[PATCH 00/17] kvm: drop support for really old kernels

[PATCH 00/17] kvm: drop support for really old kernels

[Paolo Bonzini]

This removes checks for KVM capabilities introduced in very old kernels.
I focused mostly (though not exclusively) on caps that are extremely
old or have a lot of supporting code around them.  The new inferior
limit is Linux 4.4 (or RHEL/CentOS 7) for all architectures except ARM,
for which the limit is Linux 4.19 (or RHEL/CentOS 8).

Note that sometimes older capabilities have evolved and have to
be tested anyway, for example in the case of KVM_CAP_ADJUST_CLOCK
it is possible to remove kvm_has_adjust_clock() but not
kvm_has_adjust_clock_stable().

I intentionally didn't include capabilities that map to guest-visible
features, such as KVM_CAP_TSC_DEADLINE_TIMER, instead focusing more
on very old APIs.

Kernel code technically does not require KVM_CAP_IRQFD to be always
available on architectures that support an in-kernel interrupt controller;
however this is currently true for all supported architectures, and
I plan to submit a kernel patch to ensure this equivalence.

Paolo

Paolo Bonzini (17):
  kvm: remove unnecessary stub
  kvm: require KVM_CAP_INTERNAL_ERROR_DATA
  kvm: require KVM_CAP_SIGNAL_MSI
  kvm: require KVM_CAP_IRQFD for kernel irqchip
  kvm: drop reference to KVM_CAP_PCI_2_3
  kvm: assume that many ioeventfds can be created
  kvm: require KVM_CAP_IOEVENTFD and KVM_CAP_IOEVENTFD_ANY_LENGTH
  kvm: unify listeners for PIO address space
  kvm: arm: require KVM_CAP_SET_VCPU_EVENTS
  kvm: i386: move KVM_CAP_IRQ_ROUTING detection to kvm_arch_required_capabilities
  kvm: i386: require KVM_CAP_DEBUGREGS
  kvm: i386: require KVM_CAP_XSAVE
  kvm: i386: require KVM_CAP_SET_VCPU_EVENTS and KVM_CAP_X86_ROBUST_SINGLESTEP
  kvm: i386: require KVM_CAP_MCE
  kvm: i386: require KVM_CAP_ADJUST_CLOCK
  kvm: i386: require KVM_CAP_SET_IDENTITY_MAP_ADDR
  kvm: i8254: require KVM_CAP_PIT2 and KVM_CAP_PIT_STATE2

 accel/kvm/kvm-all.c            | 239 ++++-----------------------------
 accel/stubs/kvm-stub.c         |  14 --
 hw/i386/kvm/clock.c            |   4 -
 hw/i386/kvm/i8254.c            |  38 ++----
 hw/i386/pc.c                   |   6 +-
 hw/intc/arm_gicv3_its_common.c |   3 +-
 hw/intc/arm_gicv3_its_kvm.c    |   2 +-
 hw/misc/pci-testdev.c          |   3 +-
 hw/s390x/virtio-ccw.c          |   4 -
 hw/virtio/vhost-user.c         |   7 +-
 hw/virtio/virtio-mmio.c        |   4 -
 hw/virtio/virtio-pci.c         |  23 +---
 include/sysemu/kvm.h           |  39 +-----
 include/sysemu/kvm_int.h       |   6 -
 system/memory.c                |  16 +--
 target/arm/kvm.c               |  10 +-
 target/i386/kvm/kvm.c          | 225 +++----------------------------
 target/i386/kvm/kvm_i386.h     |   2 -
 target/riscv/kvm/kvm-cpu.c     |   2 +-
 20 files changed, 77 insertions(+), 572 deletions(-)

-- 
2.41.0

[PATCH 01/17] kvm: remove unnecessary stub

[Paolo Bonzini]

This function is only invoked from hw/intc/s390_flic_kvm.c, and therefore
only if CONFIG_KVM is defined.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 accel/stubs/kvm-stub.c | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/accel/stubs/kvm-stub.c b/accel/stubs/kvm-stub.c
index 51f522e52e8..a323252f8e2 100644
--- a/accel/stubs/kvm-stub.c
+++ b/accel/stubs/kvm-stub.c
@@ -92,11 +92,6 @@ void kvm_irqchip_change_notify(void)
 {
 }
 
-int kvm_irqchip_add_adapter_route(KVMState *s, AdapterInfo *adapter)
-{
-    return -ENOSYS;
-}
-
 int kvm_irqchip_add_irqfd_notifier_gsi(KVMState *s, EventNotifier *n,
                                        EventNotifier *rn, int virq)
 {
-- 
2.41.0

[PATCH 02/17] kvm: require KVM_CAP_INTERNAL_ERROR_DATA

[Paolo Bonzini]

This was introduced in KVM in Linux 2.6.33, we can require it unconditionally.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 accel/kvm/kvm-all.c | 13 ++++++-------
 1 file changed, 6 insertions(+), 7 deletions(-)

diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index 3f7eafe08cb..8eee5042252 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -111,6 +111,7 @@ static const KVMCapabilityInfo kvm_required_capabilites[] = {
     KVM_CAP_INFO(USER_MEMORY),
     KVM_CAP_INFO(DESTROY_MEMORY_REGION_WORKS),
     KVM_CAP_INFO(JOIN_MEMORY_REGIONS_WORKS),
+    KVM_CAP_INFO(INTERNAL_ERROR_DATA),
     KVM_CAP_LAST_INFO
 };
 
@@ -2794,16 +2795,14 @@ static void kvm_handle_io(uint16_t port, MemTxAttrs attrs, void *data, int direc
 
 static int kvm_handle_internal_error(CPUState *cpu, struct kvm_run *run)
 {
+    int i;
+
     fprintf(stderr, "KVM internal error. Suberror: %d\n",
             run->internal.suberror);
 
-    if (kvm_check_extension(kvm_state, KVM_CAP_INTERNAL_ERROR_DATA)) {
-        int i;
-
-        for (i = 0; i < run->internal.ndata; ++i) {
-            fprintf(stderr, "extra data[%d]: 0x%016"PRIx64"\n",
-                    i, (uint64_t)run->internal.data[i]);
-        }
+    for (i = 0; i < run->internal.ndata; ++i) {
+        fprintf(stderr, "extra data[%d]: 0x%016"PRIx64"\n",
+                i, (uint64_t)run->internal.data[i]);
     }
     if (run->internal.suberror == KVM_INTERNAL_ERROR_EMULATION) {
         fprintf(stderr, "emulation failure\n");
-- 
2.41.0

[PATCH 03/17] kvm: require KVM_CAP_SIGNAL_MSI

[Paolo Bonzini]

This was introduced in KVM in Linux 3.5, we can require it unconditionally
in kvm_irqchip_send_msi().  However, not all architectures have to implement
it so check it only in x86, the only architecture that ever had MSI injection
but not KVM_CAP_SIGNAL_MSI.

ARM uses it to detect the presence of the ITS emulation in the kernel,
introduced in Linux 4.8.  Assume that it's there and possibly fail when
realizing the arm-its-kvm device.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 accel/kvm/kvm-all.c            | 102 +++------------------------------
 accel/stubs/kvm-stub.c         |   1 -
 hw/intc/arm_gicv3_its_common.c |   3 +-
 include/sysemu/kvm.h           |   9 ---
 include/sysemu/kvm_int.h       |   1 -
 target/i386/kvm/kvm.c          |   1 +
 6 files changed, 9 insertions(+), 108 deletions(-)

diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index 8eee5042252..0c7b0569da4 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -99,7 +99,6 @@ bool kvm_gsi_direct_mapping;
 bool kvm_allowed;
 bool kvm_readonly_mem_allowed;
 bool kvm_vm_attributes_allowed;
-bool kvm_direct_msi_allowed;
 bool kvm_ioeventfd_any_length_allowed;
 bool kvm_msi_use_devid;
 bool kvm_has_guest_debug;
@@ -1848,7 +1847,7 @@ static void clear_gsi(KVMState *s, unsigned int gsi)
 
 void kvm_init_irq_routing(KVMState *s)
 {
-    int gsi_count, i;
+    int gsi_count;
 
     gsi_count = kvm_check_extension(s, KVM_CAP_IRQ_ROUTING) - 1;
     if (gsi_count > 0) {
@@ -1860,12 +1859,6 @@ void kvm_init_irq_routing(KVMState *s)
     s->irq_routes = g_malloc0(sizeof(*s->irq_routes));
     s->nr_allocated_irq_routes = 0;
 
-    if (!kvm_direct_msi_allowed) {
-        for (i = 0; i < KVM_MSI_HASHTAB_SIZE; i++) {
-            QTAILQ_INIT(&s->msi_hashtab[i]);
-        }
-    }
-
     kvm_arch_init_irq_routing(s);
 }
 
@@ -1985,41 +1978,10 @@ void kvm_irqchip_change_notify(void)
     notifier_list_notify(&kvm_irqchip_change_notifiers, NULL);
 }
 
-static unsigned int kvm_hash_msi(uint32_t data)
-{
-    /* This is optimized for IA32 MSI layout. However, no other arch shall
-     * repeat the mistake of not providing a direct MSI injection API. */
-    return data & 0xff;
-}
-
-static void kvm_flush_dynamic_msi_routes(KVMState *s)
-{
-    KVMMSIRoute *route, *next;
-    unsigned int hash;
-
-    for (hash = 0; hash < KVM_MSI_HASHTAB_SIZE; hash++) {
-        QTAILQ_FOREACH_SAFE(route, &s->msi_hashtab[hash], entry, next) {
-            kvm_irqchip_release_virq(s, route->kroute.gsi);
-            QTAILQ_REMOVE(&s->msi_hashtab[hash], route, entry);
-            g_free(route);
-        }
-    }
-}
-
 static int kvm_irqchip_get_virq(KVMState *s)
 {
     int next_virq;
 
-    /*
-     * PIC and IOAPIC share the first 16 GSI numbers, thus the available
-     * GSI numbers are more than the number of IRQ route. Allocating a GSI
-     * number can succeed even though a new route entry cannot be added.
-     * When this happens, flush dynamic MSI entries to free IRQ route entries.
-     */
-    if (!kvm_direct_msi_allowed && s->irq_routes->nr == s->gsi_count) {
-        kvm_flush_dynamic_msi_routes(s);
-    }
-
     /* Return the lowest unused GSI in the bitmap */
     next_virq = find_first_zero_bit(s->used_gsi_bitmap, s->gsi_count);
     if (next_virq >= s->gsi_count) {
@@ -2029,63 +1991,17 @@ static int kvm_irqchip_get_virq(KVMState *s)
     }
 }
 
-static KVMMSIRoute *kvm_lookup_msi_route(KVMState *s, MSIMessage msg)
-{
-    unsigned int hash = kvm_hash_msi(msg.data);
-    KVMMSIRoute *route;
-
-    QTAILQ_FOREACH(route, &s->msi_hashtab[hash], entry) {
-        if (route->kroute.u.msi.address_lo == (uint32_t)msg.address &&
-            route->kroute.u.msi.address_hi == (msg.address >> 32) &&
-            route->kroute.u.msi.data == le32_to_cpu(msg.data)) {
-            return route;
-        }
-    }
-    return NULL;
-}
-
 int kvm_irqchip_send_msi(KVMState *s, MSIMessage msg)
 {
     struct kvm_msi msi;
-    KVMMSIRoute *route;
 
-    if (kvm_direct_msi_allowed) {
-        msi.address_lo = (uint32_t)msg.address;
-        msi.address_hi = msg.address >> 32;
-        msi.data = le32_to_cpu(msg.data);
-        msi.flags = 0;
-        memset(msi.pad, 0, sizeof(msi.pad));
+    msi.address_lo = (uint32_t)msg.address;
+    msi.address_hi = msg.address >> 32;
+    msi.data = le32_to_cpu(msg.data);
+    msi.flags = 0;
+    memset(msi.pad, 0, sizeof(msi.pad));
 
-        return kvm_vm_ioctl(s, KVM_SIGNAL_MSI, &msi);
-    }
-
-    route = kvm_lookup_msi_route(s, msg);
-    if (!route) {
-        int virq;
-
-        virq = kvm_irqchip_get_virq(s);
-        if (virq < 0) {
-            return virq;
-        }
-
-        route = g_new0(KVMMSIRoute, 1);
-        route->kroute.gsi = virq;
-        route->kroute.type = KVM_IRQ_ROUTING_MSI;
-        route->kroute.flags = 0;
-        route->kroute.u.msi.address_lo = (uint32_t)msg.address;
-        route->kroute.u.msi.address_hi = msg.address >> 32;
-        route->kroute.u.msi.data = le32_to_cpu(msg.data);
-
-        kvm_add_routing_entry(s, &route->kroute);
-        kvm_irqchip_commit_routes(s);
-
-        QTAILQ_INSERT_TAIL(&s->msi_hashtab[kvm_hash_msi(msg.data)], route,
-                           entry);
-    }
-
-    assert(route->kroute.type == KVM_IRQ_ROUTING_MSI);
-
-    return kvm_set_irq(s, route->kroute.gsi, 1);
+    return kvm_vm_ioctl(s, KVM_SIGNAL_MSI, &msi);
 }
 
 int kvm_irqchip_add_msi_route(KVMRouteChange *c, int vector, PCIDevice *dev)
@@ -2660,10 +2576,6 @@ static int kvm_init(MachineState *ms)
 
     s->max_nested_state_len = kvm_check_extension(s, KVM_CAP_NESTED_STATE);
 
-#ifdef KVM_CAP_IRQ_ROUTING
-    kvm_direct_msi_allowed = (kvm_check_extension(s, KVM_CAP_SIGNAL_MSI) > 0);
-#endif
-
     s->intx_set_mask = kvm_check_extension(s, KVM_CAP_PCI_2_3);
 
     s->irq_set_ioctl = KVM_IRQ_LINE;
diff --git a/accel/stubs/kvm-stub.c b/accel/stubs/kvm-stub.c
index a323252f8e2..bce005adad8 100644
--- a/accel/stubs/kvm-stub.c
+++ b/accel/stubs/kvm-stub.c
@@ -27,7 +27,6 @@ bool kvm_allowed;
 bool kvm_readonly_mem_allowed;
 bool kvm_ioeventfd_any_length_allowed;
 bool kvm_msi_use_devid;
-bool kvm_direct_msi_allowed;
 
 void kvm_flush_coalesced_mmio_buffer(void)
 {
diff --git a/hw/intc/arm_gicv3_its_common.c b/hw/intc/arm_gicv3_its_common.c
index abaf77057e1..fddd6d490c2 100644
--- a/hw/intc/arm_gicv3_its_common.c
+++ b/hw/intc/arm_gicv3_its_common.c
@@ -163,8 +163,7 @@ type_init(gicv3_its_common_register_types)
 const char *its_class_name(void)
 {
     if (kvm_irqchip_in_kernel()) {
-        /* KVM implementation requires this capability */
-        return kvm_direct_msi_enabled() ? "arm-its-kvm" : NULL;
+        return "arm-its-kvm";
     } else {
         /* Software emulation based model */
         return "arm-gicv3-its";
diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h
index 97a8a4f201c..93dccf5dd92 100644
--- a/include/sysemu/kvm.h
+++ b/include/sysemu/kvm.h
@@ -43,7 +43,6 @@ extern bool kvm_msi_via_irqfd_allowed;
 extern bool kvm_gsi_routing_allowed;
 extern bool kvm_gsi_direct_mapping;
 extern bool kvm_readonly_mem_allowed;
-extern bool kvm_direct_msi_allowed;
 extern bool kvm_ioeventfd_any_length_allowed;
 extern bool kvm_msi_use_devid;
 
@@ -147,13 +146,6 @@ extern bool kvm_msi_use_devid;
  */
 #define kvm_readonly_mem_enabled() (kvm_readonly_mem_allowed)
 
-/**
- * kvm_direct_msi_enabled:
- *
- * Returns: true if KVM allows direct MSI injection.
- */
-#define kvm_direct_msi_enabled() (kvm_direct_msi_allowed)
-
 /**
  * kvm_ioeventfd_any_length_enabled:
  * Returns: true if KVM allows any length io eventfd.
@@ -181,7 +173,6 @@ extern bool kvm_msi_use_devid;
 #define kvm_gsi_routing_allowed() (false)
 #define kvm_gsi_direct_mapping() (false)
 #define kvm_readonly_mem_enabled() (false)
-#define kvm_direct_msi_enabled() (false)
 #define kvm_ioeventfd_any_length_enabled() (false)
 #define kvm_msi_devid_required() (false)
 
diff --git a/include/sysemu/kvm_int.h b/include/sysemu/kvm_int.h
index 075939a3c4f..a7dacd12d6c 100644
--- a/include/sysemu/kvm_int.h
+++ b/include/sysemu/kvm_int.h
@@ -103,7 +103,6 @@ struct KVMState
     int nr_allocated_irq_routes;
     unsigned long *used_gsi_bitmap;
     unsigned int gsi_count;
-    QTAILQ_HEAD(, KVMMSIRoute) msi_hashtab[KVM_MSI_HASHTAB_SIZE];
 #endif
     KVMMemoryListener memory_listener;
     QLIST_HEAD(, KVMParkedVcpu) kvm_parked_vcpus;
diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c
index ab72bcdfad1..ab7d39d5399 100644
--- a/target/i386/kvm/kvm.c
+++ b/target/i386/kvm/kvm.c
@@ -91,6 +91,7 @@ const KVMCapabilityInfo kvm_arch_required_capabilities[] = {
     KVM_CAP_INFO(SET_TSS_ADDR),
     KVM_CAP_INFO(EXT_CPUID),
     KVM_CAP_INFO(MP_STATE),
+    KVM_CAP_INFO(SIGNAL_MSI),
     KVM_CAP_LAST_INFO
 };
 
-- 
2.41.0

[PATCH 04/17] kvm: require KVM_CAP_IRQFD for kernel irqchip

[Paolo Bonzini]

KVM_CAP_IRQFD is always available on architectures that support an in-kernel
interrupt controller, and was introduced in Linux 2.6.32.  We can require
it unconditionally.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 accel/kvm/kvm-all.c         | 13 +++++--------
 accel/stubs/kvm-stub.c      |  1 -
 hw/intc/arm_gicv3_its_kvm.c |  2 +-
 include/sysemu/kvm.h        |  6 +++---
 target/riscv/kvm/kvm-cpu.c  |  2 +-
 5 files changed, 10 insertions(+), 14 deletions(-)

diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index 0c7b0569da4..be50d47f7b4 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -91,7 +91,6 @@ bool kvm_split_irqchip;
 bool kvm_async_interrupts_allowed;
 bool kvm_halt_in_kernel_allowed;
 bool kvm_eventfds_allowed;
-bool kvm_irqfds_allowed;
 bool kvm_resamplefds_allowed;
 bool kvm_msi_via_irqfd_allowed;
 bool kvm_gsi_routing_allowed;
@@ -2128,10 +2127,6 @@ static int kvm_irqchip_assign_irqfd(KVMState *s, EventNotifier *event,
         }
     }
 
-    if (!kvm_irqfds_enabled()) {
-        return -ENOSYS;
-    }
-
     return kvm_vm_ioctl(s, KVM_IRQFD, &irqfd);
 }
 
@@ -2292,6 +2287,11 @@ static void kvm_irqchip_create(KVMState *s)
         return;
     }
 
+    if (kvm_check_extension(s, KVM_CAP_IRQFD) <= 0) {
+        fprintf(stderr, "kvm: irqfd not implemented\n");
+        exit(1);
+    }
+
     /* First probe and see if there's a arch-specific hook to create the
      * in-kernel irqchip for us */
     ret = kvm_arch_irqchip_create(s);
@@ -2589,9 +2589,6 @@ static int kvm_init(MachineState *ms)
     kvm_eventfds_allowed =
         (kvm_check_extension(s, KVM_CAP_IOEVENTFD) > 0);
 
-    kvm_irqfds_allowed =
-        (kvm_check_extension(s, KVM_CAP_IRQFD) > 0);
-
     kvm_resamplefds_allowed =
         (kvm_check_extension(s, KVM_CAP_IRQFD_RESAMPLE) > 0);
 
diff --git a/accel/stubs/kvm-stub.c b/accel/stubs/kvm-stub.c
index bce005adad8..19d58f2778f 100644
--- a/accel/stubs/kvm-stub.c
+++ b/accel/stubs/kvm-stub.c
@@ -18,7 +18,6 @@ KVMState *kvm_state;
 bool kvm_kernel_irqchip;
 bool kvm_async_interrupts_allowed;
 bool kvm_eventfds_allowed;
-bool kvm_irqfds_allowed;
 bool kvm_resamplefds_allowed;
 bool kvm_msi_via_irqfd_allowed;
 bool kvm_gsi_routing_allowed;
diff --git a/hw/intc/arm_gicv3_its_kvm.c b/hw/intc/arm_gicv3_its_kvm.c
index 7eda9fb86ea..178f6d7de90 100644
--- a/hw/intc/arm_gicv3_its_kvm.c
+++ b/hw/intc/arm_gicv3_its_kvm.c
@@ -124,7 +124,7 @@ static void kvm_arm_its_realize(DeviceState *dev, Error **errp)
 
     kvm_msi_use_devid = true;
     kvm_gsi_direct_mapping = false;
-    kvm_msi_via_irqfd_allowed = kvm_irqfds_enabled();
+    kvm_msi_via_irqfd_allowed = true;
 }
 
 /**
diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h
index 93dccf5dd92..575dee53b39 100644
--- a/include/sysemu/kvm.h
+++ b/include/sysemu/kvm.h
@@ -37,7 +37,6 @@ extern bool kvm_split_irqchip;
 extern bool kvm_async_interrupts_allowed;
 extern bool kvm_halt_in_kernel_allowed;
 extern bool kvm_eventfds_allowed;
-extern bool kvm_irqfds_allowed;
 extern bool kvm_resamplefds_allowed;
 extern bool kvm_msi_via_irqfd_allowed;
 extern bool kvm_gsi_routing_allowed;
@@ -102,8 +101,10 @@ extern bool kvm_msi_use_devid;
  * Returns: true if we can use irqfds to inject interrupts into
  * a KVM CPU (ie the kernel supports irqfds and we are running
  * with a configuration where it is meaningful to use them).
+ *
+ * Always available if running with in-kernel irqchip.
  */
-#define kvm_irqfds_enabled() (kvm_irqfds_allowed)
+#define kvm_irqfds_enabled() kvm_irqchip_in_kernel()
 
 /**
  * kvm_resamplefds_enabled:
@@ -167,7 +168,6 @@ extern bool kvm_msi_use_devid;
 #define kvm_async_interrupts_enabled() (false)
 #define kvm_halt_in_kernel() (false)
 #define kvm_eventfds_enabled() (false)
-#define kvm_irqfds_enabled() (false)
 #define kvm_resamplefds_enabled() (false)
 #define kvm_msi_via_irqfd_enabled() (false)
 #define kvm_gsi_routing_allowed() (false)
diff --git a/target/riscv/kvm/kvm-cpu.c b/target/riscv/kvm/kvm-cpu.c
index 090d6176275..26e68c7ab45 100644
--- a/target/riscv/kvm/kvm-cpu.c
+++ b/target/riscv/kvm/kvm-cpu.c
@@ -1420,7 +1420,7 @@ void kvm_riscv_aia_create(MachineState *machine, uint64_t group_shift,
         exit(1);
     }
 
-    kvm_msi_via_irqfd_allowed = kvm_irqfds_enabled();
+    kvm_msi_via_irqfd_allowed = true;
 }
 
 static void kvm_cpu_instance_init(CPUState *cs)
-- 
2.41.0

[PATCH 05/17] kvm: drop reference to KVM_CAP_PCI_2_3

[Paolo Bonzini]

This is a remnant of pre-VFIO device assignment; it is not defined
anymore by Linux and not used by QEMU.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 accel/kvm/kvm-all.c      | 7 -------
 include/sysemu/kvm.h     | 1 -
 include/sysemu/kvm_int.h | 1 -
 3 files changed, 9 deletions(-)

diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index be50d47f7b4..50717a0d634 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -2576,8 +2576,6 @@ static int kvm_init(MachineState *ms)
 
     s->max_nested_state_len = kvm_check_extension(s, KVM_CAP_NESTED_STATE);
 
-    s->intx_set_mask = kvm_check_extension(s, KVM_CAP_PCI_2_3);
-
     s->irq_set_ioctl = KVM_IRQ_LINE;
     if (kvm_check_extension(s, KVM_CAP_IRQ_INJECT_STATUS)) {
         s->irq_set_ioctl = KVM_IRQ_LINE_STATUS;
@@ -3237,11 +3235,6 @@ int kvm_has_gsi_routing(void)
 #endif
 }
 
-int kvm_has_intx_set_mask(void)
-{
-    return kvm_state->intx_set_mask;
-}
-
 bool kvm_arm_supports_user_irq(void)
 {
     return kvm_check_extension(kvm_state, KVM_CAP_ARM_USER_IRQ);
diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h
index 575dee53b39..aa299396fba 100644
--- a/include/sysemu/kvm.h
+++ b/include/sysemu/kvm.h
@@ -215,7 +215,6 @@ int kvm_has_debugregs(void);
 int kvm_max_nested_state_length(void);
 int kvm_has_many_ioeventfds(void);
 int kvm_has_gsi_routing(void);
-int kvm_has_intx_set_mask(void);
 
 /**
  * kvm_arm_supports_user_irq
diff --git a/include/sysemu/kvm_int.h b/include/sysemu/kvm_int.h
index a7dacd12d6c..817238b9582 100644
--- a/include/sysemu/kvm_int.h
+++ b/include/sysemu/kvm_int.h
@@ -85,7 +85,6 @@ struct KVMState
 #endif
     int max_nested_state_len;
     int many_ioeventfds;
-    int intx_set_mask;
     int kvm_shadow_mem;
     bool kernel_irqchip_allowed;
     bool kernel_irqchip_required;
-- 
2.41.0

[PATCH 06/17] kvm: assume that many ioeventfds can be created

[Paolo Bonzini]

NR_IOBUS_DEVS was increased to 200 in Linux 2.6.34.  By Linux 3.5 it had
increased to 1000 and later ioeventfds were changed to not count against
the limit.  But the earlier limit of 200 would already be enough for
kvm_check_many_ioeventfds() to be true, so remove the check.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 accel/kvm/kvm-all.c      | 47 ----------------------------------------
 accel/stubs/kvm-stub.c   |  5 -----
 hw/virtio/virtio-pci.c   |  4 ----
 include/sysemu/kvm.h     |  1 -
 include/sysemu/kvm_int.h |  1 -
 5 files changed, 58 deletions(-)

diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index 50717a0d634..05be687be1b 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -1253,43 +1253,6 @@ static int kvm_set_ioeventfd_pio(int fd, uint16_t addr, uint16_t val,
 }
 
 
-static int kvm_check_many_ioeventfds(void)
-{
-    /* Userspace can use ioeventfd for io notification.  This requires a host
-     * that supports eventfd(2) and an I/O thread; since eventfd does not
-     * support SIGIO it cannot interrupt the vcpu.
-     *
-     * Older kernels have a 6 device limit on the KVM io bus.  Find out so we
-     * can avoid creating too many ioeventfds.
-     */
-#if defined(CONFIG_EVENTFD)
-    int ioeventfds[7];
-    int i, ret = 0;
-    for (i = 0; i < ARRAY_SIZE(ioeventfds); i++) {
-        ioeventfds[i] = eventfd(0, EFD_CLOEXEC);
-        if (ioeventfds[i] < 0) {
-            break;
-        }
-        ret = kvm_set_ioeventfd_pio(ioeventfds[i], 0, i, true, 2, true);
-        if (ret < 0) {
-            close(ioeventfds[i]);
-            break;
-        }
-    }
-
-    /* Decide whether many devices are supported or not */
-    ret = i == ARRAY_SIZE(ioeventfds);
-
-    while (i-- > 0) {
-        kvm_set_ioeventfd_pio(ioeventfds[i], 0, i, false, 2, true);
-        close(ioeventfds[i]);
-    }
-    return ret;
-#else
-    return 0;
-#endif
-}
-
 static const KVMCapabilityInfo *
 kvm_check_extension_list(KVMState *s, const KVMCapabilityInfo *list)
 {
@@ -2648,8 +2611,6 @@ static int kvm_init(MachineState *ms)
     memory_listener_register(&kvm_coalesced_pio_listener,
                              &address_space_io);
 
-    s->many_ioeventfds = kvm_check_many_ioeventfds();
-
     s->sync_mmu = !!kvm_vm_check_extension(kvm_state, KVM_CAP_SYNC_MMU);
     if (!s->sync_mmu) {
         ret = ram_block_discard_disable(true);
@@ -3218,14 +3179,6 @@ int kvm_max_nested_state_length(void)
     return kvm_state->max_nested_state_len;
 }
 
-int kvm_has_many_ioeventfds(void)
-{
-    if (!kvm_enabled()) {
-        return 0;
-    }
-    return kvm_state->many_ioeventfds;
-}
-
 int kvm_has_gsi_routing(void)
 {
 #ifdef KVM_CAP_IRQ_ROUTING
diff --git a/accel/stubs/kvm-stub.c b/accel/stubs/kvm-stub.c
index 19d58f2778f..b2d8885853f 100644
--- a/accel/stubs/kvm-stub.c
+++ b/accel/stubs/kvm-stub.c
@@ -40,11 +40,6 @@ bool kvm_has_sync_mmu(void)
     return false;
 }
 
-int kvm_has_many_ioeventfds(void)
-{
-    return 0;
-}
-
 int kvm_on_sigbus_vcpu(CPUState *cpu, int code, void *addr)
 {
     return 1;
diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
index abebd0075a6..a851c67bdd3 100644
--- a/hw/virtio/virtio-pci.c
+++ b/hw/virtio/virtio-pci.c
@@ -2096,10 +2096,6 @@ static void virtio_pci_realize(PCIDevice *pci_dev, Error **errp)
     bool pcie_port = pci_bus_is_express(pci_get_bus(pci_dev)) &&
                      !pci_bus_is_root(pci_get_bus(pci_dev));
 
-    if (kvm_enabled() && !kvm_has_many_ioeventfds()) {
-        proxy->flags &= ~VIRTIO_PCI_FLAG_USE_IOEVENTFD;
-    }
-
     /* fd-based ioevents can't be synchronized in record/replay */
     if (replay_mode != REPLAY_MODE_NONE) {
         proxy->flags &= ~VIRTIO_PCI_FLAG_USE_IOEVENTFD;
diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h
index aa299396fba..15c427f6e09 100644
--- a/include/sysemu/kvm.h
+++ b/include/sysemu/kvm.h
@@ -213,7 +213,6 @@ int kvm_has_vcpu_events(void);
 int kvm_has_robust_singlestep(void);
 int kvm_has_debugregs(void);
 int kvm_max_nested_state_length(void);
-int kvm_has_many_ioeventfds(void);
 int kvm_has_gsi_routing(void);
 
 /**
diff --git a/include/sysemu/kvm_int.h b/include/sysemu/kvm_int.h
index 817238b9582..840b905a2e9 100644
--- a/include/sysemu/kvm_int.h
+++ b/include/sysemu/kvm_int.h
@@ -84,7 +84,6 @@ struct KVMState
     QTAILQ_HEAD(, kvm_sw_breakpoint) kvm_sw_breakpoints;
 #endif
     int max_nested_state_len;
-    int many_ioeventfds;
     int kvm_shadow_mem;
     bool kernel_irqchip_allowed;
     bool kernel_irqchip_required;
-- 
2.41.0

[PATCH 07/17] kvm: require KVM_CAP_IOEVENTFD and KVM_CAP_IOEVENTFD_ANY_LENGTH

[Paolo Bonzini]

KVM_CAP_IOEVENTFD_ANY_LENGTH was added in Linux 4.4, released in 2016.
Assume that it is present.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 accel/kvm/kvm-all.c     | 22 ++++++----------------
 accel/stubs/kvm-stub.c  |  2 --
 hw/misc/pci-testdev.c   |  3 +--
 hw/s390x/virtio-ccw.c   |  4 ----
 hw/virtio/vhost-user.c  |  7 +------
 hw/virtio/virtio-mmio.c |  4 ----
 hw/virtio/virtio-pci.c  | 19 ++++---------------
 include/sysemu/kvm.h    | 19 -------------------
 system/memory.c         | 16 ++++++----------
 9 files changed, 18 insertions(+), 78 deletions(-)

diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index 05be687be1b..120051da649 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -90,7 +90,6 @@ bool kvm_kernel_irqchip;
 bool kvm_split_irqchip;
 bool kvm_async_interrupts_allowed;
 bool kvm_halt_in_kernel_allowed;
-bool kvm_eventfds_allowed;
 bool kvm_resamplefds_allowed;
 bool kvm_msi_via_irqfd_allowed;
 bool kvm_gsi_routing_allowed;
@@ -98,7 +97,6 @@ bool kvm_gsi_direct_mapping;
 bool kvm_allowed;
 bool kvm_readonly_mem_allowed;
 bool kvm_vm_attributes_allowed;
-bool kvm_ioeventfd_any_length_allowed;
 bool kvm_msi_use_devid;
 bool kvm_has_guest_debug;
 static int kvm_sstep_flags;
@@ -110,6 +108,8 @@ static const KVMCapabilityInfo kvm_required_capabilites[] = {
     KVM_CAP_INFO(DESTROY_MEMORY_REGION_WORKS),
     KVM_CAP_INFO(JOIN_MEMORY_REGIONS_WORKS),
     KVM_CAP_INFO(INTERNAL_ERROR_DATA),
+    KVM_CAP_INFO(IOEVENTFD),
+    KVM_CAP_INFO(IOEVENTFD_ANY_LENGTH),
     KVM_CAP_LAST_INFO
 };
 
@@ -2547,18 +2547,12 @@ static int kvm_init(MachineState *ms)
     kvm_readonly_mem_allowed =
         (kvm_check_extension(s, KVM_CAP_READONLY_MEM) > 0);
 
-    kvm_eventfds_allowed =
-        (kvm_check_extension(s, KVM_CAP_IOEVENTFD) > 0);
-
     kvm_resamplefds_allowed =
         (kvm_check_extension(s, KVM_CAP_IRQFD_RESAMPLE) > 0);
 
     kvm_vm_attributes_allowed =
         (kvm_check_extension(s, KVM_CAP_VM_ATTRIBUTES) > 0);
 
-    kvm_ioeventfd_any_length_allowed =
-        (kvm_check_extension(s, KVM_CAP_IOEVENTFD_ANY_LENGTH) > 0);
-
 #ifdef KVM_CAP_SET_GUEST_DEBUG
     kvm_has_guest_debug =
         (kvm_check_extension(s, KVM_CAP_SET_GUEST_DEBUG) > 0);
@@ -2595,19 +2589,15 @@ static int kvm_init(MachineState *ms)
         kvm_irqchip_create(s);
     }
 
-    if (kvm_eventfds_allowed) {
-        s->memory_listener.listener.eventfd_add = kvm_mem_ioeventfd_add;
-        s->memory_listener.listener.eventfd_del = kvm_mem_ioeventfd_del;
-    }
+    s->memory_listener.listener.eventfd_add = kvm_mem_ioeventfd_add;
+    s->memory_listener.listener.eventfd_del = kvm_mem_ioeventfd_del;
     s->memory_listener.listener.coalesced_io_add = kvm_coalesce_mmio_region;
     s->memory_listener.listener.coalesced_io_del = kvm_uncoalesce_mmio_region;
 
     kvm_memory_listener_register(s, &s->memory_listener,
                                  &address_space_memory, 0, "kvm-memory");
-    if (kvm_eventfds_allowed) {
-        memory_listener_register(&kvm_io_listener,
-                                 &address_space_io);
-    }
+    memory_listener_register(&kvm_io_listener,
+                             &address_space_io);
     memory_listener_register(&kvm_coalesced_pio_listener,
                              &address_space_io);
 
diff --git a/accel/stubs/kvm-stub.c b/accel/stubs/kvm-stub.c
index b2d8885853f..1b37d9a302c 100644
--- a/accel/stubs/kvm-stub.c
+++ b/accel/stubs/kvm-stub.c
@@ -17,14 +17,12 @@
 KVMState *kvm_state;
 bool kvm_kernel_irqchip;
 bool kvm_async_interrupts_allowed;
-bool kvm_eventfds_allowed;
 bool kvm_resamplefds_allowed;
 bool kvm_msi_via_irqfd_allowed;
 bool kvm_gsi_routing_allowed;
 bool kvm_gsi_direct_mapping;
 bool kvm_allowed;
 bool kvm_readonly_mem_allowed;
-bool kvm_ioeventfd_any_length_allowed;
 bool kvm_msi_use_devid;
 
 void kvm_flush_coalesced_mmio_buffer(void)
diff --git a/hw/misc/pci-testdev.c b/hw/misc/pci-testdev.c
index 49303134e4c..acedd0f82bc 100644
--- a/hw/misc/pci-testdev.c
+++ b/hw/misc/pci-testdev.c
@@ -245,7 +245,6 @@ static void pci_testdev_realize(PCIDevice *pci_dev, Error **errp)
     uint8_t *pci_conf;
     char *name;
     int r, i;
-    bool fastmmio = kvm_ioeventfd_any_length_enabled();
 
     pci_conf = pci_dev->config;
 
@@ -279,7 +278,7 @@ static void pci_testdev_realize(PCIDevice *pci_dev, Error **errp)
         g_free(name);
         test->hdr->offset = cpu_to_le32(IOTEST_SIZE(i) + i * IOTEST_ACCESS_WIDTH);
         test->match_data = strcmp(IOTEST_TEST(i), "wildcard-eventfd");
-        if (fastmmio && IOTEST_IS_MEM(i) && !test->match_data) {
+        if (IOTEST_IS_MEM(i) && !test->match_data) {
             test->size = 0;
         } else {
             test->size = IOTEST_ACCESS_WIDTH;
diff --git a/hw/s390x/virtio-ccw.c b/hw/s390x/virtio-ccw.c
index 17c548b84ff..80453718a33 100644
--- a/hw/s390x/virtio-ccw.c
+++ b/hw/s390x/virtio-ccw.c
@@ -768,10 +768,6 @@ static void virtio_ccw_device_realize(VirtioCcwDevice *dev, Error **errp)
         sch->cssid, sch->ssid, sch->schid, sch->devno,
         ccw_dev->devno.valid ? "user-configured" : "auto-configured");
 
-    if (kvm_enabled() && !kvm_eventfds_enabled()) {
-        dev->flags &= ~VIRTIO_CCW_FLAG_USE_IOEVENTFD;
-    }
-
     /* fd-based ioevents can't be synchronized in record/replay */
     if (replay_mode != REPLAY_MODE_NONE) {
         dev->flags &= ~VIRTIO_CCW_FLAG_USE_IOEVENTFD;
diff --git a/hw/virtio/vhost-user.c b/hw/virtio/vhost-user.c
index 68eb1f0c99c..873d2b2eaae 100644
--- a/hw/virtio/vhost-user.c
+++ b/hw/virtio/vhost-user.c
@@ -264,11 +264,6 @@ struct scrub_regions {
     int fd_idx;
 };
 
-static bool ioeventfd_enabled(void)
-{
-    return !kvm_enabled() || kvm_eventfds_enabled();
-}
-
 static int vhost_user_read_header(struct vhost_dev *dev, VhostUserMsg *msg)
 {
     struct vhost_user *u = dev->opaque;
@@ -1218,7 +1213,7 @@ static int vhost_set_vring_file(struct vhost_dev *dev,
         .hdr.size = sizeof(msg.payload.u64),
     };
 
-    if (ioeventfd_enabled() && file->fd > 0) {
+    if (file->fd > 0) {
         fds[fd_num++] = file->fd;
     } else {
         msg.payload.u64 |= VHOST_USER_VRING_NOFD_MASK;
diff --git a/hw/virtio/virtio-mmio.c b/hw/virtio/virtio-mmio.c
index c2c6d854750..22f15e1e02f 100644
--- a/hw/virtio/virtio-mmio.c
+++ b/hw/virtio/virtio-mmio.c
@@ -761,10 +761,6 @@ static void virtio_mmio_realizefn(DeviceState *d, Error **errp)
     qbus_init(&proxy->bus, sizeof(proxy->bus), TYPE_VIRTIO_MMIO_BUS, d, NULL);
     sysbus_init_irq(sbd, &proxy->irq);
 
-    if (!kvm_eventfds_enabled()) {
-        proxy->flags &= ~VIRTIO_IOMMIO_FLAG_USE_IOEVENTFD;
-    }
-
     /* fd-based ioevents can't be synchronized in record/replay */
     if (replay_mode != REPLAY_MODE_NONE) {
         proxy->flags &= ~VIRTIO_IOMMIO_FLAG_USE_IOEVENTFD;
diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
index a851c67bdd3..4af5e146c6a 100644
--- a/hw/virtio/virtio-pci.c
+++ b/hw/virtio/virtio-pci.c
@@ -332,7 +332,6 @@ static int virtio_pci_ioeventfd_assign(DeviceState *d, EventNotifier *notifier,
     VirtQueue *vq = virtio_get_queue(vdev, n);
     bool legacy = virtio_pci_legacy(proxy);
     bool modern = virtio_pci_modern(proxy);
-    bool fast_mmio = kvm_ioeventfd_any_length_enabled();
     bool modern_pio = proxy->flags & VIRTIO_PCI_FLAG_MODERN_PIO_NOTIFY;
     MemoryRegion *modern_mr = &proxy->notify.mr;
     MemoryRegion *modern_notify_mr = &proxy->notify_pio.mr;
@@ -343,13 +342,8 @@ static int virtio_pci_ioeventfd_assign(DeviceState *d, EventNotifier *notifier,
 
     if (assign) {
         if (modern) {
-            if (fast_mmio) {
-                memory_region_add_eventfd(modern_mr, modern_addr, 0,
-                                          false, n, notifier);
-            } else {
-                memory_region_add_eventfd(modern_mr, modern_addr, 2,
-                                          false, n, notifier);
-            }
+            memory_region_add_eventfd(modern_mr, modern_addr, 0,
+                                      false, n, notifier);
             if (modern_pio) {
                 memory_region_add_eventfd(modern_notify_mr, 0, 2,
                                               true, n, notifier);
@@ -361,13 +355,8 @@ static int virtio_pci_ioeventfd_assign(DeviceState *d, EventNotifier *notifier,
         }
     } else {
         if (modern) {
-            if (fast_mmio) {
-                memory_region_del_eventfd(modern_mr, modern_addr, 0,
-                                          false, n, notifier);
-            } else {
-                memory_region_del_eventfd(modern_mr, modern_addr, 2,
-                                          false, n, notifier);
-            }
+            memory_region_del_eventfd(modern_mr, modern_addr, 0,
+                                      false, n, notifier);
             if (modern_pio) {
                 memory_region_del_eventfd(modern_notify_mr, 0, 2,
                                           true, n, notifier);
diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h
index 15c427f6e09..c06982fd64c 100644
--- a/include/sysemu/kvm.h
+++ b/include/sysemu/kvm.h
@@ -36,13 +36,11 @@ extern bool kvm_kernel_irqchip;
 extern bool kvm_split_irqchip;
 extern bool kvm_async_interrupts_allowed;
 extern bool kvm_halt_in_kernel_allowed;
-extern bool kvm_eventfds_allowed;
 extern bool kvm_resamplefds_allowed;
 extern bool kvm_msi_via_irqfd_allowed;
 extern bool kvm_gsi_routing_allowed;
 extern bool kvm_gsi_direct_mapping;
 extern bool kvm_readonly_mem_allowed;
-extern bool kvm_ioeventfd_any_length_allowed;
 extern bool kvm_msi_use_devid;
 
 #define kvm_enabled()           (kvm_allowed)
@@ -86,15 +84,6 @@ extern bool kvm_msi_use_devid;
  */
 #define kvm_halt_in_kernel() (kvm_halt_in_kernel_allowed)
 
-/**
- * kvm_eventfds_enabled:
- *
- * Returns: true if we can use eventfds to receive notifications
- * from a KVM CPU (ie the kernel supports eventds and we are running
- * with a configuration where it is meaningful to use them).
- */
-#define kvm_eventfds_enabled() (kvm_eventfds_allowed)
-
 /**
  * kvm_irqfds_enabled:
  *
@@ -147,12 +136,6 @@ extern bool kvm_msi_use_devid;
  */
 #define kvm_readonly_mem_enabled() (kvm_readonly_mem_allowed)
 
-/**
- * kvm_ioeventfd_any_length_enabled:
- * Returns: true if KVM allows any length io eventfd.
- */
-#define kvm_ioeventfd_any_length_enabled() (kvm_ioeventfd_any_length_allowed)
-
 /**
  * kvm_msi_devid_required:
  * Returns: true if KVM requires a device id to be provided while
@@ -167,13 +150,11 @@ extern bool kvm_msi_use_devid;
 #define kvm_irqchip_is_split() (false)
 #define kvm_async_interrupts_enabled() (false)
 #define kvm_halt_in_kernel() (false)
-#define kvm_eventfds_enabled() (false)
 #define kvm_resamplefds_enabled() (false)
 #define kvm_msi_via_irqfd_enabled() (false)
 #define kvm_gsi_routing_allowed() (false)
 #define kvm_gsi_direct_mapping() (false)
 #define kvm_readonly_mem_enabled() (false)
-#define kvm_ioeventfd_any_length_enabled() (false)
 #define kvm_msi_devid_required() (false)
 
 #endif  /* CONFIG_KVM_IS_POSSIBLE */
diff --git a/system/memory.c b/system/memory.c
index a800fbc9e59..4928f2525de 100644
--- a/system/memory.c
+++ b/system/memory.c
@@ -1535,7 +1535,12 @@ MemTxResult memory_region_dispatch_write(MemoryRegion *mr,
 
     adjust_endianness(mr, &data, op);
 
-    if ((!kvm_eventfds_enabled()) &&
+    /*
+     * FIXME: it's not clear why under KVM the write would be processed
+     * directly, instead of going through eventfd.  This probably should
+     * test "tcg_enabled() || qtest_enabled()", or should just go away.
+     */
+    if (!kvm_enabled() &&
         memory_region_dispatch_write_eventfds(mr, addr, data, size, attrs)) {
         return MEMTX_OK;
     }
@@ -2550,8 +2555,6 @@ void memory_region_clear_flush_coalesced(MemoryRegion *mr)
     }
 }
 
-static bool userspace_eventfd_warning;
-
 void memory_region_add_eventfd(MemoryRegion *mr,
                                hwaddr addr,
                                unsigned size,
@@ -2568,13 +2571,6 @@ void memory_region_add_eventfd(MemoryRegion *mr,
     };
     unsigned i;
 
-    if (kvm_enabled() && (!(kvm_eventfds_enabled() ||
-                            userspace_eventfd_warning))) {
-        userspace_eventfd_warning = true;
-        error_report("Using eventfd without MMIO binding in KVM. "
-                     "Suboptimal performance expected");
-    }
-
     if (size) {
         adjust_endianness(mr, &mrfd.data, size_memop(size) | MO_TE);
     }
-- 
2.41.0

[PATCH 08/17] kvm: unify listeners for PIO address space

[Paolo Bonzini]

Since we now assume that ioeventfds are present, kvm_io_listener is always
registered.  Merge it with kvm_coalesced_pio_listener in a single
listener.  Since PIO space does not have KVM memslots attached to it,
the priority is irrelevant.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 accel/kvm/kvm-all.c | 11 ++---------
 1 file changed, 2 insertions(+), 9 deletions(-)

diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index 120051da649..b59a48da924 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -1105,13 +1105,6 @@ static void kvm_coalesce_pio_del(MemoryListener *listener,
      }
 }
 
-static MemoryListener kvm_coalesced_pio_listener = {
-    .name = "kvm-coalesced-pio",
-    .coalesced_io_add = kvm_coalesce_pio_add,
-    .coalesced_io_del = kvm_coalesce_pio_del,
-    .priority = MEMORY_LISTENER_PRIORITY_MIN,
-};
-
 int kvm_check_extension(KVMState *s, unsigned int extension)
 {
     int ret;
@@ -1768,6 +1761,8 @@ void kvm_memory_listener_register(KVMState *s, KVMMemoryListener *kml,
 
 static MemoryListener kvm_io_listener = {
     .name = "kvm-io",
+    .coalesced_io_add = kvm_coalesce_pio_add,
+    .coalesced_io_del = kvm_coalesce_pio_del,
     .eventfd_add = kvm_io_ioeventfd_add,
     .eventfd_del = kvm_io_ioeventfd_del,
     .priority = MEMORY_LISTENER_PRIORITY_DEV_BACKEND,
@@ -2598,8 +2593,6 @@ static int kvm_init(MachineState *ms)
                                  &address_space_memory, 0, "kvm-memory");
     memory_listener_register(&kvm_io_listener,
                              &address_space_io);
-    memory_listener_register(&kvm_coalesced_pio_listener,
-                             &address_space_io);
 
     s->sync_mmu = !!kvm_vm_check_extension(kvm_state, KVM_CAP_SYNC_MMU);
     if (!s->sync_mmu) {
-- 
2.41.0

[PATCH 09/17] kvm: arm: require KVM_CAP_SET_VCPU_EVENTS

[Paolo Bonzini]

KVM_CAP_SET_VCPU_EVENTS was added in Linux 4.19, released in 2018.
It was backported to RHEL8's 4.18 kernel and is in other supported distros
as well.

While it is nice that QEMU can run even on older kernels, for example
on boards that can only run old vendor kernels, I don't think this should
extend to ancient versions of KVM/ARM which, at the time, was still
moving pretty fast.  As a reference, 4.19 still had 32-bit
KVM support.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 target/arm/kvm.c | 10 +---------
 1 file changed, 1 insertion(+), 9 deletions(-)

diff --git a/target/arm/kvm.c b/target/arm/kvm.c
index b66b936a958..a50e975181f 100644
--- a/target/arm/kvm.c
+++ b/target/arm/kvm.c
@@ -34,6 +34,7 @@
 #include "qemu/log.h"
 
 const KVMCapabilityInfo kvm_arch_required_capabilities[] = {
+    KVM_CAP_INFO(VCPU_EVENTS),
     KVM_CAP_LAST_INFO
 };
 
@@ -756,10 +757,6 @@ int kvm_put_vcpu_events(ARMCPU *cpu)
     struct kvm_vcpu_events events;
     int ret;
 
-    if (!kvm_has_vcpu_events()) {
-        return 0;
-    }
-
     memset(&events, 0, sizeof(events));
     events.exception.serror_pending = env->serror.pending;
 
@@ -785,10 +782,6 @@ int kvm_get_vcpu_events(ARMCPU *cpu)
     struct kvm_vcpu_events events;
     int ret;
 
-    if (!kvm_has_vcpu_events()) {
-        return 0;
-    }
-
     memset(&events, 0, sizeof(events));
     ret = kvm_vcpu_ioctl(CPU(cpu), KVM_GET_VCPU_EVENTS, &events);
     if (ret) {
@@ -927,7 +920,6 @@ static int kvm_arm_handle_dabt_nisv(CPUState *cs, uint64_t esr_iss,
          * synchronization can be exceptionally skipped.
          */
         events.exception.ext_dabt_pending = 1;
-        /* KVM_CAP_ARM_INJECT_EXT_DABT implies KVM_CAP_VCPU_EVENTS */
         if (!kvm_vcpu_ioctl(cs, KVM_SET_VCPU_EVENTS, &events)) {
             env->ext_dabt_raised = 1;
             return 0;
-- 
2.41.0

[PATCH 10/17] kvm: i386: move KVM_CAP_IRQ_ROUTING detection to kvm_arch_required_capabilities

[Paolo Bonzini]

Simple code cleanup.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 target/i386/kvm/kvm.c | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c
index ab7d39d5399..91dd43eaa9b 100644
--- a/target/i386/kvm/kvm.c
+++ b/target/i386/kvm/kvm.c
@@ -92,6 +92,7 @@ const KVMCapabilityInfo kvm_arch_required_capabilities[] = {
     KVM_CAP_INFO(EXT_CPUID),
     KVM_CAP_INFO(MP_STATE),
     KVM_CAP_INFO(SIGNAL_MSI),
+    KVM_CAP_INFO(IRQ_ROUTING),
     KVM_CAP_LAST_INFO
 };
 
@@ -2590,11 +2591,6 @@ int kvm_arch_init(MachineState *ms, KVMState *s)
         return ret;
     }
 
-    if (!kvm_check_extension(s, KVM_CAP_IRQ_ROUTING)) {
-        error_report("kvm: KVM_CAP_IRQ_ROUTING not supported by KVM");
-        return -ENOTSUP;
-    }
-
     has_xsave = kvm_check_extension(s, KVM_CAP_XSAVE);
     has_xcrs = kvm_check_extension(s, KVM_CAP_XCRS);
     has_pit_state2 = kvm_check_extension(s, KVM_CAP_PIT_STATE2);
-- 
2.41.0

[PATCH 11/17] kvm: i386: require KVM_CAP_DEBUGREGS

[Paolo Bonzini]

This was introduced in KVM in Linux 2.6.35, we can require it unconditionally.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 accel/kvm/kvm-all.c      | 9 ---------
 include/sysemu/kvm.h     | 1 -
 include/sysemu/kvm_int.h | 1 -
 target/i386/kvm/kvm.c    | 9 +--------
 4 files changed, 1 insertion(+), 19 deletions(-)

diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index b59a48da924..aeda902b3e9 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -2528,10 +2528,6 @@ static int kvm_init(MachineState *ms)
     s->robust_singlestep =
         kvm_check_extension(s, KVM_CAP_X86_ROBUST_SINGLESTEP);
 
-#ifdef KVM_CAP_DEBUGREGS
-    s->debugregs = kvm_check_extension(s, KVM_CAP_DEBUGREGS);
-#endif
-
     s->max_nested_state_len = kvm_check_extension(s, KVM_CAP_NESTED_STATE);
 
     s->irq_set_ioctl = KVM_IRQ_LINE;
@@ -3152,11 +3148,6 @@ int kvm_has_robust_singlestep(void)
     return kvm_state->robust_singlestep;
 }
 
-int kvm_has_debugregs(void)
-{
-    return kvm_state->debugregs;
-}
-
 int kvm_max_nested_state_length(void)
 {
     return kvm_state->max_nested_state_len;
diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h
index c06982fd64c..16fddeef663 100644
--- a/include/sysemu/kvm.h
+++ b/include/sysemu/kvm.h
@@ -192,7 +192,6 @@ unsigned int kvm_get_free_memslots(void);
 bool kvm_has_sync_mmu(void);
 int kvm_has_vcpu_events(void);
 int kvm_has_robust_singlestep(void);
-int kvm_has_debugregs(void);
 int kvm_max_nested_state_length(void);
 int kvm_has_gsi_routing(void);
 
diff --git a/include/sysemu/kvm_int.h b/include/sysemu/kvm_int.h
index 840b905a2e9..151ecc84233 100644
--- a/include/sysemu/kvm_int.h
+++ b/include/sysemu/kvm_int.h
@@ -79,7 +79,6 @@ struct KVMState
     bool coalesced_flush_in_progress;
     int vcpu_events;
     int robust_singlestep;
-    int debugregs;
 #ifdef KVM_CAP_SET_GUEST_DEBUG
     QTAILQ_HEAD(, kvm_sw_breakpoint) kvm_sw_breakpoints;
 #endif
diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c
index 91dd43eaa9b..86af39c437a 100644
--- a/target/i386/kvm/kvm.c
+++ b/target/i386/kvm/kvm.c
@@ -93,6 +93,7 @@ const KVMCapabilityInfo kvm_arch_required_capabilities[] = {
     KVM_CAP_INFO(MP_STATE),
     KVM_CAP_INFO(SIGNAL_MSI),
     KVM_CAP_INFO(IRQ_ROUTING),
+    KVM_CAP_INFO(DEBUGREGS),
     KVM_CAP_LAST_INFO
 };
 
@@ -4601,10 +4602,6 @@ static int kvm_put_debugregs(X86CPU *cpu)
     struct kvm_debugregs dbgregs;
     int i;
 
-    if (!kvm_has_debugregs()) {
-        return 0;
-    }
-
     memset(&dbgregs, 0, sizeof(dbgregs));
     for (i = 0; i < 4; i++) {
         dbgregs.db[i] = env->dr[i];
@@ -4622,10 +4619,6 @@ static int kvm_get_debugregs(X86CPU *cpu)
     struct kvm_debugregs dbgregs;
     int i, ret;
 
-    if (!kvm_has_debugregs()) {
-        return 0;
-    }
-
     ret = kvm_vcpu_ioctl(CPU(cpu), KVM_GET_DEBUGREGS, &dbgregs);
     if (ret < 0) {
         return ret;
-- 
2.41.0

[PATCH 12/17] kvm: i386: require KVM_CAP_XSAVE

[Paolo Bonzini]

This was introduced in KVM in Linux 2.6.36, and could already be used at
the time to save/restore FPU data even on older processor.  We can require
it unconditionally and stop using KVM_GET/SET_FPU.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 target/i386/kvm/kvm.c | 70 ++-----------------------------------------
 1 file changed, 2 insertions(+), 68 deletions(-)

diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c
index 86af39c437a..f6622684024 100644
--- a/target/i386/kvm/kvm.c
+++ b/target/i386/kvm/kvm.c
@@ -94,6 +94,7 @@ const KVMCapabilityInfo kvm_arch_required_capabilities[] = {
     KVM_CAP_INFO(SIGNAL_MSI),
     KVM_CAP_INFO(IRQ_ROUTING),
     KVM_CAP_INFO(DEBUGREGS),
+    KVM_CAP_INFO(XSAVE),
     KVM_CAP_LAST_INFO
 };
 
@@ -137,7 +138,6 @@ static uint32_t has_architectural_pmu_version;
 static uint32_t num_architectural_pmu_gp_counters;
 static uint32_t num_architectural_pmu_fixed_counters;
 
-static int has_xsave;
 static int has_xsave2;
 static int has_xcrs;
 static int has_pit_state2;
@@ -1714,10 +1714,8 @@ static void kvm_init_xsave(CPUX86State *env)
 {
     if (has_xsave2) {
         env->xsave_buf_len = QEMU_ALIGN_UP(has_xsave2, 4096);
-    } else if (has_xsave) {
-        env->xsave_buf_len = sizeof(struct kvm_xsave);
     } else {
-        return;
+        env->xsave_buf_len = sizeof(struct kvm_xsave);
     }
 
     env->xsave_buf = qemu_memalign(4096, env->xsave_buf_len);
@@ -2592,7 +2590,6 @@ int kvm_arch_init(MachineState *ms, KVMState *s)
         return ret;
     }
 
-    has_xsave = kvm_check_extension(s, KVM_CAP_XSAVE);
     has_xcrs = kvm_check_extension(s, KVM_CAP_XCRS);
     has_pit_state2 = kvm_check_extension(s, KVM_CAP_PIT_STATE2);
     has_sregs2 = kvm_check_extension(s, KVM_CAP_SREGS2) > 0;
@@ -2877,40 +2874,11 @@ static int kvm_getput_regs(X86CPU *cpu, int set)
     return ret;
 }
 
-static int kvm_put_fpu(X86CPU *cpu)
-{
-    CPUX86State *env = &cpu->env;
-    struct kvm_fpu fpu;
-    int i;
-
-    memset(&fpu, 0, sizeof fpu);
-    fpu.fsw = env->fpus & ~(7 << 11);
-    fpu.fsw |= (env->fpstt & 7) << 11;
-    fpu.fcw = env->fpuc;
-    fpu.last_opcode = env->fpop;
-    fpu.last_ip = env->fpip;
-    fpu.last_dp = env->fpdp;
-    for (i = 0; i < 8; ++i) {
-        fpu.ftwx |= (!env->fptags[i]) << i;
-    }
-    memcpy(fpu.fpr, env->fpregs, sizeof env->fpregs);
-    for (i = 0; i < CPU_NB_REGS; i++) {
-        stq_p(&fpu.xmm[i][0], env->xmm_regs[i].ZMM_Q(0));
-        stq_p(&fpu.xmm[i][8], env->xmm_regs[i].ZMM_Q(1));
-    }
-    fpu.mxcsr = env->mxcsr;
-
-    return kvm_vcpu_ioctl(CPU(cpu), KVM_SET_FPU, &fpu);
-}
-
 static int kvm_put_xsave(X86CPU *cpu)
 {
     CPUX86State *env = &cpu->env;
     void *xsave = env->xsave_buf;
 
-    if (!has_xsave) {
-        return kvm_put_fpu(cpu);
-    }
     x86_cpu_xsave_all_areas(cpu, xsave, env->xsave_buf_len);
 
     return kvm_vcpu_ioctl(CPU(cpu), KVM_SET_XSAVE, xsave);
@@ -3655,46 +3623,12 @@ static int kvm_put_msrs(X86CPU *cpu, int level)
 }
 
 
-static int kvm_get_fpu(X86CPU *cpu)
-{
-    CPUX86State *env = &cpu->env;
-    struct kvm_fpu fpu;
-    int i, ret;
-
-    ret = kvm_vcpu_ioctl(CPU(cpu), KVM_GET_FPU, &fpu);
-    if (ret < 0) {
-        return ret;
-    }
-
-    env->fpstt = (fpu.fsw >> 11) & 7;
-    env->fpus = fpu.fsw;
-    env->fpuc = fpu.fcw;
-    env->fpop = fpu.last_opcode;
-    env->fpip = fpu.last_ip;
-    env->fpdp = fpu.last_dp;
-    for (i = 0; i < 8; ++i) {
-        env->fptags[i] = !((fpu.ftwx >> i) & 1);
-    }
-    memcpy(env->fpregs, fpu.fpr, sizeof env->fpregs);
-    for (i = 0; i < CPU_NB_REGS; i++) {
-        env->xmm_regs[i].ZMM_Q(0) = ldq_p(&fpu.xmm[i][0]);
-        env->xmm_regs[i].ZMM_Q(1) = ldq_p(&fpu.xmm[i][8]);
-    }
-    env->mxcsr = fpu.mxcsr;
-
-    return 0;
-}
-
 static int kvm_get_xsave(X86CPU *cpu)
 {
     CPUX86State *env = &cpu->env;
     void *xsave = env->xsave_buf;
     int type, ret;
 
-    if (!has_xsave) {
-        return kvm_get_fpu(cpu);
-    }
-
     type = has_xsave2 ? KVM_GET_XSAVE2 : KVM_GET_XSAVE;
     ret = kvm_vcpu_ioctl(CPU(cpu), type, xsave);
     if (ret < 0) {
-- 
2.41.0

[PATCH 13/17] kvm: i386: require KVM_CAP_SET_VCPU_EVENTS and KVM_CAP_X86_ROBUST_SINGLESTEP

[Paolo Bonzini]

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 accel/kvm/kvm-all.c      | 12 ------
 include/sysemu/kvm.h     |  2 -
 include/sysemu/kvm_int.h |  2 -
 target/i386/kvm/kvm.c    | 92 +---------------------------------------
 4 files changed, 2 insertions(+), 106 deletions(-)

diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index aeda902b3e9..608da41465d 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -2521,13 +2521,6 @@ static int kvm_init(MachineState *ms)
         }
     }
 
-#ifdef KVM_CAP_VCPU_EVENTS
-    s->vcpu_events = kvm_check_extension(s, KVM_CAP_VCPU_EVENTS);
-#endif
-
-    s->robust_singlestep =
-        kvm_check_extension(s, KVM_CAP_X86_ROBUST_SINGLESTEP);
-
     s->max_nested_state_len = kvm_check_extension(s, KVM_CAP_NESTED_STATE);
 
     s->irq_set_ioctl = KVM_IRQ_LINE;
@@ -3138,11 +3131,6 @@ bool kvm_has_sync_mmu(void)
     return kvm_state->sync_mmu;
 }
 
-int kvm_has_vcpu_events(void)
-{
-    return kvm_state->vcpu_events;
-}
-
 int kvm_has_robust_singlestep(void)
 {
     return kvm_state->robust_singlestep;
diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h
index 16fddeef663..a2526009c6c 100644
--- a/include/sysemu/kvm.h
+++ b/include/sysemu/kvm.h
@@ -190,8 +190,6 @@ typedef struct KVMRouteChange {
 unsigned int kvm_get_max_memslots(void);
 unsigned int kvm_get_free_memslots(void);
 bool kvm_has_sync_mmu(void);
-int kvm_has_vcpu_events(void);
-int kvm_has_robust_singlestep(void);
 int kvm_max_nested_state_length(void);
 int kvm_has_gsi_routing(void);
 
diff --git a/include/sysemu/kvm_int.h b/include/sysemu/kvm_int.h
index 151ecc84233..2dd684a363c 100644
--- a/include/sysemu/kvm_int.h
+++ b/include/sysemu/kvm_int.h
@@ -77,8 +77,6 @@ struct KVMState
     int coalesced_pio;
     struct kvm_coalesced_mmio_ring *coalesced_mmio_ring;
     bool coalesced_flush_in_progress;
-    int vcpu_events;
-    int robust_singlestep;
 #ifdef KVM_CAP_SET_GUEST_DEBUG
     QTAILQ_HEAD(, kvm_sw_breakpoint) kvm_sw_breakpoints;
 #endif
diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c
index f6622684024..5db3fd0f4c4 100644
--- a/target/i386/kvm/kvm.c
+++ b/target/i386/kvm/kvm.c
@@ -95,6 +95,8 @@ const KVMCapabilityInfo kvm_arch_required_capabilities[] = {
     KVM_CAP_INFO(IRQ_ROUTING),
     KVM_CAP_INFO(DEBUGREGS),
     KVM_CAP_INFO(XSAVE),
+    KVM_CAP_INFO(VCPU_EVENTS),
+    KVM_CAP_INFO(X86_ROBUST_SINGLESTEP),
     KVM_CAP_LAST_INFO
 };
 
@@ -690,15 +692,6 @@ void kvm_arch_on_sigbus_vcpu(CPUState *c, int code, void *addr)
     emit_hypervisor_memory_failure(MEMORY_FAILURE_ACTION_IGNORE, false);
 }
 
-static void kvm_reset_exception(CPUX86State *env)
-{
-    env->exception_nr = -1;
-    env->exception_pending = 0;
-    env->exception_injected = 0;
-    env->exception_has_payload = false;
-    env->exception_payload = 0;
-}
-
 static void kvm_queue_exception(CPUX86State *env,
                                 int32_t exception_nr,
                                 uint8_t exception_has_payload,
@@ -731,38 +724,6 @@ static void kvm_queue_exception(CPUX86State *env,
     }
 }
 
-static int kvm_inject_mce_oldstyle(X86CPU *cpu)
-{
-    CPUX86State *env = &cpu->env;
-
-    if (!kvm_has_vcpu_events() && env->exception_nr == EXCP12_MCHK) {
-        unsigned int bank, bank_num = env->mcg_cap & 0xff;
-        struct kvm_x86_mce mce;
-
-        kvm_reset_exception(env);
-
-        /*
-         * There must be at least one bank in use if an MCE is pending.
-         * Find it and use its values for the event injection.
-         */
-        for (bank = 0; bank < bank_num; bank++) {
-            if (env->mce_banks[bank * 4 + 1] & MCI_STATUS_VAL) {
-                break;
-            }
-        }
-        assert(bank < bank_num);
-
-        mce.bank = bank;
-        mce.status = env->mce_banks[bank * 4 + 1];
-        mce.mcg_status = env->mcg_status;
-        mce.addr = env->mce_banks[bank * 4 + 2];
-        mce.misc = env->mce_banks[bank * 4 + 3];
-
-        return kvm_vcpu_ioctl(CPU(cpu), KVM_X86_SET_MCE, &mce);
-    }
-    return 0;
-}
-
 static void cpu_update_state(void *opaque, bool running, RunState state)
 {
     CPUX86State *env = opaque;
@@ -4359,10 +4320,6 @@ static int kvm_put_vcpu_events(X86CPU *cpu, int level)
     CPUX86State *env = &cpu->env;
     struct kvm_vcpu_events events = {};
 
-    if (!kvm_has_vcpu_events()) {
-        return 0;
-    }
-
     events.flags = 0;
 
     if (has_exception_payload) {
@@ -4430,10 +4387,6 @@ static int kvm_get_vcpu_events(X86CPU *cpu)
     struct kvm_vcpu_events events;
     int ret;
 
-    if (!kvm_has_vcpu_events()) {
-        return 0;
-    }
-
     memset(&events, 0, sizeof(events));
     ret = kvm_vcpu_ioctl(CPU(cpu), KVM_GET_VCPU_EVENTS, &events);
     if (ret < 0) {
@@ -4499,37 +4452,6 @@ static int kvm_get_vcpu_events(X86CPU *cpu)
     return 0;
 }
 
-static int kvm_guest_debug_workarounds(X86CPU *cpu)
-{
-    CPUState *cs = CPU(cpu);
-    CPUX86State *env = &cpu->env;
-    int ret = 0;
-    unsigned long reinject_trap = 0;
-
-    if (!kvm_has_vcpu_events()) {
-        if (env->exception_nr == EXCP01_DB) {
-            reinject_trap = KVM_GUESTDBG_INJECT_DB;
-        } else if (env->exception_injected == EXCP03_INT3) {
-            reinject_trap = KVM_GUESTDBG_INJECT_BP;
-        }
-        kvm_reset_exception(env);
-    }
-
-    /*
-     * Kernels before KVM_CAP_X86_ROBUST_SINGLESTEP overwrote flags.TF
-     * injected via SET_GUEST_DEBUG while updating GP regs. Work around this
-     * by updating the debug state once again if single-stepping is on.
-     * Another reason to call kvm_update_guest_debug here is a pending debug
-     * trap raise by the guest. On kernels without SET_VCPU_EVENTS we have to
-     * reinject them via SET_GUEST_DEBUG.
-     */
-    if (reinject_trap ||
-        (!kvm_has_robust_singlestep() && cs->singlestep_enabled)) {
-        ret = kvm_update_guest_debug(cs, reinject_trap);
-    }
-    return ret;
-}
-
 static int kvm_put_debugregs(X86CPU *cpu)
 {
     CPUX86State *env = &cpu->env;
@@ -4702,11 +4624,6 @@ int kvm_arch_put_registers(CPUState *cpu, int level)
     if (ret < 0) {
         return ret;
     }
-    /* must be before kvm_put_msrs */
-    ret = kvm_inject_mce_oldstyle(x86_cpu);
-    if (ret < 0) {
-        return ret;
-    }
     ret = kvm_put_msrs(x86_cpu, level);
     if (ret < 0) {
         return ret;
@@ -4730,11 +4647,6 @@ int kvm_arch_put_registers(CPUState *cpu, int level)
     if (ret < 0) {
         return ret;
     }
-    /* must be last */
-    ret = kvm_guest_debug_workarounds(x86_cpu);
-    if (ret < 0) {
-        return ret;
-    }
     return 0;
 }
 
-- 
2.41.0

[PATCH 14/17] kvm: i386: require KVM_CAP_MCE

[Paolo Bonzini]

This was introduced in KVM in Linux 2.6.34, we can require it unconditionally.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 accel/kvm/kvm-all.c   |  5 -----
 target/i386/kvm/kvm.c | 14 ++++----------
 2 files changed, 4 insertions(+), 15 deletions(-)

diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index 608da41465d..41d388acff2 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -3131,11 +3131,6 @@ bool kvm_has_sync_mmu(void)
     return kvm_state->sync_mmu;
 }
 
-int kvm_has_robust_singlestep(void)
-{
-    return kvm_state->robust_singlestep;
-}
-
 int kvm_max_nested_state_length(void)
 {
     return kvm_state->max_nested_state_len;
diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c
index 5db3fd0f4c4..6df8d56c87e 100644
--- a/target/i386/kvm/kvm.c
+++ b/target/i386/kvm/kvm.c
@@ -97,6 +97,7 @@ const KVMCapabilityInfo kvm_arch_required_capabilities[] = {
     KVM_CAP_INFO(XSAVE),
     KVM_CAP_INFO(VCPU_EVENTS),
     KVM_CAP_INFO(X86_ROBUST_SINGLESTEP),
+    KVM_CAP_INFO(MCE),
     KVM_CAP_LAST_INFO
 };
 
@@ -582,14 +583,8 @@ uint64_t kvm_arch_get_supported_msr_feature(KVMState *s, uint32_t index)
 static int kvm_get_mce_cap_supported(KVMState *s, uint64_t *mce_cap,
                                      int *max_banks)
 {
-    int r;
-
-    r = kvm_check_extension(s, KVM_CAP_MCE);
-    if (r > 0) {
-        *max_banks = r;
-        return kvm_ioctl(s, KVM_X86_GET_MCE_CAP_SUPPORTED, mce_cap);
-    }
-    return -ENOSYS;
+    *max_banks = kvm_check_extension(s, KVM_CAP_MCE);
+    return kvm_ioctl(s, KVM_X86_GET_MCE_CAP_SUPPORTED, mce_cap);
 }
 
 static void kvm_mce_inject(X86CPU *cpu, hwaddr paddr, int code)
@@ -2116,8 +2111,7 @@ int kvm_arch_init_vcpu(CPUState *cs)
 
     if (((env->cpuid_version >> 8)&0xF) >= 6
         && (env->features[FEAT_1_EDX] & (CPUID_MCE | CPUID_MCA)) ==
-           (CPUID_MCE | CPUID_MCA)
-        && kvm_check_extension(cs->kvm_state, KVM_CAP_MCE) > 0) {
+           (CPUID_MCE | CPUID_MCA)) {
         uint64_t mcg_cap, unsupported_caps;
         int banks;
         int ret;
-- 
2.41.0

[PATCH 15/17] kvm: i386: require KVM_CAP_ADJUST_CLOCK

[Paolo Bonzini]

This was introduced in KVM in Linux 2.6.33, we can require it
unconditionally.  KVM_CLOCK_TSC_STABLE was only added in Linux 4.9,
for now do not require it (though it would allow the removal of some
pretty yucky code).

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 hw/i386/kvm/clock.c        | 4 ----
 target/i386/kvm/kvm.c      | 6 +-----
 target/i386/kvm/kvm_i386.h | 1 -
 3 files changed, 1 insertion(+), 10 deletions(-)

diff --git a/hw/i386/kvm/clock.c b/hw/i386/kvm/clock.c
index f25977d3f68..e756b0aa43f 100644
--- a/hw/i386/kvm/clock.c
+++ b/hw/i386/kvm/clock.c
@@ -333,10 +333,6 @@ void kvmclock_create(bool create_always)
     X86CPU *cpu = X86_CPU(first_cpu);
 
     assert(kvm_enabled());
-    if (!kvm_has_adjust_clock()) {
-        return;
-    }
-
     if (create_always ||
         cpu->env.features[FEAT_KVM] & ((1ULL << KVM_FEATURE_CLOCKSOURCE) |
                                        (1ULL << KVM_FEATURE_CLOCKSOURCE2))) {
diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c
index 6df8d56c87e..df131fcc8b1 100644
--- a/target/i386/kvm/kvm.c
+++ b/target/i386/kvm/kvm.c
@@ -98,6 +98,7 @@ const KVMCapabilityInfo kvm_arch_required_capabilities[] = {
     KVM_CAP_INFO(VCPU_EVENTS),
     KVM_CAP_INFO(X86_ROBUST_SINGLESTEP),
     KVM_CAP_INFO(MCE),
+    KVM_CAP_INFO(ADJUST_CLOCK),
     KVM_CAP_LAST_INFO
 };
 
@@ -177,11 +178,6 @@ bool kvm_has_adjust_clock_stable(void)
     return (ret & KVM_CLOCK_TSC_STABLE);
 }
 
-bool kvm_has_adjust_clock(void)
-{
-    return kvm_check_extension(kvm_state, KVM_CAP_ADJUST_CLOCK);
-}
-
 bool kvm_has_exception_payload(void)
 {
     return has_exception_payload;
diff --git a/target/i386/kvm/kvm_i386.h b/target/i386/kvm/kvm_i386.h
index 55d4e68c342..7e60ea4f23f 100644
--- a/target/i386/kvm/kvm_i386.h
+++ b/target/i386/kvm/kvm_i386.h
@@ -50,7 +50,6 @@ void kvm_request_xsave_components(X86CPU *cpu, uint64_t mask);
 
 #ifdef CONFIG_KVM
 
-bool kvm_has_adjust_clock(void);
 bool kvm_has_adjust_clock_stable(void);
 bool kvm_has_exception_payload(void);
 void kvm_synchronize_all_tsc(void);
-- 
2.41.0

[PATCH 16/17] kvm: i386: require KVM_CAP_SET_IDENTITY_MAP_ADDR

[Paolo Bonzini]

This was introduced in KVM in Linux 2.6.32, we can require it unconditionally.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 target/i386/kvm/kvm.c | 20 +++++++-------------
 1 file changed, 7 insertions(+), 13 deletions(-)

diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c
index df131fcc8b1..a3d52cb538f 100644
--- a/target/i386/kvm/kvm.c
+++ b/target/i386/kvm/kvm.c
@@ -99,6 +99,7 @@ const KVMCapabilityInfo kvm_arch_required_capabilities[] = {
     KVM_CAP_INFO(X86_ROBUST_SINGLESTEP),
     KVM_CAP_INFO(MCE),
     KVM_CAP_INFO(ADJUST_CLOCK),
+    KVM_CAP_INFO(SET_IDENTITY_MAP_ADDR),
     KVM_CAP_LAST_INFO
 };
 
@@ -2600,20 +2601,13 @@ int kvm_arch_init(MachineState *ms, KVMState *s)
      * In order to use vm86 mode, an EPT identity map and a TSS  are needed.
      * Since these must be part of guest physical memory, we need to allocate
      * them, both by setting their start addresses in the kernel and by
-     * creating a corresponding e820 entry. We need 4 pages before the BIOS.
-     *
-     * Older KVM versions may not support setting the identity map base. In
-     * that case we need to stick with the default, i.e. a 256K maximum BIOS
-     * size.
+     * creating a corresponding e820 entry. We need 4 pages before the BIOS,
+     * so this value allows up to 16M BIOSes.
      */
-    if (kvm_check_extension(s, KVM_CAP_SET_IDENTITY_MAP_ADDR)) {
-        /* Allows up to 16M BIOSes. */
-        identity_base = 0xfeffc000;
-
-        ret = kvm_vm_ioctl(s, KVM_SET_IDENTITY_MAP_ADDR, &identity_base);
-        if (ret < 0) {
-            return ret;
-        }
+    identity_base = 0xfeffc000;
+    ret = kvm_vm_ioctl(s, KVM_SET_IDENTITY_MAP_ADDR, &identity_base);
+    if (ret < 0) {
+        return ret;
     }
 
     /* Set TSS base one page after EPT identity map. */
-- 
2.41.0

[PATCH 17/17] kvm: i8254: require KVM_CAP_PIT2 and KVM_CAP_PIT_STATE2

[Paolo Bonzini]

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
 hw/i386/kvm/i8254.c        | 38 ++++++++++++--------------------------
 hw/i386/pc.c               |  6 +-----
 target/i386/kvm/kvm.c      |  7 -------
 target/i386/kvm/kvm_i386.h |  1 -
 4 files changed, 13 insertions(+), 39 deletions(-)

diff --git a/hw/i386/kvm/i8254.c b/hw/i386/kvm/i8254.c
index a649b2b7caf..e49b9c4b565 100644
--- a/hw/i386/kvm/i8254.c
+++ b/hw/i386/kvm/i8254.c
@@ -97,24 +97,12 @@ static void kvm_pit_get(PITCommonState *pit)
         return;
     }
 
-    if (kvm_has_pit_state2()) {
-        ret = kvm_vm_ioctl(kvm_state, KVM_GET_PIT2, &kpit);
-        if (ret < 0) {
-            fprintf(stderr, "KVM_GET_PIT2 failed: %s\n", strerror(-ret));
-            abort();
-        }
-        pit->channels[0].irq_disabled = kpit.flags & KVM_PIT_FLAGS_HPET_LEGACY;
-    } else {
-        /*
-         * kvm_pit_state2 is superset of kvm_pit_state struct,
-         * so we can use it for KVM_GET_PIT as well.
-         */
-        ret = kvm_vm_ioctl(kvm_state, KVM_GET_PIT, &kpit);
-        if (ret < 0) {
-            fprintf(stderr, "KVM_GET_PIT failed: %s\n", strerror(-ret));
-            abort();
-        }
+    ret = kvm_vm_ioctl(kvm_state, KVM_GET_PIT2, &kpit);
+    if (ret < 0) {
+        fprintf(stderr, "KVM_GET_PIT2 failed: %s\n", strerror(-ret));
+        abort();
     }
+    pit->channels[0].irq_disabled = kpit.flags & KVM_PIT_FLAGS_HPET_LEGACY;
     for (i = 0; i < 3; i++) {
         kchan = &kpit.channels[i];
         sc = &pit->channels[i];
@@ -170,12 +158,9 @@ static void kvm_pit_put(PITCommonState *pit)
         kchan->count_load_time = sc->count_load_time - s->kernel_clock_offset;
     }
 
-    ret = kvm_vm_ioctl(kvm_state,
-                       kvm_has_pit_state2() ? KVM_SET_PIT2 : KVM_SET_PIT,
-                       &kpit);
+    ret = kvm_vm_ioctl(kvm_state, KVM_SET_PIT2, &kpit);
     if (ret < 0) {
-        fprintf(stderr, "%s failed: %s\n",
-                kvm_has_pit_state2() ? "KVM_SET_PIT2" : "KVM_SET_PIT",
+        fprintf(stderr, "KVM_SET_PIT2 failed: %s\n",
                 strerror(-ret));
         abort();
     }
@@ -261,11 +246,12 @@ static void kvm_pit_realizefn(DeviceState *dev, Error **errp)
     };
     int ret;
 
-    if (kvm_check_extension(kvm_state, KVM_CAP_PIT2)) {
-        ret = kvm_vm_ioctl(kvm_state, KVM_CREATE_PIT2, &config);
-    } else {
-        ret = kvm_vm_ioctl(kvm_state, KVM_CREATE_PIT);
+    if (!kvm_check_extension(kvm_state, KVM_CAP_PIT_STATE2) ||
+        !kvm_check_extension(kvm_state, KVM_CAP_PIT2)) {
+        error_setg(errp, "In-kernel PIT not available");
     }
+
+    ret = kvm_vm_ioctl(kvm_state, KVM_CREATE_PIT2, &config);
     if (ret < 0) {
         error_setg(errp, "Create kernel PIC irqchip failed: %s",
                    strerror(-ret));
diff --git a/hw/i386/pc.c b/hw/i386/pc.c
index bb3854d1d08..f3793dd0633 100644
--- a/hw/i386/pc.c
+++ b/hw/i386/pc.c
@@ -1213,12 +1213,8 @@ void pc_basic_device_init(struct PCMachineState *pcms,
 
     /*
      * Check if an HPET shall be created.
-     *
-     * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
-     * when the HPET wants to take over. Thus we have to disable the latter.
      */
-    if (pcms->hpet_enabled && (!kvm_irqchip_in_kernel() ||
-                               kvm_has_pit_state2())) {
+    if (pcms->hpet_enabled) {
         hpet = qdev_try_new(TYPE_HPET);
         if (!hpet) {
             error_report("couldn't create HPET device");
diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c
index a3d52cb538f..cbb8db6ea70 100644
--- a/target/i386/kvm/kvm.c
+++ b/target/i386/kvm/kvm.c
@@ -145,7 +145,6 @@ static uint32_t num_architectural_pmu_fixed_counters;
 
 static int has_xsave2;
 static int has_xcrs;
-static int has_pit_state2;
 static int has_sregs2;
 static int has_exception_payload;
 static int has_triple_fault_event;
@@ -162,11 +161,6 @@ static KVMMSRHandlers msr_handlers[KVM_MSR_FILTER_MAX_RANGES];
 static RateLimit bus_lock_ratelimit_ctrl;
 static int kvm_get_one_msr(X86CPU *cpu, int index, uint64_t *value);
 
-bool kvm_has_pit_state2(void)
-{
-    return !!has_pit_state2;
-}
-
 bool kvm_has_smm(void)
 {
     return kvm_vm_check_extension(kvm_state, KVM_CAP_X86_SMM);
@@ -2543,7 +2537,6 @@ int kvm_arch_init(MachineState *ms, KVMState *s)
     }
 
     has_xcrs = kvm_check_extension(s, KVM_CAP_XCRS);
-    has_pit_state2 = kvm_check_extension(s, KVM_CAP_PIT_STATE2);
     has_sregs2 = kvm_check_extension(s, KVM_CAP_SREGS2) > 0;
 
     hv_vpindex_settable = kvm_check_extension(s, KVM_CAP_HYPERV_VP_INDEX);
diff --git a/target/i386/kvm/kvm_i386.h b/target/i386/kvm/kvm_i386.h
index 7e60ea4f23f..30fedcffea3 100644
--- a/target/i386/kvm/kvm_i386.h
+++ b/target/i386/kvm/kvm_i386.h
@@ -33,7 +33,6 @@
 bool kvm_has_smm(void);
 bool kvm_enable_x2apic(void);
 bool kvm_hv_vpindex_settable(void);
-bool kvm_has_pit_state2(void);
 
 bool kvm_enable_sgx_provisioning(KVMState *s);
 bool kvm_hyperv_expand_features(X86CPU *cpu, Error **errp);
-- 
2.41.0

Re: [PATCH 01/17] kvm: remove unnecessary stub

[Manos Pitsidianakis]

On Wed, 18 Oct 2023 19:37, Paolo Bonzini <pbonzini@redhat.com> wrote:
>This function is only invoked from hw/intc/s390_flic_kvm.c, and therefore
>only if CONFIG_KVM is defined.
>
>Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>

Re: [PATCH 02/17] kvm: require KVM_CAP_INTERNAL_ERROR_DATA

[Manos Pitsidianakis]

On Wed, 18 Oct 2023 19:37, Paolo Bonzini <pbonzini@redhat.com> wrote:
>This was introduced in KVM in Linux 2.6.33, we can require it unconditionally.
>
>Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>

Re: [PATCH 04/17] kvm: require KVM_CAP_IRQFD for kernel irqchip

[Manos Pitsidianakis]

On Wed, 18 Oct 2023 19:37, Paolo Bonzini <pbonzini@redhat.com> wrote:
>KVM_CAP_IRQFD is always available on architectures that support an in-kernel
>interrupt controller, and was introduced in Linux 2.6.32.  We can require
>it unconditionally.
>
>Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
>---
> accel/kvm/kvm-all.c         | 13 +++++--------
> accel/stubs/kvm-stub.c      |  1 -
> hw/intc/arm_gicv3_its_kvm.c |  2 +-
> include/sysemu/kvm.h        |  6 +++---
> target/riscv/kvm/kvm-cpu.c  |  2 +-
> 5 files changed, 10 insertions(+), 14 deletions(-)

Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>

Re: [PATCH 08/17] kvm: unify listeners for PIO address space

[Manos Pitsidianakis]

On Wed, 18 Oct 2023 19:37, Paolo Bonzini <pbonzini@redhat.com> wrote:
>Since we now assume that ioeventfds are present, kvm_io_listener is always
>registered.  Merge it with kvm_coalesced_pio_listener in a single
>listener.  Since PIO space does not have KVM memslots attached to it,
>the priority is irrelevant.
>
>Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
>---
> accel/kvm/kvm-all.c | 11 ++---------
> 1 file changed, 2 insertions(+), 9 deletions(-)

Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>