From 5b49ab188ff0339aa3097ce7f5309f1306092f9e Mon Sep 17 00:00:00 2001
From: Alex Williamson <alex.williamson@redhat.com>
Date: Fri, 6 Dec 2013 11:16:13 -0700
Subject: vfio-pci: Make use of new KVM-VFIO device
Add and remove groups from the KVM virtual VFIO device as we make
use of them. This allows KVM to optimize for performance and
correctness based on properties of the group.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
---
hw/misc/vfio.c | 67 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 67 insertions(+)
(limited to 'hw/misc/vfio.c')
diff --git a/hw/misc/vfio.c b/hw/misc/vfio.c
index f7f8a19ee8..f1aa4a7209 100644
--- a/hw/misc/vfio.c
+++ b/hw/misc/vfio.c
@@ -208,6 +208,17 @@ static QLIST_HEAD(, VFIOContainer)
static QLIST_HEAD(, VFIOGroup)
group_list = QLIST_HEAD_INITIALIZER(group_list);
+#ifdef CONFIG_KVM
+/*
+ * We have a single VFIO pseudo device per KVM VM. Once created it lives
+ * for the life of the VM. Closing the file descriptor only drops our
+ * reference to it and the device's reference to kvm. Therefore once
+ * initialized, this file descriptor is only released on QEMU exit and
+ * we'll re-use it should another vfio device be attached before then.
+ */
+static int vfio_kvm_device_fd = -1;
+#endif
+
static void vfio_disable_interrupts(VFIODevice *vdev);
static uint32_t vfio_pci_read_config(PCIDevice *pdev, uint32_t addr, int len);
static void vfio_pci_write_config(PCIDevice *pdev, uint32_t addr,
@@ -3041,6 +3052,59 @@ static void vfio_pci_reset_handler(void *opaque)
}
}
+static void vfio_kvm_device_add_group(VFIOGroup *group)
+{
+#ifdef CONFIG_KVM
+ struct kvm_device_attr attr = {
+ .group = KVM_DEV_VFIO_GROUP,
+ .attr = KVM_DEV_VFIO_GROUP_ADD,
+ .addr = (uint64_t)(unsigned long)&group->fd,
+ };
+
+ if (!kvm_enabled()) {
+ return;
+ }
+
+ if (vfio_kvm_device_fd < 0) {
+ struct kvm_create_device cd = {
+ .type = KVM_DEV_TYPE_VFIO,
+ };
+
+ if (kvm_vm_ioctl(kvm_state, KVM_CREATE_DEVICE, &cd)) {
+ DPRINTF("KVM_CREATE_DEVICE: %m\n");
+ return;
+ }
+
+ vfio_kvm_device_fd = cd.fd;
+ }
+
+ if (ioctl(vfio_kvm_device_fd, KVM_SET_DEVICE_ATTR, &attr)) {
+ error_report("Failed to add group %d to KVM VFIO device: %m",
+ group->groupid);
+ }
+#endif
+}
+
+static void vfio_kvm_device_del_group(VFIOGroup *group)
+{
+#ifdef CONFIG_KVM
+ struct kvm_device_attr attr = {
+ .group = KVM_DEV_VFIO_GROUP,
+ .attr = KVM_DEV_VFIO_GROUP_DEL,
+ .addr = (uint64_t)(unsigned long)&group->fd,
+ };
+
+ if (vfio_kvm_device_fd < 0) {
+ return;
+ }
+
+ if (ioctl(vfio_kvm_device_fd, KVM_SET_DEVICE_ATTR, &attr)) {
+ error_report("Failed to remove group %d to KVM VFIO device: %m",
+ group->groupid);
+ }
+#endif
+}
+
static int vfio_connect_container(VFIOGroup *group)
{
VFIOContainer *container;
@@ -3189,6 +3253,8 @@ static VFIOGroup *vfio_get_group(int groupid)
QLIST_INSERT_HEAD(&group_list, group, next);
+ vfio_kvm_device_add_group(group);
+
return group;
}
@@ -3198,6 +3264,7 @@ static void vfio_put_group(VFIOGroup *group)
return;
}
+ vfio_kvm_device_del_group(group);
vfio_disconnect_container(group);
QLIST_REMOVE(group, next);
DPRINTF("vfio_put_group: close group->fd\n");
--
cgit v1.2.3
From 96eeeba0db38b856eb2cae0e4a2a620d8d65771a Mon Sep 17 00:00:00 2001
From: Alex Williamson <alex.williamson@redhat.com>
Date: Fri, 6 Dec 2013 11:16:23 -0700
Subject: vfio-pci: Fix Nvidia MSI ACK through 0x88000 quirk
When MSI is enabled on Nvidia GeForce cards the driver seems to
acknowledge the interrupt by writing a 0xff byte to the MSI capability
ID register using the PCI config space mirror at offset 0x88000 from
BAR0. Without this, the device will only fire a single interrupt.
VFIO handles the PCI capability ID/next registers as virtual w/o write
support, so any write through config space is currently dropped. Add
a check for this and allow the write through the BAR window. The
registers are read-only anyway.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
---
hw/misc/vfio.c | 30 +++++++++++++++++++++++++++++-
1 file changed, 29 insertions(+), 1 deletion(-)
(limited to 'hw/misc/vfio.c')
diff --git a/hw/misc/vfio.c b/hw/misc/vfio.c
index f1aa4a7209..09bdddf18a 100644
--- a/hw/misc/vfio.c
+++ b/hw/misc/vfio.c
@@ -1811,6 +1811,34 @@ static void vfio_probe_nvidia_bar5_window_quirk(VFIODevice *vdev, int nr)
vdev->host.function);
}
+static void vfio_nvidia_88000_quirk_write(void *opaque, hwaddr addr,
+ uint64_t data, unsigned size)
+{
+ VFIOQuirk *quirk = opaque;
+ VFIODevice *vdev = quirk->vdev;
+ PCIDevice *pdev = &vdev->pdev;
+ hwaddr base = quirk->data.address_match & TARGET_PAGE_MASK;
+
+ vfio_generic_quirk_write(opaque, addr, data, size);
+
+ /*
+ * Nvidia seems to acknowledge MSI interrupts by writing 0xff to the
+ * MSI capability ID register. Both the ID and next register are
+ * read-only, so we allow writes covering either of those to real hw.
+ * NB - only fixed for the 0x88000 MMIO window.
+ */
+ if ((pdev->cap_present & QEMU_PCI_CAP_MSI) &&
+ vfio_range_contained(addr, size, pdev->msi_cap, PCI_MSI_FLAGS)) {
+ vfio_bar_write(&vdev->bars[quirk->data.bar], addr + base, data, size);
+ }
+}
+
+static const MemoryRegionOps vfio_nvidia_88000_quirk = {
+ .read = vfio_generic_quirk_read,
+ .write = vfio_nvidia_88000_quirk_write,
+ .endianness = DEVICE_LITTLE_ENDIAN,
+};
+
/*
* Finally, BAR0 itself. We want to redirect any accesses to either
* 0x1800 or 0x88000 through the PCI config space access functions.
@@ -1837,7 +1865,7 @@ static void vfio_probe_nvidia_bar0_88000_quirk(VFIODevice *vdev, int nr)
quirk->data.address_mask = PCIE_CONFIG_SPACE_SIZE - 1;
quirk->data.bar = nr;
- memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_generic_quirk,
+ memory_region_init_io(&quirk->mem, OBJECT(vdev), &vfio_nvidia_88000_quirk,
quirk, "vfio-nvidia-bar0-88000-quirk",
TARGET_PAGE_ALIGN(quirk->data.address_mask + 1));
memory_region_add_subregion_overlap(&vdev->bars[nr].mem,
--
cgit v1.2.3
From b3ebc10c373ed5922d4bdb5076fd0e9fd7ff8056 Mon Sep 17 00:00:00 2001
From: Alex Williamson <alex.williamson@redhat.com>
Date: Fri, 6 Dec 2013 11:16:29 -0700
Subject: vfio-pci: Add debug config options to disable MSI/X KVM support
It's sometimes useful to be able to verify interrupts are passing
through correctly.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
---
hw/misc/vfio.c | 24 ++++++++++++++++++++----
1 file changed, 20 insertions(+), 4 deletions(-)
(limited to 'hw/misc/vfio.c')
diff --git a/hw/misc/vfio.c b/hw/misc/vfio.c
index 09bdddf18a..f367537737 100644
--- a/hw/misc/vfio.c
+++ b/hw/misc/vfio.c
@@ -52,6 +52,8 @@
/* Extra debugging, trap acceleration paths for more logging */
#define VFIO_ALLOW_MMAP 1
#define VFIO_ALLOW_KVM_INTX 1
+#define VFIO_ALLOW_KVM_MSI 1
+#define VFIO_ALLOW_KVM_MSIX 1
struct VFIODevice;
@@ -590,9 +592,21 @@ static void vfio_msi_interrupt(void *opaque)
return;
}
- DPRINTF("%s(%04x:%02x:%02x.%x) vector %d\n", __func__,
+#ifdef VFIO_DEBUG
+ MSIMessage msg;
+
+ if (vdev->interrupt == VFIO_INT_MSIX) {
+ msg = msi_get_message(&vdev->pdev, nr);
+ } else if (vdev->interrupt == VFIO_INT_MSI) {
+ msg = msix_get_message(&vdev->pdev, nr);
+ } else {
+ abort();
+ }
+
+ DPRINTF("%s(%04x:%02x:%02x.%x) vector %d 0x%"PRIx64"/0x%x\n", __func__,
vdev->host.domain, vdev->host.bus, vdev->host.slot,
- vdev->host.function, nr);
+ vdev->host.function, nr, msg.address, msg.data);
+#endif
if (vdev->interrupt == VFIO_INT_MSIX) {
msix_notify(&vdev->pdev, nr);
@@ -660,7 +674,8 @@ static int vfio_msix_vector_do_use(PCIDevice *pdev, unsigned int nr,
* Attempt to enable route through KVM irqchip,
* default to userspace handling if unavailable.
*/
- vector->virq = msg ? kvm_irqchip_add_msi_route(kvm_state, *msg) : -1;
+ vector->virq = msg && VFIO_ALLOW_KVM_MSIX ?
+ kvm_irqchip_add_msi_route(kvm_state, *msg) : -1;
if (vector->virq < 0 ||
kvm_irqchip_add_irqfd_notifier(kvm_state, &vector->interrupt,
NULL, vector->virq) < 0) {
@@ -827,7 +842,8 @@ retry:
* Attempt to enable route through KVM irqchip,
* default to userspace handling if unavailable.
*/
- vector->virq = kvm_irqchip_add_msi_route(kvm_state, vector->msg);
+ vector->virq = VFIO_ALLOW_KVM_MSI ?
+ kvm_irqchip_add_msi_route(kvm_state, vector->msg) : -1;
if (vector->virq < 0 ||
kvm_irqchip_add_irqfd_notifier(kvm_state, &vector->interrupt,
NULL, vector->virq) < 0) {
--
cgit v1.2.3
From 3e40ba0faf0822fa78336fe6cd9d677ea9b14f1b Mon Sep 17 00:00:00 2001
From: Alex Williamson <alex.williamson@redhat.com>
Date: Fri, 6 Dec 2013 11:16:40 -0700
Subject: vfio-pci: Release all MSI-X vectors when disabled
We were relying on msix_unset_vector_notifiers() to release all the
vectors when we disable MSI-X, but this only happens when MSI-X is
still enabled on the device. Perform further cleanup by releasing
any remaining vectors listed as in-use after this call. This caused
a leak of IRQ routes on hotplug depending on how the guest OS prepared
the device for removal.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Cc: qemu-stable@nongnu.org
---
hw/misc/vfio.c | 12 ++++++++++++
1 file changed, 12 insertions(+)
(limited to 'hw/misc/vfio.c')
diff --git a/hw/misc/vfio.c b/hw/misc/vfio.c
index f367537737..9aecaa82bc 100644
--- a/hw/misc/vfio.c
+++ b/hw/misc/vfio.c
@@ -905,8 +905,20 @@ static void vfio_disable_msi_common(VFIODevice *vdev)
static void vfio_disable_msix(VFIODevice *vdev)
{
+ int i;
+
msix_unset_vector_notifiers(&vdev->pdev);
+ /*
+ * MSI-X will only release vectors if MSI-X is still enabled on the
+ * device, check through the rest and release it ourselves if necessary.
+ */
+ for (i = 0; i < vdev->nr_vectors; i++) {
+ if (vdev->msi_vectors[i].use) {
+ vfio_msix_vector_release(&vdev->pdev, i);
+ }
+ }
+
if (vdev->nr_vectors) {
vfio_disable_irqindex(vdev, VFIO_PCI_MSIX_IRQ_INDEX);
}
--
cgit v1.2.3