Merge tag 'pull-vfio-20231219' of https://github.com/legoater/qemu in…

…to staging vfio queue: * Introduce an IOMMU interface backend for VFIO devices * Convert IOMMU type1 and sPAPR IOMMU to respective backends * Introduce a new IOMMUFD backend for ARM, x86_64 and s390x platforms # -----BEGIN PGP SIGNATURE----- # # iQIzBAABCAAdFiEEoPZlSPBIlev+awtgUaNDx8/77KEFAmWB34AACgkQUaNDx8/7 # 7KGOMxAAqXegvAneHqIlu4c8TzTuUR2rkYgev9RdfIHRDuY2XtaX14xlWn/rpTXZ # qSgeta+iT8Cv4YV1POJeHWFDNs9E29p1w+R7nLcH1qTIIaZHtxwbVVQ3s7kAo1Vb # 1S1G0/zIznzGVI50a0lj1gO2yQJnu/79nXpnICgA5REW0CscMssnvboQODlwq17V # ZLNVM8CSAvKl6ppkmzRdfNXCfq6x7bf4MsvnuXsqda4TBbvyyTjAqdo/8sjKiGly # gSDQqhgy6cvEXIF0UUHPJzFApf0YdXUDlL8hzH90hvRVu4W/t24dPmT7UkVIX9Ek # TA7RVxv7iJlHtFDqfSTAJFr7nKO9Tm2V9N7xbD1OJUKrMoPZRT6+0R1hMKqsZ5z+ # nG6khqHGzuo/aI9n70YxYIPXt+vs/EHI4WUtslGLUTL0xv8lUzk6cxyIJupFRmDS # ix6GM9TXOV8RyOveL2knHVymlFnAR6dekkMB+6ljUTuzDwG0oco4vno8z9bi7Vct # j36bM56U3lhY+w+Ljoy0gPwgrw/FROnGG3mp1mwp1KRHqtEDnUQu8CaLbJOBsBGE # JJDP6AKAYMczdmYVkd4CvE0WaeSxtOUxW5H5NCPjtaFQt0qEcght2lA2K15g521q # jeojoJ/QK5949jnNCqm1Z66/YQVL79lPyL0E+mxEohwu+yTORk4= # =U0x5 # -----END PGP SIGNATURE----- # gpg: Signature made Tue 19 Dec 2023 13:22:56 EST # gpg: using RSA key A0F66548F04895EBFE6B0B6051A343C7CFFBECA1 # gpg: Good signature from "Cédric Le Goater <clg@redhat.com>" [unknown] # gpg: aka "Cédric Le Goater <clg@kaod.org>" [unknown] # gpg: WARNING: This key is not certified with a trusted signature! # gpg: There is no indication that the signature belongs to the owner. # Primary key fingerprint: A0F6 6548 F048 95EB FE6B 0B60 51A3 43C7 CFFB ECA1 * tag 'pull-vfio-20231219' of https://github.com/legoater/qemu: (47 commits) hw/ppc/Kconfig: Imply VFIO_PCI docs/devel: Add VFIO iommufd backend documentation vfio: Introduce a helper function to initialize VFIODevice vfio/ccw: Move VFIODevice initializations in vfio_ccw_instance_init vfio/ap: Move VFIODevice initializations in vfio_ap_instance_init vfio/platform: Move VFIODevice initializations in vfio_platform_instance_init vfio/pci: Move VFIODevice initializations in vfio_instance_init hw/i386: Activate IOMMUFD for q35 machines kconfig: Activate IOMMUFD for s390x machines hw/arm: Activate IOMMUFD for virt machines vfio: Make VFIOContainerBase poiner parameter const in VFIOIOMMUOps callbacks vfio/ccw: Make vfio cdev pre-openable by passing a file handle vfio/ccw: Allow the selection of a given iommu backend vfio/ap: Make vfio cdev pre-openable by passing a file handle vfio/ap: Allow the selection of a given iommu backend vfio/platform: Make vfio cdev pre-openable by passing a file handle vfio/platform: Allow the selection of a given iommu backend vfio/pci: Make vfio cdev pre-openable by passing a file handle vfio/pci: Allow the selection of a given iommu backend vfio/iommufd: Enable pci hot reset through iommufd cdev interface ... Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
qemu · Dec 20, 2023 · dd7d3e3 · dd7d3e3
2 parents bd00730 + 4278df9
commit dd7d3e3
Show file tree

Hide file tree

Showing 33 changed files with 2,229 additions and 514 deletions.
diff --git a/MAINTAINERS b/MAINTAINERS
@@ -2167,6 +2167,17 @@ F: hw/vfio/ap.c
 F: docs/system/s390x/vfio-ap.rst
 L: qemu-s390x@nongnu.org
 
+iommufd
+M: Yi Liu <yi.l.liu@intel.com>
+M: Eric Auger <eric.auger@redhat.com>
+M: Zhenzhong Duan <zhenzhong.duan@intel.com>
+S: Supported
+F: backends/iommufd.c
+F: include/sysemu/iommufd.h
+F: include/qemu/chardev_open.h
+F: util/chardev_open.c
+F: docs/devel/vfio-iommufd.rst
+
 vhost
 M: Michael S. Tsirkin <mst@redhat.com>
 S: Supported

diff --git a/backends/Kconfig b/backends/Kconfig
@@ -1 +1,5 @@
 source tpm/Kconfig
+
+config IOMMUFD
+    bool
+    depends on VFIO
diff --git a/backends/iommufd.c b/backends/iommufd.c
@@ -0,0 +1,245 @@
+/*
+ * 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 "sysemu/iommufd.h"
+#include "qapi/error.h"
+#include "qapi/qmp/qerror.h"
+#include "qemu/module.h"
+#include "qom/object_interfaces.h"
+#include "qemu/error-report.h"
+#include "monitor/monitor.h"
+#include "trace.h"
+#include <sys/ioctl.h>
+#include <linux/iommufd.h>
+
+static void iommufd_backend_init(Object *obj)
+{
+    IOMMUFDBackend *be = IOMMUFD_BACKEND(obj);
+
+    be->fd = -1;
+    be->users = 0;
+    be->owned = true;
+    qemu_mutex_init(&be->lock);
+}
+
+static void iommufd_backend_finalize(Object *obj)
+{
+    IOMMUFDBackend *be = IOMMUFD_BACKEND(obj);
+
+    if (be->owned) {
+        close(be->fd);
+        be->fd = -1;
+    }
+}
+
+static void iommufd_backend_set_fd(Object *obj, const char *str, Error **errp)
+{
+    IOMMUFDBackend *be = IOMMUFD_BACKEND(obj);
+    int fd = -1;
+
+    fd = monitor_fd_param(monitor_cur(), str, errp);
+    if (fd == -1) {
+        error_prepend(errp, "Could not parse remote object fd %s:", str);
+        return;
+    }
+    qemu_mutex_lock(&be->lock);
+    be->fd = fd;
+    be->owned = false;
+    qemu_mutex_unlock(&be->lock);
+    trace_iommu_backend_set_fd(be->fd);
+}
+
+static bool iommufd_backend_can_be_deleted(UserCreatable *uc)
+{
+    IOMMUFDBackend *be = IOMMUFD_BACKEND(uc);
+
+    return !be->users;
+}
+
+static void iommufd_backend_class_init(ObjectClass *oc, void *data)
+{
+    UserCreatableClass *ucc = USER_CREATABLE_CLASS(oc);
+
+    ucc->can_be_deleted = iommufd_backend_can_be_deleted;
+
+    object_class_property_add_str(oc, "fd", NULL, iommufd_backend_set_fd);
+}
+
+int iommufd_backend_connect(IOMMUFDBackend *be, Error **errp)
+{
+    int fd, ret = 0;
+
+    qemu_mutex_lock(&be->lock);
+    if (be->users == UINT32_MAX) {
+        error_setg(errp, "too many connections");
+        ret = -E2BIG;
+        goto out;
+    }
+    if (be->owned && !be->users) {
+        fd = qemu_open_old("/dev/iommu", O_RDWR);
+        if (fd < 0) {
+            error_setg_errno(errp, errno, "/dev/iommu opening failed");
+            ret = fd;
+            goto out;
+        }
+        be->fd = fd;
+    }
+    be->users++;
+out:
+    trace_iommufd_backend_connect(be->fd, be->owned,
+                                  be->users, ret);
+    qemu_mutex_unlock(&be->lock);
+    return ret;
+}
+
+void iommufd_backend_disconnect(IOMMUFDBackend *be)
+{
+    qemu_mutex_lock(&be->lock);
+    if (!be->users) {
+        goto out;
+    }
+    be->users--;
+    if (!be->users && be->owned) {
+        close(be->fd);
+        be->fd = -1;
+    }
+out:
+    trace_iommufd_backend_disconnect(be->fd, be->users);
+    qemu_mutex_unlock(&be->lock);
+}
+
+int iommufd_backend_alloc_ioas(IOMMUFDBackend *be, uint32_t *ioas_id,
+                               Error **errp)
+{
+    int ret, fd = be->fd;
+    struct iommu_ioas_alloc alloc_data  = {
+        .size = sizeof(alloc_data),
+        .flags = 0,
+    };
+
+    ret = ioctl(fd, IOMMU_IOAS_ALLOC, &alloc_data);
+    if (ret) {
+        error_setg_errno(errp, errno, "Failed to allocate ioas");
+        return ret;
+    }
+
+    *ioas_id = alloc_data.out_ioas_id;
+    trace_iommufd_backend_alloc_ioas(fd, *ioas_id, ret);
+
+    return ret;
+}
+
+void iommufd_backend_free_id(IOMMUFDBackend *be, uint32_t id)
+{
+    int ret, fd = be->fd;
+    struct iommu_destroy des = {
+        .size = sizeof(des),
+        .id = id,
+    };
+
+    ret = ioctl(fd, IOMMU_DESTROY, &des);
+    trace_iommufd_backend_free_id(fd, id, ret);
+    if (ret) {
+        error_report("Failed to free id: %u %m", id);
+    }
+}
+
+int iommufd_backend_map_dma(IOMMUFDBackend *be, uint32_t ioas_id, hwaddr iova,
+                            ram_addr_t size, void *vaddr, bool readonly)
+{
+    int ret, fd = be->fd;
+    struct iommu_ioas_map map = {
+        .size = sizeof(map),
+        .flags = IOMMU_IOAS_MAP_READABLE |
+                 IOMMU_IOAS_MAP_FIXED_IOVA,
+        .ioas_id = ioas_id,
+        .__reserved = 0,
+        .user_va = (uintptr_t)vaddr,
+        .iova = iova,
+        .length = size,
+    };
+
+    if (!readonly) {
+        map.flags |= IOMMU_IOAS_MAP_WRITEABLE;
+    }
+
+    ret = ioctl(fd, IOMMU_IOAS_MAP, &map);
+    trace_iommufd_backend_map_dma(fd, ioas_id, iova, size,
+                                  vaddr, readonly, ret);
+    if (ret) {
+        ret = -errno;
+
+        /* TODO: Not support mapping hardware PCI BAR region for now. */
+        if (errno == EFAULT) {
+            warn_report("IOMMU_IOAS_MAP failed: %m, PCI BAR?");
+        } else {
+            error_report("IOMMU_IOAS_MAP failed: %m");
+        }
+    }
+    return ret;
+}
+
+int iommufd_backend_unmap_dma(IOMMUFDBackend *be, uint32_t ioas_id,
+                              hwaddr iova, ram_addr_t size)
+{
+    int ret, fd = be->fd;
+    struct iommu_ioas_unmap unmap = {
+        .size = sizeof(unmap),
+        .ioas_id = ioas_id,
+        .iova = iova,
+        .length = size,
+    };
+
+    ret = ioctl(fd, IOMMU_IOAS_UNMAP, &unmap);
+    /*
+     * IOMMUFD takes mapping as some kind of object, unmapping
+     * nonexistent mapping is treated as deleting a nonexistent
+     * object and return ENOENT. This is different from legacy
+     * backend which allows it. vIOMMU may trigger a lot of
+     * redundant unmapping, to avoid flush the log, treat them
+     * as succeess for IOMMUFD just like legacy backend.
+     */
+    if (ret && errno == ENOENT) {
+        trace_iommufd_backend_unmap_dma_non_exist(fd, ioas_id, iova, size, ret);
+        ret = 0;
+    } else {
+        trace_iommufd_backend_unmap_dma(fd, ioas_id, iova, size, ret);
+    }
+
+    if (ret) {
+        ret = -errno;
+        error_report("IOMMU_IOAS_UNMAP failed: %m");
+    }
+    return ret;
+}
+
+static const TypeInfo iommufd_backend_info = {
+    .name = TYPE_IOMMUFD_BACKEND,
+    .parent = TYPE_OBJECT,
+    .instance_size = sizeof(IOMMUFDBackend),
+    .instance_init = iommufd_backend_init,
+    .instance_finalize = iommufd_backend_finalize,
+    .class_size = sizeof(IOMMUFDBackendClass),
+    .class_init = iommufd_backend_class_init,
+    .interfaces = (InterfaceInfo[]) {
+        { TYPE_USER_CREATABLE },
+        { }
+    }
+};
+
+static void register_types(void)
+{
+    type_register_static(&iommufd_backend_info);
+}
+
+type_init(register_types);
diff --git a/backends/meson.build b/backends/meson.build
@@ -20,6 +20,7 @@ if have_vhost_user
   system_ss.add(when: 'CONFIG_VIRTIO', if_true: files('vhost-user.c'))
 endif
 system_ss.add(when: 'CONFIG_VIRTIO_CRYPTO', if_true: files('cryptodev-vhost.c'))
+system_ss.add(when: 'CONFIG_IOMMUFD', if_true: files('iommufd.c'))
 if have_vhost_user_crypto
   system_ss.add(when: 'CONFIG_VIRTIO_CRYPTO', if_true: files('cryptodev-vhost-user.c'))
 endif

diff --git a/backends/trace-events b/backends/trace-events
@@ -5,3 +5,13 @@ dbus_vmstate_pre_save(void)
 dbus_vmstate_post_load(int version_id) "version_id: %d"
 dbus_vmstate_loading(const char *id) "id: %s"
 dbus_vmstate_saving(const char *id) "id: %s"
+
+# iommufd.c
+iommufd_backend_connect(int fd, bool owned, uint32_t users, int ret) "fd=%d owned=%d users=%d (%d)"
+iommufd_backend_disconnect(int fd, uint32_t users) "fd=%d users=%d"
+iommu_backend_set_fd(int fd) "pre-opened /dev/iommu fd=%d"
+iommufd_backend_map_dma(int iommufd, uint32_t ioas, uint64_t iova, uint64_t size, void *vaddr, bool readonly, int ret) " iommufd=%d ioas=%d iova=0x%"PRIx64" size=0x%"PRIx64" addr=%p readonly=%d (%d)"
+iommufd_backend_unmap_dma_non_exist(int iommufd, uint32_t ioas, uint64_t iova, uint64_t size, int ret) " Unmap nonexistent mapping: iommufd=%d ioas=%d iova=0x%"PRIx64" size=0x%"PRIx64" (%d)"
+iommufd_backend_unmap_dma(int iommufd, uint32_t ioas, uint64_t iova, uint64_t size, int ret) " iommufd=%d ioas=%d iova=0x%"PRIx64" size=0x%"PRIx64" (%d)"
+iommufd_backend_alloc_ioas(int iommufd, uint32_t ioas, int ret) " iommufd=%d ioas=%d (%d)"
+iommufd_backend_free_id(int iommufd, uint32_t id, int ret) " iommufd=%d id=%d (%d)"
diff --git a/docs/devel/index-internals.rst b/docs/devel/index-internals.rst
@@ -18,5 +18,6 @@ Details about QEMU's various subsystems including how to add features to them.
    s390-dasd-ipl
    tracing
    vfio-migration
+   vfio-iommufd
    writing-monitor-commands
    virtio-backends