmppc_xive_native_exit_module(void)
+{
+	;
+}
Documentation/virtual/kvm/devices/xive.txt new file +19 -0
new file mode 100644
index 000000000000..fdbd2ff92a88
@@ -0,0 +1,19 @@
+POWER9 eXternal Interrupt Virtualization Engine (XIVE Gen1)
+==========================================================
+
+Device types supported:
+  KVM_DEV_TYPE_XIVE     POWER9 XIVE Interrupt Controller generation 1
+
+This device acts as a VM interrupt controller. It provides the KVM
+interface to configure the interrupt sources of a VM in the underlying
+POWER9 XIVE interrupt controller.
+
+Only one XIVE instance may be instantiated. A guest XIVE device
+requires a POWER9 host and the guest OS should have support for the
+XIVE native exploitation interrupt mode. If not, it should run using
+the legacy interrupt mode, referred as XICS (POWER7/8).
+
+* Groups:
+
+  1. KVM_DEV_XIVE_GRP_CTRL
+  Provides global controls on the device
arch/powerpc/kvm/Makefile +1 -1
index 64f1135e7732..806cbe488410 100644
@@ -99,7 +99,7 @@ endif
 kvm-book3s_64-objs-$(CONFIG_KVM_XICS) += \
 	book3s_xics.o
 
-kvm-book3s_64-objs-$(CONFIG_KVM_XIVE) += book3s_xive.o
+kvm-book3s_64-objs-$(CONFIG_KVM_XIVE) += book3s_xive.o book3s_xive_native.o
 kvm-book3s_64-objs-$(CONFIG_SPAPR_TCE_IOMMU) += book3s_64_vio.o
 
 kvm-book3s_64-module-objs := \
#10 lore raw

PATCH v2 3/16 KVM: PPC: Book3S HV: XIVE: introduce a new capability KVM_CAP_PPC_IRQ_XIVE

In reply to Cédric Le Goater

The user interface exposes a new capability to let QEMU connect the vCPU to the XIVE KVM device if required. The capability is only advertised on a PowerNV Hypervisor as support for nested guests (pseries KVM Hypervisor) is not yet available.

Internally, the interface to the new KVM device is protected with a new interrupt mode: KVMPPC_IRQ_XIVE.

---
 arch/powerpc/include/asm/kvm_host.h   |   1 +
 arch/powerpc/include/asm/kvm_ppc.h    |  13 +++
 arch/powerpc/kvm/book3s_xive.h        |   6 ++
 include/uapi/linux/kvm.h              |   1 +
 arch/powerpc/kvm/book3s_xive.c        |  67 +++++++-----
 arch/powerpc/kvm/book3s_xive_native.c | 144 ++++++++++++++++++++++++++
 arch/powerpc/kvm/powerpc.c            |  33 ++++++
 Documentation/virtual/kvm/api.txt     |   9 ++
 8 files changed, 246 insertions(+), 28 deletions(-)
arch/powerpc/include/asm/kvm_host.h +1 -0
index 9f75a75a07f2..eb8581be0ee8 100644
@@ -448,6 +448,7 @@ struct kvmppc_passthru_irqmap {
 #define KVMPPC_IRQ_DEFAULT	0
 #define KVMPPC_IRQ_MPIC		1
 #define KVMPPC_IRQ_XICS		2 /* Includes a XIVE option */
+#define KVMPPC_IRQ_XIVE		3 /* XIVE native exploitation mode */
 
 #define MMIO_HPTE_CACHE_SIZE	4
 
arch/powerpc/include/asm/kvm_ppc.h +13 -0
index 4b72ddde7dc1..1e61877fe147 100644
@@ -594,6 +594,14 @@ extern int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 irq,
 			       int level, bool line_status);
 extern void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu);
 
+static inline int kvmppc_xive_enabled(struct kvm_vcpu *vcpu)
+{
+	return vcpu->arch.irq_type == KVMPPC_IRQ_XIVE;
+}
+
+extern int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
+					   struct kvm_vcpu *vcpu, u32 cpu);
+extern void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu);
 extern void kvmppc_xive_native_init_module(void);
 extern void kvmppc_xive_native_exit_module(void);
 
@@ -621,6 +629,11 @@ static inline int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 ir
 				      int level, bool line_status) { return -ENODEV; }
 static inline void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu) { }
 
+static inline int kvmppc_xive_enabled(struct kvm_vcpu *vcpu)
+	{ return 0; }
+static inline int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
+			  struct kvm_vcpu *vcpu, u32 cpu) { return -EBUSY; }
+static inline void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu) { }
 static inline void kvmppc_xive_native_init_module(void) { }
 static inline void kvmppc_xive_native_exit_module(void) { }
 
arch/powerpc/kvm/book3s_xive.h +6 -0
index a08ae6fd4c51..bcb1bbcf0359 100644
@@ -248,5 +248,11 @@ extern int (*__xive_vm_h_ipi)(struct kvm_vcpu *vcpu, unsigned long server,
 extern int (*__xive_vm_h_cppr)(struct kvm_vcpu *vcpu, unsigned long cppr);
 extern int (*__xive_vm_h_eoi)(struct kvm_vcpu *vcpu, unsigned long xirr);
 
+/*
+ * Common Xive routines for XICS-over-XIVE and XIVE native
+ */
+void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu);
+int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu);
+
 #endif /* CONFIG_KVM_XICS */
 #endif /* _KVM_PPC_BOOK3S_XICS_H */
include/uapi/linux/kvm.h +1 -0
index e6368163d3a0..52bf74a1616e 100644
@@ -988,6 +988,7 @@ struct kvm_ppc_resize_hpt {
 #define KVM_CAP_ARM_VM_IPA_SIZE 165
 #define KVM_CAP_MANUAL_DIRTY_LOG_PROTECT 166
 #define KVM_CAP_HYPERV_CPUID 167
+#define KVM_CAP_PPC_IRQ_XIVE 168
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
arch/powerpc/kvm/book3s_xive.c +39 -28

92 lines not on this page · raw message

arch/powerpc/kvm/book3s_xive_native.c +144 -0

160 lines not on this page · raw message

arch/powerpc/kvm/powerpc.c +33 -0
index 8c69af10f91d..a38a643a24dd 100644
@@ -570,6 +570,12 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 	case KVM_CAP_PPC_GET_CPU_CHAR:
 		r = 1;
 		break;
+#ifdef CONFIG_KVM_XIVE
+	case KVM_CAP_PPC_IRQ_XIVE:
+		/* only for PowerNV */
+		r = !!cpu_has_feature(CPU_FTR_HVMODE);
+		break;
+#endif
 
 	case KVM_CAP_PPC_ALLOC_HTAB:
 		r = hv_enabled;
@@ -753,6 +759,9 @@ void kvm_arch_vcpu_free(struct kvm_vcpu *vcpu)
 		else
 			kvmppc_xics_free_icp(vcpu);
 		break;
+	case KVMPPC_IRQ_XIVE:
+		kvmppc_xive_native_cleanup_vcpu(vcpu);
+		break;
 	}
 
 	kvmppc_core_vcpu_free(vcpu);
@@ -1941,6 +1950,30 @@ static int kvm_vcpu_ioctl_enable_cap(struct kvm_vcpu *vcpu,
 		break;
 	}
 #endif /* CONFIG_KVM_XICS */
+#ifdef CONFIG_KVM_XIVE
+	case KVM_CAP_PPC_IRQ_XIVE: {
+		struct fd f;
+		struct kvm_device *dev;
+
+		r = -EBADF;
+		f = fdget(cap->args[0]);
+		if (!f.file)
+			break;
+
+		r = -ENXIO;
+		if (!xive_enabled())
+			break;
+
+		r = -EPERM;
+		dev = kvm_device_from_filp(f.file);
+		if (dev)
+			r = kvmppc_xive_native_connect_vcpu(dev, vcpu,
+							    cap->args[1]);
+
+		fdput(f);
+		break;
+	}
+#endif /* CONFIG_KVM_XIVE */
 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
 	case KVM_CAP_PPC_FWNMI:
 		r = -EINVAL;
Documentation/virtual/kvm/api.txt +9 -0
index 356156f5c52d..1db1435769b4 100644
@@ -4458,6 +4458,15 @@ struct kvm_sync_regs {
         struct kvm_vcpu_events events;
 };
 
+6.75 KVM_CAP_PPC_IRQ_XIVE
+
+Architectures: ppc
+Target: vcpu
+Parameters: args[0] is the XIVE device fd
+            args[1] is the XIVE CPU number (server ID) for this vcpu
+
+This capability connects the vcpu to an in-kernel XIVE device.
+
 7. Capabilities that can be enabled on VMs
 ------------------------------------------
 
Signature · 1 line
2.20.1
#11 lore raw

Re: [PATCH v2 03/16] KVM: PPC: Book3S HV: XIVE: introduce a new capability KVM_CAP_PPC_IRQ_XIVE

Quoted 261 lines · Cédric Le Goater

The user interface exposes a new capability to let QEMU connect the vCPU to the XIVE KVM device if required. The capability is only advertised on a PowerNV Hypervisor as support for nested guests (pseries KVM Hypervisor) is not yet available.

Internally, the interface to the new KVM device is protected with a new interrupt mode: KVMPPC_IRQ_XIVE.

---
 arch/powerpc/include/asm/kvm_host.h   |   1 +
 arch/powerpc/include/asm/kvm_ppc.h    |  13 +++
 arch/powerpc/kvm/book3s_xive.h        |   6 ++
 include/uapi/linux/kvm.h              |   1 +
 arch/powerpc/kvm/book3s_xive.c        |  67 +++++++-----
 arch/powerpc/kvm/book3s_xive_native.c | 144 ++++++++++++++++++++++++++
 arch/powerpc/kvm/powerpc.c            |  33 ++++++
 Documentation/virtual/kvm/api.txt     |   9 ++
 8 files changed, 246 insertions(+), 28 deletions(-)
arch/powerpc/include/asm/kvm_host.h +1 -0
index 9f75a75a07f2..eb8581be0ee8 100644
@@ -448,6 +448,7 @@ struct kvmppc_passthru_irqmap {
 #define KVMPPC_IRQ_DEFAULT	0
 #define KVMPPC_IRQ_MPIC		1
 #define KVMPPC_IRQ_XICS		2 /* Includes a XIVE option */
+#define KVMPPC_IRQ_XIVE		3 /* XIVE native exploitation mode */
 
 #define MMIO_HPTE_CACHE_SIZE	4
 
arch/powerpc/include/asm/kvm_ppc.h +13 -0
index 4b72ddde7dc1..1e61877fe147 100644
@@ -594,6 +594,14 @@ extern int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 irq,
 			       int level, bool line_status);
 extern void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu);
 
+static inline int kvmppc_xive_enabled(struct kvm_vcpu *vcpu)
+{
+	return vcpu->arch.irq_type == KVMPPC_IRQ_XIVE;
+}
+
+extern int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
+					   struct kvm_vcpu *vcpu, u32 cpu);
+extern void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu);
 extern void kvmppc_xive_native_init_module(void);
 extern void kvmppc_xive_native_exit_module(void);
 
@@ -621,6 +629,11 @@ static inline int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 ir
 				      int level, bool line_status) { return -ENODEV; }
 static inline void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu) { }
 
+static inline int kvmppc_xive_enabled(struct kvm_vcpu *vcpu)
+	{ return 0; }
+static inline int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
+			  struct kvm_vcpu *vcpu, u32 cpu) { return -EBUSY; }
+static inline void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu) { }
 static inline void kvmppc_xive_native_init_module(void) { }
 static inline void kvmppc_xive_native_exit_module(void) { }
 
arch/powerpc/kvm/book3s_xive.h +6 -0
index a08ae6fd4c51..bcb1bbcf0359 100644
@@ -248,5 +248,11 @@ extern int (*__xive_vm_h_ipi)(struct kvm_vcpu *vcpu, unsigned long server,
 extern int (*__xive_vm_h_cppr)(struct kvm_vcpu *vcpu, unsigned long cppr);
 extern int (*__xive_vm_h_eoi)(struct kvm_vcpu *vcpu, unsigned long xirr);
 
+/*
+ * Common Xive routines for XICS-over-XIVE and XIVE native
+ */
+void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu);
+int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu);
+
 #endif /* CONFIG_KVM_XICS */
 #endif /* _KVM_PPC_BOOK3S_XICS_H */
include/uapi/linux/kvm.h +1 -0
index e6368163d3a0..52bf74a1616e 100644
@@ -988,6 +988,7 @@ struct kvm_ppc_resize_hpt {
 #define KVM_CAP_ARM_VM_IPA_SIZE 165
 #define KVM_CAP_MANUAL_DIRTY_LOG_PROTECT 166
 #define KVM_CAP_HYPERV_CPUID 167
+#define KVM_CAP_PPC_IRQ_XIVE 168
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
arch/powerpc/kvm/book3s_xive.c +39 -28
index f78d002f0fe0..d1cc18a5b1c4 100644
@@ -1049,7 +1049,7 @@ int kvmppc_xive_clr_mapped(struct kvm *kvm, unsigned long guest_irq,
 }
 EXPORT_SYMBOL_GPL(kvmppc_xive_clr_mapped);
 
-static void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu)
+void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
 	struct kvm *kvm = vcpu->kvm;
@@ -1883,6 +1883,43 @@ static int kvmppc_xive_create(struct kvm_device *dev, u32 type)
 	return 0;
 }
 
+int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	unsigned int i;
+
+	for (i = 0; i < KVMPPC_XIVE_Q_COUNT; i++) {
+		struct xive_q *q = &xc->queues[i];
+		u32 i0, i1, idx;
+
+		if (!q->qpage && !xc->esc_virq[i])
+			continue;
+
+		seq_printf(m, " [q%d]: ", i);
+
+		if (q->qpage) {
+			idx = q->idx;
+			i0 = be32_to_cpup(q->qpage + idx);
+			idx = (idx + 1) & q->msk;
+			i1 = be32_to_cpup(q->qpage + idx);
+			seq_printf(m, "T=%d %08x %08x...\n", q->toggle,
+				   i0, i1);
+		}
+		if (xc->esc_virq[i]) {
+			struct irq_data *d = irq_get_irq_data(xc->esc_virq[i]);
+			struct xive_irq_data *xd =
+				irq_data_get_irq_handler_data(d);
+			u64 pq = xive_vm_esb_load(xd, XIVE_ESB_GET);
+
+			seq_printf(m, "E:%c%c I(%d:%llx:%llx)",
+				   (pq & XIVE_ESB_VAL_P) ? 'P' : 'p',
+				   (pq & XIVE_ESB_VAL_Q) ? 'Q' : 'q',
+				   xc->esc_virq[i], pq, xd->eoi_page);
+			seq_puts(m, "\n");
+		}
+	}
+	return 0;
+}
 
 static int xive_debug_show(struct seq_file *m, void *private)
 {
@@ -1908,7 +1945,6 @@ static int xive_debug_show(struct seq_file *m, void *private)
 
 	kvm_for_each_vcpu(i, vcpu, kvm) {
 		struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
-		unsigned int i;
 
 		if (!xc)
 			continue;
@@ -1918,33 +1954,8 @@ static int xive_debug_show(struct seq_file *m, void *private)
 			   xc->server_num, xc->cppr, xc->hw_cppr,
 			   xc->mfrr, xc->pending,
 			   xc->stat_rm_h_xirr, xc->stat_vm_h_xirr);
-		for (i = 0; i < KVMPPC_XIVE_Q_COUNT; i++) {
-			struct xive_q *q = &xc->queues[i];
-			u32 i0, i1, idx;
 
-			if (!q->qpage && !xc->esc_virq[i])
-				continue;
-
-			seq_printf(m, " [q%d]: ", i);
-
-			if (q->qpage) {
-				idx = q->idx;
-				i0 = be32_to_cpup(q->qpage + idx);
-				idx = (idx + 1) & q->msk;
-				i1 = be32_to_cpup(q->qpage + idx);
-				seq_printf(m, "T=%d %08x %08x... \n", q->toggle, i0, i1);
-			}
-			if (xc->esc_virq[i]) {
-				struct irq_data *d = irq_get_irq_data(xc->esc_virq[i]);
-				struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);
-				u64 pq = xive_vm_esb_load(xd, XIVE_ESB_GET);
-				seq_printf(m, "E:%c%c I(%d:%llx:%llx)",
-					   (pq & XIVE_ESB_VAL_P) ? 'P' : 'p',
-					   (pq & XIVE_ESB_VAL_Q) ? 'Q' : 'q',
-					   xc->esc_virq[i], pq, xd->eoi_page);
-				seq_printf(m, "\n");
-			}
-		}
+		kvmppc_xive_debug_show_queues(m, vcpu);
 
 		t_rm_h_xirr += xc->stat_rm_h_xirr;
 		t_rm_h_ipoll += xc->stat_rm_h_ipoll;
arch/powerpc/kvm/book3s_xive_native.c +62 -0
index e475ce83ad14..1f3da47a4a6a 100644
@@ -31,6 +31,128 @@
 
 #include "book3s_xive.h"
 
+static void kvmppc_xive_native_cleanup_queue(struct kvm_vcpu *vcpu, int prio)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	struct xive_q *q = &xc->queues[prio];
+
+	xive_native_disable_queue(xc->vp_id, q, prio);
+	if (q->qpage) {
+		put_page(virt_to_page(q->qpage));
+		q->qpage = NULL;
+	}
+}
+
+void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	int i;
+
+	if (!kvmppc_xive_enabled(vcpu))
+		return;
+
+	if (!xc)
+		return;
+
+	pr_devel("native_cleanup_vcpu(cpu=%d)\n", xc->server_num);
+
+	/* Ensure no interrupt is still routed to that VP */
+	xc->valid = false;
+	kvmppc_xive_disable_vcpu_interrupts(vcpu);
+
+	/* Disable the VP */
+	xive_native_disable_vp(xc->vp_id);
+
+	/* Free the queues & associated interrupts */
+	for (i = 0; i < KVMPPC_XIVE_Q_COUNT; i++) {
+		/* Free the escalation irq */
+		if (xc->esc_virq[i]) {
+			free_irq(xc->esc_virq[i], vcpu);
+			irq_dispose_mapping(xc->esc_virq[i]);
+			kfree(xc->esc_virq_names[i]);
+			xc->esc_virq[i] = 0;
+		}
+
+		/* Free the queue */
+		kvmppc_xive_native_cleanup_queue(vcpu, i);
+	}
+
+	/* Free the VP */
+	kfree(xc);
+
+	/* Cleanup the vcpu */
+	vcpu->arch.irq_type = KVMPPC_IRQ_DEFAULT;
+	vcpu->arch.xive_vcpu = NULL;
+}
+
+int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
+				    struct kvm_vcpu *vcpu, u32 cpu)
+{
+	struct kvmppc_xive *xive = dev->private;
+	struct kvmppc_xive_vcpu *xc;
+	int rc;
+
+	pr_devel("native_connect_vcpu(cpu=%d)\n", cpu);

arch/powerpc/kvm/book3s_xive_native.c line 105

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+
+	if (dev->ops != &kvm_xive_native_ops) {
+		pr_devel("Wrong ops !\n");
+		return -EPERM;
+	}
+	if (xive->kvm != vcpu->kvm)
+		return -EPERM;
+	if (vcpu->arch.irq_type != KVMPPC_IRQ_DEFAULT)
+		return -EBUSY;
+	if (kvmppc_xive_find_server(vcpu->kvm, cpu)) {

You haven't taken the kvm->lock yet, so couldn't a race mean a duplicate server gets inserted after you make this check?

arch/powerpc/kvm/book3s_xive_native.c line 118

Quoted 13 lines
arch/powerpc/kvm/book3s_xive_native.c +13 -0
+		pr_devel("Duplicate !\n");
+		return -EEXIST;
+	}
+	if (cpu >= KVM_MAX_VCPUS) {
+		pr_devel("Out of bounds !\n");
+		return -EINVAL;
+	}
+	xc = kzalloc(sizeof(*xc), GFP_KERNEL);
+	if (!xc)
+		return -ENOMEM;
+
+	mutex_lock(&vcpu->kvm->lock);
+	vcpu->arch.xive_vcpu = xc;

Similarly you don't verify this is NULL after taking the lock, so couldn't another thread race and make a connect which gets clobbered here?

arch/powerpc/kvm/book3s_xive_native.c line 122

Quoted 4 lines
arch/powerpc/kvm/book3s_xive_native.c +4 -0
+	xc->xive = xive;
+	xc->vcpu = vcpu;
+	xc->server_num = cpu;
+	xc->vp_id = xive->vp_base + cpu;

Hrm. This ties the internal VP id to the userspace chosen server number, which isn't ideal. It puts a constraint on those server numbers that you wouldn't otherwise have.

Quoted 147 lines
arch/powerpc/kvm/book3s_xive_native.c +55 -0
+	xc->valid = true;
+
+	rc = xive_native_get_vp_info(xc->vp_id, &xc->vp_cam, &xc->vp_chip_id);
+	if (rc) {
+		pr_err("Failed to get VP info from OPAL: %d\n", rc);
+		goto bail;
+	}
+
+	/*
+	 * Enable the VP first as the single escalation mode will
+	 * affect escalation interrupts numbering
+	 */
+	rc = xive_native_enable_vp(xc->vp_id, xive->single_escalation);
+	if (rc) {
+		pr_err("Failed to enable VP in OPAL: %d\n", rc);
+		goto bail;
+	}
+
+	/* Configure VCPU fields for use by assembly push/pull */
+	vcpu->arch.xive_saved_state.w01 = cpu_to_be64(0xff000000);
+	vcpu->arch.xive_cam_word = cpu_to_be32(xc->vp_cam | TM_QW1W2_VO);
+
+	/* TODO: initialize queues ? */
+
+bail:
+	vcpu->arch.irq_type = KVMPPC_IRQ_XIVE;
+	mutex_unlock(&vcpu->kvm->lock);
+	if (rc)
+		kvmppc_xive_native_cleanup_vcpu(vcpu);
+
+	return rc;
+}
+
 static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 				       struct kvm_device_attr *attr)
 {
@@ -126,10 +248,32 @@ static int xive_native_debug_show(struct seq_file *m, void *private)
 {
 	struct kvmppc_xive *xive = m->private;
 	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
+	unsigned int i;
 
 	if (!kvm)
 		return 0;
 
+	seq_puts(m, "=========\nVCPU state\n=========\n");
+
+	kvm_for_each_vcpu(i, vcpu, kvm) {
+		struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+
+		if (!xc)
+			continue;
+
+		seq_printf(m, "cpu server %#x NSR=%02x CPPR=%02x IBP=%02x PIPR=%02x w01=%016llx w2=%08x\n",
+			   xc->server_num,
+			   vcpu->arch.xive_saved_state.nsr,
+			   vcpu->arch.xive_saved_state.cppr,
+			   vcpu->arch.xive_saved_state.ipb,
+			   vcpu->arch.xive_saved_state.pipr,
+			   vcpu->arch.xive_saved_state.w01,
+			   (u32) vcpu->arch.xive_cam_word);
+
+		kvmppc_xive_debug_show_queues(m, vcpu);
+	}
+
 	return 0;
 }
 
arch/powerpc/kvm/powerpc.c +33 -0
index 8c69af10f91d..a38a643a24dd 100644
@@ -570,6 +570,12 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 	case KVM_CAP_PPC_GET_CPU_CHAR:
 		r = 1;
 		break;
+#ifdef CONFIG_KVM_XIVE
+	case KVM_CAP_PPC_IRQ_XIVE:
+		/* only for PowerNV */
+		r = !!cpu_has_feature(CPU_FTR_HVMODE);
+		break;
+#endif
 
 	case KVM_CAP_PPC_ALLOC_HTAB:
 		r = hv_enabled;
@@ -753,6 +759,9 @@ void kvm_arch_vcpu_free(struct kvm_vcpu *vcpu)
 		else
 			kvmppc_xics_free_icp(vcpu);
 		break;
+	case KVMPPC_IRQ_XIVE:
+		kvmppc_xive_native_cleanup_vcpu(vcpu);
+		break;
 	}
 
 	kvmppc_core_vcpu_free(vcpu);
@@ -1941,6 +1950,30 @@ static int kvm_vcpu_ioctl_enable_cap(struct kvm_vcpu *vcpu,
 		break;
 	}
 #endif /* CONFIG_KVM_XICS */
+#ifdef CONFIG_KVM_XIVE
+	case KVM_CAP_PPC_IRQ_XIVE: {
+		struct fd f;
+		struct kvm_device *dev;
+
+		r = -EBADF;
+		f = fdget(cap->args[0]);
+		if (!f.file)
+			break;
+
+		r = -ENXIO;
+		if (!xive_enabled())
+			break;
+
+		r = -EPERM;
+		dev = kvm_device_from_filp(f.file);
+		if (dev)
+			r = kvmppc_xive_native_connect_vcpu(dev, vcpu,
+							    cap->args[1]);
+
+		fdput(f);
+		break;
+	}
+#endif /* CONFIG_KVM_XIVE */
 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
 	case KVM_CAP_PPC_FWNMI:
 		r = -EINVAL;
Documentation/virtual/kvm/api.txt +9 -0
index 356156f5c52d..1db1435769b4 100644
@@ -4458,6 +4458,15 @@ struct kvm_sync_regs {
         struct kvm_vcpu_events events;
 };
 
+6.75 KVM_CAP_PPC_IRQ_XIVE
+
+Architectures: ppc
+Target: vcpu
+Parameters: args[0] is the XIVE device fd
+            args[1] is the XIVE CPU number (server ID) for this vcpu
+
+This capability connects the vcpu to an in-kernel XIVE device.
+
 7. Capabilities that can be enabled on VMs
 ------------------------------------------
 
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#12 lore raw

Re: [PATCH v2 03/16] KVM: PPC: Book3S HV: XIVE: introduce a new capability KVM_CAP_PPC_IRQ_XIVE

Quoted 9 lines · David Gibson

arch/powerpc/kvm/book3s_xive_native.c line 122

Quoted 4 lines · Cédric Le Goater
arch/powerpc/kvm/book3s_xive_native.c +4 -0
+	xc->xive = xive;
+	xc->vcpu = vcpu;
+	xc->server_num = cpu;
+	xc->vp_id = xive->vp_base + cpu;

Hrm. This ties the internal VP id to the userspace chosen server number, which isn't ideal. It puts a constraint on those server numbers that you wouldn't otherwise have.

We should probably do the same as the xics-on-xive code, which is to put the server number through kvmppc_pack_vcpu_id(), which is a folding function that maps the QEMU vcpu id (which is the server number) down to the range 0..KVM_MAX_VCPUS-1, and works for the allocation patterns used in the various vSMT modes.

Paul.

#13 lore raw

Re: [PATCH v2 03/16] KVM: PPC: Book3S HV: XIVE: introduce a new capability KVM_CAP_PPC_IRQ_XIVE

Quoted 11 lines · Paul Mackerras
Quoted 9 lines · David Gibson

arch/powerpc/kvm/book3s_xive_native.c line 122

Quoted 4 lines · Cédric Le Goater
arch/powerpc/kvm/book3s_xive_native.c +4 -0
+	xc->xive = xive;
+	xc->vcpu = vcpu;
+	xc->server_num = cpu;
+	xc->vp_id = xive->vp_base + cpu;

Hrm. This ties the internal VP id to the userspace chosen server number, which isn't ideal. It puts a constraint on those server numbers that you wouldn't otherwise have.

Quoted 5 lines

We should probably do the same as the xics-on-xive code, which is to put the server number through kvmppc_pack_vcpu_id(), which is a folding function that maps the QEMU vcpu id (which is the server number) down to the range 0..KVM_MAX_VCPUS-1, and works for the allocation patterns used in the various vSMT modes.

yes. I will see how it goes.

Thanks,

C.

#14 lore raw

Re: [PATCH v2 03/16] KVM: PPC: Book3S HV: XIVE: introduce a new capability KVM_CAP_PPC_IRQ_XIVE

In reply to David Gibson
Quoted 285 lines · David Gibson
Quoted 261 lines · Cédric Le Goater

The user interface exposes a new capability to let QEMU connect the vCPU to the XIVE KVM device if required. The capability is only advertised on a PowerNV Hypervisor as support for nested guests (pseries KVM Hypervisor) is not yet available.

Internally, the interface to the new KVM device is protected with a new interrupt mode: KVMPPC_IRQ_XIVE.

---
 arch/powerpc/include/asm/kvm_host.h   |   1 +
 arch/powerpc/include/asm/kvm_ppc.h    |  13 +++
 arch/powerpc/kvm/book3s_xive.h        |   6 ++
 include/uapi/linux/kvm.h              |   1 +
 arch/powerpc/kvm/book3s_xive.c        |  67 +++++++-----
 arch/powerpc/kvm/book3s_xive_native.c | 144 ++++++++++++++++++++++++++
 arch/powerpc/kvm/powerpc.c            |  33 ++++++
 Documentation/virtual/kvm/api.txt     |   9 ++
 8 files changed, 246 insertions(+), 28 deletions(-)
arch/powerpc/include/asm/kvm_host.h +1 -0
index 9f75a75a07f2..eb8581be0ee8 100644
@@ -448,6 +448,7 @@ struct kvmppc_passthru_irqmap {
 #define KVMPPC_IRQ_DEFAULT	0
 #define KVMPPC_IRQ_MPIC		1
 #define KVMPPC_IRQ_XICS		2 /* Includes a XIVE option */
+#define KVMPPC_IRQ_XIVE		3 /* XIVE native exploitation mode */
 
 #define MMIO_HPTE_CACHE_SIZE	4
 
arch/powerpc/include/asm/kvm_ppc.h +13 -0
index 4b72ddde7dc1..1e61877fe147 100644
@@ -594,6 +594,14 @@ extern int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 irq,
 			       int level, bool line_status);
 extern void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu);
 
+static inline int kvmppc_xive_enabled(struct kvm_vcpu *vcpu)
+{
+	return vcpu->arch.irq_type == KVMPPC_IRQ_XIVE;
+}
+
+extern int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
+					   struct kvm_vcpu *vcpu, u32 cpu);
+extern void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu);
 extern void kvmppc_xive_native_init_module(void);
 extern void kvmppc_xive_native_exit_module(void);
 
@@ -621,6 +629,11 @@ static inline int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 ir
 				      int level, bool line_status) { return -ENODEV; }
 static inline void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu) { }
 
+static inline int kvmppc_xive_enabled(struct kvm_vcpu *vcpu)
+	{ return 0; }
+static inline int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
+			  struct kvm_vcpu *vcpu, u32 cpu) { return -EBUSY; }
+static inline void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu) { }
 static inline void kvmppc_xive_native_init_module(void) { }
 static inline void kvmppc_xive_native_exit_module(void) { }
 
arch/powerpc/kvm/book3s_xive.h +6 -0
index a08ae6fd4c51..bcb1bbcf0359 100644
@@ -248,5 +248,11 @@ extern int (*__xive_vm_h_ipi)(struct kvm_vcpu *vcpu, unsigned long server,
 extern int (*__xive_vm_h_cppr)(struct kvm_vcpu *vcpu, unsigned long cppr);
 extern int (*__xive_vm_h_eoi)(struct kvm_vcpu *vcpu, unsigned long xirr);
 
+/*
+ * Common Xive routines for XICS-over-XIVE and XIVE native
+ */
+void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu);
+int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu);
+
 #endif /* CONFIG_KVM_XICS */
 #endif /* _KVM_PPC_BOOK3S_XICS_H */
include/uapi/linux/kvm.h +1 -0
index e6368163d3a0..52bf74a1616e 100644
@@ -988,6 +988,7 @@ struct kvm_ppc_resize_hpt {
 #define KVM_CAP_ARM_VM_IPA_SIZE 165
 #define KVM_CAP_MANUAL_DIRTY_LOG_PROTECT 166
 #define KVM_CAP_HYPERV_CPUID 167
+#define KVM_CAP_PPC_IRQ_XIVE 168
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
arch/powerpc/kvm/book3s_xive.c +39 -28
index f78d002f0fe0..d1cc18a5b1c4 100644
@@ -1049,7 +1049,7 @@ int kvmppc_xive_clr_mapped(struct kvm *kvm, unsigned long guest_irq,
 }
 EXPORT_SYMBOL_GPL(kvmppc_xive_clr_mapped);
 
-static void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu)
+void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
 	struct kvm *kvm = vcpu->kvm;
@@ -1883,6 +1883,43 @@ static int kvmppc_xive_create(struct kvm_device *dev, u32 type)
 	return 0;
 }
 
+int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	unsigned int i;
+
+	for (i = 0; i < KVMPPC_XIVE_Q_COUNT; i++) {
+		struct xive_q *q = &xc->queues[i];
+		u32 i0, i1, idx;
+
+		if (!q->qpage && !xc->esc_virq[i])
+			continue;
+
+		seq_printf(m, " [q%d]: ", i);
+
+		if (q->qpage) {
+			idx = q->idx;
+			i0 = be32_to_cpup(q->qpage + idx);
+			idx = (idx + 1) & q->msk;
+			i1 = be32_to_cpup(q->qpage + idx);
+			seq_printf(m, "T=%d %08x %08x...\n", q->toggle,
+				   i0, i1);
+		}
+		if (xc->esc_virq[i]) {
+			struct irq_data *d = irq_get_irq_data(xc->esc_virq[i]);
+			struct xive_irq_data *xd =
+				irq_data_get_irq_handler_data(d);
+			u64 pq = xive_vm_esb_load(xd, XIVE_ESB_GET);
+
+			seq_printf(m, "E:%c%c I(%d:%llx:%llx)",
+				   (pq & XIVE_ESB_VAL_P) ? 'P' : 'p',
+				   (pq & XIVE_ESB_VAL_Q) ? 'Q' : 'q',
+				   xc->esc_virq[i], pq, xd->eoi_page);
+			seq_puts(m, "\n");
+		}
+	}
+	return 0;
+}
 
 static int xive_debug_show(struct seq_file *m, void *private)
 {
@@ -1908,7 +1945,6 @@ static int xive_debug_show(struct seq_file *m, void *private)
 
 	kvm_for_each_vcpu(i, vcpu, kvm) {
 		struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
-		unsigned int i;
 
 		if (!xc)
 			continue;
@@ -1918,33 +1954,8 @@ static int xive_debug_show(struct seq_file *m, void *private)
 			   xc->server_num, xc->cppr, xc->hw_cppr,
 			   xc->mfrr, xc->pending,
 			   xc->stat_rm_h_xirr, xc->stat_vm_h_xirr);
-		for (i = 0; i < KVMPPC_XIVE_Q_COUNT; i++) {
-			struct xive_q *q = &xc->queues[i];
-			u32 i0, i1, idx;
 
-			if (!q->qpage && !xc->esc_virq[i])
-				continue;
-
-			seq_printf(m, " [q%d]: ", i);
-
-			if (q->qpage) {
-				idx = q->idx;
-				i0 = be32_to_cpup(q->qpage + idx);
-				idx = (idx + 1) & q->msk;
-				i1 = be32_to_cpup(q->qpage + idx);
-				seq_printf(m, "T=%d %08x %08x... \n", q->toggle, i0, i1);
-			}
-			if (xc->esc_virq[i]) {
-				struct irq_data *d = irq_get_irq_data(xc->esc_virq[i]);
-				struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);
-				u64 pq = xive_vm_esb_load(xd, XIVE_ESB_GET);
-				seq_printf(m, "E:%c%c I(%d:%llx:%llx)",
-					   (pq & XIVE_ESB_VAL_P) ? 'P' : 'p',
-					   (pq & XIVE_ESB_VAL_Q) ? 'Q' : 'q',
-					   xc->esc_virq[i], pq, xd->eoi_page);
-				seq_printf(m, "\n");
-			}
-		}
+		kvmppc_xive_debug_show_queues(m, vcpu);
 
 		t_rm_h_xirr += xc->stat_rm_h_xirr;
 		t_rm_h_ipoll += xc->stat_rm_h_ipoll;
arch/powerpc/kvm/book3s_xive_native.c +62 -0
index e475ce83ad14..1f3da47a4a6a 100644
@@ -31,6 +31,128 @@
 
 #include "book3s_xive.h"
 
+static void kvmppc_xive_native_cleanup_queue(struct kvm_vcpu *vcpu, int prio)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	struct xive_q *q = &xc->queues[prio];
+
+	xive_native_disable_queue(xc->vp_id, q, prio);
+	if (q->qpage) {
+		put_page(virt_to_page(q->qpage));
+		q->qpage = NULL;
+	}
+}
+
+void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	int i;
+
+	if (!kvmppc_xive_enabled(vcpu))
+		return;
+
+	if (!xc)
+		return;
+
+	pr_devel("native_cleanup_vcpu(cpu=%d)\n", xc->server_num);
+
+	/* Ensure no interrupt is still routed to that VP */
+	xc->valid = false;
+	kvmppc_xive_disable_vcpu_interrupts(vcpu);
+
+	/* Disable the VP */
+	xive_native_disable_vp(xc->vp_id);
+
+	/* Free the queues & associated interrupts */
+	for (i = 0; i < KVMPPC_XIVE_Q_COUNT; i++) {
+		/* Free the escalation irq */
+		if (xc->esc_virq[i]) {
+			free_irq(xc->esc_virq[i], vcpu);
+			irq_dispose_mapping(xc->esc_virq[i]);
+			kfree(xc->esc_virq_names[i]);
+			xc->esc_virq[i] = 0;
+		}
+
+		/* Free the queue */
+		kvmppc_xive_native_cleanup_queue(vcpu, i);
+	}
+
+	/* Free the VP */
+	kfree(xc);
+
+	/* Cleanup the vcpu */
+	vcpu->arch.irq_type = KVMPPC_IRQ_DEFAULT;
+	vcpu->arch.xive_vcpu = NULL;
+}
+
+int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
+				    struct kvm_vcpu *vcpu, u32 cpu)
+{
+	struct kvmppc_xive *xive = dev->private;
+	struct kvmppc_xive_vcpu *xc;
+	int rc;
+
+	pr_devel("native_connect_vcpu(cpu=%d)\n", cpu);

arch/powerpc/kvm/book3s_xive_native.c line 105

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+
+	if (dev->ops != &kvm_xive_native_ops) {
+		pr_devel("Wrong ops !\n");
+		return -EPERM;
+	}
+	if (xive->kvm != vcpu->kvm)
+		return -EPERM;
+	if (vcpu->arch.irq_type != KVMPPC_IRQ_DEFAULT)
+		return -EBUSY;
+	if (kvmppc_xive_find_server(vcpu->kvm, cpu)) {

You haven't taken the kvm->lock yet, so couldn't a race mean a duplicate server gets inserted after you make this check?

Quoted 9 lines
arch/powerpc/kvm/book3s_xive_native.c +9 -0
+		pr_devel("Duplicate !\n");
+		return -EEXIST;
+	}
+	if (cpu >= KVM_MAX_VCPUS) {
+		pr_devel("Out of bounds !\n");
+		return -EINVAL;
+	}
+	xc = kzalloc(sizeof(*xc), GFP_KERNEL);
+	if (!xc)
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c line 118

Quoted 4 lines
arch/powerpc/kvm/book3s_xive_native.c +4 -0
+		return -ENOMEM;
+
+	mutex_lock(&vcpu->kvm->lock);
+	vcpu->arch.xive_vcpu = xc;

Similarly you don't verify this is NULL after taking the lock, so couldn't another thread race and make a connect which gets clobbered here?

Yes. this is not very safe ... We need to clean up all the KVM device methods doing the connection of the presenter to the vCPU AFAICT. I will fix the XIVE native one for now.

And also, this CPU parameter is useless. There is no reason to connect a vCPU from another vCPU.

Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c line 122

Quoted 4 lines
arch/powerpc/kvm/book3s_xive_native.c +4 -0
+	xc->xive = xive;
+	xc->vcpu = vcpu;
+	xc->server_num = cpu;
+	xc->vp_id = xive->vp_base + cpu;

Hrm. This ties the internal VP id to the userspace chosen server number, which isn't ideal. It puts a constraint on those server numbers that you wouldn't otherwise have.

Ah yes. I should be using the kvmppc_pack_vcpu_id() like we do for the XICS-over-XIVE device probably. I need to check that it is correct in this mode.

Thanks,

C.

Quoted 148 lines
Quoted 147 lines
arch/powerpc/kvm/book3s_xive_native.c +55 -0
+	xc->valid = true;
+
+	rc = xive_native_get_vp_info(xc->vp_id, &xc->vp_cam, &xc->vp_chip_id);
+	if (rc) {
+		pr_err("Failed to get VP info from OPAL: %d\n", rc);
+		goto bail;
+	}
+
+	/*
+	 * Enable the VP first as the single escalation mode will
+	 * affect escalation interrupts numbering
+	 */
+	rc = xive_native_enable_vp(xc->vp_id, xive->single_escalation);
+	if (rc) {
+		pr_err("Failed to enable VP in OPAL: %d\n", rc);
+		goto bail;
+	}
+
+	/* Configure VCPU fields for use by assembly push/pull */
+	vcpu->arch.xive_saved_state.w01 = cpu_to_be64(0xff000000);
+	vcpu->arch.xive_cam_word = cpu_to_be32(xc->vp_cam | TM_QW1W2_VO);
+
+	/* TODO: initialize queues ? */
+
+bail:
+	vcpu->arch.irq_type = KVMPPC_IRQ_XIVE;
+	mutex_unlock(&vcpu->kvm->lock);
+	if (rc)
+		kvmppc_xive_native_cleanup_vcpu(vcpu);
+
+	return rc;
+}
+
 static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 				       struct kvm_device_attr *attr)
 {
@@ -126,10 +248,32 @@ static int xive_native_debug_show(struct seq_file *m, void *private)
 {
 	struct kvmppc_xive *xive = m->private;
 	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
+	unsigned int i;
 
 	if (!kvm)
 		return 0;
 
+	seq_puts(m, "=========\nVCPU state\n=========\n");
+
+	kvm_for_each_vcpu(i, vcpu, kvm) {
+		struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+
+		if (!xc)
+			continue;
+
+		seq_printf(m, "cpu server %#x NSR=%02x CPPR=%02x IBP=%02x PIPR=%02x w01=%016llx w2=%08x\n",
+			   xc->server_num,
+			   vcpu->arch.xive_saved_state.nsr,
+			   vcpu->arch.xive_saved_state.cppr,
+			   vcpu->arch.xive_saved_state.ipb,
+			   vcpu->arch.xive_saved_state.pipr,
+			   vcpu->arch.xive_saved_state.w01,
+			   (u32) vcpu->arch.xive_cam_word);
+
+		kvmppc_xive_debug_show_queues(m, vcpu);
+	}
+
 	return 0;
 }
 
arch/powerpc/kvm/powerpc.c +33 -0
index 8c69af10f91d..a38a643a24dd 100644
@@ -570,6 +570,12 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 	case KVM_CAP_PPC_GET_CPU_CHAR:
 		r = 1;
 		break;
+#ifdef CONFIG_KVM_XIVE
+	case KVM_CAP_PPC_IRQ_XIVE:
+		/* only for PowerNV */
+		r = !!cpu_has_feature(CPU_FTR_HVMODE);
+		break;
+#endif
 
 	case KVM_CAP_PPC_ALLOC_HTAB:
 		r = hv_enabled;
@@ -753,6 +759,9 @@ void kvm_arch_vcpu_free(struct kvm_vcpu *vcpu)
 		else
 			kvmppc_xics_free_icp(vcpu);
 		break;
+	case KVMPPC_IRQ_XIVE:
+		kvmppc_xive_native_cleanup_vcpu(vcpu);
+		break;
 	}
 
 	kvmppc_core_vcpu_free(vcpu);
@@ -1941,6 +1950,30 @@ static int kvm_vcpu_ioctl_enable_cap(struct kvm_vcpu *vcpu,
 		break;
 	}
 #endif /* CONFIG_KVM_XICS */
+#ifdef CONFIG_KVM_XIVE
+	case KVM_CAP_PPC_IRQ_XIVE: {
+		struct fd f;
+		struct kvm_device *dev;
+
+		r = -EBADF;
+		f = fdget(cap->args[0]);
+		if (!f.file)
+			break;
+
+		r = -ENXIO;
+		if (!xive_enabled())
+			break;
+
+		r = -EPERM;
+		dev = kvm_device_from_filp(f.file);
+		if (dev)
+			r = kvmppc_xive_native_connect_vcpu(dev, vcpu,
+							    cap->args[1]);
+
+		fdput(f);
+		break;
+	}
+#endif /* CONFIG_KVM_XIVE */
 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
 	case KVM_CAP_PPC_FWNMI:
 		r = -EINVAL;
Documentation/virtual/kvm/api.txt +9 -0
index 356156f5c52d..1db1435769b4 100644
@@ -4458,6 +4458,15 @@ struct kvm_sync_regs {
         struct kvm_vcpu_events events;
 };
 
+6.75 KVM_CAP_PPC_IRQ_XIVE
+
+Architectures: ppc
+Target: vcpu
+Parameters: args[0] is the XIVE device fd
+            args[1] is the XIVE CPU number (server ID) for this vcpu
+
+This capability connects the vcpu to an in-kernel XIVE device.
+
 7. Capabilities that can be enabled on VMs
 ------------------------------------------
 
#15 lore raw

Re: [PATCH v2 03/16] KVM: PPC: Book3S HV: XIVE: introduce a new capability KVM_CAP_PPC_IRQ_XIVE

Quoted 2 lines · Cédric Le Goater
Quoted 1 line · David Gibson

On Fri, Feb 22, 2019 at 12:28:27PM +0100, Cédric Le Goater wrote:

[snip]

Quoted 40 lines
Quoted 31 lines
Quoted 8 lines
arch/powerpc/kvm/book3s_xive_native.c +8 -0
+int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
+				    struct kvm_vcpu *vcpu, u32 cpu)
+{
+	struct kvmppc_xive *xive = dev->private;
+	struct kvmppc_xive_vcpu *xc;
+	int rc;
+
+	pr_devel("native_connect_vcpu(cpu=%d)\n", cpu);

arch/powerpc/kvm/book3s_xive_native.c line 105

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+
+	if (dev->ops != &kvm_xive_native_ops) {
+		pr_devel("Wrong ops !\n");
+		return -EPERM;
+	}
+	if (xive->kvm != vcpu->kvm)
+		return -EPERM;
+	if (vcpu->arch.irq_type != KVMPPC_IRQ_DEFAULT)
+		return -EBUSY;
+	if (kvmppc_xive_find_server(vcpu->kvm, cpu)) {

You haven't taken the kvm->lock yet, so couldn't a race mean a duplicate server gets inserted after you make this check?

Quoted 9 lines
arch/powerpc/kvm/book3s_xive_native.c +9 -0
+		pr_devel("Duplicate !\n");
+		return -EEXIST;
+	}
+	if (cpu >= KVM_MAX_VCPUS) {
+		pr_devel("Out of bounds !\n");
+		return -EINVAL;
+	}
+	xc = kzalloc(sizeof(*xc), GFP_KERNEL);
+	if (!xc)
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c line 118

Quoted 4 lines
arch/powerpc/kvm/book3s_xive_native.c +4 -0
+		return -ENOMEM;
+
+	mutex_lock(&vcpu->kvm->lock);
+	vcpu->arch.xive_vcpu = xc;

Similarly you don't verify this is NULL after taking the lock, so couldn't another thread race and make a connect which gets clobbered here?

Quoted 6 lines

Yes. this is not very safe ... We need to clean up all the KVM device methods doing the connection of the presenter to the vCPU AFAICT. I will fix the XIVE native one for now.

And also, this CPU parameter is useless. There is no reason to connect a vCPU from another vCPU.

Hmm.. I thought the point of the 'cpu' parameter (not a great name) is that it lets userspace chose the guest visible irq server ID. I think that's preferable to tying it to an existing cpu id, if possible.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#16 lore raw

Re: [PATCH v2 03/16] KVM: PPC: Book3S HV: XIVE: introduce a new capability KVM_CAP_PPC_IRQ_XIVE

In reply to Cédric Le Goater
Quoted 4 lines · Cédric Le Goater

The user interface exposes a new capability to let QEMU connect the vCPU to the XIVE KVM device if required. The capability is only advertised on a PowerNV Hypervisor as support for nested guests (pseries KVM Hypervisor) is not yet available.

If a bisection happened to land on this commit, we would have KVM saying it had the ability to support guests using XIVE natively, but it wouldn't actually work since we don't have all the code that is in the following patches.

Thus, in order to avoid breaking bisection, you should either add the capability now but have it always return false until the rest of the code is in place, or else defer the addition of the capability until the end of the patch series.

arch/powerpc/kvm/powerpc.c int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) · line 576

Quoted 12 lines
arch/powerpc/kvm/powerpc.c +4 -0
index 8c69af10f91d..a38a643a24dd 100644
@@ -570,6 +570,12 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 	case KVM_CAP_PPC_GET_CPU_CHAR:
 		r = 1;
 		break;
+#ifdef CONFIG_KVM_XIVE
+	case KVM_CAP_PPC_IRQ_XIVE:
+		/* only for PowerNV */
+		r = !!cpu_has_feature(CPU_FTR_HVMODE);

Shouldn't this be r = xive_enabled() && !!cpu_has_feature(CPU_FTR_HVMODE) (or alternatively r = xics_on_xive(), though that would be confusing to the reader)?

As it stands this would report true on POWER8, unless I'm missing something.

Paul.

#17 lore raw

Re: [PATCH v2 03/16] KVM: PPC: Book3S HV: XIVE: introduce a new capability KVM_CAP_PPC_IRQ_XIVE

Quoted 10 lines · Paul Mackerras
Quoted 4 lines · Cédric Le Goater

The user interface exposes a new capability to let QEMU connect the vCPU to the XIVE KVM device if required. The capability is only advertised on a PowerNV Hypervisor as support for nested guests (pseries KVM Hypervisor) is not yet available.

If a bisection happened to land on this commit, we would have KVM saying it had the ability to support guests using XIVE natively, but it wouldn't actually work since we don't have all the code that is in the following patches.

OK. I didn't think migration was a must-have for bisection. I will move the enablement at end.

Quoted 4 lines

Thus, in order to avoid breaking bisection, you should either add the capability now but have it always return false until the rest of the code is in place, or else defer the addition of the capability until the end of the patch series.

I will introduce the capability early in the patchset and return false as you are proposing. It seems to be the best approach.

Quoted 6 lines
Quoted 6 lines
arch/powerpc/kvm/powerpc.c +0 -0
index 8c69af10f91d..a38a643a24dd 100644
@@ -570,6 +570,12 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
 	case KVM_CAP_PPC_GET_CPU_CHAR:
Quoted 8 lines

arch/powerpc/kvm/powerpc.c int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) · line 576

Quoted 6 lines
arch/powerpc/kvm/powerpc.c +4 -0
 		r = 1;
 		break;
+#ifdef CONFIG_KVM_XIVE
+	case KVM_CAP_PPC_IRQ_XIVE:
+		/* only for PowerNV */
+		r = !!cpu_has_feature(CPU_FTR_HVMODE);

Shouldn't this be r = xive_enabled() && !!cpu_has_feature(CPU_FTR_HVMODE)

yes. we need the '__xive_enabled' toggle to be set also :/

It can set to off with the "xive=off" on the command line and on old P9 skiboot. That could be simplified one day.

Quoted 2 lines

(or alternatively r = xics_on_xive(), though that would be confusing to the reader)?

This is correct. I didn't want to use the xics_on_xive() which is not the capability we are activating.

I will keep the open-coded version.

Quoted 2 lines

As it stands this would report true on POWER8, unless I'm missing something.

Ah yes. I forgot this combination also.

This should not be too complex to fix.

Thanks,

C.

#18 lore raw

Re: [PATCH v2 03/16] KVM: PPC: Book3S HV: XIVE: introduce a new capability KVM_CAP_PPC_IRQ_XIVE

Quoted 7 lines · Cédric Le Goater
Quoted 6 lines · Paul Mackerras
Quoted 4 lines · Cédric Le Goater

The user interface exposes a new capability to let QEMU connect the vCPU to the XIVE KVM device if required. The capability is only advertised on a PowerNV Hypervisor as support for nested guests (pseries KVM Hypervisor) is not yet available.

Quoted 7 lines
Quoted 4 lines

If a bisection happened to land on this commit, we would have KVM saying it had the ability to support guests using XIVE natively, but it wouldn't actually work since we don't have all the code that is in the following patches.

OK. I didn't think migration was a must-have for bisection. I will move the enablement at end.

Any temporary feature regression potentially breaks bisection, because you don't know what we'll want to bisect for. Obviously we're never going to get that perfectly right, but that doesn't mean we shouldn't try when we do see the problem in advance.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#19 lore raw

PATCH v2 4/16 KVM: PPC: Book3S HV: XIVE: add a control to initialize a source

In reply to Cédric Le Goater

The associated HW interrupt source is simply allocated at the OPAL/HW level and then MASKED. KVM only needs to know about its type: LSI or MSI.

---
 arch/powerpc/include/uapi/asm/kvm.h        |   5 +
 arch/powerpc/kvm/book3s_xive.h             |  10 ++
 arch/powerpc/kvm/book3s_xive.c             |   8 +-
 arch/powerpc/kvm/book3s_xive_native.c      | 114 +++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  15 +++
 5 files changed, 148 insertions(+), 4 deletions(-)
arch/powerpc/include/uapi/asm/kvm.h +5 -0
index b002c0c67787..a9ad99f2a11b 100644
@@ -677,5 +677,10 @@ struct kvm_ppc_cpu_char {
 
 /* POWER9 XIVE Native Interrupt Controller */
 #define KVM_DEV_XIVE_GRP_CTRL		1
+#define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
+
+/* Layout of 64-bit XIVE source attribute values */
+#define KVM_XIVE_LEVEL_SENSITIVE	(1ULL << 0)
+#define KVM_XIVE_LEVEL_ASSERTED		(1ULL << 1)
 
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive.h +10 -0
index bcb1bbcf0359..f22f2d46d0f0 100644
@@ -12,6 +12,13 @@
 #ifdef CONFIG_KVM_XICS
 #include "book3s_xics.h"
 
+/*
+ * The XIVE IRQ number space is aligned with the XICS IRQ number
+ * space, CPU IPIs being allocated in the first 4K.
+ */
+#define KVMPPC_XIVE_FIRST_IRQ	0
+#define KVMPPC_XIVE_NR_IRQS	KVMPPC_XICS_NR_IRQS
+
 /*
  * State for one guest irq source.
  *
@@ -253,6 +260,9 @@ extern int (*__xive_vm_h_eoi)(struct kvm_vcpu *vcpu, unsigned long xirr);
  */
 void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu);
 int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu);
+struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
+	struct kvmppc_xive *xive, int irq);
+void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb);
 
 #endif /* CONFIG_KVM_XICS */
 #endif /* _KVM_PPC_BOOK3S_XICS_H */
arch/powerpc/kvm/book3s_xive.c +4 -4
index d1cc18a5b1c4..6f950ecb3592 100644
@@ -1485,8 +1485,8 @@ static int xive_get_source(struct kvmppc_xive *xive, long irq, u64 addr)
 	return 0;
 }
 
-static struct kvmppc_xive_src_block *xive_create_src_block(struct kvmppc_xive *xive,
-							   int irq)
+struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
+	struct kvmppc_xive *xive, int irq)
 {
 	struct kvm *kvm = xive->kvm;
 	struct kvmppc_xive_src_block *sb;
@@ -1565,7 +1565,7 @@ static int xive_set_source(struct kvmppc_xive *xive, long irq, u64 addr)
 	sb = kvmppc_xive_find_source(xive, irq, &idx);
 	if (!sb) {
 		pr_devel("No source, creating source block...\n");
-		sb = xive_create_src_block(xive, irq);
+		sb = kvmppc_xive_create_src_block(xive, irq);
 		if (!sb) {
 			pr_devel("Failed to create block...\n");
 			return -ENOMEM;
@@ -1789,7 +1789,7 @@ static void kvmppc_xive_cleanup_irq(u32 hw_num, struct xive_irq_data *xd)
 	xive_cleanup_irq_data(xd);
 }
 
-static void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb)
+void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb)
 {
 	int i;
 
arch/powerpc/kvm/book3s_xive_native.c +114 -0

150 lines not on this page · raw message

Documentation/virtual/kvm/devices/xive.txt +15 -0
index fdbd2ff92a88..cd8bfc37b72e 100644
@@ -17,3 +17,18 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
 
   1. KVM_DEV_XIVE_GRP_CTRL
   Provides global controls on the device
+
+  2. KVM_DEV_XIVE_GRP_SOURCE (write only)
+  Initializes a new source in the XIVE device and mask it.
+  Attributes:
+    Interrupt source number  (64-bit)
+  The kvm_device_attr.addr points to a __u64 value:
+  bits:     | 63   ....  2 |   1   |   0
+  values:   |    unused    | level | type
+  - type:  0:MSI 1:LSI
+  - level: assertion level in case of an LSI.
+  Errors:
+    -E2BIG:  Interrupt source number is out of range
+    -ENOMEM: Could not create a new source block
+    -EFAULT: Invalid user pointer for attr->addr.
+    -ENXIO:  Could not allocate underlying HW interrupt
Signature · 1 line
2.20.1
#20 lore raw

Re: [PATCH v2 04/16] KVM: PPC: Book3S HV: XIVE: add a control to initialize a source

Quoted 33 lines · Cédric Le Goater

The associated HW interrupt source is simply allocated at the OPAL/HW level and then MASKED. KVM only needs to know about its type: LSI or MSI.

---
 arch/powerpc/include/uapi/asm/kvm.h        |   5 +
 arch/powerpc/kvm/book3s_xive.h             |  10 ++
 arch/powerpc/kvm/book3s_xive.c             |   8 +-
 arch/powerpc/kvm/book3s_xive_native.c      | 114 +++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  15 +++
 5 files changed, 148 insertions(+), 4 deletions(-)
arch/powerpc/include/uapi/asm/kvm.h +5 -0
index b002c0c67787..a9ad99f2a11b 100644
@@ -677,5 +677,10 @@ struct kvm_ppc_cpu_char {
 
 /* POWER9 XIVE Native Interrupt Controller */
 #define KVM_DEV_XIVE_GRP_CTRL		1
+#define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
+
+/* Layout of 64-bit XIVE source attribute values */
+#define KVM_XIVE_LEVEL_SENSITIVE	(1ULL << 0)
+#define KVM_XIVE_LEVEL_ASSERTED		(1ULL << 1)
 
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive.h +0 -0
index bcb1bbcf0359..f22f2d46d0f0 100644

arch/powerpc/kvm/book3s_xive.h line 17

Quoted 6 lines
arch/powerpc/kvm/book3s_xive.h +3 -0
@@ -12,6 +12,13 @@
 #ifdef CONFIG_KVM_XICS
 #include "book3s_xics.h"
 
+/*
+ * The XIVE IRQ number space is aligned with the XICS IRQ number
+ * space, CPU IPIs being allocated in the first 4K.

We do align these in qemu, but I don't see that the kernel part
cares: as far as it's concerned only one of XICS or XIVE is active at a time, and the irq numbers are chosen by userspace.

arch/powerpc/kvm/book3s_xive.c

Quoted 21 lines
arch/powerpc/kvm/book3s_xive.h +7 -0
+ */
+#define KVMPPC_XIVE_FIRST_IRQ	0
+#define KVMPPC_XIVE_NR_IRQS	KVMPPC_XICS_NR_IRQS
+
 /*
  * State for one guest irq source.
  *
@@ -253,6 +260,9 @@ extern int (*__xive_vm_h_eoi)(struct kvm_vcpu *vcpu, unsigned long xirr);
  */
 void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu);
 int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu);
+struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
+	struct kvmppc_xive *xive, int irq);
+void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb);
 
 #endif /* CONFIG_KVM_XICS */
 #endif /* _KVM_PPC_BOOK3S_XICS_H */
arch/powerpc/kvm/book3s_xive.c +0 -0
index d1cc18a5b1c4..6f950ecb3592 100644

I wonder if we should rename this book3s_xics_on_xive.c or something at some point, I keep getting confused because I forget that this is only dealing with host xive, not guest xive.

arch/powerpc/kvm/book3s_xive.c static int xive_get_source(struct kvmppc_xive *xive, long irq, u64 addr) · line 1492

Quoted 11 lines
arch/powerpc/kvm/book3s_xive.c +2 -2
@@ -1485,8 +1485,8 @@ static int xive_get_source(struct kvmppc_xive *xive, long irq, u64 addr)
 	return 0;
 }
 
-static struct kvmppc_xive_src_block *xive_create_src_block(struct kvmppc_xive *xive,
-							   int irq)
+struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
+	struct kvmppc_xive *xive, int irq)
 {
 	struct kvm *kvm = xive->kvm;
 	struct kvmppc_xive_src_block *sb;

It's odd that this function, now used from the xive-on-xive path as well as the xics-on-xive path references KVMPPC_XICS_ICS_SHIFT a few lines down from this change.

Quoted 75 lines
arch/powerpc/kvm/book3s_xive.c +2 -2
@@ -1565,7 +1565,7 @@ static int xive_set_source(struct kvmppc_xive *xive, long irq, u64 addr)
 	sb = kvmppc_xive_find_source(xive, irq, &idx);
 	if (!sb) {
 		pr_devel("No source, creating source block...\n");
-		sb = xive_create_src_block(xive, irq);
+		sb = kvmppc_xive_create_src_block(xive, irq);
 		if (!sb) {
 			pr_devel("Failed to create block...\n");
 			return -ENOMEM;
@@ -1789,7 +1789,7 @@ static void kvmppc_xive_cleanup_irq(u32 hw_num, struct xive_irq_data *xd)
 	xive_cleanup_irq_data(xd);
 }
 
-static void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb)
+void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb)
 {
 	int i;
 
arch/powerpc/kvm/book3s_xive_native.c +42 -0
index 1f3da47a4a6a..a9b2d2d9af99 100644
@@ -31,6 +31,29 @@
 
 #include "book3s_xive.h"
 
+/*
+ * TODO: introduce a common template file with the XIVE native layer
+ * and the XICS-on-XIVE glue for the utility functions
+ */
+#define __x_eoi_page(xd)	((void __iomem *)((xd)->eoi_mmio))
+#define __x_trig_page(xd)	((void __iomem *)((xd)->trig_mmio))
+#define __x_readq	__raw_readq
+#define __x_writeq	__raw_writeq
+
+static u8 xive_vm_esb_load(struct xive_irq_data *xd, u32 offset)
+{
+	u64 val;
+
+	if (xd->flags & XIVE_IRQ_FLAG_SHIFT_BUG)
+		offset |= offset << 4;
+
+	val = __x_readq(__x_eoi_page(xd) + offset);
+#ifdef __LITTLE_ENDIAN__
+	val >>= 64-8;
+#endif
+	return (u8)val;
+}
+
 static void kvmppc_xive_native_cleanup_queue(struct kvm_vcpu *vcpu, int prio)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
@@ -153,12 +176,89 @@ int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 	return rc;
 }
 
+static int kvmppc_xive_native_set_source(struct kvmppc_xive *xive, long irq,
+					 u64 addr)
+{
+	struct kvmppc_xive_src_block *sb;
+	struct kvmppc_xive_irq_state *state;
+	u64 __user *ubufp = (u64 __user *) addr;
+	u64 val;
+	u16 idx;
+
+	pr_devel("%s irq=0x%lx\n", __func__, irq);
+
+	if (irq < KVMPPC_XIVE_FIRST_IRQ || irq >= KVMPPC_XIVE_NR_IRQS)
+		return -E2BIG;
+
+	sb = kvmppc_xive_find_source(xive, irq, &idx);
+	if (!sb) {
+		pr_debug("No source, creating source block...\n");
+		sb = kvmppc_xive_create_src_block(xive, irq);
+		if (!sb) {

arch/powerpc/kvm/book3s_xive_native.c int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev, · line 207

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+			pr_err("Failed to create block...\n");
+			return -ENOMEM;
+		}
+	}
+	state = &sb->irq_state[idx];
+
+	if (get_user(val, ubufp)) {
+		pr_err("fault getting user info !\n");
+		return -EFAULT;
+	}

You should validate the value loaded here to check it doesn't have any bits set we don't know about.

Quoted 8 lines
arch/powerpc/kvm/book3s_xive_native.c +8 -0
+
+	/*
+	 * If the source doesn't already have an IPI, allocate
+	 * one and get the corresponding data
+	 */
+	if (!state->ipi_number) {
+		state->ipi_number = xive_native_alloc_irq();
+		if (state->ipi_number == 0) {

arch/powerpc/kvm/book3s_xive_native.c int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev, · line 225

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+			pr_err("Failed to allocate IRQ !\n");
+			return -ENXIO;
+		}
+		xive_native_populate_irq_data(state->ipi_number,
+					      &state->ipi_data);
+		pr_debug("%s allocated hw_irq=0x%x for irq=0x%lx\n", __func__,
+			 state->ipi_number, irq);
+	}
+
+	arch_spin_lock(&sb->lock);

Why the direct call to arch_spin_lock() rather than just spin_lock()?

Quoted 97 lines
arch/powerpc/kvm/book3s_xive_native.c +44 -0
+
+	/* Restore LSI state */
+	if (val & KVM_XIVE_LEVEL_SENSITIVE) {
+		state->lsi = true;
+		if (val & KVM_XIVE_LEVEL_ASSERTED)
+			state->asserted = true;
+		pr_devel("  LSI ! Asserted=%d\n", state->asserted);
+	}
+
+	/* Mask IRQ to start with */
+	state->act_server = 0;
+	state->act_priority = MASKED;
+	xive_vm_esb_load(&state->ipi_data, XIVE_ESB_SET_PQ_01);
+	xive_native_configure_irq(state->ipi_number, 0, MASKED, 0);
+
+	/* Increment the number of valid sources and mark this one valid */
+	if (!state->valid)
+		xive->src_count++;
+	state->valid = true;
+
+	arch_spin_unlock(&sb->lock);
+
+	return 0;
+}
+
 static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 				       struct kvm_device_attr *attr)
 {
+	struct kvmppc_xive *xive = dev->private;
+
 	switch (attr->group) {
 	case KVM_DEV_XIVE_GRP_CTRL:
 		break;
+	case KVM_DEV_XIVE_GRP_SOURCE:
+		return kvmppc_xive_native_set_source(xive, attr->attr,
+						     attr->addr);
 	}
 	return -ENXIO;
 }
@@ -175,6 +275,11 @@ static int kvmppc_xive_native_has_attr(struct kvm_device *dev,
 	switch (attr->group) {
 	case KVM_DEV_XIVE_GRP_CTRL:
 		break;
+	case KVM_DEV_XIVE_GRP_SOURCE:
+		if (attr->attr >= KVMPPC_XIVE_FIRST_IRQ &&
+		    attr->attr < KVMPPC_XIVE_NR_IRQS)
+			return 0;
+		break;
 	}
 	return -ENXIO;
 }
@@ -183,6 +288,7 @@ static void kvmppc_xive_native_free(struct kvm_device *dev)
 {
 	struct kvmppc_xive *xive = dev->private;
 	struct kvm *kvm = xive->kvm;
+	int i;
 
 	debugfs_remove(xive->dentry);
 
@@ -191,6 +297,14 @@ static void kvmppc_xive_native_free(struct kvm_device *dev)
 	if (kvm)
 		kvm->arch.xive = NULL;
 
+	/* Mask and free interrupts */
+	for (i = 0; i <= xive->max_sbid; i++) {
+		if (xive->src_blocks[i])
+			kvmppc_xive_free_sources(xive->src_blocks[i]);
+		kfree(xive->src_blocks[i]);
+		xive->src_blocks[i] = NULL;
+	}
+
 	if (xive->vp_base != XIVE_INVALID_VP)
 		xive_native_free_vp_block(xive->vp_base);
 
Documentation/virtual/kvm/devices/xive.txt +15 -0
index fdbd2ff92a88..cd8bfc37b72e 100644
@@ -17,3 +17,18 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
 
   1. KVM_DEV_XIVE_GRP_CTRL
   Provides global controls on the device
+
+  2. KVM_DEV_XIVE_GRP_SOURCE (write only)
+  Initializes a new source in the XIVE device and mask it.
+  Attributes:
+    Interrupt source number  (64-bit)
+  The kvm_device_attr.addr points to a __u64 value:
+  bits:     | 63   ....  2 |   1   |   0
+  values:   |    unused    | level | type
+  - type:  0:MSI 1:LSI
+  - level: assertion level in case of an LSI.
+  Errors:
+    -E2BIG:  Interrupt source number is out of range
+    -ENOMEM: Could not create a new source block
+    -EFAULT: Invalid user pointer for attr->addr.
+    -ENXIO:  Could not allocate underlying HW interrupt
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#21 lore raw

Re: [PATCH v2 04/16] KVM: PPC: Book3S HV: XIVE: add a control to initialize a source

Quoted 12 lines · David Gibson
Quoted 11 lines · Cédric Le Goater
arch/powerpc/kvm/book3s_xive_native.c +11 -0
+	/*
+	 * If the source doesn't already have an IPI, allocate
+	 * one and get the corresponding data
+	 */
+	if (!state->ipi_number) {
+		state->ipi_number = xive_native_alloc_irq();
+		if (state->ipi_number == 0) {
+			pr_err("Failed to allocate IRQ !\n");
+			return -ENXIO;
+		}
+		xive_native_populate_irq_data(state->ipi_number,
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev, · line 225

Quoted 6 lines
arch/powerpc/kvm/book3s_xive_native.c +6 -0
+					      &state->ipi_data);
+		pr_debug("%s allocated hw_irq=0x%x for irq=0x%lx\n", __func__,
+			 state->ipi_number, irq);
+	}
+
+	arch_spin_lock(&sb->lock);

Why the direct call to arch_spin_lock() rather than just spin_lock()?

He's sharing data structures with the xics-on-xive code, and that code has a real-mode variant, and in real mode we don't want to risk invoking lockdep code. Hence sb->lock is an arch_spinlock_t, and he has to use arch_spin_lock() on it.

Paul.

#22 lore raw

Re: [PATCH v2 04/16] KVM: PPC: Book3S HV: XIVE: add a control to initialize a source

Quoted 22 lines · Paul Mackerras
Quoted 12 lines · David Gibson
Quoted 11 lines · Cédric Le Goater
arch/powerpc/kvm/book3s_xive_native.c +11 -0
+	/*
+	 * If the source doesn't already have an IPI, allocate
+	 * one and get the corresponding data
+	 */
+	if (!state->ipi_number) {
+		state->ipi_number = xive_native_alloc_irq();
+		if (state->ipi_number == 0) {
+			pr_err("Failed to allocate IRQ !\n");
+			return -ENXIO;
+		}
+		xive_native_populate_irq_data(state->ipi_number,
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev, · line 225

Quoted 6 lines
arch/powerpc/kvm/book3s_xive_native.c +6 -0
+					      &state->ipi_data);
+		pr_debug("%s allocated hw_irq=0x%x for irq=0x%lx\n", __func__,
+			 state->ipi_number, irq);
+	}
+
+	arch_spin_lock(&sb->lock);

Why the direct call to arch_spin_lock() rather than just spin_lock()?

Quoted 4 lines

He's sharing data structures with the xics-on-xive code, and that code has a real-mode variant, and in real mode we don't want to risk invoking lockdep code. Hence sb->lock is an arch_spinlock_t, and he has to use arch_spin_lock() on it.

Ah, right.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#23 lore raw

Re: [PATCH v2 04/16] KVM: PPC: Book3S HV: XIVE: add a control to initialize a source

In reply to David Gibson
Quoted 37 lines · David Gibson
Quoted 36 lines · Cédric Le Goater

The associated HW interrupt source is simply allocated at the OPAL/HW level and then MASKED. KVM only needs to know about its type: LSI or MSI.

---
 arch/powerpc/include/uapi/asm/kvm.h        |   5 +
 arch/powerpc/kvm/book3s_xive.h             |  10 ++
 arch/powerpc/kvm/book3s_xive.c             |   8 +-
 arch/powerpc/kvm/book3s_xive_native.c      | 114 +++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  15 +++
 5 files changed, 148 insertions(+), 4 deletions(-)
arch/powerpc/include/uapi/asm/kvm.h +5 -0
index b002c0c67787..a9ad99f2a11b 100644
@@ -677,5 +677,10 @@ struct kvm_ppc_cpu_char {
 
 /* POWER9 XIVE Native Interrupt Controller */
 #define KVM_DEV_XIVE_GRP_CTRL		1
+#define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
+
+/* Layout of 64-bit XIVE source attribute values */
+#define KVM_XIVE_LEVEL_SENSITIVE	(1ULL << 0)
+#define KVM_XIVE_LEVEL_ASSERTED		(1ULL << 1)
 
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive.h +0 -0
index bcb1bbcf0359..f22f2d46d0f0 100644
@@ -12,6 +12,13 @@
 #ifdef CONFIG_KVM_XICS
 #include "book3s_xics.h"
 
Quoted 7 lines

arch/powerpc/kvm/book3s_xive.h line 17

Quoted 3 lines
arch/powerpc/kvm/book3s_xive.h +3 -0
+/*
+ * The XIVE IRQ number space is aligned with the XICS IRQ number
+ * space, CPU IPIs being allocated in the first 4K.

We do align these in qemu, but I don't see that the kernel part
cares: as far as it's concerned only one of XICS or XIVE is active at a time, and the irq numbers are chosen by userspace.

There is some relation with userspace nevertheless. The KVM device does not remap the numbers to some other range today and the limits are fixed values. Checks are being done in the has_attr() and the set_attr().

Quoted 15 lines
Quoted 15 lines
arch/powerpc/kvm/book3s_xive.h +7 -0
+ */
+#define KVMPPC_XIVE_FIRST_IRQ	0
+#define KVMPPC_XIVE_NR_IRQS	KVMPPC_XICS_NR_IRQS
+
 /*
  * State for one guest irq source.
  *
@@ -253,6 +260,9 @@ extern int (*__xive_vm_h_eoi)(struct kvm_vcpu *vcpu, unsigned long xirr);
  */
 void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu);
 int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu);
+struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
+	struct kvmppc_xive *xive, int irq);
+void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb);
 
Quoted 10 lines

arch/powerpc/kvm/book3s_xive.c

Quoted 6 lines
arch/powerpc/kvm/book3s_xive.h +0 -0
 #endif /* CONFIG_KVM_XICS */
 #endif /* _KVM_PPC_BOOK3S_XICS_H */
arch/powerpc/kvm/book3s_xive.c +0 -0
index d1cc18a5b1c4..6f950ecb3592 100644

I wonder if we should rename this book3s_xics_on_xive.c or something at some point, I keep getting confused because I forget that this is only dealing with host xive, not guest xive.

I am fine with renaming. Any objections ? book3s_xics_p9.c ?

Quoted 4 lines
Quoted 4 lines
arch/powerpc/kvm/book3s_xive.c +0 -0
@@ -1485,8 +1485,8 @@ static int xive_get_source(struct kvmppc_xive *xive, long irq, u64 addr)
 	return 0;
 }
 
Quoted 11 lines

arch/powerpc/kvm/book3s_xive.c static int xive_get_source(struct kvmppc_xive *xive, long irq, u64 addr) · line 1492

Quoted 7 lines
arch/powerpc/kvm/book3s_xive.c +2 -2
-static struct kvmppc_xive_src_block *xive_create_src_block(struct kvmppc_xive *xive,
-							   int irq)
+struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
+	struct kvmppc_xive *xive, int irq)
 {
 	struct kvm *kvm = xive->kvm;
 	struct kvmppc_xive_src_block *sb;

It's odd that this function, now used from the xive-on-xive path as well as the xics-on-xive path references KVMPPC_XICS_ICS_SHIFT a few lines down from this change.

Yes. This is because of the definition of the struct kvmppc_xive_src_block.

We could introduce new defines for XIVE or a common set of defines for XICS and XIVE.

Quoted 80 lines
Quoted 80 lines
arch/powerpc/kvm/book3s_xive.c +2 -2
@@ -1565,7 +1565,7 @@ static int xive_set_source(struct kvmppc_xive *xive, long irq, u64 addr)
 	sb = kvmppc_xive_find_source(xive, irq, &idx);
 	if (!sb) {
 		pr_devel("No source, creating source block...\n");
-		sb = xive_create_src_block(xive, irq);
+		sb = kvmppc_xive_create_src_block(xive, irq);
 		if (!sb) {
 			pr_devel("Failed to create block...\n");
 			return -ENOMEM;
@@ -1789,7 +1789,7 @@ static void kvmppc_xive_cleanup_irq(u32 hw_num, struct xive_irq_data *xd)
 	xive_cleanup_irq_data(xd);
 }
 
-static void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb)
+void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb)
 {
 	int i;
 
arch/powerpc/kvm/book3s_xive_native.c +47 -0
index 1f3da47a4a6a..a9b2d2d9af99 100644
@@ -31,6 +31,29 @@
 
 #include "book3s_xive.h"
 
+/*
+ * TODO: introduce a common template file with the XIVE native layer
+ * and the XICS-on-XIVE glue for the utility functions
+ */
+#define __x_eoi_page(xd)	((void __iomem *)((xd)->eoi_mmio))
+#define __x_trig_page(xd)	((void __iomem *)((xd)->trig_mmio))
+#define __x_readq	__raw_readq
+#define __x_writeq	__raw_writeq
+
+static u8 xive_vm_esb_load(struct xive_irq_data *xd, u32 offset)
+{
+	u64 val;
+
+	if (xd->flags & XIVE_IRQ_FLAG_SHIFT_BUG)
+		offset |= offset << 4;
+
+	val = __x_readq(__x_eoi_page(xd) + offset);
+#ifdef __LITTLE_ENDIAN__
+	val >>= 64-8;
+#endif
+	return (u8)val;
+}
+
 static void kvmppc_xive_native_cleanup_queue(struct kvm_vcpu *vcpu, int prio)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
@@ -153,12 +176,89 @@ int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 	return rc;
 }
 
+static int kvmppc_xive_native_set_source(struct kvmppc_xive *xive, long irq,
+					 u64 addr)
+{
+	struct kvmppc_xive_src_block *sb;
+	struct kvmppc_xive_irq_state *state;
+	u64 __user *ubufp = (u64 __user *) addr;
+	u64 val;
+	u16 idx;
+
+	pr_devel("%s irq=0x%lx\n", __func__, irq);
+
+	if (irq < KVMPPC_XIVE_FIRST_IRQ || irq >= KVMPPC_XIVE_NR_IRQS)
+		return -E2BIG;
+
+	sb = kvmppc_xive_find_source(xive, irq, &idx);
+	if (!sb) {
+		pr_debug("No source, creating source block...\n");
+		sb = kvmppc_xive_create_src_block(xive, irq);
+		if (!sb) {
+			pr_err("Failed to create block...\n");
+			return -ENOMEM;
+		}
+	}
+	state = &sb->irq_state[idx];
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev, · line 207

Quoted 5 lines
arch/powerpc/kvm/book3s_xive_native.c +5 -0
+
+	if (get_user(val, ubufp)) {
+		pr_err("fault getting user info !\n");
+		return -EFAULT;
+	}

You should validate the value loaded here to check it doesn't have any bits set we don't know about.

ok

Quoted 13 lines
Quoted 13 lines
arch/powerpc/kvm/book3s_xive_native.c +12 -0
+
+	/*
+	 * If the source doesn't already have an IPI, allocate
+	 * one and get the corresponding data
+	 */
+	if (!state->ipi_number) {
+		state->ipi_number = xive_native_alloc_irq();
+		if (state->ipi_number == 0) {
+			pr_err("Failed to allocate IRQ !\n");
+			return -ENXIO;
+		}
+		xive_native_populate_irq_data(state->ipi_number,
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev, · line 225

Quoted 6 lines
arch/powerpc/kvm/book3s_xive_native.c +6 -0
+					      &state->ipi_data);
+		pr_debug("%s allocated hw_irq=0x%x for irq=0x%lx\n", __func__,
+			 state->ipi_number, irq);
+	}
+
+	arch_spin_lock(&sb->lock);

Why the direct call to arch_spin_lock() rather than just spin_lock()?

Paul answered this question but may be I should make the effort to decouple both devices on this aspect.

Thanks,

C.

Quoted 98 lines
Quoted 97 lines
arch/powerpc/kvm/book3s_xive_native.c +44 -0
+
+	/* Restore LSI state */
+	if (val & KVM_XIVE_LEVEL_SENSITIVE) {
+		state->lsi = true;
+		if (val & KVM_XIVE_LEVEL_ASSERTED)
+			state->asserted = true;
+		pr_devel("  LSI ! Asserted=%d\n", state->asserted);
+	}
+
+	/* Mask IRQ to start with */
+	state->act_server = 0;
+	state->act_priority = MASKED;
+	xive_vm_esb_load(&state->ipi_data, XIVE_ESB_SET_PQ_01);
+	xive_native_configure_irq(state->ipi_number, 0, MASKED, 0);
+
+	/* Increment the number of valid sources and mark this one valid */
+	if (!state->valid)
+		xive->src_count++;
+	state->valid = true;
+
+	arch_spin_unlock(&sb->lock);
+
+	return 0;
+}
+
 static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 				       struct kvm_device_attr *attr)
 {
+	struct kvmppc_xive *xive = dev->private;
+
 	switch (attr->group) {
 	case KVM_DEV_XIVE_GRP_CTRL:
 		break;
+	case KVM_DEV_XIVE_GRP_SOURCE:
+		return kvmppc_xive_native_set_source(xive, attr->attr,
+						     attr->addr);
 	}
 	return -ENXIO;
 }
@@ -175,6 +275,11 @@ static int kvmppc_xive_native_has_attr(struct kvm_device *dev,
 	switch (attr->group) {
 	case KVM_DEV_XIVE_GRP_CTRL:
 		break;
+	case KVM_DEV_XIVE_GRP_SOURCE:
+		if (attr->attr >= KVMPPC_XIVE_FIRST_IRQ &&
+		    attr->attr < KVMPPC_XIVE_NR_IRQS)
+			return 0;
+		break;
 	}
 	return -ENXIO;
 }
@@ -183,6 +288,7 @@ static void kvmppc_xive_native_free(struct kvm_device *dev)
 {
 	struct kvmppc_xive *xive = dev->private;
 	struct kvm *kvm = xive->kvm;
+	int i;
 
 	debugfs_remove(xive->dentry);
 
@@ -191,6 +297,14 @@ static void kvmppc_xive_native_free(struct kvm_device *dev)
 	if (kvm)
 		kvm->arch.xive = NULL;
 
+	/* Mask and free interrupts */
+	for (i = 0; i <= xive->max_sbid; i++) {
+		if (xive->src_blocks[i])
+			kvmppc_xive_free_sources(xive->src_blocks[i]);
+		kfree(xive->src_blocks[i]);
+		xive->src_blocks[i] = NULL;
+	}
+
 	if (xive->vp_base != XIVE_INVALID_VP)
 		xive_native_free_vp_block(xive->vp_base);
 
Documentation/virtual/kvm/devices/xive.txt +15 -0
index fdbd2ff92a88..cd8bfc37b72e 100644
@@ -17,3 +17,18 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
 
   1. KVM_DEV_XIVE_GRP_CTRL
   Provides global controls on the device
+
+  2. KVM_DEV_XIVE_GRP_SOURCE (write only)
+  Initializes a new source in the XIVE device and mask it.
+  Attributes:
+    Interrupt source number  (64-bit)
+  The kvm_device_attr.addr points to a __u64 value:
+  bits:     | 63   ....  2 |   1   |   0
+  values:   |    unused    | level | type
+  - type:  0:MSI 1:LSI
+  - level: assertion level in case of an LSI.
+  Errors:
+    -E2BIG:  Interrupt source number is out of range
+    -ENOMEM: Could not create a new source block
+    -EFAULT: Invalid user pointer for attr->addr.
+    -ENXIO:  Could not allocate underlying HW interrupt
#24 lore raw

Re: [PATCH v2 04/16] KVM: PPC: Book3S HV: XIVE: add a control to initialize a source

Quoted 46 lines · Cédric Le Goater
Quoted 37 lines · David Gibson
Quoted 36 lines · Cédric Le Goater

The associated HW interrupt source is simply allocated at the OPAL/HW level and then MASKED. KVM only needs to know about its type: LSI or MSI.

---
 arch/powerpc/include/uapi/asm/kvm.h        |   5 +
 arch/powerpc/kvm/book3s_xive.h             |  10 ++
 arch/powerpc/kvm/book3s_xive.c             |   8 +-
 arch/powerpc/kvm/book3s_xive_native.c      | 114 +++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  15 +++
 5 files changed, 148 insertions(+), 4 deletions(-)
arch/powerpc/include/uapi/asm/kvm.h +5 -0
index b002c0c67787..a9ad99f2a11b 100644
@@ -677,5 +677,10 @@ struct kvm_ppc_cpu_char {
 
 /* POWER9 XIVE Native Interrupt Controller */
 #define KVM_DEV_XIVE_GRP_CTRL		1
+#define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
+
+/* Layout of 64-bit XIVE source attribute values */
+#define KVM_XIVE_LEVEL_SENSITIVE	(1ULL << 0)
+#define KVM_XIVE_LEVEL_ASSERTED		(1ULL << 1)
 
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive.h +0 -0
index bcb1bbcf0359..f22f2d46d0f0 100644
@@ -12,6 +12,13 @@
 #ifdef CONFIG_KVM_XICS
 #include "book3s_xics.h"
 
Quoted 7 lines

arch/powerpc/kvm/book3s_xive.h line 17

Quoted 3 lines
arch/powerpc/kvm/book3s_xive.h +3 -0
+/*
+ * The XIVE IRQ number space is aligned with the XICS IRQ number
+ * space, CPU IPIs being allocated in the first 4K.

We do align these in qemu, but I don't see that the kernel part
cares: as far as it's concerned only one of XICS or XIVE is active at a time, and the irq numbers are chosen by userspace.

Quoted 3 lines

There is some relation with userspace nevertheless. The KVM device does not remap the numbers to some other range today and the limits are fixed values. Checks are being done in the has_attr() and the set_attr().

Hrm. I still think the comment needs to describe what the constraint is from the point of view of the kernel, which this isn't. Maybe, "we allow irqs in the range X..Y (allowing userspace to do ...)"

Quoted 27 lines
Quoted 16 lines
Quoted 16 lines
arch/powerpc/kvm/book3s_xive.h +7 -0
+ */
+#define KVMPPC_XIVE_FIRST_IRQ	0
+#define KVMPPC_XIVE_NR_IRQS	KVMPPC_XICS_NR_IRQS
+
 /*
  * State for one guest irq source.
  *
@@ -253,6 +260,9 @@ extern int (*__xive_vm_h_eoi)(struct kvm_vcpu *vcpu, unsigned long xirr);
  */
 void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu);
 int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu);
+struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
+	struct kvmppc_xive *xive, int irq);
+void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb);
 
Quoted 10 lines

arch/powerpc/kvm/book3s_xive.c

Quoted 6 lines
arch/powerpc/kvm/book3s_xive.h +0 -0
 #endif /* CONFIG_KVM_XICS */
 #endif /* _KVM_PPC_BOOK3S_XICS_H */
arch/powerpc/kvm/book3s_xive.c +0 -0
index d1cc18a5b1c4..6f950ecb3592 100644

I wonder if we should rename this book3s_xics_on_xive.c or something at some point, I keep getting confused because I forget that this is only dealing with host xive, not guest xive.

Quoted 1 line

I am fine with renaming. Any objections ? book3s_xics_p9.c ?

Hm, maybe?

Quoted 16 lines
Quoted 4 lines
Quoted 4 lines
arch/powerpc/kvm/book3s_xive.c +0 -0
@@ -1485,8 +1485,8 @@ static int xive_get_source(struct kvmppc_xive *xive, long irq, u64 addr)
 	return 0;
 }
 
Quoted 11 lines

arch/powerpc/kvm/book3s_xive.c static int xive_get_source(struct kvmppc_xive *xive, long irq, u64 addr) · line 1492

Quoted 7 lines
arch/powerpc/kvm/book3s_xive.c +2 -2
-static struct kvmppc_xive_src_block *xive_create_src_block(struct kvmppc_xive *xive,
-							   int irq)
+struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
+	struct kvmppc_xive *xive, int irq)
 {
 	struct kvm *kvm = xive->kvm;
 	struct kvmppc_xive_src_block *sb;

It's odd that this function, now used from the xive-on-xive path as well as the xics-on-xive path references KVMPPC_XICS_ICS_SHIFT a few lines down from this change.

Quoted 4 lines

Yes. This is because of the definition of the struct kvmppc_xive_src_block.

We could introduce new defines for XIVE or a common set of defines for XICS and XIVE.

I think making common definitions would be best, if possible.

Quoted 220 lines
Quoted 81 lines
Quoted 81 lines
arch/powerpc/kvm/book3s_xive.c +2 -2
@@ -1565,7 +1565,7 @@ static int xive_set_source(struct kvmppc_xive *xive, long irq, u64 addr)
 	sb = kvmppc_xive_find_source(xive, irq, &idx);
 	if (!sb) {
 		pr_devel("No source, creating source block...\n");
-		sb = xive_create_src_block(xive, irq);
+		sb = kvmppc_xive_create_src_block(xive, irq);
 		if (!sb) {
 			pr_devel("Failed to create block...\n");
 			return -ENOMEM;
@@ -1789,7 +1789,7 @@ static void kvmppc_xive_cleanup_irq(u32 hw_num, struct xive_irq_data *xd)
 	xive_cleanup_irq_data(xd);
 }
 
-static void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb)
+void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb)
 {
 	int i;
 
arch/powerpc/kvm/book3s_xive_native.c +47 -0
index 1f3da47a4a6a..a9b2d2d9af99 100644
@@ -31,6 +31,29 @@
 
 #include "book3s_xive.h"
 
+/*
+ * TODO: introduce a common template file with the XIVE native layer
+ * and the XICS-on-XIVE glue for the utility functions
+ */
+#define __x_eoi_page(xd)	((void __iomem *)((xd)->eoi_mmio))
+#define __x_trig_page(xd)	((void __iomem *)((xd)->trig_mmio))
+#define __x_readq	__raw_readq
+#define __x_writeq	__raw_writeq
+
+static u8 xive_vm_esb_load(struct xive_irq_data *xd, u32 offset)
+{
+	u64 val;
+
+	if (xd->flags & XIVE_IRQ_FLAG_SHIFT_BUG)
+		offset |= offset << 4;
+
+	val = __x_readq(__x_eoi_page(xd) + offset);
+#ifdef __LITTLE_ENDIAN__
+	val >>= 64-8;
+#endif
+	return (u8)val;
+}
+
 static void kvmppc_xive_native_cleanup_queue(struct kvm_vcpu *vcpu, int prio)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
@@ -153,12 +176,89 @@ int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 	return rc;
 }
 
+static int kvmppc_xive_native_set_source(struct kvmppc_xive *xive, long irq,
+					 u64 addr)
+{
+	struct kvmppc_xive_src_block *sb;
+	struct kvmppc_xive_irq_state *state;
+	u64 __user *ubufp = (u64 __user *) addr;
+	u64 val;
+	u16 idx;
+
+	pr_devel("%s irq=0x%lx\n", __func__, irq);
+
+	if (irq < KVMPPC_XIVE_FIRST_IRQ || irq >= KVMPPC_XIVE_NR_IRQS)
+		return -E2BIG;
+
+	sb = kvmppc_xive_find_source(xive, irq, &idx);
+	if (!sb) {
+		pr_debug("No source, creating source block...\n");
+		sb = kvmppc_xive_create_src_block(xive, irq);
+		if (!sb) {
+			pr_err("Failed to create block...\n");
+			return -ENOMEM;
+		}
+	}
+	state = &sb->irq_state[idx];
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev, · line 207

Quoted 5 lines
arch/powerpc/kvm/book3s_xive_native.c +5 -0
+
+	if (get_user(val, ubufp)) {
+		pr_err("fault getting user info !\n");
+		return -EFAULT;
+	}

You should validate the value loaded here to check it doesn't have any bits set we don't know about.

ok

Quoted 13 lines
Quoted 13 lines
arch/powerpc/kvm/book3s_xive_native.c +12 -0
+
+	/*
+	 * If the source doesn't already have an IPI, allocate
+	 * one and get the corresponding data
+	 */
+	if (!state->ipi_number) {
+		state->ipi_number = xive_native_alloc_irq();
+		if (state->ipi_number == 0) {
+			pr_err("Failed to allocate IRQ !\n");
+			return -ENXIO;
+		}
+		xive_native_populate_irq_data(state->ipi_number,
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev, · line 225

Quoted 6 lines
arch/powerpc/kvm/book3s_xive_native.c +6 -0
+					      &state->ipi_data);
+		pr_debug("%s allocated hw_irq=0x%x for irq=0x%lx\n", __func__,
+			 state->ipi_number, irq);
+	}
+
+	arch_spin_lock(&sb->lock);

Why the direct call to arch_spin_lock() rather than just spin_lock()?

Paul answered this question but may be I should make the effort to decouple both devices on this aspect.

Thanks,

C.

Quoted 98 lines
Quoted 97 lines
arch/powerpc/kvm/book3s_xive_native.c +44 -0
+
+	/* Restore LSI state */
+	if (val & KVM_XIVE_LEVEL_SENSITIVE) {
+		state->lsi = true;
+		if (val & KVM_XIVE_LEVEL_ASSERTED)
+			state->asserted = true;
+		pr_devel("  LSI ! Asserted=%d\n", state->asserted);
+	}
+
+	/* Mask IRQ to start with */
+	state->act_server = 0;
+	state->act_priority = MASKED;
+	xive_vm_esb_load(&state->ipi_data, XIVE_ESB_SET_PQ_01);
+	xive_native_configure_irq(state->ipi_number, 0, MASKED, 0);
+
+	/* Increment the number of valid sources and mark this one valid */
+	if (!state->valid)
+		xive->src_count++;
+	state->valid = true;
+
+	arch_spin_unlock(&sb->lock);
+
+	return 0;
+}
+
 static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 				       struct kvm_device_attr *attr)
 {
+	struct kvmppc_xive *xive = dev->private;
+
 	switch (attr->group) {
 	case KVM_DEV_XIVE_GRP_CTRL:
 		break;
+	case KVM_DEV_XIVE_GRP_SOURCE:
+		return kvmppc_xive_native_set_source(xive, attr->attr,
+						     attr->addr);
 	}
 	return -ENXIO;
 }
@@ -175,6 +275,11 @@ static int kvmppc_xive_native_has_attr(struct kvm_device *dev,
 	switch (attr->group) {
 	case KVM_DEV_XIVE_GRP_CTRL:
 		break;
+	case KVM_DEV_XIVE_GRP_SOURCE:
+		if (attr->attr >= KVMPPC_XIVE_FIRST_IRQ &&
+		    attr->attr < KVMPPC_XIVE_NR_IRQS)
+			return 0;
+		break;
 	}
 	return -ENXIO;
 }
@@ -183,6 +288,7 @@ static void kvmppc_xive_native_free(struct kvm_device *dev)
 {
 	struct kvmppc_xive *xive = dev->private;
 	struct kvm *kvm = xive->kvm;
+	int i;
 
 	debugfs_remove(xive->dentry);
 
@@ -191,6 +297,14 @@ static void kvmppc_xive_native_free(struct kvm_device *dev)
 	if (kvm)
 		kvm->arch.xive = NULL;
 
+	/* Mask and free interrupts */
+	for (i = 0; i <= xive->max_sbid; i++) {
+		if (xive->src_blocks[i])
+			kvmppc_xive_free_sources(xive->src_blocks[i]);
+		kfree(xive->src_blocks[i]);
+		xive->src_blocks[i] = NULL;
+	}
+
 	if (xive->vp_base != XIVE_INVALID_VP)
 		xive_native_free_vp_block(xive->vp_base);
 
Documentation/virtual/kvm/devices/xive.txt +15 -0
index fdbd2ff92a88..cd8bfc37b72e 100644
@@ -17,3 +17,18 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
 
   1. KVM_DEV_XIVE_GRP_CTRL
   Provides global controls on the device
+
+  2. KVM_DEV_XIVE_GRP_SOURCE (write only)
+  Initializes a new source in the XIVE device and mask it.
+  Attributes:
+    Interrupt source number  (64-bit)
+  The kvm_device_attr.addr points to a __u64 value:
+  bits:     | 63   ....  2 |   1   |   0
+  values:   |    unused    | level | type
+  - type:  0:MSI 1:LSI
+  - level: assertion level in case of an LSI.
+  Errors:
+    -E2BIG:  Interrupt source number is out of range
+    -ENOMEM: Could not create a new source block
+    -EFAULT: Invalid user pointer for attr->addr.
+    -ENXIO:  Could not allocate underlying HW interrupt
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#25 lore raw

Re: [PATCH v2 04/16] KVM: PPC: Book3S HV: XIVE: add a control to initialize a source

In reply to Cédric Le Goater
Quoted 3 lines · Cédric Le Goater

The associated HW interrupt source is simply allocated at the OPAL/HW level and then MASKED. KVM only needs to know about its type: LSI or MSI.

I think it would be helpful to explain to the reader here that with XIVE, all interrupts have a hardware source, even IPIs and virtual device interrupts, for which we allocate a software-triggerable interrupt source in the XIVE hardware.

Quoted 44 lines
---
 arch/powerpc/include/uapi/asm/kvm.h        |   5 +
 arch/powerpc/kvm/book3s_xive.h             |  10 ++
 arch/powerpc/kvm/book3s_xive.c             |   8 +-
 arch/powerpc/kvm/book3s_xive_native.c      | 114 +++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  15 +++
 5 files changed, 148 insertions(+), 4 deletions(-)
arch/powerpc/include/uapi/asm/kvm.h +5 -0
index b002c0c67787..a9ad99f2a11b 100644
@@ -677,5 +677,10 @@ struct kvm_ppc_cpu_char {
 
 /* POWER9 XIVE Native Interrupt Controller */
 #define KVM_DEV_XIVE_GRP_CTRL		1
+#define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
+
+/* Layout of 64-bit XIVE source attribute values */
+#define KVM_XIVE_LEVEL_SENSITIVE	(1ULL << 0)
+#define KVM_XIVE_LEVEL_ASSERTED		(1ULL << 1)
 
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive.h +7 -0
index bcb1bbcf0359..f22f2d46d0f0 100644
@@ -12,6 +12,13 @@
 #ifdef CONFIG_KVM_XICS
 #include "book3s_xics.h"
 
+/*
+ * The XIVE IRQ number space is aligned with the XICS IRQ number
+ * space, CPU IPIs being allocated in the first 4K.
+ */
+#define KVMPPC_XIVE_FIRST_IRQ	0
+#define KVMPPC_XIVE_NR_IRQS	KVMPPC_XICS_NR_IRQS
+
 /*
  * State for one guest irq source.
  *

arch/powerpc/kvm/book3s_xive.h extern int (*__xive_vm_h_eoi)(struct kvm_vcpu *vcpu, unsigned long xirr); · line 265

Quoted 6 lines
arch/powerpc/kvm/book3s_xive.h +3 -0
@@ -253,6 +260,9 @@ extern int (*__xive_vm_h_eoi)(struct kvm_vcpu *vcpu, unsigned long xirr);
  */
 void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu);
 int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu);
+struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
+	struct kvmppc_xive *xive, int irq);
+void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb);

So we're using the same source block data structure (effectively a 2-level tree) that the XICS-on-XIVE code uses. That would be worth mentioning as a design choice in the patch description.

Quoted 51 lines
arch/powerpc/kvm/book3s_xive.h +0 -0
 #endif /* CONFIG_KVM_XICS */
 #endif /* _KVM_PPC_BOOK3S_XICS_H */
arch/powerpc/kvm/book3s_xive.c +4 -4
index d1cc18a5b1c4..6f950ecb3592 100644
@@ -1485,8 +1485,8 @@ static int xive_get_source(struct kvmppc_xive *xive, long irq, u64 addr)
 	return 0;
 }
 
-static struct kvmppc_xive_src_block *xive_create_src_block(struct kvmppc_xive *xive,
-							   int irq)
+struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
+	struct kvmppc_xive *xive, int irq)
 {
 	struct kvm *kvm = xive->kvm;
 	struct kvmppc_xive_src_block *sb;
@@ -1565,7 +1565,7 @@ static int xive_set_source(struct kvmppc_xive *xive, long irq, u64 addr)
 	sb = kvmppc_xive_find_source(xive, irq, &idx);
 	if (!sb) {
 		pr_devel("No source, creating source block...\n");
-		sb = xive_create_src_block(xive, irq);
+		sb = kvmppc_xive_create_src_block(xive, irq);
 		if (!sb) {
 			pr_devel("Failed to create block...\n");
 			return -ENOMEM;
@@ -1789,7 +1789,7 @@ static void kvmppc_xive_cleanup_irq(u32 hw_num, struct xive_irq_data *xd)
 	xive_cleanup_irq_data(xd);
 }
 
-static void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb)
+void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb)
 {
 	int i;
 
arch/powerpc/kvm/book3s_xive_native.c +7 -0
index 1f3da47a4a6a..a9b2d2d9af99 100644
@@ -31,6 +31,29 @@
 
 #include "book3s_xive.h"
 
+/*
+ * TODO: introduce a common template file with the XIVE native layer
+ * and the XICS-on-XIVE glue for the utility functions
+ */
+#define __x_eoi_page(xd)	((void __iomem *)((xd)->eoi_mmio))
+#define __x_trig_page(xd)	((void __iomem *)((xd)->trig_mmio))
+#define __x_readq	__raw_readq

arch/powerpc/kvm/book3s_xive_native.c line 50

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+#define __x_writeq	__raw_writeq
+
+static u8 xive_vm_esb_load(struct xive_irq_data *xd, u32 offset)
+{
+	u64 val;
+
+	if (xd->flags & XIVE_IRQ_FLAG_SHIFT_BUG)
+		offset |= offset << 4;
+
+	val = __x_readq(__x_eoi_page(xd) + offset);

Ben originally defined the __x_* macros so that he could use the same source code twice, once for real mode and once for virtual mode. Since you're not doing that, is there really any reason for this indirection? Why not just __raw_readq, __raw_writeq etc. directly?

Paul.

#26 lore raw

PATCH v2 5/16 KVM: PPC: Book3S HV: XIVE: add a control to configure a source

In reply to Cédric Le Goater

This control will be used by the H_INT_SET_SOURCE_CONFIG hcall from QEMU and also to restore the configuration of the source in the KVM device.

The XIVE internal IRQ structure is extended with the value of the Effective Interrupt Source Number. The EISN is the interrupt number pushed in the event queue that the guest OS will use to dispatch events internally. Caching the EISN value in KVM ease the test when checking if a reconfiguration is indeed needed.

---
 arch/powerpc/include/uapi/asm/kvm.h        | 11 +++
 arch/powerpc/kvm/book3s_xive.h             |  4 +
 arch/powerpc/kvm/book3s_xive.c             |  5 +-
 arch/powerpc/kvm/book3s_xive_native.c      | 97 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt | 20 +++++
 5 files changed, 135 insertions(+), 2 deletions(-)
arch/powerpc/include/uapi/asm/kvm.h +11 -0
index a9ad99f2a11b..91899c7f9abd 100644
@@ -678,9 +678,20 @@ struct kvm_ppc_cpu_char {
 /* POWER9 XIVE Native Interrupt Controller */
 #define KVM_DEV_XIVE_GRP_CTRL		1
 #define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
+#define KVM_DEV_XIVE_GRP_SOURCE_CONFIG	3	/* 64-bit source attributes */
 
 /* Layout of 64-bit XIVE source attribute values */
 #define KVM_XIVE_LEVEL_SENSITIVE	(1ULL << 0)
 #define KVM_XIVE_LEVEL_ASSERTED		(1ULL << 1)
 
+/* Layout of 64-bit XIVE source configuration attribute values */
+#define KVM_XIVE_SOURCE_PRIORITY_SHIFT	0
+#define KVM_XIVE_SOURCE_PRIORITY_MASK	0x7
+#define KVM_XIVE_SOURCE_SERVER_SHIFT	3
+#define KVM_XIVE_SOURCE_SERVER_MASK	0xfffffff8ULL
+#define KVM_XIVE_SOURCE_MASK_SHIFT	32
+#define KVM_XIVE_SOURCE_MASK_MASK	0x100000000ULL
+#define KVM_XIVE_SOURCE_EISN_SHIFT	33
+#define KVM_XIVE_SOURCE_EISN_MASK	0xfffffffe00000000ULL
+
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive.h +4 -0
index f22f2d46d0f0..ab3ac152980d 100644
@@ -61,6 +61,9 @@ struct kvmppc_xive_irq_state {
 	bool saved_p;
 	bool saved_q;
 	u8 saved_scan_prio;
+
+	/* Xive native */
+	u32 eisn;			/* Guest Effective IRQ number */
 };
 
 /* Select the "right" interrupt (IPI vs. passthrough) */
@@ -263,6 +266,7 @@ int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu);
 struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
 	struct kvmppc_xive *xive, int irq);
 void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb);
+int kvmppc_xive_select_target(struct kvm *kvm, u32 *server, u8 prio);
 
 #endif /* CONFIG_KVM_XICS */
 #endif /* _KVM_PPC_BOOK3S_XICS_H */
arch/powerpc/kvm/book3s_xive.c +3 -2
index 6f950ecb3592..086da91d7c6e 100644
@@ -342,7 +342,7 @@ static int xive_try_pick_queue(struct kvm_vcpu *vcpu, u8 prio)
 	return atomic_add_unless(&q->count, 1, max) ? 0 : -EBUSY;
 }
 
-static int xive_select_target(struct kvm *kvm, u32 *server, u8 prio)
+int kvmppc_xive_select_target(struct kvm *kvm, u32 *server, u8 prio)
 {
 	struct kvm_vcpu *vcpu;
 	int i, rc;
@@ -535,7 +535,7 @@ static int xive_target_interrupt(struct kvm *kvm,
 	 * priority. The count for that new target will have
 	 * already been incremented.
 	 */
-	rc = xive_select_target(kvm, &server, prio);
+	rc = kvmppc_xive_select_target(kvm, &server, prio);
 
 	/*
 	 * We failed to find a target ? Not much we can do
@@ -1509,6 +1509,7 @@ struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
 
 	for (i = 0; i < KVMPPC_XICS_IRQ_PER_ICS; i++) {
 		sb->irq_state[i].number = (bid << KVMPPC_XICS_ICS_SHIFT) | i;
+		sb->irq_state[i].eisn = 0;
 		sb->irq_state[i].guest_priority = MASKED;
 		sb->irq_state[i].saved_priority = MASKED;
 		sb->irq_state[i].act_priority = MASKED;
arch/powerpc/kvm/book3s_xive_native.c +97 -0

115 lines not on this page · raw message

Documentation/virtual/kvm/devices/xive.txt +20 -0
index cd8bfc37b72e..4f513a1880c7 100644
@@ -32,3 +32,23 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -ENOMEM: Could not create a new source block
     -EFAULT: Invalid user pointer for attr->addr.
     -ENXIO:  Could not allocate underlying HW interrupt
+
+  3. KVM_DEV_XIVE_GRP_SOURCE_CONFIG (write only)
+  Configures source targeting
+  Attributes:
+    Interrupt source number  (64-bit)
+  The kvm_device_attr.addr points to a __u64 value:
+  bits:     | 63   ....  33 |  32  | 31 .. 3 |  2 .. 0
+  values:   |    eisn       | mask |  server | priority
+  - priority: 0-7 interrupt priority level
+  - server: CPU number chosen to handle the interrupt
+  - mask: mask flag (unused)
+  - eisn: Effective Interrupt Source Number
+  Errors:
+    -ENOENT: Unknown source number
+    -EINVAL: Not initialized source number, invalid priority or
+             invalid CPU number.
+    -EFAULT: Invalid user pointer for attr->addr.
+    -ENXIO:  CPU event queues not configured or configuration of the
+             underlying HW interrupt failed
+    -EBUSY:  No CPU available to serve interrupt
Signature · 1 line
2.20.1
Reviewed-by on v2 5/16
#27 lore raw

Re: [PATCH v2 05/16] KVM: PPC: Book3S HV: XIVE: add a control to configure a source

Quoted 11 lines · Cédric Le Goater

This control will be used by the H_INT_SET_SOURCE_CONFIG hcall from QEMU and also to restore the configuration of the source in the KVM device.

The XIVE internal IRQ structure is extended with the value of the Effective Interrupt Source Number. The EISN is the interrupt number pushed in the event queue that the guest OS will use to dispatch events internally. Caching the EISN value in KVM ease the test when checking if a reconfiguration is indeed needed.

Quoted 235 lines
---
 arch/powerpc/include/uapi/asm/kvm.h        | 11 +++
 arch/powerpc/kvm/book3s_xive.h             |  4 +
 arch/powerpc/kvm/book3s_xive.c             |  5 +-
 arch/powerpc/kvm/book3s_xive_native.c      | 97 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt | 20 +++++
 5 files changed, 135 insertions(+), 2 deletions(-)
arch/powerpc/include/uapi/asm/kvm.h +11 -0
index a9ad99f2a11b..91899c7f9abd 100644
@@ -678,9 +678,20 @@ struct kvm_ppc_cpu_char {
 /* POWER9 XIVE Native Interrupt Controller */
 #define KVM_DEV_XIVE_GRP_CTRL		1
 #define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
+#define KVM_DEV_XIVE_GRP_SOURCE_CONFIG	3	/* 64-bit source attributes */
 
 /* Layout of 64-bit XIVE source attribute values */
 #define KVM_XIVE_LEVEL_SENSITIVE	(1ULL << 0)
 #define KVM_XIVE_LEVEL_ASSERTED		(1ULL << 1)
 
+/* Layout of 64-bit XIVE source configuration attribute values */
+#define KVM_XIVE_SOURCE_PRIORITY_SHIFT	0
+#define KVM_XIVE_SOURCE_PRIORITY_MASK	0x7
+#define KVM_XIVE_SOURCE_SERVER_SHIFT	3
+#define KVM_XIVE_SOURCE_SERVER_MASK	0xfffffff8ULL
+#define KVM_XIVE_SOURCE_MASK_SHIFT	32
+#define KVM_XIVE_SOURCE_MASK_MASK	0x100000000ULL
+#define KVM_XIVE_SOURCE_EISN_SHIFT	33
+#define KVM_XIVE_SOURCE_EISN_MASK	0xfffffffe00000000ULL
+
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive.h +4 -0
index f22f2d46d0f0..ab3ac152980d 100644
@@ -61,6 +61,9 @@ struct kvmppc_xive_irq_state {
 	bool saved_p;
 	bool saved_q;
 	u8 saved_scan_prio;
+
+	/* Xive native */
+	u32 eisn;			/* Guest Effective IRQ number */
 };
 
 /* Select the "right" interrupt (IPI vs. passthrough) */
@@ -263,6 +266,7 @@ int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu);
 struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
 	struct kvmppc_xive *xive, int irq);
 void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb);
+int kvmppc_xive_select_target(struct kvm *kvm, u32 *server, u8 prio);
 
 #endif /* CONFIG_KVM_XICS */
 #endif /* _KVM_PPC_BOOK3S_XICS_H */
arch/powerpc/kvm/book3s_xive.c +3 -2
index 6f950ecb3592..086da91d7c6e 100644
@@ -342,7 +342,7 @@ static int xive_try_pick_queue(struct kvm_vcpu *vcpu, u8 prio)
 	return atomic_add_unless(&q->count, 1, max) ? 0 : -EBUSY;
 }
 
-static int xive_select_target(struct kvm *kvm, u32 *server, u8 prio)
+int kvmppc_xive_select_target(struct kvm *kvm, u32 *server, u8 prio)
 {
 	struct kvm_vcpu *vcpu;
 	int i, rc;
@@ -535,7 +535,7 @@ static int xive_target_interrupt(struct kvm *kvm,
 	 * priority. The count for that new target will have
 	 * already been incremented.
 	 */
-	rc = xive_select_target(kvm, &server, prio);
+	rc = kvmppc_xive_select_target(kvm, &server, prio);
 
 	/*
 	 * We failed to find a target ? Not much we can do
@@ -1509,6 +1509,7 @@ struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
 
 	for (i = 0; i < KVMPPC_XICS_IRQ_PER_ICS; i++) {
 		sb->irq_state[i].number = (bid << KVMPPC_XICS_ICS_SHIFT) | i;
+		sb->irq_state[i].eisn = 0;
 		sb->irq_state[i].guest_priority = MASKED;
 		sb->irq_state[i].saved_priority = MASKED;
 		sb->irq_state[i].act_priority = MASKED;
arch/powerpc/kvm/book3s_xive_native.c +97 -0
index a9b2d2d9af99..cb5a5c6e05af 100644
@@ -248,6 +248,99 @@ static int kvmppc_xive_native_set_source(struct kvmppc_xive *xive, long irq,
 	return 0;
 }
 
+static int kvmppc_xive_native_update_source_config(struct kvmppc_xive *xive,
+					struct kvmppc_xive_src_block *sb,
+					struct kvmppc_xive_irq_state *state,
+					u32 server,
+					u8 priority,
+					u32 eisn)
+{
+	struct kvm *kvm = xive->kvm;
+	u32 hw_num;
+	int rc = 0;
+
+	/*
+	 * TODO: Do we need to safely mask and unmask a source ? can
+	 * we just let the guest handle the possible races ?
+	 */
+	arch_spin_lock(&sb->lock);
+
+	if (state->act_server == server && state->act_priority == priority &&
+	    state->eisn == eisn)
+		goto unlock;
+
+	pr_devel("new_act_prio=%d new_act_server=%d act_server=%d act_prio=%d\n",
+		 priority, server, state->act_server, state->act_priority);
+
+	kvmppc_xive_select_irq(state, &hw_num, NULL);
+
+	if (priority != MASKED) {
+		rc = kvmppc_xive_select_target(kvm, &server, priority);
+		if (rc)
+			goto unlock;
+
+		state->act_priority = priority;
+		state->act_server = server;
+		state->eisn = eisn;
+
+		rc = xive_native_configure_irq(hw_num, xive->vp_base + server,
+					       priority, eisn);
+	} else {
+		state->act_priority = MASKED;
+		state->act_server = 0;
+		state->eisn = 0;
+
+		rc = xive_native_configure_irq(hw_num, 0, MASKED, 0);
+	}
+
+unlock:
+	arch_spin_unlock(&sb->lock);
+	return rc;
+}
+
+static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive,
+						long irq, u64 addr)
+{
+	struct kvmppc_xive_src_block *sb;
+	struct kvmppc_xive_irq_state *state;
+	u64 __user *ubufp = (u64 __user *) addr;
+	u16 src;
+	u64 kvm_cfg;
+	u32 server;
+	u8 priority;
+	u32 eisn;
+
+	sb = kvmppc_xive_find_source(xive, irq, &src);
+	if (!sb)
+		return -ENOENT;
+
+	state = &sb->irq_state[src];
+
+	if (!state->valid)
+		return -EINVAL;
+
+	if (get_user(kvm_cfg, ubufp))
+		return -EFAULT;
+
+	pr_devel("%s irq=0x%lx cfg=%016llx\n", __func__, irq, kvm_cfg);
+
+	priority = (kvm_cfg & KVM_XIVE_SOURCE_PRIORITY_MASK) >>
+		KVM_XIVE_SOURCE_PRIORITY_SHIFT;
+	server = (kvm_cfg & KVM_XIVE_SOURCE_SERVER_MASK) >>
+		KVM_XIVE_SOURCE_SERVER_SHIFT;
+	eisn = (kvm_cfg & KVM_XIVE_SOURCE_EISN_MASK) >>
+		KVM_XIVE_SOURCE_EISN_SHIFT;
+
+	if (priority != xive_prio_from_guest(priority)) {
+		pr_err("invalid priority for queue %d for VCPU %d\n",
+		       priority, server);
+		return -EINVAL;
+	}
+
+	return kvmppc_xive_native_update_source_config(xive, sb, state, server,
+						       priority, eisn);
+}
+
 static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 				       struct kvm_device_attr *attr)
 {
@@ -259,6 +352,9 @@ static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 	case KVM_DEV_XIVE_GRP_SOURCE:
 		return kvmppc_xive_native_set_source(xive, attr->attr,
 						     attr->addr);
+	case KVM_DEV_XIVE_GRP_SOURCE_CONFIG:
+		return kvmppc_xive_native_set_source_config(xive, attr->attr,
+							    attr->addr);
 	}
 	return -ENXIO;
 }
@@ -276,6 +372,7 @@ static int kvmppc_xive_native_has_attr(struct kvm_device *dev,
 	case KVM_DEV_XIVE_GRP_CTRL:
 		break;
 	case KVM_DEV_XIVE_GRP_SOURCE:
+	case KVM_DEV_XIVE_GRP_SOURCE_CONFIG:
 		if (attr->attr >= KVMPPC_XIVE_FIRST_IRQ &&
 		    attr->attr < KVMPPC_XIVE_NR_IRQS)
 			return 0;
Documentation/virtual/kvm/devices/xive.txt +20 -0
index cd8bfc37b72e..4f513a1880c7 100644
@@ -32,3 +32,23 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -ENOMEM: Could not create a new source block
     -EFAULT: Invalid user pointer for attr->addr.
     -ENXIO:  Could not allocate underlying HW interrupt
+
+  3. KVM_DEV_XIVE_GRP_SOURCE_CONFIG (write only)
+  Configures source targeting
+  Attributes:
+    Interrupt source number  (64-bit)
+  The kvm_device_attr.addr points to a __u64 value:
+  bits:     | 63   ....  33 |  32  | 31 .. 3 |  2 .. 0
+  values:   |    eisn       | mask |  server | priority
+  - priority: 0-7 interrupt priority level
+  - server: CPU number chosen to handle the interrupt
+  - mask: mask flag (unused)
+  - eisn: Effective Interrupt Source Number
+  Errors:
+    -ENOENT: Unknown source number
+    -EINVAL: Not initialized source number, invalid priority or
+             invalid CPU number.
+    -EFAULT: Invalid user pointer for attr->addr.
+    -ENXIO:  CPU event queues not configured or configuration of the
+             underlying HW interrupt failed
+    -EBUSY:  No CPU available to serve interrupt
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#28 lore raw

PATCH v2 6/16 KVM: PPC: Book3S HV: XIVE: add controls for the EQ configuration

In reply to Cédric Le Goater

These controls will be used by the H_INT_SET_QUEUE_CONFIG and H_INT_GET_QUEUE_CONFIG hcalls from QEMU. They will also be used to restore the configuration of the XIVE EQs in the KVM device and to capture the internal runtime state of the EQs. Both 'get' and 'set' rely on an OPAL call to access from the XIVE interrupt controller the EQ toggle bit and EQ index which are updated by the HW when event notifications are enqueued in the EQ.

The value of the guest physical address of the event queue is saved in the XIVE internal xive_q structure for later use. That is when migration needs to mark the EQ pages dirty to capture a consistent memory state of the VM.

To be noted that H_INT_SET_QUEUE_CONFIG does not require the extra OPAL call setting the EQ toggle bit and EQ index to configure the EQ, but restoring the EQ state will.

---
 arch/powerpc/include/asm/xive.h            |   2 +
 arch/powerpc/include/uapi/asm/kvm.h        |  21 +++
 arch/powerpc/kvm/book3s_xive.h             |   2 +
 arch/powerpc/kvm/book3s_xive.c             |  15 +-
 arch/powerpc/kvm/book3s_xive_native.c      | 207 +++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  29 +++
 6 files changed, 270 insertions(+), 6 deletions(-)
arch/powerpc/include/asm/xive.h +2 -0
index b579a943407b..46891f321606 100644
@@ -73,6 +73,8 @@ struct xive_q {
 	u32			esc_irq;
 	atomic_t		count;
 	atomic_t		pending_count;
+	u64			guest_qpage;
+	u32			guest_qsize;
 };
 
 /* Global enable flags for the XIVE support */
arch/powerpc/include/uapi/asm/kvm.h +21 -0
index 91899c7f9abd..177e43f3edaf 100644
@@ -679,6 +679,7 @@ struct kvm_ppc_cpu_char {
 #define KVM_DEV_XIVE_GRP_CTRL		1
 #define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_SOURCE_CONFIG	3	/* 64-bit source attributes */
+#define KVM_DEV_XIVE_GRP_EQ_CONFIG	4	/* 64-bit eq attributes */
 
 /* Layout of 64-bit XIVE source attribute values */
 #define KVM_XIVE_LEVEL_SENSITIVE	(1ULL << 0)
@@ -694,4 +695,24 @@ struct kvm_ppc_cpu_char {
 #define KVM_XIVE_SOURCE_EISN_SHIFT	33
 #define KVM_XIVE_SOURCE_EISN_MASK	0xfffffffe00000000ULL
 
+/* Layout of 64-bit eq attribute */
+#define KVM_XIVE_EQ_PRIORITY_SHIFT	0
+#define KVM_XIVE_EQ_PRIORITY_MASK	0x7
+#define KVM_XIVE_EQ_SERVER_SHIFT	3
+#define KVM_XIVE_EQ_SERVER_MASK		0xfffffff8ULL
+
+/* Layout of 64-bit eq attribute values */
+struct kvm_ppc_xive_eq {
+	__u32 flags;
+	__u32 qsize;
+	__u64 qpage;
+	__u32 qtoggle;
+	__u32 qindex;
+	__u8  pad[40];
+};
+
+#define KVM_XIVE_EQ_FLAG_ENABLED	0x00000001
+#define KVM_XIVE_EQ_FLAG_ALWAYS_NOTIFY	0x00000002
+#define KVM_XIVE_EQ_FLAG_ESCALATE	0x00000004
+
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive.h +2 -0
index ab3ac152980d..6660d138c6b7 100644
@@ -267,6 +267,8 @@ struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
 	struct kvmppc_xive *xive, int irq);
 void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb);
 int kvmppc_xive_select_target(struct kvm *kvm, u32 *server, u8 prio);
+int kvmppc_xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio,
+				  bool single_escalation);
 
 #endif /* CONFIG_KVM_XICS */
 #endif /* _KVM_PPC_BOOK3S_XICS_H */
arch/powerpc/kvm/book3s_xive.c +9 -6
index 086da91d7c6e..7431e31bc541 100644
@@ -166,7 +166,8 @@ static irqreturn_t xive_esc_irq(int irq, void *data)
 	return IRQ_HANDLED;
 }
 
-static int xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio)
+int kvmppc_xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio,
+				  bool single_escalation)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
 	struct xive_q *q = &xc->queues[prio];
@@ -185,7 +186,7 @@ static int xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio)
 		return -EIO;
 	}
 
-	if (xc->xive->single_escalation)
+	if (single_escalation)
 		name = kasprintf(GFP_KERNEL, "kvm-%d-%d",
 				 vcpu->kvm->arch.lpid, xc->server_num);
 	else
@@ -217,7 +218,7 @@ static int xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio)
 	 * interrupt, thus leaving it effectively masked after
 	 * it fires once.
 	 */
-	if (xc->xive->single_escalation) {
+	if (single_escalation) {
 		struct irq_data *d = irq_get_irq_data(xc->esc_virq[prio]);
 		struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);
 
@@ -291,7 +292,8 @@ static int xive_check_provisioning(struct kvm *kvm, u8 prio)
 			continue;
 		rc = xive_provision_queue(vcpu, prio);
 		if (rc == 0 && !xive->single_escalation)
-			xive_attach_escalation(vcpu, prio);
+			kvmppc_xive_attach_escalation(vcpu, prio,
+						      xive->single_escalation);
 		if (rc)
 			return rc;
 	}
@@ -1219,7 +1221,8 @@ int kvmppc_xive_connect_vcpu(struct kvm_device *dev,
 		if (xive->qmap & (1 << i)) {
 			r = xive_provision_queue(vcpu, i);
 			if (r == 0 && !xive->single_escalation)
-				xive_attach_escalation(vcpu, i);
+				kvmppc_xive_attach_escalation(
+					vcpu, i, xive->single_escalation);
 			if (r)
 				goto bail;
 		} else {
@@ -1234,7 +1237,7 @@ int kvmppc_xive_connect_vcpu(struct kvm_device *dev,
 	}
 
 	/* If not done above, attach priority 0 escalation */
-	r = xive_attach_escalation(vcpu, 0);
+	r = kvmppc_xive_attach_escalation(vcpu, 0, xive->single_escalation);
 	if (r)
 		goto bail;
 
arch/powerpc/kvm/book3s_xive_native.c +207 -0

231 lines not on this page · raw message

Documentation/virtual/kvm/devices/xive.txt +29 -0
index 4f513a1880c7..c0b5d9bd43fb 100644
@@ -52,3 +52,32 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -ENXIO:  CPU event queues not configured or configuration of the
              underlying HW interrupt failed
     -EBUSY:  No CPU available to serve interrupt
+
+  4. KVM_DEV_XIVE_GRP_EQ_CONFIG (read-write)
+  Configures an event queue of a CPU
+  Attributes:
+    EQ descriptor identifier (64-bit)
+  The EQ descriptor identifier is a tuple (server, priority) :
+  bits:     | 63   ....  32 | 31 .. 3 |  2 .. 0
+  values:   |    unused     |  server | priority
+  The kvm_device_attr.addr points to :
+    struct kvm_ppc_xive_eq {
+	__u32 flags;
+	__u32 qsize;
+	__u64 qpage;
+	__u32 qtoggle;
+	__u32 qindex;
+	__u8  pad[40];
+    };
+  - flags: queue flags
+  - qsize: queue size (power of 2)
+  - qpage: real address of queue
+  - qtoggle: current queue toggle bit
+  - qindex: current queue index
+  - pad: reserved for future use
+  Errors:
+    -ENOENT: Invalid CPU number
+    -EINVAL: Invalid priority or invalid queue size
+    -EFAULT: Invalid user pointer for attr->addr.
+    -ENOMEM: Invalid queue address
+    -EIO:    Configuration of the underlying HW failed
Signature · 1 line
2.20.1
#29 lore raw

Re: [PATCH v2 06/16] KVM: PPC: Book3S HV: XIVE: add controls for the EQ configuration

Quoted 202 lines · Cédric Le Goater

These controls will be used by the H_INT_SET_QUEUE_CONFIG and H_INT_GET_QUEUE_CONFIG hcalls from QEMU. They will also be used to restore the configuration of the XIVE EQs in the KVM device and to capture the internal runtime state of the EQs. Both 'get' and 'set' rely on an OPAL call to access from the XIVE interrupt controller the EQ toggle bit and EQ index which are updated by the HW when event notifications are enqueued in the EQ.

The value of the guest physical address of the event queue is saved in the XIVE internal xive_q structure for later use. That is when migration needs to mark the EQ pages dirty to capture a consistent memory state of the VM.

To be noted that H_INT_SET_QUEUE_CONFIG does not require the extra OPAL call setting the EQ toggle bit and EQ index to configure the EQ, but restoring the EQ state will.

---
 arch/powerpc/include/asm/xive.h            |   2 +
 arch/powerpc/include/uapi/asm/kvm.h        |  21 +++
 arch/powerpc/kvm/book3s_xive.h             |   2 +
 arch/powerpc/kvm/book3s_xive.c             |  15 +-
 arch/powerpc/kvm/book3s_xive_native.c      | 207 +++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  29 +++
 6 files changed, 270 insertions(+), 6 deletions(-)
arch/powerpc/include/asm/xive.h +2 -0
index b579a943407b..46891f321606 100644
@@ -73,6 +73,8 @@ struct xive_q {
 	u32			esc_irq;
 	atomic_t		count;
 	atomic_t		pending_count;
+	u64			guest_qpage;
+	u32			guest_qsize;
 };
 
 /* Global enable flags for the XIVE support */
arch/powerpc/include/uapi/asm/kvm.h +21 -0
index 91899c7f9abd..177e43f3edaf 100644
@@ -679,6 +679,7 @@ struct kvm_ppc_cpu_char {
 #define KVM_DEV_XIVE_GRP_CTRL		1
 #define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_SOURCE_CONFIG	3	/* 64-bit source attributes */
+#define KVM_DEV_XIVE_GRP_EQ_CONFIG	4	/* 64-bit eq attributes */
 
 /* Layout of 64-bit XIVE source attribute values */
 #define KVM_XIVE_LEVEL_SENSITIVE	(1ULL << 0)
@@ -694,4 +695,24 @@ struct kvm_ppc_cpu_char {
 #define KVM_XIVE_SOURCE_EISN_SHIFT	33
 #define KVM_XIVE_SOURCE_EISN_MASK	0xfffffffe00000000ULL
 
+/* Layout of 64-bit eq attribute */
+#define KVM_XIVE_EQ_PRIORITY_SHIFT	0
+#define KVM_XIVE_EQ_PRIORITY_MASK	0x7
+#define KVM_XIVE_EQ_SERVER_SHIFT	3
+#define KVM_XIVE_EQ_SERVER_MASK		0xfffffff8ULL
+
+/* Layout of 64-bit eq attribute values */
+struct kvm_ppc_xive_eq {
+	__u32 flags;
+	__u32 qsize;
+	__u64 qpage;
+	__u32 qtoggle;
+	__u32 qindex;
+	__u8  pad[40];
+};
+
+#define KVM_XIVE_EQ_FLAG_ENABLED	0x00000001
+#define KVM_XIVE_EQ_FLAG_ALWAYS_NOTIFY	0x00000002
+#define KVM_XIVE_EQ_FLAG_ESCALATE	0x00000004
+
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive.h +2 -0
index ab3ac152980d..6660d138c6b7 100644
@@ -267,6 +267,8 @@ struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
 	struct kvmppc_xive *xive, int irq);
 void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb);
 int kvmppc_xive_select_target(struct kvm *kvm, u32 *server, u8 prio);
+int kvmppc_xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio,
+				  bool single_escalation);
 
 #endif /* CONFIG_KVM_XICS */
 #endif /* _KVM_PPC_BOOK3S_XICS_H */
arch/powerpc/kvm/book3s_xive.c +9 -6
index 086da91d7c6e..7431e31bc541 100644
@@ -166,7 +166,8 @@ static irqreturn_t xive_esc_irq(int irq, void *data)
 	return IRQ_HANDLED;
 }
 
-static int xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio)
+int kvmppc_xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio,
+				  bool single_escalation)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
 	struct xive_q *q = &xc->queues[prio];
@@ -185,7 +186,7 @@ static int xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio)
 		return -EIO;
 	}
 
-	if (xc->xive->single_escalation)
+	if (single_escalation)
 		name = kasprintf(GFP_KERNEL, "kvm-%d-%d",
 				 vcpu->kvm->arch.lpid, xc->server_num);
 	else
@@ -217,7 +218,7 @@ static int xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio)
 	 * interrupt, thus leaving it effectively masked after
 	 * it fires once.
 	 */
-	if (xc->xive->single_escalation) {
+	if (single_escalation) {
 		struct irq_data *d = irq_get_irq_data(xc->esc_virq[prio]);
 		struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);
 
@@ -291,7 +292,8 @@ static int xive_check_provisioning(struct kvm *kvm, u8 prio)
 			continue;
 		rc = xive_provision_queue(vcpu, prio);
 		if (rc == 0 && !xive->single_escalation)
-			xive_attach_escalation(vcpu, prio);
+			kvmppc_xive_attach_escalation(vcpu, prio,
+						      xive->single_escalation);
 		if (rc)
 			return rc;
 	}
@@ -1219,7 +1221,8 @@ int kvmppc_xive_connect_vcpu(struct kvm_device *dev,
 		if (xive->qmap & (1 << i)) {
 			r = xive_provision_queue(vcpu, i);
 			if (r == 0 && !xive->single_escalation)
-				xive_attach_escalation(vcpu, i);
+				kvmppc_xive_attach_escalation(
+					vcpu, i, xive->single_escalation);
 			if (r)
 				goto bail;
 		} else {
@@ -1234,7 +1237,7 @@ int kvmppc_xive_connect_vcpu(struct kvm_device *dev,
 	}
 
 	/* If not done above, attach priority 0 escalation */
-	r = xive_attach_escalation(vcpu, 0);
+	r = kvmppc_xive_attach_escalation(vcpu, 0, xive->single_escalation);
 	if (r)
 		goto bail;
 
arch/powerpc/kvm/book3s_xive_native.c +43 -0
index cb5a5c6e05af..34a35bcf550c 100644
@@ -341,6 +341,201 @@ static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive,
 						       priority, eisn);
 }
 
+static int xive_native_validate_queue_size(u32 qsize)
+{
+	switch (qsize) {
+	case 12:
+	case 16:
+	case 21:
+	case 24:
+	case 0:
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
+
+static int kvmppc_xive_native_set_queue_config(struct kvmppc_xive *xive,
+					       long eq_idx, u64 addr)
+{
+	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
+	struct kvmppc_xive_vcpu *xc;
+	void __user *ubufp = (u64 __user *) addr;
+	u32 server;
+	u8 priority;
+	struct kvm_ppc_xive_eq kvm_eq;
+	int rc;
+	__be32 *qaddr = 0;
+	struct page *page;
+	struct xive_q *q;
+
+	/*
+	 * Demangle priority/server tuple from the EQ index
+	 */
+	priority = (eq_idx & KVM_XIVE_EQ_PRIORITY_MASK) >>
+		KVM_XIVE_EQ_PRIORITY_SHIFT;
+	server = (eq_idx & KVM_XIVE_EQ_SERVER_MASK) >>
+		KVM_XIVE_EQ_SERVER_SHIFT;
+
+	if (copy_from_user(&kvm_eq, ubufp, sizeof(kvm_eq)))
+		return -EFAULT;
+
+	vcpu = kvmppc_xive_find_server(kvm, server);
+	if (!vcpu) {
+		pr_err("Can't find server %d\n", server);

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive, · line 396

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+		return -ENOENT;
+	}
+	xc = vcpu->arch.xive_vcpu;
+
+	if (priority != xive_prio_from_guest(priority)) {
+		pr_err("Trying to restore invalid queue %d for VCPU %d\n",
+		       priority, server);
+		return -EINVAL;
+	}
+	q = &xc->queues[priority];

You need to validate the 'flags' field (AFAICT we don't actually have any flags yet, so it's only valid it if is 0.

Quoted 27 lines
arch/powerpc/kvm/book3s_xive_native.c +27 -0
+
+	pr_devel("%s VCPU %d priority %d fl:%x sz:%d addr:%llx g:%d idx:%d\n",
+		 __func__, server, priority, kvm_eq.flags,
+		 kvm_eq.qsize, kvm_eq.qpage, kvm_eq.qtoggle, kvm_eq.qindex);
+
+	rc = xive_native_validate_queue_size(kvm_eq.qsize);
+	if (rc) {
+		pr_err("invalid queue size %d\n", kvm_eq.qsize);
+		return rc;
+	}
+
+	/* reset queue and disable queueing */
+	if (!kvm_eq.qsize) {
+		q->guest_qpage = 0;
+		q->guest_qsize = 0;
+
+		rc = xive_native_configure_queue(xc->vp_id, q, priority,
+						 NULL, 0, true);
+		if (rc) {
+			pr_err("Failed to reset queue %d for VCPU %d: %d\n",
+			       priority, xc->server_num, rc);
+			return rc;
+		}
+
+		if (q->qpage) {
+			put_page(virt_to_page(q->qpage));
+			q->qpage = NULL;

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive, · line 433

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+		}
+
+		return 0;
+	}
+
+
+	page = gfn_to_page(kvm, gpa_to_gfn(kvm_eq.qpage));
+	if (is_error_page(page)) {
+		pr_warn("Couldn't get guest page for %llx!\n", kvm_eq.qpage);
+		return -ENOMEM;

Nit: I don't think ENOMEM is the right error here. ENOMEM indicates that the kernel couldn't allocate enough memory to complete whatever you asked. Here the problem is the user supplied a bad guest address, which is a rather different error. EFAULT is closer, but still not quite right, since it could be a valid user address but not a valid guest address. There are probably existing KVM calls that could hit this problem, I wonder what they use.

Quoted 178 lines
arch/powerpc/kvm/book3s_xive_native.c +117 -0
+	}
+	qaddr = page_to_virt(page) + (kvm_eq.qpage & ~PAGE_MASK);
+
+	/* Backup queue page guest address for migration */
+	q->guest_qpage = kvm_eq.qpage;
+	q->guest_qsize = kvm_eq.qsize;
+
+	rc = xive_native_configure_queue(xc->vp_id, q, priority,
+					 (__be32 *) qaddr, kvm_eq.qsize, true);
+	if (rc) {
+		pr_err("Failed to configure queue %d for VCPU %d: %d\n",
+		       priority, xc->server_num, rc);
+		put_page(page);
+		return rc;
+	}
+
+	rc = xive_native_set_queue_state(xc->vp_id, priority, kvm_eq.qtoggle,
+					 kvm_eq.qindex);
+	if (rc)
+		goto error;
+
+	rc = kvmppc_xive_attach_escalation(vcpu, priority,
+					   xive->single_escalation);
+error:
+	if (rc)
+		kvmppc_xive_native_cleanup_queue(vcpu, priority);
+	return rc;
+}
+
+static int kvmppc_xive_native_get_queue_config(struct kvmppc_xive *xive,
+					       long eq_idx, u64 addr)
+{
+	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
+	struct kvmppc_xive_vcpu *xc;
+	struct xive_q *q;
+	void __user *ubufp = (u64 __user *) addr;
+	u32 server;
+	u8 priority;
+	struct kvm_ppc_xive_eq kvm_eq;
+	u64 qpage;
+	u64 qsize;
+	u64 qeoi_page;
+	u32 escalate_irq;
+	u64 qflags;
+	int rc;
+
+	/*
+	 * Demangle priority/server tuple from the EQ index
+	 */
+	priority = (eq_idx & KVM_XIVE_EQ_PRIORITY_MASK) >>
+		KVM_XIVE_EQ_PRIORITY_SHIFT;
+	server = (eq_idx & KVM_XIVE_EQ_SERVER_MASK) >>
+		KVM_XIVE_EQ_SERVER_SHIFT;
+
+	vcpu = kvmppc_xive_find_server(kvm, server);
+	if (!vcpu) {
+		pr_err("Can't find server %d\n", server);
+		return -ENOENT;
+	}
+	xc = vcpu->arch.xive_vcpu;
+
+	if (priority != xive_prio_from_guest(priority)) {
+		pr_err("invalid priority for queue %d for VCPU %d\n",
+		       priority, server);
+		return -EINVAL;
+	}
+	q = &xc->queues[priority];
+
+	memset(&kvm_eq, 0, sizeof(kvm_eq));
+
+	if (!q->qpage)
+		return 0;
+
+	rc = xive_native_get_queue_info(xc->vp_id, priority, &qpage, &qsize,
+					&qeoi_page, &escalate_irq, &qflags);
+	if (rc)
+		return rc;
+
+	kvm_eq.flags = 0;
+	if (qflags & OPAL_XIVE_EQ_ENABLED)
+		kvm_eq.flags |= KVM_XIVE_EQ_FLAG_ENABLED;
+	if (qflags & OPAL_XIVE_EQ_ALWAYS_NOTIFY)
+		kvm_eq.flags |= KVM_XIVE_EQ_FLAG_ALWAYS_NOTIFY;
+	if (qflags & OPAL_XIVE_EQ_ESCALATE)
+		kvm_eq.flags |= KVM_XIVE_EQ_FLAG_ESCALATE;
+
+	kvm_eq.qsize = q->guest_qsize;
+	kvm_eq.qpage = q->guest_qpage;
+
+	rc = xive_native_get_queue_state(xc->vp_id, priority, &kvm_eq.qtoggle,
+					 &kvm_eq.qindex);
+	if (rc)
+		return rc;
+
+	pr_devel("%s VCPU %d priority %d fl:%x sz:%d addr:%llx g:%d idx:%d\n",
+		 __func__, server, priority, kvm_eq.flags,
+		 kvm_eq.qsize, kvm_eq.qpage, kvm_eq.qtoggle, kvm_eq.qindex);
+
+	if (copy_to_user(ubufp, &kvm_eq, sizeof(kvm_eq)))
+		return -EFAULT;
+
+	return 0;
+}
+
 static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 				       struct kvm_device_attr *attr)
 {
@@ -355,6 +550,9 @@ static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 	case KVM_DEV_XIVE_GRP_SOURCE_CONFIG:
 		return kvmppc_xive_native_set_source_config(xive, attr->attr,
 							    attr->addr);
+	case KVM_DEV_XIVE_GRP_EQ_CONFIG:
+		return kvmppc_xive_native_set_queue_config(xive, attr->attr,
+							   attr->addr);
 	}
 	return -ENXIO;
 }
@@ -362,6 +560,13 @@ static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 static int kvmppc_xive_native_get_attr(struct kvm_device *dev,
 				       struct kvm_device_attr *attr)
 {
+	struct kvmppc_xive *xive = dev->private;
+
+	switch (attr->group) {
+	case KVM_DEV_XIVE_GRP_EQ_CONFIG:
+		return kvmppc_xive_native_get_queue_config(xive, attr->attr,
+							   attr->addr);
+	}
 	return -ENXIO;
 }
 
@@ -377,6 +582,8 @@ static int kvmppc_xive_native_has_attr(struct kvm_device *dev,
 		    attr->attr < KVMPPC_XIVE_NR_IRQS)
 			return 0;
 		break;
+	case KVM_DEV_XIVE_GRP_EQ_CONFIG:
+		return 0;
 	}
 	return -ENXIO;
 }
Documentation/virtual/kvm/devices/xive.txt +29 -0
index 4f513a1880c7..c0b5d9bd43fb 100644
@@ -52,3 +52,32 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -ENXIO:  CPU event queues not configured or configuration of the
              underlying HW interrupt failed
     -EBUSY:  No CPU available to serve interrupt
+
+  4. KVM_DEV_XIVE_GRP_EQ_CONFIG (read-write)
+  Configures an event queue of a CPU
+  Attributes:
+    EQ descriptor identifier (64-bit)
+  The EQ descriptor identifier is a tuple (server, priority) :
+  bits:     | 63   ....  32 | 31 .. 3 |  2 .. 0
+  values:   |    unused     |  server | priority
+  The kvm_device_attr.addr points to :
+    struct kvm_ppc_xive_eq {
+	__u32 flags;
+	__u32 qsize;
+	__u64 qpage;
+	__u32 qtoggle;
+	__u32 qindex;
+	__u8  pad[40];
+    };
+  - flags: queue flags
+  - qsize: queue size (power of 2)
+  - qpage: real address of queue
+  - qtoggle: current queue toggle bit
+  - qindex: current queue index
+  - pad: reserved for future use
+  Errors:
+    -ENOENT: Invalid CPU number
+    -EINVAL: Invalid priority or invalid queue size
+    -EFAULT: Invalid user pointer for attr->addr.
+    -ENOMEM: Invalid queue address
+    -EIO:    Configuration of the underlying HW failed
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#30 lore raw

Re: [PATCH v2 06/16] KVM: PPC: Book3S HV: XIVE: add controls for the EQ configuration

Quoted 9 lines · David Gibson
Quoted 8 lines · Cédric Le Goater

These controls will be used by the H_INT_SET_QUEUE_CONFIG and H_INT_GET_QUEUE_CONFIG hcalls from QEMU. They will also be used to restore the configuration of the XIVE EQs in the KVM device and to capture the internal runtime state of the EQs. Both 'get' and 'set' rely on an OPAL call to access from the XIVE interrupt controller the EQ toggle bit and EQ index which are updated by the HW when event notifications are enqueued in the EQ.

Quoted 8 lines
Quoted 8 lines

The value of the guest physical address of the event queue is saved in the XIVE internal xive_q structure for later use. That is when migration needs to mark the EQ pages dirty to capture a consistent memory state of the VM.

To be noted that H_INT_SET_QUEUE_CONFIG does not require the extra OPAL call setting the EQ toggle bit and EQ index to configure the EQ, but restoring the EQ state will.

I think we need to add some kind of flags to differentiate the hcall H_INT_SET_QUEUE_CONFIG from the restore of the EQ. The hcall does not need OPAL support call and this could help in the code transition.

But without OPAL support, we won't have migration. Would that be of any use ?

Quoted 191 lines
Quoted 191 lines
---
 arch/powerpc/include/asm/xive.h            |   2 +
 arch/powerpc/include/uapi/asm/kvm.h        |  21 +++
 arch/powerpc/kvm/book3s_xive.h             |   2 +
 arch/powerpc/kvm/book3s_xive.c             |  15 +-
 arch/powerpc/kvm/book3s_xive_native.c      | 207 +++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  29 +++
 6 files changed, 270 insertions(+), 6 deletions(-)
arch/powerpc/include/asm/xive.h +2 -0
index b579a943407b..46891f321606 100644
@@ -73,6 +73,8 @@ struct xive_q {
 	u32			esc_irq;
 	atomic_t		count;
 	atomic_t		pending_count;
+	u64			guest_qpage;
+	u32			guest_qsize;
 };
 
 /* Global enable flags for the XIVE support */
arch/powerpc/include/uapi/asm/kvm.h +21 -0
index 91899c7f9abd..177e43f3edaf 100644
@@ -679,6 +679,7 @@ struct kvm_ppc_cpu_char {
 #define KVM_DEV_XIVE_GRP_CTRL		1
 #define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_SOURCE_CONFIG	3	/* 64-bit source attributes */
+#define KVM_DEV_XIVE_GRP_EQ_CONFIG	4	/* 64-bit eq attributes */
 
 /* Layout of 64-bit XIVE source attribute values */
 #define KVM_XIVE_LEVEL_SENSITIVE	(1ULL << 0)
@@ -694,4 +695,24 @@ struct kvm_ppc_cpu_char {
 #define KVM_XIVE_SOURCE_EISN_SHIFT	33
 #define KVM_XIVE_SOURCE_EISN_MASK	0xfffffffe00000000ULL
 
+/* Layout of 64-bit eq attribute */
+#define KVM_XIVE_EQ_PRIORITY_SHIFT	0
+#define KVM_XIVE_EQ_PRIORITY_MASK	0x7
+#define KVM_XIVE_EQ_SERVER_SHIFT	3
+#define KVM_XIVE_EQ_SERVER_MASK		0xfffffff8ULL
+
+/* Layout of 64-bit eq attribute values */
+struct kvm_ppc_xive_eq {
+	__u32 flags;
+	__u32 qsize;
+	__u64 qpage;
+	__u32 qtoggle;
+	__u32 qindex;
+	__u8  pad[40];
+};
+
+#define KVM_XIVE_EQ_FLAG_ENABLED	0x00000001
+#define KVM_XIVE_EQ_FLAG_ALWAYS_NOTIFY	0x00000002
+#define KVM_XIVE_EQ_FLAG_ESCALATE	0x00000004
+
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive.h +2 -0
index ab3ac152980d..6660d138c6b7 100644
@@ -267,6 +267,8 @@ struct kvmppc_xive_src_block *kvmppc_xive_create_src_block(
 	struct kvmppc_xive *xive, int irq);
 void kvmppc_xive_free_sources(struct kvmppc_xive_src_block *sb);
 int kvmppc_xive_select_target(struct kvm *kvm, u32 *server, u8 prio);
+int kvmppc_xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio,
+				  bool single_escalation);
 
 #endif /* CONFIG_KVM_XICS */
 #endif /* _KVM_PPC_BOOK3S_XICS_H */
arch/powerpc/kvm/book3s_xive.c +9 -6
index 086da91d7c6e..7431e31bc541 100644
@@ -166,7 +166,8 @@ static irqreturn_t xive_esc_irq(int irq, void *data)
 	return IRQ_HANDLED;
 }
 
-static int xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio)
+int kvmppc_xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio,
+				  bool single_escalation)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
 	struct xive_q *q = &xc->queues[prio];
@@ -185,7 +186,7 @@ static int xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio)
 		return -EIO;
 	}
 
-	if (xc->xive->single_escalation)
+	if (single_escalation)
 		name = kasprintf(GFP_KERNEL, "kvm-%d-%d",
 				 vcpu->kvm->arch.lpid, xc->server_num);
 	else
@@ -217,7 +218,7 @@ static int xive_attach_escalation(struct kvm_vcpu *vcpu, u8 prio)
 	 * interrupt, thus leaving it effectively masked after
 	 * it fires once.
 	 */
-	if (xc->xive->single_escalation) {
+	if (single_escalation) {
 		struct irq_data *d = irq_get_irq_data(xc->esc_virq[prio]);
 		struct xive_irq_data *xd = irq_data_get_irq_handler_data(d);
 
@@ -291,7 +292,8 @@ static int xive_check_provisioning(struct kvm *kvm, u8 prio)
 			continue;
 		rc = xive_provision_queue(vcpu, prio);
 		if (rc == 0 && !xive->single_escalation)
-			xive_attach_escalation(vcpu, prio);
+			kvmppc_xive_attach_escalation(vcpu, prio,
+						      xive->single_escalation);
 		if (rc)
 			return rc;
 	}
@@ -1219,7 +1221,8 @@ int kvmppc_xive_connect_vcpu(struct kvm_device *dev,
 		if (xive->qmap & (1 << i)) {
 			r = xive_provision_queue(vcpu, i);
 			if (r == 0 && !xive->single_escalation)
-				xive_attach_escalation(vcpu, i);
+				kvmppc_xive_attach_escalation(
+					vcpu, i, xive->single_escalation);
 			if (r)
 				goto bail;
 		} else {
@@ -1234,7 +1237,7 @@ int kvmppc_xive_connect_vcpu(struct kvm_device *dev,
 	}
 
 	/* If not done above, attach priority 0 escalation */
-	r = xive_attach_escalation(vcpu, 0);
+	r = kvmppc_xive_attach_escalation(vcpu, 0, xive->single_escalation);
 	if (r)
 		goto bail;
 
arch/powerpc/kvm/book3s_xive_native.c +48 -0
index cb5a5c6e05af..34a35bcf550c 100644
@@ -341,6 +341,201 @@ static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive,
 						       priority, eisn);
 }
 
+static int xive_native_validate_queue_size(u32 qsize)
+{
+	switch (qsize) {
+	case 12:
+	case 16:
+	case 21:
+	case 24:
+	case 0:
+		return 0;
+	default:
+		return -EINVAL;
+	}
+}
+
+static int kvmppc_xive_native_set_queue_config(struct kvmppc_xive *xive,
+					       long eq_idx, u64 addr)
+{
+	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
+	struct kvmppc_xive_vcpu *xc;
+	void __user *ubufp = (u64 __user *) addr;
+	u32 server;
+	u8 priority;
+	struct kvm_ppc_xive_eq kvm_eq;
+	int rc;
+	__be32 *qaddr = 0;
+	struct page *page;
+	struct xive_q *q;
+
+	/*
+	 * Demangle priority/server tuple from the EQ index
+	 */
+	priority = (eq_idx & KVM_XIVE_EQ_PRIORITY_MASK) >>
+		KVM_XIVE_EQ_PRIORITY_SHIFT;
+	server = (eq_idx & KVM_XIVE_EQ_SERVER_MASK) >>
+		KVM_XIVE_EQ_SERVER_SHIFT;
+
+	if (copy_from_user(&kvm_eq, ubufp, sizeof(kvm_eq)))
+		return -EFAULT;
+
+	vcpu = kvmppc_xive_find_server(kvm, server);
+	if (!vcpu) {
+		pr_err("Can't find server %d\n", server);
+		return -ENOENT;
+	}
+	xc = vcpu->arch.xive_vcpu;
+
+	if (priority != xive_prio_from_guest(priority)) {
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive, · line 396

Quoted 5 lines
arch/powerpc/kvm/book3s_xive_native.c +5 -0
+		pr_err("Trying to restore invalid queue %d for VCPU %d\n",
+		       priority, server);
+		return -EINVAL;
+	}
+	q = &xc->queues[priority];

You need to validate the 'flags' field (AFAICT we don't actually have any flags yet, so it's only valid it if is 0.

Well, we should set the ESCALATE flags and ALWAYS_NOTIFY also but this is already set by xive_native_configure_queue(). I will take a closer look and document.

Quoted 38 lines
Quoted 27 lines
arch/powerpc/kvm/book3s_xive_native.c +27 -0
+
+	pr_devel("%s VCPU %d priority %d fl:%x sz:%d addr:%llx g:%d idx:%d\n",
+		 __func__, server, priority, kvm_eq.flags,
+		 kvm_eq.qsize, kvm_eq.qpage, kvm_eq.qtoggle, kvm_eq.qindex);
+
+	rc = xive_native_validate_queue_size(kvm_eq.qsize);
+	if (rc) {
+		pr_err("invalid queue size %d\n", kvm_eq.qsize);
+		return rc;
+	}
+
+	/* reset queue and disable queueing */
+	if (!kvm_eq.qsize) {
+		q->guest_qpage = 0;
+		q->guest_qsize = 0;
+
+		rc = xive_native_configure_queue(xc->vp_id, q, priority,
+						 NULL, 0, true);
+		if (rc) {
+			pr_err("Failed to reset queue %d for VCPU %d: %d\n",
+			       priority, xc->server_num, rc);
+			return rc;
+		}
+
+		if (q->qpage) {
+			put_page(virt_to_page(q->qpage));
+			q->qpage = NULL;

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive, · line 433

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+		}
+
+		return 0;
+	}
+
+
+	page = gfn_to_page(kvm, gpa_to_gfn(kvm_eq.qpage));
+	if (is_error_page(page)) {
+		pr_warn("Couldn't get guest page for %llx!\n", kvm_eq.qpage);
+		return -ENOMEM;
Quoted 7 lines

Nit: I don't think ENOMEM is the right error here. ENOMEM indicates that the kernel couldn't allocate enough memory to complete whatever you asked. Here the problem is the user supplied a bad guest address, which is a rather different error. EFAULT is closer, but still not quite right, since it could be a valid user address but not a valid guest address. There are probably existing KVM calls that could hit this problem, I wonder what they use.

I have seen a -EINVAL (s390) and a -EFAULT (x86)

Thanks,

C.

Quoted 179 lines
Quoted 178 lines
arch/powerpc/kvm/book3s_xive_native.c +117 -0
+	}
+	qaddr = page_to_virt(page) + (kvm_eq.qpage & ~PAGE_MASK);
+
+	/* Backup queue page guest address for migration */
+	q->guest_qpage = kvm_eq.qpage;
+	q->guest_qsize = kvm_eq.qsize;
+
+	rc = xive_native_configure_queue(xc->vp_id, q, priority,
+					 (__be32 *) qaddr, kvm_eq.qsize, true);
+	if (rc) {
+		pr_err("Failed to configure queue %d for VCPU %d: %d\n",
+		       priority, xc->server_num, rc);
+		put_page(page);
+		return rc;
+	}
+
+	rc = xive_native_set_queue_state(xc->vp_id, priority, kvm_eq.qtoggle,
+					 kvm_eq.qindex);
+	if (rc)
+		goto error;
+
+	rc = kvmppc_xive_attach_escalation(vcpu, priority,
+					   xive->single_escalation);
+error:
+	if (rc)
+		kvmppc_xive_native_cleanup_queue(vcpu, priority);
+	return rc;
+}
+
+static int kvmppc_xive_native_get_queue_config(struct kvmppc_xive *xive,
+					       long eq_idx, u64 addr)
+{
+	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
+	struct kvmppc_xive_vcpu *xc;
+	struct xive_q *q;
+	void __user *ubufp = (u64 __user *) addr;
+	u32 server;
+	u8 priority;
+	struct kvm_ppc_xive_eq kvm_eq;
+	u64 qpage;
+	u64 qsize;
+	u64 qeoi_page;
+	u32 escalate_irq;
+	u64 qflags;
+	int rc;
+
+	/*
+	 * Demangle priority/server tuple from the EQ index
+	 */
+	priority = (eq_idx & KVM_XIVE_EQ_PRIORITY_MASK) >>
+		KVM_XIVE_EQ_PRIORITY_SHIFT;
+	server = (eq_idx & KVM_XIVE_EQ_SERVER_MASK) >>
+		KVM_XIVE_EQ_SERVER_SHIFT;
+
+	vcpu = kvmppc_xive_find_server(kvm, server);
+	if (!vcpu) {
+		pr_err("Can't find server %d\n", server);
+		return -ENOENT;
+	}
+	xc = vcpu->arch.xive_vcpu;
+
+	if (priority != xive_prio_from_guest(priority)) {
+		pr_err("invalid priority for queue %d for VCPU %d\n",
+		       priority, server);
+		return -EINVAL;
+	}
+	q = &xc->queues[priority];
+
+	memset(&kvm_eq, 0, sizeof(kvm_eq));
+
+	if (!q->qpage)
+		return 0;
+
+	rc = xive_native_get_queue_info(xc->vp_id, priority, &qpage, &qsize,
+					&qeoi_page, &escalate_irq, &qflags);
+	if (rc)
+		return rc;
+
+	kvm_eq.flags = 0;
+	if (qflags & OPAL_XIVE_EQ_ENABLED)
+		kvm_eq.flags |= KVM_XIVE_EQ_FLAG_ENABLED;
+	if (qflags & OPAL_XIVE_EQ_ALWAYS_NOTIFY)
+		kvm_eq.flags |= KVM_XIVE_EQ_FLAG_ALWAYS_NOTIFY;
+	if (qflags & OPAL_XIVE_EQ_ESCALATE)
+		kvm_eq.flags |= KVM_XIVE_EQ_FLAG_ESCALATE;
+
+	kvm_eq.qsize = q->guest_qsize;
+	kvm_eq.qpage = q->guest_qpage;
+
+	rc = xive_native_get_queue_state(xc->vp_id, priority, &kvm_eq.qtoggle,
+					 &kvm_eq.qindex);
+	if (rc)
+		return rc;
+
+	pr_devel("%s VCPU %d priority %d fl:%x sz:%d addr:%llx g:%d idx:%d\n",
+		 __func__, server, priority, kvm_eq.flags,
+		 kvm_eq.qsize, kvm_eq.qpage, kvm_eq.qtoggle, kvm_eq.qindex);
+
+	if (copy_to_user(ubufp, &kvm_eq, sizeof(kvm_eq)))
+		return -EFAULT;
+
+	return 0;
+}
+
 static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 				       struct kvm_device_attr *attr)
 {
@@ -355,6 +550,9 @@ static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 	case KVM_DEV_XIVE_GRP_SOURCE_CONFIG:
 		return kvmppc_xive_native_set_source_config(xive, attr->attr,
 							    attr->addr);
+	case KVM_DEV_XIVE_GRP_EQ_CONFIG:
+		return kvmppc_xive_native_set_queue_config(xive, attr->attr,
+							   attr->addr);
 	}
 	return -ENXIO;
 }
@@ -362,6 +560,13 @@ static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 static int kvmppc_xive_native_get_attr(struct kvm_device *dev,
 				       struct kvm_device_attr *attr)
 {
+	struct kvmppc_xive *xive = dev->private;
+
+	switch (attr->group) {
+	case KVM_DEV_XIVE_GRP_EQ_CONFIG:
+		return kvmppc_xive_native_get_queue_config(xive, attr->attr,
+							   attr->addr);
+	}
 	return -ENXIO;
 }
 
@@ -377,6 +582,8 @@ static int kvmppc_xive_native_has_attr(struct kvm_device *dev,
 		    attr->attr < KVMPPC_XIVE_NR_IRQS)
 			return 0;
 		break;
+	case KVM_DEV_XIVE_GRP_EQ_CONFIG:
+		return 0;
 	}
 	return -ENXIO;
 }
Documentation/virtual/kvm/devices/xive.txt +29 -0
index 4f513a1880c7..c0b5d9bd43fb 100644
@@ -52,3 +52,32 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -ENXIO:  CPU event queues not configured or configuration of the
              underlying HW interrupt failed
     -EBUSY:  No CPU available to serve interrupt
+
+  4. KVM_DEV_XIVE_GRP_EQ_CONFIG (read-write)
+  Configures an event queue of a CPU
+  Attributes:
+    EQ descriptor identifier (64-bit)
+  The EQ descriptor identifier is a tuple (server, priority) :
+  bits:     | 63   ....  32 | 31 .. 3 |  2 .. 0
+  values:   |    unused     |  server | priority
+  The kvm_device_attr.addr points to :
+    struct kvm_ppc_xive_eq {
+	__u32 flags;
+	__u32 qsize;
+	__u64 qpage;
+	__u32 qtoggle;
+	__u32 qindex;
+	__u8  pad[40];
+    };
+  - flags: queue flags
+  - qsize: queue size (power of 2)
+  - qpage: real address of queue
+  - qtoggle: current queue toggle bit
+  - qindex: current queue index
+  - pad: reserved for future use
+  Errors:
+    -ENOENT: Invalid CPU number
+    -EINVAL: Invalid priority or invalid queue size
+    -EFAULT: Invalid user pointer for attr->addr.
+    -ENOMEM: Invalid queue address
+    -EIO:    Configuration of the underlying HW failed
#31 lore raw

Re: [PATCH v2 06/16] KVM: PPC: Book3S HV: XIVE: add controls for the EQ configuration

Quoted 15 lines · Cédric Le Goater
Quoted 14 lines · David Gibson
Quoted 13 lines · Cédric Le Goater

These controls will be used by the H_INT_SET_QUEUE_CONFIG and H_INT_GET_QUEUE_CONFIG hcalls from QEMU. They will also be used to restore the configuration of the XIVE EQs in the KVM device and to capture the internal runtime state of the EQs. Both 'get' and 'set' rely on an OPAL call to access from the XIVE interrupt controller the EQ toggle bit and EQ index which are updated by the HW when event notifications are enqueued in the EQ.

The value of the guest physical address of the event queue is saved in the XIVE internal xive_q structure for later use. That is when migration needs to mark the EQ pages dirty to capture a consistent memory state of the VM.

Quoted 8 lines
Quoted 3 lines
Quoted 3 lines

To be noted that H_INT_SET_QUEUE_CONFIG does not require the extra OPAL call setting the EQ toggle bit and EQ index to configure the EQ, but restoring the EQ state will.

I think we need to add some kind of flags to differentiate the hcall H_INT_SET_QUEUE_CONFIG from the restore of the EQ. The hcall does not need OPAL support call and this could help in the code transition.

Hrm. What's the actual difference in the semantics between the two cases. The guest shouldn't have awareness of whether or not OPAL is involved.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#32 lore raw

Re: [PATCH v2 06/16] KVM: PPC: Book3S HV: XIVE: add controls for the EQ configuration

Quoted 20 lines · David Gibson
Quoted 19 lines · Cédric Le Goater
Quoted 17 lines · David Gibson
Quoted 16 lines · Cédric Le Goater

These controls will be used by the H_INT_SET_QUEUE_CONFIG and H_INT_GET_QUEUE_CONFIG hcalls from QEMU. They will also be used to restore the configuration of the XIVE EQs in the KVM device and to capture the internal runtime state of the EQs. Both 'get' and 'set' rely on an OPAL call to access from the XIVE interrupt controller the EQ toggle bit and EQ index which are updated by the HW when event notifications are enqueued in the EQ.

The value of the guest physical address of the event queue is saved in the XIVE internal xive_q structure for later use. That is when migration needs to mark the EQ pages dirty to capture a consistent memory state of the VM.

To be noted that H_INT_SET_QUEUE_CONFIG does not require the extra OPAL call setting the EQ toggle bit and EQ index to configure the EQ, but restoring the EQ state will.

Quoted 7 lines
Quoted 4 lines

I think we need to add some kind of flags to differentiate the hcall H_INT_SET_QUEUE_CONFIG from the restore of the EQ. The hcall does not need OPAL support call and this could help in the code transition.

Hrm. What's the actual difference in the semantics between the two cases.

None.

But we don't need to set the EQ state in the case of the HCALL and it's (very) practical to run guests with XIVE enabled without the OPAL support. The latter is the main reason clearly.

Thinking of it, I could test the EQ toggle bit and index passed to KVM and skip the OPAL call which restores the EQ state if they are zero. This is because I know that the OPAL call configuring the EQ resets them.

That will do. No need for a flag.

Quoted 1 line

The guest shouldn't have awareness of whether or not OPAL is involved.

yes.

Thanks,

C.

#33 lore raw

Re: [PATCH v2 06/16] KVM: PPC: Book3S HV: XIVE: add controls for the EQ configuration

Quoted 35 lines · Cédric Le Goater
Quoted 20 lines · David Gibson
Quoted 19 lines · Cédric Le Goater
Quoted 17 lines · David Gibson
Quoted 16 lines · Cédric Le Goater

These controls will be used by the H_INT_SET_QUEUE_CONFIG and H_INT_GET_QUEUE_CONFIG hcalls from QEMU. They will also be used to restore the configuration of the XIVE EQs in the KVM device and to capture the internal runtime state of the EQs. Both 'get' and 'set' rely on an OPAL call to access from the XIVE interrupt controller the EQ toggle bit and EQ index which are updated by the HW when event notifications are enqueued in the EQ.

The value of the guest physical address of the event queue is saved in the XIVE internal xive_q structure for later use. That is when migration needs to mark the EQ pages dirty to capture a consistent memory state of the VM.

To be noted that H_INT_SET_QUEUE_CONFIG does not require the extra OPAL call setting the EQ toggle bit and EQ index to configure the EQ, but restoring the EQ state will.

Quoted 7 lines
Quoted 4 lines

I think we need to add some kind of flags to differentiate the hcall H_INT_SET_QUEUE_CONFIG from the restore of the EQ. The hcall does not need OPAL support call and this could help in the code transition.

Hrm. What's the actual difference in the semantics between the two cases.

None.

But we don't need to set the EQ state in the case of the HCALL and it's (very) practical to run guests with XIVE enabled without the OPAL support. The latter is the main reason clearly.

Quoted 5 lines

Thinking of it, I could test the EQ toggle bit and index passed to KVM and skip the OPAL call which restores the EQ state if they are zero. This is because I know that the OPAL call configuring the EQ resets them.

That will do. No need for a flag.

That's a much better idea.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#34 lore raw

Re: [PATCH v2 06/16] KVM: PPC: Book3S HV: XIVE: add controls for the EQ configuration

In reply to Cédric Le Goater
Quoted 16 lines · Cédric Le Goater

These controls will be used by the H_INT_SET_QUEUE_CONFIG and H_INT_GET_QUEUE_CONFIG hcalls from QEMU. They will also be used to restore the configuration of the XIVE EQs in the KVM device and to capture the internal runtime state of the EQs. Both 'get' and 'set' rely on an OPAL call to access from the XIVE interrupt controller the EQ toggle bit and EQ index which are updated by the HW when event notifications are enqueued in the EQ.

The value of the guest physical address of the event queue is saved in the XIVE internal xive_q structure for later use. That is when migration needs to mark the EQ pages dirty to capture a consistent memory state of the VM.

To be noted that H_INT_SET_QUEUE_CONFIG does not require the extra OPAL call setting the EQ toggle bit and EQ index to configure the EQ, but restoring the EQ state will.

[snip]

Quoted 5 lines
arch/powerpc/include/uapi/asm/kvm.h +5 -0
+/* Layout of 64-bit eq attribute */
+#define KVM_XIVE_EQ_PRIORITY_SHIFT	0
+#define KVM_XIVE_EQ_PRIORITY_MASK	0x7
+#define KVM_XIVE_EQ_SERVER_SHIFT	3
+#define KVM_XIVE_EQ_SERVER_MASK		0xfffffff8ULL

arch/powerpc/include/uapi/asm/kvm.h struct kvm_ppc_cpu_char { · line 712

Quoted 10 lines
arch/powerpc/include/uapi/asm/kvm.h +10 -0
+
+/* Layout of 64-bit eq attribute values */
+struct kvm_ppc_xive_eq {
+	__u32 flags;
+	__u32 qsize;
+	__u64 qpage;
+	__u32 qtoggle;
+	__u32 qindex;
+	__u8  pad[40];
+};

This is confusing. What's the difference between an "eq attribute" and an "eq attribute value"? Is the first actually a queue index or a queue identifier?

Also, the kvm_ppc_xive_eq is not 64 bits, so the comment above it is wrong. Maybe you meant "64-byte"?

[snip]

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive, · line 435

Quoted 6 lines
arch/powerpc/kvm/book3s_xive_native.c +6 -0
+	page = gfn_to_page(kvm, gpa_to_gfn(kvm_eq.qpage));
+	if (is_error_page(page)) {
+		pr_warn("Couldn't get guest page for %llx!\n", kvm_eq.qpage);
+		return -ENOMEM;
+	}
+	qaddr = page_to_virt(page) + (kvm_eq.qpage & ~PAGE_MASK);

Isn't this assuming that we can map the whole queue with a single gfn_to_page? That would only be true if kvm_eq.qsize <= PAGE_SHIFT. What happens if kvm_eq.qsize > PAGE_SHIFT?

Paul.

#35 lore raw

Re: [PATCH v2 06/16] KVM: PPC: Book3S HV: XIVE: add controls for the EQ configuration

Quoted 31 lines · Paul Mackerras
Quoted 16 lines · Cédric Le Goater

These controls will be used by the H_INT_SET_QUEUE_CONFIG and H_INT_GET_QUEUE_CONFIG hcalls from QEMU. They will also be used to restore the configuration of the XIVE EQs in the KVM device and to capture the internal runtime state of the EQs. Both 'get' and 'set' rely on an OPAL call to access from the XIVE interrupt controller the EQ toggle bit and EQ index which are updated by the HW when event notifications are enqueued in the EQ.

The value of the guest physical address of the event queue is saved in the XIVE internal xive_q structure for later use. That is when migration needs to mark the EQ pages dirty to capture a consistent memory state of the VM.

To be noted that H_INT_SET_QUEUE_CONFIG does not require the extra OPAL call setting the EQ toggle bit and EQ index to configure the EQ, but restoring the EQ state will.

[snip]

Quoted 11 lines
arch/powerpc/include/uapi/asm/kvm.h +11 -0
+/* Layout of 64-bit eq attribute */
+#define KVM_XIVE_EQ_PRIORITY_SHIFT	0
+#define KVM_XIVE_EQ_PRIORITY_MASK	0x7
+#define KVM_XIVE_EQ_SERVER_SHIFT	3
+#define KVM_XIVE_EQ_SERVER_MASK		0xfffffff8ULL
+
+/* Layout of 64-bit eq attribute values */
+struct kvm_ppc_xive_eq {
+	__u32 flags;
+	__u32 qsize;
+	__u64 qpage;
Quoted 8 lines

arch/powerpc/include/uapi/asm/kvm.h struct kvm_ppc_cpu_char { · line 712

Quoted 4 lines
arch/powerpc/include/uapi/asm/kvm.h +4 -0
+	__u32 qtoggle;
+	__u32 qindex;
+	__u8  pad[40];
+};

This is confusing. What's the difference between an "eq attribute" and an "eq attribute value"? Is the first actually a queue index or a queue identifier?

The "attribute" qualifier comes from the {get,set,has}_addr methods of the KVM device. But it is not a well chosen name for the group KVM_DEV_XIVE_GRP_EQ_CONFIG.

I should be using "eq identifier" and "eq values" or "eq state".

Quoted 2 lines

Also, the kvm_ppc_xive_eq is not 64 bits, so the comment above it is wrong. Maybe you meant "64-byte"?

That was a bad copy paste. I have padded the structure to twice the size of the XIVE END (the XIVE EQ descriptor in HW) which size is 32 bytes. I thought that one extra u64 was not enough room for future.

Quoted 13 lines

[snip]

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive, · line 435

Quoted 6 lines
arch/powerpc/kvm/book3s_xive_native.c +6 -0
+	page = gfn_to_page(kvm, gpa_to_gfn(kvm_eq.qpage));
+	if (is_error_page(page)) {
+		pr_warn("Couldn't get guest page for %llx!\n", kvm_eq.qpage);
+		return -ENOMEM;
+	}
+	qaddr = page_to_virt(page) + (kvm_eq.qpage & ~PAGE_MASK);

Isn't this assuming that we can map the whole queue with a single gfn_to_page? That would only be true if kvm_eq.qsize <= PAGE_SHIFT. What happens if kvm_eq.qsize > PAGE_SHIFT?

Ah yes. Theoretically, it should not happen because we only advertise 64K in the DT for the moment. I should at least add a check. So I will change the helper xive_native_validate_queue_size() to return -EINVAL for other page sizes.

Do you think it would be complex to support XIVE EQs using a page larger than the default one on the guest ?

Thanks,

C.

Quoted 3 lines

Paul.

#36 lore raw

Re: [PATCH v2 06/16] KVM: PPC: Book3S HV: XIVE: add controls for the EQ configuration

Quoted 41 lines · Cédric Le Goater
Quoted 31 lines · Paul Mackerras
Quoted 16 lines · Cédric Le Goater

These controls will be used by the H_INT_SET_QUEUE_CONFIG and H_INT_GET_QUEUE_CONFIG hcalls from QEMU. They will also be used to restore the configuration of the XIVE EQs in the KVM device and to capture the internal runtime state of the EQs. Both 'get' and 'set' rely on an OPAL call to access from the XIVE interrupt controller the EQ toggle bit and EQ index which are updated by the HW when event notifications are enqueued in the EQ.

The value of the guest physical address of the event queue is saved in the XIVE internal xive_q structure for later use. That is when migration needs to mark the EQ pages dirty to capture a consistent memory state of the VM.

To be noted that H_INT_SET_QUEUE_CONFIG does not require the extra OPAL call setting the EQ toggle bit and EQ index to configure the EQ, but restoring the EQ state will.

[snip]

Quoted 11 lines
arch/powerpc/include/uapi/asm/kvm.h +11 -0
+/* Layout of 64-bit eq attribute */
+#define KVM_XIVE_EQ_PRIORITY_SHIFT	0
+#define KVM_XIVE_EQ_PRIORITY_MASK	0x7
+#define KVM_XIVE_EQ_SERVER_SHIFT	3
+#define KVM_XIVE_EQ_SERVER_MASK		0xfffffff8ULL
+
+/* Layout of 64-bit eq attribute values */
+struct kvm_ppc_xive_eq {
+	__u32 flags;
+	__u32 qsize;
+	__u64 qpage;
Quoted 8 lines

arch/powerpc/include/uapi/asm/kvm.h struct kvm_ppc_cpu_char { · line 712

Quoted 4 lines
arch/powerpc/include/uapi/asm/kvm.h +4 -0
+	__u32 qtoggle;
+	__u32 qindex;
+	__u8  pad[40];
+};

This is confusing. What's the difference between an "eq attribute" and an "eq attribute value"? Is the first actually a queue index or a queue identifier?

Quoted 5 lines

The "attribute" qualifier comes from the {get,set,has}_addr methods of the KVM device. But it is not a well chosen name for the group KVM_DEV_XIVE_GRP_EQ_CONFIG.

I should be using "eq identifier" and "eq values" or "eq state".

Yeah, that seems clearer.

Quoted 21 lines
Quoted 2 lines

Also, the kvm_ppc_xive_eq is not 64 bits, so the comment above it is wrong. Maybe you meant "64-byte"?

That was a bad copy paste. I have padded the structure to twice the size of the XIVE END (the XIVE EQ descriptor in HW) which size is 32 bytes. I thought that one extra u64 was not enough room for future.

Quoted 13 lines

[snip]

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive, · line 435

Quoted 6 lines
arch/powerpc/kvm/book3s_xive_native.c +6 -0
+	page = gfn_to_page(kvm, gpa_to_gfn(kvm_eq.qpage));
+	if (is_error_page(page)) {
+		pr_warn("Couldn't get guest page for %llx!\n", kvm_eq.qpage);
+		return -ENOMEM;
+	}
+	qaddr = page_to_virt(page) + (kvm_eq.qpage & ~PAGE_MASK);

Isn't this assuming that we can map the whole queue with a single gfn_to_page? That would only be true if kvm_eq.qsize <= PAGE_SHIFT. What happens if kvm_eq.qsize > PAGE_SHIFT?

Quoted 4 lines

Ah yes. Theoretically, it should not happen because we only advertise 64K in the DT for the moment. I should at least add a check. So I will change the helper xive_native_validate_queue_size() to return -EINVAL for other page sizes.

Ok.

Quoted 2 lines

Do you think it would be complex to support XIVE EQs using a page larger than the default one on the guest ?

Hm. The queue has to be physically contiguous from the host point of view, in order for the XIVE hardware to write to it, doesn't it? If so then supporting queues bigger than the guest page size would be very difficult.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#37 lore raw

Re: [PATCH v2 06/16] KVM: PPC: Book3S HV: XIVE: add controls for the EQ configuration

Quoted 78 lines · David Gibson
Quoted 41 lines · Cédric Le Goater
Quoted 31 lines · Paul Mackerras
Quoted 16 lines · Cédric Le Goater

These controls will be used by the H_INT_SET_QUEUE_CONFIG and H_INT_GET_QUEUE_CONFIG hcalls from QEMU. They will also be used to restore the configuration of the XIVE EQs in the KVM device and to capture the internal runtime state of the EQs. Both 'get' and 'set' rely on an OPAL call to access from the XIVE interrupt controller the EQ toggle bit and EQ index which are updated by the HW when event notifications are enqueued in the EQ.

The value of the guest physical address of the event queue is saved in the XIVE internal xive_q structure for later use. That is when migration needs to mark the EQ pages dirty to capture a consistent memory state of the VM.

To be noted that H_INT_SET_QUEUE_CONFIG does not require the extra OPAL call setting the EQ toggle bit and EQ index to configure the EQ, but restoring the EQ state will.

[snip]

Quoted 11 lines
arch/powerpc/include/uapi/asm/kvm.h +11 -0
+/* Layout of 64-bit eq attribute */
+#define KVM_XIVE_EQ_PRIORITY_SHIFT	0
+#define KVM_XIVE_EQ_PRIORITY_MASK	0x7
+#define KVM_XIVE_EQ_SERVER_SHIFT	3
+#define KVM_XIVE_EQ_SERVER_MASK		0xfffffff8ULL
+
+/* Layout of 64-bit eq attribute values */
+struct kvm_ppc_xive_eq {
+	__u32 flags;
+	__u32 qsize;
+	__u64 qpage;
Quoted 8 lines

arch/powerpc/include/uapi/asm/kvm.h struct kvm_ppc_cpu_char { · line 712

Quoted 4 lines
arch/powerpc/include/uapi/asm/kvm.h +4 -0
+	__u32 qtoggle;
+	__u32 qindex;
+	__u8  pad[40];
+};

This is confusing. What's the difference between an "eq attribute" and an "eq attribute value"? Is the first actually a queue index or a queue identifier?

Quoted 5 lines

The "attribute" qualifier comes from the {get,set,has}_addr methods of the KVM device. But it is not a well chosen name for the group KVM_DEV_XIVE_GRP_EQ_CONFIG.

I should be using "eq identifier" and "eq values" or "eq state".

Yeah, that seems clearer.

Quoted 21 lines
Quoted 2 lines

Also, the kvm_ppc_xive_eq is not 64 bits, so the comment above it is wrong. Maybe you meant "64-byte"?

That was a bad copy paste. I have padded the structure to twice the size of the XIVE END (the XIVE EQ descriptor in HW) which size is 32 bytes. I thought that one extra u64 was not enough room for future.

Quoted 13 lines

[snip]

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive, · line 435

Quoted 6 lines
arch/powerpc/kvm/book3s_xive_native.c +6 -0
+	page = gfn_to_page(kvm, gpa_to_gfn(kvm_eq.qpage));
+	if (is_error_page(page)) {
+		pr_warn("Couldn't get guest page for %llx!\n", kvm_eq.qpage);
+		return -ENOMEM;
+	}
+	qaddr = page_to_virt(page) + (kvm_eq.qpage & ~PAGE_MASK);

Isn't this assuming that we can map the whole queue with a single gfn_to_page? That would only be true if kvm_eq.qsize <= PAGE_SHIFT. What happens if kvm_eq.qsize > PAGE_SHIFT?

Quoted 4 lines

Ah yes. Theoretically, it should not happen because we only advertise 64K in the DT for the moment. I should at least add a check. So I will change the helper xive_native_validate_queue_size() to return -EINVAL for other page sizes.

Ok.

Quoted 7 lines
Quoted 2 lines

Do you think it would be complex to support XIVE EQs using a page larger than the default one on the guest ?

Hm. The queue has to be physically contiguous from the host point of view, in order for the XIVE hardware to write to it, doesn't it? If so then supporting queues bigger than the guest page size would be very difficult.

The queue is only *one* page.

C.

#38 lore raw

Re: [PATCH v2 06/16] KVM: PPC: Book3S HV: XIVE: add controls for the EQ configuration

Quoted 82 lines · Cédric Le Goater
Quoted 81 lines · David Gibson
Quoted 41 lines · Cédric Le Goater
Quoted 31 lines · Paul Mackerras
Quoted 16 lines · Cédric Le Goater

These controls will be used by the H_INT_SET_QUEUE_CONFIG and H_INT_GET_QUEUE_CONFIG hcalls from QEMU. They will also be used to restore the configuration of the XIVE EQs in the KVM device and to capture the internal runtime state of the EQs. Both 'get' and 'set' rely on an OPAL call to access from the XIVE interrupt controller the EQ toggle bit and EQ index which are updated by the HW when event notifications are enqueued in the EQ.

The value of the guest physical address of the event queue is saved in the XIVE internal xive_q structure for later use. That is when migration needs to mark the EQ pages dirty to capture a consistent memory state of the VM.

To be noted that H_INT_SET_QUEUE_CONFIG does not require the extra OPAL call setting the EQ toggle bit and EQ index to configure the EQ, but restoring the EQ state will.

[snip]

Quoted 11 lines
arch/powerpc/include/uapi/asm/kvm.h +11 -0
+/* Layout of 64-bit eq attribute */
+#define KVM_XIVE_EQ_PRIORITY_SHIFT	0
+#define KVM_XIVE_EQ_PRIORITY_MASK	0x7
+#define KVM_XIVE_EQ_SERVER_SHIFT	3
+#define KVM_XIVE_EQ_SERVER_MASK		0xfffffff8ULL
+
+/* Layout of 64-bit eq attribute values */
+struct kvm_ppc_xive_eq {
+	__u32 flags;
+	__u32 qsize;
+	__u64 qpage;
Quoted 8 lines

arch/powerpc/include/uapi/asm/kvm.h struct kvm_ppc_cpu_char { · line 712

Quoted 4 lines
arch/powerpc/include/uapi/asm/kvm.h +4 -0
+	__u32 qtoggle;
+	__u32 qindex;
+	__u8  pad[40];
+};

This is confusing. What's the difference between an "eq attribute" and an "eq attribute value"? Is the first actually a queue index or a queue identifier?

Quoted 5 lines

The "attribute" qualifier comes from the {get,set,has}_addr methods of the KVM device. But it is not a well chosen name for the group KVM_DEV_XIVE_GRP_EQ_CONFIG.

I should be using "eq identifier" and "eq values" or "eq state".

Yeah, that seems clearer.

Quoted 21 lines
Quoted 2 lines

Also, the kvm_ppc_xive_eq is not 64 bits, so the comment above it is wrong. Maybe you meant "64-byte"?

That was a bad copy paste. I have padded the structure to twice the size of the XIVE END (the XIVE EQ descriptor in HW) which size is 32 bytes. I thought that one extra u64 was not enough room for future.

Quoted 13 lines

[snip]

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive, · line 435

Quoted 6 lines
arch/powerpc/kvm/book3s_xive_native.c +6 -0
+	page = gfn_to_page(kvm, gpa_to_gfn(kvm_eq.qpage));
+	if (is_error_page(page)) {
+		pr_warn("Couldn't get guest page for %llx!\n", kvm_eq.qpage);
+		return -ENOMEM;
+	}
+	qaddr = page_to_virt(page) + (kvm_eq.qpage & ~PAGE_MASK);

Isn't this assuming that we can map the whole queue with a single gfn_to_page? That would only be true if kvm_eq.qsize <= PAGE_SHIFT. What happens if kvm_eq.qsize > PAGE_SHIFT?

Quoted 4 lines

Ah yes. Theoretically, it should not happen because we only advertise 64K in the DT for the moment. I should at least add a check. So I will change the helper xive_native_validate_queue_size() to return -EINVAL for other page sizes.

Ok.

Quoted 2 lines

Do you think it would be complex to support XIVE EQs using a page larger than the default one on the guest ?

Quoted 6 lines
Quoted 4 lines

Hm. The queue has to be physically contiguous from the host point of view, in order for the XIVE hardware to write to it, doesn't it? If so then supporting queues bigger than the guest page size would be very difficult.

The queue is only *one* page.

Right, but it's one *host* page, right, which is by nature host physically contiguous. If the guest page size is different a single guest page might not be host physically contiguous.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#39 lore raw

PATCH v2 7/16 KVM: PPC: Book3S HV: XIVE: add a global reset control

In reply to Cédric Le Goater

This control is to be used by the H_INT_RESET hcall from QEMU. Its purpose is to clear all configuration of the sources and EQs. This is necessary in case of a kexec (for a kdump kernel for instance) to make sure that no remaining configuration is left from the previous boot setup so that the new kernel can start safely from a clean state.

The queue 7 is ignored when the KVM device is configured to run in single escalation mode. Prio 7 is used by escalations.

The XIVE VP is kept enabled as the vCPU is still active and connected to the XIVE device.

---
 arch/powerpc/include/uapi/asm/kvm.h        |  1 +
 arch/powerpc/kvm/book3s_xive_native.c      | 82 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  5 ++
 3 files changed, 88 insertions(+)
arch/powerpc/include/uapi/asm/kvm.h +1 -0
index 177e43f3edaf..7ae8cb22af7d 100644
@@ -677,6 +677,7 @@ struct kvm_ppc_cpu_char {
 
 /* POWER9 XIVE Native Interrupt Controller */
 #define KVM_DEV_XIVE_GRP_CTRL		1
+#define   KVM_DEV_XIVE_RESET		1
 #define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_SOURCE_CONFIG	3	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_EQ_CONFIG	4	/* 64-bit eq attributes */
arch/powerpc/kvm/book3s_xive_native.c +82 -0

100 lines not on this page · raw message

Documentation/virtual/kvm/devices/xive.txt +5 -0
index c0b5d9bd43fb..f1d007f485a9 100644
@@ -17,6 +17,11 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
 
   1. KVM_DEV_XIVE_GRP_CTRL
   Provides global controls on the device
+  Attributes:
+    1.1 KVM_DEV_XIVE_RESET (write only)
+    Resets the interrupt controller configuration for sources and event
+    queues. To be used by kexec and kdump.
+    Errors: none
 
   2. KVM_DEV_XIVE_GRP_SOURCE (write only)
   Initializes a new source in the XIVE device and mask it.
Signature · 1 line
2.20.1
#40 lore raw

Re: [PATCH v2 07/16] KVM: PPC: Book3S HV: XIVE: add a global reset control

Quoted 40 lines · Cédric Le Goater

This control is to be used by the H_INT_RESET hcall from QEMU. Its purpose is to clear all configuration of the sources and EQs. This is necessary in case of a kexec (for a kdump kernel for instance) to make sure that no remaining configuration is left from the previous boot setup so that the new kernel can start safely from a clean state.

The queue 7 is ignored when the KVM device is configured to run in single escalation mode. Prio 7 is used by escalations.

The XIVE VP is kept enabled as the vCPU is still active and connected to the XIVE device.

---
 arch/powerpc/include/uapi/asm/kvm.h        |  1 +
 arch/powerpc/kvm/book3s_xive_native.c      | 82 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  5 ++
 3 files changed, 88 insertions(+)
arch/powerpc/include/uapi/asm/kvm.h +1 -0
index 177e43f3edaf..7ae8cb22af7d 100644
@@ -677,6 +677,7 @@ struct kvm_ppc_cpu_char {
 
 /* POWER9 XIVE Native Interrupt Controller */
 #define KVM_DEV_XIVE_GRP_CTRL		1
+#define   KVM_DEV_XIVE_RESET		1
 #define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_SOURCE_CONFIG	3	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_EQ_CONFIG	4	/* 64-bit eq attributes */
arch/powerpc/kvm/book3s_xive_native.c +1 -0
index 34a35bcf550c..bb3e121c918a 100644
@@ -536,6 +536,80 @@ static int kvmppc_xive_native_get_queue_config(struct kvmppc_xive *xive,
 	return 0;
 }
 
+static void kvmppc_xive_reset_sources(struct kvmppc_xive_src_block *sb)

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_get_queue_config(struct kvmppc_xive *xive, · line 549

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+{
+	int i;
+
+	for (i = 0; i < KVMPPC_XICS_IRQ_PER_ICS; i++) {
+		struct kvmppc_xive_irq_state *state = &sb->irq_state[i];
+
+		if (!state->valid)
+			continue;
+
+		if (state->act_priority == MASKED)

You're checking some things in state before you take the lock. Could this race?

Quoted 104 lines
arch/powerpc/kvm/book3s_xive_native.c +71 -0
+			continue;
+
+		arch_spin_lock(&sb->lock);
+		state->eisn = 0;
+		state->act_server = 0;
+		state->act_priority = MASKED;
+		xive_vm_esb_load(&state->ipi_data, XIVE_ESB_SET_PQ_01);
+		xive_native_configure_irq(state->ipi_number, 0, MASKED, 0);
+		if (state->pt_number) {
+			xive_vm_esb_load(state->pt_data, XIVE_ESB_SET_PQ_01);
+			xive_native_configure_irq(state->pt_number,
+						  0, MASKED, 0);
+		}
+		arch_spin_unlock(&sb->lock);
+	}
+}
+
+static int kvmppc_xive_reset(struct kvmppc_xive *xive)
+{
+	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
+	unsigned int i;
+
+	pr_devel("%s\n", __func__);
+
+	mutex_lock(&kvm->lock);
+
+	kvm_for_each_vcpu(i, vcpu, kvm) {
+		struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+		unsigned int prio;
+
+		if (!xc)
+			continue;
+
+		kvmppc_xive_disable_vcpu_interrupts(vcpu);
+
+		for (prio = 0; prio < KVMPPC_XIVE_Q_COUNT; prio++) {
+
+			/* Single escalation, no queue 7 */
+			if (prio == 7 && xive->single_escalation)
+				break;
+
+			if (xc->esc_virq[prio]) {
+				free_irq(xc->esc_virq[prio], vcpu);
+				irq_dispose_mapping(xc->esc_virq[prio]);
+				kfree(xc->esc_virq_names[prio]);
+				xc->esc_virq[prio] = 0;
+			}
+
+			kvmppc_xive_native_cleanup_queue(vcpu, prio);
+		}
+	}
+
+	for (i = 0; i <= xive->max_sbid; i++) {
+		if (xive->src_blocks[i])
+			kvmppc_xive_reset_sources(xive->src_blocks[i]);
+	}
+
+	mutex_unlock(&kvm->lock);
+
+	return 0;
+}
+
 static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 				       struct kvm_device_attr *attr)
 {
@@ -543,6 +617,10 @@ static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 
 	switch (attr->group) {
 	case KVM_DEV_XIVE_GRP_CTRL:
+		switch (attr->attr) {
+		case KVM_DEV_XIVE_RESET:
+			return kvmppc_xive_reset(xive);
+		}
 		break;
 	case KVM_DEV_XIVE_GRP_SOURCE:
 		return kvmppc_xive_native_set_source(xive, attr->attr,
@@ -575,6 +653,10 @@ static int kvmppc_xive_native_has_attr(struct kvm_device *dev,
 {
 	switch (attr->group) {
 	case KVM_DEV_XIVE_GRP_CTRL:
+		switch (attr->attr) {
+		case KVM_DEV_XIVE_RESET:
+			return 0;
+		}
 		break;
 	case KVM_DEV_XIVE_GRP_SOURCE:
 	case KVM_DEV_XIVE_GRP_SOURCE_CONFIG:
Documentation/virtual/kvm/devices/xive.txt +5 -0
index c0b5d9bd43fb..f1d007f485a9 100644
@@ -17,6 +17,11 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
 
   1. KVM_DEV_XIVE_GRP_CTRL
   Provides global controls on the device
+  Attributes:
+    1.1 KVM_DEV_XIVE_RESET (write only)
+    Resets the interrupt controller configuration for sources and event
+    queues. To be used by kexec and kdump.
+    Errors: none
 
   2. KVM_DEV_XIVE_GRP_SOURCE (write only)
   Initializes a new source in the XIVE device and mask it.
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#41 lore raw

PATCH v2 8/16 KVM: PPC: Book3S HV: XIVE: add a control to sync the sources

In reply to Cédric Le Goater

This control will be used by the H_INT_SYNC hcall from QEMU.

---
 arch/powerpc/include/uapi/asm/kvm.h        |  1 +
 arch/powerpc/kvm/book3s_xive_native.c      | 34 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  9 ++++++
 3 files changed, 44 insertions(+)
arch/powerpc/include/uapi/asm/kvm.h +1 -0
index 7ae8cb22af7d..289c504b7c1d 100644
@@ -681,6 +681,7 @@ struct kvm_ppc_cpu_char {
 #define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_SOURCE_CONFIG	3	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_EQ_CONFIG	4	/* 64-bit eq attributes */
+#define KVM_DEV_XIVE_GRP_SOURCE_SYNC	5       /* 64-bit source attributes */
 
 /* Layout of 64-bit XIVE source attribute values */
 #define KVM_XIVE_LEVEL_SENSITIVE	(1ULL << 0)
arch/powerpc/kvm/book3s_xive_native.c +34 -0
index bb3e121c918a..dd2a9d411fe7 100644
@@ -341,6 +341,36 @@ static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive,
 						       priority, eisn);
 }
 
+static int kvmppc_xive_native_sync_source(struct kvmppc_xive *xive,
+					  long irq, u64 addr)
+{
+	struct kvmppc_xive_src_block *sb;
+	struct kvmppc_xive_irq_state *state;
+	struct xive_irq_data *xd;
+	u32 hw_num;
+	u16 src;
+
+	pr_devel("%s irq=0x%lx", __func__, irq);
+
+	sb = kvmppc_xive_find_source(xive, irq, &src);
+	if (!sb)
+		return -ENOENT;
+
+	state = &sb->irq_state[src];
+
+	if (!state->valid)
+		return -EINVAL;
+
+	arch_spin_lock(&sb->lock);
+
+	kvmppc_xive_select_irq(state, &hw_num, &xd);
+	xive_native_sync_source(hw_num);
+
+	arch_spin_unlock(&sb->lock);
+
+	return 0;
+}
+
 static int xive_native_validate_queue_size(u32 qsize)
 {
 	switch (qsize) {
@@ -631,6 +661,9 @@ static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 	case KVM_DEV_XIVE_GRP_EQ_CONFIG:
 		return kvmppc_xive_native_set_queue_config(xive, attr->attr,
 							   attr->addr);
+	case KVM_DEV_XIVE_GRP_SOURCE_SYNC:
+		return kvmppc_xive_native_sync_source(xive, attr->attr,
+						      attr->addr);
 	}
 	return -ENXIO;
 }
@@ -660,6 +693,7 @@ static int kvmppc_xive_native_has_attr(struct kvm_device *dev,
 		break;
 	case KVM_DEV_XIVE_GRP_SOURCE:
 	case KVM_DEV_XIVE_GRP_SOURCE_CONFIG:
+	case KVM_DEV_XIVE_GRP_SOURCE_SYNC:
 		if (attr->attr >= KVMPPC_XIVE_FIRST_IRQ &&
 		    attr->attr < KVMPPC_XIVE_NR_IRQS)
 			return 0;
Documentation/virtual/kvm/devices/xive.txt +9 -0
index f1d007f485a9..267634eae9e0 100644
@@ -86,3 +86,12 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -EFAULT: Invalid user pointer for attr->addr.
     -ENOMEM: Invalid queue address
     -EIO:    Configuration of the underlying HW failed
+
+  5. KVM_DEV_XIVE_GRP_SOURCE_SYNC (write only)
+  Synchronize the source to flush event notification
+  Attributes:
+    Interrupt source number  (64-bit)
+  Errors:
+    -ENOENT: Unknown source number
+    -EINVAL: Not initialized source number, invalid priority or
+             invalid CPU number.
Signature · 1 line
2.20.1
#42 lore raw

Re: [PATCH v2 08/16] KVM: PPC: Book3S HV: XIVE: add a control to sync the sources

Quoted 37 lines · Cédric Le Goater

This control will be used by the H_INT_SYNC hcall from QEMU.

---
 arch/powerpc/include/uapi/asm/kvm.h        |  1 +
 arch/powerpc/kvm/book3s_xive_native.c      | 34 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  9 ++++++
 3 files changed, 44 insertions(+)
arch/powerpc/include/uapi/asm/kvm.h +1 -0
index 7ae8cb22af7d..289c504b7c1d 100644
@@ -681,6 +681,7 @@ struct kvm_ppc_cpu_char {
 #define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_SOURCE_CONFIG	3	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_EQ_CONFIG	4	/* 64-bit eq attributes */
+#define KVM_DEV_XIVE_GRP_SOURCE_SYNC	5       /* 64-bit source attributes */
 
 /* Layout of 64-bit XIVE source attribute values */
 #define KVM_XIVE_LEVEL_SENSITIVE	(1ULL << 0)
arch/powerpc/kvm/book3s_xive_native.c +8 -0
index bb3e121c918a..dd2a9d411fe7 100644
@@ -341,6 +341,36 @@ static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive,
 						       priority, eisn);
 }
 
+static int kvmppc_xive_native_sync_source(struct kvmppc_xive *xive,
+					  long irq, u64 addr)
+{
+	struct kvmppc_xive_src_block *sb;
+	struct kvmppc_xive_irq_state *state;
+	struct xive_irq_data *xd;
+	u32 hw_num;
+	u16 src;

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_set_source_config(struct kvmppc_xive *xive, · line 361

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+
+	pr_devel("%s irq=0x%lx", __func__, irq);
+
+	sb = kvmppc_xive_find_source(xive, irq, &src);
+	if (!sb)
+		return -ENOENT;
+
+	state = &sb->irq_state[src];
+
+	if (!state->valid)

Is testing this before you take the lock safe?

Quoted 50 lines
arch/powerpc/kvm/book3s_xive_native.c +16 -0
+		return -EINVAL;
+
+	arch_spin_lock(&sb->lock);
+
+	kvmppc_xive_select_irq(state, &hw_num, &xd);
+	xive_native_sync_source(hw_num);
+
+	arch_spin_unlock(&sb->lock);
+
+	return 0;
+}
+
 static int xive_native_validate_queue_size(u32 qsize)
 {
 	switch (qsize) {
@@ -631,6 +661,9 @@ static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 	case KVM_DEV_XIVE_GRP_EQ_CONFIG:
 		return kvmppc_xive_native_set_queue_config(xive, attr->attr,
 							   attr->addr);
+	case KVM_DEV_XIVE_GRP_SOURCE_SYNC:
+		return kvmppc_xive_native_sync_source(xive, attr->attr,
+						      attr->addr);
 	}
 	return -ENXIO;
 }
@@ -660,6 +693,7 @@ static int kvmppc_xive_native_has_attr(struct kvm_device *dev,
 		break;
 	case KVM_DEV_XIVE_GRP_SOURCE:
 	case KVM_DEV_XIVE_GRP_SOURCE_CONFIG:
+	case KVM_DEV_XIVE_GRP_SOURCE_SYNC:
 		if (attr->attr >= KVMPPC_XIVE_FIRST_IRQ &&
 		    attr->attr < KVMPPC_XIVE_NR_IRQS)
 			return 0;
Documentation/virtual/kvm/devices/xive.txt +9 -0
index f1d007f485a9..267634eae9e0 100644
@@ -86,3 +86,12 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -EFAULT: Invalid user pointer for attr->addr.
     -ENOMEM: Invalid queue address
     -EIO:    Configuration of the underlying HW failed
+
+  5. KVM_DEV_XIVE_GRP_SOURCE_SYNC (write only)
+  Synchronize the source to flush event notification
+  Attributes:
+    Interrupt source number  (64-bit)
+  Errors:
+    -ENOENT: Unknown source number
+    -EINVAL: Not initialized source number, invalid priority or
+             invalid CPU number.
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#43 lore raw

PATCH v2 9/16 KVM: PPC: Book3S HV: XIVE: add a control to dirty the XIVE EQ pages

In reply to Cédric Le Goater

When migration of a VM is initiated, a first copy of the RAM is transferred to the destination before the VM is stopped, but there is no guarantee that the EQ pages in which the event notification are queued have not been modified.

To make sure migration will capture a consistent memory state, the XIVE device should perform a XIVE quiesce sequence to stop the flow of event notifications and stabilize the EQs. This is the purpose of the KVM_DEV_XIVE_EQ_SYNC control which will also marks the EQ pages dirty to force their transfer.

---
 arch/powerpc/include/uapi/asm/kvm.h        |  1 +
 arch/powerpc/kvm/book3s_xive_native.c      | 67 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt | 29 ++++++++++
 3 files changed, 97 insertions(+)
arch/powerpc/include/uapi/asm/kvm.h +1 -0
index 289c504b7c1d..cd78ad1020fe 100644
@@ -678,6 +678,7 @@ struct kvm_ppc_cpu_char {
 /* POWER9 XIVE Native Interrupt Controller */
 #define KVM_DEV_XIVE_GRP_CTRL		1
 #define   KVM_DEV_XIVE_RESET		1
+#define   KVM_DEV_XIVE_EQ_SYNC		2
 #define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_SOURCE_CONFIG	3	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_EQ_CONFIG	4	/* 64-bit eq attributes */
arch/powerpc/kvm/book3s_xive_native.c +67 -0

85 lines not on this page · raw message

Documentation/virtual/kvm/devices/xive.txt +29 -0
index 267634eae9e0..a26be635cff9 100644
@@ -23,6 +23,12 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     queues. To be used by kexec and kdump.
     Errors: none
 
+    1.2 KVM_DEV_XIVE_EQ_SYNC (write only)
+    Sync all the sources and queues and mark the EQ pages dirty. This
+    to make sure that a consistent memory state is captured when
+    migrating the VM.
+    Errors: none
+
   2. KVM_DEV_XIVE_GRP_SOURCE (write only)
   Initializes a new source in the XIVE device and mask it.
   Attributes:
@@ -95,3 +101,26 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -ENOENT: Unknown source number
     -EINVAL: Not initialized source number, invalid priority or
              invalid CPU number.
+
+* Migration:
+
+  Saving the state of a VM using the XIVE native exploitation mode
+  should follow a specific sequence. When the VM is stopped :
+
+  1. Mask all sources (PQ=01) to stop the flow of events.
+
+  2. Sync the XIVE device with the KVM control KVM_DEV_XIVE_EQ_SYNC to
+  flush any in-flight event notification and to stabilize the EQs. At
+  this stage, the EQ pages are marked dirty to make sure they are
+  transferred in the migration sequence.
+
+  3. Capture the state of the source targeting, the EQs configuration
+  and the state of thread interrupt context registers.
+
+  Restore is similar :
+
+  1. Restore the EQ configuration. As targeting depends on it.
+  2. Restore targeting
+  3. Restore the thread interrupt contexts
+  4. Restore the source states
+  5. Let the vCPU run
Signature · 1 line
2.20.1
#44 lore raw

Re: [PATCH v2 09/16] KVM: PPC: Book3S HV: XIVE: add a control to dirty the XIVE EQ pages

Quoted 40 lines · Cédric Le Goater

When migration of a VM is initiated, a first copy of the RAM is transferred to the destination before the VM is stopped, but there is no guarantee that the EQ pages in which the event notification are queued have not been modified.

To make sure migration will capture a consistent memory state, the XIVE device should perform a XIVE quiesce sequence to stop the flow of event notifications and stabilize the EQs. This is the purpose of the KVM_DEV_XIVE_EQ_SYNC control which will also marks the EQ pages dirty to force their transfer.

---
 arch/powerpc/include/uapi/asm/kvm.h        |  1 +
 arch/powerpc/kvm/book3s_xive_native.c      | 67 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt | 29 ++++++++++
 3 files changed, 97 insertions(+)
arch/powerpc/include/uapi/asm/kvm.h +1 -0
index 289c504b7c1d..cd78ad1020fe 100644
@@ -678,6 +678,7 @@ struct kvm_ppc_cpu_char {
 /* POWER9 XIVE Native Interrupt Controller */
 #define KVM_DEV_XIVE_GRP_CTRL		1
 #define   KVM_DEV_XIVE_RESET		1
+#define   KVM_DEV_XIVE_EQ_SYNC		2
 #define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_SOURCE_CONFIG	3	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_EQ_CONFIG	4	/* 64-bit eq attributes */
arch/powerpc/kvm/book3s_xive_native.c +2 -0
index dd2a9d411fe7..3debc876d5a0 100644
@@ -640,6 +640,70 @@ static int kvmppc_xive_reset(struct kvmppc_xive *xive)
 	return 0;
 }
 
+static void kvmppc_xive_native_sync_sources(struct kvmppc_xive_src_block *sb)
+{

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_reset(struct kvmppc_xive *xive) · line 654

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+	int j;
+
+	for (j = 0; j < KVMPPC_XICS_IRQ_PER_ICS; j++) {
+		struct kvmppc_xive_irq_state *state = &sb->irq_state[j];
+		struct xive_irq_data *xd;
+		u32 hw_num;
+
+		if (!state->valid)
+			continue;
+		if (state->act_priority == MASKED)

Is this correct? If you masked an irq, then immediately did a sync, couldn't there still be some of the irqs in flight? I thought the reason we needed a sync was that masking and other such operations _didn't_ implicitly synchronize.

Quoted 116 lines
arch/powerpc/kvm/book3s_xive_native.c +55 -0
+			continue;
+
+		arch_spin_lock(&sb->lock);
+		kvmppc_xive_select_irq(state, &hw_num, &xd);
+		xive_native_sync_source(hw_num);
+		xive_native_sync_queue(hw_num);
+		arch_spin_unlock(&sb->lock);
+	}
+}
+
+static int kvmppc_xive_native_vcpu_eq_sync(struct kvm_vcpu *vcpu)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	unsigned int prio;
+
+	if (!xc)
+		return -ENOENT;
+
+	for (prio = 0; prio < KVMPPC_XIVE_Q_COUNT; prio++) {
+		struct xive_q *q = &xc->queues[prio];
+
+		if (!q->qpage)
+			continue;
+
+		/* Mark EQ page dirty for migration */
+		mark_page_dirty(vcpu->kvm, gpa_to_gfn(q->guest_qpage));
+	}
+	return 0;
+}
+
+static int kvmppc_xive_native_eq_sync(struct kvmppc_xive *xive)
+{
+	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
+	unsigned int i;
+
+	pr_devel("%s\n", __func__);
+
+	for (i = 0; i <= xive->max_sbid; i++) {
+		if (xive->src_blocks[i])
+			kvmppc_xive_native_sync_sources(xive->src_blocks[i]);
+	}
+
+	mutex_lock(&kvm->lock);
+	kvm_for_each_vcpu(i, vcpu, kvm) {
+		kvmppc_xive_native_vcpu_eq_sync(vcpu);
+	}
+	mutex_unlock(&kvm->lock);
+
+	return 0;
+}
+
 static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 				       struct kvm_device_attr *attr)
 {
@@ -650,6 +714,8 @@ static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 		switch (attr->attr) {
 		case KVM_DEV_XIVE_RESET:
 			return kvmppc_xive_reset(xive);
+		case KVM_DEV_XIVE_EQ_SYNC:
+			return kvmppc_xive_native_eq_sync(xive);
 		}
 		break;
 	case KVM_DEV_XIVE_GRP_SOURCE:
@@ -688,6 +754,7 @@ static int kvmppc_xive_native_has_attr(struct kvm_device *dev,
 	case KVM_DEV_XIVE_GRP_CTRL:
 		switch (attr->attr) {
 		case KVM_DEV_XIVE_RESET:
+		case KVM_DEV_XIVE_EQ_SYNC:
 			return 0;
 		}
 		break;
Documentation/virtual/kvm/devices/xive.txt +29 -0
index 267634eae9e0..a26be635cff9 100644
@@ -23,6 +23,12 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     queues. To be used by kexec and kdump.
     Errors: none
 
+    1.2 KVM_DEV_XIVE_EQ_SYNC (write only)
+    Sync all the sources and queues and mark the EQ pages dirty. This
+    to make sure that a consistent memory state is captured when
+    migrating the VM.
+    Errors: none
+
   2. KVM_DEV_XIVE_GRP_SOURCE (write only)
   Initializes a new source in the XIVE device and mask it.
   Attributes:
@@ -95,3 +101,26 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -ENOENT: Unknown source number
     -EINVAL: Not initialized source number, invalid priority or
              invalid CPU number.
+
+* Migration:
+
+  Saving the state of a VM using the XIVE native exploitation mode
+  should follow a specific sequence. When the VM is stopped :
+
+  1. Mask all sources (PQ=01) to stop the flow of events.
+
+  2. Sync the XIVE device with the KVM control KVM_DEV_XIVE_EQ_SYNC to
+  flush any in-flight event notification and to stabilize the EQs. At
+  this stage, the EQ pages are marked dirty to make sure they are
+  transferred in the migration sequence.
+
+  3. Capture the state of the source targeting, the EQs configuration
+  and the state of thread interrupt context registers.
+
+  Restore is similar :
+
+  1. Restore the EQ configuration. As targeting depends on it.
+  2. Restore targeting
+  3. Restore the thread interrupt contexts
+  4. Restore the source states
+  5. Let the vCPU run
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#45 lore raw

Re: [PATCH v2 09/16] KVM: PPC: Book3S HV: XIVE: add a control to dirty the XIVE EQ pages

Quoted 48 lines · David Gibson
Quoted 47 lines · Cédric Le Goater

When migration of a VM is initiated, a first copy of the RAM is transferred to the destination before the VM is stopped, but there is no guarantee that the EQ pages in which the event notification are queued have not been modified.

To make sure migration will capture a consistent memory state, the XIVE device should perform a XIVE quiesce sequence to stop the flow of event notifications and stabilize the EQs. This is the purpose of the KVM_DEV_XIVE_EQ_SYNC control which will also marks the EQ pages dirty to force their transfer.

---
 arch/powerpc/include/uapi/asm/kvm.h        |  1 +
 arch/powerpc/kvm/book3s_xive_native.c      | 67 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt | 29 ++++++++++
 3 files changed, 97 insertions(+)
arch/powerpc/include/uapi/asm/kvm.h +1 -0
index 289c504b7c1d..cd78ad1020fe 100644
@@ -678,6 +678,7 @@ struct kvm_ppc_cpu_char {
 /* POWER9 XIVE Native Interrupt Controller */
 #define KVM_DEV_XIVE_GRP_CTRL		1
 #define   KVM_DEV_XIVE_RESET		1
+#define   KVM_DEV_XIVE_EQ_SYNC		2
 #define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_SOURCE_CONFIG	3	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_EQ_CONFIG	4	/* 64-bit eq attributes */
arch/powerpc/kvm/book3s_xive_native.c +9 -0
index dd2a9d411fe7..3debc876d5a0 100644
@@ -640,6 +640,70 @@ static int kvmppc_xive_reset(struct kvmppc_xive *xive)
 	return 0;
 }
 
+static void kvmppc_xive_native_sync_sources(struct kvmppc_xive_src_block *sb)
+{
+	int j;
+
+	for (j = 0; j < KVMPPC_XICS_IRQ_PER_ICS; j++) {
+		struct kvmppc_xive_irq_state *state = &sb->irq_state[j];
+		struct xive_irq_data *xd;
+		u32 hw_num;
+
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_reset(struct kvmppc_xive *xive) · line 654

Quoted 3 lines
arch/powerpc/kvm/book3s_xive_native.c +3 -0
+		if (!state->valid)
+			continue;
+		if (state->act_priority == MASKED)

Is this correct? If you masked an irq, then immediately did a sync, couldn't there still be some of the irqs in flight? I thought the reason we needed a sync was that masking and other such operations _didn't_ implicitly synchronize.

The struct kvmppc_xive_irq_state reflects the state of the EAS configuration and not the state of the source. The source is masked setting the PQ bits to '-Q', which is what is being done before calling the KVM_DEV_XIVE_EQ_SYNC control.

If a source EAS is configured, OPAL syncs the XIVE IC of the source and the XIVE IC of the previous target if any.

So I think we are fine.

C.

Quoted 117 lines
Quoted 116 lines
arch/powerpc/kvm/book3s_xive_native.c +55 -0
+			continue;
+
+		arch_spin_lock(&sb->lock);
+		kvmppc_xive_select_irq(state, &hw_num, &xd);
+		xive_native_sync_source(hw_num);
+		xive_native_sync_queue(hw_num);
+		arch_spin_unlock(&sb->lock);
+	}
+}
+
+static int kvmppc_xive_native_vcpu_eq_sync(struct kvm_vcpu *vcpu)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	unsigned int prio;
+
+	if (!xc)
+		return -ENOENT;
+
+	for (prio = 0; prio < KVMPPC_XIVE_Q_COUNT; prio++) {
+		struct xive_q *q = &xc->queues[prio];
+
+		if (!q->qpage)
+			continue;
+
+		/* Mark EQ page dirty for migration */
+		mark_page_dirty(vcpu->kvm, gpa_to_gfn(q->guest_qpage));
+	}
+	return 0;
+}
+
+static int kvmppc_xive_native_eq_sync(struct kvmppc_xive *xive)
+{
+	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
+	unsigned int i;
+
+	pr_devel("%s\n", __func__);
+
+	for (i = 0; i <= xive->max_sbid; i++) {
+		if (xive->src_blocks[i])
+			kvmppc_xive_native_sync_sources(xive->src_blocks[i]);
+	}
+
+	mutex_lock(&kvm->lock);
+	kvm_for_each_vcpu(i, vcpu, kvm) {
+		kvmppc_xive_native_vcpu_eq_sync(vcpu);
+	}
+	mutex_unlock(&kvm->lock);
+
+	return 0;
+}
+
 static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 				       struct kvm_device_attr *attr)
 {
@@ -650,6 +714,8 @@ static int kvmppc_xive_native_set_attr(struct kvm_device *dev,
 		switch (attr->attr) {
 		case KVM_DEV_XIVE_RESET:
 			return kvmppc_xive_reset(xive);
+		case KVM_DEV_XIVE_EQ_SYNC:
+			return kvmppc_xive_native_eq_sync(xive);
 		}
 		break;
 	case KVM_DEV_XIVE_GRP_SOURCE:
@@ -688,6 +754,7 @@ static int kvmppc_xive_native_has_attr(struct kvm_device *dev,
 	case KVM_DEV_XIVE_GRP_CTRL:
 		switch (attr->attr) {
 		case KVM_DEV_XIVE_RESET:
+		case KVM_DEV_XIVE_EQ_SYNC:
 			return 0;
 		}
 		break;
Documentation/virtual/kvm/devices/xive.txt +29 -0
index 267634eae9e0..a26be635cff9 100644
@@ -23,6 +23,12 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     queues. To be used by kexec and kdump.
     Errors: none
 
+    1.2 KVM_DEV_XIVE_EQ_SYNC (write only)
+    Sync all the sources and queues and mark the EQ pages dirty. This
+    to make sure that a consistent memory state is captured when
+    migrating the VM.
+    Errors: none
+
   2. KVM_DEV_XIVE_GRP_SOURCE (write only)
   Initializes a new source in the XIVE device and mask it.
   Attributes:
@@ -95,3 +101,26 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -ENOENT: Unknown source number
     -EINVAL: Not initialized source number, invalid priority or
              invalid CPU number.
+
+* Migration:
+
+  Saving the state of a VM using the XIVE native exploitation mode
+  should follow a specific sequence. When the VM is stopped :
+
+  1. Mask all sources (PQ=01) to stop the flow of events.
+
+  2. Sync the XIVE device with the KVM control KVM_DEV_XIVE_EQ_SYNC to
+  flush any in-flight event notification and to stabilize the EQs. At
+  this stage, the EQ pages are marked dirty to make sure they are
+  transferred in the migration sequence.
+
+  3. Capture the state of the source targeting, the EQs configuration
+  and the state of thread interrupt context registers.
+
+  Restore is similar :
+
+  1. Restore the EQ configuration. As targeting depends on it.
+  2. Restore targeting
+  3. Restore the thread interrupt contexts
+  4. Restore the source states
+  5. Let the vCPU run
#46 lore raw

Re: [PATCH v2 09/16] KVM: PPC: Book3S HV: XIVE: add a control to dirty the XIVE EQ pages

Quoted 63 lines · Cédric Le Goater
Quoted 48 lines · David Gibson
Quoted 47 lines · Cédric Le Goater

When migration of a VM is initiated, a first copy of the RAM is transferred to the destination before the VM is stopped, but there is no guarantee that the EQ pages in which the event notification are queued have not been modified.

To make sure migration will capture a consistent memory state, the XIVE device should perform a XIVE quiesce sequence to stop the flow of event notifications and stabilize the EQs. This is the purpose of the KVM_DEV_XIVE_EQ_SYNC control which will also marks the EQ pages dirty to force their transfer.

---
 arch/powerpc/include/uapi/asm/kvm.h        |  1 +
 arch/powerpc/kvm/book3s_xive_native.c      | 67 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt | 29 ++++++++++
 3 files changed, 97 insertions(+)
arch/powerpc/include/uapi/asm/kvm.h +1 -0
index 289c504b7c1d..cd78ad1020fe 100644
@@ -678,6 +678,7 @@ struct kvm_ppc_cpu_char {
 /* POWER9 XIVE Native Interrupt Controller */
 #define KVM_DEV_XIVE_GRP_CTRL		1
 #define   KVM_DEV_XIVE_RESET		1
+#define   KVM_DEV_XIVE_EQ_SYNC		2
 #define KVM_DEV_XIVE_GRP_SOURCE		2	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_SOURCE_CONFIG	3	/* 64-bit source attributes */
 #define KVM_DEV_XIVE_GRP_EQ_CONFIG	4	/* 64-bit eq attributes */
arch/powerpc/kvm/book3s_xive_native.c +9 -0
index dd2a9d411fe7..3debc876d5a0 100644
@@ -640,6 +640,70 @@ static int kvmppc_xive_reset(struct kvmppc_xive *xive)
 	return 0;
 }
 
+static void kvmppc_xive_native_sync_sources(struct kvmppc_xive_src_block *sb)
+{
+	int j;
+
+	for (j = 0; j < KVMPPC_XICS_IRQ_PER_ICS; j++) {
+		struct kvmppc_xive_irq_state *state = &sb->irq_state[j];
+		struct xive_irq_data *xd;
+		u32 hw_num;
+
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_reset(struct kvmppc_xive *xive) · line 654

Quoted 3 lines
arch/powerpc/kvm/book3s_xive_native.c +3 -0
+		if (!state->valid)
+			continue;
+		if (state->act_priority == MASKED)

Is this correct? If you masked an irq, then immediately did a sync, couldn't there still be some of the irqs in flight? I thought the reason we needed a sync was that masking and other such operations _didn't_ implicitly synchronize.

The struct kvmppc_xive_irq_state reflects the state of the EAS configuration and not the state of the source. The source is masked setting the PQ bits to '-Q', which is what is being done before calling the KVM_DEV_XIVE_EQ_SYNC control.

Quoted 4 lines

If a source EAS is configured, OPAL syncs the XIVE IC of the source and the XIVE IC of the previous target if any.

So I think we are fine.

Ok.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#47 lore raw

PATCH v2 10/16 KVM: PPC: Book3S HV: XIVE: add get/set accessors for the VP XIVE state

In reply to Cédric Le Goater

At a VCPU level, the state of the thread interrupt management registers needs to be collected. These registers are cached under the 'xive_saved_state.w01' field of the VCPU when the VPCU context is pulled from the HW thread. An OPAL call retrieves the backup of the IPB register in the underlying XIVE NVT structure and merges it in the KVM state.

The structures of the interface between QEMU and KVM provisions some extra room (two u64) for further extensions if more state needs to be transferred back to QEMU.

---
 arch/powerpc/include/asm/kvm_ppc.h         | 11 +++
 arch/powerpc/include/uapi/asm/kvm.h        |  2 +
 arch/powerpc/kvm/book3s.c                  | 24 +++++++
 arch/powerpc/kvm/book3s_xive_native.c      | 82 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt | 19 +++++
 5 files changed, 138 insertions(+)
arch/powerpc/include/asm/kvm_ppc.h +11 -0
index 1e61877fe147..664c65051612 100644
@@ -272,6 +272,7 @@ union kvmppc_one_reg {
 		u64	addr;
 		u64	length;
 	}	vpaval;
+	u64	xive_timaval[4];
 };
 
 struct kvmppc_ops {
@@ -604,6 +605,10 @@ extern int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 extern void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu);
 extern void kvmppc_xive_native_init_module(void);
 extern void kvmppc_xive_native_exit_module(void);
+extern int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu,
+				     union kvmppc_one_reg *val);
+extern int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu,
+				     union kvmppc_one_reg *val);
 
 #else
 static inline int kvmppc_xive_set_xive(struct kvm *kvm, u32 irq, u32 server,
@@ -636,6 +641,12 @@ static inline int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 static inline void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu) { }
 static inline void kvmppc_xive_native_init_module(void) { }
 static inline void kvmppc_xive_native_exit_module(void) { }
+static inline int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu,
+					    union kvmppc_one_reg *val)
+{ return 0; }
+static inline int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu,
+					    union kvmppc_one_reg *val)
+{ return -ENOENT; }
 
 #endif /* CONFIG_KVM_XIVE */
 
arch/powerpc/include/uapi/asm/kvm.h +2 -0
index cd78ad1020fe..42d4ef93ec2d 100644
@@ -480,6 +480,8 @@ struct kvm_ppc_cpu_char {
 #define  KVM_REG_PPC_ICP_PPRI_SHIFT	16	/* pending irq priority */
 #define  KVM_REG_PPC_ICP_PPRI_MASK	0xff
 
+#define KVM_REG_PPC_VP_STATE	(KVM_REG_PPC | KVM_REG_SIZE_U256 | 0x8d)
+
 /* Device control API: PPC-specific devices */
 #define KVM_DEV_MPIC_GRP_MISC		1
 #define   KVM_DEV_MPIC_BASE_ADDR	0	/* 64-bit */
arch/powerpc/kvm/book3s.c +24 -0
index 96d43f091255..f85a9211f30c 100644
@@ -641,6 +641,18 @@ int kvmppc_get_one_reg(struct kvm_vcpu *vcpu, u64 id,
 				*val = get_reg_val(id, kvmppc_xics_get_icp(vcpu));
 			break;
 #endif /* CONFIG_KVM_XICS */
+#ifdef CONFIG_KVM_XIVE
+		case KVM_REG_PPC_VP_STATE:
+			if (!vcpu->arch.xive_vcpu) {
+				r = -ENXIO;
+				break;
+			}
+			if (xive_enabled())
+				r = kvmppc_xive_native_get_vp(vcpu, val);
+			else
+				r = -ENXIO;
+			break;
+#endif /* CONFIG_KVM_XIVE */
 		case KVM_REG_PPC_FSCR:
 			*val = get_reg_val(id, vcpu->arch.fscr);
 			break;
@@ -714,6 +726,18 @@ int kvmppc_set_one_reg(struct kvm_vcpu *vcpu, u64 id,
 				r = kvmppc_xics_set_icp(vcpu, set_reg_val(id, *val));
 			break;
 #endif /* CONFIG_KVM_XICS */
+#ifdef CONFIG_KVM_XIVE
+		case KVM_REG_PPC_VP_STATE:
+			if (!vcpu->arch.xive_vcpu) {
+				r = -ENXIO;
+				break;
+			}
+			if (xive_enabled())
+				r = kvmppc_xive_native_set_vp(vcpu, val);
+			else
+				r = -ENXIO;
+			break;
+#endif /* CONFIG_KVM_XIVE */
 		case KVM_REG_PPC_FSCR:
 			vcpu->arch.fscr = set_reg_val(id, *val);
 			break;
arch/powerpc/kvm/book3s_xive_native.c +82 -0

88 lines not on this page · raw message

Documentation/virtual/kvm/devices/xive.txt +19 -0
index a26be635cff9..1b8957c50c53 100644
@@ -102,6 +102,25 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -EINVAL: Not initialized source number, invalid priority or
              invalid CPU number.
 
+* VCPU state
+
+  The XIVE IC maintains VP interrupt state in an internal structure
+  called the NVT. When a VP is not dispatched on a HW processor
+  thread, this structure can be updated by HW if the VP is the target
+  of an event notification.
+
+  It is important for migration to capture the cached IPB from the NVT
+  as it synthesizes the priorities of the pending interrupts. We
+  capture a bit more to report debug information.
+
+  KVM_REG_PPC_VP_STATE (4 * 64bits)
+  bits:     |  63  ....  32  |  31  ....  0  |
+  values:   |   TIMA word0   |   TIMA word1  |
+  bits:     | 127       ..........       64  |
+  values:   |         VP CAM Line            |
+  bits:     | 255       ..........      128  |
+  values:   |            unused              |
+
 * Migration:
 
   Saving the state of a VM using the XIVE native exploitation mode
Signature · 1 line
2.20.1
#48 lore raw

Re: [PATCH v2 10/16] KVM: PPC: Book3S HV: XIVE: add get/set accessors for the VP XIVE state

Quoted 25 lines · Cédric Le Goater

At a VCPU level, the state of the thread interrupt management registers needs to be collected. These registers are cached under the 'xive_saved_state.w01' field of the VCPU when the VPCU context is pulled from the HW thread. An OPAL call retrieves the backup of the IPB register in the underlying XIVE NVT structure and merges it in the KVM state.

The structures of the interface between QEMU and KVM provisions some extra room (two u64) for further extensions if more state needs to be transferred back to QEMU.

---
 arch/powerpc/include/asm/kvm_ppc.h         | 11 +++
 arch/powerpc/include/uapi/asm/kvm.h        |  2 +
 arch/powerpc/kvm/book3s.c                  | 24 +++++++
 arch/powerpc/kvm/book3s_xive_native.c      | 82 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt | 19 +++++
 5 files changed, 138 insertions(+)
arch/powerpc/include/asm/kvm_ppc.h +0 -0
index 1e61877fe147..664c65051612 100644

arch/powerpc/include/asm/kvm_ppc.h union kvmppc_one_reg { · line 275

Quoted 4 lines
arch/powerpc/include/asm/kvm_ppc.h +1 -0
@@ -272,6 +272,7 @@ union kvmppc_one_reg {
 		u64	addr;
 		u64	length;
 	}	vpaval;
+	u64	xive_timaval[4];

This is doubling the size of the userspace visible one_reg union. Is that safe?

Quoted 106 lines
arch/powerpc/include/asm/kvm_ppc.h +10 -0
 };
 
 struct kvmppc_ops {
@@ -604,6 +605,10 @@ extern int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 extern void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu);
 extern void kvmppc_xive_native_init_module(void);
 extern void kvmppc_xive_native_exit_module(void);
+extern int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu,
+				     union kvmppc_one_reg *val);
+extern int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu,
+				     union kvmppc_one_reg *val);
 
 #else
 static inline int kvmppc_xive_set_xive(struct kvm *kvm, u32 irq, u32 server,
@@ -636,6 +641,12 @@ static inline int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 static inline void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu) { }
 static inline void kvmppc_xive_native_init_module(void) { }
 static inline void kvmppc_xive_native_exit_module(void) { }
+static inline int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu,
+					    union kvmppc_one_reg *val)
+{ return 0; }
+static inline int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu,
+					    union kvmppc_one_reg *val)
+{ return -ENOENT; }
 
 #endif /* CONFIG_KVM_XIVE */
 
arch/powerpc/include/uapi/asm/kvm.h +2 -0
index cd78ad1020fe..42d4ef93ec2d 100644
@@ -480,6 +480,8 @@ struct kvm_ppc_cpu_char {
 #define  KVM_REG_PPC_ICP_PPRI_SHIFT	16	/* pending irq priority */
 #define  KVM_REG_PPC_ICP_PPRI_MASK	0xff
 
+#define KVM_REG_PPC_VP_STATE	(KVM_REG_PPC | KVM_REG_SIZE_U256 | 0x8d)
+
 /* Device control API: PPC-specific devices */
 #define KVM_DEV_MPIC_GRP_MISC		1
 #define   KVM_DEV_MPIC_BASE_ADDR	0	/* 64-bit */
arch/powerpc/kvm/book3s.c +24 -0
index 96d43f091255..f85a9211f30c 100644
@@ -641,6 +641,18 @@ int kvmppc_get_one_reg(struct kvm_vcpu *vcpu, u64 id,
 				*val = get_reg_val(id, kvmppc_xics_get_icp(vcpu));
 			break;
 #endif /* CONFIG_KVM_XICS */
+#ifdef CONFIG_KVM_XIVE
+		case KVM_REG_PPC_VP_STATE:
+			if (!vcpu->arch.xive_vcpu) {
+				r = -ENXIO;
+				break;
+			}
+			if (xive_enabled())
+				r = kvmppc_xive_native_get_vp(vcpu, val);
+			else
+				r = -ENXIO;
+			break;
+#endif /* CONFIG_KVM_XIVE */
 		case KVM_REG_PPC_FSCR:
 			*val = get_reg_val(id, vcpu->arch.fscr);
 			break;
@@ -714,6 +726,18 @@ int kvmppc_set_one_reg(struct kvm_vcpu *vcpu, u64 id,
 				r = kvmppc_xics_set_icp(vcpu, set_reg_val(id, *val));
 			break;
 #endif /* CONFIG_KVM_XICS */
+#ifdef CONFIG_KVM_XIVE
+		case KVM_REG_PPC_VP_STATE:
+			if (!vcpu->arch.xive_vcpu) {
+				r = -ENXIO;
+				break;
+			}
+			if (xive_enabled())
+				r = kvmppc_xive_native_set_vp(vcpu, val);
+			else
+				r = -ENXIO;
+			break;
+#endif /* CONFIG_KVM_XIVE */
 		case KVM_REG_PPC_FSCR:
 			vcpu->arch.fscr = set_reg_val(id, *val);
 			break;
arch/powerpc/kvm/book3s_xive_native.c +16 -0
index 3debc876d5a0..132bff52d70a 100644
@@ -845,6 +845,88 @@ static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type)
 	return ret;
 }
 
+/*
+ * Interrupt Pending Buffer (IPB) offset
+ */
+#define TM_IPB_SHIFT 40
+#define TM_IPB_MASK  (((u64) 0xFF) << TM_IPB_SHIFT)
+
+int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu, union kvmppc_one_reg *val)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	u64 opal_state;
+	int rc;
+
+	if (!kvmppc_xive_enabled(vcpu))
+		return -EPERM;
+
+	if (!xc)

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 873

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+		return -ENOENT;
+
+	/* Thread context registers. We only care about IPB and CPPR */
+	val->xive_timaval[0] = vcpu->arch.xive_saved_state.w01;
+
+	/*
+	 * Return the OS CAM line to print out the VP identifier in
+	 * the QEMU monitor. This is not restored.
+	 */
+	val->xive_timaval[1] = vcpu->arch.xive_cam_word;

I'm pretty dubious about this mixing of vital state information with what's basically debug information. Doubly so since it requires changing the ABI to increase the one_reg union's size.

Might be better to have this control only return the 0th and 2nd u64s from the TIMA, with the CAM debug information returned via some other mechanism.

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+
+	/* Get the VP state from OPAL */
+	rc = xive_native_get_vp_state(xc->vp_id, &opal_state);
+	if (rc)
+		return rc;
+
+	/*
+	 * Capture the backup of IPB register in the NVT structure and
+	 * merge it in our KVM VP state.
+	 */

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 893

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+	val->xive_timaval[0] |= cpu_to_be64(opal_state & TM_IPB_MASK);
+
+	pr_devel("%s NSR=%02x CPPR=%02x IBP=%02x PIPR=%02x w01=%016llx w2=%08x opal=%016llx\n",
+		 __func__,
+		 vcpu->arch.xive_saved_state.nsr,
+		 vcpu->arch.xive_saved_state.cppr,
+		 vcpu->arch.xive_saved_state.ipb,
+		 vcpu->arch.xive_saved_state.pipr,
+		 vcpu->arch.xive_saved_state.w01,
+		 (u32) vcpu->arch.xive_cam_word, opal_state);

Hrm.. except you don't seem to be using the last half of the timaval field anyway.

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+
+	return 0;
+}
+
+int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu, union kvmppc_one_reg *val)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	struct kvmppc_xive *xive = vcpu->kvm->arch.xive;
+
+	pr_devel("%s w01=%016llx vp=%016llx\n", __func__,

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 913

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+		 val->xive_timaval[0], val->xive_timaval[1]);
+
+	if (!kvmppc_xive_enabled(vcpu))
+		return -EPERM;
+
+	if (!xc || !xive)
+		return -ENOENT;
+
+	/* We can't update the state of a "pushed" VCPU	 */
+	if (WARN_ON(vcpu->arch.xive_pushed))

What prevents userspace from tripping this WARN_ON()?

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 914

Quoted 1 line
arch/powerpc/kvm/book3s_xive_native.c +1 -0
+		return -EIO;

EBUSY might be more appropriate here.

Quoted 48 lines
arch/powerpc/kvm/book3s_xive_native.c +15 -0
+
+	/*
+	 * Restore the thread context registers. IPB and CPPR should
+	 * be the only ones that matter.
+	 */
+	vcpu->arch.xive_saved_state.w01 = val->xive_timaval[0];
+
+	/*
+	 * There is no need to restore the XIVE internal state (IPB
+	 * stored in the NVT) as the IPB register was merged in KVM VP
+	 * state when captured.
+	 */
+	return 0;
+}
+
 static int xive_native_debug_show(struct seq_file *m, void *private)
 {
 	struct kvmppc_xive *xive = m->private;
Documentation/virtual/kvm/devices/xive.txt +19 -0
index a26be635cff9..1b8957c50c53 100644
@@ -102,6 +102,25 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -EINVAL: Not initialized source number, invalid priority or
              invalid CPU number.
 
+* VCPU state
+
+  The XIVE IC maintains VP interrupt state in an internal structure
+  called the NVT. When a VP is not dispatched on a HW processor
+  thread, this structure can be updated by HW if the VP is the target
+  of an event notification.
+
+  It is important for migration to capture the cached IPB from the NVT
+  as it synthesizes the priorities of the pending interrupts. We
+  capture a bit more to report debug information.
+
+  KVM_REG_PPC_VP_STATE (4 * 64bits)
+  bits:     |  63  ....  32  |  31  ....  0  |
+  values:   |   TIMA word0   |   TIMA word1  |
+  bits:     | 127       ..........       64  |
+  values:   |         VP CAM Line            |
+  bits:     | 255       ..........      128  |
+  values:   |            unused              |
+
 * Migration:
 
   Saving the state of a VM using the XIVE native exploitation mode
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#49 lore raw

Re: [PATCH v2 10/16] KVM: PPC: Book3S HV: XIVE: add get/set accessors for the VP XIVE state

Quoted 21 lines · David Gibson
Quoted 20 lines · Cédric Le Goater

At a VCPU level, the state of the thread interrupt management registers needs to be collected. These registers are cached under the 'xive_saved_state.w01' field of the VCPU when the VPCU context is pulled from the HW thread. An OPAL call retrieves the backup of the IPB register in the underlying XIVE NVT structure and merges it in the KVM state.

The structures of the interface between QEMU and KVM provisions some extra room (two u64) for further extensions if more state needs to be transferred back to QEMU.

---
 arch/powerpc/include/asm/kvm_ppc.h         | 11 +++
 arch/powerpc/include/uapi/asm/kvm.h        |  2 +
 arch/powerpc/kvm/book3s.c                  | 24 +++++++
 arch/powerpc/kvm/book3s_xive_native.c      | 82 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt | 19 +++++
 5 files changed, 138 insertions(+)
Quoted 12 lines

arch/powerpc/include/asm/kvm_ppc.h union kvmppc_one_reg { · line 275

Quoted 9 lines
arch/powerpc/include/asm/kvm_ppc.h +1 -0
index 1e61877fe147..664c65051612 100644
@@ -272,6 +272,7 @@ union kvmppc_one_reg {
 		u64	addr;
 		u64	length;
 	}	vpaval;
+	u64	xive_timaval[4];

This is doubling the size of the userspace visible one_reg union. Is that safe?

'safe' as in compatibility on an older KVM which would still use the old kvmppc_one_reg definition ?

It should be fine as KVM_REG_PPC_VP_STATE would not be handled. Am I wrong ?

Quoted 111 lines
Quoted 111 lines
arch/powerpc/include/asm/kvm_ppc.h +10 -0
 };
 
 struct kvmppc_ops {
@@ -604,6 +605,10 @@ extern int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 extern void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu);
 extern void kvmppc_xive_native_init_module(void);
 extern void kvmppc_xive_native_exit_module(void);
+extern int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu,
+				     union kvmppc_one_reg *val);
+extern int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu,
+				     union kvmppc_one_reg *val);
 
 #else
 static inline int kvmppc_xive_set_xive(struct kvm *kvm, u32 irq, u32 server,
@@ -636,6 +641,12 @@ static inline int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 static inline void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu) { }
 static inline void kvmppc_xive_native_init_module(void) { }
 static inline void kvmppc_xive_native_exit_module(void) { }
+static inline int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu,
+					    union kvmppc_one_reg *val)
+{ return 0; }
+static inline int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu,
+					    union kvmppc_one_reg *val)
+{ return -ENOENT; }
 
 #endif /* CONFIG_KVM_XIVE */
 
arch/powerpc/include/uapi/asm/kvm.h +2 -0
index cd78ad1020fe..42d4ef93ec2d 100644
@@ -480,6 +480,8 @@ struct kvm_ppc_cpu_char {
 #define  KVM_REG_PPC_ICP_PPRI_SHIFT	16	/* pending irq priority */
 #define  KVM_REG_PPC_ICP_PPRI_MASK	0xff
 
+#define KVM_REG_PPC_VP_STATE	(KVM_REG_PPC | KVM_REG_SIZE_U256 | 0x8d)
+
 /* Device control API: PPC-specific devices */
 #define KVM_DEV_MPIC_GRP_MISC		1
 #define   KVM_DEV_MPIC_BASE_ADDR	0	/* 64-bit */
arch/powerpc/kvm/book3s.c +24 -0
index 96d43f091255..f85a9211f30c 100644
@@ -641,6 +641,18 @@ int kvmppc_get_one_reg(struct kvm_vcpu *vcpu, u64 id,
 				*val = get_reg_val(id, kvmppc_xics_get_icp(vcpu));
 			break;
 #endif /* CONFIG_KVM_XICS */
+#ifdef CONFIG_KVM_XIVE
+		case KVM_REG_PPC_VP_STATE:
+			if (!vcpu->arch.xive_vcpu) {
+				r = -ENXIO;
+				break;
+			}
+			if (xive_enabled())
+				r = kvmppc_xive_native_get_vp(vcpu, val);
+			else
+				r = -ENXIO;
+			break;
+#endif /* CONFIG_KVM_XIVE */
 		case KVM_REG_PPC_FSCR:
 			*val = get_reg_val(id, vcpu->arch.fscr);
 			break;
@@ -714,6 +726,18 @@ int kvmppc_set_one_reg(struct kvm_vcpu *vcpu, u64 id,
 				r = kvmppc_xics_set_icp(vcpu, set_reg_val(id, *val));
 			break;
 #endif /* CONFIG_KVM_XICS */
+#ifdef CONFIG_KVM_XIVE
+		case KVM_REG_PPC_VP_STATE:
+			if (!vcpu->arch.xive_vcpu) {
+				r = -ENXIO;
+				break;
+			}
+			if (xive_enabled())
+				r = kvmppc_xive_native_set_vp(vcpu, val);
+			else
+				r = -ENXIO;
+			break;
+#endif /* CONFIG_KVM_XIVE */
 		case KVM_REG_PPC_FSCR:
 			vcpu->arch.fscr = set_reg_val(id, *val);
 			break;
arch/powerpc/kvm/book3s_xive_native.c +21 -0
index 3debc876d5a0..132bff52d70a 100644
@@ -845,6 +845,88 @@ static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type)
 	return ret;
 }
 
+/*
+ * Interrupt Pending Buffer (IPB) offset
+ */
+#define TM_IPB_SHIFT 40
+#define TM_IPB_MASK  (((u64) 0xFF) << TM_IPB_SHIFT)
+
+int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu, union kvmppc_one_reg *val)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	u64 opal_state;
+	int rc;
+
+	if (!kvmppc_xive_enabled(vcpu))
+		return -EPERM;
+
+	if (!xc)
+		return -ENOENT;
+
+	/* Thread context registers. We only care about IPB and CPPR */
+	val->xive_timaval[0] = vcpu->arch.xive_saved_state.w01;
+
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 873

Quoted 5 lines
arch/powerpc/kvm/book3s_xive_native.c +5 -0
+	/*
+	 * Return the OS CAM line to print out the VP identifier in
+	 * the QEMU monitor. This is not restored.
+	 */
+	val->xive_timaval[1] = vcpu->arch.xive_cam_word;

I'm pretty dubious about this mixing of vital state information with what's basically debug information.

I think QEMU deserves to know about the OS CAM line value. I was even thinking about adding the POOL CAM line value for future use (nested)

Quoted 2 lines

Doubly so since it requires changing the ABI to increase the one_reg union's size.

OK. That's one argument.

Quoted 3 lines

Might be better to have this control only return the 0th and 2nd u64s from the TIMA, with the CAM debug information returned via some other mechanism.

Like an extra reg : KVM_REG_PPC_VP_CAM ?

Quoted 15 lines
Quoted 15 lines
arch/powerpc/kvm/book3s_xive_native.c +15 -0
+
+	/* Get the VP state from OPAL */
+	rc = xive_native_get_vp_state(xc->vp_id, &opal_state);
+	if (rc)
+		return rc;
+
+	/*
+	 * Capture the backup of IPB register in the NVT structure and
+	 * merge it in our KVM VP state.
+	 */
+	val->xive_timaval[0] |= cpu_to_be64(opal_state & TM_IPB_MASK);
+
+	pr_devel("%s NSR=%02x CPPR=%02x IBP=%02x PIPR=%02x w01=%016llx w2=%08x opal=%016llx\n",
+		 __func__,
+		 vcpu->arch.xive_saved_state.nsr,
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 893

Quoted 5 lines
arch/powerpc/kvm/book3s_xive_native.c +5 -0
+		 vcpu->arch.xive_saved_state.cppr,
+		 vcpu->arch.xive_saved_state.ipb,
+		 vcpu->arch.xive_saved_state.pipr,
+		 vcpu->arch.xive_saved_state.w01,
+		 (u32) vcpu->arch.xive_cam_word, opal_state);

Hrm.. except you don't seem to be using the last half of the timaval field anyway.

Yes. The two u64 are extras. We can do without.

Would that be ok if I stored the w01 regs in the first u64, the CAM line(s) in the second and remove the extra two u64 ?

Quoted 14 lines
Quoted 14 lines
arch/powerpc/kvm/book3s_xive_native.c +14 -0
+
+	return 0;
+}
+
+int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu, union kvmppc_one_reg *val)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	struct kvmppc_xive *xive = vcpu->kvm->arch.xive;
+
+	pr_devel("%s w01=%016llx vp=%016llx\n", __func__,
+		 val->xive_timaval[0], val->xive_timaval[1]);
+
+	if (!kvmppc_xive_enabled(vcpu))
+		return -EPERM;
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 913

Quoted 6 lines
arch/powerpc/kvm/book3s_xive_native.c +6 -0
+
+	if (!xc || !xive)
+		return -ENOENT;
+
+	/* We can't update the state of a "pushed" VCPU	 */
+	if (WARN_ON(vcpu->arch.xive_pushed))

What prevents userspace from tripping this WARN_ON()?

if the vCPU is executing a vCPU ioctl, it means that it exited the guest and that its interrupt context has been pulled out of XIVE.

Quoted 3 lines

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 914

Quoted 1 line
arch/powerpc/kvm/book3s_xive_native.c +1 -0
+		return -EIO;

EBUSY might be more appropriate here.

OK.

Thanks,

C.

Quoted 50 lines
Quoted 49 lines
arch/powerpc/kvm/book3s_xive_native.c +15 -0
+
+	/*
+	 * Restore the thread context registers. IPB and CPPR should
+	 * be the only ones that matter.
+	 */
+	vcpu->arch.xive_saved_state.w01 = val->xive_timaval[0];
+
+	/*
+	 * There is no need to restore the XIVE internal state (IPB
+	 * stored in the NVT) as the IPB register was merged in KVM VP
+	 * state when captured.
+	 */
+	return 0;
+}
+
 static int xive_native_debug_show(struct seq_file *m, void *private)
 {
 	struct kvmppc_xive *xive = m->private;
Documentation/virtual/kvm/devices/xive.txt +19 -0
index a26be635cff9..1b8957c50c53 100644
@@ -102,6 +102,25 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -EINVAL: Not initialized source number, invalid priority or
              invalid CPU number.
 
+* VCPU state
+
+  The XIVE IC maintains VP interrupt state in an internal structure
+  called the NVT. When a VP is not dispatched on a HW processor
+  thread, this structure can be updated by HW if the VP is the target
+  of an event notification.
+
+  It is important for migration to capture the cached IPB from the NVT
+  as it synthesizes the priorities of the pending interrupts. We
+  capture a bit more to report debug information.
+
+  KVM_REG_PPC_VP_STATE (4 * 64bits)
+  bits:     |  63  ....  32  |  31  ....  0  |
+  values:   |   TIMA word0   |   TIMA word1  |
+  bits:     | 127       ..........       64  |
+  values:   |         VP CAM Line            |
+  bits:     | 255       ..........      128  |
+  values:   |            unused              |
+
 * Migration:
 
   Saving the state of a VM using the XIVE native exploitation mode
#50 lore raw

Re: [PATCH v2 10/16] KVM: PPC: Book3S HV: XIVE: add get/set accessors for the VP XIVE state

Quoted 35 lines · Cédric Le Goater
Quoted 21 lines · David Gibson
Quoted 20 lines · Cédric Le Goater

At a VCPU level, the state of the thread interrupt management registers needs to be collected. These registers are cached under the 'xive_saved_state.w01' field of the VCPU when the VPCU context is pulled from the HW thread. An OPAL call retrieves the backup of the IPB register in the underlying XIVE NVT structure and merges it in the KVM state.

The structures of the interface between QEMU and KVM provisions some extra room (two u64) for further extensions if more state needs to be transferred back to QEMU.

---
 arch/powerpc/include/asm/kvm_ppc.h         | 11 +++
 arch/powerpc/include/uapi/asm/kvm.h        |  2 +
 arch/powerpc/kvm/book3s.c                  | 24 +++++++
 arch/powerpc/kvm/book3s_xive_native.c      | 82 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt | 19 +++++
 5 files changed, 138 insertions(+)
Quoted 12 lines

arch/powerpc/include/asm/kvm_ppc.h union kvmppc_one_reg { · line 275

Quoted 9 lines
arch/powerpc/include/asm/kvm_ppc.h +1 -0
index 1e61877fe147..664c65051612 100644
@@ -272,6 +272,7 @@ union kvmppc_one_reg {
 		u64	addr;
 		u64	length;
 	}	vpaval;
+	u64	xive_timaval[4];

This is doubling the size of the userspace visible one_reg union. Is that safe?

Quoted 2 lines

'safe' as in compatibility on an older KVM which would still use the old kvmppc_one_reg definition ?

I was more thinking of old qemu with a new kernel.

Quoted 2 lines

It should be fine as KVM_REG_PPC_VP_STATE would not be handled. Am I wrong ?

Looks like it should be ok, because we only partially copy the structure to/from userspace due to the one_reg_size() logic. If the whole union was always copied, it would be hilariously unsafe.

Quoted 127 lines
Quoted 112 lines
Quoted 112 lines
arch/powerpc/include/asm/kvm_ppc.h +10 -0
 };
 
 struct kvmppc_ops {
@@ -604,6 +605,10 @@ extern int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 extern void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu);
 extern void kvmppc_xive_native_init_module(void);
 extern void kvmppc_xive_native_exit_module(void);
+extern int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu,
+				     union kvmppc_one_reg *val);
+extern int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu,
+				     union kvmppc_one_reg *val);
 
 #else
 static inline int kvmppc_xive_set_xive(struct kvm *kvm, u32 irq, u32 server,
@@ -636,6 +641,12 @@ static inline int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 static inline void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu) { }
 static inline void kvmppc_xive_native_init_module(void) { }
 static inline void kvmppc_xive_native_exit_module(void) { }
+static inline int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu,
+					    union kvmppc_one_reg *val)
+{ return 0; }
+static inline int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu,
+					    union kvmppc_one_reg *val)
+{ return -ENOENT; }
 
 #endif /* CONFIG_KVM_XIVE */
 
arch/powerpc/include/uapi/asm/kvm.h +2 -0
index cd78ad1020fe..42d4ef93ec2d 100644
@@ -480,6 +480,8 @@ struct kvm_ppc_cpu_char {
 #define  KVM_REG_PPC_ICP_PPRI_SHIFT	16	/* pending irq priority */
 #define  KVM_REG_PPC_ICP_PPRI_MASK	0xff
 
+#define KVM_REG_PPC_VP_STATE	(KVM_REG_PPC | KVM_REG_SIZE_U256 | 0x8d)
+
 /* Device control API: PPC-specific devices */
 #define KVM_DEV_MPIC_GRP_MISC		1
 #define   KVM_DEV_MPIC_BASE_ADDR	0	/* 64-bit */
arch/powerpc/kvm/book3s.c +24 -0
index 96d43f091255..f85a9211f30c 100644
@@ -641,6 +641,18 @@ int kvmppc_get_one_reg(struct kvm_vcpu *vcpu, u64 id,
 				*val = get_reg_val(id, kvmppc_xics_get_icp(vcpu));
 			break;
 #endif /* CONFIG_KVM_XICS */
+#ifdef CONFIG_KVM_XIVE
+		case KVM_REG_PPC_VP_STATE:
+			if (!vcpu->arch.xive_vcpu) {
+				r = -ENXIO;
+				break;
+			}
+			if (xive_enabled())
+				r = kvmppc_xive_native_get_vp(vcpu, val);
+			else
+				r = -ENXIO;
+			break;
+#endif /* CONFIG_KVM_XIVE */
 		case KVM_REG_PPC_FSCR:
 			*val = get_reg_val(id, vcpu->arch.fscr);
 			break;
@@ -714,6 +726,18 @@ int kvmppc_set_one_reg(struct kvm_vcpu *vcpu, u64 id,
 				r = kvmppc_xics_set_icp(vcpu, set_reg_val(id, *val));
 			break;
 #endif /* CONFIG_KVM_XICS */
+#ifdef CONFIG_KVM_XIVE
+		case KVM_REG_PPC_VP_STATE:
+			if (!vcpu->arch.xive_vcpu) {
+				r = -ENXIO;
+				break;
+			}
+			if (xive_enabled())
+				r = kvmppc_xive_native_set_vp(vcpu, val);
+			else
+				r = -ENXIO;
+			break;
+#endif /* CONFIG_KVM_XIVE */
 		case KVM_REG_PPC_FSCR:
 			vcpu->arch.fscr = set_reg_val(id, *val);
 			break;
arch/powerpc/kvm/book3s_xive_native.c +21 -0
index 3debc876d5a0..132bff52d70a 100644
@@ -845,6 +845,88 @@ static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type)
 	return ret;
 }
 
+/*
+ * Interrupt Pending Buffer (IPB) offset
+ */
+#define TM_IPB_SHIFT 40
+#define TM_IPB_MASK  (((u64) 0xFF) << TM_IPB_SHIFT)
+
+int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu, union kvmppc_one_reg *val)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	u64 opal_state;
+	int rc;
+
+	if (!kvmppc_xive_enabled(vcpu))
+		return -EPERM;
+
+	if (!xc)
+		return -ENOENT;
+
+	/* Thread context registers. We only care about IPB and CPPR */
+	val->xive_timaval[0] = vcpu->arch.xive_saved_state.w01;
+
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 873

Quoted 5 lines
arch/powerpc/kvm/book3s_xive_native.c +5 -0
+	/*
+	 * Return the OS CAM line to print out the VP identifier in
+	 * the QEMU monitor. This is not restored.
+	 */
+	val->xive_timaval[1] = vcpu->arch.xive_cam_word;

I'm pretty dubious about this mixing of vital state information with what's basically debug information.

I think QEMU deserves to know about the OS CAM line value. I was even thinking about adding the POOL CAM line value for future use (nested)

Quoted 2 lines

Doubly so since it requires changing the ABI to increase the one_reg union's size.

Quoted 7 lines

OK. That's one argument.

Quoted 3 lines

Might be better to have this control only return the 0th and 2nd u64s from the TIMA, with the CAM debug information returned via some other mechanism.

Like an extra reg : KVM_REG_PPC_VP_CAM ?

That would be the obvious choice, yes.

Quoted 24 lines
Quoted 15 lines
Quoted 15 lines
arch/powerpc/kvm/book3s_xive_native.c +15 -0
+
+	/* Get the VP state from OPAL */
+	rc = xive_native_get_vp_state(xc->vp_id, &opal_state);
+	if (rc)
+		return rc;
+
+	/*
+	 * Capture the backup of IPB register in the NVT structure and
+	 * merge it in our KVM VP state.
+	 */
+	val->xive_timaval[0] |= cpu_to_be64(opal_state & TM_IPB_MASK);
+
+	pr_devel("%s NSR=%02x CPPR=%02x IBP=%02x PIPR=%02x w01=%016llx w2=%08x opal=%016llx\n",
+		 __func__,
+		 vcpu->arch.xive_saved_state.nsr,
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 893

Quoted 5 lines
arch/powerpc/kvm/book3s_xive_native.c +5 -0
+		 vcpu->arch.xive_saved_state.cppr,
+		 vcpu->arch.xive_saved_state.ipb,
+		 vcpu->arch.xive_saved_state.pipr,
+		 vcpu->arch.xive_saved_state.w01,
+		 (u32) vcpu->arch.xive_cam_word, opal_state);

Hrm.. except you don't seem to be using the last half of the timaval field anyway.

Quoted 4 lines

Yes. The two u64 are extras. We can do without.

Would that be ok if I stored the w01 regs in the first u64, the CAM line(s) in the second and remove the extra two u64 ?

I'd still prefer them in separate regs. They kind of belong to different categories of information, and I can't think of any particular reason you'd have to update or fetch them as a unit.

Quoted 24 lines
Quoted 15 lines
Quoted 15 lines
arch/powerpc/kvm/book3s_xive_native.c +14 -0
+
+	return 0;
+}
+
+int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu, union kvmppc_one_reg *val)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	struct kvmppc_xive *xive = vcpu->kvm->arch.xive;
+
+	pr_devel("%s w01=%016llx vp=%016llx\n", __func__,
+		 val->xive_timaval[0], val->xive_timaval[1]);
+
+	if (!kvmppc_xive_enabled(vcpu))
+		return -EPERM;
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 913

Quoted 6 lines
arch/powerpc/kvm/book3s_xive_native.c +6 -0
+
+	if (!xc || !xive)
+		return -ENOENT;
+
+	/* We can't update the state of a "pushed" VCPU	 */
+	if (WARN_ON(vcpu->arch.xive_pushed))

What prevents userspace from tripping this WARN_ON()?

Quoted 2 lines

if the vCPU is executing a vCPU ioctl, it means that it exited the guest and that its interrupt context has been pulled out of XIVE.

But couldn't one user thread call the vcpu ioctl() while another is inside the guest?

Quoted 61 lines
Quoted 3 lines

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 914

Quoted 1 line
arch/powerpc/kvm/book3s_xive_native.c +1 -0
+		return -EIO;

EBUSY might be more appropriate here.

OK.

Thanks,

C.

Quoted 50 lines
Quoted 49 lines
arch/powerpc/kvm/book3s_xive_native.c +15 -0
+
+	/*
+	 * Restore the thread context registers. IPB and CPPR should
+	 * be the only ones that matter.
+	 */
+	vcpu->arch.xive_saved_state.w01 = val->xive_timaval[0];
+
+	/*
+	 * There is no need to restore the XIVE internal state (IPB
+	 * stored in the NVT) as the IPB register was merged in KVM VP
+	 * state when captured.
+	 */
+	return 0;
+}
+
 static int xive_native_debug_show(struct seq_file *m, void *private)
 {
 	struct kvmppc_xive *xive = m->private;
Documentation/virtual/kvm/devices/xive.txt +19 -0
index a26be635cff9..1b8957c50c53 100644
@@ -102,6 +102,25 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -EINVAL: Not initialized source number, invalid priority or
              invalid CPU number.
 
+* VCPU state
+
+  The XIVE IC maintains VP interrupt state in an internal structure
+  called the NVT. When a VP is not dispatched on a HW processor
+  thread, this structure can be updated by HW if the VP is the target
+  of an event notification.
+
+  It is important for migration to capture the cached IPB from the NVT
+  as it synthesizes the priorities of the pending interrupts. We
+  capture a bit more to report debug information.
+
+  KVM_REG_PPC_VP_STATE (4 * 64bits)
+  bits:     |  63  ....  32  |  31  ....  0  |
+  values:   |   TIMA word0   |   TIMA word1  |
+  bits:     | 127       ..........       64  |
+  values:   |         VP CAM Line            |
+  bits:     | 255       ..........      128  |
+  values:   |            unused              |
+
 * Migration:
 
   Saving the state of a VM using the XIVE native exploitation mode
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#51 lore raw

Re: [PATCH v2 10/16] KVM: PPC: Book3S HV: XIVE: add get/set accessors for the VP XIVE state

Quoted 177 lines · David Gibson
Quoted 35 lines · Cédric Le Goater
Quoted 21 lines · David Gibson
Quoted 20 lines · Cédric Le Goater

At a VCPU level, the state of the thread interrupt management registers needs to be collected. These registers are cached under the 'xive_saved_state.w01' field of the VCPU when the VPCU context is pulled from the HW thread. An OPAL call retrieves the backup of the IPB register in the underlying XIVE NVT structure and merges it in the KVM state.

The structures of the interface between QEMU and KVM provisions some extra room (two u64) for further extensions if more state needs to be transferred back to QEMU.

---
 arch/powerpc/include/asm/kvm_ppc.h         | 11 +++
 arch/powerpc/include/uapi/asm/kvm.h        |  2 +
 arch/powerpc/kvm/book3s.c                  | 24 +++++++
 arch/powerpc/kvm/book3s_xive_native.c      | 82 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt | 19 +++++
 5 files changed, 138 insertions(+)
Quoted 12 lines

arch/powerpc/include/asm/kvm_ppc.h union kvmppc_one_reg { · line 275

Quoted 9 lines
arch/powerpc/include/asm/kvm_ppc.h +1 -0
index 1e61877fe147..664c65051612 100644
@@ -272,6 +272,7 @@ union kvmppc_one_reg {
 		u64	addr;
 		u64	length;
 	}	vpaval;
+	u64	xive_timaval[4];

This is doubling the size of the userspace visible one_reg union. Is that safe?

Quoted 2 lines

'safe' as in compatibility on an older KVM which would still use the old kvmppc_one_reg definition ?

I was more thinking of old qemu with a new kernel.

Quoted 2 lines

It should be fine as KVM_REG_PPC_VP_STATE would not be handled. Am I wrong ?

Looks like it should be ok, because we only partially copy the structure to/from userspace due to the one_reg_size() logic. If the whole union was always copied, it would be hilariously unsafe.

Quoted 129 lines
Quoted 112 lines
Quoted 112 lines
arch/powerpc/include/asm/kvm_ppc.h +10 -0
 };
 
 struct kvmppc_ops {
@@ -604,6 +605,10 @@ extern int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 extern void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu);
 extern void kvmppc_xive_native_init_module(void);
 extern void kvmppc_xive_native_exit_module(void);
+extern int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu,
+				     union kvmppc_one_reg *val);
+extern int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu,
+				     union kvmppc_one_reg *val);
 
 #else
 static inline int kvmppc_xive_set_xive(struct kvm *kvm, u32 irq, u32 server,
@@ -636,6 +641,12 @@ static inline int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 static inline void kvmppc_xive_native_cleanup_vcpu(struct kvm_vcpu *vcpu) { }
 static inline void kvmppc_xive_native_init_module(void) { }
 static inline void kvmppc_xive_native_exit_module(void) { }
+static inline int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu,
+					    union kvmppc_one_reg *val)
+{ return 0; }
+static inline int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu,
+					    union kvmppc_one_reg *val)
+{ return -ENOENT; }
 
 #endif /* CONFIG_KVM_XIVE */
 
arch/powerpc/include/uapi/asm/kvm.h +2 -0
index cd78ad1020fe..42d4ef93ec2d 100644
@@ -480,6 +480,8 @@ struct kvm_ppc_cpu_char {
 #define  KVM_REG_PPC_ICP_PPRI_SHIFT	16	/* pending irq priority */
 #define  KVM_REG_PPC_ICP_PPRI_MASK	0xff
 
+#define KVM_REG_PPC_VP_STATE	(KVM_REG_PPC | KVM_REG_SIZE_U256 | 0x8d)
+
 /* Device control API: PPC-specific devices */
 #define KVM_DEV_MPIC_GRP_MISC		1
 #define   KVM_DEV_MPIC_BASE_ADDR	0	/* 64-bit */
arch/powerpc/kvm/book3s.c +24 -0
index 96d43f091255..f85a9211f30c 100644
@@ -641,6 +641,18 @@ int kvmppc_get_one_reg(struct kvm_vcpu *vcpu, u64 id,
 				*val = get_reg_val(id, kvmppc_xics_get_icp(vcpu));
 			break;
 #endif /* CONFIG_KVM_XICS */
+#ifdef CONFIG_KVM_XIVE
+		case KVM_REG_PPC_VP_STATE:
+			if (!vcpu->arch.xive_vcpu) {
+				r = -ENXIO;
+				break;
+			}
+			if (xive_enabled())
+				r = kvmppc_xive_native_get_vp(vcpu, val);
+			else
+				r = -ENXIO;
+			break;
+#endif /* CONFIG_KVM_XIVE */
 		case KVM_REG_PPC_FSCR:
 			*val = get_reg_val(id, vcpu->arch.fscr);
 			break;
@@ -714,6 +726,18 @@ int kvmppc_set_one_reg(struct kvm_vcpu *vcpu, u64 id,
 				r = kvmppc_xics_set_icp(vcpu, set_reg_val(id, *val));
 			break;
 #endif /* CONFIG_KVM_XICS */
+#ifdef CONFIG_KVM_XIVE
+		case KVM_REG_PPC_VP_STATE:
+			if (!vcpu->arch.xive_vcpu) {
+				r = -ENXIO;
+				break;
+			}
+			if (xive_enabled())
+				r = kvmppc_xive_native_set_vp(vcpu, val);
+			else
+				r = -ENXIO;
+			break;
+#endif /* CONFIG_KVM_XIVE */
 		case KVM_REG_PPC_FSCR:
 			vcpu->arch.fscr = set_reg_val(id, *val);
 			break;
arch/powerpc/kvm/book3s_xive_native.c +21 -0
index 3debc876d5a0..132bff52d70a 100644
@@ -845,6 +845,88 @@ static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type)
 	return ret;
 }
 
+/*
+ * Interrupt Pending Buffer (IPB) offset
+ */
+#define TM_IPB_SHIFT 40
+#define TM_IPB_MASK  (((u64) 0xFF) << TM_IPB_SHIFT)
+
+int kvmppc_xive_native_get_vp(struct kvm_vcpu *vcpu, union kvmppc_one_reg *val)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	u64 opal_state;
+	int rc;
+
+	if (!kvmppc_xive_enabled(vcpu))
+		return -EPERM;
+
+	if (!xc)
+		return -ENOENT;
+
+	/* Thread context registers. We only care about IPB and CPPR */
+	val->xive_timaval[0] = vcpu->arch.xive_saved_state.w01;
+
Quoted 8 lines

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 873

Quoted 5 lines
arch/powerpc/kvm/book3s_xive_native.c +5 -0
+	/*
+	 * Return the OS CAM line to print out the VP identifier in
+	 * the QEMU monitor. This is not restored.
+	 */
+	val->xive_timaval[1] = vcpu->arch.xive_cam_word;

I'm pretty dubious about this mixing of vital state information with what's basically debug information.

I think QEMU deserves to know about the OS CAM line value. I was even thinking about adding the POOL CAM line value for future use (nested)

Quoted 2 lines

Doubly so since it requires changing the ABI to increase the one_reg union's size.

OK. That's one argument.

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

Might be better to have this control only return the 0th and 2nd u64s from the TIMA, with the CAM debug information returned via some other mechanism.

Like an extra reg : KVM_REG_PPC_VP_CAM ?

That would be the obvious choice, yes.

OK. Let's keep that in mind but I think it is overkill. I would rather have one reg per ring instead.

Quoted 24 lines
Quoted 24 lines
Quoted 23 lines
Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+
+	/* Get the VP state from OPAL */
+	rc = xive_native_get_vp_state(xc->vp_id, &opal_state);
+	if (rc)
+		return rc;
+
+	/*
+	 * Capture the backup of IPB register in the NVT structure and
+	 * merge it in our KVM VP state.
+	 */

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 893

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+	val->xive_timaval[0] |= cpu_to_be64(opal_state & TM_IPB_MASK);
+
+	pr_devel("%s NSR=%02x CPPR=%02x IBP=%02x PIPR=%02x w01=%016llx w2=%08x opal=%016llx\n",
+		 __func__,
+		 vcpu->arch.xive_saved_state.nsr,
+		 vcpu->arch.xive_saved_state.cppr,
+		 vcpu->arch.xive_saved_state.ipb,
+		 vcpu->arch.xive_saved_state.pipr,
+		 vcpu->arch.xive_saved_state.w01,
+		 (u32) vcpu->arch.xive_cam_word, opal_state);

Hrm.. except you don't seem to be using the last half of the timaval field anyway.

Quoted 8 lines
Quoted 4 lines

Yes. The two u64 are extras. We can do without.

Would that be ok if I stored the w01 regs in the first u64, the CAM line(s) in the second and remove the extra two u64 ?

I'd still prefer them in separate regs. They kind of belong to different categories of information, and I can't think of any particular reason you'd have to update or fetch them as a unit.

Because they belong to the same thread interrupt context and the same ring (OS) even if only the hypervisor can set the OS CAM line. The OS can only set the CPPR. QEMU operates at the hypervisor level so it is not violating any privilege level.

Quoted 24 lines
Quoted 24 lines
Quoted 23 lines
Quoted 11 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+
+	return 0;
+}
+
+int kvmppc_xive_native_set_vp(struct kvm_vcpu *vcpu, union kvmppc_one_reg *val)
+{
+	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
+	struct kvmppc_xive *xive = vcpu->kvm->arch.xive;
+
+	pr_devel("%s w01=%016llx vp=%016llx\n", __func__,

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 913

Quoted 10 lines
arch/powerpc/kvm/book3s_xive_native.c +10 -0
+		 val->xive_timaval[0], val->xive_timaval[1]);
+
+	if (!kvmppc_xive_enabled(vcpu))
+		return -EPERM;
+
+	if (!xc || !xive)
+		return -ENOENT;
+
+	/* We can't update the state of a "pushed" VCPU	 */
+	if (WARN_ON(vcpu->arch.xive_pushed))

What prevents userspace from tripping this WARN_ON()?

Quoted 5 lines
Quoted 2 lines

if the vCPU is executing a vCPU ioctl, it means that it exited the guest and that its interrupt context has been pulled out of XIVE.

But couldn't one user thread call the vcpu ioctl() while another is inside the guest?

Not while setting the VP state. The guest is not resumed.

Thanks,

C.

Quoted 63 lines
Quoted 62 lines
Quoted 4 lines

arch/powerpc/kvm/book3s_xive_native.c static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type) · line 914

Quoted 2 lines
arch/powerpc/kvm/book3s_xive_native.c +1 -0
+		return -EIO;

EBUSY might be more appropriate here.

OK.

Thanks,

C.

Quoted 50 lines
Quoted 49 lines
arch/powerpc/kvm/book3s_xive_native.c +15 -0
+
+	/*
+	 * Restore the thread context registers. IPB and CPPR should
+	 * be the only ones that matter.
+	 */
+	vcpu->arch.xive_saved_state.w01 = val->xive_timaval[0];
+
+	/*
+	 * There is no need to restore the XIVE internal state (IPB
+	 * stored in the NVT) as the IPB register was merged in KVM VP
+	 * state when captured.
+	 */
+	return 0;
+}
+
 static int xive_native_debug_show(struct seq_file *m, void *private)
 {
 	struct kvmppc_xive *xive = m->private;
Documentation/virtual/kvm/devices/xive.txt +19 -0
index a26be635cff9..1b8957c50c53 100644
@@ -102,6 +102,25 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
     -EINVAL: Not initialized source number, invalid priority or
              invalid CPU number.
 
+* VCPU state
+
+  The XIVE IC maintains VP interrupt state in an internal structure
+  called the NVT. When a VP is not dispatched on a HW processor
+  thread, this structure can be updated by HW if the VP is the target
+  of an event notification.
+
+  It is important for migration to capture the cached IPB from the NVT
+  as it synthesizes the priorities of the pending interrupts. We
+  capture a bit more to report debug information.
+
+  KVM_REG_PPC_VP_STATE (4 * 64bits)
+  bits:     |  63  ....  32  |  31  ....  0  |
+  values:   |   TIMA word0   |   TIMA word1  |
+  bits:     | 127       ..........       64  |
+  values:   |         VP CAM Line            |
+  bits:     | 255       ..........      128  |
+  values:   |            unused              |
+
 * Migration:
 
   Saving the state of a VM using the XIVE native exploitation mode
#52 lore raw

PATCH v2 11/16 KVM: introduce a 'mmap' method for KVM devices

In reply to Cédric Le Goater

Some KVM devices will want to handle special mappings related to the underlying HW. For instance, the XIVE interrupt controller of the POWER9 processor has MMIO pages for thread interrupt management and for interrupt source control that need to be exposed to the guest when the OS has the required support.

---
 include/linux/kvm_host.h |  1 +
 virt/kvm/kvm_main.c      | 11 +++++++++++
 2 files changed, 12 insertions(+)
include/linux/kvm_host.h +1 -0
index c38cc5eb7e73..cbf81487b69f 100644
@@ -1223,6 +1223,7 @@ struct kvm_device_ops {
 	int (*has_attr)(struct kvm_device *dev, struct kvm_device_attr *attr);
 	long (*ioctl)(struct kvm_device *dev, unsigned int ioctl,
 		      unsigned long arg);
+	int (*mmap)(struct kvm_device *dev, struct vm_area_struct *vma);
 };
 
 void kvm_device_get(struct kvm_device *dev);
virt/kvm/kvm_main.c +11 -0
index 585845203db8..84717d8cb5e4 100644
@@ -2878,6 +2878,16 @@ static long kvm_vcpu_compat_ioctl(struct file *filp,
 }
 #endif
 
+static int kvm_device_mmap(struct file *filp, struct vm_area_struct *vma)
+{
+	struct kvm_device *dev = filp->private_data;
+
+	if (dev->ops->mmap)
+		return dev->ops->mmap(dev, vma);
+
+	return -ENODEV;
+}
+
 static int kvm_device_ioctl_attr(struct kvm_device *dev,
 				 int (*accessor)(struct kvm_device *dev,
 						 struct kvm_device_attr *attr),
@@ -2927,6 +2937,7 @@ static const struct file_operations kvm_device_fops = {
 	.unlocked_ioctl = kvm_device_ioctl,
 	.release = kvm_device_release,
 	KVM_COMPAT(kvm_device_ioctl),
+	.mmap = kvm_device_mmap,
 };
 
 struct kvm_device *kvm_device_from_filp(struct file *filp)
Signature · 1 line
2.20.1
#53 lore raw

Re: [PATCH v2 11/16] KVM: introduce a 'mmap' method for KVM devices

Quoted 7 lines · Cédric Le Goater

Some KVM devices will want to handle special mappings related to the underlying HW. For instance, the XIVE interrupt controller of the POWER9 processor has MMIO pages for thread interrupt management and for interrupt source control that need to be exposed to the guest when the OS has the required support.

Ah, when I suggested mmap() on the base device fd, I hadn't realized there wasn't a facility for that yet.

Have you discussed this with Paolo? We'll need some core KVM buy in to merge this.

Quoted 46 lines
---
 include/linux/kvm_host.h |  1 +
 virt/kvm/kvm_main.c      | 11 +++++++++++
 2 files changed, 12 insertions(+)
include/linux/kvm_host.h +1 -0
index c38cc5eb7e73..cbf81487b69f 100644
@@ -1223,6 +1223,7 @@ struct kvm_device_ops {
 	int (*has_attr)(struct kvm_device *dev, struct kvm_device_attr *attr);
 	long (*ioctl)(struct kvm_device *dev, unsigned int ioctl,
 		      unsigned long arg);
+	int (*mmap)(struct kvm_device *dev, struct vm_area_struct *vma);
 };
 
 void kvm_device_get(struct kvm_device *dev);
virt/kvm/kvm_main.c +11 -0
index 585845203db8..84717d8cb5e4 100644
@@ -2878,6 +2878,16 @@ static long kvm_vcpu_compat_ioctl(struct file *filp,
 }
 #endif
 
+static int kvm_device_mmap(struct file *filp, struct vm_area_struct *vma)
+{
+	struct kvm_device *dev = filp->private_data;
+
+	if (dev->ops->mmap)
+		return dev->ops->mmap(dev, vma);
+
+	return -ENODEV;
+}
+
 static int kvm_device_ioctl_attr(struct kvm_device *dev,
 				 int (*accessor)(struct kvm_device *dev,
 						 struct kvm_device_attr *attr),
@@ -2927,6 +2937,7 @@ static const struct file_operations kvm_device_fops = {
 	.unlocked_ioctl = kvm_device_ioctl,
 	.release = kvm_device_release,
 	KVM_COMPAT(kvm_device_ioctl),
+	.mmap = kvm_device_mmap,
 };
 
 struct kvm_device *kvm_device_from_filp(struct file *filp)
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#54 lore raw

Re: [PATCH v2 11/16] KVM: introduce a 'mmap' method for KVM devices

Hello Paolo,

Quoted 7 lines · David Gibson
Quoted 6 lines · Cédric Le Goater

Some KVM devices will want to handle special mappings related to the underlying HW. For instance, the XIVE interrupt controller of the POWER9 processor has MMIO pages for thread interrupt management and for interrupt source control that need to be exposed to the guest when the OS has the required support.

Quoted 6 lines
Quoted 1 line

Ah, when I suggested mmap() on the base device fd, I hadn't realized there wasn't a facility for that yet.

Have you discussed this with Paolo?

Not yet.

Quoted 1 line

We'll need some core KVM buy in to merge this.

Here is an extension of the KVM device to allow special mappings. Something we would need for the support of the POWER9 XIVE interrupt controller.

There are two MMIOs we need to expose to the guest :

1. HW MMIO controlling of the interrupt presenter registers (TIMA)
2. HW MMIO of the interrupt sources for interrupt management (ESB)

The TIMA could have been exposed with a page offset in the vCPU mapping but as it only makes sense when the XIVE interrupt mode is active, we chose to use directly the KVM device fd for that. Is that ok ?

An alternate solution is to use a device ioctl to allocate an anon fd and do the mapping, but that seems like extra fuss for the same result.

Thanks,

C.

Quoted 47 lines
Quoted 46 lines
---
 include/linux/kvm_host.h |  1 +
 virt/kvm/kvm_main.c      | 11 +++++++++++
 2 files changed, 12 insertions(+)
include/linux/kvm_host.h +1 -0
index c38cc5eb7e73..cbf81487b69f 100644
@@ -1223,6 +1223,7 @@ struct kvm_device_ops {
 	int (*has_attr)(struct kvm_device *dev, struct kvm_device_attr *attr);
 	long (*ioctl)(struct kvm_device *dev, unsigned int ioctl,
 		      unsigned long arg);
+	int (*mmap)(struct kvm_device *dev, struct vm_area_struct *vma);
 };
 
 void kvm_device_get(struct kvm_device *dev);
virt/kvm/kvm_main.c +11 -0
index 585845203db8..84717d8cb5e4 100644
@@ -2878,6 +2878,16 @@ static long kvm_vcpu_compat_ioctl(struct file *filp,
 }
 #endif
 
+static int kvm_device_mmap(struct file *filp, struct vm_area_struct *vma)
+{
+	struct kvm_device *dev = filp->private_data;
+
+	if (dev->ops->mmap)
+		return dev->ops->mmap(dev, vma);
+
+	return -ENODEV;
+}
+
 static int kvm_device_ioctl_attr(struct kvm_device *dev,
 				 int (*accessor)(struct kvm_device *dev,
 						 struct kvm_device_attr *attr),
@@ -2927,6 +2937,7 @@ static const struct file_operations kvm_device_fops = {
 	.unlocked_ioctl = kvm_device_ioctl,
 	.release = kvm_device_release,
 	KVM_COMPAT(kvm_device_ioctl),
+	.mmap = kvm_device_mmap,
 };
 
 struct kvm_device *kvm_device_from_filp(struct file *filp)
maintainer
#55 lore raw

Re: [PATCH v2 11/16] KVM: introduce a 'mmap' method for KVM devices

Quoted 30 lines · Cédric Le Goater

Hello Paolo,

Quoted 7 lines · David Gibson
Quoted 6 lines · Cédric Le Goater

Some KVM devices will want to handle special mappings related to the underlying HW. For instance, the XIVE interrupt controller of the POWER9 processor has MMIO pages for thread interrupt management and for interrupt source control that need to be exposed to the guest when the OS has the required support.

Quoted 6 lines
Quoted 1 line

Ah, when I suggested mmap() on the base device fd, I hadn't realized there wasn't a facility for that yet.

Have you discussed this with Paolo?

Not yet.

Quoted 1 line

We'll need some core KVM buy in to merge this.

Here is an extension of the KVM device to allow special mappings. Something we would need for the support of the POWER9 XIVE interrupt controller.

There are two MMIOs we need to expose to the guest :

1. HW MMIO controlling of the interrupt presenter registers (TIMA)
2. HW MMIO of the interrupt sources for interrupt management (ESB)

Quoted 6 lines

The TIMA could have been exposed with a page offset in the vCPU mapping but as it only makes sense when the XIVE interrupt mode is active, we chose to use directly the KVM device fd for that. Is that ok ?

An alternate solution is to use a device ioctl to allocate an anon fd and do the mapping, but that seems like extra fuss for the same result.

It's okay, it's a natural extension to dev_ops - but thanks for asking anyway. :)

Paolo

#56 lore raw

Re: [PATCH v2 11/16] KVM: introduce a 'mmap' method for KVM devices

Quoted 35 lines · Paolo Bonzini
Quoted 34 lines · Cédric Le Goater

Hello Paolo,

Quoted 7 lines · David Gibson
Quoted 6 lines · Cédric Le Goater

Some KVM devices will want to handle special mappings related to the underlying HW. For instance, the XIVE interrupt controller of the POWER9 processor has MMIO pages for thread interrupt management and for interrupt source control that need to be exposed to the guest when the OS has the required support.

Quoted 6 lines
Quoted 1 line

Ah, when I suggested mmap() on the base device fd, I hadn't realized there wasn't a facility for that yet.

Have you discussed this with Paolo?

Not yet.

Quoted 1 line

We'll need some core KVM buy in to merge this.

Here is an extension of the KVM device to allow special mappings. Something we would need for the support of the POWER9 XIVE interrupt controller.

There are two MMIOs we need to expose to the guest :

1. HW MMIO controlling of the interrupt presenter registers (TIMA)
2. HW MMIO of the interrupt sources for interrupt management (ESB)

The TIMA could have been exposed with a page offset in the vCPU mapping but as it only makes sense when the XIVE interrupt mode is active, we chose to use directly the KVM device fd for that. Is that ok ?

Quoted 5 lines
Quoted 2 lines

An alternate solution is to use a device ioctl to allocate an anon fd and do the mapping, but that seems like extra fuss for the same result.

It's okay, it's a natural extension to dev_ops - but thanks for asking anyway. :)

Ok, cool.

Given that, do you want to merge directly - since this looks sound enough, even though the rest of the series needs some polish? Or would you prefer it to come in via Paulus' tree?

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#57 lore raw

PATCH v2 12/16 KVM: PPC: Book3S HV: XIVE: add a TIMA mapping

In reply to Cédric Le Goater

Each thread has an associated Thread Interrupt Management context composed of a set of registers. These registers let the thread handle priority management and interrupt acknowledgment. The most important are :

- Interrupt Pending Buffer (IPB)
- Current Processor Priority (CPPR)
- Notification Source Register (NSR)

They are exposed to software in four different pages each proposing a view with a different privilege. The first page is for the physical thread context and the second for the hypervisor. Only the third (operating system) and the fourth (user level) are exposed the guest.

A custom VM fault handler will populate the VMA with the appropriate pages, which should only be the OS page for now.

---
 arch/powerpc/include/asm/xive.h            |  1 +
 arch/powerpc/include/uapi/asm/kvm.h        |  2 ++
 arch/powerpc/kvm/book3s_xive_native.c      | 39 ++++++++++++++++++++++
 arch/powerpc/sysdev/xive/native.c          | 11 ++++++
 Documentation/virtual/kvm/devices/xive.txt | 23 +++++++++++++
 5 files changed, 76 insertions(+)
arch/powerpc/include/asm/xive.h +1 -0
index 46891f321606..eb6d302082da 100644
@@ -23,6 +23,7 @@
  * same offset regardless of where the code is executing
  */
 extern void __iomem *xive_tima;
+extern unsigned long xive_tima_os;
 
 /*
  * Offset in the TM area of our current execution level (provided by
arch/powerpc/include/uapi/asm/kvm.h +2 -0
index 42d4ef93ec2d..be9b255e061d 100644
@@ -720,4 +720,6 @@ struct kvm_ppc_xive_eq {
 #define KVM_XIVE_EQ_FLAG_ALWAYS_NOTIFY	0x00000002
 #define KVM_XIVE_EQ_FLAG_ESCALATE	0x00000004
 
+#define KVM_XIVE_TIMA_PAGE_OFFSET	0
+
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive_native.c +39 -0
index 132bff52d70a..c6ac818a13b2 100644
@@ -176,6 +176,44 @@ int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 	return rc;
 }
 
+static int xive_native_tima_fault(struct vm_fault *vmf)
+{
+	struct vm_area_struct *vma = vmf->vma;
+
+	switch (vmf->pgoff - vma->vm_pgoff) {
+	case 0: /* HW - forbid access */
+	case 1: /* HV - forbid access */
+		return VM_FAULT_SIGBUS;
+	case 2: /* OS */
+		vmf_insert_pfn(vma, vmf->address, xive_tima_os >> PAGE_SHIFT);
+		return VM_FAULT_NOPAGE;
+	case 3: /* USER - TODO */
+	default:
+		return VM_FAULT_SIGBUS;
+	}
+}
+
+static const struct vm_operations_struct xive_native_tima_vmops = {
+	.fault = xive_native_tima_fault,
+};
+
+static int kvmppc_xive_native_mmap(struct kvm_device *dev,
+				   struct vm_area_struct *vma)
+{
+	/* We only allow mappings at fixed offset for now */
+	if (vma->vm_pgoff == KVM_XIVE_TIMA_PAGE_OFFSET) {
+		if (vma_pages(vma) > 4)
+			return -EINVAL;
+		vma->vm_ops = &xive_native_tima_vmops;
+	} else {
+		return -EINVAL;
+	}
+
+	vma->vm_flags |= VM_IO | VM_PFNMAP;
+	vma->vm_page_prot = pgprot_noncached_wc(vma->vm_page_prot);
+	return 0;
+}
+
 static int kvmppc_xive_native_set_source(struct kvmppc_xive *xive, long irq,
 					 u64 addr)
 {
@@ -1005,6 +1043,7 @@ struct kvm_device_ops kvm_xive_native_ops = {
 	.set_attr = kvmppc_xive_native_set_attr,
 	.get_attr = kvmppc_xive_native_get_attr,
 	.has_attr = kvmppc_xive_native_has_attr,
+	.mmap = kvmppc_xive_native_mmap,
 };
 
 void kvmppc_xive_native_init_module(void)
arch/powerpc/sysdev/xive/native.c +11 -0
index 0c037e933e55..7782201e5fe8 100644
@@ -521,6 +521,9 @@ u32 xive_native_default_eq_shift(void)
 }
 EXPORT_SYMBOL_GPL(xive_native_default_eq_shift);
 
+unsigned long xive_tima_os;
+EXPORT_SYMBOL_GPL(xive_tima_os);
+
 bool __init xive_native_init(void)
 {
 	struct device_node *np;
@@ -573,6 +576,14 @@ bool __init xive_native_init(void)
 	for_each_possible_cpu(cpu)
 		kvmppc_set_xive_tima(cpu, r.start, tima);
 
+	/* Resource 2 is OS window */
+	if (of_address_to_resource(np, 2, &r)) {
+		pr_err("Failed to get thread mgmnt area resource\n");
+		return false;
+	}
+
+	xive_tima_os = r.start;
+
 	/* Grab size of provisionning pages */
 	xive_parse_provisioning(np);
 
Documentation/virtual/kvm/devices/xive.txt +23 -0
index 1b8957c50c53..4d6b41609fd9 100644
@@ -13,6 +13,29 @@ requires a POWER9 host and the guest OS should have support for the
 XIVE native exploitation interrupt mode. If not, it should run using
 the legacy interrupt mode, referred as XICS (POWER7/8).
 
+* Device Mappings
+
+  The KVM device exposes different MMIO ranges of the XIVE HW which
+  are required for interrupt management. These are exposed to the
+  guest in VMAs populated with a custom VM fault handler.
+
+  1. Thread Interrupt Management Area (TIMA)
+
+  Each thread has an associated Thread Interrupt Management context
+  composed of a set of registers. These registers let the thread
+  handle priority management and interrupt acknowledgment. The most
+  important are :
+
+      - Interrupt Pending Buffer     (IPB)
+      - Current Processor Priority   (CPPR)
+      - Notification Source Register (NSR)
+
+  They are exposed to software in four different pages each proposing
+  a view with a different privilege. The first page is for the
+  physical thread context and the second for the hypervisor. Only the
+  third (operating system) and the fourth (user level) are exposed the
+  guest.
+
 * Groups:
 
   1. KVM_DEV_XIVE_GRP_CTRL
Signature · 1 line
2.20.1
Reviewed-by on v2 12/16
#58 lore raw

Re: [PATCH v2 12/16] KVM: PPC: Book3S HV: XIVE: add a TIMA mapping

Quoted 14 lines · Cédric Le Goater

Each thread has an associated Thread Interrupt Management context composed of a set of registers. These registers let the thread handle priority management and interrupt acknowledgment. The most important are :

- Interrupt Pending Buffer (IPB)
- Current Processor Priority (CPPR)
- Notification Source Register (NSR)

They are exposed to software in four different pages each proposing a view with a different privilege. The first page is for the physical thread context and the second for the hypervisor. Only the third (operating system) and the fourth (user level) are exposed the guest.

Quoted 4 lines

A custom VM fault handler will populate the VMA with the appropriate pages, which should only be the OS page for now.

Subject to possible modification depending on whether we go with the generic change to allow mmap() on kvm devices.

Quoted 151 lines
---
 arch/powerpc/include/asm/xive.h            |  1 +
 arch/powerpc/include/uapi/asm/kvm.h        |  2 ++
 arch/powerpc/kvm/book3s_xive_native.c      | 39 ++++++++++++++++++++++
 arch/powerpc/sysdev/xive/native.c          | 11 ++++++
 Documentation/virtual/kvm/devices/xive.txt | 23 +++++++++++++
 5 files changed, 76 insertions(+)
arch/powerpc/include/asm/xive.h +1 -0
index 46891f321606..eb6d302082da 100644
@@ -23,6 +23,7 @@
  * same offset regardless of where the code is executing
  */
 extern void __iomem *xive_tima;
+extern unsigned long xive_tima_os;
 
 /*
  * Offset in the TM area of our current execution level (provided by
arch/powerpc/include/uapi/asm/kvm.h +2 -0
index 42d4ef93ec2d..be9b255e061d 100644
@@ -720,4 +720,6 @@ struct kvm_ppc_xive_eq {
 #define KVM_XIVE_EQ_FLAG_ALWAYS_NOTIFY	0x00000002
 #define KVM_XIVE_EQ_FLAG_ESCALATE	0x00000004
 
+#define KVM_XIVE_TIMA_PAGE_OFFSET	0
+
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive_native.c +39 -0
index 132bff52d70a..c6ac818a13b2 100644
@@ -176,6 +176,44 @@ int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 	return rc;
 }
 
+static int xive_native_tima_fault(struct vm_fault *vmf)
+{
+	struct vm_area_struct *vma = vmf->vma;
+
+	switch (vmf->pgoff - vma->vm_pgoff) {
+	case 0: /* HW - forbid access */
+	case 1: /* HV - forbid access */
+		return VM_FAULT_SIGBUS;
+	case 2: /* OS */
+		vmf_insert_pfn(vma, vmf->address, xive_tima_os >> PAGE_SHIFT);
+		return VM_FAULT_NOPAGE;
+	case 3: /* USER - TODO */
+	default:
+		return VM_FAULT_SIGBUS;
+	}
+}
+
+static const struct vm_operations_struct xive_native_tima_vmops = {
+	.fault = xive_native_tima_fault,
+};
+
+static int kvmppc_xive_native_mmap(struct kvm_device *dev,
+				   struct vm_area_struct *vma)
+{
+	/* We only allow mappings at fixed offset for now */
+	if (vma->vm_pgoff == KVM_XIVE_TIMA_PAGE_OFFSET) {
+		if (vma_pages(vma) > 4)
+			return -EINVAL;
+		vma->vm_ops = &xive_native_tima_vmops;
+	} else {
+		return -EINVAL;
+	}
+
+	vma->vm_flags |= VM_IO | VM_PFNMAP;
+	vma->vm_page_prot = pgprot_noncached_wc(vma->vm_page_prot);
+	return 0;
+}
+
 static int kvmppc_xive_native_set_source(struct kvmppc_xive *xive, long irq,
 					 u64 addr)
 {
@@ -1005,6 +1043,7 @@ struct kvm_device_ops kvm_xive_native_ops = {
 	.set_attr = kvmppc_xive_native_set_attr,
 	.get_attr = kvmppc_xive_native_get_attr,
 	.has_attr = kvmppc_xive_native_has_attr,
+	.mmap = kvmppc_xive_native_mmap,
 };
 
 void kvmppc_xive_native_init_module(void)
arch/powerpc/sysdev/xive/native.c +11 -0
index 0c037e933e55..7782201e5fe8 100644
@@ -521,6 +521,9 @@ u32 xive_native_default_eq_shift(void)
 }
 EXPORT_SYMBOL_GPL(xive_native_default_eq_shift);
 
+unsigned long xive_tima_os;
+EXPORT_SYMBOL_GPL(xive_tima_os);
+
 bool __init xive_native_init(void)
 {
 	struct device_node *np;
@@ -573,6 +576,14 @@ bool __init xive_native_init(void)
 	for_each_possible_cpu(cpu)
 		kvmppc_set_xive_tima(cpu, r.start, tima);
 
+	/* Resource 2 is OS window */
+	if (of_address_to_resource(np, 2, &r)) {
+		pr_err("Failed to get thread mgmnt area resource\n");
+		return false;
+	}
+
+	xive_tima_os = r.start;
+
 	/* Grab size of provisionning pages */
 	xive_parse_provisioning(np);
 
Documentation/virtual/kvm/devices/xive.txt +23 -0
index 1b8957c50c53..4d6b41609fd9 100644
@@ -13,6 +13,29 @@ requires a POWER9 host and the guest OS should have support for the
 XIVE native exploitation interrupt mode. If not, it should run using
 the legacy interrupt mode, referred as XICS (POWER7/8).
 
+* Device Mappings
+
+  The KVM device exposes different MMIO ranges of the XIVE HW which
+  are required for interrupt management. These are exposed to the
+  guest in VMAs populated with a custom VM fault handler.
+
+  1. Thread Interrupt Management Area (TIMA)
+
+  Each thread has an associated Thread Interrupt Management context
+  composed of a set of registers. These registers let the thread
+  handle priority management and interrupt acknowledgment. The most
+  important are :
+
+      - Interrupt Pending Buffer     (IPB)
+      - Current Processor Priority   (CPPR)
+      - Notification Source Register (NSR)
+
+  They are exposed to software in four different pages each proposing
+  a view with a different privilege. The first page is for the
+  physical thread context and the second for the hypervisor. Only the
+  third (operating system) and the fourth (user level) are exposed the
+  guest.
+
 * Groups:
 
   1. KVM_DEV_XIVE_GRP_CTRL
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#59 lore raw

PATCH v2 13/16 KVM: PPC: Book3S HV: XIVE: add a mapping for the source ESB pages

In reply to Cédric Le Goater

Each source is associated with an Event State Buffer (ESB) with a even/odd pair of pages which provides commands to manage the source:
to trigger, to EOI, to turn off the source for instance.

The custom VM fault handler will deduce the guest IRQ number from the offset of the fault, and the ESB page of the associated XIVE interrupt will be inserted into the VMA using the internal structure caching information on the interrupts.

---
 arch/powerpc/include/uapi/asm/kvm.h        |  1 +
 arch/powerpc/kvm/book3s_xive_native.c      | 57 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  7 +++
 3 files changed, 65 insertions(+)
arch/powerpc/include/uapi/asm/kvm.h +1 -0
index be9b255e061d..d8990e9044e3 100644
@@ -721,5 +721,6 @@ struct kvm_ppc_xive_eq {
 #define KVM_XIVE_EQ_FLAG_ESCALATE	0x00000004
 
 #define KVM_XIVE_TIMA_PAGE_OFFSET	0
+#define KVM_XIVE_ESB_PAGE_OFFSET	4
 
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive_native.c +57 -0

69 lines not on this page · raw message

Documentation/virtual/kvm/devices/xive.txt +7 -0
index 4d6b41609fd9..be5000b2eb5a 100644
@@ -36,6 +36,13 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
   third (operating system) and the fourth (user level) are exposed the
   guest.
 
+  2. Event State Buffer (ESB)
+
+  Each source is associated with an Event State Buffer (ESB) with
+  either a pair of even/odd pair of pages which provides commands to
+  manage the source: to trigger, to EOI, to turn off the source for
+  instance.
+
 * Groups:
 
   1. KVM_DEV_XIVE_GRP_CTRL
Signature · 1 line
2.20.1
Reviewed-by on v2 13/16
#60 lore raw

Re: [PATCH v2 13/16] KVM: PPC: Book3S HV: XIVE: add a mapping for the source ESB pages

Quoted 10 lines · Cédric Le Goater

Each source is associated with an Event State Buffer (ESB) with a even/odd pair of pages which provides commands to manage the source:
to trigger, to EOI, to turn off the source for instance.

The custom VM fault handler will deduce the guest IRQ number from the offset of the fault, and the ESB page of the associated XIVE interrupt will be inserted into the VMA using the internal structure caching information on the interrupts.

With the same caveat as the previous patch.

Quoted 110 lines
---
 arch/powerpc/include/uapi/asm/kvm.h        |  1 +
 arch/powerpc/kvm/book3s_xive_native.c      | 57 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt |  7 +++
 3 files changed, 65 insertions(+)
arch/powerpc/include/uapi/asm/kvm.h +1 -0
index be9b255e061d..d8990e9044e3 100644
@@ -721,5 +721,6 @@ struct kvm_ppc_xive_eq {
 #define KVM_XIVE_EQ_FLAG_ESCALATE	0x00000004
 
 #define KVM_XIVE_TIMA_PAGE_OFFSET	0
+#define KVM_XIVE_ESB_PAGE_OFFSET	4
 
 #endif /* __LINUX_KVM_POWERPC_H */
arch/powerpc/kvm/book3s_xive_native.c +57 -0
index c6ac818a13b2..92cab6409e8e 100644
@@ -176,6 +176,59 @@ int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 	return rc;
 }
 
+static int xive_native_esb_fault(struct vm_fault *vmf)
+{
+	struct vm_area_struct *vma = vmf->vma;
+	struct kvm_device *dev = vma->vm_file->private_data;
+	struct kvmppc_xive *xive = dev->private;
+	struct kvmppc_xive_src_block *sb;
+	struct kvmppc_xive_irq_state *state;
+	struct xive_irq_data *xd;
+	u32 hw_num;
+	u16 src;
+	u64 page;
+	unsigned long irq;
+	u64 page_offset;
+
+	/*
+	 * Linux/KVM uses a two pages ESB setting, one for trigger and
+	 * one for EOI
+	 */
+	page_offset = vmf->pgoff - vma->vm_pgoff;
+	irq = page_offset / 2;
+
+	sb = kvmppc_xive_find_source(xive, irq, &src);
+	if (!sb) {
+		pr_devel("%s: source %lx not found !\n", __func__, irq);
+		return VM_FAULT_SIGBUS;
+	}
+
+	state = &sb->irq_state[src];
+	kvmppc_xive_select_irq(state, &hw_num, &xd);
+
+	arch_spin_lock(&sb->lock);
+
+	/*
+	 * first/even page is for trigger
+	 * second/odd page is for EOI and management.
+	 */
+	page = page_offset % 2 ? xd->eoi_page : xd->trig_page;
+	arch_spin_unlock(&sb->lock);
+
+	if (WARN_ON(!page)) {
+		pr_err("%s: acessing invalid ESB page for source %lx !\n",
+		       __func__, irq);
+		return VM_FAULT_SIGBUS;
+	}
+
+	vmf_insert_pfn(vma, vmf->address, page >> PAGE_SHIFT);
+	return VM_FAULT_NOPAGE;
+}
+
+static const struct vm_operations_struct xive_native_esb_vmops = {
+	.fault = xive_native_esb_fault,
+};
+
 static int xive_native_tima_fault(struct vm_fault *vmf)
 {
 	struct vm_area_struct *vma = vmf->vma;
@@ -205,6 +258,10 @@ static int kvmppc_xive_native_mmap(struct kvm_device *dev,
 		if (vma_pages(vma) > 4)
 			return -EINVAL;
 		vma->vm_ops = &xive_native_tima_vmops;
+	} else if (vma->vm_pgoff == KVM_XIVE_ESB_PAGE_OFFSET) {
+		if (vma_pages(vma) > KVMPPC_XIVE_NR_IRQS * 2)
+			return -EINVAL;
+		vma->vm_ops = &xive_native_esb_vmops;
 	} else {
 		return -EINVAL;
 	}
Documentation/virtual/kvm/devices/xive.txt +7 -0
index 4d6b41609fd9..be5000b2eb5a 100644
@@ -36,6 +36,13 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
   third (operating system) and the fourth (user level) are exposed the
   guest.
 
+  2. Event State Buffer (ESB)
+
+  Each source is associated with an Event State Buffer (ESB) with
+  either a pair of even/odd pair of pages which provides commands to
+  manage the source: to trigger, to EOI, to turn off the source for
+  instance.
+
 * Groups:
 
   1. KVM_DEV_XIVE_GRP_CTRL
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#61 lore raw

PATCH v2 14/16 KVM: PPC: Book3S HV: XIVE: add passthrough support

In reply to Cédric Le Goater

The KVM XICS-over-XIVE device and the proposed KVM XIVE native device implement an IRQ space for the guest using the generic IPI interrupts of the XIVE IC controller. These interrupts are allocated at the OPAL level and "mapped" into the guest IRQ number space in the range 0-0x1FFF. Interrupt management is performed in the XIVE way: using loads and stores on the addresses of the XIVE IPI interrupt ESB pages.

Both KVM devices share the same internal structure caching information on the interrupts, among which the xive_irq_data struct containing the addresses of the IPI ESB pages and an extra one in case of passthrough. The later contains the addresses of the ESB pages of the underlying HW controller interrupts, PHB4 in all cases for now.

A guest, when running in the XICS legacy interrupt mode, lets the KVM XICS-over-XIVE device "handle" interrupt management, that is to perform the loads and stores on the addresses of the ESB pages of the guest interrupts. However, when running in XIVE native exploitation mode, the KVM XIVE native device exposes the interrupt ESB pages to the guest and lets the guest perform directly the loads and stores.

The VMA exposing the ESB pages make use of a custom VM fault handler which role is to populate the VMA with appropriate pages. When a fault occurs, the guest IRQ number is deduced from the offset, and the ESB pages of associated XIVE IPI interrupt are inserted in the VMA (using the internal structure caching information on the interrupts).

Supporting device passthrough in the guest running in XIVE native exploitation mode adds some extra refinements because the ESB pages of a different HW controller (PHB4) need to be exposed to the guest along with the initial IPI ESB pages of the XIVE IC controller. But the overall mechanic is the same.

When the device HW irqs are mapped into or unmapped from the guest IRQ number space, the passthru_irq helpers, kvmppc_xive_set_mapped() and kvmppc_xive_clr_mapped(), are called to record or clear the passthrough interrupt information and to perform the switch.

The approach taken by this patch is to clear the ESB pages of the guest IRQ number being mapped and let the VM fault handler repopulate. The handler will insert the ESB page corresponding to the HW interrupt of the device being passed-through or the initial IPI ESB page if the device is being removed.

---
 arch/powerpc/kvm/book3s_xive.h             |  9 +++++
 arch/powerpc/kvm/book3s_xive.c             | 15 ++++++++
 arch/powerpc/kvm/book3s_xive_native.c      | 41 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt | 15 ++++++++
 4 files changed, 80 insertions(+)
arch/powerpc/kvm/book3s_xive.h +9 -0
index 6660d138c6b7..d1f832a53811 100644
@@ -94,6 +94,11 @@ struct kvmppc_xive_src_block {
 	struct kvmppc_xive_irq_state irq_state[KVMPPC_XICS_IRQ_PER_ICS];
 };
 
+struct kvmppc_xive;
+
+struct kvmppc_xive_ops {
+	int (*reset_mapped)(struct kvm *kvm, unsigned long guest_irq);
+};
 
 struct kvmppc_xive {
 	struct kvm *kvm;
@@ -132,6 +137,10 @@ struct kvmppc_xive {
 
 	/* Flags */
 	u8	single_escalation;
+
+	struct kvmppc_xive_ops *ops;
+	struct address_space   *mapping;
+	struct mutex mapping_lock;
 };
 
 #define KVMPPC_XIVE_Q_COUNT	8
arch/powerpc/kvm/book3s_xive.c +15 -0
index 7431e31bc541..7a14512b8944 100644
@@ -942,6 +942,13 @@ int kvmppc_xive_set_mapped(struct kvm *kvm, unsigned long guest_irq,
 	/* Turn the IPI hard off */
 	xive_vm_esb_load(&state->ipi_data, XIVE_ESB_SET_PQ_01);
 
+	/*
+	 * Reset ESB guest mapping. Needed when ESB pages are exposed
+	 * to the guest in XIVE native mode
+	 */
+	if (xive->ops && xive->ops->reset_mapped)
+		xive->ops->reset_mapped(kvm, guest_irq);
+
 	/* Grab info about irq */
 	state->pt_number = hw_irq;
 	state->pt_data = irq_data_get_irq_handler_data(host_data);
@@ -1027,6 +1034,14 @@ int kvmppc_xive_clr_mapped(struct kvm *kvm, unsigned long guest_irq,
 	state->pt_number = 0;
 	state->pt_data = NULL;
 
+	/*
+	 * Reset ESB guest mapping. Needed when ESB pages are exposed
+	 * to the guest in XIVE native mode
+	 */
+	if (xive->ops && xive->ops->reset_mapped) {
+		xive->ops->reset_mapped(kvm, guest_irq);
+	}
+
 	/* Reconfigure the IPI */
 	xive_native_configure_irq(state->ipi_number,
 				  xive_vp(xive, state->act_server),
arch/powerpc/kvm/book3s_xive_native.c +41 -0

77 lines not on this page · raw message

Documentation/virtual/kvm/devices/xive.txt +15 -0
index be5000b2eb5a..7a242cb07e7c 100644
@@ -43,6 +43,21 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
   manage the source: to trigger, to EOI, to turn off the source for
   instance.
 
+  3. Device passthrough
+
+  When a device is passed-through into the guest, the source
+  interrupts are from a different HW controller (PHB4) and the ESB
+  pages exposed to the guest should accommadate this change.
+
+  The passthru_irq helpers, kvmppc_xive_set_mapped() and
+  kvmppc_xive_clr_mapped() are called when the device HW irqs are
+  mapped into or unmapped from the guest IRQ number space. The KVM
+  device extends these helpers to clear the ESB pages of the guest IRQ
+  number being mapped and then lets the VM fault handler repopulate.
+  The handler will insert the ESB page corresponding to the HW
+  interrupt of the device being passed-through or the initial IPI ESB
+  page if the device has being removed.
+
 * Groups:
 
   1. KVM_DEV_XIVE_GRP_CTRL
Signature · 1 line
2.20.1
#62 lore raw

Re: [PATCH v2 14/16] KVM: PPC: Book3S HV: XIVE: add passthrough support

Quoted 210 lines · Cédric Le Goater

The KVM XICS-over-XIVE device and the proposed KVM XIVE native device implement an IRQ space for the guest using the generic IPI interrupts of the XIVE IC controller. These interrupts are allocated at the OPAL level and "mapped" into the guest IRQ number space in the range 0-0x1FFF. Interrupt management is performed in the XIVE way: using loads and stores on the addresses of the XIVE IPI interrupt ESB pages.

Both KVM devices share the same internal structure caching information on the interrupts, among which the xive_irq_data struct containing the addresses of the IPI ESB pages and an extra one in case of passthrough. The later contains the addresses of the ESB pages of the underlying HW controller interrupts, PHB4 in all cases for now.

A guest, when running in the XICS legacy interrupt mode, lets the KVM XICS-over-XIVE device "handle" interrupt management, that is to perform the loads and stores on the addresses of the ESB pages of the guest interrupts. However, when running in XIVE native exploitation mode, the KVM XIVE native device exposes the interrupt ESB pages to the guest and lets the guest perform directly the loads and stores.

The VMA exposing the ESB pages make use of a custom VM fault handler which role is to populate the VMA with appropriate pages. When a fault occurs, the guest IRQ number is deduced from the offset, and the ESB pages of associated XIVE IPI interrupt are inserted in the VMA (using the internal structure caching information on the interrupts).

Supporting device passthrough in the guest running in XIVE native exploitation mode adds some extra refinements because the ESB pages of a different HW controller (PHB4) need to be exposed to the guest along with the initial IPI ESB pages of the XIVE IC controller. But the overall mechanic is the same.

When the device HW irqs are mapped into or unmapped from the guest IRQ number space, the passthru_irq helpers, kvmppc_xive_set_mapped() and kvmppc_xive_clr_mapped(), are called to record or clear the passthrough interrupt information and to perform the switch.

The approach taken by this patch is to clear the ESB pages of the guest IRQ number being mapped and let the VM fault handler repopulate. The handler will insert the ESB page corresponding to the HW interrupt of the device being passed-through or the initial IPI ESB page if the device is being removed.

---
 arch/powerpc/kvm/book3s_xive.h             |  9 +++++
 arch/powerpc/kvm/book3s_xive.c             | 15 ++++++++
 arch/powerpc/kvm/book3s_xive_native.c      | 41 ++++++++++++++++++++++
 Documentation/virtual/kvm/devices/xive.txt | 15 ++++++++
 4 files changed, 80 insertions(+)
arch/powerpc/kvm/book3s_xive.h +9 -0
index 6660d138c6b7..d1f832a53811 100644
@@ -94,6 +94,11 @@ struct kvmppc_xive_src_block {
 	struct kvmppc_xive_irq_state irq_state[KVMPPC_XICS_IRQ_PER_ICS];
 };
 
+struct kvmppc_xive;
+
+struct kvmppc_xive_ops {
+	int (*reset_mapped)(struct kvm *kvm, unsigned long guest_irq);
+};
 
 struct kvmppc_xive {
 	struct kvm *kvm;
@@ -132,6 +137,10 @@ struct kvmppc_xive {
 
 	/* Flags */
 	u8	single_escalation;
+
+	struct kvmppc_xive_ops *ops;
+	struct address_space   *mapping;
+	struct mutex mapping_lock;
 };
 
 #define KVMPPC_XIVE_Q_COUNT	8
arch/powerpc/kvm/book3s_xive.c +15 -0
index 7431e31bc541..7a14512b8944 100644
@@ -942,6 +942,13 @@ int kvmppc_xive_set_mapped(struct kvm *kvm, unsigned long guest_irq,
 	/* Turn the IPI hard off */
 	xive_vm_esb_load(&state->ipi_data, XIVE_ESB_SET_PQ_01);
 
+	/*
+	 * Reset ESB guest mapping. Needed when ESB pages are exposed
+	 * to the guest in XIVE native mode
+	 */
+	if (xive->ops && xive->ops->reset_mapped)
+		xive->ops->reset_mapped(kvm, guest_irq);
+
 	/* Grab info about irq */
 	state->pt_number = hw_irq;
 	state->pt_data = irq_data_get_irq_handler_data(host_data);
@@ -1027,6 +1034,14 @@ int kvmppc_xive_clr_mapped(struct kvm *kvm, unsigned long guest_irq,
 	state->pt_number = 0;
 	state->pt_data = NULL;
 
+	/*
+	 * Reset ESB guest mapping. Needed when ESB pages are exposed
+	 * to the guest in XIVE native mode
+	 */
+	if (xive->ops && xive->ops->reset_mapped) {
+		xive->ops->reset_mapped(kvm, guest_irq);
+	}
+
 	/* Reconfigure the IPI */
 	xive_native_configure_irq(state->ipi_number,
 				  xive_vp(xive, state->act_server),
arch/powerpc/kvm/book3s_xive_native.c +41 -0
index 92cab6409e8e..bf60870144f1 100644
@@ -14,6 +14,7 @@
 #include <linux/delay.h>
 #include <linux/percpu.h>
 #include <linux/cpumask.h>
+#include <linux/file.h>
 #include <asm/uaccess.h>
 #include <asm/kvm_book3s.h>
 #include <asm/kvm_ppc.h>
@@ -176,6 +177,35 @@ int kvmppc_xive_native_connect_vcpu(struct kvm_device *dev,
 	return rc;
 }
 
+/*
+ * Device passthrough support
+ */
+static int kvmppc_xive_native_reset_mapped(struct kvm *kvm, unsigned long irq)
+{
+	struct kvmppc_xive *xive = kvm->arch.xive;
+
+	if (irq >= KVMPPC_XIVE_NR_IRQS)
+		return -EINVAL;
+
+	/*
+	 * Clear the ESB pages of the IRQ number being mapped (or
+	 * unmapped) into the guest and let the the VM fault handler
+	 * repopulate with the appropriate ESB pages (device or IC)
+	 */
+	pr_debug("clearing esb pages for girq 0x%lx\n", irq);
+	mutex_lock(&xive->mapping_lock);
+	if (xive->mapping)
+		unmap_mapping_range(xive->mapping,
+				    irq * (2ull << PAGE_SHIFT),
+				    2ull << PAGE_SHIFT, 1);
+	mutex_unlock(&xive->mapping_lock);
+	return 0;
+}
+
+static struct kvmppc_xive_ops kvmppc_xive_native_ops =  {
+	.reset_mapped = kvmppc_xive_native_reset_mapped,
+};
+
 static int xive_native_esb_fault(struct vm_fault *vmf)
 {
 	struct vm_area_struct *vma = vmf->vma;
@@ -253,6 +283,8 @@ static const struct vm_operations_struct xive_native_tima_vmops = {
 static int kvmppc_xive_native_mmap(struct kvm_device *dev,
 				   struct vm_area_struct *vma)
 {
+	struct kvmppc_xive *xive = dev->private;
+
 	/* We only allow mappings at fixed offset for now */
 	if (vma->vm_pgoff == KVM_XIVE_TIMA_PAGE_OFFSET) {
 		if (vma_pages(vma) > 4)
@@ -268,6 +300,13 @@ static int kvmppc_xive_native_mmap(struct kvm_device *dev,
 
 	vma->vm_flags |= VM_IO | VM_PFNMAP;
 	vma->vm_page_prot = pgprot_noncached_wc(vma->vm_page_prot);
+
+	/*
+	 * Grab the KVM device file address_space to be able to clear
+	 * the ESB pages mapping when a device is passed-through into
+	 * the guest.
+	 */
+	xive->mapping = vma->vm_file->f_mapping;
 	return 0;
 }
 
@@ -913,6 +952,7 @@ static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type)
 	xive->dev = dev;
 	xive->kvm = kvm;
 	kvm->arch.xive = xive;
+	mutex_init(&xive->mapping_lock);
 
 	/* We use the default queue size set by the host */
 	xive->q_order = xive_native_default_eq_shift();
@@ -933,6 +973,7 @@ static int kvmppc_xive_native_create(struct kvm_device *dev, u32 type)
 		ret = -ENOMEM;
 
 	xive->single_escalation = xive_native_has_single_escalation();
+	xive->ops = &kvmppc_xive_native_ops;
 
 	if (ret)
 		kfree(xive);
Documentation/virtual/kvm/devices/xive.txt +4 -0
index be5000b2eb5a..7a242cb07e7c 100644
@@ -43,6 +43,21 @@ the legacy interrupt mode, referred as XICS (POWER7/8).
   manage the source: to trigger, to EOI, to turn off the source for
   instance.
 
+  3. Device passthrough
+
+  When a device is passed-through into the guest, the source
+  interrupts are from a different HW controller (PHB4) and the ESB

Documentation/virtual/kvm/devices/xive.txt the legacy interrupt mode, referred as XICS (POWER7/8). · line 59

Quoted 10 lines
Documentation/virtual/kvm/devices/xive.txt +10 -0
+  pages exposed to the guest should accommadate this change.
+
+  The passthru_irq helpers, kvmppc_xive_set_mapped() and
+  kvmppc_xive_clr_mapped() are called when the device HW irqs are
+  mapped into or unmapped from the guest IRQ number space. The KVM
+  device extends these helpers to clear the ESB pages of the guest IRQ
+  number being mapped and then lets the VM fault handler repopulate.
+  The handler will insert the ESB page corresponding to the HW
+  interrupt of the device being passed-through or the initial IPI ESB
+  page if the device has being removed.

I think it might be worth emphasizing that this all happens with KVM and userspace / the guest doesn't need to do anything about this remapping. Really this is an informational aside, not something a user of the device actually needs to know.

Quoted 3 lines
Documentation/virtual/kvm/devices/xive.txt +0 -0
 * Groups:
 
   1. KVM_DEV_XIVE_GRP_CTRL
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#63 lore raw

PATCH v2 15/16 KVM: introduce a KVM_DESTROY_DEVICE ioctl

In reply to Cédric Le Goater

The 'destroy' method is currently used to destroy all devices when the VM is destroyed after the vCPUs have been freed.

This new KVM ioctl exposes the same KVM device method. It acts as a software reset of the VM to 'destroy' selected devices when necessary and perform the required cleanups on the vCPUs. Called with the kvm->lock.

The 'destroy' method could be improved by returning an error code.

---
 include/uapi/linux/kvm.h          |  7 ++++++
 virt/kvm/kvm_main.c               | 38 +++++++++++++++++++++++++++++++
 Documentation/virtual/kvm/api.txt | 19 ++++++++++++++++
 3 files changed, 64 insertions(+)
include/uapi/linux/kvm.h +7 -0
index 52bf74a1616e..d78fafa54274 100644
@@ -1183,6 +1183,11 @@ struct kvm_create_device {
 	__u32	flags;	/* in: KVM_CREATE_DEVICE_xxx */
 };
 
+struct kvm_destroy_device {
+	__u32	fd;	/* in: device handle */
+	__u32	flags;	/* in: unused */
+};
+
 struct kvm_device_attr {
 	__u32	flags;		/* no flags currently defined */
 	__u32	group;		/* device-defined */
@@ -1331,6 +1336,8 @@ struct kvm_s390_ucas_mapping {
 #define KVM_GET_DEVICE_ATTR	  _IOW(KVMIO,  0xe2, struct kvm_device_attr)
 #define KVM_HAS_DEVICE_ATTR	  _IOW(KVMIO,  0xe3, struct kvm_device_attr)
 
+#define KVM_DESTROY_DEVICE	  _IOWR(KVMIO,  0xf0, struct kvm_destroy_device)
+
 /*
  * ioctls for vcpu fds
  */
virt/kvm/kvm_main.c +38 -0
index 84717d8cb5e4..983d5c37f710 100644
@@ -3026,6 +3026,30 @@ static int kvm_ioctl_create_device(struct kvm *kvm,
 	return 0;
 }
 
+static int kvm_ioctl_destroy_device(struct kvm *kvm,
+				    struct kvm_destroy_device *dd)
+{
+	struct fd f;
+	struct kvm_device *dev;
+
+	f = fdget(dd->fd);
+	if (!f.file)
+		return -EBADF;
+
+	dev = kvm_device_from_filp(f.file);
+	fdput(f);
+
+	if (!dev)
+		return -ENODEV;
+
+	mutex_lock(&kvm->lock);
+	list_del(&dev->vm_node);
+	dev->ops->destroy(dev);
+	mutex_unlock(&kvm->lock);
+
+	return 0;
+}
+
 static long kvm_vm_ioctl_check_extension_generic(struct kvm *kvm, long arg)
 {
 	switch (arg) {
@@ -3270,6 +3294,20 @@ static long kvm_vm_ioctl(struct file *filp,
 		r = 0;
 		break;
 	}
+	case KVM_DESTROY_DEVICE: {
+		struct kvm_destroy_device dd;
+
+		r = -EFAULT;
+		if (copy_from_user(&dd, argp, sizeof(dd)))
+			goto out;
+
+		r = kvm_ioctl_destroy_device(kvm, &dd);
+		if (r)
+			goto out;
+
+		r = 0;
+		break;
+	}
 	case KVM_CHECK_EXTENSION:
 		r = kvm_vm_ioctl_check_extension_generic(kvm, arg);
 		break;
Documentation/virtual/kvm/api.txt +19 -0
index 1db1435769b4..c2ba18279298 100644
@@ -3857,6 +3857,25 @@ number of valid entries in the 'entries' array, which is then filled.
 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
 userspace should not expect to get any particular value there.
 
+4.119 KVM_DESTROY_DEVICE
+
+Capability: KVM_CAP_DEVICE_CTRL
+Type: vm ioctl
+Parameters: struct kvm_destroy_device (in)
+Returns: 0 on success, -1 on error
+Errors:
+  ENODEV: The device type is unknown or unsupported
+
+  Other error conditions may be defined by individual device types or
+  have their standard meanings.
+
+Destroys an emulated device in the kernel.
+
+struct kvm_destroy_device {
+	__u32	fd;	/* in: device handle */
+	__u32	flags;	/* unused */
+};
+
 5. The kvm_run structure
 ------------------------
 
Signature · 1 line
2.20.1
#64 lore raw

Re: [PATCH v2 15/16] KVM: introduce a KVM_DESTROY_DEVICE ioctl

Quoted 11 lines · Cédric Le Goater

The 'destroy' method is currently used to destroy all devices when the VM is destroyed after the vCPUs have been freed.

This new KVM ioctl exposes the same KVM device method. It acts as a software reset of the VM to 'destroy' selected devices when necessary and perform the required cleanups on the vCPUs. Called with the kvm->lock.

The 'destroy' method could be improved by returning an error code.

Again, has this been discussed with Paolo and/or other KVM core people?

Quoted 44 lines
---
 include/uapi/linux/kvm.h          |  7 ++++++
 virt/kvm/kvm_main.c               | 38 +++++++++++++++++++++++++++++++
 Documentation/virtual/kvm/api.txt | 19 ++++++++++++++++
 3 files changed, 64 insertions(+)
include/uapi/linux/kvm.h +7 -0
index 52bf74a1616e..d78fafa54274 100644
@@ -1183,6 +1183,11 @@ struct kvm_create_device {
 	__u32	flags;	/* in: KVM_CREATE_DEVICE_xxx */
 };
 
+struct kvm_destroy_device {
+	__u32	fd;	/* in: device handle */
+	__u32	flags;	/* in: unused */
+};
+
 struct kvm_device_attr {
 	__u32	flags;		/* no flags currently defined */
 	__u32	group;		/* device-defined */
@@ -1331,6 +1336,8 @@ struct kvm_s390_ucas_mapping {
 #define KVM_GET_DEVICE_ATTR	  _IOW(KVMIO,  0xe2, struct kvm_device_attr)
 #define KVM_HAS_DEVICE_ATTR	  _IOW(KVMIO,  0xe3, struct kvm_device_attr)
 
+#define KVM_DESTROY_DEVICE	  _IOWR(KVMIO,  0xf0, struct kvm_destroy_device)
+
 /*
  * ioctls for vcpu fds
  */
virt/kvm/kvm_main.c +5 -0
index 84717d8cb5e4..983d5c37f710 100644
@@ -3026,6 +3026,30 @@ static int kvm_ioctl_create_device(struct kvm *kvm,
 	return 0;
 }
 
+static int kvm_ioctl_destroy_device(struct kvm *kvm,
+				    struct kvm_destroy_device *dd)
+{
+	struct fd f;
+	struct kvm_device *dev;

virt/kvm/kvm_main.c static int kvm_ioctl_create_device(struct kvm *kvm, · line 3043

Quoted 10 lines
virt/kvm/kvm_main.c +10 -0
+
+	f = fdget(dd->fd);
+	if (!f.file)
+		return -EBADF;
+
+	dev = kvm_device_from_filp(f.file);
+	fdput(f);
+
+	if (!dev)
+		return -ENODEV;

Don't you need to verify that the device belongs to this KVM instance?

Quoted 62 lines
virt/kvm/kvm_main.c +22 -0
+	mutex_lock(&kvm->lock);
+	list_del(&dev->vm_node);
+	dev->ops->destroy(dev);
+	mutex_unlock(&kvm->lock);
+
+	return 0;
+}
+
 static long kvm_vm_ioctl_check_extension_generic(struct kvm *kvm, long arg)
 {
 	switch (arg) {
@@ -3270,6 +3294,20 @@ static long kvm_vm_ioctl(struct file *filp,
 		r = 0;
 		break;
 	}
+	case KVM_DESTROY_DEVICE: {
+		struct kvm_destroy_device dd;
+
+		r = -EFAULT;
+		if (copy_from_user(&dd, argp, sizeof(dd)))
+			goto out;
+
+		r = kvm_ioctl_destroy_device(kvm, &dd);
+		if (r)
+			goto out;
+
+		r = 0;
+		break;
+	}
 	case KVM_CHECK_EXTENSION:
 		r = kvm_vm_ioctl_check_extension_generic(kvm, arg);
 		break;
Documentation/virtual/kvm/api.txt +19 -0
index 1db1435769b4..c2ba18279298 100644
@@ -3857,6 +3857,25 @@ number of valid entries in the 'entries' array, which is then filled.
 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
 userspace should not expect to get any particular value there.
 
+4.119 KVM_DESTROY_DEVICE
+
+Capability: KVM_CAP_DEVICE_CTRL
+Type: vm ioctl
+Parameters: struct kvm_destroy_device (in)
+Returns: 0 on success, -1 on error
+Errors:
+  ENODEV: The device type is unknown or unsupported
+
+  Other error conditions may be defined by individual device types or
+  have their standard meanings.
+
+Destroys an emulated device in the kernel.
+
+struct kvm_destroy_device {
+	__u32	fd;	/* in: device handle */
+	__u32	flags;	/* unused */
+};
+
 5. The kvm_run structure
 ------------------------
 
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#65 lore raw

Re: [PATCH v2 15/16] KVM: introduce a KVM_DESTROY_DEVICE ioctl

Quoted 9 lines · David Gibson
Quoted 8 lines · Cédric Le Goater

The 'destroy' method is currently used to destroy all devices when the VM is destroyed after the vCPUs have been freed.

This new KVM ioctl exposes the same KVM device method. It acts as a software reset of the VM to 'destroy' selected devices when necessary and perform the required cleanups on the vCPUs. Called with the kvm->lock.

Quoted 6 lines
Quoted 3 lines

The 'destroy' method could be improved by returning an error code.

Again, has this been discussed with Paolo and/or other KVM core people?

Not yet. Adding Paolo for feedback.

Quoted 48 lines
Quoted 48 lines
---
 include/uapi/linux/kvm.h          |  7 ++++++
 virt/kvm/kvm_main.c               | 38 +++++++++++++++++++++++++++++++
 Documentation/virtual/kvm/api.txt | 19 ++++++++++++++++
 3 files changed, 64 insertions(+)
include/uapi/linux/kvm.h +7 -0
index 52bf74a1616e..d78fafa54274 100644
@@ -1183,6 +1183,11 @@ struct kvm_create_device {
 	__u32	flags;	/* in: KVM_CREATE_DEVICE_xxx */
 };
 
+struct kvm_destroy_device {
+	__u32	fd;	/* in: device handle */
+	__u32	flags;	/* in: unused */
+};
+
 struct kvm_device_attr {
 	__u32	flags;		/* no flags currently defined */
 	__u32	group;		/* device-defined */
@@ -1331,6 +1336,8 @@ struct kvm_s390_ucas_mapping {
 #define KVM_GET_DEVICE_ATTR	  _IOW(KVMIO,  0xe2, struct kvm_device_attr)
 #define KVM_HAS_DEVICE_ATTR	  _IOW(KVMIO,  0xe3, struct kvm_device_attr)
 
+#define KVM_DESTROY_DEVICE	  _IOWR(KVMIO,  0xf0, struct kvm_destroy_device)
+
 /*
  * ioctls for vcpu fds
  */
virt/kvm/kvm_main.c +9 -0
index 84717d8cb5e4..983d5c37f710 100644
@@ -3026,6 +3026,30 @@ static int kvm_ioctl_create_device(struct kvm *kvm,
 	return 0;
 }
 
+static int kvm_ioctl_destroy_device(struct kvm *kvm,
+				    struct kvm_destroy_device *dd)
+{
+	struct fd f;
+	struct kvm_device *dev;
+
+	f = fdget(dd->fd);
+	if (!f.file)
+		return -EBADF;
Quoted 8 lines

virt/kvm/kvm_main.c static int kvm_ioctl_create_device(struct kvm *kvm, · line 3043

Quoted 6 lines
virt/kvm/kvm_main.c +6 -0
+
+	dev = kvm_device_from_filp(f.file);
+	fdput(f);
+
+	if (!dev)
+		return -ENODEV;

Don't you need to verify that the device belongs to this KVM instance?

ah yes. I should at least check : dev->kvm == kvm

Quoted 4 lines

virt/kvm/kvm_main.c static int kvm_ioctl_create_device(struct kvm *kvm, · line 3048

Quoted 4 lines
virt/kvm/kvm_main.c +4 -0
+	mutex_lock(&kvm->lock);
+	list_del(&dev->vm_node);
+	dev->ops->destroy(dev);
+	mutex_unlock(&kvm->lock);

And there, I should problably drop a ref count on the VM. I am not sure of that.

Thanks,

C.

Quoted 58 lines
Quoted 57 lines
virt/kvm/kvm_main.c +17 -0
+	return 0;
+}
+
 static long kvm_vm_ioctl_check_extension_generic(struct kvm *kvm, long arg)
 {
 	switch (arg) {
@@ -3270,6 +3294,20 @@ static long kvm_vm_ioctl(struct file *filp,
 		r = 0;
 		break;
 	}
+	case KVM_DESTROY_DEVICE: {
+		struct kvm_destroy_device dd;
+
+		r = -EFAULT;
+		if (copy_from_user(&dd, argp, sizeof(dd)))
+			goto out;
+
+		r = kvm_ioctl_destroy_device(kvm, &dd);
+		if (r)
+			goto out;
+
+		r = 0;
+		break;
+	}
 	case KVM_CHECK_EXTENSION:
 		r = kvm_vm_ioctl_check_extension_generic(kvm, arg);
 		break;
Documentation/virtual/kvm/api.txt +19 -0
index 1db1435769b4..c2ba18279298 100644
@@ -3857,6 +3857,25 @@ number of valid entries in the 'entries' array, which is then filled.
 'index' and 'flags' fields in 'struct kvm_cpuid_entry2' are currently reserved,
 userspace should not expect to get any particular value there.
 
+4.119 KVM_DESTROY_DEVICE
+
+Capability: KVM_CAP_DEVICE_CTRL
+Type: vm ioctl
+Parameters: struct kvm_destroy_device (in)
+Returns: 0 on success, -1 on error
+Errors:
+  ENODEV: The device type is unknown or unsupported
+
+  Other error conditions may be defined by individual device types or
+  have their standard meanings.
+
+Destroys an emulated device in the kernel.
+
+struct kvm_destroy_device {
+	__u32	fd;	/* in: device handle */
+	__u32	flags;	/* unused */
+};
+
 5. The kvm_run structure
 ------------------------
 
maintainer
#66 lore raw

Re: [PATCH v2 15/16] KVM: introduce a KVM_DESTROY_DEVICE ioctl

Quoted 11 lines · Cédric Le Goater

The 'destroy' method is currently used to destroy all devices when the VM is destroyed after the vCPUs have been freed.

This new KVM ioctl exposes the same KVM device method. It acts as a software reset of the VM to 'destroy' selected devices when necessary and perform the required cleanups on the vCPUs. Called with the kvm->lock.

The 'destroy' method could be improved by returning an error code.

Looks good to me.

Paolo

#67 lore raw

PATCH v2 16/16 KVM: PPC: Book3S HV: XIVE: clear the vCPU interrupt presenters

In reply to Cédric Le Goater

When the VM boots, the CAS negotiation process determines which interrupt mode to use and invokes a machine reset. At that time, the previous KVM interrupt device is 'destroyed' before the chosen one is created. Upon destruction, the vCPU interrupt presenters using the KVM device should be cleared first, the machine will reconnect them later to the new device after it is created.

When using the KVM device, there is still a race window with the early checks in kvmppc_native_connect_vcpu(). Yet to be fixed.

---
 arch/powerpc/kvm/book3s_xics.c        | 19 +++++++++++++
 arch/powerpc/kvm/book3s_xive.c        | 39 +++++++++++++++++++++++++--
 arch/powerpc/kvm/book3s_xive_native.c | 16 +++++++++++
 3 files changed, 72 insertions(+), 2 deletions(-)
arch/powerpc/kvm/book3s_xics.c +19 -0
index f27ee57ab46e..81cdabf4295f 100644
@@ -1342,6 +1342,25 @@ static void kvmppc_xics_free(struct kvm_device *dev)
 	struct kvmppc_xics *xics = dev->private;
 	int i;
 	struct kvm *kvm = xics->kvm;
+	struct kvm_vcpu *vcpu;
+
+	/*
+	 * When destroying the VM, the vCPUs are destroyed first and
+	 * the vCPU list should be empty. If this is not the case,
+	 * then we are simply destroying the device and we should
+	 * clean up the vCPU interrupt presenters first.
+	 */
+	if (atomic_read(&kvm->online_vcpus) != 0) {
+		/*
+		 * call kick_all_cpus_sync() to ensure that all CPUs
+		 * have executed any pending interrupts
+		 */
+		if (is_kvmppc_hv_enabled(kvm))
+			kick_all_cpus_sync();
+
+		kvm_for_each_vcpu(i, vcpu, kvm)
+			kvmppc_xics_free_icp(vcpu);
+	}
 
 	debugfs_remove(xics->dentry);
 
arch/powerpc/kvm/book3s_xive.c +37 -2
index 7a14512b8944..0a1c11d6881c 100644
@@ -1105,11 +1105,19 @@ void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu)
 void kvmppc_xive_cleanup_vcpu(struct kvm_vcpu *vcpu)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
-	struct kvmppc_xive *xive = xc->xive;
+	struct kvmppc_xive *xive;
 	int i;
 
+	if (!kvmppc_xics_enabled(vcpu))
+		return;
+
+	if (!xc)
+		return;
+
 	pr_devel("cleanup_vcpu(cpu=%d)\n", xc->server_num);
 
+	xive = xc->xive;
+
 	/* Ensure no interrupt is still routed to that VP */
 	xc->valid = false;
 	kvmppc_xive_disable_vcpu_interrupts(vcpu);
@@ -1146,6 +1154,10 @@ void kvmppc_xive_cleanup_vcpu(struct kvm_vcpu *vcpu)
 	}
 	/* Free the VP */
 	kfree(xc);
+
+	/* Cleanup the vcpu */
+	vcpu->arch.irq_type = KVMPPC_IRQ_DEFAULT;
+	vcpu->arch.xive_vcpu = NULL;
 }
 
 int kvmppc_xive_connect_vcpu(struct kvm_device *dev,
@@ -1163,7 +1175,7 @@ int kvmppc_xive_connect_vcpu(struct kvm_device *dev,
 	}
 	if (xive->kvm != vcpu->kvm)
 		return -EPERM;
-	if (vcpu->arch.irq_type)
+	if (vcpu->arch.irq_type != KVMPPC_IRQ_DEFAULT)
 		return -EBUSY;
 	if (kvmppc_xive_find_server(vcpu->kvm, cpu)) {
 		pr_devel("Duplicate !\n");
@@ -1833,8 +1845,31 @@ static void kvmppc_xive_free(struct kvm_device *dev)
 {
 	struct kvmppc_xive *xive = dev->private;
 	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
 	int i;
 
+	/*
+	 * When destroying the VM, the vCPUs are destroyed first and
+	 * the vCPU list should be empty. If this is not the case,
+	 * then we are simply destroying the device and we should
+	 * clean up the vCPU interrupt presenters first.
+	 */
+	if (atomic_read(&kvm->online_vcpus) != 0) {
+		/*
+		 * call kick_all_cpus_sync() to ensure that all CPUs
+		 * have executed any pending interrupts
+		 */
+		if (is_kvmppc_hv_enabled(kvm))
+			kick_all_cpus_sync();
+
+		/*
+		 * TODO: There is still a race window with the early
+		 * checks in kvmppc_native_connect_vcpu()
+		 */
+		kvm_for_each_vcpu(i, vcpu, kvm)
+			kvmppc_xive_cleanup_vcpu(vcpu);
+	}
+
 	debugfs_remove(xive->dentry);
 
 	if (kvm)
arch/powerpc/kvm/book3s_xive_native.c +16 -0
index bf60870144f1..c0655164d9af 100644
@@ -909,8 +909,24 @@ static void kvmppc_xive_native_free(struct kvm_device *dev)
 {
 	struct kvmppc_xive *xive = dev->private;
 	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
 	int i;
 
+	/*
+	 * When destroying the VM, the vCPUs are destroyed first and
+	 * the vCPU list should be empty. If this is not the case,
+	 * then we are simply destroying the device and we should
+	 * clean up the vCPU interrupt presenters first.
+	 */
+	if (atomic_read(&kvm->online_vcpus) != 0) {
+		/*
+		 * TODO: There is still a race window with the early
+		 * checks in kvmppc_xive_native_connect_vcpu()
+		 */
+		kvm_for_each_vcpu(i, vcpu, kvm)
+			kvmppc_xive_native_cleanup_vcpu(vcpu);
+	}
+
 	debugfs_remove(xive->dentry);
 
 	pr_devel("Destroying xive native device\n");
Signature · 1 line
2.20.1
#68 lore raw

Re: [PATCH v2 16/16] KVM: PPC: Book3S HV: XIVE: clear the vCPU interrupt presenters

Quoted 52 lines · Cédric Le Goater

When the VM boots, the CAS negotiation process determines which interrupt mode to use and invokes a machine reset. At that time, the previous KVM interrupt device is 'destroyed' before the chosen one is created. Upon destruction, the vCPU interrupt presenters using the KVM device should be cleared first, the machine will reconnect them later to the new device after it is created.

When using the KVM device, there is still a race window with the early checks in kvmppc_native_connect_vcpu(). Yet to be fixed.

---
 arch/powerpc/kvm/book3s_xics.c        | 19 +++++++++++++
 arch/powerpc/kvm/book3s_xive.c        | 39 +++++++++++++++++++++++++--
 arch/powerpc/kvm/book3s_xive_native.c | 16 +++++++++++
 3 files changed, 72 insertions(+), 2 deletions(-)
arch/powerpc/kvm/book3s_xics.c +19 -0
index f27ee57ab46e..81cdabf4295f 100644
@@ -1342,6 +1342,25 @@ static void kvmppc_xics_free(struct kvm_device *dev)
 	struct kvmppc_xics *xics = dev->private;
 	int i;
 	struct kvm *kvm = xics->kvm;
+	struct kvm_vcpu *vcpu;
+
+	/*
+	 * When destroying the VM, the vCPUs are destroyed first and
+	 * the vCPU list should be empty. If this is not the case,
+	 * then we are simply destroying the device and we should
+	 * clean up the vCPU interrupt presenters first.
+	 */
+	if (atomic_read(&kvm->online_vcpus) != 0) {
+		/*
+		 * call kick_all_cpus_sync() to ensure that all CPUs
+		 * have executed any pending interrupts
+		 */
+		if (is_kvmppc_hv_enabled(kvm))
+			kick_all_cpus_sync();
+
+		kvm_for_each_vcpu(i, vcpu, kvm)
+			kvmppc_xics_free_icp(vcpu);
+	}
 
 	debugfs_remove(xics->dentry);
 
arch/powerpc/kvm/book3s_xive.c +0 -0
index 7a14512b8944..0a1c11d6881c 100644

arch/powerpc/kvm/book3s_xive.c void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu) · line 1111

Quoted 8 lines
arch/powerpc/kvm/book3s_xive.c +2 -1
@@ -1105,11 +1105,19 @@ void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu)
 void kvmppc_xive_cleanup_vcpu(struct kvm_vcpu *vcpu)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
-	struct kvmppc_xive *xive = xc->xive;
+	struct kvmppc_xive *xive;
 	int i;
 
+	if (!kvmppc_xics_enabled(vcpu))

This should be kvmppc_xive_enabled(), no?

Quoted 93 lines
arch/powerpc/kvm/book3s_xive.c +35 -1
+		return;
+
+	if (!xc)
+		return;
+
 	pr_devel("cleanup_vcpu(cpu=%d)\n", xc->server_num);
 
+	xive = xc->xive;
+
 	/* Ensure no interrupt is still routed to that VP */
 	xc->valid = false;
 	kvmppc_xive_disable_vcpu_interrupts(vcpu);
@@ -1146,6 +1154,10 @@ void kvmppc_xive_cleanup_vcpu(struct kvm_vcpu *vcpu)
 	}
 	/* Free the VP */
 	kfree(xc);
+
+	/* Cleanup the vcpu */
+	vcpu->arch.irq_type = KVMPPC_IRQ_DEFAULT;
+	vcpu->arch.xive_vcpu = NULL;
 }
 
 int kvmppc_xive_connect_vcpu(struct kvm_device *dev,
@@ -1163,7 +1175,7 @@ int kvmppc_xive_connect_vcpu(struct kvm_device *dev,
 	}
 	if (xive->kvm != vcpu->kvm)
 		return -EPERM;
-	if (vcpu->arch.irq_type)
+	if (vcpu->arch.irq_type != KVMPPC_IRQ_DEFAULT)
 		return -EBUSY;
 	if (kvmppc_xive_find_server(vcpu->kvm, cpu)) {
 		pr_devel("Duplicate !\n");
@@ -1833,8 +1845,31 @@ static void kvmppc_xive_free(struct kvm_device *dev)
 {
 	struct kvmppc_xive *xive = dev->private;
 	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
 	int i;
 
+	/*
+	 * When destroying the VM, the vCPUs are destroyed first and
+	 * the vCPU list should be empty. If this is not the case,
+	 * then we are simply destroying the device and we should
+	 * clean up the vCPU interrupt presenters first.
+	 */
+	if (atomic_read(&kvm->online_vcpus) != 0) {
+		/*
+		 * call kick_all_cpus_sync() to ensure that all CPUs
+		 * have executed any pending interrupts
+		 */
+		if (is_kvmppc_hv_enabled(kvm))
+			kick_all_cpus_sync();
+
+		/*
+		 * TODO: There is still a race window with the early
+		 * checks in kvmppc_native_connect_vcpu()
+		 */
+		kvm_for_each_vcpu(i, vcpu, kvm)
+			kvmppc_xive_cleanup_vcpu(vcpu);
+	}
+
 	debugfs_remove(xive->dentry);
 
 	if (kvm)
arch/powerpc/kvm/book3s_xive_native.c +16 -0
index bf60870144f1..c0655164d9af 100644
@@ -909,8 +909,24 @@ static void kvmppc_xive_native_free(struct kvm_device *dev)
 {
 	struct kvmppc_xive *xive = dev->private;
 	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
 	int i;
 
+	/*
+	 * When destroying the VM, the vCPUs are destroyed first and
+	 * the vCPU list should be empty. If this is not the case,
+	 * then we are simply destroying the device and we should
+	 * clean up the vCPU interrupt presenters first.
+	 */
+	if (atomic_read(&kvm->online_vcpus) != 0) {
+		/*
+		 * TODO: There is still a race window with the early
+		 * checks in kvmppc_xive_native_connect_vcpu()
+		 */
+		kvm_for_each_vcpu(i, vcpu, kvm)
+			kvmppc_xive_native_cleanup_vcpu(vcpu);
+	}
+
 	debugfs_remove(xive->dentry);
 
 	pr_devel("Destroying xive native device\n");
Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson
#69 lore raw

Re: [PATCH v2 16/16] KVM: PPC: Book3S HV: XIVE: clear the vCPU interrupt presenters

Quoted 60 lines · David Gibson
Quoted 59 lines · Cédric Le Goater

When the VM boots, the CAS negotiation process determines which interrupt mode to use and invokes a machine reset. At that time, the previous KVM interrupt device is 'destroyed' before the chosen one is created. Upon destruction, the vCPU interrupt presenters using the KVM device should be cleared first, the machine will reconnect them later to the new device after it is created.

When using the KVM device, there is still a race window with the early checks in kvmppc_native_connect_vcpu(). Yet to be fixed.

---
 arch/powerpc/kvm/book3s_xics.c        | 19 +++++++++++++
 arch/powerpc/kvm/book3s_xive.c        | 39 +++++++++++++++++++++++++--
 arch/powerpc/kvm/book3s_xive_native.c | 16 +++++++++++
 3 files changed, 72 insertions(+), 2 deletions(-)
arch/powerpc/kvm/book3s_xics.c +19 -0
index f27ee57ab46e..81cdabf4295f 100644
@@ -1342,6 +1342,25 @@ static void kvmppc_xics_free(struct kvm_device *dev)
 	struct kvmppc_xics *xics = dev->private;
 	int i;
 	struct kvm *kvm = xics->kvm;
+	struct kvm_vcpu *vcpu;
+
+	/*
+	 * When destroying the VM, the vCPUs are destroyed first and
+	 * the vCPU list should be empty. If this is not the case,
+	 * then we are simply destroying the device and we should
+	 * clean up the vCPU interrupt presenters first.
+	 */
+	if (atomic_read(&kvm->online_vcpus) != 0) {
+		/*
+		 * call kick_all_cpus_sync() to ensure that all CPUs
+		 * have executed any pending interrupts
+		 */
+		if (is_kvmppc_hv_enabled(kvm))
+			kick_all_cpus_sync();
+
+		kvm_for_each_vcpu(i, vcpu, kvm)
+			kvmppc_xics_free_icp(vcpu);
+	}
 
 	debugfs_remove(xics->dentry);
 
arch/powerpc/kvm/book3s_xive.c +1 -1
index 7a14512b8944..0a1c11d6881c 100644
@@ -1105,11 +1105,19 @@ void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu)
 void kvmppc_xive_cleanup_vcpu(struct kvm_vcpu *vcpu)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
-	struct kvmppc_xive *xive = xc->xive;
+	struct kvmppc_xive *xive;
 	int i;
 
Quoted 3 lines

arch/powerpc/kvm/book3s_xive.c void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu) · line 1111

Quoted 1 line
arch/powerpc/kvm/book3s_xive.c +1 -0
+	if (!kvmppc_xics_enabled(vcpu))

This should be kvmppc_xive_enabled(), no?

This is the KVM XICS-on-XIVE device and its IRQ type is KVMPPC_IRQ_XICS. So this is correct :/

May be we should introduce a KVMPPC_IRQ_XICS_ON_XIVE macro to clarify.

Quoted 95 lines
Quoted 94 lines
arch/powerpc/kvm/book3s_xive.c +35 -1
+		return;
+
+	if (!xc)
+		return;
+
 	pr_devel("cleanup_vcpu(cpu=%d)\n", xc->server_num);
 
+	xive = xc->xive;
+
 	/* Ensure no interrupt is still routed to that VP */
 	xc->valid = false;
 	kvmppc_xive_disable_vcpu_interrupts(vcpu);
@@ -1146,6 +1154,10 @@ void kvmppc_xive_cleanup_vcpu(struct kvm_vcpu *vcpu)
 	}
 	/* Free the VP */
 	kfree(xc);
+
+	/* Cleanup the vcpu */
+	vcpu->arch.irq_type = KVMPPC_IRQ_DEFAULT;
+	vcpu->arch.xive_vcpu = NULL;
 }
 
 int kvmppc_xive_connect_vcpu(struct kvm_device *dev,
@@ -1163,7 +1175,7 @@ int kvmppc_xive_connect_vcpu(struct kvm_device *dev,
 	}
 	if (xive->kvm != vcpu->kvm)
 		return -EPERM;
-	if (vcpu->arch.irq_type)
+	if (vcpu->arch.irq_type != KVMPPC_IRQ_DEFAULT)
 		return -EBUSY;
 	if (kvmppc_xive_find_server(vcpu->kvm, cpu)) {
 		pr_devel("Duplicate !\n");
@@ -1833,8 +1845,31 @@ static void kvmppc_xive_free(struct kvm_device *dev)
 {
 	struct kvmppc_xive *xive = dev->private;
 	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
 	int i;
 
+	/*
+	 * When destroying the VM, the vCPUs are destroyed first and
+	 * the vCPU list should be empty. If this is not the case,
+	 * then we are simply destroying the device and we should
+	 * clean up the vCPU interrupt presenters first.
+	 */
+	if (atomic_read(&kvm->online_vcpus) != 0) {
+		/*
+		 * call kick_all_cpus_sync() to ensure that all CPUs
+		 * have executed any pending interrupts
+		 */
+		if (is_kvmppc_hv_enabled(kvm))
+			kick_all_cpus_sync();
+
+		/*
+		 * TODO: There is still a race window with the early
+		 * checks in kvmppc_native_connect_vcpu()
+		 */
+		kvm_for_each_vcpu(i, vcpu, kvm)
+			kvmppc_xive_cleanup_vcpu(vcpu);
+	}
+
 	debugfs_remove(xive->dentry);
 
 	if (kvm)
arch/powerpc/kvm/book3s_xive_native.c +16 -0
index bf60870144f1..c0655164d9af 100644
@@ -909,8 +909,24 @@ static void kvmppc_xive_native_free(struct kvm_device *dev)
 {
 	struct kvmppc_xive *xive = dev->private;
 	struct kvm *kvm = xive->kvm;
+	struct kvm_vcpu *vcpu;
 	int i;
 
+	/*
+	 * When destroying the VM, the vCPUs are destroyed first and
+	 * the vCPU list should be empty. If this is not the case,
+	 * then we are simply destroying the device and we should
+	 * clean up the vCPU interrupt presenters first.
+	 */
+	if (atomic_read(&kvm->online_vcpus) != 0) {
+		/*
+		 * TODO: There is still a race window with the early
+		 * checks in kvmppc_xive_native_connect_vcpu()
+		 */
+		kvm_for_each_vcpu(i, vcpu, kvm)
+			kvmppc_xive_native_cleanup_vcpu(vcpu);
+	}
+
 	debugfs_remove(xive->dentry);
 
 	pr_devel("Destroying xive native device\n");
#70 lore raw

Re: [PATCH v2 16/16] KVM: PPC: Book3S HV: XIVE: clear the vCPU interrupt presenters

Quoted 61 lines · Cédric Le Goater
Quoted 60 lines · David Gibson
Quoted 59 lines · Cédric Le Goater

When the VM boots, the CAS negotiation process determines which interrupt mode to use and invokes a machine reset. At that time, the previous KVM interrupt device is 'destroyed' before the chosen one is created. Upon destruction, the vCPU interrupt presenters using the KVM device should be cleared first, the machine will reconnect them later to the new device after it is created.

When using the KVM device, there is still a race window with the early checks in kvmppc_native_connect_vcpu(). Yet to be fixed.

---
 arch/powerpc/kvm/book3s_xics.c        | 19 +++++++++++++
 arch/powerpc/kvm/book3s_xive.c        | 39 +++++++++++++++++++++++++--
 arch/powerpc/kvm/book3s_xive_native.c | 16 +++++++++++
 3 files changed, 72 insertions(+), 2 deletions(-)
arch/powerpc/kvm/book3s_xics.c +19 -0
index f27ee57ab46e..81cdabf4295f 100644
@@ -1342,6 +1342,25 @@ static void kvmppc_xics_free(struct kvm_device *dev)
 	struct kvmppc_xics *xics = dev->private;
 	int i;
 	struct kvm *kvm = xics->kvm;
+	struct kvm_vcpu *vcpu;
+
+	/*
+	 * When destroying the VM, the vCPUs are destroyed first and
+	 * the vCPU list should be empty. If this is not the case,
+	 * then we are simply destroying the device and we should
+	 * clean up the vCPU interrupt presenters first.
+	 */
+	if (atomic_read(&kvm->online_vcpus) != 0) {
+		/*
+		 * call kick_all_cpus_sync() to ensure that all CPUs
+		 * have executed any pending interrupts
+		 */
+		if (is_kvmppc_hv_enabled(kvm))
+			kick_all_cpus_sync();
+
+		kvm_for_each_vcpu(i, vcpu, kvm)
+			kvmppc_xics_free_icp(vcpu);
+	}
 
 	debugfs_remove(xics->dentry);
 
arch/powerpc/kvm/book3s_xive.c +1 -1
index 7a14512b8944..0a1c11d6881c 100644
@@ -1105,11 +1105,19 @@ void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu)
 void kvmppc_xive_cleanup_vcpu(struct kvm_vcpu *vcpu)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
-	struct kvmppc_xive *xive = xc->xive;
+	struct kvmppc_xive *xive;
 	int i;
 
Quoted 6 lines
Quoted 3 lines

arch/powerpc/kvm/book3s_xive.c void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu) · line 1111

Quoted 1 line
arch/powerpc/kvm/book3s_xive.c +1 -0
+	if (!kvmppc_xics_enabled(vcpu))

This should be kvmppc_xive_enabled(), no?

This is the KVM XICS-on-XIVE device and its IRQ type is KVMPPC_IRQ_XICS. So this is correct :/

Ah, right, sorry.

Quoted 2 lines

May be we should introduce a KVMPPC_IRQ_XICS_ON_XIVE macro to clarify.

Yeah, maybe.

Signature · 4 lines
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson