* [Qemu-devel] [v3 1/6] hw/msi: Allow platform devices to use explicit SID
2016-08-31 16:25 [Qemu-devel] [v3 0/6] AMD IOMMU David Kiarie
@ 2016-08-31 16:25 ` David Kiarie
2016-08-31 16:25 ` [Qemu-devel] [v3 2/6] hw/i386: enforce SID verification David Kiarie
` (5 subsequent siblings)
6 siblings, 0 replies; 8+ messages in thread
From: David Kiarie @ 2016-08-31 16:25 UTC (permalink / raw)
To: qemu-devel
Cc: ehabkost, rkrcmar, peterx, jan.kiszka, mst, valentine.sinitsyn,
David Kiarie
When using IOMMU platform devices like IOAPIC are required to make
interrupt remapping requests using explicit SID.We affiliate an MSI
route with a requester ID and a PCI device if present which ensures
that platform devices can call IOMMU interrupt remapping code with
explicit SID while maintaining compatility with the original code
which mainly dealt with PCI devices.
Signed-off-by: David Kiarie <davidkiarie4@gmail.com>
---
hw/i386/intel_iommu.c | 3 +++
hw/i386/kvm/pci-assign.c | 12 ++++++++----
hw/intc/ioapic.c | 25 +++++++++++++++++++++----
hw/misc/ivshmem.c | 6 ++++--
hw/vfio/pci.c | 6 ++++--
hw/virtio/virtio-pci.c | 7 +++++--
include/hw/i386/ioapic_internal.h | 1 +
include/hw/i386/x86-iommu.h | 1 +
include/sysemu/kvm.h | 25 ++++++++++++++-----------
kvm-all.c | 10 ++++++----
kvm-stub.c | 5 +++--
qemu-version.h | 1 +
target-i386/kvm.c | 15 +++++++++------
13 files changed, 80 insertions(+), 37 deletions(-)
create mode 100644 qemu-version.h
diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c
index d6e02c8..496d836 100644
--- a/hw/i386/intel_iommu.c
+++ b/hw/i386/intel_iommu.c
@@ -2466,6 +2466,9 @@ static void vtd_realize(DeviceState *dev, Error **errp)
vtd_init(s);
sysbus_mmio_map(SYS_BUS_DEVICE(s), 0, Q35_HOST_BRIDGE_IOMMU_ADDR);
pci_setup_iommu(bus, vtd_host_dma_iommu, dev);
+ /* IOMMU expected IOAPIC SID */
+ x86_iommu->ioapic_bdf = PCI_BUILD_BDF(Q35_PSEUDO_DEVFN_IOAPIC,
+ Q35_PSEUDO_DEVFN_IOAPIC);
/* Pseudo address space under root PCI bus. */
pcms->ioapic_as = vtd_host_dma_iommu(bus, s, Q35_PSEUDO_DEVFN_IOAPIC);
diff --git a/hw/i386/kvm/pci-assign.c b/hw/i386/kvm/pci-assign.c
index 8238fbc..3f26be1 100644
--- a/hw/i386/kvm/pci-assign.c
+++ b/hw/i386/kvm/pci-assign.c
@@ -976,7 +976,8 @@ static void assigned_dev_update_msi(PCIDevice *pci_dev)
if (ctrl_byte & PCI_MSI_FLAGS_ENABLE) {
int virq;
- virq = kvm_irqchip_add_msi_route(kvm_state, 0, pci_dev);
+ virq = kvm_irqchip_add_msi_route(kvm_state, 0, pci_dev,
+ pci_requester_id(pci_dev));
if (virq < 0) {
perror("assigned_dev_update_msi: kvm_irqchip_add_msi_route");
return;
@@ -1014,7 +1015,8 @@ static void assigned_dev_update_msi_msg(PCIDevice *pci_dev)
}
kvm_irqchip_update_msi_route(kvm_state, assigned_dev->msi_virq[0],
- msi_get_message(pci_dev, 0), pci_dev);
+ msi_get_message(pci_dev, 0), pci_dev,
+ pci_requester_id(pci_dev));
kvm_irqchip_commit_routes(kvm_state);
}
@@ -1078,7 +1080,8 @@ static int assigned_dev_update_msix_mmio(PCIDevice *pci_dev)
continue;
}
- r = kvm_irqchip_add_msi_route(kvm_state, i, pci_dev);
+ r = kvm_irqchip_add_msi_route(kvm_state, i, pci_dev,
+ pci_requester_id(pci_dev));
if (r < 0) {
return r;
}
@@ -1599,7 +1602,8 @@ static void assigned_dev_msix_mmio_write(void *opaque, hwaddr addr,
ret = kvm_irqchip_update_msi_route(kvm_state,
adev->msi_virq[i], msg,
- pdev);
+ pdev,
+ pci_requester_id(pdev));
if (ret) {
error_report("Error updating irq routing entry (%d)", ret);
}
diff --git a/hw/intc/ioapic.c b/hw/intc/ioapic.c
index 31791b0..b8b2f33 100644
--- a/hw/intc/ioapic.c
+++ b/hw/intc/ioapic.c
@@ -95,9 +95,17 @@ static void ioapic_entry_parse(uint64_t entry, struct ioapic_entry_info *info)
(info->delivery_mode << MSI_DATA_DELIVERY_MODE_SHIFT);
}
-static void ioapic_service(IOAPICCommonState *s)
+static void ioapic_as_write(IOAPICCommonState *s, uint32_t data, uint64_t addr)
{
AddressSpace *ioapic_as = PC_MACHINE(qdev_get_machine())->ioapic_as;
+ MemTxAttrs attrs;
+
+ attrs.requester_id = s->devid;
+ address_space_stl_le(ioapic_as, addr, data, attrs, NULL);
+}
+
+static void ioapic_service(IOAPICCommonState *s)
+{
struct ioapic_entry_info info;
uint8_t i;
uint32_t mask;
@@ -141,7 +149,7 @@ static void ioapic_service(IOAPICCommonState *s)
* the IOAPIC message into a MSI one, and its
* address space will decide whether we need a
* translation. */
- stl_le_phys(ioapic_as, info.addr, info.data);
+ ioapic_as_write(s, info.data, info.addr);
}
}
}
@@ -197,7 +205,7 @@ static void ioapic_update_kvm_routes(IOAPICCommonState *s)
ioapic_entry_parse(s->ioredtbl[i], &info);
msg.address = info.addr;
msg.data = info.data;
- kvm_irqchip_update_msi_route(kvm_state, i, msg, NULL);
+ kvm_irqchip_update_msi_route(kvm_state, i, msg, NULL, s->devid);
}
kvm_irqchip_commit_routes(kvm_state);
}
@@ -385,12 +393,22 @@ static void ioapic_machine_done_notify(Notifier *notifier, void *data)
if (kvm_irqchip_is_split()) {
X86IOMMUState *iommu = x86_iommu_get_default();
+ MSIMessage msg = {0, 0};
+ int i;
+
if (iommu) {
/* Register this IOAPIC with IOMMU IEC notifier, so that
* when there are IR invalidates, we can be notified to
* update kernel IR cache. */
x86_iommu_iec_register_notifier(iommu, ioapic_iec_notifier, s);
+ s->devid = iommu->ioapic_bdf;
+ /* update IOAPIC routes to the right SID */
+ for (i = 0; i < IOAPIC_NUM_PINS; i++) {
+ kvm_irqchip_update_msi_route(kvm_state, i, msg, NULL, s->devid);
+ }
+ kvm_irqchip_commit_routes(kvm_state);
}
+
}
#endif
}
@@ -407,7 +425,6 @@ static void ioapic_realize(DeviceState *dev, Error **errp)
memory_region_init_io(&s->io_memory, OBJECT(s), &ioapic_io_ops, s,
"ioapic", 0x1000);
-
qdev_init_gpio_in(dev, ioapic_set_irq, IOAPIC_NUM_PINS);
ioapics[ioapic_no] = s;
diff --git a/hw/misc/ivshmem.c b/hw/misc/ivshmem.c
index 40a2ebc..9d5133f 100644
--- a/hw/misc/ivshmem.c
+++ b/hw/misc/ivshmem.c
@@ -318,7 +318,8 @@ static int ivshmem_vector_unmask(PCIDevice *dev, unsigned vector,
IVSHMEM_DPRINTF("vector unmask %p %d\n", dev, vector);
- ret = kvm_irqchip_update_msi_route(kvm_state, v->virq, msg, dev);
+ ret = kvm_irqchip_update_msi_route(kvm_state, v->virq, msg, dev,
+ pci_requester_id(dev));
if (ret < 0) {
return ret;
}
@@ -447,7 +448,8 @@ static void ivshmem_add_kvm_msi_virq(IVShmemState *s, int vector,
IVSHMEM_DPRINTF("ivshmem_add_kvm_msi_virq vector:%d\n", vector);
assert(!s->msi_vectors[vector].pdev);
- ret = kvm_irqchip_add_msi_route(kvm_state, vector, pdev);
+ ret = kvm_irqchip_add_msi_route(kvm_state, vector, pdev,
+ pci_requester_id(pdev));
if (ret < 0) {
error_setg(errp, "kvm_irqchip_add_msi_route failed");
return;
diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
index 7bfa17c..a745e07 100644
--- a/hw/vfio/pci.c
+++ b/hw/vfio/pci.c
@@ -429,7 +429,8 @@ static void vfio_add_kvm_msi_virq(VFIOPCIDevice *vdev, VFIOMSIVector *vector,
return;
}
- virq = kvm_irqchip_add_msi_route(kvm_state, vector_n, &vdev->pdev);
+ virq = kvm_irqchip_add_msi_route(kvm_state, vector_n, &vdev->pdev,
+ pci_requester_id(&vdev->pdev));
if (virq < 0) {
event_notifier_cleanup(&vector->kvm_interrupt);
return;
@@ -457,7 +458,8 @@ static void vfio_remove_kvm_msi_virq(VFIOMSIVector *vector)
static void vfio_update_kvm_msi_virq(VFIOMSIVector *vector, MSIMessage msg,
PCIDevice *pdev)
{
- kvm_irqchip_update_msi_route(kvm_state, vector->virq, msg, pdev);
+ kvm_irqchip_update_msi_route(kvm_state, vector->virq, msg, pdev,
+ pci_requester_id(pdev));
kvm_irqchip_commit_routes(kvm_state);
}
diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
index 755f921..658d904 100644
--- a/hw/virtio/virtio-pci.c
+++ b/hw/virtio/virtio-pci.c
@@ -705,7 +705,8 @@ static int kvm_virtio_pci_vq_vector_use(VirtIOPCIProxy *proxy,
int ret;
if (irqfd->users == 0) {
- ret = kvm_irqchip_add_msi_route(kvm_state, vector, &proxy->pci_dev);
+ ret = kvm_irqchip_add_msi_route(kvm_state, vector, &proxy->pci_dev,
+ pci_requester_id(&proxy->pci_dev));
if (ret < 0) {
return ret;
}
@@ -833,12 +834,14 @@ static int virtio_pci_vq_vector_unmask(VirtIOPCIProxy *proxy,
EventNotifier *n = virtio_queue_get_guest_notifier(vq);
VirtIOIRQFD *irqfd;
int ret = 0;
+ uint16_t requester_id = pci_requester_id(&proxy->pci_dev);
if (proxy->vector_irqfd) {
irqfd = &proxy->vector_irqfd[vector];
if (irqfd->msg.data != msg.data || irqfd->msg.address != msg.address) {
ret = kvm_irqchip_update_msi_route(kvm_state, irqfd->virq, msg,
- &proxy->pci_dev);
+ &proxy->pci_dev,
+ requester_id);
if (ret < 0) {
return ret;
}
diff --git a/include/hw/i386/ioapic_internal.h b/include/hw/i386/ioapic_internal.h
index a11d86d..d68a24f 100644
--- a/include/hw/i386/ioapic_internal.h
+++ b/include/hw/i386/ioapic_internal.h
@@ -103,6 +103,7 @@ typedef struct IOAPICCommonClass {
struct IOAPICCommonState {
SysBusDevice busdev;
MemoryRegion io_memory;
+ uint16_t devid;
uint8_t id;
uint8_t ioregsel;
uint32_t irr;
diff --git a/include/hw/i386/x86-iommu.h b/include/hw/i386/x86-iommu.h
index 0c89d98..5d05865 100644
--- a/include/hw/i386/x86-iommu.h
+++ b/include/hw/i386/x86-iommu.h
@@ -72,6 +72,7 @@ typedef struct IEC_Notifier IEC_Notifier;
struct X86IOMMUState {
SysBusDevice busdev;
+ uint16_t ioapic_bdf; /* expected IOAPIC SID */
bool intr_supported; /* Whether vIOMMU supports IR */
IommuType type; /* IOMMU type - AMD/Intel */
QLIST_HEAD(, IEC_Notifier) iec_notifiers; /* IEC notify list */
diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h
index c9c2436..ae81f92 100644
--- a/include/sysemu/kvm.h
+++ b/include/sysemu/kvm.h
@@ -357,7 +357,8 @@ int kvm_arch_on_sigbus(int code, void *addr);
void kvm_arch_init_irq_routing(KVMState *s);
int kvm_arch_fixup_msi_route(struct kvm_irq_routing_entry *route,
- uint64_t address, uint32_t data, PCIDevice *dev);
+ uint64_t address, uint32_t data,
+ uint16_t requester_id);
/* Notify arch about newly added MSI routes */
int kvm_arch_add_msi_route_post(struct kvm_irq_routing_entry *route,
@@ -481,18 +482,20 @@ static inline void cpu_synchronize_post_init(CPUState *cpu)
/**
* kvm_irqchip_add_msi_route - Add MSI route for specific vector
- * @s: KVM state
- * @vector: which vector to add. This can be either MSI/MSIX
- * vector. The function will automatically detect whether
- * MSI/MSIX is enabled, and fetch corresponding MSI
- * message.
- * @dev: Owner PCI device to add the route. If @dev is specified
- * as @NULL, an empty MSI message will be inited.
- * @return: virq (>=0) when success, errno (<0) when failed.
+ * @s: KVM state
+ * @vector: which vector to add. This can be either MSI/MSIX
+ * vector. The function will automatically detect whether
+ * MSI/MSIX is enabled, and fetch corresponding MSI
+ * message.
+ * @dev: Owner PCI device to add the route. If @dev is specified
+ * as @NULL, an empty MSI message will be inited.
+ * @requester_id: SID when calling IOMMU code
+ * @return: virq (>=0) when success, errno (<0) when failed.
*/
-int kvm_irqchip_add_msi_route(KVMState *s, int vector, PCIDevice *dev);
+int kvm_irqchip_add_msi_route(KVMState *s, int vector, PCIDevice *dev,
+ uint16_t requester_id);
int kvm_irqchip_update_msi_route(KVMState *s, int virq, MSIMessage msg,
- PCIDevice *dev);
+ PCIDevice *dev, uint16_t requester_id);
void kvm_irqchip_commit_routes(KVMState *s);
void kvm_irqchip_release_virq(KVMState *s, int virq);
diff --git a/kvm-all.c b/kvm-all.c
index ebf35b0..9e888ab 100644
--- a/kvm-all.c
+++ b/kvm-all.c
@@ -1246,7 +1246,7 @@ int kvm_irqchip_send_msi(KVMState *s, MSIMessage msg)
return kvm_set_irq(s, route->kroute.gsi, 1);
}
-int kvm_irqchip_add_msi_route(KVMState *s, int vector, PCIDevice *dev)
+int kvm_irqchip_add_msi_route(KVMState *s, int vector, PCIDevice *dev, uint16_t requester_id)
{
struct kvm_irq_routing_entry kroute = {};
int virq;
@@ -1275,7 +1275,8 @@ int kvm_irqchip_add_msi_route(KVMState *s, int vector, PCIDevice *dev)
kroute.u.msi.address_lo = (uint32_t)msg.address;
kroute.u.msi.address_hi = msg.address >> 32;
kroute.u.msi.data = le32_to_cpu(msg.data);
- if (kvm_arch_fixup_msi_route(&kroute, msg.address, msg.data, dev)) {
+ if (kvm_arch_fixup_msi_route(&kroute, msg.address, msg.data,
+ requester_id)) {
kvm_irqchip_release_virq(s, virq);
return -EINVAL;
}
@@ -1290,7 +1291,7 @@ int kvm_irqchip_add_msi_route(KVMState *s, int vector, PCIDevice *dev)
}
int kvm_irqchip_update_msi_route(KVMState *s, int virq, MSIMessage msg,
- PCIDevice *dev)
+ PCIDevice *dev, uint16_t requester_id)
{
struct kvm_irq_routing_entry kroute = {};
@@ -1308,7 +1309,8 @@ int kvm_irqchip_update_msi_route(KVMState *s, int virq, MSIMessage msg,
kroute.u.msi.address_lo = (uint32_t)msg.address;
kroute.u.msi.address_hi = msg.address >> 32;
kroute.u.msi.data = le32_to_cpu(msg.data);
- if (kvm_arch_fixup_msi_route(&kroute, msg.address, msg.data, dev)) {
+ if (kvm_arch_fixup_msi_route(&kroute, msg.address, msg.data,
+ requester_id)) {
return -EINVAL;
}
diff --git a/kvm-stub.c b/kvm-stub.c
index 64e23f6..9a49ce0 100644
--- a/kvm-stub.c
+++ b/kvm-stub.c
@@ -116,7 +116,8 @@ int kvm_on_sigbus(int code, void *addr)
}
#ifndef CONFIG_USER_ONLY
-int kvm_irqchip_add_msi_route(KVMState *s, int vector, PCIDevice *dev)
+int kvm_irqchip_add_msi_route(KVMState *s, int vector, PCIDevice *dev,
+ uint16_t requester_id)
{
return -ENOSYS;
}
@@ -130,7 +131,7 @@ void kvm_irqchip_release_virq(KVMState *s, int virq)
}
int kvm_irqchip_update_msi_route(KVMState *s, int virq, MSIMessage msg,
- PCIDevice *dev)
+ PCIDevice *dev, uint16_t requester_id)
{
return -ENOSYS;
}
diff --git a/qemu-version.h b/qemu-version.h
new file mode 100644
index 0000000..0e06b3e
--- /dev/null
+++ b/qemu-version.h
@@ -0,0 +1 @@
+#define QEMU_PKGVERSION " (v2.3.0-9603-g33203eb-dirty)"
diff --git a/target-i386/kvm.c b/target-i386/kvm.c
index d1a25c5..723bd67 100644
--- a/target-i386/kvm.c
+++ b/target-i386/kvm.c
@@ -3243,7 +3243,7 @@ void kvm_arch_init_irq_routing(KVMState *s)
/* If the ioapic is in QEMU and the lapics are in KVM, reserve
MSI routes for signaling interrupts to the local apics. */
for (i = 0; i < IOAPIC_NUM_PINS; i++) {
- if (kvm_irqchip_add_msi_route(s, 0, NULL) < 0) {
+ if (kvm_irqchip_add_msi_route(s, 0, NULL, 0) < 0) {
error_report("Could not enable split IRQ mode.");
exit(1);
}
@@ -3411,7 +3411,8 @@ int kvm_device_msix_deassign(KVMState *s, uint32_t dev_id)
}
int kvm_arch_fixup_msi_route(struct kvm_irq_routing_entry *route,
- uint64_t address, uint32_t data, PCIDevice *dev)
+ uint64_t address, uint32_t data,
+ uint16_t requester_id)
{
X86IOMMUState *iommu = x86_iommu_get_default();
@@ -3425,9 +3426,8 @@ int kvm_arch_fixup_msi_route(struct kvm_irq_routing_entry *route,
src.address |= route->u.msi.address_lo;
src.data = route->u.msi.data;
- ret = class->int_remap(iommu, &src, &dst, dev ? \
- pci_requester_id(dev) : \
- X86_IOMMU_SID_INVALID);
+ ret = class->int_remap(iommu, &src, &dst, requester_id);
+
if (ret) {
trace_kvm_x86_fixup_msi_error(route->gsi);
return 1;
@@ -3445,6 +3445,7 @@ typedef struct MSIRouteEntry MSIRouteEntry;
struct MSIRouteEntry {
PCIDevice *dev; /* Device pointer */
+ uint16_t requester_id; /* Requesting SID */
int vector; /* MSI/MSIX vector index */
int virq; /* Virtual IRQ index */
QLIST_ENTRY(MSIRouteEntry) list;
@@ -3465,7 +3466,8 @@ static void kvm_update_msi_routes_all(void *private, bool global,
cnt++;
msg = pci_get_msi_message(entry->dev, entry->vector);
kvm_irqchip_update_msi_route(kvm_state, entry->virq,
- msg, entry->dev);
+ msg, entry->dev,
+ entry->requester_id);
}
kvm_irqchip_commit_routes(kvm_state);
trace_kvm_x86_update_msi_routes(cnt);
@@ -3486,6 +3488,7 @@ int kvm_arch_add_msi_route_post(struct kvm_irq_routing_entry *route,
entry = g_new0(MSIRouteEntry, 1);
entry->dev = dev;
+ entry->requester_id = pci_requester_id(dev);
entry->vector = vector;
entry->virq = route->gsi;
QLIST_INSERT_HEAD(&msi_route_list, entry, list);
--
2.1.4
^ permalink raw reply related [flat|nested] 8+ messages in thread* [Qemu-devel] [v3 2/6] hw/i386: enforce SID verification
2016-08-31 16:25 [Qemu-devel] [v3 0/6] AMD IOMMU David Kiarie
2016-08-31 16:25 ` [Qemu-devel] [v3 1/6] hw/msi: Allow platform devices to use explicit SID David Kiarie
@ 2016-08-31 16:25 ` David Kiarie
2016-08-31 16:25 ` [Qemu-devel] [v3 3/6] hw/iommu: Prepare for AMD IOMMU interrupt remapping David Kiarie
` (4 subsequent siblings)
6 siblings, 0 replies; 8+ messages in thread
From: David Kiarie @ 2016-08-31 16:25 UTC (permalink / raw)
To: qemu-devel
Cc: ehabkost, rkrcmar, peterx, jan.kiszka, mst, valentine.sinitsyn,
David Kiarie
Platform device are now able to make interrupt request with
explicit SIDs hence we can safely expect triggered AddressSpace ID
to match the requesting ID
Signed-off-by: David Kiarie <davidkiarie4@gmail.com>
---
hw/i386/intel_iommu.c | 77 ++++++++++++++++++++++++++-------------------------
1 file changed, 39 insertions(+), 38 deletions(-)
diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c
index 496d836..e4bad6a 100644
--- a/hw/i386/intel_iommu.c
+++ b/hw/i386/intel_iommu.c
@@ -2043,43 +2043,41 @@ static int vtd_irte_get(IntelIOMMUState *iommu, uint16_t index,
return -VTD_FR_IR_IRTE_RSVD;
}
- if (sid != X86_IOMMU_SID_INVALID) {
- /* Validate IRTE SID */
- source_id = le32_to_cpu(entry->irte.source_id);
- switch (entry->irte.sid_vtype) {
- case VTD_SVT_NONE:
- VTD_DPRINTF(IR, "No SID validation for IRTE index %d", index);
- break;
-
- case VTD_SVT_ALL:
- mask = vtd_svt_mask[entry->irte.sid_q];
- if ((source_id & mask) != (sid & mask)) {
- VTD_DPRINTF(GENERAL, "SID validation for IRTE index "
- "%d failed (reqid 0x%04x sid 0x%04x)", index,
- sid, source_id);
- return -VTD_FR_IR_SID_ERR;
- }
- break;
+ /* Validate IRTE SID */
+ source_id = le32_to_cpu(entry->irte.source_id);
+ switch (entry->irte.sid_vtype) {
+ case VTD_SVT_NONE:
+ VTD_DPRINTF(IR, "No SID validation for IRTE index %d", index);
+ break;
- case VTD_SVT_BUS:
- bus_max = source_id >> 8;
- bus_min = source_id & 0xff;
- bus = sid >> 8;
- if (bus > bus_max || bus < bus_min) {
- VTD_DPRINTF(GENERAL, "SID validation for IRTE index %d "
- "failed (bus %d outside %d-%d)", index, bus,
- bus_min, bus_max);
- return -VTD_FR_IR_SID_ERR;
- }
- break;
+ case VTD_SVT_ALL:
+ mask = vtd_svt_mask[entry->irte.sid_q];
+ if ((source_id & mask) != (sid & mask)) {
+ VTD_DPRINTF(GENERAL, "SID validation for IRTE index "
+ "%d failed (reqid 0x%04x sid 0x%04x)", index,
+ sid, source_id);
+ return -VTD_FR_IR_SID_ERR;
+ }
+ break;
- default:
- VTD_DPRINTF(GENERAL, "Invalid SVT bits (0x%x) in IRTE index "
- "%d", entry->irte.sid_vtype, index);
- /* Take this as verification failure. */
+ case VTD_SVT_BUS:
+ bus_max = source_id >> 8;
+ bus_min = source_id & 0xff;
+ bus = sid >> 8;
+ if (bus > bus_max || bus < bus_min) {
+ VTD_DPRINTF(GENERAL, "SID validation for IRTE index %d "
+ "failed (bus %d outside %d-%d)", index, bus,
+ bus_min, bus_max);
return -VTD_FR_IR_SID_ERR;
- break;
}
+ break;
+
+ default:
+ VTD_DPRINTF(GENERAL, "Invalid SVT bits (0x%x) in IRTE index "
+ "%d", entry->irte.sid_vtype, index);
+ /* Take this as verification failure. */
+ return -VTD_FR_IR_SID_ERR;
+ break;
}
return 0;
@@ -2252,14 +2250,17 @@ static MemTxResult vtd_mem_ir_write(void *opaque, hwaddr addr,
{
int ret = 0;
MSIMessage from = {}, to = {};
- uint16_t sid = X86_IOMMU_SID_INVALID;
+ VTDAddressSpace *as = opaque;
+ uint16_t sid = PCI_BUILD_BDF(pci_bus_num(as->bus), as->devfn);
from.address = (uint64_t) addr + VTD_INTERRUPT_ADDR_FIRST;
from.data = (uint32_t) value;
- if (!attrs.unspecified) {
- /* We have explicit Source ID */
- sid = attrs.requester_id;
+ if (attrs.requester_id != sid) {
+ VTD_DPRINTF(GENERAL, "int remap request for sid 0x%04x"
+ " requester_id 0x%04x couldn't be verified",
+ sid, attrs.requester_id);
+ return MEMTX_ERROR;
}
ret = vtd_interrupt_remap_msi(opaque, &from, &to, sid);
@@ -2325,7 +2326,7 @@ VTDAddressSpace *vtd_find_add_as(IntelIOMMUState *s, PCIBus *bus, int devfn)
memory_region_init_iommu(&vtd_dev_as->iommu, OBJECT(s),
&s->iommu_ops, "intel_iommu", UINT64_MAX);
memory_region_init_io(&vtd_dev_as->iommu_ir, OBJECT(s),
- &vtd_mem_ir_ops, s, "intel_iommu_ir",
+ &vtd_mem_ir_ops, vtd_dev_as, "intel_iommu_ir",
VTD_INTERRUPT_ADDR_SIZE);
memory_region_add_subregion(&vtd_dev_as->iommu, VTD_INTERRUPT_ADDR_FIRST,
&vtd_dev_as->iommu_ir);
--
2.1.4
^ permalink raw reply related [flat|nested] 8+ messages in thread* [Qemu-devel] [v3 3/6] hw/iommu: Prepare for AMD IOMMU interrupt remapping
2016-08-31 16:25 [Qemu-devel] [v3 0/6] AMD IOMMU David Kiarie
2016-08-31 16:25 ` [Qemu-devel] [v3 1/6] hw/msi: Allow platform devices to use explicit SID David Kiarie
2016-08-31 16:25 ` [Qemu-devel] [v3 2/6] hw/i386: enforce SID verification David Kiarie
@ 2016-08-31 16:25 ` David Kiarie
2016-08-31 16:25 ` [Qemu-devel] [v3 4/6] hw/iommu: " David Kiarie
` (3 subsequent siblings)
6 siblings, 0 replies; 8+ messages in thread
From: David Kiarie @ 2016-08-31 16:25 UTC (permalink / raw)
To: qemu-devel
Cc: ehabkost, rkrcmar, peterx, jan.kiszka, mst, valentine.sinitsyn,
David Kiarie
Introduce macros and trace events for use in AMD IOMMU
interrupt remapping
Signed-off-by: David Kiarie <davidkiarie4@gmail.com>
---
hw/i386/amd_iommu.h | 80 ++++++++++++++++++++++++++++++++++++++++++++++++++++
hw/i386/trace-events | 7 +++++
2 files changed, 87 insertions(+)
diff --git a/hw/i386/amd_iommu.h b/hw/i386/amd_iommu.h
index 884926e..5c4a13b 100644
--- a/hw/i386/amd_iommu.h
+++ b/hw/i386/amd_iommu.h
@@ -177,6 +177,68 @@
#define AMDVI_IOTLB_MAX_SIZE 1024
#define AMDVI_DEVID_SHIFT 36
+/* interrupt types */
+#define AMDVI_MT_FIXED 0x0
+#define AMDVI_MT_ARBIT 0x1
+#define AMDVI_MT_SMI 0x2
+#define AMDVI_MT_NMI 0x3
+#define AMDVI_MT_INIT 0x4
+#define AMDVI_MT_EXTINT 0x6
+#define AMDVI_MT_LINT1 0xb
+#define AMDVI_MT_LINT0 0xe
+
+/* MSI interrupt type mask */
+#define AMDVI_IR_TYPE_MASK 0x300
+
+/* interrupt destination mode */
+#define AMDVI_IRDEST_MODE_MASK 0x2
+
+/* select MSI data 10:0 bits */
+#define AMDVI_IRTE_INDEX_MASK 0x7ff
+
+/* bits determining whether specific interrupts should be passed
+ * split DTE into 64-bit chunks
+ */
+#define AMDVI_DTE_INTPASS_LSHIFT 56
+#define AMDVI_DTE_EINTPASS_LSHIFT 57
+#define AMDVI_DTE_NMIPASS_LSHIFT 58
+#define AMDVI_DTE_INTCTL_RSHIFT 60
+#define AMDVI_DTE_LINT0PASS_LSHIFT 62
+#define AMDVI_DTE_LINT1PASS_LSHIFT 63
+
+/* INTCTL expected values */
+#define AMDVI_INTCTL_ABORT 0x0
+#define AMDVI_INTCTL_PASS 0x1
+#define AMDVI_INTCTL_REMAP 0x2
+#define AMDVI_INTCTL_RSVD 0x3
+
+/* interrupt data valid */
+#define AMDVI_IR_VALID (1UL << 0)
+
+/* interrupt root table mask */
+#define AMDVI_IRTEROOT_MASK 0xffffffffffffc0
+
+/* default IRTE size */
+#define AMDVI_DEFAULT_IRTE_SIZE 0x4
+
+#define AMDVI_IR_TABLE_SIZE_MASK 0xfe
+
+/* offsets into MSI data */
+#define AMDVI_MSI_DATA_DM_RSHIFT 0x8
+#define AMDVI_MSI_DATA_LEVEL_RSHIFT 0xe
+#define AMDVI_MSI_DATA_TRM_RSHIFT 0xf
+
+/* offsets into MSI address */
+#define AMDVI_MSI_ADDR_DM_RSHIFT 0x2
+#define AMDVI_MSI_ADDR_RH_RSHIFT 0x3
+#define AMDVI_MSI_ADDR_DEST_RSHIFT 0xc
+
+#define AMDVI_BUS_NUM 0x0
+/* AMD-Vi specific IOAPIC Device function */
+#define AMDVI_DEVFN_IOAPIC 0xa0
+
+#define AMDVI_LOCAL_APIC_ADDR 0xfee00000
+
/* extended feature support */
#define AMDVI_EXT_FEATURES (AMDVI_FEATURE_PREFETCH | AMDVI_FEATURE_PPR | \
AMDVI_FEATURE_IA | AMDVI_FEATURE_GT | AMDVI_FEATURE_HE | \
@@ -214,6 +276,24 @@
#define AMDVI_INT_ADDR_FIRST 0xfee00000
#define AMDVI_INT_ADDR_LAST 0xfeefffff
+#define AMDVI_INT_ADDR_SIZE ((AMDVI_INT_ADDR_LAST - \
+ AMDVI_INT_ADDR_FIRST) + 1)
+/* AMD IOMMU errors */
+#define AMDVI_ILLEG_DEV_TAB 0x1
+#define AMDVI_IOPF_ 0x2
+#define AMDVI_DEV_TAB_HW 0x3
+#define AMDVI_PAGE_TAB_HW 0x4
+#define AMDVI_ILLEG_COM 0x5
+#define AMDVI_COM_HW 0x6
+#define AMDVI_IOTLB_TIMEOUT 0x7
+#define AMDVI_INVAL_DEV_REQ 0x8
+#define AMDVI_INVAL_PPR_REQ 0x9
+#define AMDVI_EVT_COUNT_ZERO 0xa
+
+/* represent target and master aborts error state */
+#define AMDVI_TARGET_ABORT 0xb
+#define AMDVI_MASTER_ABORT 0xc
+
#define TYPE_AMD_IOMMU_DEVICE "amd-iommu"
#define AMD_IOMMU_DEVICE(obj)\
OBJECT_CHECK(AMDVIState, (obj), TYPE_AMD_IOMMU_DEVICE)
diff --git a/hw/i386/trace-events b/hw/i386/trace-events
index 60bdf6a..344c2f6 100644
--- a/hw/i386/trace-events
+++ b/hw/i386/trace-events
@@ -42,3 +42,10 @@ amdvi_mode_invalid(unsigned level, uint64_t addr)"error: translation level 0x%"P
amdvi_page_fault(uint64_t addr) "error: page fault accessing guest physical address 0x%"PRIx64
amdvi_iotlb_hit(uint16_t bus, uint16_t slot, uint16_t func, uint64_t addr, uint64_t txaddr) "hit iotlb devid %02x:%02x.%x gpa 0x%"PRIx64 " hpa 0x%"PRIx64
amdvi_translation_result(uint16_t bus, uint16_t slot, uint16_t func, uint64_t addr, uint64_t txaddr) "devid: %02x:%02x.%x gpa 0x%"PRIx64 " hpa 0x%"PRIx64
+amdvi_irte_get_fail(uint64_t addr, uint64_t offset) "couldn't access device table entry 0x%"PRIx64" + offset 0x%"PRIx64
+amdvi_invalid_irte_entry(uint16_t devid, uint64_t offset) "devid %x requested IRTE offset 0x%"PRIx64" Outside IR table range"
+amdvi_ir_request(uint32_t data, uint64_t addr, uint16_t sid) "IR request data 0x%"PRIx32" address 0x%"PRIx64" SID %x"
+amdvi_ir_remap(uint32_t data, uint64_t addr, uint16_t sid) "IR remap data 0x%"PRIx32" address 0x%"PRIx64" SID %x"
+amdvi_ir_target_abort(uint32_t data, uint64_t addr, uint16_t sid) "IR target abort data 0x%"PRIx32" address 0x%"PRIx64" SID %x"
+amdvi_ir_write_fail(uint64_t addr, uint32_t data) "fail to write to addr 0x%"PRIx64 " value 0x%"PRIx32
+amdvi_ir_read_fail(uint64_t addr) " fail to read from addr 0x%"PRIx64
--
2.1.4
^ permalink raw reply related [flat|nested] 8+ messages in thread* [Qemu-devel] [v3 4/6] hw/iommu: AMD IOMMU interrupt remapping
2016-08-31 16:25 [Qemu-devel] [v3 0/6] AMD IOMMU David Kiarie
` (2 preceding siblings ...)
2016-08-31 16:25 ` [Qemu-devel] [v3 3/6] hw/iommu: Prepare for AMD IOMMU interrupt remapping David Kiarie
@ 2016-08-31 16:25 ` David Kiarie
2016-08-31 16:25 ` [Qemu-devel] [v3 5/6] hw/acpi: report IOAPIC on IVRS David Kiarie
` (2 subsequent siblings)
6 siblings, 0 replies; 8+ messages in thread
From: David Kiarie @ 2016-08-31 16:25 UTC (permalink / raw)
To: qemu-devel
Cc: ehabkost, rkrcmar, peterx, jan.kiszka, mst, valentine.sinitsyn,
David Kiarie
Introduce AMD IOMMU interrupt remapping and hook it onto
the existing interrupt remapping infrastructure
Signed-off-by: David Kiarie <davidkiarie4@gmail.com>
---
hw/i386/amd_iommu.c | 241 +++++++++++++++++++++++++++++++++++++++++++++++++++-
hw/i386/amd_iommu.h | 4 +-
hw/intc/ioapic.c | 9 +-
3 files changed, 249 insertions(+), 5 deletions(-)
diff --git a/hw/i386/amd_iommu.c b/hw/i386/amd_iommu.c
index 226fea5..54519bb 100644
--- a/hw/i386/amd_iommu.c
+++ b/hw/i386/amd_iommu.c
@@ -18,8 +18,10 @@
* with this program; if not, see <http://www.gnu.org/licenses/>.
*
* Cache implementation inspired by hw/i386/intel_iommu.c
+ *
*/
#include "qemu/osdep.h"
+#include "qemu/error-report.h"
#include "hw/i386/amd_iommu.h"
#include "trace.h"
@@ -255,6 +257,31 @@ typedef struct QEMU_PACKED {
uint32_t reserved_5:16;
} CMDCompletePPR;
+typedef union IRTE {
+ struct {
+#ifdef HOST_WORDS_BIGENDIAN
+ uint32_t destination:8;
+ uint32_t rsvd_1:1;
+ uint32_t dm:1;
+ uint32_t rq_eoi:1;
+ uint32_t int_type:3;
+ uint32_t no_fault:1;
+ uint32_t valid:1;
+#else
+ uint32_t valid:1;
+ uint32_t no_fault:1;
+ uint32_t int_type:3;
+ uint32_t rq_eoi:1;
+ uint32_t dm:1;
+ uint32_t rsvd_1:1;
+ uint32_t destination:8;
+#endif
+ uint32_t vector:8;
+ uint32_t rsvd_2:8;
+ } bits;
+ uint32_t data;
+} IRTE;
+
/* configure MMIO registers at startup/reset */
static void amdvi_set_quad(AMDVIState *s, hwaddr addr, uint64_t val,
uint64_t romask, uint64_t w1cmask)
@@ -641,6 +668,11 @@ static void amdvi_inval_inttable(AMDVIState *s, CMDInvalIntrTable *inval)
amdvi_log_illegalcom_error(s, inval->type, s->cmdbuf + s->cmdbuf_head);
return;
}
+
+ if (s->ir_cache) {
+ x86_iommu_iec_notify_all(X86_IOMMU_DEVICE(s), true, 0, 0);
+ }
+
trace_amdvi_intr_inval();
}
@@ -1203,6 +1235,197 @@ static IOMMUTLBEntry amdvi_translate(MemoryRegion *iommu, hwaddr addr,
return ret;
}
+static inline int amdvi_ir_handle_non_vectored(MSIMessage *src,
+ MSIMessage *dst, uint8_t bitpos,
+ uint64_t dte)
+{
+ if ((dte & (1UL << bitpos))) {
+ /* passing interrupt enabled */
+ memcpy(dst, src, sizeof(*dst));
+ } else {
+ /* should be target aborted */
+ return -AMDVI_TARGET_ABORT;
+ }
+ return 0;
+}
+
+static int amdvi_remap_ir_intctl(uint64_t dte, IRTE irte,
+ MSIMessage *src, MSIMessage *dst)
+{
+ int ret = 0;
+
+ switch ((dte >> AMDVI_DTE_INTCTL_RSHIFT) & 3UL) {
+ case AMDVI_INTCTL_PASS:
+ /* pass */
+ memcpy(dst, src, sizeof(*dst));
+ break;
+ case AMDVI_INTCTL_REMAP:
+ /* remap */
+ if (irte.bits.valid) {
+ /* LOCAL APIC address */
+ dst->address = AMDVI_LOCAL_APIC_ADDR;
+ /* destination mode */
+ dst->address |= ((uint64_t)irte.bits.dm) <<
+ AMDVI_MSI_ADDR_DM_RSHIFT;
+ /* RH */
+ dst->address |= ((uint64_t)irte.bits.rq_eoi) <<
+ AMDVI_MSI_ADDR_RH_RSHIFT;
+ /* Destination ID */
+ dst->address |= ((uint64_t)irte.bits.destination) <<
+ AMDVI_MSI_ADDR_DEST_RSHIFT;
+ /* construct data - vector */
+ dst->data |= irte.bits.vector;
+ /* Interrupt type */
+ dst->data |= ((uint64_t)irte.bits.int_type) <<
+ AMDVI_MSI_DATA_DM_RSHIFT;
+ } else {
+ ret = -AMDVI_TARGET_ABORT;
+ }
+ break;
+ case AMDVI_INTCTL_ABORT:
+ case AMDVI_INTCTL_RSVD:
+ ret = -AMDVI_TARGET_ABORT;
+ }
+ return ret;
+}
+
+static int amdvi_irte_get(AMDVIState *s, MSIMessage *src, IRTE *irte,
+ uint64_t *dte, uint16_t devid)
+{
+ uint64_t irte_root, offset = devid * AMDVI_DEVTAB_ENTRY_SIZE,
+ ir_table_size;
+
+ irte_root = dte[2] & AMDVI_IRTEROOT_MASK;
+ offset = (src->data & AMDVI_IRTE_INDEX_MASK) << 2;
+ ir_table_size = 1UL << (dte[2] & AMDVI_IR_TABLE_SIZE_MASK);
+ /* enforce IR table size */
+ if (offset > (ir_table_size * AMDVI_DEFAULT_IRTE_SIZE)) {
+ trace_amdvi_invalid_irte_entry(offset, ir_table_size);
+ return -AMDVI_TARGET_ABORT;
+ }
+ /* read IRTE */
+ if (dma_memory_read(&address_space_memory, irte_root + offset,
+ irte, sizeof(*irte))) {
+ trace_amdvi_irte_get_fail(irte_root, offset);
+ return -AMDVI_DEV_TAB_HW;
+ }
+ return 0;
+}
+
+static int amdvi_int_remap(X86IOMMUState *iommu, MSIMessage *src,
+ MSIMessage *dst, uint16_t sid)
+{
+ trace_amdvi_ir_request(src->data, src->address, sid);
+
+ AMDVIState *s = AMD_IOMMU_DEVICE(iommu);
+ int ret = 0;
+ uint64_t dte[4];
+ uint32_t bitpos;
+ IRTE irte;
+
+ amdvi_get_dte(s, sid, dte);
+
+ /* interrupt remapping disabled */
+ if (!(dte[2] & AMDVI_IR_VALID)) {
+ memcpy(dst, src, sizeof(*src));
+ return ret;
+ }
+
+ ret = amdvi_irte_get(s, src, &irte, dte, sid);
+ if (ret < 0) {
+ goto no_remap;
+ }
+ switch (src->data & AMDVI_IR_TYPE_MASK) {
+ case AMDVI_MT_FIXED:
+ case AMDVI_MT_ARBIT:
+ ret = amdvi_remap_ir_intctl(dte[2], irte, src, dst);
+ if (ret < 0) {
+ goto no_remap;
+ } else {
+ s->ir_cache = true;
+ trace_amdvi_ir_remap(dst->data, dst->address, sid);
+ return ret;
+ }
+ /* not handling SMI currently */
+ case AMDVI_MT_SMI:
+ error_report("SMI interrupts not currently handled");
+ goto no_remap;
+ case AMDVI_MT_NMI:
+ bitpos = AMDVI_DTE_NMIPASS_LSHIFT;
+ break;
+ case AMDVI_MT_INIT:
+ bitpos = AMDVI_DTE_INTPASS_LSHIFT;
+ break;
+ case AMDVI_MT_EXTINT:
+ bitpos = AMDVI_DTE_EINTPASS_LSHIFT;
+ break;
+ case AMDVI_MT_LINT1:
+ bitpos = AMDVI_DTE_LINT1PASS_LSHIFT;
+ break;
+ case AMDVI_MT_LINT0:
+ bitpos = AMDVI_DTE_LINT0PASS_LSHIFT;
+ default:
+ goto no_remap;
+ }
+
+ ret = amdvi_ir_handle_non_vectored(src, dst, bitpos, dte[2]);
+ if (ret < 0){
+ goto no_remap;
+ }
+ s->ir_cache = true;
+ trace_amdvi_ir_remap(dst->data, dst->address, sid);
+ return ret;
+no_remap:
+ memcpy(dst, src, sizeof(*src));
+ trace_amdvi_ir_target_abort(dst->data, dst->address, sid);
+ return ret;
+}
+
+static MemTxResult amdvi_ir_read(void *opaque, hwaddr addr,
+ uint64_t *data, unsigned size,
+ MemTxAttrs attrs)
+{
+ return MEMTX_OK;
+}
+
+static MemTxResult amdvi_ir_write(void *opaque, hwaddr addr, uint64_t val,
+ unsigned size, MemTxAttrs attrs)
+{
+ AMDVIAddressSpace *as = opaque;
+ MSIMessage from = { addr + AMDVI_INT_ADDR_FIRST, val }, to = { 0, 0};
+ int ret = 0;
+
+ ret = amdvi_int_remap(X86_IOMMU_DEVICE(as->iommu_state), &from, &to,
+ attrs.requester_id);
+
+ if (ret < 0) {
+ trace_amdvi_ir_target_abort(from.data, from.address,
+ attrs.requester_id);
+ return MEMTX_ERROR;
+ }
+
+ if(dma_memory_write(&address_space_memory, to.address, &to.data, size)) {
+ trace_amdvi_ir_write_fail(to.address, to.data);
+ return MEMTX_ERROR;
+ }
+
+ return MEMTX_OK;
+}
+
+static const MemoryRegionOps amdvi_ir_ops = {
+ .read_with_attrs = amdvi_ir_read,
+ .write_with_attrs = amdvi_ir_write,
+ .endianness = DEVICE_LITTLE_ENDIAN,
+ .impl = {
+ .min_access_size = 4,
+ .max_access_size = 4,
+ },
+ .valid = {
+ .min_access_size = 4,
+ .max_access_size = 4,
+ }
+};
+
static AddressSpace *amdvi_host_dma_iommu(PCIBus *bus, void *opaque, int devfn)
{
AMDVIState *s = opaque;
@@ -1226,6 +1449,12 @@ static AddressSpace *amdvi_host_dma_iommu(PCIBus *bus, void *opaque, int devfn)
memory_region_init_iommu(&iommu_as[devfn]->iommu, OBJECT(s),
&s->iommu_ops, "amd-iommu", UINT64_MAX);
+ memory_region_init_io(&iommu_as[devfn]->iommu_ir, OBJECT(s),
+ &amdvi_ir_ops, iommu_as[devfn], "amd-iommu-ir",
+ AMDVI_INT_ADDR_SIZE);
+ memory_region_add_subregion(&iommu_as[devfn]->iommu,
+ AMDVI_INT_ADDR_FIRST,
+ &iommu_as[devfn]->iommu_ir);
address_space_init(&iommu_as[devfn]->as, &iommu_as[devfn]->iommu,
"amd-iommu");
}
@@ -1274,6 +1503,7 @@ static void amdvi_init(AMDVIState *s)
s->enabled = false;
s->ats_enabled = false;
s->cmdbuf_enabled = false;
+ s->ir_cache = false;
/* reset MMIO */
memset(s->mmior, 0, AMDVI_MMIO_SIZE);
@@ -1313,11 +1543,15 @@ static void amdvi_realize(DeviceState *dev, Error **err)
AMDVIState *s = AMD_IOMMU_DEVICE(dev);
X86IOMMUState *x86_iommu = X86_IOMMU_DEVICE(dev);
PCIBus *bus = PC_MACHINE(qdev_get_machine())->bus;
+ PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
s->iotlb = g_hash_table_new_full(amdvi_uint64_hash,
amdvi_uint64_equal, g_free, g_free);
- /* This device should take care of IOMMU PCI properties */
+ /* AMD IOMMU has Interrupt Remapping on by default */
+ x86_iommu->intr_supported = true;
x86_iommu->type = TYPE_AMD;
+
+ /* This device should take care of IOMMU PCI properties */
qdev_set_parent_bus(DEVICE(&s->pci), &bus->qbus);
object_property_set_bool(OBJECT(&s->pci), true, "realized", err);
s->capab_offset = pci_add_capability(&s->pci.dev, AMDVI_CAPAB_ID_SEC, 0,
@@ -1329,9 +1563,13 @@ static void amdvi_realize(DeviceState *dev, Error **err)
memory_region_init_io(&s->mmio, OBJECT(s), &mmio_mem_ops, s, "amdvi-mmio",
AMDVI_MMIO_SIZE);
+ x86_iommu->ioapic_bdf = PCI_BUILD_BDF(AMDVI_BUS_NUM,
+ AMDVI_SB_IOAPIC_ID);
+
sysbus_init_mmio(SYS_BUS_DEVICE(s), &s->mmio);
sysbus_mmio_map(SYS_BUS_DEVICE(s), 0, AMDVI_BASE_ADDR);
pci_setup_iommu(bus, amdvi_host_dma_iommu, s);
+ pcms->ioapic_as = amdvi_host_dma_iommu(bus, s, AMDVI_SB_IOAPIC_ID);
s->devid = object_property_get_int(OBJECT(&s->pci), "addr", err);
msi_init(&s->pci.dev, 0, 1, true, false, err);
amdvi_init(s);
@@ -1358,6 +1596,7 @@ static void amdvi_class_init(ObjectClass *klass, void* data)
dc->vmsd = &vmstate_amdvi;
dc->hotpluggable = false;
dc_class->realize = amdvi_realize;
+ dc_class->int_remap = amdvi_int_remap;
}
static const TypeInfo amdvi = {
diff --git a/hw/i386/amd_iommu.h b/hw/i386/amd_iommu.h
index 5c4a13b..28b9603 100644
--- a/hw/i386/amd_iommu.h
+++ b/hw/i386/amd_iommu.h
@@ -235,7 +235,7 @@
#define AMDVI_BUS_NUM 0x0
/* AMD-Vi specific IOAPIC Device function */
-#define AMDVI_DEVFN_IOAPIC 0xa0
+#define AMDVI_SB_IOAPIC_ID 0xa0
#define AMDVI_LOCAL_APIC_ADDR 0xfee00000
@@ -338,6 +338,8 @@ typedef struct AMDVIState {
uint32_t evtlog_len; /* event log length */
uint32_t evtlog_head; /* current IOMMU write position */
uint32_t evtlog_tail; /* current Software read position */
+ /* whether we have remapped any interrupts and hence IR cache */
+ bool ir_cache;
/* unused for now */
hwaddr excl_base; /* base DVA - IOMMU exclusion range */
diff --git a/hw/intc/ioapic.c b/hw/intc/ioapic.c
index b8b2f33..55129d8 100644
--- a/hw/intc/ioapic.c
+++ b/hw/intc/ioapic.c
@@ -390,9 +390,14 @@ static void ioapic_machine_done_notify(Notifier *notifier, void *data)
#ifdef CONFIG_KVM
IOAPICCommonState *s = container_of(notifier, IOAPICCommonState,
machine_done);
+ X86IOMMUState *iommu = x86_iommu_get_default();
+
+ /* when kernel_irqchip=off */
+ if (iommu) {
+ s->devid = iommu->ioapic_bdf;
+ }
if (kvm_irqchip_is_split()) {
- X86IOMMUState *iommu = x86_iommu_get_default();
MSIMessage msg = {0, 0};
int i;
@@ -401,14 +406,12 @@ static void ioapic_machine_done_notify(Notifier *notifier, void *data)
* when there are IR invalidates, we can be notified to
* update kernel IR cache. */
x86_iommu_iec_register_notifier(iommu, ioapic_iec_notifier, s);
- s->devid = iommu->ioapic_bdf;
/* update IOAPIC routes to the right SID */
for (i = 0; i < IOAPIC_NUM_PINS; i++) {
kvm_irqchip_update_msi_route(kvm_state, i, msg, NULL, s->devid);
}
kvm_irqchip_commit_routes(kvm_state);
}
-
}
#endif
}
--
2.1.4
^ permalink raw reply related [flat|nested] 8+ messages in thread* [Qemu-devel] [v3 5/6] hw/acpi: report IOAPIC on IVRS
2016-08-31 16:25 [Qemu-devel] [v3 0/6] AMD IOMMU David Kiarie
` (3 preceding siblings ...)
2016-08-31 16:25 ` [Qemu-devel] [v3 4/6] hw/iommu: " David Kiarie
@ 2016-08-31 16:25 ` David Kiarie
2016-08-31 16:25 ` [Qemu-devel] [v3 6/6] hw/iommu: share common code between IOMMUs David Kiarie
2016-09-09 18:22 ` [Qemu-devel] [v3 0/6] AMD IOMMU Michael S. Tsirkin
6 siblings, 0 replies; 8+ messages in thread
From: David Kiarie @ 2016-08-31 16:25 UTC (permalink / raw)
To: qemu-devel
Cc: ehabkost, rkrcmar, peterx, jan.kiszka, mst, valentine.sinitsyn,
David Kiarie
Report IOAPIC via IVRS which effectively allows linux AMD-Vi
driver to enable interrupt remapping
Signed-off-by: David Kiarie <davidkiarie4@gmail.com>
---
hw/i386/acpi-build.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/hw/i386/acpi-build.c b/hw/i386/acpi-build.c
index 49bd183..c2559ff 100644
--- a/hw/i386/acpi-build.c
+++ b/hw/i386/acpi-build.c
@@ -2615,6 +2615,8 @@ build_amd_iommu(GArray *table_data, BIOSLinker *linker)
* Refer to Spec - Table 95:IVHD Device Entry Type Codes(4-byte)
*/
build_append_int_noprefix(table_data, 0x0000001, 4);
+ /* IOAPIC represented as an 8-byte entry. Spec v2.62 Tables 97 */
+ build_append_int_noprefix(table_data, 0x0100a000cf000048, 8);
build_header(linker, table_data, (void *)(table_data->data + iommu_start),
"IVRS", table_data->len - iommu_start, 1, NULL, NULL);
--
2.1.4
^ permalink raw reply related [flat|nested] 8+ messages in thread* [Qemu-devel] [v3 6/6] hw/iommu: share common code between IOMMUs
2016-08-31 16:25 [Qemu-devel] [v3 0/6] AMD IOMMU David Kiarie
` (4 preceding siblings ...)
2016-08-31 16:25 ` [Qemu-devel] [v3 5/6] hw/acpi: report IOAPIC on IVRS David Kiarie
@ 2016-08-31 16:25 ` David Kiarie
2016-09-09 18:22 ` [Qemu-devel] [v3 0/6] AMD IOMMU Michael S. Tsirkin
6 siblings, 0 replies; 8+ messages in thread
From: David Kiarie @ 2016-08-31 16:25 UTC (permalink / raw)
To: qemu-devel
Cc: ehabkost, rkrcmar, peterx, jan.kiszka, mst, valentine.sinitsyn,
David Kiarie
Enabling interrupt remapping with kernel_irqchip=on should result
in an error for both VT-d and AMD-Vi
Signed-off-by: David Kiarie <davidkiarie4@gmail.com>
---
hw/i386/intel_iommu.c | 9 ---------
hw/i386/x86-iommu.c | 8 ++++++++
2 files changed, 8 insertions(+), 9 deletions(-)
diff --git a/hw/i386/intel_iommu.c b/hw/i386/intel_iommu.c
index e4bad6a..bf86dcc 100644
--- a/hw/i386/intel_iommu.c
+++ b/hw/i386/intel_iommu.c
@@ -30,7 +30,6 @@
#include "hw/boards.h"
#include "hw/i386/x86-iommu.h"
#include "hw/pci-host/q35.h"
-#include "sysemu/kvm.h"
/*#define DEBUG_INTEL_IOMMU*/
#ifdef DEBUG_INTEL_IOMMU
@@ -2472,14 +2471,6 @@ static void vtd_realize(DeviceState *dev, Error **errp)
Q35_PSEUDO_DEVFN_IOAPIC);
/* Pseudo address space under root PCI bus. */
pcms->ioapic_as = vtd_host_dma_iommu(bus, s, Q35_PSEUDO_DEVFN_IOAPIC);
-
- /* Currently Intel IOMMU IR only support "kernel-irqchip={off|split}" */
- if (x86_iommu->intr_supported && kvm_irqchip_in_kernel() &&
- !kvm_irqchip_is_split()) {
- error_report("Intel Interrupt Remapping cannot work with "
- "kernel-irqchip=on, please use 'split|off'.");
- exit(1);
- }
}
static void vtd_class_init(ObjectClass *klass, void *data)
diff --git a/hw/i386/x86-iommu.c b/hw/i386/x86-iommu.c
index 2278af7..66510f7 100644
--- a/hw/i386/x86-iommu.c
+++ b/hw/i386/x86-iommu.c
@@ -21,6 +21,7 @@
#include "hw/sysbus.h"
#include "hw/boards.h"
#include "hw/i386/x86-iommu.h"
+#include "sysemu/kvm.h"
#include "qemu/error-report.h"
#include "trace.h"
@@ -84,6 +85,13 @@ static void x86_iommu_realize(DeviceState *dev, Error **errp)
if (x86_class->realize) {
x86_class->realize(dev, errp);
}
+ /* Currently IOMMU IR only support "kernel-irqchip={off|split}" */
+ if (x86_iommu->intr_supported && kvm_irqchip_in_kernel() &&
+ !kvm_irqchip_is_split()) {
+ error_report("Interrupt Remapping cannot work with "
+ "kernel-irqchip=on, please use 'split|off'.");
+ exit(1);
+ }
x86_iommu_set_default(X86_IOMMU_DEVICE(dev));
}
--
2.1.4
^ permalink raw reply related [flat|nested] 8+ messages in thread* Re: [Qemu-devel] [v3 0/6] AMD IOMMU
2016-08-31 16:25 [Qemu-devel] [v3 0/6] AMD IOMMU David Kiarie
` (5 preceding siblings ...)
2016-08-31 16:25 ` [Qemu-devel] [v3 6/6] hw/iommu: share common code between IOMMUs David Kiarie
@ 2016-09-09 18:22 ` Michael S. Tsirkin
6 siblings, 0 replies; 8+ messages in thread
From: Michael S. Tsirkin @ 2016-09-09 18:22 UTC (permalink / raw)
To: David Kiarie
Cc: qemu-devel, ehabkost, rkrcmar, peterx, jan.kiszka,
valentine.sinitsyn
On Wed, Aug 31, 2016 at 07:25:11PM +0300, David Kiarie wrote:
> Hello all,
>
> Changes since V2
> -formating fixes.
> -fixed an issue where the right IOAPIC id was not being correctly set when using kernel_irqchip=off
>
> The following patchset implements AMD-Vi interrupt remapping logic and hooks it onto existing IR infrastucture.
>
> I have bundled this patchset together with the "Explicit SID for IOAPIC"."Explicit SID for IOAPIC" functions to
> affiliate MSI routes with a requester ID and a PCI device if present which enables platform devices like IOAPIC to
> make interrupt requests using an explicit SID as required by both VT-d and AMD-Vi.
Could you please Cc maintainers of all affected subsystems?
E.g. I see changes in VFIO and kvm and neither Paolo nor Alex
were Cc'd.
> David Kiarie (6):
> hw/msi: Allow platform devices to use explicit SID
> hw/i386: enforce SID verification
> hw/iommu: Prepare for AMD IOMMU interrupt remapping
> hw/iommu: AMD IOMMU interrupt remapping
> hw/acpi: report IOAPIC on IVRS
> hw/iommu: share common code between IOMMUs
>
> hw/i386/acpi-build.c | 2 +
> hw/i386/amd_iommu.c | 241 +++++++++++++++++++++++++++++++++++++-
> hw/i386/amd_iommu.h | 82 +++++++++++++
> hw/i386/intel_iommu.c | 89 +++++++-------
> hw/i386/kvm/pci-assign.c | 12 +-
> hw/i386/trace-events | 7 ++
> hw/i386/x86-iommu.c | 8 ++
> hw/intc/ioapic.c | 30 ++++-
> hw/misc/ivshmem.c | 6 +-
> hw/vfio/pci.c | 6 +-
> hw/virtio/virtio-pci.c | 7 +-
> include/hw/i386/ioapic_internal.h | 1 +
> include/hw/i386/x86-iommu.h | 1 +
> include/sysemu/kvm.h | 25 ++--
> kvm-all.c | 10 +-
> kvm-stub.c | 5 +-
> qemu-version.h | 1 +
> target-i386/kvm.c | 15 ++-
> 18 files changed, 462 insertions(+), 86 deletions(-)
> create mode 100644 qemu-version.h
>
> --
> 2.1.4
^ permalink raw reply [flat|nested] 8+ messages in thread