rpms
/
qemu-kvm
7
0
Fork 0
Code Issues Pull Requests Releases Activity
qemu-kvm/kvm-vfio-iommufd-Implement-...

562 lines
18 KiB
Diff
Raw Blame History

From f018d0b686406256c2b5e823e4227316ee1394e9 Mon Sep 17 00:00:00 2001
From: Yi Liu <yi.l.liu@intel.com>
Date: Tue, 21 Nov 2023 16:44:03 +0800
Subject: [PATCH 024/101] vfio/iommufd: Implement the iommufd backend
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
RH-Author: Eric Auger <eric.auger@redhat.com>
RH-MergeRequest: 211: IOMMUFD backend backport
RH-Jira: RHEL-19302 RHEL-21057
RH-Acked-by: Cédric Le Goater <clg@redhat.com>
RH-Acked-by: Sebastian Ott <sebott@redhat.com>
RH-Commit: [23/67] d11046654117a690542a1e2b48b9d1994f778b2d (eauger1/centos-qemu-kvm)
The iommufd backend is implemented based on the new /dev/iommu user API.
This backend obviously depends on CONFIG_IOMMUFD.
So far, the iommufd backend doesn't support dirty page sync yet.
Co-authored-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Tested-by: Eric Auger <eric.auger@redhat.com>
Tested-by: Nicolin Chen <nicolinc@nvidia.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
(cherry picked from commit 5ee3dc7af7859e7b8aa34c10c21778101c15e812)
Signed-off-by: Eric Auger <eric.auger@redhat.com>
---
hw/vfio/common.c | 6 +
hw/vfio/iommufd.c | 422 ++++++++++++++++++++++++++++++++++
hw/vfio/meson.build | 3 +
hw/vfio/trace-events | 10 +
include/hw/vfio/vfio-common.h | 11 +
5 files changed, 452 insertions(+)
create mode 100644 hw/vfio/iommufd.c
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 934f4f5446..6569732b7a 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -19,6 +19,7 @@
*/
#include "qemu/osdep.h"
+#include CONFIG_DEVICES /* CONFIG_IOMMUFD */
#include <sys/ioctl.h>
#ifdef CONFIG_KVM
#include <linux/kvm.h>
@@ -1503,6 +1504,11 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
{
const VFIOIOMMUOps *ops = &vfio_legacy_ops;
+#ifdef CONFIG_IOMMUFD
+ if (vbasedev->iommufd) {
+ ops = &vfio_iommufd_ops;
+ }
+#endif
return ops->attach_device(name, vbasedev, as, errp);
}
diff --git a/hw/vfio/iommufd.c b/hw/vfio/iommufd.c
new file mode 100644
index 0000000000..6d31aeac7b
--- /dev/null
+++ b/hw/vfio/iommufd.c
@@ -0,0 +1,422 @@
+/*
+ * iommufd container backend
+ *
+ * Copyright (C) 2023 Intel Corporation.
+ * Copyright Red Hat, Inc. 2023
+ *
+ * Authors: Yi Liu <yi.l.liu@intel.com>
+ * Eric Auger <eric.auger@redhat.com>
+ *
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ */
+
+#include "qemu/osdep.h"
+#include <sys/ioctl.h>
+#include <linux/vfio.h>
+#include <linux/iommufd.h>
+
+#include "hw/vfio/vfio-common.h"
+#include "qemu/error-report.h"
+#include "trace.h"
+#include "qapi/error.h"
+#include "sysemu/iommufd.h"
+#include "hw/qdev-core.h"
+#include "sysemu/reset.h"
+#include "qemu/cutils.h"
+#include "qemu/chardev_open.h"
+
+static int iommufd_cdev_map(VFIOContainerBase *bcontainer, hwaddr iova,
+ ram_addr_t size, void *vaddr, bool readonly)
+{
+ VFIOIOMMUFDContainer *container =
+ container_of(bcontainer, VFIOIOMMUFDContainer, bcontainer);
+
+ return iommufd_backend_map_dma(container->be,
+ container->ioas_id,
+ iova, size, vaddr, readonly);
+}
+
+static int iommufd_cdev_unmap(VFIOContainerBase *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ IOMMUTLBEntry *iotlb)
+{
+ VFIOIOMMUFDContainer *container =
+ container_of(bcontainer, VFIOIOMMUFDContainer, bcontainer);
+
+ /* TODO: Handle dma_unmap_bitmap with iotlb args (migration) */
+ return iommufd_backend_unmap_dma(container->be,
+ container->ioas_id, iova, size);
+}
+
+static int iommufd_cdev_kvm_device_add(VFIODevice *vbasedev, Error **errp)
+{
+ return vfio_kvm_device_add_fd(vbasedev->fd, errp);
+}
+
+static void iommufd_cdev_kvm_device_del(VFIODevice *vbasedev)
+{
+ Error *err = NULL;
+
+ if (vfio_kvm_device_del_fd(vbasedev->fd, &err)) {
+ error_report_err(err);
+ }
+}
+
+static int iommufd_cdev_connect_and_bind(VFIODevice *vbasedev, Error **errp)
+{
+ IOMMUFDBackend *iommufd = vbasedev->iommufd;
+ struct vfio_device_bind_iommufd bind = {
+ .argsz = sizeof(bind),
+ .flags = 0,
+ };
+ int ret;
+
+ ret = iommufd_backend_connect(iommufd, errp);
+ if (ret) {
+ return ret;
+ }
+
+ /*
+ * Add device to kvm-vfio to be prepared for the tracking
+ * in KVM. Especially for some emulated devices, it requires
+ * to have kvm information in the device open.
+ */
+ ret = iommufd_cdev_kvm_device_add(vbasedev, errp);
+ if (ret) {
+ goto err_kvm_device_add;
+ }
+
+ /* Bind device to iommufd */
+ bind.iommufd = iommufd->fd;
+ ret = ioctl(vbasedev->fd, VFIO_DEVICE_BIND_IOMMUFD, &bind);
+ if (ret) {
+ error_setg_errno(errp, errno, "error bind device fd=%d to iommufd=%d",
+ vbasedev->fd, bind.iommufd);
+ goto err_bind;
+ }
+
+ vbasedev->devid = bind.out_devid;
+ trace_iommufd_cdev_connect_and_bind(bind.iommufd, vbasedev->name,
+ vbasedev->fd, vbasedev->devid);
+ return ret;
+err_bind:
+ iommufd_cdev_kvm_device_del(vbasedev);
+err_kvm_device_add:
+ iommufd_backend_disconnect(iommufd);
+ return ret;
+}
+
+static void iommufd_cdev_unbind_and_disconnect(VFIODevice *vbasedev)
+{
+ /* Unbind is automatically conducted when device fd is closed */
+ iommufd_cdev_kvm_device_del(vbasedev);
+ iommufd_backend_disconnect(vbasedev->iommufd);
+}
+
+static int iommufd_cdev_getfd(const char *sysfs_path, Error **errp)
+{
+ long int ret = -ENOTTY;
+ char *path, *vfio_dev_path = NULL, *vfio_path = NULL;
+ DIR *dir = NULL;
+ struct dirent *dent;
+ gchar *contents;
+ struct stat st;
+ gsize length;
+ int major, minor;
+ dev_t vfio_devt;
+
+ path = g_strdup_printf("%s/vfio-dev", sysfs_path);
+ if (stat(path, &st) < 0) {
+ error_setg_errno(errp, errno, "no such host device");
+ goto out_free_path;
+ }
+
+ dir = opendir(path);
+ if (!dir) {
+ error_setg_errno(errp, errno, "couldn't open directory %s", path);
+ goto out_free_path;
+ }
+
+ while ((dent = readdir(dir))) {
+ if (!strncmp(dent->d_name, "vfio", 4)) {
+ vfio_dev_path = g_strdup_printf("%s/%s/dev", path, dent->d_name);
+ break;
+ }
+ }
+
+ if (!vfio_dev_path) {
+ error_setg(errp, "failed to find vfio-dev/vfioX/dev");
+ goto out_close_dir;
+ }
+
+ if (!g_file_get_contents(vfio_dev_path, &contents, &length, NULL)) {
+ error_setg(errp, "failed to load \"%s\"", vfio_dev_path);
+ goto out_free_dev_path;
+ }
+
+ if (sscanf(contents, "%d:%d", &major, &minor) != 2) {
+ error_setg(errp, "failed to get major:minor for \"%s\"", vfio_dev_path);
+ goto out_free_dev_path;
+ }
+ g_free(contents);
+ vfio_devt = makedev(major, minor);
+
+ vfio_path = g_strdup_printf("/dev/vfio/devices/%s", dent->d_name);
+ ret = open_cdev(vfio_path, vfio_devt);
+ if (ret < 0) {
+ error_setg(errp, "Failed to open %s", vfio_path);
+ }
+
+ trace_iommufd_cdev_getfd(vfio_path, ret);
+ g_free(vfio_path);
+
+out_free_dev_path:
+ g_free(vfio_dev_path);
+out_close_dir:
+ closedir(dir);
+out_free_path:
+ if (*errp) {
+ error_prepend(errp, VFIO_MSG_PREFIX, path);
+ }
+ g_free(path);
+
+ return ret;
+}
+
+static int iommufd_cdev_attach_ioas_hwpt(VFIODevice *vbasedev, uint32_t id,
+ Error **errp)
+{
+ int ret, iommufd = vbasedev->iommufd->fd;
+ struct vfio_device_attach_iommufd_pt attach_data = {
+ .argsz = sizeof(attach_data),
+ .flags = 0,
+ .pt_id = id,
+ };
+
+ /* Attach device to an IOAS or hwpt within iommufd */
+ ret = ioctl(vbasedev->fd, VFIO_DEVICE_ATTACH_IOMMUFD_PT, &attach_data);
+ if (ret) {
+ error_setg_errno(errp, errno,
+ "[iommufd=%d] error attach %s (%d) to id=%d",
+ iommufd, vbasedev->name, vbasedev->fd, id);
+ } else {
+ trace_iommufd_cdev_attach_ioas_hwpt(iommufd, vbasedev->name,
+ vbasedev->fd, id);
+ }
+ return ret;
+}
+
+static int iommufd_cdev_detach_ioas_hwpt(VFIODevice *vbasedev, Error **errp)
+{
+ int ret, iommufd = vbasedev->iommufd->fd;
+ struct vfio_device_detach_iommufd_pt detach_data = {
+ .argsz = sizeof(detach_data),
+ .flags = 0,
+ };
+
+ ret = ioctl(vbasedev->fd, VFIO_DEVICE_DETACH_IOMMUFD_PT, &detach_data);
+ if (ret) {
+ error_setg_errno(errp, errno, "detach %s failed", vbasedev->name);
+ } else {
+ trace_iommufd_cdev_detach_ioas_hwpt(iommufd, vbasedev->name);
+ }
+ return ret;
+}
+
+static int iommufd_cdev_attach_container(VFIODevice *vbasedev,
+ VFIOIOMMUFDContainer *container,
+ Error **errp)
+{
+ return iommufd_cdev_attach_ioas_hwpt(vbasedev, container->ioas_id, errp);
+}
+
+static void iommufd_cdev_detach_container(VFIODevice *vbasedev,
+ VFIOIOMMUFDContainer *container)
+{
+ Error *err = NULL;
+
+ if (iommufd_cdev_detach_ioas_hwpt(vbasedev, &err)) {
+ error_report_err(err);
+ }
+}
+
+static void iommufd_cdev_container_destroy(VFIOIOMMUFDContainer *container)
+{
+ VFIOContainerBase *bcontainer = &container->bcontainer;
+
+ if (!QLIST_EMPTY(&bcontainer->device_list)) {
+ return;
+ }
+ memory_listener_unregister(&bcontainer->listener);
+ vfio_container_destroy(bcontainer);
+ iommufd_backend_free_id(container->be, container->ioas_id);
+ g_free(container);
+}
+
+static int iommufd_cdev_ram_block_discard_disable(bool state)
+{
+ /*
+ * We support coordinated discarding of RAM via the RamDiscardManager.
+ */
+ return ram_block_uncoordinated_discard_disable(state);
+}
+
+static int iommufd_cdev_attach(const char *name, VFIODevice *vbasedev,
+ AddressSpace *as, Error **errp)
+{
+ VFIOContainerBase *bcontainer;
+ VFIOIOMMUFDContainer *container;
+ VFIOAddressSpace *space;
+ struct vfio_device_info dev_info = { .argsz = sizeof(dev_info) };
+ int ret, devfd;
+ uint32_t ioas_id;
+ Error *err = NULL;
+
+ devfd = iommufd_cdev_getfd(vbasedev->sysfsdev, errp);
+ if (devfd < 0) {
+ return devfd;
+ }
+ vbasedev->fd = devfd;
+
+ ret = iommufd_cdev_connect_and_bind(vbasedev, errp);
+ if (ret) {
+ goto err_connect_bind;
+ }
+
+ space = vfio_get_address_space(as);
+
+ /* try to attach to an existing container in this space */
+ QLIST_FOREACH(bcontainer, &space->containers, next) {
+ container = container_of(bcontainer, VFIOIOMMUFDContainer, bcontainer);
+ if (bcontainer->ops != &vfio_iommufd_ops ||
+ vbasedev->iommufd != container->be) {
+ continue;
+ }
+ if (iommufd_cdev_attach_container(vbasedev, container, &err)) {
+ const char *msg = error_get_pretty(err);
+
+ trace_iommufd_cdev_fail_attach_existing_container(msg);
+ error_free(err);
+ err = NULL;
+ } else {
+ ret = iommufd_cdev_ram_block_discard_disable(true);
+ if (ret) {
+ error_setg(errp,
+ "Cannot set discarding of RAM broken (%d)", ret);
+ goto err_discard_disable;
+ }
+ goto found_container;
+ }
+ }
+
+ /* Need to allocate a new dedicated container */
+ ret = iommufd_backend_alloc_ioas(vbasedev->iommufd, &ioas_id, errp);
+ if (ret < 0) {
+ goto err_alloc_ioas;
+ }
+
+ trace_iommufd_cdev_alloc_ioas(vbasedev->iommufd->fd, ioas_id);
+
+ container = g_malloc0(sizeof(*container));
+ container->be = vbasedev->iommufd;
+ container->ioas_id = ioas_id;
+
+ bcontainer = &container->bcontainer;
+ vfio_container_init(bcontainer, space, &vfio_iommufd_ops);
+ QLIST_INSERT_HEAD(&space->containers, bcontainer, next);
+
+ ret = iommufd_cdev_attach_container(vbasedev, container, errp);
+ if (ret) {
+ goto err_attach_container;
+ }
+
+ ret = iommufd_cdev_ram_block_discard_disable(true);
+ if (ret) {
+ goto err_discard_disable;
+ }
+
+ bcontainer->pgsizes = qemu_real_host_page_size();
+
+ bcontainer->listener = vfio_memory_listener;
+ memory_listener_register(&bcontainer->listener, bcontainer->space->as);
+
+ if (bcontainer->error) {
+ ret = -1;
+ error_propagate_prepend(errp, bcontainer->error,
+ "memory listener initialization failed: ");
+ goto err_listener_register;
+ }
+
+ bcontainer->initialized = true;
+
+found_container:
+ ret = ioctl(devfd, VFIO_DEVICE_GET_INFO, &dev_info);
+ if (ret) {
+ error_setg_errno(errp, errno, "error getting device info");
+ goto err_listener_register;
+ }
+
+ /*
+ * TODO: examine RAM_BLOCK_DISCARD stuff, should we do group level
+ * for discarding incompatibility check as well?
+ */
+ if (vbasedev->ram_block_discard_allowed) {
+ iommufd_cdev_ram_block_discard_disable(false);
+ }
+
+ vbasedev->group = 0;
+ vbasedev->num_irqs = dev_info.num_irqs;
+ vbasedev->num_regions = dev_info.num_regions;
+ vbasedev->flags = dev_info.flags;
+ vbasedev->reset_works = !!(dev_info.flags & VFIO_DEVICE_FLAGS_RESET);
+ vbasedev->bcontainer = bcontainer;
+ QLIST_INSERT_HEAD(&bcontainer->device_list, vbasedev, container_next);
+ QLIST_INSERT_HEAD(&vfio_device_list, vbasedev, global_next);
+
+ trace_iommufd_cdev_device_info(vbasedev->name, devfd, vbasedev->num_irqs,
+ vbasedev->num_regions, vbasedev->flags);
+ return 0;
+
+err_listener_register:
+ iommufd_cdev_ram_block_discard_disable(false);
+err_discard_disable:
+ iommufd_cdev_detach_container(vbasedev, container);
+err_attach_container:
+ iommufd_cdev_container_destroy(container);
+err_alloc_ioas:
+ vfio_put_address_space(space);
+ iommufd_cdev_unbind_and_disconnect(vbasedev);
+err_connect_bind:
+ close(vbasedev->fd);
+ return ret;
+}
+
+static void iommufd_cdev_detach(VFIODevice *vbasedev)
+{
+ VFIOContainerBase *bcontainer = vbasedev->bcontainer;
+ VFIOAddressSpace *space = bcontainer->space;
+ VFIOIOMMUFDContainer *container = container_of(bcontainer,
+ VFIOIOMMUFDContainer,
+ bcontainer);
+ QLIST_REMOVE(vbasedev, global_next);
+ QLIST_REMOVE(vbasedev, container_next);
+ vbasedev->bcontainer = NULL;
+
+ if (!vbasedev->ram_block_discard_allowed) {
+ iommufd_cdev_ram_block_discard_disable(false);
+ }
+
+ iommufd_cdev_detach_container(vbasedev, container);
+ iommufd_cdev_container_destroy(container);
+ vfio_put_address_space(space);
+
+ iommufd_cdev_unbind_and_disconnect(vbasedev);
+ close(vbasedev->fd);
+}
+
+const VFIOIOMMUOps vfio_iommufd_ops = {
+ .dma_map = iommufd_cdev_map,
+ .dma_unmap = iommufd_cdev_unmap,
+ .attach_device = iommufd_cdev_attach,
+ .detach_device = iommufd_cdev_detach,
+};
diff --git a/hw/vfio/meson.build b/hw/vfio/meson.build
index eb6ce6229d..e5d98b6adc 100644
--- a/hw/vfio/meson.build
+++ b/hw/vfio/meson.build
@@ -7,6 +7,9 @@ vfio_ss.add(files(
'spapr.c',
'migration.c',
))
+vfio_ss.add(when: 'CONFIG_IOMMUFD', if_true: files(
+ 'iommufd.c',
+))
vfio_ss.add(when: 'CONFIG_VFIO_PCI', if_true: files(
'display.c',
'pci-quirks.c',
diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
index 08a1f9dfa4..3340c93af0 100644
--- a/hw/vfio/trace-events
+++ b/hw/vfio/trace-events
@@ -164,3 +164,13 @@ vfio_state_pending_estimate(const char *name, uint64_t precopy, uint64_t postcop
vfio_state_pending_exact(const char *name, uint64_t precopy, uint64_t postcopy, uint64_t stopcopy_size, uint64_t precopy_init_size, uint64_t precopy_dirty_size) " (%s) precopy 0x%"PRIx64" postcopy 0x%"PRIx64" stopcopy size 0x%"PRIx64" precopy initial size 0x%"PRIx64" precopy dirty size 0x%"PRIx64
vfio_vmstate_change(const char *name, int running, const char *reason, const char *dev_state) " (%s) running %d reason %s device state %s"
vfio_vmstate_change_prepare(const char *name, int running, const char *reason, const char *dev_state) " (%s) running %d reason %s device state %s"
+
+#iommufd.c
+
+iommufd_cdev_connect_and_bind(int iommufd, const char *name, int devfd, int devid) " [iommufd=%d] Successfully bound device %s (fd=%d): output devid=%d"
+iommufd_cdev_getfd(const char *dev, int devfd) " %s (fd=%d)"
+iommufd_cdev_attach_ioas_hwpt(int iommufd, const char *name, int devfd, int id) " [iommufd=%d] Successfully attached device %s (%d) to id=%d"
+iommufd_cdev_detach_ioas_hwpt(int iommufd, const char *name) " [iommufd=%d] Successfully detached %s"
+iommufd_cdev_fail_attach_existing_container(const char *msg) " %s"
+iommufd_cdev_alloc_ioas(int iommufd, int ioas_id) " [iommufd=%d] new IOMMUFD container with ioasid=%d"
+iommufd_cdev_device_info(char *name, int devfd, int num_irqs, int num_regions, int flags) " %s (%d) num_irqs=%d num_regions=%d flags=%d"
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 24ecc0e7ee..3dac5c167e 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -89,6 +89,14 @@ typedef struct VFIOHostDMAWindow {
QLIST_ENTRY(VFIOHostDMAWindow) hostwin_next;
} VFIOHostDMAWindow;
+typedef struct IOMMUFDBackend IOMMUFDBackend;
+
+typedef struct VFIOIOMMUFDContainer {
+ VFIOContainerBase bcontainer;
+ IOMMUFDBackend *be;
+ uint32_t ioas_id;
+} VFIOIOMMUFDContainer;
+
typedef struct VFIODeviceOps VFIODeviceOps;
typedef struct VFIODevice {
@@ -116,6 +124,8 @@ typedef struct VFIODevice {
OnOffAuto pre_copy_dirty_page_tracking;
bool dirty_pages_supported;
bool dirty_tracking;
+ int devid;
+ IOMMUFDBackend *iommufd;
} VFIODevice;
struct VFIODeviceOps {
@@ -201,6 +211,7 @@ typedef QLIST_HEAD(VFIODeviceList, VFIODevice) VFIODeviceList;
extern VFIOGroupList vfio_group_list;
extern VFIODeviceList vfio_device_list;
extern const VFIOIOMMUOps vfio_legacy_ops;
+extern const VFIOIOMMUOps vfio_iommufd_ops;
extern const MemoryListener vfio_memory_listener;
extern int vfio_kvm_device_fd;
--
2.39.3
Reference in New Issue View Git Blame Copy Permalink