Cédric Le Goater <c...@redhat.com> writes:
> On 11/2/23 08:12, Zhenzhong Duan wrote:
>> From: Eric Auger <eric.au...@redhat.com>
>> Introduce an iommufd object which allows the interaction
>> with the host /dev/iommu device.
>> The /dev/iommu can have been already pre-opened outside of qemu,
>> in which case the fd can be passed directly along with the
>> iommufd object:
>> This allows the iommufd object to be shared accross several
>> subsystems (VFIO, VDPA, ...). For example, libvirt would open
>> the /dev/iommu once.
>> If no fd is passed along with the iommufd object, the /dev/iommu
>> is opened by the qemu code.
>> The CONFIG_IOMMUFD option must be set to compile this new object.
>> Suggested-by: Alex Williamson <alex.william...@redhat.com>
>> Signed-off-by: Eric Auger <eric.au...@redhat.com>
>> Signed-off-by: Yi Liu <yi.l....@intel.com>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.d...@intel.com>
>> ---
>> v4: add CONFIG_IOMMUFD check, document default case
>> MAINTAINERS | 7 ++
>> qapi/qom.json | 22 ++++
>> include/sysemu/iommufd.h | 46 +++++++
>> backends/iommufd-stub.c | 59 +++++++++
>> backends/iommufd.c | 257 +++++++++++++++++++++++++++++++++++++++
>> backends/Kconfig | 4 +
>> backends/meson.build | 5 +
>> backends/trace-events | 12 ++
>> qemu-options.hx | 13 ++
>> 9 files changed, 425 insertions(+)
>> create mode 100644 include/sysemu/iommufd.h
>> create mode 100644 backends/iommufd-stub.c
>> create mode 100644 backends/iommufd.c
>> diff --git a/MAINTAINERS b/MAINTAINERS
>> index cd8d6b140f..6f35159255 100644
>> --- a/MAINTAINERS
>> +++ b/MAINTAINERS
>> @@ -2135,6 +2135,13 @@ F: hw/vfio/ap.c
>> F: docs/system/s390x/vfio-ap.rst
>> L: qemu-s3...@nongnu.org
>> +iommufd
>> +M: Yi Liu <yi.l....@intel.com>
>> +M: Eric Auger <eric.au...@redhat.com>
>> +S: Supported
>> +F: backends/iommufd.c
>> +F: include/sysemu/iommufd.h
>> +
>> vhost
>> M: Michael S. Tsirkin <m...@redhat.com>
>> S: Supported
>> diff --git a/qapi/qom.json b/qapi/qom.json
>> index c53ef978ff..27300add48 100644
>> --- a/qapi/qom.json
>> +++ b/qapi/qom.json
>> @@ -794,6 +794,24 @@
>> { 'struct': 'VfioUserServerProperties',
>> 'data': { 'socket': 'SocketAddress', 'device': 'str' } }
>> +##
>> +# @IOMMUFDProperties:
>> +#
>> +# Properties for iommufd objects.
>> +#
>> +# @fd: file descriptor name previously passed via 'getfd' command,
>> +# which represents a pre-opened /dev/iommu. This allows the
>> +# iommufd object to be shared accross several subsystems
>> +# (VFIO, VDPA, ...), and the file descriptor to be shared
>> +# with other process, e.g. DPDK. (default: QEMU opens
>> +# /dev/iommu by itself)
>> +#
>> +# Since: 8.2
>> +##
>> +{ 'struct': 'IOMMUFDProperties',
>> + 'data': { '*fd': 'str' },
>> + 'if': 'CONFIG_IOMMUFD' }
>
>
> Activating or not IOMMUFD on a platform is a configuration choice
> and it is not a dependency on an external resource. I would make
> things simpler and drop all the #ifdef in the documentation files.
What exactly are you proposing?
The use of 'if': 'CONFIG_IOMMUFD' in the QAPI schema enables
introspection with query-qmp-schema: when ObjectType @iommufd exists,
QEMU supports creating the object. Or am I confused?
> There might be a way to remove the documentation also. Not a big
> issue for now.
>
>
>> +
>> ##
>> # @RngProperties:
>> #
>> @@ -934,6 +952,8 @@
>> 'input-barrier',
>> { 'name': 'input-linux',
>> 'if': 'CONFIG_LINUX' },
>> + { 'name': 'iommufd',
>> + 'if': 'CONFIG_IOMMUFD' },
>> 'iothread',
>> 'main-loop',
>> { 'name': 'memory-backend-epc',
>> @@ -1003,6 +1023,8 @@
>> 'input-barrier': 'InputBarrierProperties',
>> 'input-linux': { 'type': 'InputLinuxProperties',
>> 'if': 'CONFIG_LINUX' },
>> + 'iommufd': { 'type': 'IOMMUFDProperties',
>> + 'if': 'CONFIG_IOMMUFD' },
>> 'iothread': 'IothreadProperties',
>> 'main-loop': 'MainLoopProperties',
>> 'memory-backend-epc': { 'type': 'MemoryBackendEpcProperties',
[...]
