[KVM PATCH v5 0/2] irqfd

[KVM PATCH v5 0/2] irqfd

[Gregory Haskins]

(Applies to kvm.git:66b0aed4, plus you will also need Davide Libenzi's
eventfd_file_create() patch, which you can find here:

http://www.mail-archive.com/kvm@vger.kernel.org/msg13923.html

You can find my complete tree with kvm.git, Davide's patch, and this series
here:

http://git.kernel.org/?p=linux/kernel/git/ghaskins/vbus/linux-2.6.git;a=shortlog;h=irqfd

----------------------

irqfd, v5

This series implements a mechanism called "irqfd".  It lets you create
an eventfd based file-desriptor to inject interrupts to a kvm guest. For
more details, please see the prologue for patch 2/2.

I am reasonably satisfied with this series, so Avi please consider for
inclusion contingent on Davide formally submitting his eventfd_file_create()
patch (and pending any new review comments, of course).

[ Changelog:

   v5:
        *) Added padding to the ioctl structure
        *) Added proper ref-count increment to the file before returning
           success. (Needs review by Al Viro, Davide Libenzi)
        *) Cleaned up error-handling path to make sure we remove ourself
           from the waitq if necessary.
        *) Make sure we only add ourselves to kvm->irqfds if successful
           creating the irqfd in the first place.
        *) Rebased to kvm.git:66b0aed4 

   v4:
        *) Changed allocation model to create the new fd last, after
           we get past the last potential error point by using Davide's
           new eventfd_file_create interface (Al Viro, Davide Libenzi)
	*) We no longer export sys_eventfd2() since it is replaced
           functionally with eventfd_file_create();
        *) Rebased to kvm.git:7da2e3ba

   v3:
        *) The kernel now allocates the eventfd (need to export sys_eventfd2)
        *) Added a flags field for future expansion to kvm_irqfd()
        *) We properly toggle the irq level 1+0.
        *) We re-use the USERSPACE_SRC_ID instead of creating our own
        *) Properly check for failures establishing a poll-table with eventfd
	*) Fixed fd/file leaks on failure
	*) Rebased to lateste kvm.git::41b76d8d04

   v2:
	*) Dropped notifier_chain based callbacks in favor of
	   wait_queue_t::func and file::poll based callbacks (Thanks to
	   Davide for the suggestion)

   v1:
        *) Initial release

--------

We do not have a user of this interface in this series, though note
future version of virtual-bus (v4 and above) will be based on this.

Note that this series requires userspace patches for qemu-kvm.git, v3, which
you can find here: http://patchwork.kernel.org/patch/20213/

-Greg

---

Gregory Haskins (2):
      kvm: add support for irqfd via eventfd-notification interface
      eventfd: export eventfd interfaces for module use


 arch/x86/kvm/Makefile    |    2 -
 arch/x86/kvm/x86.c       |    1 
 fs/eventfd.c             |    4 +
 include/linux/kvm.h      |    8 ++
 include/linux/kvm_host.h |    4 +
 virt/kvm/irqfd.c         |  163 ++++++++++++++++++++++++++++++++++++++++++++++
 virt/kvm/kvm_main.c      |   11 +++
 7 files changed, 192 insertions(+), 1 deletions(-)
 create mode 100644 virt/kvm/irqfd.c

-- 

[KVM PATCH v5 1/2] eventfd: export eventfd interfaces for module use

[Gregory Haskins]

We will re-use eventfd for implmenting irqfd later in the series, and the
irqfd users will potentially live in modules.

Signed-off-by: Gregory Haskins <ghaskins@novell.com>
---

 fs/eventfd.c |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/fs/eventfd.c b/fs/eventfd.c
index 0de6ebb..2e1c2ff 100644
--- a/fs/eventfd.c
+++ b/fs/eventfd.c
@@ -16,6 +16,7 @@
 #include <linux/anon_inodes.h>
 #include <linux/eventfd.h>
 #include <linux/syscalls.h>
+#include <linux/module.h>
 
 struct eventfd_ctx {
 	wait_queue_head_t wqh;
@@ -56,6 +57,7 @@ int eventfd_signal(struct file *file, int n)
 
 	return n;
 }
+EXPORT_SYMBOL_GPL(eventfd_signal);
 
 static int eventfd_release(struct inode *inode, struct file *file)
 {
@@ -197,6 +199,7 @@ struct file *eventfd_fget(int fd)
 
 	return file;
 }
+EXPORT_SYMBOL_GPL(eventfd_fget);
 
 struct file *eventfd_file_create(unsigned int count, int flags)
 {
@@ -225,6 +228,7 @@ struct file *eventfd_file_create(unsigned int count, int flags)
 
 	return file;
 }
+EXPORT_SYMBOL_GPL(eventfd_file_create);
 
 SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags)
 {

[KVM PATCH v5 2/2] kvm: add support for irqfd via eventfd-notification interface

[Gregory Haskins]

KVM provides a complete virtual system environment for guests, including
support for injecting interrupts modeled after the real exception/interrupt
facilities present on the native platform (such as the IDT on x86).
Virtual interrupts can come from a variety of sources (emulated devices,
pass-through devices, etc) but all must be injected to the guest via
the KVM infrastructure.  This patch adds a new mechanism to inject a specific
interrupt to a guest using a decoupled eventfd mechnanism:  Any legal signal
on the irqfd (using eventfd semantics from either userspace or kernel) will
translate into an injected interrupt in the guest at the next available
interrupt window.

Signed-off-by: Gregory Haskins <ghaskins@novell.com>
---

 arch/x86/kvm/Makefile    |    2 -
 arch/x86/kvm/x86.c       |    1 
 include/linux/kvm.h      |    8 ++
 include/linux/kvm_host.h |    4 +
 virt/kvm/irqfd.c         |  163 ++++++++++++++++++++++++++++++++++++++++++++++
 virt/kvm/kvm_main.c      |   11 +++
 6 files changed, 188 insertions(+), 1 deletions(-)
 create mode 100644 virt/kvm/irqfd.c

diff --git a/arch/x86/kvm/Makefile b/arch/x86/kvm/Makefile
index b43c4ef..d5fff51 100644
--- a/arch/x86/kvm/Makefile
+++ b/arch/x86/kvm/Makefile
@@ -3,7 +3,7 @@
 #
 
 common-objs = $(addprefix ../../../virt/kvm/, kvm_main.o ioapic.o \
-                coalesced_mmio.o irq_comm.o)
+                coalesced_mmio.o irq_comm.o irqfd.o)
 ifeq ($(CONFIG_KVM_TRACE),y)
 common-objs += $(addprefix ../../../virt/kvm/, kvm_trace.o)
 endif
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 308d8e9..e68e415 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -1026,6 +1026,7 @@ int kvm_dev_ioctl_check_extension(long ext)
 	case KVM_CAP_REINJECT_CONTROL:
 	case KVM_CAP_IRQ_INJECT_STATUS:
 	case KVM_CAP_ASSIGN_DEV_IRQ:
+	case KVM_CAP_IRQFD:
 		r = 1;
 		break;
 	case KVM_CAP_COALESCED_MMIO:
diff --git a/include/linux/kvm.h b/include/linux/kvm.h
index 3db5d8d..ff8a08a 100644
--- a/include/linux/kvm.h
+++ b/include/linux/kvm.h
@@ -415,6 +415,7 @@ struct kvm_trace_rec {
 #define KVM_CAP_ASSIGN_DEV_IRQ 29
 /* Another bug in KVM_SET_USER_MEMORY_REGION fixed: */
 #define KVM_CAP_JOIN_MEMORY_REGIONS_WORKS 30
+#define KVM_CAP_IRQFD 31
 
 #ifdef KVM_CAP_IRQ_ROUTING
 
@@ -454,6 +455,12 @@ struct kvm_irq_routing {
 
 #endif
 
+struct kvm_irqfd {
+	__u32 gsi;
+	__u32 flags;
+	__u8  pad[24];
+};
+
 /*
  * ioctls for VM fds
  */
@@ -498,6 +505,7 @@ struct kvm_irq_routing {
 #define KVM_ASSIGN_SET_MSIX_ENTRY \
 			_IOW(KVMIO, 0x74, struct kvm_assigned_msix_entry)
 #define KVM_DEASSIGN_DEV_IRQ       _IOW(KVMIO, 0x75, struct kvm_assigned_irq)
+#define KVM_IRQFD               _IOW(KVMIO, 0x76, struct kvm_irqfd)
 
 /*
  * ioctls for vcpu fds
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 095ebb6..6a8d1c1 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -134,6 +134,7 @@ struct kvm {
 	struct list_head vm_list;
 	struct kvm_io_bus mmio_bus;
 	struct kvm_io_bus pio_bus;
+	struct list_head irqfds;
 	struct kvm_vm_stat stat;
 	struct kvm_arch arch;
 	atomic_t users_count;
@@ -524,4 +525,7 @@ static inline void kvm_free_irq_routing(struct kvm *kvm) {}
 
 #endif
 
+int kvm_irqfd(struct kvm *kvm, int gsi, int flags);
+void kvm_irqfd_release(struct kvm *kvm);
+
 #endif
diff --git a/virt/kvm/irqfd.c b/virt/kvm/irqfd.c
new file mode 100644
index 0000000..962e90e
--- /dev/null
+++ b/virt/kvm/irqfd.c
@@ -0,0 +1,163 @@
+/*
+ * irqfd: Allows an eventfd to be used to inject an interrupt to the guest
+ *
+ * Credit goes to Avi Kivity for the original idea.
+ *
+ * Copyright 2009 Novell.  All Rights Reserved.
+ *
+ * Author:
+ *	Gregory Haskins <ghaskins@novell.com>
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of version 2 of the GNU General Public License
+ * as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.	 See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA.
+ */
+
+#include <linux/kvm_host.h>
+#include <linux/eventfd.h>
+#include <linux/workqueue.h>
+#include <linux/syscalls.h>
+#include <linux/wait.h>
+#include <linux/poll.h>
+#include <linux/file.h>
+#include <linux/list.h>
+
+struct _irqfd {
+	struct kvm               *kvm;
+	int                       gsi;
+	struct file              *file;
+	struct list_head          list;
+	poll_table                pt;
+	wait_queue_head_t        *wqh;
+	wait_queue_t              wait;
+	struct work_struct        work;
+};
+
+static void
+irqfd_inject(struct work_struct *work)
+{
+	struct _irqfd *irqfd = container_of(work, struct _irqfd, work);
+	struct kvm *kvm = irqfd->kvm;
+
+	mutex_lock(&kvm->lock);
+	kvm_set_irq(kvm, KVM_USERSPACE_IRQ_SOURCE_ID, irqfd->gsi, 1);
+	kvm_set_irq(kvm, KVM_USERSPACE_IRQ_SOURCE_ID, irqfd->gsi, 0);
+	mutex_unlock(&kvm->lock);
+}
+
+static int
+irqfd_wakeup(wait_queue_t *wait, unsigned mode, int sync, void *key)
+{
+	struct _irqfd *irqfd = container_of(wait, struct _irqfd, wait);
+
+	/*
+	 * The eventfd calls its wake_up with interrupts disabled, and
+	 * eventfd_signal() may be called from interrupt context. Therefore
+	 * we need to defer the IRQ injection until later since we need to
+	 * acquire the kvm->lock to do so.
+	 */
+	schedule_work(&irqfd->work);
+
+	return 0;
+}
+
+static void
+irqfd_ptable_queue_proc(struct file *file, wait_queue_head_t *wqh,
+			poll_table *pt)
+{
+	struct _irqfd *irqfd = container_of(pt, struct _irqfd, pt);
+
+	irqfd->wqh = wqh;
+	add_wait_queue(wqh, &irqfd->wait);
+}
+
+int
+kvm_irqfd(struct kvm *kvm, int gsi, int flags)
+{
+	struct _irqfd *irqfd;
+	struct file *file = NULL;
+	int fd = -1;
+	int ret;
+
+	irqfd = kzalloc(sizeof(*irqfd), GFP_KERNEL);
+	if (!irqfd)
+		return -ENOMEM;
+
+	irqfd->kvm = kvm;
+	irqfd->gsi = gsi;
+	INIT_LIST_HEAD(&irqfd->list);
+	INIT_WORK(&irqfd->work, irqfd_inject);
+
+	/*
+	 * We re-use eventfd for irqfd, and therefore will embed the eventfd
+	 * lifetime in the irqfd.
+	 */
+	file = eventfd_file_create(0, 0);
+	if (IS_ERR(file)) {
+		ret = PTR_ERR(file);
+		goto fail;
+	}
+
+	/*
+	 * Install our own custom wake-up handling so we are notified via
+	 * a callback whenever someone signals the underlying eventfd
+	 */
+	init_waitqueue_func_entry(&irqfd->wait, irqfd_wakeup);
+	init_poll_funcptr(&irqfd->pt, irqfd_ptable_queue_proc);
+
+	ret = file->f_op->poll(file, &irqfd->pt);
+	if (ret < 0)
+		goto fail;
+
+	fd = get_unused_fd();
+	if (fd < 0) {
+		ret = fd;
+		goto fail;
+	}
+
+	fd_install(fd, file);
+
+	get_file(file);
+	irqfd->file = file;
+
+	mutex_lock(&kvm->lock);
+	list_add_tail(&irqfd->list, &kvm->irqfds);
+	mutex_unlock(&kvm->lock);
+
+	return fd;
+
+fail:
+	if (irqfd->wqh)
+		remove_wait_queue(irqfd->wqh, &irqfd->wait);
+
+	if (file && !IS_ERR(file))
+		fput(file);
+
+	kfree(irqfd);
+	return ret;
+}
+
+void
+kvm_irqfd_release(struct kvm *kvm)
+{
+	struct _irqfd *irqfd, *tmp;
+
+	list_for_each_entry_safe(irqfd, tmp, &kvm->irqfds, list) {
+		remove_wait_queue(irqfd->wqh, &irqfd->wait);
+
+		flush_work(&irqfd->work);
+		fput(irqfd->file);
+
+		list_del(&irqfd->list);
+		kfree(irqfd);
+	}
+}
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 2b73e19..8b3b06a 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -972,6 +972,7 @@ static struct kvm *kvm_create_vm(void)
 	atomic_inc(&kvm->mm->mm_count);
 	spin_lock_init(&kvm->mmu_lock);
 	kvm_io_bus_init(&kvm->pio_bus);
+	INIT_LIST_HEAD(&kvm->irqfds);
 	mutex_init(&kvm->lock);
 	kvm_io_bus_init(&kvm->mmio_bus);
 	init_rwsem(&kvm->slots_lock);
@@ -1023,6 +1024,7 @@ static void kvm_destroy_vm(struct kvm *kvm)
 	spin_lock(&kvm_lock);
 	list_del(&kvm->vm_list);
 	spin_unlock(&kvm_lock);
+	kvm_irqfd_release(kvm);
 	kvm_free_irq_routing(kvm);
 	kvm_io_bus_destroy(&kvm->pio_bus);
 	kvm_io_bus_destroy(&kvm->mmio_bus);
@@ -2197,6 +2199,15 @@ static long kvm_vm_ioctl(struct file *filp,
 	}
 #endif
 #endif /* KVM_CAP_IRQ_ROUTING */
+	case KVM_IRQFD: {
+		struct kvm_irqfd data;
+
+		r = -EFAULT;
+		if (copy_from_user(&data, argp, sizeof data))
+			goto out;
+		r = kvm_irqfd(kvm, data.gsi, data.flags);
+		break;
+	}
 	default:
 		r = kvm_arch_vm_ioctl(filp, ioctl, arg);
 	}