summary refs log tree commit diff stats
path: root/hw/vfio/iommufd.c (follow)
Commit message (Collapse)AuthorAgeFilesLines
...
* vfio/container: Remove vfio_container_init()Cédric Le Goater2024-06-241-1/+0
| | | | | | | | | It's now empty. Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Eric Auger <eric.auger@redhat.com> Tested-by: Eric Auger <eric.auger@redhat.com> Signed-off-by: Cédric Le Goater <clg@redhat.com>
* vfio/container: Remove VFIOContainerBase::opsCédric Le Goater2024-06-241-2/+2
| | | | | | | | | Instead, use VFIO_IOMMU_GET_CLASS() to get the class pointer. Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Eric Auger <eric.auger@redhat.com> Tested-by: Eric Auger <eric.auger@redhat.com> Signed-off-by: Cédric Le Goater <clg@redhat.com>
* vfio/container: Switch to QOMCédric Le Goater2024-06-241-2/+2
| | | | | | | | | | Instead of allocating the container struct, create a QOM object of the appropriate type. Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Eric Auger <eric.auger@redhat.com> Tested-by: Eric Auger <eric.auger@redhat.com> Signed-off-by: Cédric Le Goater <clg@redhat.com>
* vfio/container: Change VFIOContainerBase to use QOMCédric Le Goater2024-06-241-0/+1
| | | | | | | | | | | | | | VFIOContainerBase was made a QOM interface because we believed that a QOM object would expose all the IOMMU backends to the QEMU machine and human interface. This only applies to user creatable devices or objects. Change the VFIOContainerBase nature from interface to object and make the necessary adjustments in the VFIO_IOMMU hierarchy. Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Eric Auger <eric.auger@redhat.com> Tested-by: Eric Auger <eric.auger@redhat.com> Signed-off-by: Cédric Le Goater <clg@redhat.com>
* vfio/container: Simplify vfio_container_init()Cédric Le Goater2024-06-241-1/+1
| | | | | | | | | | | | | | | Assign the base container VFIOAddressSpace 'space' pointer in vfio_address_space_insert(). The ultimate goal is to remove vfio_container_init() and instead rely on an .instance_init() handler to perfom the initialization of VFIOContainerBase. To be noted that vfio_connect_container() will assign the 'space' pointer later in the execution flow. This should not have any consequence. Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Tested-by: Eric Auger <eric.auger@redhat.com> Signed-off-by: Cédric Le Goater <clg@redhat.com>
* vfio/container: Introduce vfio_address_space_insert()Cédric Le Goater2024-06-241-1/+1
| | | | | | | | | | | It prepares ground for a future change initializing the 'space' pointer of VFIOContainerBase. The goal is to replace vfio_container_init() by an .instance_init() handler when VFIOContainerBase is QOMified. Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Eric Auger <eric.auger@redhat.com> Tested-by: Eric Auger <eric.auger@redhat.com> Signed-off-by: Cédric Le Goater <clg@redhat.com>
* HostIOMMUDevice: Introduce get_iova_ranges callbackEric Auger2024-06-241-0/+16
| | | | | | | | | | | | | Introduce a new HostIOMMUDevice callback that allows to retrieve the usable IOVA ranges. Implement this callback in the legacy VFIO and IOMMUFD VFIO host iommu devices. This relies on the VFIODevice agent's base container iova_ranges resource. Signed-off-by: Eric Auger <eric.auger@redhat.com> Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
* HostIOMMUDevice: Store the VFIO/VDPA agentEric Auger2024-06-241-0/+2
| | | | | | | | | Store the agent device (VFIO or VDPA) in the host IOMMU device. This will allow easy access to some of its resources. Signed-off-by: Eric Auger <eric.auger@redhat.com> Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
* vfio: Create host IOMMU device instanceZhenzhong Duan2024-06-241-0/+2
| | | | | | | | | | | | | | Create host IOMMU device instance in vfio_attach_device() and call .realize() to initialize it further. Introuduce attribute VFIOIOMMUClass::hiod_typename and initialize it based on VFIO backend type. It will facilitate HostIOMMUDevice creation in vfio_attach_device(). Suggested-by: Cédric Le Goater <clg@redhat.com> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Eric Auger <eric.auger@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
* vfio/iommufd: Implement HostIOMMUDeviceClass::realize() handlerZhenzhong Duan2024-06-241-0/+30
| | | | | | | | | | | | | | It calls iommufd_backend_get_device_info() to get host IOMMU related information and translate it into HostIOMMUDeviceCaps for query with .get_cap(). For aw_bits, use the same way as legacy backend by calling vfio_device_get_aw_bits() which is common for different vendor IOMMU. Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Eric Auger <eric.auger@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
* backends/iommufd: Introduce TYPE_HOST_IOMMU_DEVICE_IOMMUFD[_VFIO] devicesZhenzhong Duan2024-06-241-1/+4
| | | | | | | | | | | | | | | | | | | | | | | TYPE_HOST_IOMMU_DEVICE_IOMMUFD represents a host IOMMU device under iommufd backend. It is abstract, because it is going to be derived into VFIO or VDPA type'd device. It will have its own .get_cap() implementation. TYPE_HOST_IOMMU_DEVICE_IOMMUFD_VFIO is a sub-class of TYPE_HOST_IOMMU_DEVICE_IOMMUFD, represents a VFIO type'd host IOMMU device under iommufd backend. It will be created during VFIO device attaching and passed to vIOMMU. It will have its own .realize() implementation. Opportunistically, add missed header to include/sysemu/iommufd.h. Suggested-by: Cédric Le Goater <clg@redhat.com> Signed-off-by: Yi Liu <yi.l.liu@intel.com> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Eric Auger <eric.auger@redhat.com> Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
* backends/iommufd: Make iommufd_backend_*() return boolZhenzhong Duan2024-05-161-3/+2
| | | | | | | | | | | | | | | | | | | This is to follow the coding standand to return bool if 'Error **' is used to pass error. The changed functions include: iommufd_backend_connect iommufd_backend_alloc_ioas By this chance, simplify the functions a bit by avoiding duplicate recordings, e.g., log through either error interface or trace, not both. Suggested-by: Cédric Le Goater <clg@redhat.com> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Cédric Le Goater <clg@redhat.com> Signed-off-by: Cédric Le Goater <clg@redhat.com>
* vfio/cpr: Make vfio_cpr_register_container() return boolZhenzhong Duan2024-05-161-2/+1
| | | | | | | | | | This is to follow the coding standand to return bool if 'Error **' is used to pass error. Suggested-by: Cédric Le Goater <clg@redhat.com> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Cédric Le Goater <clg@redhat.com> Signed-off-by: Cédric Le Goater <clg@redhat.com>
* vfio/iommufd: Make iommufd_cdev_*() return boolZhenzhong Duan2024-05-161-49/+39
| | | | | | | | | | | | | | | | | | | | | | This is to follow the coding standand to return bool if 'Error **' is used to pass error. The changed functions include: iommufd_cdev_kvm_device_add iommufd_cdev_connect_and_bind iommufd_cdev_attach_ioas_hwpt iommufd_cdev_detach_ioas_hwpt iommufd_cdev_attach_container iommufd_cdev_get_info_iova_range After the change, all functions in hw/vfio/iommufd.c follows the standand. Suggested-by: Cédric Le Goater <clg@redhat.com> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Cédric Le Goater <clg@redhat.com> Signed-off-by: Cédric Le Goater <clg@redhat.com>
* vfio: Make VFIOIOMMUClass::attach_device() and its wrapper return boolZhenzhong Duan2024-05-161-6/+5
| | | | | | | | | | | | | Make VFIOIOMMUClass::attach_device() and its wrapper function vfio_attach_device() return bool. This is to follow the coding standand to return bool if 'Error **' is used to pass error. Suggested-by: Cédric Le Goater <clg@redhat.com> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Cédric Le Goater <clg@redhat.com> Signed-off-by: Cédric Le Goater <clg@redhat.com>
* vfio/pci: Use g_autofree in iommufd_cdev_get_info_iova_range()Zhenzhong Duan2024-05-161-3/+1
| | | | | | | | | | | Local pointer info is freed before return from iommufd_cdev_get_info_iova_range(). Use 'g_autofree' to avoid the g_free() calls. Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Cédric Le Goater <clg@redhat.com> Signed-off-by: Cédric Le Goater <clg@redhat.com>
* vfio/iommufd: Fix memory leakCédric Le Goater2024-03-191-11/+8
| | | | | | | | | | | | | | Coverity reported a memory leak on variable 'contents' in routine iommufd_cdev_getfd(). Use g_autofree variables to simplify the exit path and get rid of g_free() calls. Cc: Eric Auger <eric.auger@redhat.com> Cc: Yi Liu <yi.l.liu@intel.com> Fixes: CID 1540007 Fixes: 5ee3dc7af785 ("vfio/iommufd: Implement the iommufd backend") Suggested-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Signed-off-by: Cédric Le Goater <clg@redhat.com>
* Merge tag 'pull-request-2024-03-12' of https://gitlab.com/thuth/qemu into ↵Peter Maydell2024-03-121-0/+1
|\ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | staging * Add missing ERRP_GUARD() statements in functions that need it * Prefer fast cpu_env() over slower CPU QOM cast macro # -----BEGIN PGP SIGNATURE----- # # iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmXwPhYRHHRodXRoQHJl # ZGhhdC5jb20ACgkQLtnXdP5wLbWHvBAAgKx5LHFjz3xREVA+LkDTQ49mz0lK3s32 # SGvNlIHjiaDGVttVYhVC4sinBWUruG4Lyv/2QN72OJBzn6WUsEUQE3KPH1d7Y3/s # wS9X7mj70n4kugWJqeIJP5AXSRasHmWoQ4QJLVQRJd6+Eb9jqwep0x7bYkI1de6D # bL1Q7bIfkFeNQBXaiPWAm2i+hqmT4C1r8HEAGZIjAsMFrjy/hzBEjNV+pnh6ZSq9 # Vp8BsPWRfLU2XHm4WX0o8d89WUMAfUGbVkddEl/XjIHDrUD+Zbd1HAhLyfhsmrnE # jXIwSzm+ML1KX4MoF5ilGtg8Oo0gQDEBy9/xck6G0HCm9lIoLKlgTxK9glr2vdT8 # yxZmrM9Hder7F9hKKxmb127xgU6AmL7rYmVqsoQMNAq22D6Xr4UDpgFRXNk2/wO6 # zZZBkfZ4H4MpZXbd/KJpXvYH5mQA4IpkOy8LJdE+dbcHX7Szy9ksZdPA+Z10hqqf # zqS13qTs3abxymy2Q/tO3hPKSJCk1+vCGUkN60Wm+9VoLWGoU43qMc7gnY/pCS7m # 0rFKtvfwFHhokX1orK0lP/ppVzPv/5oFIeK8YDY9if+N+dU2LCwVZHIuf2/VJPRq # wmgH2vAn3JDoRKPxTGX9ly6AMxuZaeP92qBTOPap0gDhihYzIpaCq9ecEBoTakI7 # tdFhV0iRr08= # =NiP4 # -----END PGP SIGNATURE----- # gpg: Signature made Tue 12 Mar 2024 11:35:50 GMT # gpg: using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5 # gpg: issuer "thuth@redhat.com" # gpg: Good signature from "Thomas Huth <th.huth@gmx.de>" [full] # gpg: aka "Thomas Huth <thuth@redhat.com>" [full] # gpg: aka "Thomas Huth <huth@tuxfamily.org>" [full] # gpg: aka "Thomas Huth <th.huth@posteo.de>" [unknown] # Primary key fingerprint: 27B8 8847 EEE0 2501 18F3 EAB9 2ED9 D774 FE70 2DB5 * tag 'pull-request-2024-03-12' of https://gitlab.com/thuth/qemu: (55 commits) user: Prefer fast cpu_env() over slower CPU QOM cast macro target/xtensa: Prefer fast cpu_env() over slower CPU QOM cast macro target/tricore: Prefer fast cpu_env() over slower CPU QOM cast macro target/sparc: Prefer fast cpu_env() over slower CPU QOM cast macro target/sh4: Prefer fast cpu_env() over slower CPU QOM cast macro target/rx: Prefer fast cpu_env() over slower CPU QOM cast macro target/ppc: Prefer fast cpu_env() over slower CPU QOM cast macro target/openrisc: Prefer fast cpu_env() over slower CPU QOM cast macro target/nios2: Prefer fast cpu_env() over slower CPU QOM cast macro target/mips: Prefer fast cpu_env() over slower CPU QOM cast macro target/microblaze: Prefer fast cpu_env() over slower CPU QOM cast macro target/m68k: Prefer fast cpu_env() over slower CPU QOM cast macro target/loongarch: Prefer fast cpu_env() over slower CPU QOM cast macro target/i386/hvf: Use CPUState typedef target/hexagon: Prefer fast cpu_env() over slower CPU QOM cast macro target/cris: Prefer fast cpu_env() over slower CPU QOM cast macro target/avr: Prefer fast cpu_env() over slower CPU QOM cast macro target/alpha: Prefer fast cpu_env() over slower CPU QOM cast macro target: Replace CPU_GET_CLASS(cpu -> obj) in cpu_reset_hold() handler bulk: Call in place single use cpu_env() ... Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
| * hw/vfio/iommufd: Fix missing ERRP_GUARD() for error_prepend()Zhao Liu2024-03-121-0/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | As the comment in qapi/error, passing @errp to error_prepend() requires ERRP_GUARD(): * = Why, when and how to use ERRP_GUARD() = * * Without ERRP_GUARD(), use of the @errp parameter is restricted: ... * - It should not be passed to error_prepend(), error_vprepend() or * error_append_hint(), because that doesn't work with &error_fatal. * ERRP_GUARD() lifts these restrictions. * * To use ERRP_GUARD(), add it right at the beginning of the function. * @errp can then be used without worrying about the argument being * NULL or &error_fatal. ERRP_GUARD() could avoid the case when @errp is &error_fatal, the user can't see this additional information, because exit() happens in error_setg earlier than information is added [1]. The iommufd_cdev_getfd() passes @errp to error_prepend(). Its @errp is from vfio_attach_device(), and there are too many possible callers to check the impact of this defect; it may or may not be harmless. Thus it is necessary to protect @errp with ERRP_GUARD(). To avoid the issue like [1] said, add missing ERRP_GUARD() at the beginning of this function. [1]: Issue description in the commit message of commit ae7c80a7bd73 ("error: New macro ERRP_GUARD()"). Cc: Alex Williamson <alex.williamson@redhat.com> Cc: Cédric Le Goater <clg@redhat.com> Signed-off-by: Zhao Liu <zhao1.liu@intel.com> Reviewed-by: Cédric Le Goater <clg@redhat.com> Message-ID: <20240311033822.3142585-22-zhao1.liu@linux.intel.com> Signed-off-by: Thomas Huth <thuth@redhat.com>
* | vfio: register container for cprSteve Sistare2024-03-081-0/+6
|/ | | | | | | | Define entry points to perform per-container cpr-specific initialization and teardown. Signed-off-by: Steve Sistare <steven.sistare@oracle.com> Reviewed-by: Cédric Le Goater <clg@redhat.com>
* vfio/iommufd: Remove the use of stat() to check file existenceCédric Le Goater2024-01-051-6/+0
| | | | | | | | | | | | Using stat() before opening a file or a directory can lead to a time-of-check to time-of-use (TOCTOU) filesystem race, which is reported by coverity as a Security best practices violations. The sequence could be replaced by open and fdopendir but it doesn't add much in this case. Simply use opendir to avoid the race. Fixes: CID 1531551 Signed-off-by: Cédric Le Goater <clg@redhat.com> Reviewed-by: Zhenzhong Duan <Zhenzhong.duan@intel.com>
* vfio/iommufd: Introduce a VFIOIOMMU iommufd QOM interfaceCédric Le Goater2024-01-051-9/+26
| | | | | | | | | | | As previously done for the sPAPR and legacy IOMMU backends, convert the VFIOIOMMUOps struct to a QOM interface. The set of of operations for this backend can be referenced with a literal typename instead of a C struct. Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Tested-by: Eric Farman <farman@linux.ibm.com> Signed-off-by: Cédric Le Goater <clg@redhat.com>
* vfio: Make VFIOContainerBase poiner parameter const in VFIOIOMMUOps callbacksZhenzhong Duan2023-12-191-4/+4
| | | | | | | | | | | | | | | | | | Some of the callbacks in VFIOIOMMUOps pass VFIOContainerBase poiner, those callbacks only need read access to the sub object of VFIOContainerBase. So make VFIOContainerBase, VFIOContainer and VFIOIOMMUFDContainer as const in these callbacks. Local functions called by those callbacks also need same changes to avoid build error. Suggested-by: Cédric Le Goater <clg@redhat.com> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Cédric Le Goater <clg@redhat.com> Reviewed-by: Eric Auger <eric.auger@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>
* vfio/pci: Make vfio cdev pre-openable by passing a file handleZhenzhong Duan2023-12-191-4/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | | This gives management tools like libvirt a chance to open the vfio cdev with privilege and pass FD to qemu. This way qemu never needs to have privilege to open a VFIO or iommu cdev node. Together with the earlier support of pre-opening /dev/iommu device, now we have full support of passing a vfio device to unprivileged qemu by management tool. This mode is no more considered for the legacy backend. So let's remove the "TODO" comment. Add helper functions vfio_device_set_fd() and vfio_device_get_name() to set fd and get device name, they will also be used by other vfio devices. There is no easy way to check if a device is mdev with FD passing, so fail the x-balloon-allowed check unconditionally in this case. There is also no easy way to get BDF as name with FD passing, so we fake a name by VFIO_FD[fd]. 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>
* vfio/iommufd: Enable pci hot reset through iommufd cdev interfaceZhenzhong Duan2023-12-191-0/+150
| | | | | | | | | | | | Implement the newly introduced pci_hot_reset callback named iommufd_cdev_pci_hot_reset to do iommufd specific check and reset operation. Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Eric Auger <eric.auger@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>
* vfio/iommufd: Add support for iova_ranges and pgsizesZhenzhong Duan2023-12-191-1/+55
| | | | | | | | | | | | | | | | | | | | Some vIOMMU such as virtio-iommu use IOVA ranges from host side to setup reserved ranges for passthrough device, so that guest will not use an IOVA range beyond host support. Use an uAPI of IOMMUFD to get IOVA ranges of host side and pass to vIOMMU just like the legacy backend, if this fails, fallback to 64bit IOVA range. Also use out_iova_alignment returned from uAPI as pgsizes instead of qemu_real_host_page_size() as a fallback. Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> Reviewed-by: Cédric Le Goater <clg@redhat.com> Reviewed-by: Eric Auger <eric.auger@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>
* vfio/iommufd: Implement the iommufd backendYi Liu2023-12-191-0/+422
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>