LinuxLists.cc - [PATCH v2 0/6] kvm: level triggered irqfd support

2012-06-27 05:08:56

Subject: [PATCH v2 0/6] kvm: level triggered irqfd support

Ok, let's see how this flies. I actually quite like this, so be
gentle tearing it apart ;)

I just couldn't bring myself to contort KVM_IRQFD into something
that either sets up an irqfd or specifies a nearly unrelated EOI
eventfd. The solution I've come up with, that also avoids exposing
irq_source_ids to userspace, is to work through the irqfd. If we
setup a level irqfd, we can optionally associate an eoifd with
the same irq_source_id, by passing the irqfd. To do this, we just
need to create a new reference counted object for the source ID
so we don't run into problems ordering release. This means we
end up with a KVM_EOIFD ioctl that has both general usefulness and
can still tie into an irqfd. In patch 6/6 I also include an
alternate de-assert mechanism via an irqfd with the opposite
polarity. I don't currently use this, but it's pretty trivial
and at least available in archives now.

I don't address whether injecting an edge irqfd really needs an assert
followed by de-assert (I don't know). This new interface really
unties itself from caring. We might be able to consolidate inject
functions at some future point, but it doesn't change how we'd name
flags as it did in the previous version. Thanks,

Alex

---

Alex Williamson (6):
kvm: Level IRQ de-assert for KVM_IRQFD
kvm: KVM_EOIFD, an eventfd for EOIs
kvm: Extend irqfd to support level interrupts
kvm: Sanitize KVM_IRQFD flags
kvm: Add missing KVM_IRQFD API documentation
kvm: Pass kvm_irqfd to functions

Documentation/virtual/kvm/api.txt | 53 ++++++
arch/x86/kvm/x86.c | 2
include/linux/kvm.h | 23 +++
include/linux/kvm_host.h | 17 ++
virt/kvm/eventfd.c | 323 ++++++++++++++++++++++++++++++++++++-
virt/kvm/kvm_main.c | 13 +
6 files changed, 414 insertions(+), 17 deletions(-)

2012-06-27 05:09:12

by Alex Williamson

[permalink] [raw]

Subject: [PATCH v2 1/6] kvm: Pass kvm_irqfd to functions

Prune this down to just the struct kvm_irqfd so we can avoid
changing function definition for every flag or field we use.

Signed-off-by: Alex Williamson <[email protected]>
---

include/linux/kvm_host.h | 4 ++--
virt/kvm/eventfd.c | 20 ++++++++++----------
virt/kvm/kvm_main.c | 2 +-
3 files changed, 13 insertions(+), 13 deletions(-)

diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 27ac8a4..ae3b426 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -824,7 +824,7 @@ static inline void kvm_free_irq_routing(struct kvm *kvm) {}
#ifdef CONFIG_HAVE_KVM_EVENTFD

void kvm_eventfd_init(struct kvm *kvm);
-int kvm_irqfd(struct kvm *kvm, int fd, int gsi, int flags);
+int kvm_irqfd(struct kvm *kvm, struct kvm_irqfd *args);
void kvm_irqfd_release(struct kvm *kvm);
void kvm_irq_routing_update(struct kvm *, struct kvm_irq_routing_table *);
int kvm_ioeventfd(struct kvm *kvm, struct kvm_ioeventfd *args);
@@ -833,7 +833,7 @@ int kvm_ioeventfd(struct kvm *kvm, struct kvm_ioeventfd *args);

static inline void kvm_eventfd_init(struct kvm *kvm) {}

-static inline int kvm_irqfd(struct kvm *kvm, int fd, int gsi, int flags)
+static inline int kvm_irqfd(struct kvm *kvm, struct kvm_irqfd *args)
{
return -EINVAL;
}
diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c
index f59c1e8..c307c24 100644
--- a/virt/kvm/eventfd.c
+++ b/virt/kvm/eventfd.c
@@ -198,7 +198,7 @@ static void irqfd_update(struct kvm *kvm, struct _irqfd *irqfd,
}

static int
-kvm_irqfd_assign(struct kvm *kvm, int fd, int gsi)
+kvm_irqfd_assign(struct kvm *kvm, struct kvm_irqfd *args)
{
struct kvm_irq_routing_table *irq_rt;
struct _irqfd *irqfd, *tmp;
@@ -212,12 +212,12 @@ kvm_irqfd_assign(struct kvm *kvm, int fd, int gsi)
return -ENOMEM;

irqfd->kvm = kvm;
- irqfd->gsi = gsi;
+ irqfd->gsi = args->gsi;
INIT_LIST_HEAD(&irqfd->list);
INIT_WORK(&irqfd->inject, irqfd_inject);
INIT_WORK(&irqfd->shutdown, irqfd_shutdown);

- file = eventfd_fget(fd);
+ file = eventfd_fget(args->fd);
if (IS_ERR(file)) {
ret = PTR_ERR(file);
goto fail;
@@ -298,19 +298,19 @@ kvm_eventfd_init(struct kvm *kvm)
* shutdown any irqfd's that match fd+gsi
*/
static int
-kvm_irqfd_deassign(struct kvm *kvm, int fd, int gsi)
+kvm_irqfd_deassign(struct kvm *kvm, struct kvm_irqfd *args)
{
struct _irqfd *irqfd, *tmp;
struct eventfd_ctx *eventfd;

- eventfd = eventfd_ctx_fdget(fd);
+ eventfd = eventfd_ctx_fdget(args->fd);
if (IS_ERR(eventfd))
return PTR_ERR(eventfd);

spin_lock_irq(&kvm->irqfds.lock);

list_for_each_entry_safe(irqfd, tmp, &kvm->irqfds.items, list) {
- if (irqfd->eventfd == eventfd && irqfd->gsi == gsi) {
+ if (irqfd->eventfd == eventfd && irqfd->gsi == args->gsi) {
/*
* This rcu_assign_pointer is needed for when
* another thread calls kvm_irq_routing_update before
@@ -338,12 +338,12 @@ kvm_irqfd_deassign(struct kvm *kvm, int fd, int gsi)
}

int
-kvm_irqfd(struct kvm *kvm, int fd, int gsi, int flags)
+kvm_irqfd(struct kvm *kvm, struct kvm_irqfd *args)
{
- if (flags & KVM_IRQFD_FLAG_DEASSIGN)
- return kvm_irqfd_deassign(kvm, fd, gsi);
+ if (args->flags & KVM_IRQFD_FLAG_DEASSIGN)
+ return kvm_irqfd_deassign(kvm, args);

- return kvm_irqfd_assign(kvm, fd, gsi);
+ return kvm_irqfd_assign(kvm, args);
}

/*
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 02cb440..b4ad14cc 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -2059,7 +2059,7 @@ static long kvm_vm_ioctl(struct file *filp,
r = -EFAULT;
if (copy_from_user(&data, argp, sizeof data))
goto out;
- r = kvm_irqfd(kvm, data.fd, data.gsi, data.flags);
+ r = kvm_irqfd(kvm, &data);
break;
}
case KVM_IOEVENTFD: {

2012-06-27 05:09:23

by Alex Williamson

[permalink] [raw]

Subject: [PATCH v2 2/6] kvm: Add missing KVM_IRQFD API documentation

Signed-off-by: Alex Williamson <[email protected]>
---

Documentation/virtual/kvm/api.txt | 16 ++++++++++++++++
1 file changed, 16 insertions(+)

diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
index 310fe50..ea9edce 100644
--- a/Documentation/virtual/kvm/api.txt
+++ b/Documentation/virtual/kvm/api.txt
@@ -1965,6 +1965,22 @@ return the hash table order in the parameter. (If the guest is using
the virtualized real-mode area (VRMA) facility, the kernel will
re-create the VMRA HPTEs on the next KVM_RUN of any vcpu.)

+4.76 KVM_IRQFD
+
+Capability: KVM_CAP_IRQFD
+Architectures: x86
+Type: vm ioctl
+Parameters: struct kvm_irqfd (in)
+Returns: 0 on success, -1 on error
+
+Allows setting an eventfd to directly trigger a guest interrupt.
+kvm_irqfd.fd specifies the file descriptor to use as the eventfd and
+kvm_irqfd.gsi specifies the irqchip pin toggled by this event. When
+an event is tiggered on the eventfd, an interrupt is injected into
+the guest using the specified gsi pin. The irqfd is removed using
+the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
+and kvm_irqfd.gsi.
+

5. The kvm_run structure
------------------------

2012-06-27 05:09:38

by Alex Williamson

[permalink] [raw]

Subject: [PATCH v2 3/6] kvm: Sanitize KVM_IRQFD flags

We only know of one so far.

Signed-off-by: Alex Williamson <[email protected]>
---

virt/kvm/eventfd.c | 3 +++
1 file changed, 3 insertions(+)

diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c
index c307c24..7d7e2aa 100644
--- a/virt/kvm/eventfd.c
+++ b/virt/kvm/eventfd.c
@@ -340,6 +340,9 @@ kvm_irqfd_deassign(struct kvm *kvm, struct kvm_irqfd *args)
int
kvm_irqfd(struct kvm *kvm, struct kvm_irqfd *args)
{
+ if (args->flags & ~KVM_IRQFD_FLAG_DEASSIGN)
+ return -EINVAL;
+
if (args->flags & KVM_IRQFD_FLAG_DEASSIGN)
return kvm_irqfd_deassign(kvm, args);

2012-06-27 05:09:52

by Alex Williamson

[permalink] [raw]

Subject: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

In order to inject an interrupt from an external source using an
irqfd, we need to allocate a new irq_source_id. This allows us to
assert and (later) de-assert an interrupt line independently from
users of KVM_IRQ_LINE and avoid lost interrupts.

We also add what may appear like a bit of excessive infrastructure
around an object for storing this irq_source_id. However, notice
that we only provide a way to assert the interrupt here. A follow-on
interface will make use of the same irq_source_id to allow de-assert.

Signed-off-by: Alex Williamson <[email protected]>
---

Documentation/virtual/kvm/api.txt | 5 ++
arch/x86/kvm/x86.c | 1
include/linux/kvm.h | 3 +
virt/kvm/eventfd.c | 95 +++++++++++++++++++++++++++++++++++--
4 files changed, 99 insertions(+), 5 deletions(-)

diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
index ea9edce..b216709 100644
--- a/Documentation/virtual/kvm/api.txt
+++ b/Documentation/virtual/kvm/api.txt
@@ -1981,6 +1981,11 @@ the guest using the specified gsi pin. The irqfd is removed using
the KVM_IRQFD_FLAG_DEASSIGN flag, specifying both kvm_irqfd.fd
and kvm_irqfd.gsi.

+With KVM_IRQFD_FLAG_LEVEL KVM_IRQFD allocates a new IRQ source ID for
+the requested irqfd. This is necessary to share level triggered
+interrupts with those injected through KVM_IRQ_LINE. IRQFDs created
+with KVM_IRQFD_FLAG_LEVEL must also set this flag when de-assiging.
+KVM_IRQFD_FLAG_LEVEL support is indicated by KVM_CAP_IRQFD_LEVEL.

5. The kvm_run structure
------------------------
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index a01a424..80bed07 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -2148,6 +2148,7 @@ int kvm_dev_ioctl_check_extension(long ext)
case KVM_CAP_GET_TSC_KHZ:
case KVM_CAP_PCI_2_3:
case KVM_CAP_KVMCLOCK_CTRL:
+ case KVM_CAP_IRQFD_LEVEL:
r = 1;
break;
case KVM_CAP_COALESCED_MMIO:
diff --git a/include/linux/kvm.h b/include/linux/kvm.h
index 2ce09aa..b2e6e4f 100644
--- a/include/linux/kvm.h
+++ b/include/linux/kvm.h
@@ -618,6 +618,7 @@ struct kvm_ppc_smmu_info {
#define KVM_CAP_PPC_GET_SMMU_INFO 78
#define KVM_CAP_S390_COW 79
#define KVM_CAP_PPC_ALLOC_HTAB 80
+#define KVM_CAP_IRQFD_LEVEL 81

#ifdef KVM_CAP_IRQ_ROUTING

@@ -683,6 +684,8 @@ struct kvm_xen_hvm_config {
#endif

#define KVM_IRQFD_FLAG_DEASSIGN (1 << 0)
+/* Available with KVM_CAP_IRQFD_LEVEL */
+#define KVM_IRQFD_FLAG_LEVEL (1 << 1)

struct kvm_irqfd {
__u32 fd;
diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c
index 7d7e2aa..18cc284 100644
--- a/virt/kvm/eventfd.c
+++ b/virt/kvm/eventfd.c
@@ -36,6 +36,64 @@
#include "iodev.h"

/*
+ * An irq_source_id can be created from KVM_IRQFD for level interrupt
+ * injections and shared with other interfaces for EOI or de-assert.
+ * Create an object with reference counting to make it easy to use.
+ */
+struct _irq_source {
+ int id;
+ struct kvm *kvm;
+ struct kref kref;
+};
+
+static void release_irq_source(struct kref *kref)
+{
+ struct _irq_source *source;
+
+ source = container_of(kref, struct _irq_source, kref);
+
+ kvm_free_irq_source_id(source->kvm, source->id);
+ kfree(source);
+}
+
+static void put_irq_source(struct _irq_source *source)
+{
+ if (source)
+ kref_put(&source->kref, release_irq_source);
+}
+
+static struct _irq_source *__attribute__ ((used)) /* white lie for now */
+get_irq_source(struct _irq_source *source)
+{
+ if (source)
+ kref_get(&source->kref);
+
+ return source;
+}
+
+static struct _irq_source *new_irq_source(struct kvm *kvm)
+{
+ struct _irq_source *source;
+ int id;
+
+ source = kzalloc(sizeof(*source), GFP_KERNEL);
+ if (!source)
+ return ERR_PTR(-ENOMEM);
+
+ id = kvm_request_irq_source_id(kvm);
+ if (id < 0) {
+ kfree(source);
+ return ERR_PTR(id);
+ }
+
+ kref_init(&source->kref);
+ source->kvm = kvm;
+ source->id = id;
+
+ return source;
+}
+
+/*
* --------------------------------------------------------------------
* irqfd: Allows an fd to be used to inject an interrupt to the guest
*
@@ -52,6 +110,7 @@ struct _irqfd {
/* Used for level IRQ fast-path */
int gsi;
struct work_struct inject;
+ struct _irq_source *source;
/* Used for setup/shutdown */
struct eventfd_ctx *eventfd;
struct list_head list;
@@ -62,7 +121,7 @@ struct _irqfd {
static struct workqueue_struct *irqfd_cleanup_wq;

static void
-irqfd_inject(struct work_struct *work)
+irqfd_inject_edge(struct work_struct *work)
{
struct _irqfd *irqfd = container_of(work, struct _irqfd, inject);
struct kvm *kvm = irqfd->kvm;
@@ -71,6 +130,14 @@ irqfd_inject(struct work_struct *work)
kvm_set_irq(kvm, KVM_USERSPACE_IRQ_SOURCE_ID, irqfd->gsi, 0);
}

+static void
+irqfd_inject_level(struct work_struct *work)
+{
+ struct _irqfd *irqfd = container_of(work, struct _irqfd, inject);
+
+ kvm_set_irq(irqfd->kvm, irqfd->source->id, irqfd->gsi, 1);
+}
+
/*
* Race-free decouple logic (ordering is critical)
*/
@@ -96,6 +163,9 @@ irqfd_shutdown(struct work_struct *work)
* It is now safe to release the object's resources
*/
eventfd_ctx_put(irqfd->eventfd);
+
+ put_irq_source(irqfd->source);
+
kfree(irqfd);
}

@@ -214,7 +284,19 @@ kvm_irqfd_assign(struct kvm *kvm, struct kvm_irqfd *args)
irqfd->kvm = kvm;
irqfd->gsi = args->gsi;
INIT_LIST_HEAD(&irqfd->list);
- INIT_WORK(&irqfd->inject, irqfd_inject);
+
+ if (args->flags & KVM_IRQFD_FLAG_LEVEL) {
+ irqfd->source = new_irq_source(kvm);
+ if (IS_ERR(irqfd->source)) {
+ ret = PTR_ERR(irqfd->source);
+ irqfd->source = NULL;
+ goto fail;
+ }
+
+ INIT_WORK(&irqfd->inject, irqfd_inject_level);
+ } else
+ INIT_WORK(&irqfd->inject, irqfd_inject_edge);
+
INIT_WORK(&irqfd->shutdown, irqfd_shutdown);

file = eventfd_fget(args->fd);
@@ -279,9 +361,10 @@ fail:
if (eventfd && !IS_ERR(eventfd))
eventfd_ctx_put(eventfd);

- if (!IS_ERR(file))
+ if (file && !IS_ERR(file))
fput(file);

+ put_irq_source(irqfd->source);
kfree(irqfd);
return ret;
}
@@ -302,6 +385,7 @@ kvm_irqfd_deassign(struct kvm *kvm, struct kvm_irqfd *args)
{
struct _irqfd *irqfd, *tmp;
struct eventfd_ctx *eventfd;
+ bool is_level = (args->flags & KVM_IRQFD_FLAG_LEVEL) != 0;

eventfd = eventfd_ctx_fdget(args->fd);
if (IS_ERR(eventfd))
@@ -310,7 +394,8 @@ kvm_irqfd_deassign(struct kvm *kvm, struct kvm_irqfd *args)
spin_lock_irq(&kvm->irqfds.lock);

list_for_each_entry_safe(irqfd, tmp, &kvm->irqfds.items, list) {
- if (irqfd->eventfd == eventfd && irqfd->gsi == args->gsi) {
+ if (irqfd->eventfd == eventfd && irqfd->gsi == args->gsi &&
+ is_level == (irqfd->source != NULL)) {
/*
* This rcu_assign_pointer is needed for when
* another thread calls kvm_irq_routing_update before
@@ -340,7 +425,7 @@ kvm_irqfd_deassign(struct kvm *kvm, struct kvm_irqfd *args)
int
kvm_irqfd(struct kvm *kvm, struct kvm_irqfd *args)
{
- if (args->flags & ~KVM_IRQFD_FLAG_DEASSIGN)
+ if (args->flags & ~(KVM_IRQFD_FLAG_DEASSIGN | KVM_IRQFD_FLAG_LEVEL))
return -EINVAL;

if (args->flags & KVM_IRQFD_FLAG_DEASSIGN)

2012-06-27 05:10:15

by Alex Williamson

[permalink] [raw]

Subject: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

This new ioctl enables an eventfd to be triggered when an EOI is
written for a specified irqchip pin. By default this is a simple
notification, but we can also tie the eoifd to a level irqfd, which
enables the irqchip pin to be automatically de-asserted on EOI.
This mode is particularly useful for device-assignment applications
where the unmask and notify triggers a hardware unmask. The default
mode is most applicable to simple notify with no side-effects for
userspace usage, such as Qemu.

Here we make use of the reference counting of the _irq_source
object allowing us to share it with an irqfd and cleanup regardless
of the release order.

Signed-off-by: Alex Williamson <[email protected]>
---

Documentation/virtual/kvm/api.txt | 24 +++++
arch/x86/kvm/x86.c | 1
include/linux/kvm.h | 14 +++
include/linux/kvm_host.h | 13 +++
virt/kvm/eventfd.c | 189 +++++++++++++++++++++++++++++++++++++
virt/kvm/kvm_main.c | 11 ++
6 files changed, 250 insertions(+), 2 deletions(-)

diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
index b216709..87a2558 100644
--- a/Documentation/virtual/kvm/api.txt
+++ b/Documentation/virtual/kvm/api.txt
@@ -1987,6 +1987,30 @@ interrupts with those injected through KVM_IRQ_LINE. IRQFDs created
with KVM_IRQFD_FLAG_LEVEL must also set this flag when de-assiging.
KVM_IRQFD_FLAG_LEVEL support is indicated by KVM_CAP_IRQFD_LEVEL.

+4.77 KVM_EOIFD
+
+Capability: KVM_CAP_EOIFD
+Architectures: x86
+Type: vm ioctl
+Parameters: struct kvm_eoifd (in)
+Returns: 0 on success, -1 on error
+
+KVM_EOIFD allows userspace to receive EOI notification through an
+eventfd for level triggered irqchip interrupts. Behavior for edge
+triggered interrupts is undefined. kvm_eoifd.fd specifies the eventfd
+used for notification and kvm_eoifd.gsi specifies the irchip pin,
+similar to KVM_IRQFD. KVM_EOIFD_FLAG_DEASSIGN is used to deassign
+a previously enabled eoifd and should also set fd and gsi to match.
+
+The KVM_EOIFD_FLAG_LEVEL_IRQFD flag indicates that the EOI is for
+a level triggered EOI and the kvm_eoifd structure includes
+kvm_eoifd.irqfd, which must be previously configured using KVM_IRQFD
+with the KVM_IRQFD_FLAG_LEVEL flag. This allows both EOI notification
+through kvm_eoifd.fd as well as automatically de-asserting level
+irqfds on EOI. Both KVM_EOIFD_FLAG_DEASSIGN and
+KVM_EOIFD_FLAG_LEVEL_IRQFD should be used to de-assign an eoifd
+initially setup with KVM_EOIFD_FLAG_LEVEL_IRQFD.
+
5. The kvm_run structure
------------------------

diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 80bed07..62d6eca 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -2149,6 +2149,7 @@ int kvm_dev_ioctl_check_extension(long ext)
case KVM_CAP_PCI_2_3:
case KVM_CAP_KVMCLOCK_CTRL:
case KVM_CAP_IRQFD_LEVEL:
+ case KVM_CAP_EOIFD:
r = 1;
break;
case KVM_CAP_COALESCED_MMIO:
diff --git a/include/linux/kvm.h b/include/linux/kvm.h
index b2e6e4f..7567e7d 100644
--- a/include/linux/kvm.h
+++ b/include/linux/kvm.h
@@ -619,6 +619,7 @@ struct kvm_ppc_smmu_info {
#define KVM_CAP_S390_COW 79
#define KVM_CAP_PPC_ALLOC_HTAB 80
#define KVM_CAP_IRQFD_LEVEL 81
+#define KVM_CAP_EOIFD 82

#ifdef KVM_CAP_IRQ_ROUTING

@@ -694,6 +695,17 @@ struct kvm_irqfd {
__u8 pad[20];
};

+#define KVM_EOIFD_FLAG_DEASSIGN (1 << 0)
+#define KVM_EOIFD_FLAG_LEVEL_IRQFD (1 << 1)
+
+struct kvm_eoifd {
+ __u32 fd;
+ __u32 gsi;
+ __u32 flags;
+ __u32 irqfd;
+ __u8 pad[16];
+};
+
struct kvm_clock_data {
__u64 clock;
__u32 flags;
@@ -834,6 +846,8 @@ struct kvm_s390_ucas_mapping {
#define KVM_PPC_GET_SMMU_INFO _IOR(KVMIO, 0xa6, struct kvm_ppc_smmu_info)
/* Available with KVM_CAP_PPC_ALLOC_HTAB */
#define KVM_PPC_ALLOCATE_HTAB _IOWR(KVMIO, 0xa7, __u32)
+/* Available with KVM_CAP_EOIFD */
+#define KVM_EOIFD _IOW(KVMIO, 0xa8, struct kvm_eoifd)

/*
* ioctls for vcpu fds
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index ae3b426..83472eb 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -285,6 +285,10 @@ struct kvm {
struct list_head items;
} irqfds;
struct list_head ioeventfds;
+ struct {
+ spinlock_t lock;
+ struct list_head items;
+ } eoifds;
#endif
struct kvm_vm_stat stat;
struct kvm_arch arch;
@@ -828,6 +832,8 @@ int kvm_irqfd(struct kvm *kvm, struct kvm_irqfd *args);
void kvm_irqfd_release(struct kvm *kvm);
void kvm_irq_routing_update(struct kvm *, struct kvm_irq_routing_table *);
int kvm_ioeventfd(struct kvm *kvm, struct kvm_ioeventfd *args);
+int kvm_eoifd(struct kvm *kvm, struct kvm_eoifd *args);
+void kvm_eoifd_release(struct kvm *kvm);

#else

@@ -853,6 +859,13 @@ static inline int kvm_ioeventfd(struct kvm *kvm, struct kvm_ioeventfd *args)
return -ENOSYS;
}

+static inline int kvm_eoifd(struct kvm *kvm, struct kvm_eoifd *args)
+{
+ return -ENOSYS;
+}
+
+static inline void kvm_eoifd_release(struct kvm *kvm) {}
+
#endif /* CONFIG_HAVE_KVM_EVENTFD */

#ifdef CONFIG_KVM_APIC_ARCHITECTURE
diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c
index 18cc284..02ca50f 100644
--- a/virt/kvm/eventfd.c
+++ b/virt/kvm/eventfd.c
@@ -62,8 +62,7 @@ static void put_irq_source(struct _irq_source *source)
kref_put(&source->kref, release_irq_source);
}

-static struct _irq_source *__attribute__ ((used)) /* white lie for now */
-get_irq_source(struct _irq_source *source)
+static struct _irq_source *get_irq_source(struct _irq_source *source)
{
if (source)
kref_get(&source->kref);
@@ -118,6 +117,41 @@ struct _irqfd {
struct work_struct shutdown;
};

+static struct _irq_source *get_irq_source_from_irqfd(struct kvm *kvm, int fd)
+{
+ struct file *file;
+ struct eventfd_ctx *eventfd;
+ struct _irqfd *irqfd;
+ struct _irq_source *source = NULL;
+
+ file = fget(fd);
+ if (!file)
+ return ERR_PTR(-EBADF);
+
+ eventfd = eventfd_ctx_fileget(file);
+ if (IS_ERR(eventfd)) {
+ fput(file);
+ return (struct _irq_source *)eventfd;
+ }
+
+ spin_lock_irq(&kvm->irqfds.lock);
+
+ list_for_each_entry(irqfd, &kvm->irqfds.items, list) {
+ if (irqfd->eventfd != eventfd)
+ continue;
+
+ source = get_irq_source(irqfd->source);
+ break;
+ }
+
+ spin_unlock_irq(&kvm->irqfds.lock);
+
+ eventfd_ctx_put(eventfd);
+ fput(file);
+
+ return source ? source : ERR_PTR(-ENODEV);
+}
+
static struct workqueue_struct *irqfd_cleanup_wq;

static void
@@ -375,6 +409,8 @@ kvm_eventfd_init(struct kvm *kvm)
spin_lock_init(&kvm->irqfds.lock);
INIT_LIST_HEAD(&kvm->irqfds.items);
INIT_LIST_HEAD(&kvm->ioeventfds);
+ spin_lock_init(&kvm->eoifds.lock);
+ INIT_LIST_HEAD(&kvm->eoifds.items);
}

/*
@@ -743,3 +779,152 @@ kvm_ioeventfd(struct kvm *kvm, struct kvm_ioeventfd *args)

return kvm_assign_ioeventfd(kvm, args);
}
+
+/*
+ * --------------------------------------------------------------------
+ * eoifd: Translate KVM APIC/IOAPIC EOI into eventfd signal.
+ *
+ * userspace can register GSIs with an eventfd for receiving
+ * notification when an EOI occurs.
+ * --------------------------------------------------------------------
+ */
+
+struct _eoifd {
+ struct kvm *kvm;
+ struct eventfd_ctx *eventfd;
+ struct _irq_source *source;
+ struct kvm_irq_ack_notifier notifier;
+ struct list_head list;
+};
+
+static void eoifd_event(struct kvm_irq_ack_notifier *notifier)
+{
+ struct _eoifd *eoifd;
+
+ eoifd = container_of(notifier, struct _eoifd, notifier);
+
+ if (eoifd->source)
+ kvm_set_irq(eoifd->kvm, eoifd->source->id,
+ eoifd->notifier.gsi, 0);
+
+ eventfd_signal(eoifd->eventfd, 1);
+}
+
+static int kvm_assign_eoifd(struct kvm *kvm, struct kvm_eoifd *args)
+{
+ struct eventfd_ctx *eventfd;
+ struct _eoifd *eoifd, *tmp;
+ struct _irq_source *source = NULL;
+
+ if (args->flags & KVM_EOIFD_FLAG_LEVEL_IRQFD) {
+ source = get_irq_source_from_irqfd(kvm, args->irqfd);
+ if (IS_ERR(source))
+ return PTR_ERR(source);
+ }
+
+ eventfd = eventfd_ctx_fdget(args->fd);
+ if (IS_ERR(eventfd)) {
+ put_irq_source(source);
+ return PTR_ERR(eventfd);
+ }
+
+ eoifd = kzalloc(sizeof(*eoifd), GFP_KERNEL);
+ if (!eoifd) {
+ put_irq_source(source);
+ eventfd_ctx_put(eventfd);
+ return -ENOMEM;
+ }
+
+ INIT_LIST_HEAD(&eoifd->list);
+ eoifd->kvm = kvm;
+ eoifd->eventfd = eventfd;
+ eoifd->source = source;
+ eoifd->notifier.gsi = args->gsi;
+ eoifd->notifier.irq_acked = eoifd_event;
+
+ spin_lock_irq(&kvm->eoifds.lock);
+
+ list_for_each_entry(tmp, &kvm->eoifds.items, list) {
+ if (eoifd->eventfd != tmp->eventfd)
+ continue;
+
+ put_irq_source(source);
+ eventfd_ctx_put(eventfd);
+ kfree(eoifd);
+ return -EBUSY;
+ }
+
+ list_add_tail(&eoifd->list, &kvm->eoifds.items);
+ kvm_register_irq_ack_notifier(kvm, &eoifd->notifier);
+
+ spin_unlock_irq(&kvm->eoifds.lock);
+
+ return 0;
+}
+
+static void eoifd_deactivate(struct kvm *kvm, struct _eoifd *eoifd)
+{
+ list_del(&eoifd->list);
+ kvm_unregister_irq_ack_notifier(kvm, &eoifd->notifier);
+ put_irq_source(eoifd->source);
+ eventfd_ctx_put(eoifd->eventfd);
+ kfree(eoifd);
+}
+
+void kvm_eoifd_release(struct kvm *kvm)
+{
+ struct _eoifd *eoifd, *tmp;
+
+ spin_lock_irq(&kvm->eoifds.lock);
+
+ list_for_each_entry_safe(eoifd, tmp, &kvm->eoifds.items, list)
+ eoifd_deactivate(kvm, eoifd);
+
+ spin_unlock_irq(&kvm->eoifds.lock);
+}
+
+static int kvm_deassign_eoifd(struct kvm *kvm, struct kvm_eoifd *args)
+{
+ struct eventfd_ctx *eventfd;
+ struct _eoifd *eoifd;
+ bool uses_source = (args->flags & KVM_EOIFD_FLAG_LEVEL_IRQFD) != 0;
+ int ret = -ENODEV;
+
+ eventfd = eventfd_ctx_fdget(args->fd);
+ if (IS_ERR(eventfd))
+ return PTR_ERR(eventfd);
+
+ spin_lock_irq(&kvm->eoifds.lock);
+
+ list_for_each_entry(eoifd, &kvm->eoifds.items, list) {
+ /*
+ * Matching eventfd is unique since we don't allow dulicates,
+ * the rest is sanitizing the calling parameters.
+ */
+ if (eoifd->eventfd == eventfd &&
+ eoifd->notifier.gsi == args->gsi &&
+ uses_source == (eoifd->source != NULL)) {
+ eoifd_deactivate(kvm, eoifd);
+ ret = 0;
+ break;
+ }
+ }
+
+ spin_unlock_irq(&kvm->eoifds.lock);
+
+ eventfd_ctx_put(eventfd);
+
+ return ret;
+}
+
+int kvm_eoifd(struct kvm *kvm, struct kvm_eoifd *args)
+{
+ if (args->flags & ~(KVM_EOIFD_FLAG_DEASSIGN |
+ KVM_EOIFD_FLAG_LEVEL_IRQFD))
+ return -EINVAL;
+
+ if (args->flags & KVM_EOIFD_FLAG_DEASSIGN)
+ return kvm_deassign_eoifd(kvm, args);
+
+ return kvm_assign_eoifd(kvm, args);
+}
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index b4ad14cc..5b41df1 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -620,6 +620,8 @@ static int kvm_vm_release(struct inode *inode, struct file *filp)

kvm_irqfd_release(kvm);

+ kvm_eoifd_release(kvm);
+
kvm_put_kvm(kvm);
return 0;
}
@@ -2093,6 +2095,15 @@ static long kvm_vm_ioctl(struct file *filp,
break;
}
#endif
+ case KVM_EOIFD: {
+ struct kvm_eoifd data;
+
+ r = -EFAULT;
+ if (copy_from_user(&data, argp, sizeof data))
+ goto out;
+ r = kvm_eoifd(kvm, &data);
+ break;
+ }
default:
r = kvm_arch_vm_ioctl(filp, ioctl, arg);
if (r == -ENOTTY)

2012-06-27 05:10:26

by Alex Williamson

[permalink] [raw]

Subject: [PATCH v2 6/6] kvm: Level IRQ de-assert for KVM_IRQFD

This is an alternate level irqfd de-assert mode that's potentially
useful for emulated drivers. It's included here to show how easy it
is to implement with the new level irqfd and eoifd support. It's
possible this mode might also prove interesting for device-assignment
where we inject via level irqfd, receive an EOI (w/o de-assert), and
use the level de-assert irqfd here.

Signed-off-by: Alex Williamson <[email protected]>
---

Documentation/virtual/kvm/api.txt | 8 ++++++++
include/linux/kvm.h | 6 +++++-
virt/kvm/eventfd.c | 28 ++++++++++++++++++++++++++--
3 files changed, 39 insertions(+), 3 deletions(-)

diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt
index 87a2558..b356937 100644
--- a/Documentation/virtual/kvm/api.txt
+++ b/Documentation/virtual/kvm/api.txt
@@ -1987,6 +1987,14 @@ interrupts with those injected through KVM_IRQ_LINE. IRQFDs created
with KVM_IRQFD_FLAG_LEVEL must also set this flag when de-assiging.
KVM_IRQFD_FLAG_LEVEL support is indicated by KVM_CAP_IRQFD_LEVEL.

+The KVM_IRQFD_FLAG_LEVEL_DEASSERT flag creates an irqfd similar to
+KVM_IRQFD_FLAG_LEVEL, except the irqfd de-asserts the irqchip pin
+rather than asserts it. The level irqfd must first be created as
+aboved and passed to this ioctl in kvm_irqfd.irqfd. This ensures
+the de-assert irqfd uses the same IRQ source ID as the assert irqfd.
+This flag should also be specified on de-assign. This feature is
+present when KVM_CAP_IRQFD_LEVEL_DEASSERT is available.
+
4.77 KVM_EOIFD

Capability: KVM_CAP_EOIFD
diff --git a/include/linux/kvm.h b/include/linux/kvm.h
index 7567e7d..0bbfd47 100644
--- a/include/linux/kvm.h
+++ b/include/linux/kvm.h
@@ -620,6 +620,7 @@ struct kvm_ppc_smmu_info {
#define KVM_CAP_PPC_ALLOC_HTAB 80
#define KVM_CAP_IRQFD_LEVEL 81
#define KVM_CAP_EOIFD 82
+#define KVM_CAP_IRQFD_LEVEL_DEASSERT 83

#ifdef KVM_CAP_IRQ_ROUTING

@@ -687,12 +688,15 @@ struct kvm_xen_hvm_config {
#define KVM_IRQFD_FLAG_DEASSIGN (1 << 0)
/* Available with KVM_CAP_IRQFD_LEVEL */
#define KVM_IRQFD_FLAG_LEVEL (1 << 1)
+/* Available with KVM_CAP_IRQFD_LEVEL_DEASSERT */
+#define KVM_IRQFD_FLAG_LEVEL_DEASSERT (1 << 2)

struct kvm_irqfd {
__u32 fd;
__u32 gsi;
__u32 flags;
- __u8 pad[20];
+ __u32 irqfd;
+ __u8 pad[16];
};

#define KVM_EOIFD_FLAG_DEASSIGN (1 << 0)
diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c
index 02ca50f..50ace0e 100644
--- a/virt/kvm/eventfd.c
+++ b/virt/kvm/eventfd.c
@@ -172,6 +172,14 @@ irqfd_inject_level(struct work_struct *work)
kvm_set_irq(irqfd->kvm, irqfd->source->id, irqfd->gsi, 1);
}

+static void
+irqfd_inject_level_deassert(struct work_struct *work)
+{
+ struct _irqfd *irqfd = container_of(work, struct _irqfd, inject);
+
+ kvm_set_irq(irqfd->kvm, irqfd->source->id, irqfd->gsi, 0);
+}
+
/*
* Race-free decouple logic (ordering is critical)
*/
@@ -320,6 +328,9 @@ kvm_irqfd_assign(struct kvm *kvm, struct kvm_irqfd *args)
INIT_LIST_HEAD(&irqfd->list);

if (args->flags & KVM_IRQFD_FLAG_LEVEL) {
+ if (args->flags & KVM_IRQFD_FLAG_LEVEL_DEASSERT)
+ return -EINVAL; /* mutually exclusive */
+
irqfd->source = new_irq_source(kvm);
if (IS_ERR(irqfd->source)) {
ret = PTR_ERR(irqfd->source);
@@ -328,6 +339,16 @@ kvm_irqfd_assign(struct kvm *kvm, struct kvm_irqfd *args)
}

INIT_WORK(&irqfd->inject, irqfd_inject_level);
+
+ } else if (args->flags & KVM_IRQFD_FLAG_LEVEL_DEASSERT) {
+ irqfd->source = get_irq_source_from_irqfd(kvm, args->irqfd);
+ if (IS_ERR(irqfd->source)) {
+ ret = PTR_ERR(irqfd->source);
+ irqfd->source = NULL;
+ goto fail;
+ }
+
+ INIT_WORK(&irqfd->inject, irqfd_inject_level_deassert);
} else
INIT_WORK(&irqfd->inject, irqfd_inject_edge);

@@ -421,7 +442,8 @@ kvm_irqfd_deassign(struct kvm *kvm, struct kvm_irqfd *args)
{
struct _irqfd *irqfd, *tmp;
struct eventfd_ctx *eventfd;
- bool is_level = (args->flags & KVM_IRQFD_FLAG_LEVEL) != 0;
+ bool is_level = (args->flags & (KVM_IRQFD_FLAG_LEVEL |
+ KVM_IRQFD_FLAG_LEVEL_DEASSERT)) != 0;

eventfd = eventfd_ctx_fdget(args->fd);
if (IS_ERR(eventfd))
@@ -461,7 +483,9 @@ kvm_irqfd_deassign(struct kvm *kvm, struct kvm_irqfd *args)
int
kvm_irqfd(struct kvm *kvm, struct kvm_irqfd *args)
{
- if (args->flags & ~(KVM_IRQFD_FLAG_DEASSIGN | KVM_IRQFD_FLAG_LEVEL))
+ if (args->flags & ~(KVM_IRQFD_FLAG_DEASSIGN |
+ KVM_IRQFD_FLAG_LEVEL |
+ KVM_IRQFD_FLAG_LEVEL_DEASSERT))
return -EINVAL;

if (args->flags & KVM_IRQFD_FLAG_DEASSIGN)

2012-06-27 09:15:41

Subject: [PATCH v2 0/6] kvm: level triggered irqfd support

Subject: [PATCH v2 1/6] kvm: Pass kvm_irqfd to functions

Subject: [PATCH v2 2/6] kvm: Add missing KVM_IRQFD API documentation

Subject: [PATCH v2 3/6] kvm: Sanitize KVM_IRQFD flags

Subject: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

Subject: [PATCH v2 6/6] kvm: Level IRQ de-assert for KVM_IRQFD

Subject: Re: [PATCH v2 0/6] kvm: level triggered irqfd support

Subject: Re: [PATCH v2 3/6] kvm: Sanitize KVM_IRQFD flags

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 1/6] kvm: Pass kvm_irqfd to functions

Subject: Re: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 2/6] kvm: Add missing KVM_IRQFD API documentation

Subject: Re: [PATCH v2 0/6] kvm: level triggered irqfd support

Subject: Re: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

Subject: Re: [PATCH v2 1/6] kvm: Pass kvm_irqfd to functions

Subject: Re: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

Subject: Re: [PATCH v2 1/6] kvm: Pass kvm_irqfd to functions

Subject: Re: [PATCH v2 0/6] kvm: level triggered irqfd support

Subject: Re: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

Subject: Re: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 3/6] kvm: Sanitize KVM_IRQFD flags

Subject: Re: [PATCH v2 3/6] kvm: Sanitize KVM_IRQFD flags

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 1/6] kvm: Pass kvm_irqfd to functions

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 0/6] kvm: level triggered irqfd support

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 1/6] kvm: Pass kvm_irqfd to functions

Subject: Re: [PATCH v2 1/6] kvm: Pass kvm_irqfd to functions

Subject: Re: [PATCH v2 1/6] kvm: Pass kvm_irqfd to functions

Subject: Re: [PATCH v2 1/6] kvm: Pass kvm_irqfd to functions

Subject: Re: [PATCH v2 3/6] kvm: Sanitize KVM_IRQFD flags

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 6/6] kvm: Level IRQ de-assert for KVM_IRQFD

Subject: Re: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

Subject: Re: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

Subject: Re: [PATCH v2 1/6] kvm: Pass kvm_irqfd to functions

Subject: Re: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

Subject: Re: [PATCH v2 1/6] kvm: Pass kvm_irqfd to functions

Subject: Re: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

Subject: Re: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

Subject: Re: [PATCH v2 5/6] kvm: KVM_EOIFD, an eventfd for EOIs

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts

Subject: Re: [PATCH v2 1/6] kvm: Pass kvm_irqfd to functions

Subject: Re: [PATCH v2 6/6] kvm: Level IRQ de-assert for KVM_IRQFD

Subject: Re: [PATCH v2 4/6] kvm: Extend irqfd to support level interrupts