* [PATCH v2 00/27] vfio: Adopt iommufd
@ 2023-10-16 8:31 Zhenzhong Duan
2023-10-16 8:31 ` [PATCH v2 01/27] vfio: Rename VFIOContainer into VFIOLegacyContainer Zhenzhong Duan
` (26 more replies)
0 siblings, 27 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:31 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
Hi,
Thanks all for giving guides and comments on previous series, here is
the pure iommufd support part.
PATCH 1-15: Abstract out base container
PATCH 16: Add --enable/--disable-iommufd config support
PATCH 17: Introduce iommufd object
PATCH 18-21: add IOMMUFD container and cdev support
PATCH 22-27: fd passing for IOMMUFD object and cdev
We have done wide test with different combinations, e.g:
- PCI device were tested
- FD passing and hot reset with some trick.
- device hotplug test with legacy and iommufd backends
- with or without vIOMMU for legacy and iommufd backends
- devices linked to different iommufd backends
- VFIO migration with a E800 net card(no dirty sync support) passthrough
- platform, ccw and ap were only compile-tested due to environment limit
Given some iommufd kernel limitations, the iommufd backend is
not yet fully on par with the legacy backend w.r.t. features like:
- p2p mappings (you will see related error traces)
- dirty page sync
- and etc.
qemu code: https://github.com/yiliu1765/qemu/commits/zhenzhong/iommufd_cdev_v2
--------------------------------------------------------------------------
Below are some background and graph about the design:
With the introduction of iommufd, the Linux kernel provides a generic
interface for userspace drivers to propagate their DMA mappings to kernel
for assigned devices. This series does the porting of the VFIO devices
onto the /dev/iommu uapi and let it coexist with the legacy implementation.
At QEMU level, interactions with the /dev/iommu are abstracted by a new
iommufd object (compiled in with the CONFIG_IOMMUFD option).
Any QEMU device (e.g. vfio device) wishing to use /dev/iommu must be
linked with an iommufd object. In this series, the vfio-pci device is
granted with such capability (other VFIO devices are not yet ready):
It gets a new optional parameter named iommufd which allows to pass
an iommufd object:
-object iommufd,id=iommufd0
-device vfio-pci,host=0000:02:00.0,iommufd=iommufd0
Note the /dev/iommu and vfio cdev can be externally opened by a
management layer. In such a case the fd is passed:
-object iommufd,id=iommufd0,fd=22
-device vfio-pci,iommufd=iommufd0,fd=23
If the fd parameter is not passed, the fd is opened by QEMU.
See https://www.mail-archive.com/qemu-devel@nongnu.org/msg937155.html
for detailed discuss on this requirement.
If no iommufd option is passed to the vfio-pci device, iommufd is not
used and the end-user gets the behavior based on the legacy vfio iommu
interfaces:
-device vfio-pci,host=0000:02:00.0
While the legacy kernel interface is group-centric, the new iommufd
interface is device-centric, relying on device fd and iommufd.
To support both interfaces in the QEMU VFIO device we reworked the vfio
container abstraction so that the generic VFIO code can use either
backend.
The VFIOContainer object becomes a base object derived into
a) the legacy VFIO container and
b) the new iommufd based container.
The base object implements generic code such as code related to
memory_listener and address space management whereas the derived
objects implement callbacks specific to either BE, legacy and
iommufd. Indeed each backend has its own way to setup secure context
and dma management interface. The below diagram shows how it looks
like with both BEs.
VFIO AddressSpace/Memory
+-------+ +----------+ +-----+ +-----+
| pci | | platform | | ap | | ccw |
+---+---+ +----+-----+ +--+--+ +--+--+ +----------------------+
| | | | | AddressSpace |
| | | | +------------+---------+
+---V-----------V-----------V--------V----+ /
| VFIOAddressSpace | <------------+
| | | MemoryListener
| VFIOContainer list |
+-------+----------------------------+----+
| |
| |
+-------V------+ +--------V----------+
| iommufd | | vfio legacy |
| container | | container |
+-------+------+ +--------+----------+
| |
| /dev/iommu | /dev/vfio/vfio
| /dev/vfio/devices/vfioX | /dev/vfio/$group_id
Userspace | |
============+============================+===========================
Kernel | device fd |
+---------------+ | group/container fd
| (BIND_IOMMUFD | | (SET_CONTAINER/SET_IOMMU)
| ATTACH_IOAS) | | device fd
| | |
| +-------V------------V-----------------+
iommufd | | vfio |
(map/unmap | +---------+--------------------+-------+
ioas_copy) | | | map/unmap
| | |
+------V------+ +-----V------+ +------V--------+
| iommfd core | | device | | vfio iommu |
+-------------+ +------------+ +---------------+
[Secure Context setup]
- iommufd BE: uses device fd and iommufd to setup secure context
(bind_iommufd, attach_ioas)
- vfio legacy BE: uses group fd and container fd to setup secure context
(set_container, set_iommu)
[Device access]
- iommufd BE: device fd is opened through /dev/vfio/devices/vfioX
- vfio legacy BE: device fd is retrieved from group fd ioctl
[DMA Mapping flow]
1. VFIOAddressSpace receives MemoryRegion add/del via MemoryListener
2. VFIO populates DMA map/unmap via the container BEs
*) iommufd BE: uses iommufd
*) vfio legacy BE: uses container fd
Changelog:
v2:
- rebase to vfio-next
- split PATCH "vfio: Add base container" in v1 to PATCH1-15 per Cédric
- add fd passing to platform/ap/ccw vfio device
- add (uintptr_t) cast in iommufd_backend_map_dma() per Cédric
- rename char_dev.h to chardev_open.h for same naming scheme per Daniel
- add full copyright per Daniel and Jason
Note changelog below are from full IOMMUFD series:
v1:
- Alloc hwpt instead of using auto hwpt
- elaborate iommufd code per Nicolin
- consolidate two patches and drop as.c
- typo error fix and function rename
rfcv4:
- rebase on top of v8.0.3
- Add one patch from Yi which is about vfio device add in kvm
- Remove IOAS_COPY optimization and focus on functions in this patchset
- Fix wrong name issue reported and fix suggested by Matthew
- Fix compilation issue reported and fix sugggsted by Nicolin
- Use query_dirty_bitmap callback to replace get_dirty_bitmap for better
granularity
- Add dev_iter_next() callback to avoid adding so many callback
at container scope, add VFIODevice.hwpt to support that
- Restore all functions back to common from container whenever possible,
mainly migration and reset related functions
- Add --enable/disable-iommufd config option, enabled by default in linux
- Remove VFIODevice.hwpt_next as it's redundant with VFIODevice.next
- Adapt new VFIO_DEVICE_PCI_HOT_RESET uAPI for IOMMUFD backed device
- vfio_kvm_device_add/del_group call vfio_kvm_device_add/del_fd to remove
redundant code
- Add FD passing support for vfio device backed by IOMMUFD
- Fix hot unplug resource leak issue in vfio_legacy_detach_device()
- Fix FD leak in vfio_get_devicefd()
rfcv3:
- rebase on top of v7.2.0
- Fix the compilation with CONFIG_IOMMUFD unset by using true classes for
VFIO backends
- Fix use after free in error path, reported by Alister
- Split common.c in several steps to ease the review
rfcv2:
- remove the first three patches of rfcv1
- add open cdev helper suggested by Jason
- remove the QOMification of the VFIOContainer and simply use standard ops
(David)
- add "-object iommufd" suggested by Alex
Thanks
Zhenzhong
Eric Auger (16):
vfio: Rename VFIOContainer into VFIOLegacyContainer
vfio: Introduce base object for VFIOContainer and targetted interface
VFIO/container: Introduce dummy VFIOContainerClass implementation
vfio/container: Switch to dma_map|unmap API
vfio/common: Move giommu_list in base container
vfio/container: Move space field to base container
vfio/container: switch to IOMMU BE add/del_section_window
vfio/container: Move hostwin_list in base container
vfio/container: Switch to IOMMU BE
set_dirty_page_tracking/query_dirty_bitmap API
vfio/container: Convert functions to base container
vfio/container: Move vrdl_list, pgsizes and dma_max_mappings to base
container
vfio/container: Move listener to base container
vfio/container: Move dirty_pgsizes and max_dirty_bitmap_size to base
container
vfio/container: Implement attach/detach_device
backends/iommufd: Introduce the iommufd object
vfio/pci: Allow the selection of a given iommu backend
Yi Liu (2):
util/char_dev: Add open_cdev()
vfio/iommufd: Implement the iommufd backend
Zhenzhong Duan (9):
vfio/container: Move per container device list in base container
Add iommufd configure option
vfio/container: Bypass EEH if iommufd backend
vfio/pci: Adapt vfio pci hot reset support with iommufd BE
vfio/pci: Make vfio cdev pre-openable by passing a file handle
vfio: Allow the selection of a given iommu backend for platform ap and
ccw
vfio/platform: Make vfio cdev pre-openable by passing a file handle
vfio/ap: Make vfio cdev pre-openable by passing a file handle
vfio/ccw: Make vfio cdev pre-openable by passing a file handle
MAINTAINERS | 13 +
meson.build | 6 +
qapi/qom.json | 18 +-
include/hw/vfio/vfio-common.h | 110 ++----
include/hw/vfio/vfio-container-base.h | 153 ++++++++
include/hw/vfio/vfio-platform.h | 1 +
include/qemu/chardev_open.h | 16 +
include/sysemu/iommufd.h | 46 +++
backends/iommufd-stub.c | 59 +++
backends/iommufd.c | 268 +++++++++++++
hw/vfio/ap.c | 37 +-
hw/vfio/ccw.c | 39 +-
hw/vfio/common.c | 274 +++++++------
hw/vfio/container-base.c | 150 +++++++
hw/vfio/container.c | 243 +++++++-----
hw/vfio/helpers.c | 33 ++
hw/vfio/iommufd.c | 539 ++++++++++++++++++++++++++
hw/vfio/pci.c | 257 ++++++++++--
hw/vfio/platform.c | 45 ++-
hw/vfio/spapr.c | 23 +-
util/chardev_open.c | 81 ++++
backends/Kconfig | 4 +
backends/meson.build | 5 +
backends/trace-events | 12 +
hw/vfio/meson.build | 4 +
hw/vfio/trace-events | 17 +-
meson_options.txt | 2 +
qemu-options.hx | 13 +
scripts/meson-buildoptions.sh | 3 +
util/meson.build | 1 +
30 files changed, 2122 insertions(+), 350 deletions(-)
create mode 100644 include/hw/vfio/vfio-container-base.h
create mode 100644 include/qemu/chardev_open.h
create mode 100644 include/sysemu/iommufd.h
create mode 100644 backends/iommufd-stub.c
create mode 100644 backends/iommufd.c
create mode 100644 hw/vfio/container-base.c
create mode 100644 hw/vfio/iommufd.c
create mode 100644 util/chardev_open.c
--
2.34.1
^ permalink raw reply [flat|nested] 43+ messages in thread
* [PATCH v2 01/27] vfio: Rename VFIOContainer into VFIOLegacyContainer
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
@ 2023-10-16 8:31 ` Zhenzhong Duan
2023-10-17 15:50 ` Cédric Le Goater
2023-10-16 8:31 ` [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface Zhenzhong Duan
` (25 subsequent siblings)
26 siblings, 1 reply; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:31 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
From: Eric Auger <eric.auger@redhat.com>
In the prospect to introduce a base object for the VFIOContainer
and derive into the existing legacy container and the iommufd
based container, let's rename the existing one into
VFIOLegacyContainer. This is just an incremental step to ease
the migration. Soon there won't be any reference to the legacy
container in the common.c code. Only the container.c should
handle the VFIOLegacyContainer object.
No functional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 46 ++++++++++++-------------
hw/vfio/common.c | 63 ++++++++++++++++++++---------------
hw/vfio/container.c | 45 +++++++++++++------------
hw/vfio/spapr.c | 12 +++----
4 files changed, 89 insertions(+), 77 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 7780b9073a..34648e518e 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -74,13 +74,13 @@ typedef struct VFIOMigration {
typedef struct VFIOAddressSpace {
AddressSpace *as;
- QLIST_HEAD(, VFIOContainer) containers;
+ QLIST_HEAD(, VFIOLegacyContainer) containers;
QLIST_ENTRY(VFIOAddressSpace) list;
} VFIOAddressSpace;
struct VFIOGroup;
-typedef struct VFIOContainer {
+typedef struct VFIOLegacyContainer {
VFIOAddressSpace *space;
int fd; /* /dev/vfio/vfio, empowered by the attached groups */
MemoryListener listener;
@@ -97,12 +97,12 @@ typedef struct VFIOContainer {
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_HEAD(, VFIOGroup) group_list;
QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
- QLIST_ENTRY(VFIOContainer) next;
+ QLIST_ENTRY(VFIOLegacyContainer) next;
QLIST_HEAD(, VFIODevice) device_list;
-} VFIOContainer;
+} VFIOLegacyContainer;
typedef struct VFIOGuestIOMMU {
- VFIOContainer *container;
+ VFIOLegacyContainer *container;
IOMMUMemoryRegion *iommu_mr;
hwaddr iommu_offset;
IOMMUNotifier n;
@@ -110,7 +110,7 @@ typedef struct VFIOGuestIOMMU {
} VFIOGuestIOMMU;
typedef struct VFIORamDiscardListener {
- VFIOContainer *container;
+ VFIOLegacyContainer *container;
MemoryRegion *mr;
hwaddr offset_within_address_space;
hwaddr size;
@@ -133,7 +133,7 @@ typedef struct VFIODevice {
QLIST_ENTRY(VFIODevice) container_next;
QLIST_ENTRY(VFIODevice) global_next;
struct VFIOGroup *group;
- VFIOContainer *container;
+ VFIOLegacyContainer *container;
char *sysfsdev;
char *name;
DeviceState *dev;
@@ -167,7 +167,7 @@ struct VFIODeviceOps {
typedef struct VFIOGroup {
int fd;
int groupid;
- VFIOContainer *container;
+ VFIOLegacyContainer *container;
QLIST_HEAD(, VFIODevice) device_list;
QLIST_ENTRY(VFIOGroup) next;
QLIST_ENTRY(VFIOGroup) container_next;
@@ -206,28 +206,28 @@ typedef struct {
hwaddr pages;
} VFIOBitmap;
-void vfio_host_win_add(VFIOContainer *container,
+void vfio_host_win_add(VFIOLegacyContainer *container,
hwaddr min_iova, hwaddr max_iova,
uint64_t iova_pgsizes);
-int vfio_host_win_del(VFIOContainer *container, hwaddr min_iova,
+int vfio_host_win_del(VFIOLegacyContainer *container, hwaddr min_iova,
hwaddr max_iova);
VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
void vfio_put_address_space(VFIOAddressSpace *space);
-bool vfio_devices_all_running_and_saving(VFIOContainer *container);
+bool vfio_devices_all_running_and_saving(VFIOLegacyContainer *container);
/* container->fd */
-int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
+int vfio_dma_unmap(VFIOLegacyContainer *container, hwaddr iova,
ram_addr_t size, IOMMUTLBEntry *iotlb);
-int vfio_dma_map(VFIOContainer *container, hwaddr iova,
+int vfio_dma_map(VFIOLegacyContainer *container, hwaddr iova,
ram_addr_t size, void *vaddr, bool readonly);
-int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start);
-int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
+int vfio_set_dirty_page_tracking(VFIOLegacyContainer *container, bool start);
+int vfio_query_dirty_bitmap(VFIOLegacyContainer *container, VFIOBitmap *vbmap,
hwaddr iova, hwaddr size);
-int vfio_container_add_section_window(VFIOContainer *container,
+int vfio_container_add_section_window(VFIOLegacyContainer *container,
MemoryRegionSection *section,
Error **errp);
-void vfio_container_del_section_window(VFIOContainer *container,
+void vfio_container_del_section_window(VFIOLegacyContainer *container,
MemoryRegionSection *section);
void vfio_disable_irqindex(VFIODevice *vbasedev, int index);
@@ -290,21 +290,21 @@ vfio_get_cap(void *ptr, uint32_t cap_offset, uint16_t id);
#endif
extern const MemoryListener vfio_prereg_listener;
-int vfio_spapr_create_window(VFIOContainer *container,
+int vfio_spapr_create_window(VFIOLegacyContainer *container,
MemoryRegionSection *section,
hwaddr *pgsize);
-int vfio_spapr_remove_window(VFIOContainer *container,
+int vfio_spapr_remove_window(VFIOLegacyContainer *container,
hwaddr offset_within_address_space);
bool vfio_migration_realize(VFIODevice *vbasedev, Error **errp);
void vfio_migration_exit(VFIODevice *vbasedev);
int vfio_bitmap_alloc(VFIOBitmap *vbmap, hwaddr size);
-bool vfio_devices_all_running_and_mig_active(VFIOContainer *container);
-bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container);
-int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
+bool vfio_devices_all_running_and_mig_active(VFIOLegacyContainer *container);
+bool vfio_devices_all_device_dirty_tracking(VFIOLegacyContainer *container);
+int vfio_devices_query_dirty_bitmap(VFIOLegacyContainer *container,
VFIOBitmap *vbmap, hwaddr iova,
hwaddr size);
-int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
+int vfio_get_dirty_bitmap(VFIOLegacyContainer *container, uint64_t iova,
uint64_t size, ram_addr_t ram_addr);
#endif /* HW_VFIO_VFIO_COMMON_H */
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 5ff5acf1d8..b51ef3a15a 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -184,7 +184,7 @@ bool vfio_device_state_is_precopy(VFIODevice *vbasedev)
migration->device_state == VFIO_DEVICE_STATE_PRE_COPY_P2P;
}
-static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
+static bool vfio_devices_all_dirty_tracking(VFIOLegacyContainer *container)
{
VFIODevice *vbasedev;
MigrationState *ms = migrate_get_current();
@@ -210,7 +210,7 @@ static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
return true;
}
-bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container)
+bool vfio_devices_all_device_dirty_tracking(VFIOLegacyContainer *container)
{
VFIODevice *vbasedev;
@@ -227,7 +227,7 @@ bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container)
* Check if all VFIO devices are running and migration is active, which is
* essentially equivalent to the migration being in pre-copy phase.
*/
-bool vfio_devices_all_running_and_mig_active(VFIOContainer *container)
+bool vfio_devices_all_running_and_mig_active(VFIOLegacyContainer *container)
{
VFIODevice *vbasedev;
@@ -252,7 +252,7 @@ bool vfio_devices_all_running_and_mig_active(VFIOContainer *container)
return true;
}
-void vfio_host_win_add(VFIOContainer *container, hwaddr min_iova,
+void vfio_host_win_add(VFIOLegacyContainer *container, hwaddr min_iova,
hwaddr max_iova, uint64_t iova_pgsizes)
{
VFIOHostDMAWindow *hostwin;
@@ -274,7 +274,7 @@ void vfio_host_win_add(VFIOContainer *container, hwaddr min_iova,
QLIST_INSERT_HEAD(&container->hostwin_list, hostwin, hostwin_next);
}
-int vfio_host_win_del(VFIOContainer *container,
+int vfio_host_win_del(VFIOLegacyContainer *container,
hwaddr min_iova, hwaddr max_iova)
{
VFIOHostDMAWindow *hostwin;
@@ -337,7 +337,7 @@ static bool vfio_get_xlat_addr(IOMMUTLBEntry *iotlb, void **vaddr,
static void vfio_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
{
VFIOGuestIOMMU *giommu = container_of(n, VFIOGuestIOMMU, n);
- VFIOContainer *container = giommu->container;
+ VFIOLegacyContainer *container = giommu->container;
hwaddr iova = iotlb->iova + giommu->iommu_offset;
void *vaddr;
int ret;
@@ -441,7 +441,7 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
return 0;
}
-static void vfio_register_ram_discard_listener(VFIOContainer *container,
+static void vfio_register_ram_discard_listener(VFIOLegacyContainer *container,
MemoryRegionSection *section)
{
RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
@@ -515,7 +515,7 @@ static void vfio_register_ram_discard_listener(VFIOContainer *container,
}
}
-static void vfio_unregister_ram_discard_listener(VFIOContainer *container,
+static void vfio_unregister_ram_discard_listener(VFIOLegacyContainer *container,
MemoryRegionSection *section)
{
RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
@@ -538,7 +538,7 @@ static void vfio_unregister_ram_discard_listener(VFIOContainer *container,
g_free(vrdl);
}
-static VFIOHostDMAWindow *vfio_find_hostwin(VFIOContainer *container,
+static VFIOHostDMAWindow *vfio_find_hostwin(VFIOLegacyContainer *container,
hwaddr iova, hwaddr end)
{
VFIOHostDMAWindow *hostwin;
@@ -599,7 +599,7 @@ static bool vfio_listener_valid_section(MemoryRegionSection *section,
return true;
}
-static bool vfio_get_section_iova_range(VFIOContainer *container,
+static bool vfio_get_section_iova_range(VFIOLegacyContainer *container,
MemoryRegionSection *section,
hwaddr *out_iova, hwaddr *out_end,
Int128 *out_llend)
@@ -627,7 +627,9 @@ static bool vfio_get_section_iova_range(VFIOContainer *container,
static void vfio_listener_region_add(MemoryListener *listener,
MemoryRegionSection *section)
{
- VFIOContainer *container = container_of(listener, VFIOContainer, listener);
+ VFIOLegacyContainer *container = container_of(listener,
+ VFIOLegacyContainer,
+ listener);
hwaddr iova, end;
Int128 llend, llsize;
void *vaddr;
@@ -788,7 +790,9 @@ fail:
static void vfio_listener_region_del(MemoryListener *listener,
MemoryRegionSection *section)
{
- VFIOContainer *container = container_of(listener, VFIOContainer, listener);
+ VFIOLegacyContainer *container = container_of(listener,
+ VFIOLegacyContainer,
+ listener);
hwaddr iova, end;
Int128 llend, llsize;
int ret;
@@ -881,13 +885,13 @@ typedef struct VFIODirtyRanges {
} VFIODirtyRanges;
typedef struct VFIODirtyRangesListener {
- VFIOContainer *container;
+ VFIOLegacyContainer *container;
VFIODirtyRanges ranges;
MemoryListener listener;
} VFIODirtyRangesListener;
static bool vfio_section_is_vfio_pci(MemoryRegionSection *section,
- VFIOContainer *container)
+ VFIOLegacyContainer *container)
{
VFIOPCIDevice *pcidev;
VFIODevice *vbasedev;
@@ -966,7 +970,7 @@ static const MemoryListener vfio_dirty_tracking_listener = {
.region_add = vfio_dirty_tracking_update,
};
-static void vfio_dirty_tracking_init(VFIOContainer *container,
+static void vfio_dirty_tracking_init(VFIOLegacyContainer *container,
VFIODirtyRanges *ranges)
{
VFIODirtyRangesListener dirty;
@@ -991,7 +995,7 @@ static void vfio_dirty_tracking_init(VFIOContainer *container,
memory_listener_unregister(&dirty.listener);
}
-static void vfio_devices_dma_logging_stop(VFIOContainer *container)
+static void vfio_devices_dma_logging_stop(VFIOLegacyContainer *container)
{
uint64_t buf[DIV_ROUND_UP(sizeof(struct vfio_device_feature),
sizeof(uint64_t))] = {};
@@ -1016,7 +1020,7 @@ static void vfio_devices_dma_logging_stop(VFIOContainer *container)
}
static struct vfio_device_feature *
-vfio_device_feature_dma_logging_start_create(VFIOContainer *container,
+vfio_device_feature_dma_logging_start_create(VFIOLegacyContainer *container,
VFIODirtyRanges *tracking)
{
struct vfio_device_feature *feature;
@@ -1089,7 +1093,7 @@ static void vfio_device_feature_dma_logging_start_destroy(
g_free(feature);
}
-static int vfio_devices_dma_logging_start(VFIOContainer *container)
+static int vfio_devices_dma_logging_start(VFIOLegacyContainer *container)
{
struct vfio_device_feature *feature;
VFIODirtyRanges ranges;
@@ -1130,7 +1134,9 @@ out:
static void vfio_listener_log_global_start(MemoryListener *listener)
{
- VFIOContainer *container = container_of(listener, VFIOContainer, listener);
+ VFIOLegacyContainer *container = container_of(listener,
+ VFIOLegacyContainer,
+ listener);
int ret;
if (vfio_devices_all_device_dirty_tracking(container)) {
@@ -1148,7 +1154,9 @@ static void vfio_listener_log_global_start(MemoryListener *listener)
static void vfio_listener_log_global_stop(MemoryListener *listener)
{
- VFIOContainer *container = container_of(listener, VFIOContainer, listener);
+ VFIOLegacyContainer *container = container_of(listener,
+ VFIOLegacyContainer,
+ listener);
int ret = 0;
if (vfio_devices_all_device_dirty_tracking(container)) {
@@ -1190,7 +1198,7 @@ static int vfio_device_dma_logging_report(VFIODevice *vbasedev, hwaddr iova,
return 0;
}
-int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
+int vfio_devices_query_dirty_bitmap(VFIOLegacyContainer *container,
VFIOBitmap *vbmap, hwaddr iova,
hwaddr size)
{
@@ -1213,7 +1221,7 @@ int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
return 0;
}
-int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
+int vfio_get_dirty_bitmap(VFIOLegacyContainer *container, uint64_t iova,
uint64_t size, ram_addr_t ram_addr)
{
bool all_device_dirty_tracking =
@@ -1265,7 +1273,7 @@ static void vfio_iommu_map_dirty_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
vfio_giommu_dirty_notifier *gdn = container_of(n,
vfio_giommu_dirty_notifier, n);
VFIOGuestIOMMU *giommu = gdn->giommu;
- VFIOContainer *container = giommu->container;
+ VFIOLegacyContainer *container = giommu->container;
hwaddr iova = iotlb->iova + giommu->iommu_offset;
ram_addr_t translated_addr;
int ret = -EINVAL;
@@ -1313,7 +1321,8 @@ static int vfio_ram_discard_get_dirty_bitmap(MemoryRegionSection *section,
return vfio_get_dirty_bitmap(vrdl->container, iova, size, ram_addr);
}
-static int vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainer *container,
+static int
+vfio_sync_ram_discard_listener_dirty_bitmap(VFIOLegacyContainer *container,
MemoryRegionSection *section)
{
RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
@@ -1340,7 +1349,7 @@ static int vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainer *container,
&vrdl);
}
-static int vfio_sync_dirty_bitmap(VFIOContainer *container,
+static int vfio_sync_dirty_bitmap(VFIOLegacyContainer *container,
MemoryRegionSection *section)
{
ram_addr_t ram_addr;
@@ -1386,7 +1395,9 @@ static int vfio_sync_dirty_bitmap(VFIOContainer *container,
static void vfio_listener_log_sync(MemoryListener *listener,
MemoryRegionSection *section)
{
- VFIOContainer *container = container_of(listener, VFIOContainer, listener);
+ VFIOLegacyContainer *container = container_of(listener,
+ VFIOLegacyContainer,
+ listener);
int ret;
if (vfio_listener_skipped_section(section)) {
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index adc467210f..8fde302ae9 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -42,7 +42,8 @@
VFIOGroupList vfio_group_list =
QLIST_HEAD_INITIALIZER(vfio_group_list);
-static int vfio_ram_block_discard_disable(VFIOContainer *container, bool state)
+static int vfio_ram_block_discard_disable(VFIOLegacyContainer *container,
+ bool state)
{
switch (container->iommu_type) {
case VFIO_TYPE1v2_IOMMU:
@@ -65,7 +66,7 @@ static int vfio_ram_block_discard_disable(VFIOContainer *container, bool state)
}
}
-static int vfio_dma_unmap_bitmap(VFIOContainer *container,
+static int vfio_dma_unmap_bitmap(VFIOLegacyContainer *container,
hwaddr iova, ram_addr_t size,
IOMMUTLBEntry *iotlb)
{
@@ -120,7 +121,7 @@ unmap_exit:
/*
* DMA - Mapping and unmapping for the "type1" IOMMU interface used on x86
*/
-int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
+int vfio_dma_unmap(VFIOLegacyContainer *container, hwaddr iova,
ram_addr_t size, IOMMUTLBEntry *iotlb)
{
struct vfio_iommu_type1_dma_unmap unmap = {
@@ -175,7 +176,7 @@ int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
return 0;
}
-int vfio_dma_map(VFIOContainer *container, hwaddr iova,
+int vfio_dma_map(VFIOLegacyContainer *container, hwaddr iova,
ram_addr_t size, void *vaddr, bool readonly)
{
struct vfio_iommu_type1_dma_map map = {
@@ -205,7 +206,7 @@ int vfio_dma_map(VFIOContainer *container, hwaddr iova,
return -errno;
}
-int vfio_container_add_section_window(VFIOContainer *container,
+int vfio_container_add_section_window(VFIOLegacyContainer *container,
MemoryRegionSection *section,
Error **errp)
{
@@ -273,7 +274,7 @@ int vfio_container_add_section_window(VFIOContainer *container,
return 0;
}
-void vfio_container_del_section_window(VFIOContainer *container,
+void vfio_container_del_section_window(VFIOLegacyContainer *container,
MemoryRegionSection *section)
{
if (container->iommu_type != VFIO_SPAPR_TCE_v2_IOMMU) {
@@ -291,7 +292,7 @@ void vfio_container_del_section_window(VFIOContainer *container,
}
}
-int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start)
+int vfio_set_dirty_page_tracking(VFIOLegacyContainer *container, bool start)
{
int ret;
struct vfio_iommu_type1_dirty_bitmap dirty = {
@@ -318,7 +319,7 @@ int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start)
return ret;
}
-int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
+int vfio_query_dirty_bitmap(VFIOLegacyContainer *container, VFIOBitmap *vbmap,
hwaddr iova, hwaddr size)
{
struct vfio_iommu_type1_dirty_bitmap *dbitmap;
@@ -355,7 +356,7 @@ int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
return ret;
}
-static void vfio_listener_release(VFIOContainer *container)
+static void vfio_listener_release(VFIOLegacyContainer *container)
{
memory_listener_unregister(&container->listener);
if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
@@ -415,7 +416,7 @@ static void vfio_kvm_device_del_group(VFIOGroup *group)
/*
* vfio_get_iommu_type - selects the richest iommu_type (v2 first)
*/
-static int vfio_get_iommu_type(VFIOContainer *container,
+static int vfio_get_iommu_type(VFIOLegacyContainer *container,
Error **errp)
{
int iommu_types[] = { VFIO_TYPE1v2_IOMMU, VFIO_TYPE1_IOMMU,
@@ -431,7 +432,7 @@ static int vfio_get_iommu_type(VFIOContainer *container,
return -EINVAL;
}
-static int vfio_init_container(VFIOContainer *container, int group_fd,
+static int vfio_init_container(VFIOLegacyContainer *container, int group_fd,
Error **errp)
{
int iommu_type, ret;
@@ -466,7 +467,7 @@ static int vfio_init_container(VFIOContainer *container, int group_fd,
return 0;
}
-static int vfio_get_iommu_info(VFIOContainer *container,
+static int vfio_get_iommu_info(VFIOLegacyContainer *container,
struct vfio_iommu_type1_info **info)
{
@@ -510,7 +511,7 @@ vfio_get_iommu_info_cap(struct vfio_iommu_type1_info *info, uint16_t id)
return NULL;
}
-static void vfio_get_iommu_info_migration(VFIOContainer *container,
+static void vfio_get_iommu_info_migration(VFIOLegacyContainer *container,
struct vfio_iommu_type1_info *info)
{
struct vfio_info_cap_header *hdr;
@@ -538,7 +539,7 @@ static void vfio_get_iommu_info_migration(VFIOContainer *container,
static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
Error **errp)
{
- VFIOContainer *container;
+ VFIOLegacyContainer *container;
int ret, fd;
VFIOAddressSpace *space;
@@ -778,7 +779,7 @@ put_space_exit:
static void vfio_disconnect_container(VFIOGroup *group)
{
- VFIOContainer *container = group->container;
+ VFIOLegacyContainer *container = group->container;
QLIST_REMOVE(group, container_next);
group->container = NULL;
@@ -978,7 +979,7 @@ static void vfio_put_base_device(VFIODevice *vbasedev)
/*
* Interfaces for IBM EEH (Enhanced Error Handling)
*/
-static bool vfio_eeh_container_ok(VFIOContainer *container)
+static bool vfio_eeh_container_ok(VFIOLegacyContainer *container)
{
/*
* As of 2016-03-04 (linux-4.5) the host kernel EEH/VFIO
@@ -1006,7 +1007,7 @@ static bool vfio_eeh_container_ok(VFIOContainer *container)
return true;
}
-static int vfio_eeh_container_op(VFIOContainer *container, uint32_t op)
+static int vfio_eeh_container_op(VFIOLegacyContainer *container, uint32_t op)
{
struct vfio_eeh_pe_op pe_op = {
.argsz = sizeof(pe_op),
@@ -1029,10 +1030,10 @@ static int vfio_eeh_container_op(VFIOContainer *container, uint32_t op)
return ret;
}
-static VFIOContainer *vfio_eeh_as_container(AddressSpace *as)
+static VFIOLegacyContainer *vfio_eeh_as_container(AddressSpace *as)
{
VFIOAddressSpace *space = vfio_get_address_space(as);
- VFIOContainer *container = NULL;
+ VFIOLegacyContainer *container = NULL;
if (QLIST_EMPTY(&space->containers)) {
/* No containers to act on */
@@ -1057,14 +1058,14 @@ out:
bool vfio_eeh_as_ok(AddressSpace *as)
{
- VFIOContainer *container = vfio_eeh_as_container(as);
+ VFIOLegacyContainer *container = vfio_eeh_as_container(as);
return (container != NULL) && vfio_eeh_container_ok(container);
}
int vfio_eeh_as_op(AddressSpace *as, uint32_t op)
{
- VFIOContainer *container = vfio_eeh_as_container(as);
+ VFIOLegacyContainer *container = vfio_eeh_as_container(as);
if (!container) {
return -ENODEV;
@@ -1109,7 +1110,7 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
int groupid = vfio_device_groupid(vbasedev, errp);
VFIODevice *vbasedev_iter;
VFIOGroup *group;
- VFIOContainer *container;
+ VFIOLegacyContainer *container;
int ret;
if (groupid < 0) {
diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
index 9ec1e95f6d..683252c506 100644
--- a/hw/vfio/spapr.c
+++ b/hw/vfio/spapr.c
@@ -39,8 +39,8 @@ static void *vfio_prereg_gpa_to_vaddr(MemoryRegionSection *section, hwaddr gpa)
static void vfio_prereg_listener_region_add(MemoryListener *listener,
MemoryRegionSection *section)
{
- VFIOContainer *container = container_of(listener, VFIOContainer,
- prereg_listener);
+ VFIOLegacyContainer *container = container_of(listener, VFIOLegacyContainer,
+ prereg_listener);
const hwaddr gpa = section->offset_within_address_space;
hwaddr end;
int ret;
@@ -97,8 +97,8 @@ static void vfio_prereg_listener_region_add(MemoryListener *listener,
static void vfio_prereg_listener_region_del(MemoryListener *listener,
MemoryRegionSection *section)
{
- VFIOContainer *container = container_of(listener, VFIOContainer,
- prereg_listener);
+ VFIOLegacyContainer *container = container_of(listener, VFIOLegacyContainer,
+ prereg_listener);
const hwaddr gpa = section->offset_within_address_space;
hwaddr end;
int ret;
@@ -141,7 +141,7 @@ const MemoryListener vfio_prereg_listener = {
.region_del = vfio_prereg_listener_region_del,
};
-int vfio_spapr_create_window(VFIOContainer *container,
+int vfio_spapr_create_window(VFIOLegacyContainer *container,
MemoryRegionSection *section,
hwaddr *pgsize)
{
@@ -233,7 +233,7 @@ int vfio_spapr_create_window(VFIOContainer *container,
return 0;
}
-int vfio_spapr_remove_window(VFIOContainer *container,
+int vfio_spapr_remove_window(VFIOLegacyContainer *container,
hwaddr offset_within_address_space)
{
struct vfio_iommu_spapr_tce_remove remove = {
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
2023-10-16 8:31 ` [PATCH v2 01/27] vfio: Rename VFIOContainer into VFIOLegacyContainer Zhenzhong Duan
@ 2023-10-16 8:31 ` Zhenzhong Duan
2023-10-17 15:51 ` Cédric Le Goater
2023-10-16 8:31 ` [PATCH v2 03/27] VFIO/container: Introduce dummy VFIOContainerClass implementation Zhenzhong Duan
` (24 subsequent siblings)
26 siblings, 1 reply; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:31 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
Introduce a dumb VFIOContainer base object and its targetted interface.
This is willingly not a QOM object because we don't want it to be
visible from the user interface. The VFIOContainer will be smoothly
populated in subsequent patches as well as interfaces.
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 8 +--
include/hw/vfio/vfio-container-base.h | 82 +++++++++++++++++++++++++++
2 files changed, 84 insertions(+), 6 deletions(-)
create mode 100644 include/hw/vfio/vfio-container-base.h
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 34648e518e..9651cf921c 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -30,6 +30,7 @@
#include <linux/vfio.h>
#endif
#include "sysemu/sysemu.h"
+#include "hw/vfio/vfio-container-base.h"
#define VFIO_MSG_PREFIX "vfio %s: "
@@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
struct VFIOGroup;
typedef struct VFIOLegacyContainer {
+ VFIOContainer bcontainer;
VFIOAddressSpace *space;
int fd; /* /dev/vfio/vfio, empowered by the attached groups */
MemoryListener listener;
@@ -200,12 +202,6 @@ typedef struct VFIODisplay {
} dmabuf;
} VFIODisplay;
-typedef struct {
- unsigned long *bitmap;
- hwaddr size;
- hwaddr pages;
-} VFIOBitmap;
-
void vfio_host_win_add(VFIOLegacyContainer *container,
hwaddr min_iova, hwaddr max_iova,
uint64_t iova_pgsizes);
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
new file mode 100644
index 0000000000..afc8543d22
--- /dev/null
+++ b/include/hw/vfio/vfio-container-base.h
@@ -0,0 +1,82 @@
+/*
+ * VFIO BASE CONTAINER
+ *
+ * Copyright (C) 2023 Intel Corporation.
+ * Copyright Red Hat, Inc. 2023
+ *
+ * Authors: Yi Liu <yi.l.liu@intel.com>
+ * Eric Auger <eric.auger@redhat.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
+#define HW_VFIO_VFIO_BASE_CONTAINER_H
+
+#include "exec/memory.h"
+#ifndef CONFIG_USER_ONLY
+#include "exec/hwaddr.h"
+#endif
+
+typedef struct VFIOContainer VFIOContainer;
+typedef struct VFIODevice VFIODevice;
+typedef struct VFIOIOMMUBackendOpsClass VFIOIOMMUBackendOpsClass;
+
+typedef struct {
+ unsigned long *bitmap;
+ hwaddr size;
+ hwaddr pages;
+} VFIOBitmap;
+
+/*
+ * This is the base object for vfio container backends
+ */
+struct VFIOContainer {
+ VFIOIOMMUBackendOpsClass *ops;
+};
+
+#define TYPE_VFIO_IOMMU_BACKEND_OPS "vfio-iommu-backend-ops"
+
+DECLARE_CLASS_CHECKERS(VFIOIOMMUBackendOpsClass,
+ VFIO_IOMMU_BACKEND_OPS, TYPE_VFIO_IOMMU_BACKEND_OPS)
+
+struct VFIOIOMMUBackendOpsClass {
+ /*< private >*/
+ ObjectClass parent_class;
+
+ /*< public >*/
+ /* required */
+ int (*dma_map)(VFIOContainer *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ void *vaddr, bool readonly);
+ int (*dma_unmap)(VFIOContainer *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ IOMMUTLBEntry *iotlb);
+ int (*attach_device)(char *name, VFIODevice *vbasedev,
+ AddressSpace *as, Error **errp);
+ void (*detach_device)(VFIODevice *vbasedev);
+ /* migration feature */
+ int (*set_dirty_page_tracking)(VFIOContainer *bcontainer, bool start);
+ int (*query_dirty_bitmap)(VFIOContainer *bcontainer, VFIOBitmap *vbmap,
+ hwaddr iova, hwaddr size);
+
+ /* SPAPR specific */
+ int (*add_window)(VFIOContainer *bcontainer,
+ MemoryRegionSection *section,
+ Error **errp);
+ void (*del_window)(VFIOContainer *bcontainer,
+ MemoryRegionSection *section);
+};
+
+#endif /* HW_VFIO_VFIO_BASE_CONTAINER_H */
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 03/27] VFIO/container: Introduce dummy VFIOContainerClass implementation
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
2023-10-16 8:31 ` [PATCH v2 01/27] vfio: Rename VFIOContainer into VFIOLegacyContainer Zhenzhong Duan
2023-10-16 8:31 ` [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface Zhenzhong Duan
@ 2023-10-16 8:31 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 04/27] vfio/container: Switch to dma_map|unmap API Zhenzhong Duan
` (23 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:31 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
Let's instantiate a dummy VFIOContainerClass implementation whose
functions are not yet implemented.
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-container-base.h | 1 +
hw/vfio/container-base.c | 40 +++++++++++++++++++++++++++
hw/vfio/container.c | 22 +++++++++++++++
hw/vfio/meson.build | 1 +
4 files changed, 64 insertions(+)
create mode 100644 hw/vfio/container-base.c
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index afc8543d22..226e960fb5 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -46,6 +46,7 @@ struct VFIOContainer {
VFIOIOMMUBackendOpsClass *ops;
};
+#define TYPE_VFIO_IOMMU_BACKEND_LEGACY_OPS "vfio-iommu-backend-legacy-ops"
#define TYPE_VFIO_IOMMU_BACKEND_OPS "vfio-iommu-backend-ops"
DECLARE_CLASS_CHECKERS(VFIOIOMMUBackendOpsClass,
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
new file mode 100644
index 0000000000..0c21e77039
--- /dev/null
+++ b/hw/vfio/container-base.c
@@ -0,0 +1,40 @@
+/*
+ * VFIO BASE CONTAINER
+ *
+ * Copyright (C) 2023 Intel Corporation.
+ * Copyright Red Hat, Inc. 2023
+ *
+ * Authors: Yi Liu <yi.l.liu@intel.com>
+ * Eric Auger <eric.auger@redhat.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "qemu/osdep.h"
+#include "qapi/error.h"
+#include "qemu/error-report.h"
+#include "hw/vfio/vfio-container-base.h"
+
+static const TypeInfo vfio_iommu_backend_ops_type_info = {
+ .name = TYPE_VFIO_IOMMU_BACKEND_OPS,
+ .parent = TYPE_OBJECT,
+ .abstract = true,
+ .class_size = sizeof(VFIOIOMMUBackendOpsClass),
+};
+
+static void vfio_iommu_backend_ops_register_types(void)
+{
+ type_register_static(&vfio_iommu_backend_ops_type_info);
+}
+type_init(vfio_iommu_backend_ops_register_types);
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 8fde302ae9..acc4a6bf8a 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -539,6 +539,9 @@ static void vfio_get_iommu_info_migration(VFIOLegacyContainer *container,
static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
Error **errp)
{
+ VFIOIOMMUBackendOpsClass *ops = VFIO_IOMMU_BACKEND_OPS_CLASS(
+ object_class_by_name(TYPE_VFIO_IOMMU_BACKEND_LEGACY_OPS));
+ VFIOContainer *bcontainer;
VFIOLegacyContainer *container;
int ret, fd;
VFIOAddressSpace *space;
@@ -620,6 +623,8 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
QLIST_INIT(&container->giommu_list);
QLIST_INIT(&container->hostwin_list);
QLIST_INIT(&container->vrdl_list);
+ bcontainer = &container->bcontainer;
+ bcontainer->ops = ops;
ret = vfio_init_container(container, group->fd, errp);
if (ret) {
@@ -1160,3 +1165,20 @@ void vfio_detach_device(VFIODevice *vbasedev)
vfio_put_base_device(vbasedev);
vfio_put_group(group);
}
+
+static void vfio_iommu_backend_legacy_ops_class_init(ObjectClass *oc,
+ void *data) {
+}
+
+static const TypeInfo vfio_iommu_backend_legacy_ops_type = {
+ .name = TYPE_VFIO_IOMMU_BACKEND_LEGACY_OPS,
+
+ .parent = TYPE_VFIO_IOMMU_BACKEND_OPS,
+ .class_init = vfio_iommu_backend_legacy_ops_class_init,
+ .abstract = true,
+};
+static void vfio_iommu_backend_legacy_ops_register_types(void)
+{
+ type_register_static(&vfio_iommu_backend_legacy_ops_type);
+}
+type_init(vfio_iommu_backend_legacy_ops_register_types);
diff --git a/hw/vfio/meson.build b/hw/vfio/meson.build
index 2a6912c940..eb6ce6229d 100644
--- a/hw/vfio/meson.build
+++ b/hw/vfio/meson.build
@@ -2,6 +2,7 @@ vfio_ss = ss.source_set()
vfio_ss.add(files(
'helpers.c',
'common.c',
+ 'container-base.c',
'container.c',
'spapr.c',
'migration.c',
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 04/27] vfio/container: Switch to dma_map|unmap API
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (2 preceding siblings ...)
2023-10-16 8:31 ` [PATCH v2 03/27] VFIO/container: Introduce dummy VFIOContainerClass implementation Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 05/27] vfio/common: Move giommu_list in base container Zhenzhong Duan
` (22 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 4 ---
include/hw/vfio/vfio-container-base.h | 7 +++++
hw/vfio/common.c | 45 +++++++++++++++------------
hw/vfio/container-base.c | 22 +++++++++++++
hw/vfio/container.c | 25 +++++++++++----
hw/vfio/trace-events | 2 +-
6 files changed, 74 insertions(+), 31 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 9651cf921c..f2aa122c47 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -212,10 +212,6 @@ void vfio_put_address_space(VFIOAddressSpace *space);
bool vfio_devices_all_running_and_saving(VFIOLegacyContainer *container);
/* container->fd */
-int vfio_dma_unmap(VFIOLegacyContainer *container, hwaddr iova,
- ram_addr_t size, IOMMUTLBEntry *iotlb);
-int vfio_dma_map(VFIOLegacyContainer *container, hwaddr iova,
- ram_addr_t size, void *vaddr, bool readonly);
int vfio_set_dirty_page_tracking(VFIOLegacyContainer *container, bool start);
int vfio_query_dirty_bitmap(VFIOLegacyContainer *container, VFIOBitmap *vbmap,
hwaddr iova, hwaddr size);
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 226e960fb5..1483e77441 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -46,6 +46,13 @@ struct VFIOContainer {
VFIOIOMMUBackendOpsClass *ops;
};
+int vfio_container_dma_map(VFIOContainer *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ void *vaddr, bool readonly);
+int vfio_container_dma_unmap(VFIOContainer *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ IOMMUTLBEntry *iotlb);
+
#define TYPE_VFIO_IOMMU_BACKEND_LEGACY_OPS "vfio-iommu-backend-legacy-ops"
#define TYPE_VFIO_IOMMU_BACKEND_OPS "vfio-iommu-backend-ops"
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index b51ef3a15a..6be1526d79 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -337,7 +337,7 @@ static bool vfio_get_xlat_addr(IOMMUTLBEntry *iotlb, void **vaddr,
static void vfio_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
{
VFIOGuestIOMMU *giommu = container_of(n, VFIOGuestIOMMU, n);
- VFIOLegacyContainer *container = giommu->container;
+ VFIOContainer *bcontainer = &giommu->container->bcontainer;
hwaddr iova = iotlb->iova + giommu->iommu_offset;
void *vaddr;
int ret;
@@ -367,21 +367,22 @@ static void vfio_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
* of vaddr will always be there, even if the memory object is
* destroyed and its backing memory munmap-ed.
*/
- ret = vfio_dma_map(container, iova,
- iotlb->addr_mask + 1, vaddr,
- read_only);
+ ret = vfio_container_dma_map(bcontainer, iova,
+ iotlb->addr_mask + 1, vaddr,
+ read_only);
if (ret) {
- error_report("vfio_dma_map(%p, 0x%"HWADDR_PRIx", "
+ error_report("vfio_container_dma_map(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx", %p) = %d (%s)",
- container, iova,
+ bcontainer, iova,
iotlb->addr_mask + 1, vaddr, ret, strerror(-ret));
}
} else {
- ret = vfio_dma_unmap(container, iova, iotlb->addr_mask + 1, iotlb);
+ ret = vfio_container_dma_unmap(bcontainer, iova,
+ iotlb->addr_mask + 1, iotlb);
if (ret) {
- error_report("vfio_dma_unmap(%p, 0x%"HWADDR_PRIx", "
+ error_report("vfio_container_dma_unmap(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
- container, iova,
+ bcontainer, iova,
iotlb->addr_mask + 1, ret, strerror(-ret));
vfio_set_migration_error(ret);
}
@@ -400,9 +401,10 @@ static void vfio_ram_discard_notify_discard(RamDiscardListener *rdl,
int ret;
/* Unmap with a single call. */
- ret = vfio_dma_unmap(vrdl->container, iova, size , NULL);
+ ret = vfio_container_dma_unmap(&vrdl->container->bcontainer,
+ iova, size , NULL);
if (ret) {
- error_report("%s: vfio_dma_unmap() failed: %s", __func__,
+ error_report("%s: vfio_container_dma_unmap() failed: %s", __func__,
strerror(-ret));
}
}
@@ -430,8 +432,8 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
section->offset_within_address_space;
vaddr = memory_region_get_ram_ptr(section->mr) + start;
- ret = vfio_dma_map(vrdl->container, iova, next - start,
- vaddr, section->readonly);
+ ret = vfio_container_dma_map(&vrdl->container->bcontainer, iova,
+ next - start, vaddr, section->readonly);
if (ret) {
/* Rollback */
vfio_ram_discard_notify_discard(rdl, section);
@@ -746,10 +748,11 @@ static void vfio_listener_region_add(MemoryListener *listener,
}
}
- ret = vfio_dma_map(container, iova, int128_get64(llsize),
- vaddr, section->readonly);
+ ret = vfio_container_dma_map(&container->bcontainer,
+ iova, int128_get64(llsize), vaddr,
+ section->readonly);
if (ret) {
- error_setg(&err, "vfio_dma_map(%p, 0x%"HWADDR_PRIx", "
+ error_setg(&err, "vfio_container_dma_map(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx", %p) = %d (%s)",
container, iova, int128_get64(llsize), vaddr, ret,
strerror(-ret));
@@ -852,18 +855,20 @@ static void vfio_listener_region_del(MemoryListener *listener,
if (int128_eq(llsize, int128_2_64())) {
/* The unmap ioctl doesn't accept a full 64-bit span. */
llsize = int128_rshift(llsize, 1);
- ret = vfio_dma_unmap(container, iova, int128_get64(llsize), NULL);
+ ret = vfio_container_dma_unmap(&container->bcontainer, iova,
+ int128_get64(llsize), NULL);
if (ret) {
- error_report("vfio_dma_unmap(%p, 0x%"HWADDR_PRIx", "
+ error_report("vfio_container_dma_unmap(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
container, iova, int128_get64(llsize), ret,
strerror(-ret));
}
iova += int128_get64(llsize);
}
- ret = vfio_dma_unmap(container, iova, int128_get64(llsize), NULL);
+ ret = vfio_container_dma_unmap(&container->bcontainer, iova,
+ int128_get64(llsize), NULL);
if (ret) {
- error_report("vfio_dma_unmap(%p, 0x%"HWADDR_PRIx", "
+ error_report("vfio_container_dma_unmap(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
container, iova, int128_get64(llsize), ret,
strerror(-ret));
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index 0c21e77039..78329935f6 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -26,6 +26,28 @@
#include "qemu/error-report.h"
#include "hw/vfio/vfio-container-base.h"
+int vfio_container_dma_map(VFIOContainer *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ void *vaddr, bool readonly)
+{
+ if (!bcontainer->ops->dma_map) {
+ return -EINVAL;
+ }
+
+ return bcontainer->ops->dma_map(bcontainer, iova, size, vaddr, readonly);
+}
+
+int vfio_container_dma_unmap(VFIOContainer *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ IOMMUTLBEntry *iotlb)
+{
+ if (!bcontainer->ops->dma_unmap) {
+ return -EINVAL;
+ }
+
+ return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
+}
+
static const TypeInfo vfio_iommu_backend_ops_type_info = {
.name = TYPE_VFIO_IOMMU_BACKEND_OPS,
.parent = TYPE_OBJECT,
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index acc4a6bf8a..80aafa21ed 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -121,9 +121,13 @@ unmap_exit:
/*
* DMA - Mapping and unmapping for the "type1" IOMMU interface used on x86
*/
-int vfio_dma_unmap(VFIOLegacyContainer *container, hwaddr iova,
- ram_addr_t size, IOMMUTLBEntry *iotlb)
+static int vfio_legacy_dma_unmap(VFIOContainer *bcontainer, hwaddr iova,
+ ram_addr_t size, IOMMUTLBEntry *iotlb)
{
+ VFIOLegacyContainer *container = container_of(bcontainer,
+ VFIOLegacyContainer,
+ bcontainer);
+
struct vfio_iommu_type1_dma_unmap unmap = {
.argsz = sizeof(unmap),
.flags = 0,
@@ -157,7 +161,7 @@ int vfio_dma_unmap(VFIOLegacyContainer *container, hwaddr iova,
*/
if (errno == EINVAL && unmap.size && !(unmap.iova + unmap.size) &&
container->iommu_type == VFIO_TYPE1v2_IOMMU) {
- trace_vfio_dma_unmap_overflow_workaround();
+ trace_vfio_legacy_dma_unmap_overflow_workaround();
unmap.size -= 1ULL << ctz64(container->pgsizes);
continue;
}
@@ -176,9 +180,13 @@ int vfio_dma_unmap(VFIOLegacyContainer *container, hwaddr iova,
return 0;
}
-int vfio_dma_map(VFIOLegacyContainer *container, hwaddr iova,
- ram_addr_t size, void *vaddr, bool readonly)
+static int vfio_legacy_dma_map(VFIOContainer *bcontainer, hwaddr iova,
+ ram_addr_t size, void *vaddr, bool readonly)
{
+ VFIOLegacyContainer *container = container_of(bcontainer,
+ VFIOLegacyContainer,
+ bcontainer);
+
struct vfio_iommu_type1_dma_map map = {
.argsz = sizeof(map),
.flags = VFIO_DMA_MAP_FLAG_READ,
@@ -197,7 +205,8 @@ int vfio_dma_map(VFIOLegacyContainer *container, hwaddr iova,
* the VGA ROM space.
*/
if (ioctl(container->fd, VFIO_IOMMU_MAP_DMA, &map) == 0 ||
- (errno == EBUSY && vfio_dma_unmap(container, iova, size, NULL) == 0 &&
+ (errno == EBUSY &&
+ vfio_legacy_dma_unmap(bcontainer, iova, size, NULL) == 0 &&
ioctl(container->fd, VFIO_IOMMU_MAP_DMA, &map) == 0)) {
return 0;
}
@@ -1168,6 +1177,10 @@ void vfio_detach_device(VFIODevice *vbasedev)
static void vfio_iommu_backend_legacy_ops_class_init(ObjectClass *oc,
void *data) {
+ VFIOIOMMUBackendOpsClass *ops = VFIO_IOMMU_BACKEND_OPS_CLASS(oc);
+
+ ops->dma_map = vfio_legacy_dma_map;
+ ops->dma_unmap = vfio_legacy_dma_unmap;
}
static const TypeInfo vfio_iommu_backend_legacy_ops_type = {
diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
index 0eb2387cf2..9f7fedee98 100644
--- a/hw/vfio/trace-events
+++ b/hw/vfio/trace-events
@@ -116,7 +116,7 @@ vfio_region_unmap(const char *name, unsigned long offset, unsigned long end) "Re
vfio_region_sparse_mmap_header(const char *name, int index, int nr_areas) "Device %s region %d: %d sparse mmap entries"
vfio_region_sparse_mmap_entry(int i, unsigned long start, unsigned long end) "sparse entry %d [0x%lx - 0x%lx]"
vfio_get_dev_region(const char *name, int index, uint32_t type, uint32_t subtype) "%s index %d, %08x/%08x"
-vfio_dma_unmap_overflow_workaround(void) ""
+vfio_legacy_dma_unmap_overflow_workaround(void) ""
vfio_get_dirty_bitmap(int fd, uint64_t iova, uint64_t size, uint64_t bitmap_size, uint64_t start, uint64_t dirty_pages) "container fd=%d, iova=0x%"PRIx64" size= 0x%"PRIx64" bitmap_size=0x%"PRIx64" start=0x%"PRIx64" dirty_pages=%"PRIu64
vfio_iommu_map_dirty_notify(uint64_t iova_start, uint64_t iova_end) "iommu dirty @ 0x%"PRIx64" - 0x%"PRIx64
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 05/27] vfio/common: Move giommu_list in base container
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (3 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 04/27] vfio/container: Switch to dma_map|unmap API Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 06/27] vfio/container: Move space field to " Zhenzhong Duan
` (21 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
Move the giommu_list field in the base object and store the
base container in the VFIOGuestIOMMU.
We introduce vfio_container_init/destroy helper on the base
container.
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 9 ---------
include/hw/vfio/vfio-container-base.h | 13 +++++++++++++
hw/vfio/common.c | 18 ++++++++++++------
hw/vfio/container-base.c | 19 +++++++++++++++++++
hw/vfio/container.c | 13 +++----------
5 files changed, 47 insertions(+), 25 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index f2aa122c47..884d1627f4 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -95,7 +95,6 @@ typedef struct VFIOLegacyContainer {
uint64_t max_dirty_bitmap_size;
unsigned long pgsizes;
unsigned int dma_max_mappings;
- QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_HEAD(, VFIOGroup) group_list;
QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
@@ -103,14 +102,6 @@ typedef struct VFIOLegacyContainer {
QLIST_HEAD(, VFIODevice) device_list;
} VFIOLegacyContainer;
-typedef struct VFIOGuestIOMMU {
- VFIOLegacyContainer *container;
- IOMMUMemoryRegion *iommu_mr;
- hwaddr iommu_offset;
- IOMMUNotifier n;
- QLIST_ENTRY(VFIOGuestIOMMU) giommu_next;
-} VFIOGuestIOMMU;
-
typedef struct VFIORamDiscardListener {
VFIOLegacyContainer *container;
MemoryRegion *mr;
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 1483e77441..b6c8eb2313 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -33,6 +33,14 @@ typedef struct VFIOContainer VFIOContainer;
typedef struct VFIODevice VFIODevice;
typedef struct VFIOIOMMUBackendOpsClass VFIOIOMMUBackendOpsClass;
+typedef struct VFIOGuestIOMMU {
+ VFIOContainer *bcontainer;
+ IOMMUMemoryRegion *iommu_mr;
+ hwaddr iommu_offset;
+ IOMMUNotifier n;
+ QLIST_ENTRY(VFIOGuestIOMMU) giommu_next;
+} VFIOGuestIOMMU;
+
typedef struct {
unsigned long *bitmap;
hwaddr size;
@@ -44,6 +52,7 @@ typedef struct {
*/
struct VFIOContainer {
VFIOIOMMUBackendOpsClass *ops;
+ QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
};
int vfio_container_dma_map(VFIOContainer *bcontainer,
@@ -53,6 +62,10 @@ int vfio_container_dma_unmap(VFIOContainer *bcontainer,
hwaddr iova, ram_addr_t size,
IOMMUTLBEntry *iotlb);
+void vfio_container_init(VFIOContainer *bcontainer,
+ struct VFIOIOMMUBackendOpsClass *ops);
+void vfio_container_destroy(VFIOContainer *bcontainer);
+
#define TYPE_VFIO_IOMMU_BACKEND_LEGACY_OPS "vfio-iommu-backend-legacy-ops"
#define TYPE_VFIO_IOMMU_BACKEND_OPS "vfio-iommu-backend-ops"
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 6be1526d79..1adfdca4f5 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -337,7 +337,7 @@ static bool vfio_get_xlat_addr(IOMMUTLBEntry *iotlb, void **vaddr,
static void vfio_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
{
VFIOGuestIOMMU *giommu = container_of(n, VFIOGuestIOMMU, n);
- VFIOContainer *bcontainer = &giommu->container->bcontainer;
+ VFIOContainer *bcontainer = giommu->bcontainer;
hwaddr iova = iotlb->iova + giommu->iommu_offset;
void *vaddr;
int ret;
@@ -632,6 +632,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
VFIOLegacyContainer *container = container_of(listener,
VFIOLegacyContainer,
listener);
+ VFIOContainer *bcontainer = &container->bcontainer;
hwaddr iova, end;
Int128 llend, llsize;
void *vaddr;
@@ -683,7 +684,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
giommu->iommu_mr = iommu_mr;
giommu->iommu_offset = section->offset_within_address_space -
section->offset_within_region;
- giommu->container = container;
+ giommu->bcontainer = bcontainer;
llend = int128_add(int128_make64(section->offset_within_region),
section->size);
llend = int128_sub(llend, int128_one());
@@ -709,7 +710,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
g_free(giommu);
goto fail;
}
- QLIST_INSERT_HEAD(&container->giommu_list, giommu, giommu_next);
+ QLIST_INSERT_HEAD(&bcontainer->giommu_list, giommu, giommu_next);
memory_region_iommu_replay(giommu->iommu_mr, &giommu->n);
return;
@@ -796,6 +797,7 @@ static void vfio_listener_region_del(MemoryListener *listener,
VFIOLegacyContainer *container = container_of(listener,
VFIOLegacyContainer,
listener);
+ VFIOContainer *bcontainer = &container->bcontainer;
hwaddr iova, end;
Int128 llend, llsize;
int ret;
@@ -808,7 +810,7 @@ static void vfio_listener_region_del(MemoryListener *listener,
if (memory_region_is_iommu(section->mr)) {
VFIOGuestIOMMU *giommu;
- QLIST_FOREACH(giommu, &container->giommu_list, giommu_next) {
+ QLIST_FOREACH(giommu, &bcontainer->giommu_list, giommu_next) {
if (MEMORY_REGION(giommu->iommu_mr) == section->mr &&
giommu->n.start == section->offset_within_region) {
memory_region_unregister_iommu_notifier(section->mr,
@@ -1278,7 +1280,10 @@ static void vfio_iommu_map_dirty_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
vfio_giommu_dirty_notifier *gdn = container_of(n,
vfio_giommu_dirty_notifier, n);
VFIOGuestIOMMU *giommu = gdn->giommu;
- VFIOLegacyContainer *container = giommu->container;
+ VFIOContainer *bcontainer = giommu->bcontainer;
+ VFIOLegacyContainer *container = container_of(bcontainer,
+ VFIOLegacyContainer,
+ bcontainer);
hwaddr iova = iotlb->iova + giommu->iommu_offset;
ram_addr_t translated_addr;
int ret = -EINVAL;
@@ -1357,12 +1362,13 @@ vfio_sync_ram_discard_listener_dirty_bitmap(VFIOLegacyContainer *container,
static int vfio_sync_dirty_bitmap(VFIOLegacyContainer *container,
MemoryRegionSection *section)
{
+ VFIOContainer *bcontainer = &container->bcontainer;
ram_addr_t ram_addr;
if (memory_region_is_iommu(section->mr)) {
VFIOGuestIOMMU *giommu;
- QLIST_FOREACH(giommu, &container->giommu_list, giommu_next) {
+ QLIST_FOREACH(giommu, &bcontainer->giommu_list, giommu_next) {
if (MEMORY_REGION(giommu->iommu_mr) == section->mr &&
giommu->n.start == section->offset_within_region) {
Int128 llend;
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index 78329935f6..6da50e8151 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -48,6 +48,25 @@ int vfio_container_dma_unmap(VFIOContainer *bcontainer,
return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
}
+void vfio_container_init(VFIOContainer *bcontainer,
+ struct VFIOIOMMUBackendOpsClass *ops)
+{
+ bcontainer->ops = ops;
+ QLIST_INIT(&bcontainer->giommu_list);
+}
+
+void vfio_container_destroy(VFIOContainer *bcontainer)
+{
+ VFIOGuestIOMMU *giommu, *tmp;
+
+ QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next, tmp) {
+ memory_region_unregister_iommu_notifier(
+ MEMORY_REGION(giommu->iommu_mr), &giommu->n);
+ QLIST_REMOVE(giommu, giommu_next);
+ g_free(giommu);
+ }
+}
+
static const TypeInfo vfio_iommu_backend_ops_type_info = {
.name = TYPE_VFIO_IOMMU_BACKEND_OPS,
.parent = TYPE_OBJECT,
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 80aafa21ed..de6b018eeb 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -629,11 +629,10 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container->error = NULL;
container->dirty_pages_supported = false;
container->dma_max_mappings = 0;
- QLIST_INIT(&container->giommu_list);
QLIST_INIT(&container->hostwin_list);
QLIST_INIT(&container->vrdl_list);
bcontainer = &container->bcontainer;
- bcontainer->ops = ops;
+ vfio_container_init(bcontainer, ops);
ret = vfio_init_container(container, group->fd, errp);
if (ret) {
@@ -794,6 +793,7 @@ put_space_exit:
static void vfio_disconnect_container(VFIOGroup *group)
{
VFIOLegacyContainer *container = group->container;
+ VFIOContainer *bcontainer = &container->bcontainer;
QLIST_REMOVE(group, container_next);
group->container = NULL;
@@ -814,17 +814,10 @@ static void vfio_disconnect_container(VFIOGroup *group)
if (QLIST_EMPTY(&container->group_list)) {
VFIOAddressSpace *space = container->space;
- VFIOGuestIOMMU *giommu, *tmp;
VFIOHostDMAWindow *hostwin, *next;
QLIST_REMOVE(container, next);
-
- QLIST_FOREACH_SAFE(giommu, &container->giommu_list, giommu_next, tmp) {
- memory_region_unregister_iommu_notifier(
- MEMORY_REGION(giommu->iommu_mr), &giommu->n);
- QLIST_REMOVE(giommu, giommu_next);
- g_free(giommu);
- }
+ vfio_container_destroy(bcontainer);
QLIST_FOREACH_SAFE(hostwin, &container->hostwin_list, hostwin_next,
next) {
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 06/27] vfio/container: Move space field to base container
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (4 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 05/27] vfio/common: Move giommu_list in base container Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 07/27] vfio/container: switch to IOMMU BE add/del_section_window Zhenzhong Duan
` (20 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
Move the space field to the base object. Also the VFIOAddressSpace
now contains a list of base containers.
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 8 --------
include/hw/vfio/vfio-container-base.h | 9 +++++++++
hw/vfio/common.c | 4 ++--
hw/vfio/container-base.c | 4 ++++
hw/vfio/container.c | 28 +++++++++++++--------------
5 files changed, 29 insertions(+), 24 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 884d1627f4..33f475957c 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -73,17 +73,10 @@ typedef struct VFIOMigration {
bool initial_data_sent;
} VFIOMigration;
-typedef struct VFIOAddressSpace {
- AddressSpace *as;
- QLIST_HEAD(, VFIOLegacyContainer) containers;
- QLIST_ENTRY(VFIOAddressSpace) list;
-} VFIOAddressSpace;
-
struct VFIOGroup;
typedef struct VFIOLegacyContainer {
VFIOContainer bcontainer;
- VFIOAddressSpace *space;
int fd; /* /dev/vfio/vfio, empowered by the attached groups */
MemoryListener listener;
MemoryListener prereg_listener;
@@ -98,7 +91,6 @@ typedef struct VFIOLegacyContainer {
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_HEAD(, VFIOGroup) group_list;
QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
- QLIST_ENTRY(VFIOLegacyContainer) next;
QLIST_HEAD(, VFIODevice) device_list;
} VFIOLegacyContainer;
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index b6c8eb2313..9504564f4e 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -33,6 +33,12 @@ typedef struct VFIOContainer VFIOContainer;
typedef struct VFIODevice VFIODevice;
typedef struct VFIOIOMMUBackendOpsClass VFIOIOMMUBackendOpsClass;
+typedef struct VFIOAddressSpace {
+ AddressSpace *as;
+ QLIST_HEAD(, VFIOContainer) containers;
+ QLIST_ENTRY(VFIOAddressSpace) list;
+} VFIOAddressSpace;
+
typedef struct VFIOGuestIOMMU {
VFIOContainer *bcontainer;
IOMMUMemoryRegion *iommu_mr;
@@ -52,7 +58,9 @@ typedef struct {
*/
struct VFIOContainer {
VFIOIOMMUBackendOpsClass *ops;
+ VFIOAddressSpace *space;
QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
+ QLIST_ENTRY(VFIOContainer) next;
};
int vfio_container_dma_map(VFIOContainer *bcontainer,
@@ -63,6 +71,7 @@ int vfio_container_dma_unmap(VFIOContainer *bcontainer,
IOMMUTLBEntry *iotlb);
void vfio_container_init(VFIOContainer *bcontainer,
+ VFIOAddressSpace *space,
struct VFIOIOMMUBackendOpsClass *ops);
void vfio_container_destroy(VFIOContainer *bcontainer);
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 1adfdca4f5..c92af34eed 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -152,7 +152,7 @@ void vfio_unblock_multiple_devices_migration(void)
bool vfio_viommu_preset(VFIODevice *vbasedev)
{
- return vbasedev->container->space->as != &address_space_memory;
+ return vbasedev->container->bcontainer.space->as != &address_space_memory;
}
static void vfio_set_migration_error(int err)
@@ -990,7 +990,7 @@ static void vfio_dirty_tracking_init(VFIOLegacyContainer *container,
dirty.container = container;
memory_listener_register(&dirty.listener,
- container->space->as);
+ container->bcontainer.space->as);
*ranges = dirty.ranges;
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index 6da50e8151..e1056dd78e 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -49,9 +49,11 @@ int vfio_container_dma_unmap(VFIOContainer *bcontainer,
}
void vfio_container_init(VFIOContainer *bcontainer,
+ VFIOAddressSpace *space,
struct VFIOIOMMUBackendOpsClass *ops)
{
bcontainer->ops = ops;
+ bcontainer->space = space;
QLIST_INIT(&bcontainer->giommu_list);
}
@@ -59,6 +61,8 @@ void vfio_container_destroy(VFIOContainer *bcontainer)
{
VFIOGuestIOMMU *giommu, *tmp;
+ QLIST_REMOVE(bcontainer, next);
+
QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next, tmp) {
memory_region_unregister_iommu_notifier(
MEMORY_REGION(giommu->iommu_mr), &giommu->n);
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index de6b018eeb..fd2d602fb9 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -588,7 +588,8 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
* details once we know which type of IOMMU we are using.
*/
- QLIST_FOREACH(container, &space->containers, next) {
+ QLIST_FOREACH(bcontainer, &space->containers, next) {
+ container = container_of(bcontainer, VFIOLegacyContainer, bcontainer);
if (!ioctl(group->fd, VFIO_GROUP_SET_CONTAINER, &container->fd)) {
ret = vfio_ram_block_discard_disable(container, true);
if (ret) {
@@ -624,7 +625,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
}
container = g_malloc0(sizeof(*container));
- container->space = space;
container->fd = fd;
container->error = NULL;
container->dirty_pages_supported = false;
@@ -632,7 +632,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
QLIST_INIT(&container->hostwin_list);
QLIST_INIT(&container->vrdl_list);
bcontainer = &container->bcontainer;
- vfio_container_init(bcontainer, ops);
+ vfio_container_init(bcontainer, space, ops);
ret = vfio_init_container(container, group->fd, errp);
if (ret) {
@@ -750,14 +750,15 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
vfio_kvm_device_add_group(group);
QLIST_INIT(&container->group_list);
- QLIST_INSERT_HEAD(&space->containers, container, next);
+ QLIST_INSERT_HEAD(&space->containers, bcontainer, next);
group->container = container;
QLIST_INSERT_HEAD(&container->group_list, group, container_next);
container->listener = vfio_memory_listener;
- memory_listener_register(&container->listener, container->space->as);
+ memory_listener_register(&container->listener,
+ container->bcontainer.space->as);
if (container->error) {
ret = -1;
@@ -771,7 +772,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
return 0;
listener_release_exit:
QLIST_REMOVE(group, container_next);
- QLIST_REMOVE(container, next);
+ QLIST_REMOVE(bcontainer, next);
vfio_kvm_device_del_group(group);
vfio_listener_release(container);
@@ -813,10 +814,9 @@ static void vfio_disconnect_container(VFIOGroup *group)
}
if (QLIST_EMPTY(&container->group_list)) {
- VFIOAddressSpace *space = container->space;
+ VFIOAddressSpace *space = container->bcontainer.space;
VFIOHostDMAWindow *hostwin, *next;
- QLIST_REMOVE(container, next);
vfio_container_destroy(bcontainer);
QLIST_FOREACH_SAFE(hostwin, &container->hostwin_list, hostwin_next,
@@ -842,7 +842,7 @@ static VFIOGroup *vfio_get_group(int groupid, AddressSpace *as, Error **errp)
QLIST_FOREACH(group, &vfio_group_list, next) {
if (group->groupid == groupid) {
/* Found it. Now is it already in the right context? */
- if (group->container->space->as == as) {
+ if (group->container->bcontainer.space->as == as) {
return group;
} else {
error_setg(errp, "group %d used in multiple address spaces",
@@ -1040,27 +1040,27 @@ static int vfio_eeh_container_op(VFIOLegacyContainer *container, uint32_t op)
static VFIOLegacyContainer *vfio_eeh_as_container(AddressSpace *as)
{
VFIOAddressSpace *space = vfio_get_address_space(as);
- VFIOLegacyContainer *container = NULL;
+ VFIOContainer *bcontainer = NULL;
if (QLIST_EMPTY(&space->containers)) {
/* No containers to act on */
goto out;
}
- container = QLIST_FIRST(&space->containers);
+ bcontainer = QLIST_FIRST(&space->containers);
- if (QLIST_NEXT(container, next)) {
+ if (QLIST_NEXT(bcontainer, next)) {
/*
* We don't yet have logic to synchronize EEH state across
* multiple containers
*/
- container = NULL;
+ bcontainer = NULL;
goto out;
}
out:
vfio_put_address_space(space);
- return container;
+ return container_of(bcontainer, VFIOLegacyContainer, bcontainer);
}
bool vfio_eeh_as_ok(AddressSpace *as)
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 07/27] vfio/container: switch to IOMMU BE add/del_section_window
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (5 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 06/27] vfio/container: Move space field to " Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 08/27] vfio/container: Move hostwin_list in base container Zhenzhong Duan
` (19 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 6 ------
include/hw/vfio/vfio-container-base.h | 5 +++++
hw/vfio/common.c | 4 ++--
hw/vfio/container-base.c | 21 +++++++++++++++++++++
hw/vfio/container.c | 19 ++++++++++++++-----
5 files changed, 42 insertions(+), 13 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 33f475957c..b83ae4b3b6 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -199,12 +199,6 @@ int vfio_set_dirty_page_tracking(VFIOLegacyContainer *container, bool start);
int vfio_query_dirty_bitmap(VFIOLegacyContainer *container, VFIOBitmap *vbmap,
hwaddr iova, hwaddr size);
-int vfio_container_add_section_window(VFIOLegacyContainer *container,
- MemoryRegionSection *section,
- Error **errp);
-void vfio_container_del_section_window(VFIOLegacyContainer *container,
- MemoryRegionSection *section);
-
void vfio_disable_irqindex(VFIODevice *vbasedev, int index);
void vfio_unmask_single_irqindex(VFIODevice *vbasedev, int index);
void vfio_mask_single_irqindex(VFIODevice *vbasedev, int index);
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 9504564f4e..1f6d5fd229 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -69,6 +69,11 @@ int vfio_container_dma_map(VFIOContainer *bcontainer,
int vfio_container_dma_unmap(VFIOContainer *bcontainer,
hwaddr iova, ram_addr_t size,
IOMMUTLBEntry *iotlb);
+int vfio_container_add_section_window(VFIOContainer *bcontainer,
+ MemoryRegionSection *section,
+ Error **errp);
+void vfio_container_del_section_window(VFIOContainer *bcontainer,
+ MemoryRegionSection *section);
void vfio_container_init(VFIOContainer *bcontainer,
VFIOAddressSpace *space,
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index c92af34eed..49cb5b6958 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -655,7 +655,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
return;
}
- if (vfio_container_add_section_window(container, section, &err)) {
+ if (vfio_container_add_section_window(bcontainer, section, &err)) {
goto fail;
}
@@ -879,7 +879,7 @@ static void vfio_listener_region_del(MemoryListener *listener,
memory_region_unref(section->mr);
- vfio_container_del_section_window(container, section);
+ vfio_container_del_section_window(&container->bcontainer, section);
}
typedef struct VFIODirtyRanges {
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index e1056dd78e..f2a9a33465 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -48,6 +48,27 @@ int vfio_container_dma_unmap(VFIOContainer *bcontainer,
return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
}
+int vfio_container_add_section_window(VFIOContainer *bcontainer,
+ MemoryRegionSection *section,
+ Error **errp)
+{
+ if (!bcontainer->ops->add_window) {
+ return 0;
+ }
+
+ return bcontainer->ops->add_window(bcontainer, section, errp);
+}
+
+void vfio_container_del_section_window(VFIOContainer *bcontainer,
+ MemoryRegionSection *section)
+{
+ if (!bcontainer->ops->del_window) {
+ return;
+ }
+
+ return bcontainer->ops->del_window(bcontainer, section);
+}
+
void vfio_container_init(VFIOContainer *bcontainer,
VFIOAddressSpace *space,
struct VFIOIOMMUBackendOpsClass *ops)
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index fd2d602fb9..7ca61a7d36 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -215,10 +215,13 @@ static int vfio_legacy_dma_map(VFIOContainer *bcontainer, hwaddr iova,
return -errno;
}
-int vfio_container_add_section_window(VFIOLegacyContainer *container,
- MemoryRegionSection *section,
- Error **errp)
+static int vfio_legacy_add_section_window(VFIOContainer *bcontainer,
+ MemoryRegionSection *section,
+ Error **errp)
{
+ VFIOLegacyContainer *container = container_of(bcontainer,
+ VFIOLegacyContainer,
+ bcontainer);
VFIOHostDMAWindow *hostwin;
hwaddr pgsize = 0;
int ret;
@@ -283,9 +286,13 @@ int vfio_container_add_section_window(VFIOLegacyContainer *container,
return 0;
}
-void vfio_container_del_section_window(VFIOLegacyContainer *container,
- MemoryRegionSection *section)
+static void vfio_legacy_del_section_window(VFIOContainer *bcontainer,
+ MemoryRegionSection *section)
{
+ VFIOLegacyContainer *container = container_of(bcontainer,
+ VFIOLegacyContainer,
+ bcontainer);
+
if (container->iommu_type != VFIO_SPAPR_TCE_v2_IOMMU) {
return;
}
@@ -1174,6 +1181,8 @@ static void vfio_iommu_backend_legacy_ops_class_init(ObjectClass *oc,
ops->dma_map = vfio_legacy_dma_map;
ops->dma_unmap = vfio_legacy_dma_unmap;
+ ops->add_window = vfio_legacy_add_section_window;
+ ops->del_window = vfio_legacy_del_section_window;
}
static const TypeInfo vfio_iommu_backend_legacy_ops_type = {
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 08/27] vfio/container: Move hostwin_list in base container
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (6 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 07/27] vfio/container: switch to IOMMU BE add/del_section_window Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 09/27] vfio/container: Switch to IOMMU BE set_dirty_page_tracking/query_dirty_bitmap API Zhenzhong Duan
` (18 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
Move hostwin_list in base container. This conducts to
passing a base container to vfio_host_win_add/del and
vfio_find_hostwin.
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 12 ++----------
include/hw/vfio/vfio-container-base.h | 8 ++++++++
hw/vfio/common.c | 18 +++++++++---------
hw/vfio/container-base.c | 8 ++++++++
hw/vfio/container.c | 18 +++++-------------
5 files changed, 32 insertions(+), 32 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index b83ae4b3b6..85dbda296a 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -88,7 +88,6 @@ typedef struct VFIOLegacyContainer {
uint64_t max_dirty_bitmap_size;
unsigned long pgsizes;
unsigned int dma_max_mappings;
- QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_HEAD(, VFIOGroup) group_list;
QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
QLIST_HEAD(, VFIODevice) device_list;
@@ -104,13 +103,6 @@ typedef struct VFIORamDiscardListener {
QLIST_ENTRY(VFIORamDiscardListener) next;
} VFIORamDiscardListener;
-typedef struct VFIOHostDMAWindow {
- hwaddr min_iova;
- hwaddr max_iova;
- uint64_t iova_pgsizes;
- QLIST_ENTRY(VFIOHostDMAWindow) hostwin_next;
-} VFIOHostDMAWindow;
-
typedef struct VFIODeviceOps VFIODeviceOps;
typedef struct VFIODevice {
@@ -185,10 +177,10 @@ typedef struct VFIODisplay {
} dmabuf;
} VFIODisplay;
-void vfio_host_win_add(VFIOLegacyContainer *container,
+void vfio_host_win_add(VFIOContainer *bcontainer,
hwaddr min_iova, hwaddr max_iova,
uint64_t iova_pgsizes);
-int vfio_host_win_del(VFIOLegacyContainer *container, hwaddr min_iova,
+int vfio_host_win_del(VFIOContainer *bcontainer, hwaddr min_iova,
hwaddr max_iova);
VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
void vfio_put_address_space(VFIOAddressSpace *space);
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 1f6d5fd229..03bffbff73 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -47,6 +47,13 @@ typedef struct VFIOGuestIOMMU {
QLIST_ENTRY(VFIOGuestIOMMU) giommu_next;
} VFIOGuestIOMMU;
+typedef struct VFIOHostDMAWindow {
+ hwaddr min_iova;
+ hwaddr max_iova;
+ uint64_t iova_pgsizes;
+ QLIST_ENTRY(VFIOHostDMAWindow) hostwin_next;
+} VFIOHostDMAWindow;
+
typedef struct {
unsigned long *bitmap;
hwaddr size;
@@ -60,6 +67,7 @@ struct VFIOContainer {
VFIOIOMMUBackendOpsClass *ops;
VFIOAddressSpace *space;
QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
+ QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_ENTRY(VFIOContainer) next;
};
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 49cb5b6958..511f538c00 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -252,12 +252,12 @@ bool vfio_devices_all_running_and_mig_active(VFIOLegacyContainer *container)
return true;
}
-void vfio_host_win_add(VFIOLegacyContainer *container, hwaddr min_iova,
+void vfio_host_win_add(VFIOContainer *bcontainer, hwaddr min_iova,
hwaddr max_iova, uint64_t iova_pgsizes)
{
VFIOHostDMAWindow *hostwin;
- QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
+ QLIST_FOREACH(hostwin, &bcontainer->hostwin_list, hostwin_next) {
if (ranges_overlap(hostwin->min_iova,
hostwin->max_iova - hostwin->min_iova + 1,
min_iova,
@@ -271,15 +271,15 @@ void vfio_host_win_add(VFIOLegacyContainer *container, hwaddr min_iova,
hostwin->min_iova = min_iova;
hostwin->max_iova = max_iova;
hostwin->iova_pgsizes = iova_pgsizes;
- QLIST_INSERT_HEAD(&container->hostwin_list, hostwin, hostwin_next);
+ QLIST_INSERT_HEAD(&bcontainer->hostwin_list, hostwin, hostwin_next);
}
-int vfio_host_win_del(VFIOLegacyContainer *container,
+int vfio_host_win_del(VFIOContainer *bcontainer,
hwaddr min_iova, hwaddr max_iova)
{
VFIOHostDMAWindow *hostwin;
- QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
+ QLIST_FOREACH(hostwin, &bcontainer->hostwin_list, hostwin_next) {
if (hostwin->min_iova == min_iova && hostwin->max_iova == max_iova) {
QLIST_REMOVE(hostwin, hostwin_next);
g_free(hostwin);
@@ -540,13 +540,13 @@ static void vfio_unregister_ram_discard_listener(VFIOLegacyContainer *container,
g_free(vrdl);
}
-static VFIOHostDMAWindow *vfio_find_hostwin(VFIOLegacyContainer *container,
+static VFIOHostDMAWindow *vfio_find_hostwin(VFIOContainer *bcontainer,
hwaddr iova, hwaddr end)
{
VFIOHostDMAWindow *hostwin;
bool hostwin_found = false;
- QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
+ QLIST_FOREACH(hostwin, &bcontainer->hostwin_list, hostwin_next) {
if (hostwin->min_iova <= iova && end <= hostwin->max_iova) {
hostwin_found = true;
break;
@@ -659,7 +659,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
goto fail;
}
- hostwin = vfio_find_hostwin(container, iova, end);
+ hostwin = vfio_find_hostwin(bcontainer, iova, end);
if (!hostwin) {
error_setg(&err, "Container %p can't map guest IOVA region"
" 0x%"HWADDR_PRIx"..0x%"HWADDR_PRIx, container, iova, end);
@@ -842,7 +842,7 @@ static void vfio_listener_region_del(MemoryListener *listener,
hwaddr pgmask;
VFIOHostDMAWindow *hostwin;
- hostwin = vfio_find_hostwin(container, iova, end);
+ hostwin = vfio_find_hostwin(bcontainer, iova, end);
assert(hostwin); /* or region_add() would have failed */
pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index f2a9a33465..12b256c70e 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -76,11 +76,13 @@ void vfio_container_init(VFIOContainer *bcontainer,
bcontainer->ops = ops;
bcontainer->space = space;
QLIST_INIT(&bcontainer->giommu_list);
+ QLIST_INIT(&bcontainer->hostwin_list);
}
void vfio_container_destroy(VFIOContainer *bcontainer)
{
VFIOGuestIOMMU *giommu, *tmp;
+ VFIOHostDMAWindow *hostwin, *next;
QLIST_REMOVE(bcontainer, next);
@@ -90,6 +92,12 @@ void vfio_container_destroy(VFIOContainer *bcontainer)
QLIST_REMOVE(giommu, giommu_next);
g_free(giommu);
}
+
+ QLIST_FOREACH_SAFE(hostwin, &bcontainer->hostwin_list, hostwin_next,
+ next) {
+ QLIST_REMOVE(hostwin, hostwin_next);
+ g_free(hostwin);
+ }
}
static const TypeInfo vfio_iommu_backend_ops_type_info = {
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 7ca61a7d36..5d111f69c9 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -231,7 +231,7 @@ static int vfio_legacy_add_section_window(VFIOContainer *bcontainer,
}
/* For now intersections are not allowed, we may relax this later */
- QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
+ QLIST_FOREACH(hostwin, &bcontainer->hostwin_list, hostwin_next) {
if (ranges_overlap(hostwin->min_iova,
hostwin->max_iova - hostwin->min_iova + 1,
section->offset_within_address_space,
@@ -253,7 +253,7 @@ static int vfio_legacy_add_section_window(VFIOContainer *bcontainer,
return ret;
}
- vfio_host_win_add(container, section->offset_within_address_space,
+ vfio_host_win_add(bcontainer, section->offset_within_address_space,
section->offset_within_address_space +
int128_get64(section->size) - 1, pgsize);
#ifdef CONFIG_KVM
@@ -299,7 +299,7 @@ static void vfio_legacy_del_section_window(VFIOContainer *bcontainer,
vfio_spapr_remove_window(container,
section->offset_within_address_space);
- if (vfio_host_win_del(container,
+ if (vfio_host_win_del(bcontainer,
section->offset_within_address_space,
section->offset_within_address_space +
int128_get64(section->size) - 1) < 0) {
@@ -636,7 +636,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container->error = NULL;
container->dirty_pages_supported = false;
container->dma_max_mappings = 0;
- QLIST_INIT(&container->hostwin_list);
QLIST_INIT(&container->vrdl_list);
bcontainer = &container->bcontainer;
vfio_container_init(bcontainer, space, ops);
@@ -681,7 +680,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
* information to get the actual window extent rather than assume
* a 64-bit IOVA address space.
*/
- vfio_host_win_add(container, 0, (hwaddr)-1, container->pgsizes);
+ vfio_host_win_add(bcontainer, 0, (hwaddr)-1, container->pgsizes);
break;
}
@@ -746,7 +745,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
} else {
/* The default table uses 4K pages */
container->pgsizes = 0x1000;
- vfio_host_win_add(container, info.dma32_window_start,
+ vfio_host_win_add(bcontainer, info.dma32_window_start,
info.dma32_window_start +
info.dma32_window_size - 1,
0x1000);
@@ -822,16 +821,9 @@ static void vfio_disconnect_container(VFIOGroup *group)
if (QLIST_EMPTY(&container->group_list)) {
VFIOAddressSpace *space = container->bcontainer.space;
- VFIOHostDMAWindow *hostwin, *next;
vfio_container_destroy(bcontainer);
- QLIST_FOREACH_SAFE(hostwin, &container->hostwin_list, hostwin_next,
- next) {
- QLIST_REMOVE(hostwin, hostwin_next);
- g_free(hostwin);
- }
-
trace_vfio_disconnect_container(container->fd);
close(container->fd);
g_free(container);
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 09/27] vfio/container: Switch to IOMMU BE set_dirty_page_tracking/query_dirty_bitmap API
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (7 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 08/27] vfio/container: Move hostwin_list in base container Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 10/27] vfio/container: Move per container device list in base container Zhenzhong Duan
` (17 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
dirty_pages_supported field is also moved to the base container
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 6 ------
include/hw/vfio/vfio-container-base.h | 6 ++++++
hw/vfio/common.c | 12 ++++++++----
hw/vfio/container-base.c | 23 +++++++++++++++++++++++
hw/vfio/container.c | 23 ++++++++++++++++-------
5 files changed, 53 insertions(+), 17 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 85dbda296a..39bcc7ec33 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -83,7 +83,6 @@ typedef struct VFIOLegacyContainer {
unsigned iommu_type;
Error *error;
bool initialized;
- bool dirty_pages_supported;
uint64_t dirty_pgsizes;
uint64_t max_dirty_bitmap_size;
unsigned long pgsizes;
@@ -186,11 +185,6 @@ VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
void vfio_put_address_space(VFIOAddressSpace *space);
bool vfio_devices_all_running_and_saving(VFIOLegacyContainer *container);
-/* container->fd */
-int vfio_set_dirty_page_tracking(VFIOLegacyContainer *container, bool start);
-int vfio_query_dirty_bitmap(VFIOLegacyContainer *container, VFIOBitmap *vbmap,
- hwaddr iova, hwaddr size);
-
void vfio_disable_irqindex(VFIODevice *vbasedev, int index);
void vfio_unmask_single_irqindex(VFIODevice *vbasedev, int index);
void vfio_mask_single_irqindex(VFIODevice *vbasedev, int index);
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 03bffbff73..5ab52774b5 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -66,6 +66,7 @@ typedef struct {
struct VFIOContainer {
VFIOIOMMUBackendOpsClass *ops;
VFIOAddressSpace *space;
+ bool dirty_pages_supported;
QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_ENTRY(VFIOContainer) next;
@@ -77,6 +78,11 @@ int vfio_container_dma_map(VFIOContainer *bcontainer,
int vfio_container_dma_unmap(VFIOContainer *bcontainer,
hwaddr iova, ram_addr_t size,
IOMMUTLBEntry *iotlb);
+int vfio_container_set_dirty_page_tracking(VFIOContainer *bcontainer,
+ bool start);
+int vfio_container_query_dirty_bitmap(VFIOContainer *bcontainer,
+ VFIOBitmap *vbmap,
+ hwaddr iova, hwaddr size);
int vfio_container_add_section_window(VFIOContainer *bcontainer,
MemoryRegionSection *section,
Error **errp);
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 511f538c00..855d6d82d0 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -1149,7 +1149,8 @@ static void vfio_listener_log_global_start(MemoryListener *listener)
if (vfio_devices_all_device_dirty_tracking(container)) {
ret = vfio_devices_dma_logging_start(container);
} else {
- ret = vfio_set_dirty_page_tracking(container, true);
+ ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
+ true);
}
if (ret) {
@@ -1169,7 +1170,8 @@ static void vfio_listener_log_global_stop(MemoryListener *listener)
if (vfio_devices_all_device_dirty_tracking(container)) {
vfio_devices_dma_logging_stop(container);
} else {
- ret = vfio_set_dirty_page_tracking(container, false);
+ ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
+ false);
}
if (ret) {
@@ -1237,7 +1239,8 @@ int vfio_get_dirty_bitmap(VFIOLegacyContainer *container, uint64_t iova,
VFIOBitmap vbmap;
int ret;
- if (!container->dirty_pages_supported && !all_device_dirty_tracking) {
+ if (!container->bcontainer.dirty_pages_supported &&
+ !all_device_dirty_tracking) {
cpu_physical_memory_set_dirty_range(ram_addr, size,
tcg_enabled() ? DIRTY_CLIENTS_ALL :
DIRTY_CLIENTS_NOCODE);
@@ -1252,7 +1255,8 @@ int vfio_get_dirty_bitmap(VFIOLegacyContainer *container, uint64_t iova,
if (all_device_dirty_tracking) {
ret = vfio_devices_query_dirty_bitmap(container, &vbmap, iova, size);
} else {
- ret = vfio_query_dirty_bitmap(container, &vbmap, iova, size);
+ ret = vfio_container_query_dirty_bitmap(&container->bcontainer, &vbmap,
+ iova, size);
}
if (ret) {
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index 12b256c70e..530ad42c0d 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -48,6 +48,28 @@ int vfio_container_dma_unmap(VFIOContainer *bcontainer,
return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
}
+int vfio_container_set_dirty_page_tracking(VFIOContainer *bcontainer,
+ bool start)
+{
+ /* Fallback to all pages dirty if dirty page sync isn't supported */
+ if (!bcontainer->ops->set_dirty_page_tracking) {
+ return 0;
+ }
+
+ return bcontainer->ops->set_dirty_page_tracking(bcontainer, start);
+}
+
+int vfio_container_query_dirty_bitmap(VFIOContainer *bcontainer,
+ VFIOBitmap *vbmap,
+ hwaddr iova, hwaddr size)
+{
+ if (!bcontainer->ops->query_dirty_bitmap) {
+ return -EINVAL;
+ }
+
+ return bcontainer->ops->query_dirty_bitmap(bcontainer, vbmap, iova, size);
+}
+
int vfio_container_add_section_window(VFIOContainer *bcontainer,
MemoryRegionSection *section,
Error **errp)
@@ -75,6 +97,7 @@ void vfio_container_init(VFIOContainer *bcontainer,
{
bcontainer->ops = ops;
bcontainer->space = space;
+ bcontainer->dirty_pages_supported = false;
QLIST_INIT(&bcontainer->giommu_list);
QLIST_INIT(&bcontainer->hostwin_list);
}
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 5d111f69c9..26617afaa9 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -139,7 +139,7 @@ static int vfio_legacy_dma_unmap(VFIOContainer *bcontainer, hwaddr iova,
if (iotlb && vfio_devices_all_running_and_mig_active(container)) {
if (!vfio_devices_all_device_dirty_tracking(container) &&
- container->dirty_pages_supported) {
+ container->bcontainer.dirty_pages_supported) {
return vfio_dma_unmap_bitmap(container, iova, size, iotlb);
}
@@ -308,14 +308,18 @@ static void vfio_legacy_del_section_window(VFIOContainer *bcontainer,
}
}
-int vfio_set_dirty_page_tracking(VFIOLegacyContainer *container, bool start)
+static int vfio_legacy_set_dirty_page_tracking(VFIOContainer *bcontainer,
+ bool start)
{
+ VFIOLegacyContainer *container = container_of(bcontainer,
+ VFIOLegacyContainer,
+ bcontainer);
int ret;
struct vfio_iommu_type1_dirty_bitmap dirty = {
.argsz = sizeof(dirty),
};
- if (!container->dirty_pages_supported) {
+ if (!bcontainer->dirty_pages_supported) {
return 0;
}
@@ -335,9 +339,13 @@ int vfio_set_dirty_page_tracking(VFIOLegacyContainer *container, bool start)
return ret;
}
-int vfio_query_dirty_bitmap(VFIOLegacyContainer *container, VFIOBitmap *vbmap,
- hwaddr iova, hwaddr size)
+static int vfio_legacy_query_dirty_bitmap(VFIOContainer *bcontainer,
+ VFIOBitmap *vbmap,
+ hwaddr iova, hwaddr size)
{
+ VFIOLegacyContainer *container = container_of(bcontainer,
+ VFIOLegacyContainer,
+ bcontainer);
struct vfio_iommu_type1_dirty_bitmap *dbitmap;
struct vfio_iommu_type1_dirty_bitmap_get *range;
int ret;
@@ -546,7 +554,7 @@ static void vfio_get_iommu_info_migration(VFIOLegacyContainer *container,
* qemu_real_host_page_size to mark those dirty.
*/
if (cap_mig->pgsize_bitmap & qemu_real_host_page_size()) {
- container->dirty_pages_supported = true;
+ container->bcontainer.dirty_pages_supported = true;
container->max_dirty_bitmap_size = cap_mig->max_dirty_bitmap_size;
container->dirty_pgsizes = cap_mig->pgsize_bitmap;
}
@@ -634,7 +642,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container = g_malloc0(sizeof(*container));
container->fd = fd;
container->error = NULL;
- container->dirty_pages_supported = false;
container->dma_max_mappings = 0;
QLIST_INIT(&container->vrdl_list);
bcontainer = &container->bcontainer;
@@ -1173,6 +1180,8 @@ static void vfio_iommu_backend_legacy_ops_class_init(ObjectClass *oc,
ops->dma_map = vfio_legacy_dma_map;
ops->dma_unmap = vfio_legacy_dma_unmap;
+ ops->set_dirty_page_tracking = vfio_legacy_set_dirty_page_tracking;
+ ops->query_dirty_bitmap = vfio_legacy_query_dirty_bitmap;
ops->add_window = vfio_legacy_add_section_window;
ops->del_window = vfio_legacy_del_section_window;
}
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 10/27] vfio/container: Move per container device list in base container
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (8 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 09/27] vfio/container: Switch to IOMMU BE set_dirty_page_tracking/query_dirty_bitmap API Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 11/27] vfio/container: Convert functions to " Zhenzhong Duan
` (16 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
VFIO Device is also changed to point to base container instead of
legacy container.
No fucntional change intended.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 3 +--
include/hw/vfio/vfio-container-base.h | 1 +
hw/vfio/common.c | 23 +++++++++++++++--------
hw/vfio/container.c | 12 ++++++------
4 files changed, 23 insertions(+), 16 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 39bcc7ec33..6979359457 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -89,7 +89,6 @@ typedef struct VFIOLegacyContainer {
unsigned int dma_max_mappings;
QLIST_HEAD(, VFIOGroup) group_list;
QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
- QLIST_HEAD(, VFIODevice) device_list;
} VFIOLegacyContainer;
typedef struct VFIORamDiscardListener {
@@ -109,7 +108,7 @@ typedef struct VFIODevice {
QLIST_ENTRY(VFIODevice) container_next;
QLIST_ENTRY(VFIODevice) global_next;
struct VFIOGroup *group;
- VFIOLegacyContainer *container;
+ VFIOContainer *bcontainer;
char *sysfsdev;
char *name;
DeviceState *dev;
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 5ab52774b5..49637a1e6c 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -70,6 +70,7 @@ struct VFIOContainer {
QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_ENTRY(VFIOContainer) next;
+ QLIST_HEAD(, VFIODevice) device_list;
};
int vfio_container_dma_map(VFIOContainer *bcontainer,
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 855d6d82d0..7350af038a 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -152,7 +152,7 @@ void vfio_unblock_multiple_devices_migration(void)
bool vfio_viommu_preset(VFIODevice *vbasedev)
{
- return vbasedev->container->bcontainer.space->as != &address_space_memory;
+ return vbasedev->bcontainer->space->as != &address_space_memory;
}
static void vfio_set_migration_error(int err)
@@ -186,6 +186,7 @@ bool vfio_device_state_is_precopy(VFIODevice *vbasedev)
static bool vfio_devices_all_dirty_tracking(VFIOLegacyContainer *container)
{
+ VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
MigrationState *ms = migrate_get_current();
@@ -194,7 +195,7 @@ static bool vfio_devices_all_dirty_tracking(VFIOLegacyContainer *container)
return false;
}
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
VFIOMigration *migration = vbasedev->migration;
if (!migration) {
@@ -212,9 +213,10 @@ static bool vfio_devices_all_dirty_tracking(VFIOLegacyContainer *container)
bool vfio_devices_all_device_dirty_tracking(VFIOLegacyContainer *container)
{
+ VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
if (!vbasedev->dirty_pages_supported) {
return false;
}
@@ -229,13 +231,14 @@ bool vfio_devices_all_device_dirty_tracking(VFIOLegacyContainer *container)
*/
bool vfio_devices_all_running_and_mig_active(VFIOLegacyContainer *container)
{
+ VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
if (!migration_is_active(migrate_get_current())) {
return false;
}
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
VFIOMigration *migration = vbasedev->migration;
if (!migration) {
@@ -901,12 +904,13 @@ static bool vfio_section_is_vfio_pci(MemoryRegionSection *section,
VFIOLegacyContainer *container)
{
VFIOPCIDevice *pcidev;
+ VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
Object *owner;
owner = memory_region_owner(section->mr);
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
if (vbasedev->type != VFIO_DEVICE_TYPE_PCI) {
continue;
}
@@ -1007,13 +1011,14 @@ static void vfio_devices_dma_logging_stop(VFIOLegacyContainer *container)
uint64_t buf[DIV_ROUND_UP(sizeof(struct vfio_device_feature),
sizeof(uint64_t))] = {};
struct vfio_device_feature *feature = (struct vfio_device_feature *)buf;
+ VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
feature->argsz = sizeof(buf);
feature->flags = VFIO_DEVICE_FEATURE_SET |
VFIO_DEVICE_FEATURE_DMA_LOGGING_STOP;
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
if (!vbasedev->dirty_tracking) {
continue;
}
@@ -1104,6 +1109,7 @@ static int vfio_devices_dma_logging_start(VFIOLegacyContainer *container)
{
struct vfio_device_feature *feature;
VFIODirtyRanges ranges;
+ VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
int ret = 0;
@@ -1114,7 +1120,7 @@ static int vfio_devices_dma_logging_start(VFIOLegacyContainer *container)
return -errno;
}
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
if (vbasedev->dirty_tracking) {
continue;
}
@@ -1211,10 +1217,11 @@ int vfio_devices_query_dirty_bitmap(VFIOLegacyContainer *container,
VFIOBitmap *vbmap, hwaddr iova,
hwaddr size)
{
+ VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
int ret;
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
ret = vfio_device_dma_logging_report(vbasedev, iova, size,
vbmap->bitmap);
if (ret) {
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 26617afaa9..edcdee2904 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -1123,7 +1123,7 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
int groupid = vfio_device_groupid(vbasedev, errp);
VFIODevice *vbasedev_iter;
VFIOGroup *group;
- VFIOLegacyContainer *container;
+ VFIOContainer *bcontainer;
int ret;
if (groupid < 0) {
@@ -1150,9 +1150,9 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
return ret;
}
- container = group->container;
- vbasedev->container = container;
- QLIST_INSERT_HEAD(&container->device_list, vbasedev, container_next);
+ bcontainer = &group->container->bcontainer;
+ vbasedev->bcontainer = bcontainer;
+ QLIST_INSERT_HEAD(&bcontainer->device_list, vbasedev, container_next);
QLIST_INSERT_HEAD(&vfio_device_list, vbasedev, global_next);
return ret;
@@ -1162,13 +1162,13 @@ void vfio_detach_device(VFIODevice *vbasedev)
{
VFIOGroup *group = vbasedev->group;
- if (!vbasedev->container) {
+ if (!vbasedev->bcontainer) {
return;
}
QLIST_REMOVE(vbasedev, global_next);
QLIST_REMOVE(vbasedev, container_next);
- vbasedev->container = NULL;
+ vbasedev->bcontainer = NULL;
trace_vfio_detach_device(vbasedev->name, group->groupid);
vfio_put_base_device(vbasedev);
vfio_put_group(group);
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 11/27] vfio/container: Convert functions to base container
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (9 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 10/27] vfio/container: Move per container device list in base container Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 12/27] vfio/container: Move vrdl_list, pgsizes and dma_max_mappings " Zhenzhong Duan
` (15 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
In the prospect to get rid of VFIOLegacyContainer refs
in common.c lets convert misc functions to use the base
container object instead:
vfio_devices_all_dirty_tracking
vfio_devices_all_device_dirty_tracking
vfio_devices_all_running_and_mig_active
vfio_devices_query_dirty_bitmap
vfio_get_dirty_bitmap
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 9 ++++----
hw/vfio/common.c | 42 +++++++++++++++--------------------
hw/vfio/container.c | 6 ++---
hw/vfio/trace-events | 2 +-
4 files changed, 26 insertions(+), 33 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 6979359457..7bb75bc7cd 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -182,7 +182,6 @@ int vfio_host_win_del(VFIOContainer *bcontainer, hwaddr min_iova,
hwaddr max_iova);
VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
void vfio_put_address_space(VFIOAddressSpace *space);
-bool vfio_devices_all_running_and_saving(VFIOLegacyContainer *container);
void vfio_disable_irqindex(VFIODevice *vbasedev, int index);
void vfio_unmask_single_irqindex(VFIODevice *vbasedev, int index);
@@ -254,11 +253,11 @@ bool vfio_migration_realize(VFIODevice *vbasedev, Error **errp);
void vfio_migration_exit(VFIODevice *vbasedev);
int vfio_bitmap_alloc(VFIOBitmap *vbmap, hwaddr size);
-bool vfio_devices_all_running_and_mig_active(VFIOLegacyContainer *container);
-bool vfio_devices_all_device_dirty_tracking(VFIOLegacyContainer *container);
-int vfio_devices_query_dirty_bitmap(VFIOLegacyContainer *container,
+bool vfio_devices_all_running_and_mig_active(VFIOContainer *bcontainer);
+bool vfio_devices_all_device_dirty_tracking(VFIOContainer *bcontainer);
+int vfio_devices_query_dirty_bitmap(VFIOContainer *bcontainer,
VFIOBitmap *vbmap, hwaddr iova,
hwaddr size);
-int vfio_get_dirty_bitmap(VFIOLegacyContainer *container, uint64_t iova,
+int vfio_get_dirty_bitmap(VFIOContainer *bcontainer, uint64_t iova,
uint64_t size, ram_addr_t ram_addr);
#endif /* HW_VFIO_VFIO_COMMON_H */
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 7350af038a..1c47bcc478 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -184,9 +184,8 @@ bool vfio_device_state_is_precopy(VFIODevice *vbasedev)
migration->device_state == VFIO_DEVICE_STATE_PRE_COPY_P2P;
}
-static bool vfio_devices_all_dirty_tracking(VFIOLegacyContainer *container)
+static bool vfio_devices_all_dirty_tracking(VFIOContainer *bcontainer)
{
- VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
MigrationState *ms = migrate_get_current();
@@ -211,9 +210,8 @@ static bool vfio_devices_all_dirty_tracking(VFIOLegacyContainer *container)
return true;
}
-bool vfio_devices_all_device_dirty_tracking(VFIOLegacyContainer *container)
+bool vfio_devices_all_device_dirty_tracking(VFIOContainer *bcontainer)
{
- VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
@@ -229,9 +227,8 @@ bool vfio_devices_all_device_dirty_tracking(VFIOLegacyContainer *container)
* Check if all VFIO devices are running and migration is active, which is
* essentially equivalent to the migration being in pre-copy phase.
*/
-bool vfio_devices_all_running_and_mig_active(VFIOLegacyContainer *container)
+bool vfio_devices_all_running_and_mig_active(VFIOContainer *bcontainer)
{
- VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
if (!migration_is_active(migrate_get_current())) {
@@ -1152,7 +1149,7 @@ static void vfio_listener_log_global_start(MemoryListener *listener)
listener);
int ret;
- if (vfio_devices_all_device_dirty_tracking(container)) {
+ if (vfio_devices_all_device_dirty_tracking(&container->bcontainer)) {
ret = vfio_devices_dma_logging_start(container);
} else {
ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
@@ -1173,7 +1170,7 @@ static void vfio_listener_log_global_stop(MemoryListener *listener)
listener);
int ret = 0;
- if (vfio_devices_all_device_dirty_tracking(container)) {
+ if (vfio_devices_all_device_dirty_tracking(&container->bcontainer)) {
vfio_devices_dma_logging_stop(container);
} else {
ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
@@ -1213,11 +1210,10 @@ static int vfio_device_dma_logging_report(VFIODevice *vbasedev, hwaddr iova,
return 0;
}
-int vfio_devices_query_dirty_bitmap(VFIOLegacyContainer *container,
+int vfio_devices_query_dirty_bitmap(VFIOContainer *bcontainer,
VFIOBitmap *vbmap, hwaddr iova,
hwaddr size)
{
- VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
int ret;
@@ -1237,17 +1233,16 @@ int vfio_devices_query_dirty_bitmap(VFIOLegacyContainer *container,
return 0;
}
-int vfio_get_dirty_bitmap(VFIOLegacyContainer *container, uint64_t iova,
+int vfio_get_dirty_bitmap(VFIOContainer *bcontainer, uint64_t iova,
uint64_t size, ram_addr_t ram_addr)
{
bool all_device_dirty_tracking =
- vfio_devices_all_device_dirty_tracking(container);
+ vfio_devices_all_device_dirty_tracking(bcontainer);
uint64_t dirty_pages;
VFIOBitmap vbmap;
int ret;
- if (!container->bcontainer.dirty_pages_supported &&
- !all_device_dirty_tracking) {
+ if (!bcontainer->dirty_pages_supported && !all_device_dirty_tracking) {
cpu_physical_memory_set_dirty_range(ram_addr, size,
tcg_enabled() ? DIRTY_CLIENTS_ALL :
DIRTY_CLIENTS_NOCODE);
@@ -1260,10 +1255,9 @@ int vfio_get_dirty_bitmap(VFIOLegacyContainer *container, uint64_t iova,
}
if (all_device_dirty_tracking) {
- ret = vfio_devices_query_dirty_bitmap(container, &vbmap, iova, size);
+ ret = vfio_devices_query_dirty_bitmap(bcontainer, &vbmap, iova, size);
} else {
- ret = vfio_container_query_dirty_bitmap(&container->bcontainer, &vbmap,
- iova, size);
+ ret = vfio_container_query_dirty_bitmap(bcontainer, &vbmap, iova, size);
}
if (ret) {
@@ -1273,8 +1267,7 @@ int vfio_get_dirty_bitmap(VFIOLegacyContainer *container, uint64_t iova,
dirty_pages = cpu_physical_memory_set_dirty_lebitmap(vbmap.bitmap, ram_addr,
vbmap.pages);
- trace_vfio_get_dirty_bitmap(container->fd, iova, size, vbmap.size,
- ram_addr, dirty_pages);
+ trace_vfio_get_dirty_bitmap(iova, size, vbmap.size, ram_addr, dirty_pages);
out:
g_free(vbmap.bitmap);
@@ -1309,8 +1302,8 @@ static void vfio_iommu_map_dirty_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
rcu_read_lock();
if (vfio_get_xlat_addr(iotlb, NULL, &translated_addr, NULL)) {
- ret = vfio_get_dirty_bitmap(container, iova, iotlb->addr_mask + 1,
- translated_addr);
+ ret = vfio_get_dirty_bitmap(&container->bcontainer, iova,
+ iotlb->addr_mask + 1, translated_addr);
if (ret) {
error_report("vfio_iommu_map_dirty_notify(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
@@ -1339,7 +1332,8 @@ static int vfio_ram_discard_get_dirty_bitmap(MemoryRegionSection *section,
* Sync the whole mapped region (spanning multiple individual mappings)
* in one go.
*/
- return vfio_get_dirty_bitmap(vrdl->container, iova, size, ram_addr);
+ return vfio_get_dirty_bitmap(&vrdl->container->bcontainer, iova, size,
+ ram_addr);
}
static int
@@ -1409,7 +1403,7 @@ static int vfio_sync_dirty_bitmap(VFIOLegacyContainer *container,
ram_addr = memory_region_get_ram_addr(section->mr) +
section->offset_within_region;
- return vfio_get_dirty_bitmap(container,
+ return vfio_get_dirty_bitmap(&container->bcontainer,
REAL_HOST_PAGE_ALIGN(section->offset_within_address_space),
int128_get64(section->size), ram_addr);
}
@@ -1426,7 +1420,7 @@ static void vfio_listener_log_sync(MemoryListener *listener,
return;
}
- if (vfio_devices_all_dirty_tracking(container)) {
+ if (vfio_devices_all_dirty_tracking(&container->bcontainer)) {
ret = vfio_sync_dirty_bitmap(container, section);
if (ret) {
error_report("vfio: Failed to sync dirty bitmap, err: %d (%s)", ret,
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index edcdee2904..e278321c0a 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -137,8 +137,8 @@ static int vfio_legacy_dma_unmap(VFIOContainer *bcontainer, hwaddr iova,
bool need_dirty_sync = false;
int ret;
- if (iotlb && vfio_devices_all_running_and_mig_active(container)) {
- if (!vfio_devices_all_device_dirty_tracking(container) &&
+ if (iotlb && vfio_devices_all_running_and_mig_active(bcontainer)) {
+ if (!vfio_devices_all_device_dirty_tracking(bcontainer) &&
container->bcontainer.dirty_pages_supported) {
return vfio_dma_unmap_bitmap(container, iova, size, iotlb);
}
@@ -170,7 +170,7 @@ static int vfio_legacy_dma_unmap(VFIOContainer *bcontainer, hwaddr iova,
}
if (need_dirty_sync) {
- ret = vfio_get_dirty_bitmap(container, iova, size,
+ ret = vfio_get_dirty_bitmap(bcontainer, iova, size,
iotlb->translated_addr);
if (ret) {
return ret;
diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
index 9f7fedee98..08a1f9dfa4 100644
--- a/hw/vfio/trace-events
+++ b/hw/vfio/trace-events
@@ -117,7 +117,7 @@ vfio_region_sparse_mmap_header(const char *name, int index, int nr_areas) "Devic
vfio_region_sparse_mmap_entry(int i, unsigned long start, unsigned long end) "sparse entry %d [0x%lx - 0x%lx]"
vfio_get_dev_region(const char *name, int index, uint32_t type, uint32_t subtype) "%s index %d, %08x/%08x"
vfio_legacy_dma_unmap_overflow_workaround(void) ""
-vfio_get_dirty_bitmap(int fd, uint64_t iova, uint64_t size, uint64_t bitmap_size, uint64_t start, uint64_t dirty_pages) "container fd=%d, iova=0x%"PRIx64" size= 0x%"PRIx64" bitmap_size=0x%"PRIx64" start=0x%"PRIx64" dirty_pages=%"PRIu64
+vfio_get_dirty_bitmap(uint64_t iova, uint64_t size, uint64_t bitmap_size, uint64_t start, uint64_t dirty_pages) "iova=0x%"PRIx64" size= 0x%"PRIx64" bitmap_size=0x%"PRIx64" start=0x%"PRIx64" dirty_pages=%"PRIu64
vfio_iommu_map_dirty_notify(uint64_t iova_start, uint64_t iova_end) "iommu dirty @ 0x%"PRIx64" - 0x%"PRIx64
# platform.c
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 12/27] vfio/container: Move vrdl_list, pgsizes and dma_max_mappings to base container
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (10 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 11/27] vfio/container: Convert functions to " Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 13/27] vfio/container: Move listener " Zhenzhong Duan
` (14 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
From: Eric Auger <eric.auger@redhat.com>
Move vrdl_list, pgsizes and dma_max_mappings to the base
container object
No functional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 13 --------
include/hw/vfio/vfio-container-base.h | 13 ++++++++
hw/vfio/common.c | 48 +++++++++++++--------------
hw/vfio/container-base.c | 12 +++++++
hw/vfio/container.c | 18 +++++-----
hw/vfio/spapr.c | 4 +--
6 files changed, 59 insertions(+), 49 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 7bb75bc7cd..18dd676a2a 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -85,22 +85,9 @@ typedef struct VFIOLegacyContainer {
bool initialized;
uint64_t dirty_pgsizes;
uint64_t max_dirty_bitmap_size;
- unsigned long pgsizes;
- unsigned int dma_max_mappings;
QLIST_HEAD(, VFIOGroup) group_list;
- QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
} VFIOLegacyContainer;
-typedef struct VFIORamDiscardListener {
- VFIOLegacyContainer *container;
- MemoryRegion *mr;
- hwaddr offset_within_address_space;
- hwaddr size;
- uint64_t granularity;
- RamDiscardListener listener;
- QLIST_ENTRY(VFIORamDiscardListener) next;
-} VFIORamDiscardListener;
-
typedef struct VFIODeviceOps VFIODeviceOps;
typedef struct VFIODevice {
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 49637a1e6c..d6ffd7efc4 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -47,6 +47,16 @@ typedef struct VFIOGuestIOMMU {
QLIST_ENTRY(VFIOGuestIOMMU) giommu_next;
} VFIOGuestIOMMU;
+typedef struct VFIORamDiscardListener {
+ VFIOContainer *bcontainer;
+ MemoryRegion *mr;
+ hwaddr offset_within_address_space;
+ hwaddr size;
+ uint64_t granularity;
+ RamDiscardListener listener;
+ QLIST_ENTRY(VFIORamDiscardListener) next;
+} VFIORamDiscardListener;
+
typedef struct VFIOHostDMAWindow {
hwaddr min_iova;
hwaddr max_iova;
@@ -66,9 +76,12 @@ typedef struct {
struct VFIOContainer {
VFIOIOMMUBackendOpsClass *ops;
VFIOAddressSpace *space;
+ unsigned long pgsizes;
+ unsigned int dma_max_mappings;
bool dirty_pages_supported;
QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
+ QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
QLIST_ENTRY(VFIOContainer) next;
QLIST_HEAD(, VFIODevice) device_list;
};
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 1c47bcc478..b833def682 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -396,13 +396,13 @@ static void vfio_ram_discard_notify_discard(RamDiscardListener *rdl,
{
VFIORamDiscardListener *vrdl = container_of(rdl, VFIORamDiscardListener,
listener);
+ VFIOContainer *bcontainer = vrdl->bcontainer;
const hwaddr size = int128_get64(section->size);
const hwaddr iova = section->offset_within_address_space;
int ret;
/* Unmap with a single call. */
- ret = vfio_container_dma_unmap(&vrdl->container->bcontainer,
- iova, size , NULL);
+ ret = vfio_container_dma_unmap(bcontainer, iova, size , NULL);
if (ret) {
error_report("%s: vfio_container_dma_unmap() failed: %s", __func__,
strerror(-ret));
@@ -414,6 +414,7 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
{
VFIORamDiscardListener *vrdl = container_of(rdl, VFIORamDiscardListener,
listener);
+ VFIOContainer *bcontainer = vrdl->bcontainer;
const hwaddr end = section->offset_within_region +
int128_get64(section->size);
hwaddr start, next, iova;
@@ -432,8 +433,8 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
section->offset_within_address_space;
vaddr = memory_region_get_ram_ptr(section->mr) + start;
- ret = vfio_container_dma_map(&vrdl->container->bcontainer, iova,
- next - start, vaddr, section->readonly);
+ ret = vfio_container_dma_map(bcontainer, iova, next - start,
+ vaddr, section->readonly);
if (ret) {
/* Rollback */
vfio_ram_discard_notify_discard(rdl, section);
@@ -443,7 +444,7 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
return 0;
}
-static void vfio_register_ram_discard_listener(VFIOLegacyContainer *container,
+static void vfio_register_ram_discard_listener(VFIOContainer *bcontainer,
MemoryRegionSection *section)
{
RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
@@ -456,7 +457,7 @@ static void vfio_register_ram_discard_listener(VFIOLegacyContainer *container,
g_assert(QEMU_IS_ALIGNED(int128_get64(section->size), TARGET_PAGE_SIZE));
vrdl = g_new0(VFIORamDiscardListener, 1);
- vrdl->container = container;
+ vrdl->bcontainer = bcontainer;
vrdl->mr = section->mr;
vrdl->offset_within_address_space = section->offset_within_address_space;
vrdl->size = int128_get64(section->size);
@@ -464,14 +465,14 @@ static void vfio_register_ram_discard_listener(VFIOLegacyContainer *container,
section->mr);
g_assert(vrdl->granularity && is_power_of_2(vrdl->granularity));
- g_assert(container->pgsizes &&
- vrdl->granularity >= 1ULL << ctz64(container->pgsizes));
+ g_assert(bcontainer->pgsizes &&
+ vrdl->granularity >= 1ULL << ctz64(bcontainer->pgsizes));
ram_discard_listener_init(&vrdl->listener,
vfio_ram_discard_notify_populate,
vfio_ram_discard_notify_discard, true);
ram_discard_manager_register_listener(rdm, &vrdl->listener, section);
- QLIST_INSERT_HEAD(&container->vrdl_list, vrdl, next);
+ QLIST_INSERT_HEAD(&bcontainer->vrdl_list, vrdl, next);
/*
* Sanity-check if we have a theoretically problematic setup where we could
@@ -486,7 +487,7 @@ static void vfio_register_ram_discard_listener(VFIOLegacyContainer *container,
* number of sections in the address space we could have over time,
* also consuming DMA mappings.
*/
- if (container->dma_max_mappings) {
+ if (bcontainer->dma_max_mappings) {
unsigned int vrdl_count = 0, vrdl_mappings = 0, max_memslots = 512;
#ifdef CONFIG_KVM
@@ -495,7 +496,7 @@ static void vfio_register_ram_discard_listener(VFIOLegacyContainer *container,
}
#endif
- QLIST_FOREACH(vrdl, &container->vrdl_list, next) {
+ QLIST_FOREACH(vrdl, &bcontainer->vrdl_list, next) {
hwaddr start, end;
start = QEMU_ALIGN_DOWN(vrdl->offset_within_address_space,
@@ -507,23 +508,23 @@ static void vfio_register_ram_discard_listener(VFIOLegacyContainer *container,
}
if (vrdl_mappings + max_memslots - vrdl_count >
- container->dma_max_mappings) {
+ bcontainer->dma_max_mappings) {
warn_report("%s: possibly running out of DMA mappings. E.g., try"
" increasing the 'block-size' of virtio-mem devies."
" Maximum possible DMA mappings: %d, Maximum possible"
- " memslots: %d", __func__, container->dma_max_mappings,
+ " memslots: %d", __func__, bcontainer->dma_max_mappings,
max_memslots);
}
}
}
-static void vfio_unregister_ram_discard_listener(VFIOLegacyContainer *container,
+static void vfio_unregister_ram_discard_listener(VFIOContainer *bcontainer,
MemoryRegionSection *section)
{
RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
VFIORamDiscardListener *vrdl = NULL;
- QLIST_FOREACH(vrdl, &container->vrdl_list, next) {
+ QLIST_FOREACH(vrdl, &bcontainer->vrdl_list, next) {
if (vrdl->mr == section->mr &&
vrdl->offset_within_address_space ==
section->offset_within_address_space) {
@@ -697,7 +698,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
iommu_idx);
ret = memory_region_iommu_set_page_size_mask(giommu->iommu_mr,
- container->pgsizes,
+ bcontainer->pgsizes,
&err);
if (ret) {
g_free(giommu);
@@ -724,7 +725,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
* about changes.
*/
if (memory_region_has_ram_discard_manager(section->mr)) {
- vfio_register_ram_discard_listener(container, section);
+ vfio_register_ram_discard_listener(bcontainer, section);
return;
}
@@ -848,7 +849,7 @@ static void vfio_listener_region_del(MemoryListener *listener,
pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
try_unmap = !((iova & pgmask) || (int128_get64(llsize) & pgmask));
} else if (memory_region_has_ram_discard_manager(section->mr)) {
- vfio_unregister_ram_discard_listener(container, section);
+ vfio_unregister_ram_discard_listener(bcontainer, section);
/* Unregistering will trigger an unmap. */
try_unmap = false;
}
@@ -1332,18 +1333,17 @@ static int vfio_ram_discard_get_dirty_bitmap(MemoryRegionSection *section,
* Sync the whole mapped region (spanning multiple individual mappings)
* in one go.
*/
- return vfio_get_dirty_bitmap(&vrdl->container->bcontainer, iova, size,
- ram_addr);
+ return vfio_get_dirty_bitmap(vrdl->bcontainer, iova, size, ram_addr);
}
static int
-vfio_sync_ram_discard_listener_dirty_bitmap(VFIOLegacyContainer *container,
- MemoryRegionSection *section)
+vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainer *bcontainer,
+ MemoryRegionSection *section)
{
RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
VFIORamDiscardListener *vrdl = NULL;
- QLIST_FOREACH(vrdl, &container->vrdl_list, next) {
+ QLIST_FOREACH(vrdl, &bcontainer->vrdl_list, next) {
if (vrdl->mr == section->mr &&
vrdl->offset_within_address_space ==
section->offset_within_address_space) {
@@ -1397,7 +1397,7 @@ static int vfio_sync_dirty_bitmap(VFIOLegacyContainer *container,
}
return 0;
} else if (memory_region_has_ram_discard_manager(section->mr)) {
- return vfio_sync_ram_discard_listener_dirty_bitmap(container, section);
+ return vfio_sync_ram_discard_listener_dirty_bitmap(bcontainer, section);
}
ram_addr = memory_region_get_ram_addr(section->mr) +
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index 530ad42c0d..c5a4c5afed 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -98,17 +98,29 @@ void vfio_container_init(VFIOContainer *bcontainer,
bcontainer->ops = ops;
bcontainer->space = space;
bcontainer->dirty_pages_supported = false;
+ bcontainer->dma_max_mappings = 0;
QLIST_INIT(&bcontainer->giommu_list);
QLIST_INIT(&bcontainer->hostwin_list);
+ QLIST_INIT(&bcontainer->vrdl_list);
}
void vfio_container_destroy(VFIOContainer *bcontainer)
{
+ VFIORamDiscardListener *vrdl, *vrdl_tmp;
VFIOGuestIOMMU *giommu, *tmp;
VFIOHostDMAWindow *hostwin, *next;
QLIST_REMOVE(bcontainer, next);
+ QLIST_FOREACH_SAFE(vrdl, &bcontainer->vrdl_list, next, vrdl_tmp) {
+ RamDiscardManager *rdm;
+
+ rdm = memory_region_get_ram_discard_manager(vrdl->mr);
+ ram_discard_manager_unregister_listener(rdm, &vrdl->listener);
+ QLIST_REMOVE(vrdl, next);
+ g_free(vrdl);
+ }
+
QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next, tmp) {
memory_region_unregister_iommu_notifier(
MEMORY_REGION(giommu->iommu_mr), &giommu->n);
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index e278321c0a..66fad1c280 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -162,7 +162,7 @@ static int vfio_legacy_dma_unmap(VFIOContainer *bcontainer, hwaddr iova,
if (errno == EINVAL && unmap.size && !(unmap.iova + unmap.size) &&
container->iommu_type == VFIO_TYPE1v2_IOMMU) {
trace_vfio_legacy_dma_unmap_overflow_workaround();
- unmap.size -= 1ULL << ctz64(container->pgsizes);
+ unmap.size -= 1ULL << ctz64(container->bcontainer.pgsizes);
continue;
}
error_report("VFIO_UNMAP_DMA failed: %s", strerror(errno));
@@ -642,8 +642,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container = g_malloc0(sizeof(*container));
container->fd = fd;
container->error = NULL;
- container->dma_max_mappings = 0;
- QLIST_INIT(&container->vrdl_list);
bcontainer = &container->bcontainer;
vfio_container_init(bcontainer, space, ops);
@@ -671,13 +669,13 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
}
if (info->flags & VFIO_IOMMU_INFO_PGSIZES) {
- container->pgsizes = info->iova_pgsizes;
+ container->bcontainer.pgsizes = info->iova_pgsizes;
} else {
- container->pgsizes = qemu_real_host_page_size();
+ container->bcontainer.pgsizes = qemu_real_host_page_size();
}
- if (!vfio_get_info_dma_avail(info, &container->dma_max_mappings)) {
- container->dma_max_mappings = 65535;
+ if (!vfio_get_info_dma_avail(info, &bcontainer->dma_max_mappings)) {
+ container->bcontainer.dma_max_mappings = 65535;
}
vfio_get_iommu_info_migration(container, info);
g_free(info);
@@ -687,7 +685,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
* information to get the actual window extent rather than assume
* a 64-bit IOVA address space.
*/
- vfio_host_win_add(bcontainer, 0, (hwaddr)-1, container->pgsizes);
+ vfio_host_win_add(bcontainer, 0, (hwaddr)-1, bcontainer->pgsizes);
break;
}
@@ -736,7 +734,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
}
if (v2) {
- container->pgsizes = info.ddw.pgsizes;
+ container->bcontainer.pgsizes = info.ddw.pgsizes;
/*
* There is a default window in just created container.
* To make region_add/del simpler, we better remove this
@@ -751,7 +749,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
}
} else {
/* The default table uses 4K pages */
- container->pgsizes = 0x1000;
+ container->bcontainer.pgsizes = 0x1000;
vfio_host_win_add(bcontainer, info.dma32_window_start,
info.dma32_window_start +
info.dma32_window_size - 1,
diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
index 683252c506..3fdad9d227 100644
--- a/hw/vfio/spapr.c
+++ b/hw/vfio/spapr.c
@@ -159,13 +159,13 @@ int vfio_spapr_create_window(VFIOLegacyContainer *container,
if (pagesize > rampagesize) {
pagesize = rampagesize;
}
- pgmask = container->pgsizes & (pagesize | (pagesize - 1));
+ pgmask = container->bcontainer.pgsizes & (pagesize | (pagesize - 1));
pagesize = pgmask ? (1ULL << (63 - clz64(pgmask))) : 0;
if (!pagesize) {
error_report("Host doesn't support page size 0x%"PRIx64
", the supported mask is 0x%lx",
memory_region_iommu_get_min_page_size(iommu_mr),
- container->pgsizes);
+ container->bcontainer.pgsizes);
return -EINVAL;
}
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 13/27] vfio/container: Move listener to base container
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (11 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 12/27] vfio/container: Move vrdl_list, pgsizes and dma_max_mappings " Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 14/27] vfio/container: Move dirty_pgsizes and max_dirty_bitmap_size " Zhenzhong Duan
` (13 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
From: Eric Auger <eric.auger@redhat.com>
Move listener to base container. Also error and initialized fields
are moved at the same time.
No functional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 3 -
include/hw/vfio/vfio-container-base.h | 3 +
hw/vfio/common.c | 116 +++++++++++---------------
hw/vfio/container-base.c | 1 +
hw/vfio/container.c | 31 +++----
hw/vfio/spapr.c | 7 +-
6 files changed, 72 insertions(+), 89 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 18dd676a2a..8771160849 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -78,11 +78,8 @@ struct VFIOGroup;
typedef struct VFIOLegacyContainer {
VFIOContainer bcontainer;
int fd; /* /dev/vfio/vfio, empowered by the attached groups */
- MemoryListener listener;
MemoryListener prereg_listener;
unsigned iommu_type;
- Error *error;
- bool initialized;
uint64_t dirty_pgsizes;
uint64_t max_dirty_bitmap_size;
QLIST_HEAD(, VFIOGroup) group_list;
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index d6ffd7efc4..96d33495c1 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -76,6 +76,9 @@ typedef struct {
struct VFIOContainer {
VFIOIOMMUBackendOpsClass *ops;
VFIOAddressSpace *space;
+ MemoryListener listener;
+ Error *error;
+ bool initialized;
unsigned long pgsizes;
unsigned int dma_max_mappings;
bool dirty_pages_supported;
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index b833def682..da1d64efca 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -602,7 +602,7 @@ static bool vfio_listener_valid_section(MemoryRegionSection *section,
return true;
}
-static bool vfio_get_section_iova_range(VFIOLegacyContainer *container,
+static bool vfio_get_section_iova_range(VFIOContainer *bcontainer,
MemoryRegionSection *section,
hwaddr *out_iova, hwaddr *out_end,
Int128 *out_llend)
@@ -630,10 +630,7 @@ static bool vfio_get_section_iova_range(VFIOLegacyContainer *container,
static void vfio_listener_region_add(MemoryListener *listener,
MemoryRegionSection *section)
{
- VFIOLegacyContainer *container = container_of(listener,
- VFIOLegacyContainer,
- listener);
- VFIOContainer *bcontainer = &container->bcontainer;
+ VFIOContainer *bcontainer = container_of(listener, VFIOContainer, listener);
hwaddr iova, end;
Int128 llend, llsize;
void *vaddr;
@@ -645,7 +642,8 @@ static void vfio_listener_region_add(MemoryListener *listener,
return;
}
- if (!vfio_get_section_iova_range(container, section, &iova, &end, &llend)) {
+ if (!vfio_get_section_iova_range(bcontainer, section, &iova, &end,
+ &llend)) {
if (memory_region_is_ram_device(section->mr)) {
trace_vfio_listener_region_add_no_dma_map(
memory_region_name(section->mr),
@@ -663,7 +661,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
hostwin = vfio_find_hostwin(bcontainer, iova, end);
if (!hostwin) {
error_setg(&err, "Container %p can't map guest IOVA region"
- " 0x%"HWADDR_PRIx"..0x%"HWADDR_PRIx, container, iova, end);
+ " 0x%"HWADDR_PRIx"..0x%"HWADDR_PRIx, bcontainer, iova, end);
goto fail;
}
@@ -750,13 +748,12 @@ static void vfio_listener_region_add(MemoryListener *listener,
}
}
- ret = vfio_container_dma_map(&container->bcontainer,
- iova, int128_get64(llsize), vaddr,
- section->readonly);
+ ret = vfio_container_dma_map(bcontainer, iova, int128_get64(llsize),
+ vaddr, section->readonly);
if (ret) {
error_setg(&err, "vfio_container_dma_map(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx", %p) = %d (%s)",
- container, iova, int128_get64(llsize), vaddr, ret,
+ bcontainer, iova, int128_get64(llsize), vaddr, ret,
strerror(-ret));
if (memory_region_is_ram_device(section->mr)) {
/* Allow unexpected mappings not to be fatal for RAM devices */
@@ -778,9 +775,9 @@ fail:
* can gracefully fail. Runtime, there's not much we can do other
* than throw a hardware error.
*/
- if (!container->initialized) {
- if (!container->error) {
- error_propagate_prepend(&container->error, err,
+ if (!bcontainer->initialized) {
+ if (!bcontainer->error) {
+ error_propagate_prepend(&bcontainer->error, err,
"Region %s: ",
memory_region_name(section->mr));
} else {
@@ -795,10 +792,7 @@ fail:
static void vfio_listener_region_del(MemoryListener *listener,
MemoryRegionSection *section)
{
- VFIOLegacyContainer *container = container_of(listener,
- VFIOLegacyContainer,
- listener);
- VFIOContainer *bcontainer = &container->bcontainer;
+ VFIOContainer *bcontainer = container_of(listener, VFIOContainer, listener);
hwaddr iova, end;
Int128 llend, llsize;
int ret;
@@ -831,7 +825,8 @@ static void vfio_listener_region_del(MemoryListener *listener,
*/
}
- if (!vfio_get_section_iova_range(container, section, &iova, &end, &llend)) {
+ if (!vfio_get_section_iova_range(bcontainer, section, &iova, &end,
+ &llend)) {
return;
}
@@ -858,29 +853,29 @@ static void vfio_listener_region_del(MemoryListener *listener,
if (int128_eq(llsize, int128_2_64())) {
/* The unmap ioctl doesn't accept a full 64-bit span. */
llsize = int128_rshift(llsize, 1);
- ret = vfio_container_dma_unmap(&container->bcontainer, iova,
+ ret = vfio_container_dma_unmap(bcontainer, iova,
int128_get64(llsize), NULL);
if (ret) {
error_report("vfio_container_dma_unmap(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
- container, iova, int128_get64(llsize), ret,
+ bcontainer, iova, int128_get64(llsize), ret,
strerror(-ret));
}
iova += int128_get64(llsize);
}
- ret = vfio_container_dma_unmap(&container->bcontainer, iova,
+ ret = vfio_container_dma_unmap(bcontainer, iova,
int128_get64(llsize), NULL);
if (ret) {
error_report("vfio_container_dma_unmap(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
- container, iova, int128_get64(llsize), ret,
+ bcontainer, iova, int128_get64(llsize), ret,
strerror(-ret));
}
}
memory_region_unref(section->mr);
- vfio_container_del_section_window(&container->bcontainer, section);
+ vfio_container_del_section_window(bcontainer, section);
}
typedef struct VFIODirtyRanges {
@@ -893,16 +888,15 @@ typedef struct VFIODirtyRanges {
} VFIODirtyRanges;
typedef struct VFIODirtyRangesListener {
- VFIOLegacyContainer *container;
+ VFIOContainer *bcontainer;
VFIODirtyRanges ranges;
MemoryListener listener;
} VFIODirtyRangesListener;
static bool vfio_section_is_vfio_pci(MemoryRegionSection *section,
- VFIOLegacyContainer *container)
+ VFIOContainer *bcontainer)
{
VFIOPCIDevice *pcidev;
- VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
Object *owner;
@@ -931,7 +925,7 @@ static void vfio_dirty_tracking_update(MemoryListener *listener,
hwaddr iova, end, *min, *max;
if (!vfio_listener_valid_section(section, "tracking_update") ||
- !vfio_get_section_iova_range(dirty->container, section,
+ !vfio_get_section_iova_range(dirty->bcontainer, section,
&iova, &end, NULL)) {
return;
}
@@ -955,7 +949,7 @@ static void vfio_dirty_tracking_update(MemoryListener *listener,
* The alternative would be an IOVATree but that has a much bigger runtime
* overhead and unnecessary complexity.
*/
- if (vfio_section_is_vfio_pci(section, dirty->container) &&
+ if (vfio_section_is_vfio_pci(section, dirty->bcontainer) &&
iova >= UINT32_MAX) {
min = &range->minpci64;
max = &range->maxpci64;
@@ -979,7 +973,7 @@ static const MemoryListener vfio_dirty_tracking_listener = {
.region_add = vfio_dirty_tracking_update,
};
-static void vfio_dirty_tracking_init(VFIOLegacyContainer *container,
+static void vfio_dirty_tracking_init(VFIOContainer *bcontainer,
VFIODirtyRanges *ranges)
{
VFIODirtyRangesListener dirty;
@@ -989,10 +983,10 @@ static void vfio_dirty_tracking_init(VFIOLegacyContainer *container,
dirty.ranges.min64 = UINT64_MAX;
dirty.ranges.minpci64 = UINT64_MAX;
dirty.listener = vfio_dirty_tracking_listener;
- dirty.container = container;
+ dirty.bcontainer = bcontainer;
memory_listener_register(&dirty.listener,
- container->bcontainer.space->as);
+ bcontainer->space->as);
*ranges = dirty.ranges;
@@ -1004,12 +998,11 @@ static void vfio_dirty_tracking_init(VFIOLegacyContainer *container,
memory_listener_unregister(&dirty.listener);
}
-static void vfio_devices_dma_logging_stop(VFIOLegacyContainer *container)
+static void vfio_devices_dma_logging_stop(VFIOContainer *bcontainer)
{
uint64_t buf[DIV_ROUND_UP(sizeof(struct vfio_device_feature),
sizeof(uint64_t))] = {};
struct vfio_device_feature *feature = (struct vfio_device_feature *)buf;
- VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
feature->argsz = sizeof(buf);
@@ -1030,7 +1023,7 @@ static void vfio_devices_dma_logging_stop(VFIOLegacyContainer *container)
}
static struct vfio_device_feature *
-vfio_device_feature_dma_logging_start_create(VFIOLegacyContainer *container,
+vfio_device_feature_dma_logging_start_create(VFIOContainer *bcontainer,
VFIODirtyRanges *tracking)
{
struct vfio_device_feature *feature;
@@ -1103,16 +1096,15 @@ static void vfio_device_feature_dma_logging_start_destroy(
g_free(feature);
}
-static int vfio_devices_dma_logging_start(VFIOLegacyContainer *container)
+static int vfio_devices_dma_logging_start(VFIOContainer *bcontainer)
{
struct vfio_device_feature *feature;
VFIODirtyRanges ranges;
- VFIOContainer *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
int ret = 0;
- vfio_dirty_tracking_init(container, &ranges);
- feature = vfio_device_feature_dma_logging_start_create(container,
+ vfio_dirty_tracking_init(bcontainer, &ranges);
+ feature = vfio_device_feature_dma_logging_start_create(bcontainer,
&ranges);
if (!feature) {
return -errno;
@@ -1135,7 +1127,7 @@ static int vfio_devices_dma_logging_start(VFIOLegacyContainer *container)
out:
if (ret) {
- vfio_devices_dma_logging_stop(container);
+ vfio_devices_dma_logging_stop(bcontainer);
}
vfio_device_feature_dma_logging_start_destroy(feature);
@@ -1145,16 +1137,13 @@ out:
static void vfio_listener_log_global_start(MemoryListener *listener)
{
- VFIOLegacyContainer *container = container_of(listener,
- VFIOLegacyContainer,
- listener);
+ VFIOContainer *bcontainer = container_of(listener, VFIOContainer, listener);
int ret;
- if (vfio_devices_all_device_dirty_tracking(&container->bcontainer)) {
- ret = vfio_devices_dma_logging_start(container);
+ if (vfio_devices_all_device_dirty_tracking(bcontainer)) {
+ ret = vfio_devices_dma_logging_start(bcontainer);
} else {
- ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
- true);
+ ret = vfio_container_set_dirty_page_tracking(bcontainer, true);
}
if (ret) {
@@ -1166,16 +1155,13 @@ static void vfio_listener_log_global_start(MemoryListener *listener)
static void vfio_listener_log_global_stop(MemoryListener *listener)
{
- VFIOLegacyContainer *container = container_of(listener,
- VFIOLegacyContainer,
- listener);
+ VFIOContainer *bcontainer = container_of(listener, VFIOContainer, listener);
int ret = 0;
- if (vfio_devices_all_device_dirty_tracking(&container->bcontainer)) {
- vfio_devices_dma_logging_stop(container);
+ if (vfio_devices_all_device_dirty_tracking(bcontainer)) {
+ vfio_devices_dma_logging_stop(bcontainer);
} else {
- ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
- false);
+ ret = vfio_container_set_dirty_page_tracking(bcontainer, false);
}
if (ret) {
@@ -1286,9 +1272,6 @@ static void vfio_iommu_map_dirty_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
vfio_giommu_dirty_notifier, n);
VFIOGuestIOMMU *giommu = gdn->giommu;
VFIOContainer *bcontainer = giommu->bcontainer;
- VFIOLegacyContainer *container = container_of(bcontainer,
- VFIOLegacyContainer,
- bcontainer);
hwaddr iova = iotlb->iova + giommu->iommu_offset;
ram_addr_t translated_addr;
int ret = -EINVAL;
@@ -1303,12 +1286,12 @@ static void vfio_iommu_map_dirty_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
rcu_read_lock();
if (vfio_get_xlat_addr(iotlb, NULL, &translated_addr, NULL)) {
- ret = vfio_get_dirty_bitmap(&container->bcontainer, iova,
- iotlb->addr_mask + 1, translated_addr);
+ ret = vfio_get_dirty_bitmap(bcontainer, iova, iotlb->addr_mask + 1,
+ translated_addr);
if (ret) {
error_report("vfio_iommu_map_dirty_notify(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
- container, iova, iotlb->addr_mask + 1, ret,
+ bcontainer, iova, iotlb->addr_mask + 1, ret,
strerror(-ret));
}
}
@@ -1364,10 +1347,9 @@ vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainer *bcontainer,
&vrdl);
}
-static int vfio_sync_dirty_bitmap(VFIOLegacyContainer *container,
+static int vfio_sync_dirty_bitmap(VFIOContainer *bcontainer,
MemoryRegionSection *section)
{
- VFIOContainer *bcontainer = &container->bcontainer;
ram_addr_t ram_addr;
if (memory_region_is_iommu(section->mr)) {
@@ -1403,7 +1385,7 @@ static int vfio_sync_dirty_bitmap(VFIOLegacyContainer *container,
ram_addr = memory_region_get_ram_addr(section->mr) +
section->offset_within_region;
- return vfio_get_dirty_bitmap(&container->bcontainer,
+ return vfio_get_dirty_bitmap(bcontainer,
REAL_HOST_PAGE_ALIGN(section->offset_within_address_space),
int128_get64(section->size), ram_addr);
}
@@ -1411,17 +1393,15 @@ static int vfio_sync_dirty_bitmap(VFIOLegacyContainer *container,
static void vfio_listener_log_sync(MemoryListener *listener,
MemoryRegionSection *section)
{
- VFIOLegacyContainer *container = container_of(listener,
- VFIOLegacyContainer,
- listener);
+ VFIOContainer *bcontainer = container_of(listener, VFIOContainer, listener);
int ret;
if (vfio_listener_skipped_section(section)) {
return;
}
- if (vfio_devices_all_dirty_tracking(&container->bcontainer)) {
- ret = vfio_sync_dirty_bitmap(container, section);
+ if (vfio_devices_all_dirty_tracking(bcontainer)) {
+ ret = vfio_sync_dirty_bitmap(bcontainer, section);
if (ret) {
error_report("vfio: Failed to sync dirty bitmap, err: %d (%s)", ret,
strerror(-ret));
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index c5a4c5afed..29cf954019 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -97,6 +97,7 @@ void vfio_container_init(VFIOContainer *bcontainer,
{
bcontainer->ops = ops;
bcontainer->space = space;
+ bcontainer->error = NULL;
bcontainer->dirty_pages_supported = false;
bcontainer->dma_max_mappings = 0;
QLIST_INIT(&bcontainer->giommu_list);
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 66fad1c280..5b14a9b307 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -382,7 +382,9 @@ static int vfio_legacy_query_dirty_bitmap(VFIOContainer *bcontainer,
static void vfio_listener_release(VFIOLegacyContainer *container)
{
- memory_listener_unregister(&container->listener);
+ VFIOContainer *bcontainer = &container->bcontainer;
+
+ memory_listener_unregister(&bcontainer->listener);
if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
memory_listener_unregister(&container->prereg_listener);
}
@@ -540,6 +542,7 @@ static void vfio_get_iommu_info_migration(VFIOLegacyContainer *container,
{
struct vfio_info_cap_header *hdr;
struct vfio_iommu_type1_info_cap_migration *cap_mig;
+ VFIOContainer *bcontainer = &container->bcontainer;
hdr = vfio_get_iommu_info_cap(info, VFIO_IOMMU_TYPE1_INFO_CAP_MIGRATION);
if (!hdr) {
@@ -554,7 +557,7 @@ static void vfio_get_iommu_info_migration(VFIOLegacyContainer *container,
* qemu_real_host_page_size to mark those dirty.
*/
if (cap_mig->pgsize_bitmap & qemu_real_host_page_size()) {
- container->bcontainer.dirty_pages_supported = true;
+ bcontainer->dirty_pages_supported = true;
container->max_dirty_bitmap_size = cap_mig->max_dirty_bitmap_size;
container->dirty_pgsizes = cap_mig->pgsize_bitmap;
}
@@ -641,7 +644,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container = g_malloc0(sizeof(*container));
container->fd = fd;
- container->error = NULL;
bcontainer = &container->bcontainer;
vfio_container_init(bcontainer, space, ops);
@@ -669,9 +671,9 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
}
if (info->flags & VFIO_IOMMU_INFO_PGSIZES) {
- container->bcontainer.pgsizes = info->iova_pgsizes;
+ bcontainer->pgsizes = info->iova_pgsizes;
} else {
- container->bcontainer.pgsizes = qemu_real_host_page_size();
+ bcontainer->pgsizes = qemu_real_host_page_size();
}
if (!vfio_get_info_dma_avail(info, &bcontainer->dma_max_mappings)) {
@@ -712,10 +714,10 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
memory_listener_register(&container->prereg_listener,
&address_space_memory);
- if (container->error) {
+ if (bcontainer->error) {
memory_listener_unregister(&container->prereg_listener);
ret = -1;
- error_propagate_prepend(errp, container->error,
+ error_propagate_prepend(errp, bcontainer->error,
"RAM memory listener initialization failed: ");
goto enable_discards_exit;
}
@@ -734,7 +736,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
}
if (v2) {
- container->bcontainer.pgsizes = info.ddw.pgsizes;
+ bcontainer->pgsizes = info.ddw.pgsizes;
/*
* There is a default window in just created container.
* To make region_add/del simpler, we better remove this
@@ -749,7 +751,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
}
} else {
/* The default table uses 4K pages */
- container->bcontainer.pgsizes = 0x1000;
+ bcontainer->pgsizes = 0x1000;
vfio_host_win_add(bcontainer, info.dma32_window_start,
info.dma32_window_start +
info.dma32_window_size - 1,
@@ -766,19 +768,18 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
group->container = container;
QLIST_INSERT_HEAD(&container->group_list, group, container_next);
- container->listener = vfio_memory_listener;
+ bcontainer->listener = vfio_memory_listener;
- memory_listener_register(&container->listener,
- container->bcontainer.space->as);
+ memory_listener_register(&bcontainer->listener, bcontainer->space->as);
- if (container->error) {
+ if (bcontainer->error) {
ret = -1;
- error_propagate_prepend(errp, container->error,
+ error_propagate_prepend(errp, bcontainer->error,
"memory listener initialization failed: ");
goto listener_release_exit;
}
- container->initialized = true;
+ bcontainer->initialized = true;
return 0;
listener_release_exit:
diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
index 3fdad9d227..7df2a7a672 100644
--- a/hw/vfio/spapr.c
+++ b/hw/vfio/spapr.c
@@ -41,6 +41,7 @@ static void vfio_prereg_listener_region_add(MemoryListener *listener,
{
VFIOLegacyContainer *container = container_of(listener, VFIOLegacyContainer,
prereg_listener);
+ VFIOContainer *bcontainer = &container->bcontainer;
const hwaddr gpa = section->offset_within_address_space;
hwaddr end;
int ret;
@@ -83,9 +84,9 @@ static void vfio_prereg_listener_region_add(MemoryListener *listener,
* can gracefully fail. Runtime, there's not much we can do other
* than throw a hardware error.
*/
- if (!container->initialized) {
- if (!container->error) {
- error_setg_errno(&container->error, -ret,
+ if (!bcontainer->initialized) {
+ if (!bcontainer->error) {
+ error_setg_errno(&bcontainer->error, -ret,
"Memory registering failed");
}
} else {
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 14/27] vfio/container: Move dirty_pgsizes and max_dirty_bitmap_size to base container
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (12 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 13/27] vfio/container: Move listener " Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 15/27] vfio/container: Implement attach/detach_device Zhenzhong Duan
` (12 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
No functional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 2 --
include/hw/vfio/vfio-container-base.h | 2 ++
hw/vfio/container.c | 11 ++++++-----
3 files changed, 8 insertions(+), 7 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 8771160849..9f2b86581b 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -80,8 +80,6 @@ typedef struct VFIOLegacyContainer {
int fd; /* /dev/vfio/vfio, empowered by the attached groups */
MemoryListener prereg_listener;
unsigned iommu_type;
- uint64_t dirty_pgsizes;
- uint64_t max_dirty_bitmap_size;
QLIST_HEAD(, VFIOGroup) group_list;
} VFIOLegacyContainer;
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 96d33495c1..9a5971a00a 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -79,6 +79,8 @@ struct VFIOContainer {
MemoryListener listener;
Error *error;
bool initialized;
+ uint64_t dirty_pgsizes;
+ uint64_t max_dirty_bitmap_size;
unsigned long pgsizes;
unsigned int dma_max_mappings;
bool dirty_pages_supported;
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 5b14a9b307..9d5be749c7 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -70,6 +70,7 @@ static int vfio_dma_unmap_bitmap(VFIOLegacyContainer *container,
hwaddr iova, ram_addr_t size,
IOMMUTLBEntry *iotlb)
{
+ VFIOContainer *bcontainer = &container->bcontainer;
struct vfio_iommu_type1_dma_unmap *unmap;
struct vfio_bitmap *bitmap;
VFIOBitmap vbmap;
@@ -97,7 +98,7 @@ static int vfio_dma_unmap_bitmap(VFIOLegacyContainer *container,
bitmap->size = vbmap.size;
bitmap->data = (__u64 *)vbmap.bitmap;
- if (vbmap.size > container->max_dirty_bitmap_size) {
+ if (vbmap.size > bcontainer->max_dirty_bitmap_size) {
error_report("UNMAP: Size of bitmap too big 0x%"PRIx64, vbmap.size);
ret = -E2BIG;
goto unmap_exit;
@@ -139,7 +140,7 @@ static int vfio_legacy_dma_unmap(VFIOContainer *bcontainer, hwaddr iova,
if (iotlb && vfio_devices_all_running_and_mig_active(bcontainer)) {
if (!vfio_devices_all_device_dirty_tracking(bcontainer) &&
- container->bcontainer.dirty_pages_supported) {
+ bcontainer->dirty_pages_supported) {
return vfio_dma_unmap_bitmap(container, iova, size, iotlb);
}
@@ -162,7 +163,7 @@ static int vfio_legacy_dma_unmap(VFIOContainer *bcontainer, hwaddr iova,
if (errno == EINVAL && unmap.size && !(unmap.iova + unmap.size) &&
container->iommu_type == VFIO_TYPE1v2_IOMMU) {
trace_vfio_legacy_dma_unmap_overflow_workaround();
- unmap.size -= 1ULL << ctz64(container->bcontainer.pgsizes);
+ unmap.size -= 1ULL << ctz64(bcontainer->pgsizes);
continue;
}
error_report("VFIO_UNMAP_DMA failed: %s", strerror(errno));
@@ -558,8 +559,8 @@ static void vfio_get_iommu_info_migration(VFIOLegacyContainer *container,
*/
if (cap_mig->pgsize_bitmap & qemu_real_host_page_size()) {
bcontainer->dirty_pages_supported = true;
- container->max_dirty_bitmap_size = cap_mig->max_dirty_bitmap_size;
- container->dirty_pgsizes = cap_mig->pgsize_bitmap;
+ bcontainer->max_dirty_bitmap_size = cap_mig->max_dirty_bitmap_size;
+ bcontainer->dirty_pgsizes = cap_mig->pgsize_bitmap;
}
}
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 15/27] vfio/container: Implement attach/detach_device
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (13 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 14/27] vfio/container: Move dirty_pgsizes and max_dirty_bitmap_size " Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 16/27] Add iommufd configure option Zhenzhong Duan
` (11 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
hw/vfio/common.c | 22 ++++++++++++++++++++++
hw/vfio/container.c | 12 +++++-------
2 files changed, 27 insertions(+), 7 deletions(-)
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index da1d64efca..ee2ebf4be9 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -1552,3 +1552,25 @@ retry:
return info;
}
+
+int vfio_attach_device(char *name, VFIODevice *vbasedev,
+ AddressSpace *as, Error **errp)
+{
+ const VFIOIOMMUBackendOpsClass *ops;
+
+ ops = VFIO_IOMMU_BACKEND_OPS_CLASS(
+ object_class_by_name(TYPE_VFIO_IOMMU_BACKEND_LEGACY_OPS));
+ if (!ops) {
+ error_setg(errp, "VFIO IOMMU Backend not found!");
+ return -ENODEV;
+ }
+ return ops->attach_device(name, vbasedev, as, errp);
+}
+
+void vfio_detach_device(VFIODevice *vbasedev)
+{
+ if (!vbasedev->bcontainer) {
+ return;
+ }
+ vbasedev->bcontainer->ops->detach_device(vbasedev);
+}
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 9d5be749c7..c86accdb38 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -1117,8 +1117,8 @@ static int vfio_device_groupid(VFIODevice *vbasedev, Error **errp)
* @name and @vbasedev->name are likely to be different depending
* on the type of the device, hence the need for passing @name
*/
-int vfio_attach_device(char *name, VFIODevice *vbasedev,
- AddressSpace *as, Error **errp)
+static int vfio_legacy_attach_device(char *name, VFIODevice *vbasedev,
+ AddressSpace *as, Error **errp)
{
int groupid = vfio_device_groupid(vbasedev, errp);
VFIODevice *vbasedev_iter;
@@ -1158,14 +1158,10 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
return ret;
}
-void vfio_detach_device(VFIODevice *vbasedev)
+static void vfio_legacy_detach_device(VFIODevice *vbasedev)
{
VFIOGroup *group = vbasedev->group;
- if (!vbasedev->bcontainer) {
- return;
- }
-
QLIST_REMOVE(vbasedev, global_next);
QLIST_REMOVE(vbasedev, container_next);
vbasedev->bcontainer = NULL;
@@ -1180,6 +1176,8 @@ static void vfio_iommu_backend_legacy_ops_class_init(ObjectClass *oc,
ops->dma_map = vfio_legacy_dma_map;
ops->dma_unmap = vfio_legacy_dma_unmap;
+ ops->attach_device = vfio_legacy_attach_device;
+ ops->detach_device = vfio_legacy_detach_device;
ops->set_dirty_page_tracking = vfio_legacy_set_dirty_page_tracking;
ops->query_dirty_bitmap = vfio_legacy_query_dirty_bitmap;
ops->add_window = vfio_legacy_add_section_window;
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 16/27] Add iommufd configure option
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (14 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 15/27] vfio/container: Implement attach/detach_device Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 17/27] backends/iommufd: Introduce the iommufd object Zhenzhong Duan
` (10 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Paolo Bonzini, Marc-André Lureau,
Daniel P. Berrangé, Thomas Huth, Philippe Mathieu-Daudé
This adds "--enable-iommufd/--disable-iommufd" to enable or disable
iommufd support, enabled by default.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
meson.build | 6 ++++++
meson_options.txt | 2 ++
scripts/meson-buildoptions.sh | 3 +++
3 files changed, 11 insertions(+)
diff --git a/meson.build b/meson.build
index 79aef19bdc..e8d285aa5b 100644
--- a/meson.build
+++ b/meson.build
@@ -560,6 +560,10 @@ have_tpm = get_option('tpm') \
.require(targetos != 'windows', error_message: 'TPM emulation only available on POSIX systems') \
.allowed()
+have_iommufd = get_option('iommufd') \
+ .require(targetos == 'linux', error_message: 'iommufd is supported only on Linux') \
+ .allowed()
+
# vhost
have_vhost_user = get_option('vhost_user') \
.disable_auto_if(targetos != 'linux') \
@@ -2126,6 +2130,7 @@ if get_option('tcg').allowed()
endif
config_host_data.set('CONFIG_TPM', have_tpm)
config_host_data.set('CONFIG_TSAN', get_option('tsan'))
+config_host_data.set('CONFIG_IOMMUFD', have_iommufd)
config_host_data.set('CONFIG_USB_LIBUSB', libusb.found())
config_host_data.set('CONFIG_VDE', vde.found())
config_host_data.set('CONFIG_VHOST_NET', have_vhost_net)
@@ -4061,6 +4066,7 @@ summary_info += {'vhost-user-crypto support': have_vhost_user_crypto}
summary_info += {'vhost-user-blk server support': have_vhost_user_blk_server}
summary_info += {'vhost-vdpa support': have_vhost_vdpa}
summary_info += {'build guest agent': have_ga}
+summary_info += {'iommufd support': have_iommufd}
summary(summary_info, bool_yn: true, section: 'Configurable features')
# Compilation information
diff --git a/meson_options.txt b/meson_options.txt
index 6a17b90968..62bd75284b 100644
--- a/meson_options.txt
+++ b/meson_options.txt
@@ -107,6 +107,8 @@ option('dbus_display', type: 'feature', value: 'auto',
description: '-display dbus support')
option('tpm', type : 'feature', value : 'auto',
description: 'TPM support')
+option('iommufd', type : 'feature', value : 'auto',
+ description: 'iommufd support')
# Do not enable it by default even for Mingw32, because it doesn't
# work on Wine.
diff --git a/scripts/meson-buildoptions.sh b/scripts/meson-buildoptions.sh
index 2a74b0275b..86909dc2cc 100644
--- a/scripts/meson-buildoptions.sh
+++ b/scripts/meson-buildoptions.sh
@@ -114,6 +114,7 @@ meson_options_help() {
printf "%s\n" ' guest-agent-msi Build MSI package for the QEMU Guest Agent'
printf "%s\n" ' hvf HVF acceleration support'
printf "%s\n" ' iconv Font glyph conversion support'
+ printf "%s\n" ' iommufd iommufd support'
printf "%s\n" ' jack JACK sound support'
printf "%s\n" ' keyring Linux keyring support'
printf "%s\n" ' kvm KVM acceleration support'
@@ -327,6 +328,8 @@ _meson_option_parse() {
--enable-install-blobs) printf "%s" -Dinstall_blobs=true ;;
--disable-install-blobs) printf "%s" -Dinstall_blobs=false ;;
--interp-prefix=*) quote_sh "-Dinterp_prefix=$2" ;;
+ --enable-iommufd) printf "%s" -Diommufd=enabled ;;
+ --disable-iommufd) printf "%s" -Diommufd=disabled ;;
--enable-jack) printf "%s" -Djack=enabled ;;
--disable-jack) printf "%s" -Djack=disabled ;;
--enable-keyring) printf "%s" -Dkeyring=enabled ;;
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 17/27] backends/iommufd: Introduce the iommufd object
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (15 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 16/27] Add iommufd configure option Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 10:00 ` Markus Armbruster
2023-10-16 8:32 ` [PATCH v2 18/27] util/char_dev: Add open_cdev() Zhenzhong Duan
` (9 subsequent siblings)
26 siblings, 1 reply; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Paolo Bonzini, Daniel P. Berrangé,
Eduardo Habkost, Eric Blake, Markus Armbruster
From: Eric Auger <eric.auger@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.williamson@redhat.com>
Signed-off-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>
---
MAINTAINERS | 7 +
qapi/qom.json | 18 ++-
include/sysemu/iommufd.h | 46 +++++++
backends/iommufd-stub.c | 59 +++++++++
backends/iommufd.c | 268 +++++++++++++++++++++++++++++++++++++++
backends/Kconfig | 4 +
backends/meson.build | 5 +
backends/trace-events | 12 ++
qemu-options.hx | 13 ++
9 files changed, 431 insertions(+), 1 deletion(-)
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 9e7dec4a58..a7cdeb7825 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -2081,6 +2081,13 @@ 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>
+S: Supported
+F: backends/iommufd.c
+F: include/sysemu/iommufd.h
+
vhost
M: Michael S. Tsirkin <mst@redhat.com>
S: Supported
diff --git a/qapi/qom.json b/qapi/qom.json
index c53ef978ff..3f964e57f5 100644
--- a/qapi/qom.json
+++ b/qapi/qom.json
@@ -794,6 +794,18 @@
{ 'struct': 'VfioUserServerProperties',
'data': { 'socket': 'SocketAddress', 'device': 'str' } }
+##
+# @IOMMUFDProperties:
+#
+# Properties for IOMMUFDbackend objects.
+#
+# fd: file descriptor name
+#
+# Since: 7.2
+##
+{ 'struct': 'IOMMUFDProperties',
+ 'data': { '*fd': 'str' } }
+
##
# @RngProperties:
#
@@ -948,6 +960,8 @@
'qtest',
'rng-builtin',
'rng-egd',
+ { 'name': 'iommufd',
+ 'if': 'CONFIG_IOMMUFD' },
{ 'name': 'rng-random',
'if': 'CONFIG_POSIX' },
'secret',
@@ -1029,7 +1043,9 @@
'tls-creds-x509': 'TlsCredsX509Properties',
'tls-cipher-suites': 'TlsCredsProperties',
'x-remote-object': 'RemoteObjectProperties',
- 'x-vfio-user-server': 'VfioUserServerProperties'
+ 'x-vfio-user-server': 'VfioUserServerProperties',
+ 'iommufd': { 'type': 'IOMMUFDProperties',
+ 'if': 'CONFIG_IOMMUFD' }
} }
##
diff --git a/include/sysemu/iommufd.h b/include/sysemu/iommufd.h
new file mode 100644
index 0000000000..f0e5c7eeb8
--- /dev/null
+++ b/include/sysemu/iommufd.h
@@ -0,0 +1,46 @@
+#ifndef SYSEMU_IOMMUFD_H
+#define SYSEMU_IOMMUFD_H
+
+#include "qom/object.h"
+#include "qemu/thread.h"
+#include "exec/hwaddr.h"
+#include "exec/cpu-common.h"
+
+#define TYPE_IOMMUFD_BACKEND "iommufd"
+OBJECT_DECLARE_TYPE(IOMMUFDBackend, IOMMUFDBackendClass,
+ IOMMUFD_BACKEND)
+#define IOMMUFD_BACKEND(obj) \
+ OBJECT_CHECK(IOMMUFDBackend, (obj), TYPE_IOMMUFD_BACKEND)
+#define IOMMUFD_BACKEND_GET_CLASS(obj) \
+ OBJECT_GET_CLASS(IOMMUFDBackendClass, (obj), TYPE_IOMMUFD_BACKEND)
+#define IOMMUFD_BACKEND_CLASS(klass) \
+ OBJECT_CLASS_CHECK(IOMMUFDBackendClass, (klass), TYPE_IOMMUFD_BACKEND)
+struct IOMMUFDBackendClass {
+ ObjectClass parent_class;
+};
+
+struct IOMMUFDBackend {
+ Object parent;
+
+ /*< protected >*/
+ int fd; /* /dev/iommu file descriptor */
+ bool owned; /* is the /dev/iommu opened internally */
+ QemuMutex lock;
+ uint32_t users;
+
+ /*< public >*/
+};
+
+int iommufd_backend_connect(IOMMUFDBackend *be, Error **errp);
+void iommufd_backend_disconnect(IOMMUFDBackend *be);
+
+int iommufd_backend_get_ioas(IOMMUFDBackend *be, uint32_t *ioas_id);
+void iommufd_backend_put_ioas(IOMMUFDBackend *be, uint32_t ioas_id);
+void iommufd_backend_free_id(int fd, uint32_t id);
+int iommufd_backend_map_dma(IOMMUFDBackend *be, uint32_t ioas_id, hwaddr iova,
+ ram_addr_t size, void *vaddr, bool readonly);
+int iommufd_backend_unmap_dma(IOMMUFDBackend *be, uint32_t ioas_id,
+ hwaddr iova, ram_addr_t size);
+int iommufd_backend_alloc_hwpt(int iommufd, uint32_t dev_id,
+ uint32_t pt_id, uint32_t *out_hwpt);
+#endif
diff --git a/backends/iommufd-stub.c b/backends/iommufd-stub.c
new file mode 100644
index 0000000000..cfb9a87859
--- /dev/null
+++ b/backends/iommufd-stub.c
@@ -0,0 +1,59 @@
+/*
+ * iommufd container backend stub
+ *
+ * Copyright (C) 2023 Intel Corporation.
+ * Copyright Red Hat, Inc. 2023
+ *
+ * Authors: Yi Liu <yi.l....@intel.com>
+ * Eric Auger <eric.au...@redhat.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "qemu/osdep.h"
+#include "sysemu/iommufd.h"
+#include "qemu/error-report.h"
+
+int iommufd_backend_connect(IOMMUFDBackend *be, Error **errp)
+{
+ return 0;
+}
+void iommufd_backend_disconnect(IOMMUFDBackend *be)
+{
+}
+void iommufd_backend_free_id(int fd, uint32_t id)
+{
+}
+int iommufd_backend_get_ioas(IOMMUFDBackend *be, uint32_t *ioas_id)
+{
+ return 0;
+}
+void iommufd_backend_put_ioas(IOMMUFDBackend *be, uint32_t ioas_id)
+{
+}
+int iommufd_backend_map_dma(IOMMUFDBackend *be, uint32_t ioas_id, hwaddr iova,
+ ram_addr_t size, void *vaddr, bool readonly)
+{
+ return 0;
+}
+int iommufd_backend_unmap_dma(IOMMUFDBackend *be, uint32_t ioas_id,
+ hwaddr iova, ram_addr_t size)
+{
+ return 0;
+}
+int iommufd_backend_alloc_hwpt(int iommufd, uint32_t dev_id,
+ uint32_t pt_id, uint32_t *out_hwpt)
+{
+ return 0;
+}
diff --git a/backends/iommufd.c b/backends/iommufd.c
new file mode 100644
index 0000000000..3f0ed37847
--- /dev/null
+++ b/backends/iommufd.c
@@ -0,0 +1,268 @@
+/*
+ * 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>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#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 void iommufd_backend_class_init(ObjectClass *oc, void *data)
+{
+ 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);
+}
+
+static int iommufd_backend_alloc_ioas(int fd, uint32_t *ioas_id)
+{
+ int ret;
+ struct iommu_ioas_alloc alloc_data = {
+ .size = sizeof(alloc_data),
+ .flags = 0,
+ };
+
+ ret = ioctl(fd, IOMMU_IOAS_ALLOC, &alloc_data);
+ if (ret) {
+ error_report("Failed to allocate ioas %m");
+ }
+
+ *ioas_id = alloc_data.out_ioas_id;
+ trace_iommufd_backend_alloc_ioas(fd, *ioas_id, ret);
+
+ return ret;
+}
+
+void iommufd_backend_free_id(int fd, uint32_t id)
+{
+ int ret;
+ 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_get_ioas(IOMMUFDBackend *be, uint32_t *ioas_id)
+{
+ int ret;
+
+ ret = iommufd_backend_alloc_ioas(be->fd, ioas_id);
+ trace_iommufd_backend_get_ioas(be->fd, *ioas_id, ret);
+ return ret;
+}
+
+void iommufd_backend_put_ioas(IOMMUFDBackend *be, uint32_t ioas_id)
+{
+ iommufd_backend_free_id(be->fd, ioas_id);
+ trace_iommufd_backend_put_ioas(be->fd, ioas_id);
+}
+
+int iommufd_backend_map_dma(IOMMUFDBackend *be, uint32_t ioas_id, hwaddr iova,
+ ram_addr_t size, void *vaddr, bool readonly)
+{
+ int ret;
+ 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(be->fd, IOMMU_IOAS_MAP, &map);
+ trace_iommufd_backend_map_dma(be->fd, ioas_id, iova, size,
+ vaddr, readonly, ret);
+ if (ret) {
+ error_report("IOMMU_IOAS_MAP failed: %m");
+ }
+ return !ret ? 0 : -errno;
+}
+
+int iommufd_backend_unmap_dma(IOMMUFDBackend *be, uint32_t ioas_id,
+ hwaddr iova, ram_addr_t size)
+{
+ int ret;
+ struct iommu_ioas_unmap unmap = {
+ .size = sizeof(unmap),
+ .ioas_id = ioas_id,
+ .iova = iova,
+ .length = size,
+ };
+
+ ret = ioctl(be->fd, IOMMU_IOAS_UNMAP, &unmap);
+ trace_iommufd_backend_unmap_dma(be->fd, ioas_id, iova, size, ret);
+ /*
+ * TODO: IOMMUFD doesn't support mapping PCI BARs for now.
+ * It's not a problem if there is no p2p dma, relax it here
+ * and avoid many noisy trigger from vIOMMU side.
+ */
+ if (ret && errno == ENOENT) {
+ ret = 0;
+ }
+ if (ret) {
+ error_report("IOMMU_IOAS_UNMAP failed: %m");
+ }
+ return !ret ? 0 : -errno;
+}
+
+int iommufd_backend_alloc_hwpt(int iommufd, uint32_t dev_id,
+ uint32_t pt_id, uint32_t *out_hwpt)
+{
+ int ret;
+ struct iommu_hwpt_alloc alloc_hwpt = {
+ .size = sizeof(struct iommu_hwpt_alloc),
+ .flags = 0,
+ .dev_id = dev_id,
+ .pt_id = pt_id,
+ .__reserved = 0,
+ };
+
+ ret = ioctl(iommufd, IOMMU_HWPT_ALLOC, &alloc_hwpt);
+ trace_iommufd_backend_alloc_hwpt(iommufd, dev_id, pt_id,
+ alloc_hwpt.out_hwpt_id, ret);
+
+ if (ret) {
+ error_report("IOMMU_HWPT_ALLOC failed: %m");
+ } else {
+ *out_hwpt = alloc_hwpt.out_hwpt_id;
+ }
+ return !ret ? 0 : -errno;
+}
+
+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/Kconfig b/backends/Kconfig
index f35abc1609..2cb23f62fa 100644
--- a/backends/Kconfig
+++ b/backends/Kconfig
@@ -1 +1,5 @@
source tpm/Kconfig
+
+config IOMMUFD
+ bool
+ depends on VFIO
diff --git a/backends/meson.build b/backends/meson.build
index 914c7c4afb..05ac57ff15 100644
--- a/backends/meson.build
+++ b/backends/meson.build
@@ -20,6 +20,11 @@ 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'))
+if have_iommufd
+ system_ss.add(files('iommufd.c'))
+else
+ system_ss.add(files('iommufd-stub.c'))
+endif
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
index 652eb76a57..e5f828bca2 100644
--- a/backends/trace-events
+++ b/backends/trace-events
@@ -5,3 +5,15 @@ 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_get_ioas(int iommufd, uint32_t ioas, int ret) " iommufd=%d ioas=%d (%d)"
+iommufd_backend_put_ioas(int iommufd, uint32_t ioas) " iommufd=%d ioas=%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(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)"
+iommufd_backend_alloc_hwpt(int iommufd, uint32_t dev_id, uint32_t pt_id, uint32_t out_hwpt_id, int ret) " iommufd=%d dev_id=%u pt_id=%u out_hwpt=%u (%d)"
diff --git a/qemu-options.hx b/qemu-options.hx
index 54a7e94970..0af0d379a6 100644
--- a/qemu-options.hx
+++ b/qemu-options.hx
@@ -5207,6 +5207,19 @@ SRST
The ``share`` boolean option is on by default with memfd.
+#ifdef CONFIG_IOMMUFD
+ ``-object iommufd,id=id[,fd=fd]``
+ Creates an iommufd backend which allows control of DMA mapping
+ through the /dev/iommu device.
+
+ The ``id`` parameter is a unique ID which frontends (such as
+ vfio-pci of vdpa) will use to connect with the iommufd backend.
+
+ The ``fd`` parameter is an optional pre-opened file descriptor
+ resulting from /dev/iommu opening. Usually the iommufd is shared
+ across all subsystems, bringing the benefit of centralized
+ reference counting.
+#endif
``-object rng-builtin,id=id``
Creates a random number generator backend which obtains entropy
from QEMU builtin functions. The ``id`` parameter is a unique ID
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 18/27] util/char_dev: Add open_cdev()
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (16 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 17/27] backends/iommufd: Introduce the iommufd object Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 19/27] vfio/iommufd: Implement the iommufd backend Zhenzhong Duan
` (8 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
From: Yi Liu <yi.l.liu@intel.com>
/dev/vfio/devices/vfioX may not exist. In that case it is still possible
to open /dev/char/$major:$minor instead. Add helper function to abstract
the cdev open.
Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
MAINTAINERS | 6 +++
include/qemu/chardev_open.h | 16 ++++++++
util/chardev_open.c | 81 +++++++++++++++++++++++++++++++++++++
util/meson.build | 1 +
4 files changed, 104 insertions(+)
create mode 100644 include/qemu/chardev_open.h
create mode 100644 util/chardev_open.c
diff --git a/MAINTAINERS b/MAINTAINERS
index a7cdeb7825..eb6b7d274c 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -3408,6 +3408,12 @@ S: Maintained
F: include/qemu/iova-tree.h
F: util/iova-tree.c
+cdev Open
+M: Yi Liu <yi.l.liu@intel.com>
+S: Maintained
+F: include/qemu/chardev_open.h
+F: util/chardev_open.c
+
elf2dmp
M: Viktor Prutyanov <viktor.prutyanov@phystech.edu>
S: Maintained
diff --git a/include/qemu/chardev_open.h b/include/qemu/chardev_open.h
new file mode 100644
index 0000000000..6580d351c6
--- /dev/null
+++ b/include/qemu/chardev_open.h
@@ -0,0 +1,16 @@
+/*
+ * QEMU Chardev Helper
+ *
+ * Copyright (C) 2023 Intel Corporation.
+ *
+ * Authors: Yi Liu <yi.l.liu@intel.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2. See
+ * the COPYING file in the top-level directory.
+ */
+
+#ifndef QEMU_CHARDEV_HELPERS_H
+#define QEMU_CHARDEV_HELPERS_H
+
+int open_cdev(const char *devpath, dev_t cdev);
+#endif
diff --git a/util/chardev_open.c b/util/chardev_open.c
new file mode 100644
index 0000000000..005d2b81bd
--- /dev/null
+++ b/util/chardev_open.c
@@ -0,0 +1,81 @@
+/*
+ * Copyright (c) 2019, Mellanox Technologies. All rights reserved.
+ * Copyright (C) 2023 Intel Corporation.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Authors: Yi Liu <yi.l.liu@intel.com>
+ *
+ * Copied from
+ * https://github.com/linux-rdma/rdma-core/blob/master/util/open_cdev.c
+ *
+ */
+
+#include "qemu/osdep.h"
+#include "qemu/chardev_open.h"
+
+static int open_cdev_internal(const char *path, dev_t cdev)
+{
+ struct stat st;
+ int fd;
+
+ fd = qemu_open_old(path, O_RDWR);
+ if (fd == -1) {
+ return -1;
+ }
+ if (fstat(fd, &st) || !S_ISCHR(st.st_mode) ||
+ (cdev != 0 && st.st_rdev != cdev)) {
+ close(fd);
+ return -1;
+ }
+ return fd;
+}
+
+static int open_cdev_robust(dev_t cdev)
+{
+ g_autofree char *devpath;
+
+ /*
+ * This assumes that udev is being used and is creating the /dev/char/
+ * symlinks.
+ */
+ devpath = g_strdup_printf("/dev/char/%u:%u", major(cdev), minor(cdev));
+ return open_cdev_internal(devpath, cdev);
+}
+
+int open_cdev(const char *devpath, dev_t cdev)
+{
+ int fd;
+
+ fd = open_cdev_internal(devpath, cdev);
+ if (fd == -1 && cdev != 0) {
+ return open_cdev_robust(cdev);
+ }
+ return fd;
+}
diff --git a/util/meson.build b/util/meson.build
index c4827fd70a..654f4528fb 100644
--- a/util/meson.build
+++ b/util/meson.build
@@ -106,6 +106,7 @@ if have_block
util_ss.add(files('filemonitor-stub.c'))
endif
util_ss.add(when: 'CONFIG_LINUX', if_true: files('vfio-helpers.c'))
+ util_ss.add(when: 'CONFIG_LINUX', if_true: files('chardev_open.c'))
endif
if cpu == 'aarch64'
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 19/27] vfio/iommufd: Implement the iommufd backend
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (17 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 18/27] util/char_dev: Add open_cdev() Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 20/27] vfio/container: Bypass EEH if " Zhenzhong Duan
` (7 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
From: Yi Liu <yi.l.liu@intel.com>
Add the iommufd backend. The IOMMUFD container class 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 due
to missing support in the host kernel.
Co-authored-by: Eric Auger <eric.auger@redhat.com>
Signed-off-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>
---
include/hw/vfio/vfio-common.h | 22 ++
include/hw/vfio/vfio-container-base.h | 3 +
hw/vfio/common.c | 19 +-
hw/vfio/iommufd.c | 535 ++++++++++++++++++++++++++
hw/vfio/meson.build | 3 +
hw/vfio/trace-events | 12 +
6 files changed, 590 insertions(+), 4 deletions(-)
create mode 100644 hw/vfio/iommufd.c
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 9f2b86581b..e72f5962ee 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -83,6 +83,23 @@ typedef struct VFIOLegacyContainer {
QLIST_HEAD(, VFIOGroup) group_list;
} VFIOLegacyContainer;
+#ifdef CONFIG_IOMMUFD
+typedef struct VFIOIOASHwpt {
+ uint32_t hwpt_id;
+ QLIST_HEAD(, VFIODevice) device_list;
+ QLIST_ENTRY(VFIOIOASHwpt) next;
+} VFIOIOASHwpt;
+
+typedef struct IOMMUFDBackend IOMMUFDBackend;
+
+typedef struct VFIOIOMMUFDContainer {
+ VFIOContainer bcontainer;
+ IOMMUFDBackend *be;
+ uint32_t ioas_id;
+ QLIST_HEAD(, VFIOIOASHwpt) hwpt_list;
+} VFIOIOMMUFDContainer;
+#endif
+
typedef struct VFIODeviceOps VFIODeviceOps;
typedef struct VFIODevice {
@@ -110,6 +127,11 @@ typedef struct VFIODevice {
OnOffAuto pre_copy_dirty_page_tracking;
bool dirty_pages_supported;
bool dirty_tracking;
+#ifdef CONFIG_IOMMUFD
+ int devid;
+ VFIOIOASHwpt *hwpt;
+ IOMMUFDBackend *iommufd;
+#endif
} VFIODevice;
struct VFIODeviceOps {
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 9a5971a00a..5345986993 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -114,6 +114,9 @@ void vfio_container_init(VFIOContainer *bcontainer,
void vfio_container_destroy(VFIOContainer *bcontainer);
#define TYPE_VFIO_IOMMU_BACKEND_LEGACY_OPS "vfio-iommu-backend-legacy-ops"
+#ifdef CONFIG_IOMMUFD
+#define TYPE_VFIO_IOMMU_BACKEND_IOMMUFD_OPS "vfio-iommu-backend-iommufd-ops"
+#endif
#define TYPE_VFIO_IOMMU_BACKEND_OPS "vfio-iommu-backend-ops"
DECLARE_CLASS_CHECKERS(VFIOIOMMUBackendOpsClass,
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index ee2ebf4be9..6901573c32 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -1520,10 +1520,13 @@ VFIOAddressSpace *vfio_get_address_space(AddressSpace *as)
void vfio_put_address_space(VFIOAddressSpace *space)
{
- if (QLIST_EMPTY(&space->containers)) {
- QLIST_REMOVE(space, list);
- g_free(space);
+ if (!QLIST_EMPTY(&space->containers)) {
+ return;
}
+
+ QLIST_REMOVE(space, list);
+ g_free(space);
+
if (QLIST_EMPTY(&vfio_address_spaces)) {
qemu_unregister_reset(vfio_reset_handler, NULL);
}
@@ -1558,8 +1561,16 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
{
const VFIOIOMMUBackendOpsClass *ops;
- ops = VFIO_IOMMU_BACKEND_OPS_CLASS(
+#ifdef CONFIG_IOMMUFD
+ if (vbasedev->iommufd) {
+ ops = VFIO_IOMMU_BACKEND_OPS_CLASS(
+ object_class_by_name(TYPE_VFIO_IOMMU_BACKEND_IOMMUFD_OPS));
+ } else
+#endif
+ {
+ ops = VFIO_IOMMU_BACKEND_OPS_CLASS(
object_class_by_name(TYPE_VFIO_IOMMU_BACKEND_LEGACY_OPS));
+ }
if (!ops) {
error_setg(errp, "VFIO IOMMU Backend not found!");
return -ENODEV;
diff --git a/hw/vfio/iommufd.c b/hw/vfio/iommufd.c
new file mode 100644
index 0000000000..ee8c4620b6
--- /dev/null
+++ b/hw/vfio/iommufd.c
@@ -0,0 +1,535 @@
+/*
+ * 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>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#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_map(VFIOContainer *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_unmap(VFIOContainer *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 void vfio_kvm_device_add_device(VFIODevice *vbasedev)
+{
+ Error *err = NULL;
+
+ if (vfio_kvm_device_add_fd(vbasedev->fd, &err)) {
+ error_report_err(err);
+ }
+}
+
+static void vfio_kvm_device_del_device(VFIODevice *vbasedev)
+{
+ Error *err = NULL;
+
+ if (vfio_kvm_device_del_fd(vbasedev->fd, &err)) {
+ error_report_err(err);
+ }
+}
+
+static int iommufd_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.
+ */
+ vfio_kvm_device_add_device(vbasedev);
+
+ /* 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_vfio_iommufd_bind_device(bind.iommufd, vbasedev->name,
+ vbasedev->fd, vbasedev->devid);
+ return ret;
+err_bind:
+ vfio_kvm_device_del_device(vbasedev);
+ iommufd_backend_disconnect(iommufd);
+ return ret;
+}
+
+static void iommufd_unbind_and_disconnect(VFIODevice *vbasedev)
+{
+ /* Unbind is automatically conducted when device fd is closed */
+ vfio_kvm_device_del_device(vbasedev);
+ iommufd_backend_disconnect(vbasedev->iommufd);
+}
+
+static int vfio_get_devicefd(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 dirrectory %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_vfio_iommufd_get_devicefd(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 VFIOIOASHwpt *vfio_container_get_hwpt(VFIOIOMMUFDContainer *container,
+ uint32_t hwpt_id)
+{
+ VFIOIOASHwpt *hwpt;
+
+ QLIST_FOREACH(hwpt, &container->hwpt_list, next) {
+ if (hwpt->hwpt_id == hwpt_id) {
+ return hwpt;
+ }
+ }
+
+ hwpt = g_malloc0(sizeof(*hwpt));
+
+ hwpt->hwpt_id = hwpt_id;
+ QLIST_INIT(&hwpt->device_list);
+ QLIST_INSERT_HEAD(&container->hwpt_list, hwpt, next);
+
+ return hwpt;
+}
+
+static void vfio_container_put_hwpt(IOMMUFDBackend *be, VFIOIOASHwpt *hwpt)
+{
+ QLIST_REMOVE(hwpt, next);
+ iommufd_backend_free_id(be->fd, hwpt->hwpt_id);
+ g_free(hwpt);
+}
+
+static int __vfio_device_attach_hwpt(VFIODevice *vbasedev, uint32_t hwpt_id,
+ Error **errp)
+{
+ struct vfio_device_attach_iommufd_pt attach_data = {
+ .argsz = sizeof(attach_data),
+ .flags = 0,
+ .pt_id = hwpt_id,
+ };
+ int ret;
+
+ 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 hwpt_id=%d",
+ vbasedev->iommufd->fd, vbasedev->name, vbasedev->fd,
+ hwpt_id);
+ }
+ return ret;
+}
+
+static int __vfio_device_detach_hwpt(VFIODevice *vbasedev, Error **errp)
+{
+ struct vfio_device_detach_iommufd_pt detach_data = {
+ .argsz = sizeof(detach_data),
+ .flags = 0,
+ };
+ int ret;
+
+ ret = ioctl(vbasedev->fd, VFIO_DEVICE_DETACH_IOMMUFD_PT, &detach_data);
+ if (ret) {
+ error_setg_errno(errp, errno, "detach %s from ioas failed",
+ vbasedev->name);
+ }
+ return ret;
+}
+
+static int vfio_device_attach_container(VFIODevice *vbasedev,
+ VFIOIOMMUFDContainer *container,
+ Error **errp)
+{
+ int ret, iommufd = vbasedev->iommufd->fd;
+ VFIOIOASHwpt *hwpt;
+ uint32_t hwpt_id;
+ Error *err = NULL;
+
+ /* try to attach to an existing hwpt in this container */
+ QLIST_FOREACH(hwpt, &container->hwpt_list, next) {
+ ret = __vfio_device_attach_hwpt(vbasedev, hwpt->hwpt_id, &err);
+ if (ret) {
+ const char *msg = error_get_pretty(err);
+
+ trace_vfio_iommufd_fail_attach_existing_hwpt(msg);
+ error_free(err);
+ err = NULL;
+ } else {
+ goto found_hwpt;
+ }
+ }
+
+ ret = iommufd_backend_alloc_hwpt(iommufd, vbasedev->devid,
+ container->ioas_id, &hwpt_id);
+
+ if (ret) {
+ error_setg_errno(errp, errno, "error alloc shadow hwpt");
+ return ret;
+ }
+
+ /* Attach device to an hwpt within iommufd */
+ ret = __vfio_device_attach_hwpt(vbasedev, hwpt_id, errp);
+ if (ret) {
+ iommufd_backend_free_id(iommufd, hwpt_id);
+ return ret;
+ }
+
+ hwpt = vfio_container_get_hwpt(container, hwpt_id);
+found_hwpt:
+ QLIST_INSERT_HEAD(&hwpt->device_list, vbasedev, next);
+ vbasedev->hwpt = hwpt;
+
+ trace_vfio_iommufd_attach_device(iommufd, vbasedev->name, vbasedev->fd,
+ container->ioas_id, hwpt->hwpt_id);
+ return ret;
+}
+
+static void vfio_device_detach_container(VFIODevice *vbasedev,
+ VFIOIOMMUFDContainer *container)
+{
+ VFIOIOASHwpt *hwpt = vbasedev->hwpt;
+ Error *err = NULL;
+ int ret;
+
+ ret = __vfio_device_detach_hwpt(vbasedev, &err);
+ if (ret) {
+ error_report_err(err);
+ }
+
+ QLIST_REMOVE(vbasedev, next);
+ vbasedev->hwpt = NULL;
+ if (QLIST_EMPTY(&hwpt->device_list)) {
+ vfio_container_put_hwpt(vbasedev->iommufd, hwpt);
+ }
+
+ trace_vfio_iommufd_detach_device(container->be->fd, vbasedev->name,
+ container->ioas_id);
+}
+
+static void vfio_iommufd_container_destroy(VFIOIOMMUFDContainer *container)
+{
+ VFIOContainer *bcontainer = &container->bcontainer;
+
+ if (!QLIST_EMPTY(&container->hwpt_list)) {
+ return;
+ }
+ memory_listener_unregister(&bcontainer->listener);
+ vfio_container_destroy(bcontainer);
+ iommufd_backend_put_ioas(container->be, container->ioas_id);
+ g_free(container);
+}
+
+static int vfio_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_attach_device(char *name, VFIODevice *vbasedev,
+ AddressSpace *as, Error **errp)
+{
+ VFIOIOMMUBackendOpsClass *ops = VFIO_IOMMU_BACKEND_OPS_CLASS(
+ object_class_by_name(TYPE_VFIO_IOMMU_BACKEND_IOMMUFD_OPS));
+ VFIOContainer *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 = vfio_get_devicefd(vbasedev->sysfsdev, errp);
+ if (devfd < 0) {
+ return devfd;
+ }
+ vbasedev->fd = devfd;
+
+ ret = iommufd_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 != ops || vbasedev->iommufd != container->be) {
+ continue;
+ }
+ if (vfio_device_attach_container(vbasedev, container, &err)) {
+ const char *msg = error_get_pretty(err);
+
+ trace_vfio_iommufd_fail_attach_existing_container(msg);
+ error_free(err);
+ err = NULL;
+ } else {
+ ret = vfio_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_get_ioas(vbasedev->iommufd, &ioas_id);
+ if (ret < 0) {
+ error_setg_errno(errp, errno, "Failed to alloc ioas");
+ goto err_get_ioas;
+ }
+
+ trace_vfio_iommufd_alloc_ioas(vbasedev->iommufd->fd, ioas_id);
+
+ container = g_malloc0(sizeof(*container));
+ container->be = vbasedev->iommufd;
+ container->ioas_id = ioas_id;
+ QLIST_INIT(&container->hwpt_list);
+
+ bcontainer = &container->bcontainer;
+ vfio_container_init(bcontainer, space, ops);
+ QLIST_INSERT_HEAD(&space->containers, bcontainer, next);
+
+ ret = vfio_device_attach_container(vbasedev, container, errp);
+ if (ret) {
+ goto err_attach_container;
+ }
+
+ ret = vfio_ram_block_discard_disable(true);
+ if (ret) {
+ goto err_discard_disable;
+ }
+
+ /*
+ * TODO: for now iommufd BE is on par with vfio iommu type1, so it's
+ * fine to add the whole range as window. For SPAPR, below code
+ * should be updated.
+ */
+ vfio_host_win_add(bcontainer, 0, (hwaddr)-1, 4096);
+ bcontainer->pgsizes = 4096;
+
+ 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) {
+ vfio_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_vfio_iommufd_device_info(vbasedev->name, devfd, vbasedev->num_irqs,
+ vbasedev->num_regions, vbasedev->flags);
+ return 0;
+
+err_listener_register:
+ vfio_ram_block_discard_disable(false);
+err_discard_disable:
+ vfio_device_detach_container(vbasedev, container);
+err_attach_container:
+ vfio_iommufd_container_destroy(container);
+err_get_ioas:
+ vfio_put_address_space(space);
+ iommufd_unbind_and_disconnect(vbasedev);
+err_connect_bind:
+ close(vbasedev->fd);
+ return ret;
+}
+
+static void iommufd_detach_device(VFIODevice *vbasedev)
+{
+ VFIOContainer *bcontainer = vbasedev->bcontainer;
+ VFIOIOMMUFDContainer *container;
+ VFIOAddressSpace *space = bcontainer->space;
+
+ QLIST_REMOVE(vbasedev, global_next);
+ QLIST_REMOVE(vbasedev, container_next);
+ vbasedev->bcontainer = NULL;
+
+ if (!vbasedev->ram_block_discard_allowed) {
+ vfio_ram_block_discard_disable(false);
+ }
+
+ container = container_of(bcontainer, VFIOIOMMUFDContainer, bcontainer);
+ vfio_device_detach_container(vbasedev, container);
+ vfio_iommufd_container_destroy(container);
+ vfio_put_address_space(space);
+
+ iommufd_unbind_and_disconnect(vbasedev);
+ close(vbasedev->fd);
+}
+
+static void vfio_iommu_backend_iommufd_ops_class_init(ObjectClass *oc,
+ void *data) {
+ VFIOIOMMUBackendOpsClass *ops = VFIO_IOMMU_BACKEND_OPS_CLASS(oc);
+
+ ops->dma_map = iommufd_map;
+ ops->dma_unmap = iommufd_unmap;
+ ops->attach_device = iommufd_attach_device;
+ ops->detach_device = iommufd_detach_device;
+}
+
+static const TypeInfo vfio_iommu_backend_iommufd_ops_type = {
+ .name = TYPE_VFIO_IOMMU_BACKEND_IOMMUFD_OPS,
+
+ .parent = TYPE_VFIO_IOMMU_BACKEND_OPS,
+ .class_init = vfio_iommu_backend_iommufd_ops_class_init,
+ .abstract = true,
+};
+static void vfio_iommu_backend_iommufd_ops_register_types(void)
+{
+ type_register_static(&vfio_iommu_backend_iommufd_ops_type);
+}
+type_init(vfio_iommu_backend_iommufd_ops_register_types);
diff --git a/hw/vfio/meson.build b/hw/vfio/meson.build
index eb6ce6229d..9cae2c9e21 100644
--- a/hw/vfio/meson.build
+++ b/hw/vfio/meson.build
@@ -7,6 +7,9 @@ vfio_ss.add(files(
'spapr.c',
'migration.c',
))
+if have_iommufd
+ vfio_ss.add(files('iommufd.c'))
+endif
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..9b180cf77c 100644
--- a/hw/vfio/trace-events
+++ b/hw/vfio/trace-events
@@ -164,3 +164,15 @@ 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
+
+vfio_iommufd_get_devicefd(const char *dev, int devfd) " %s (fd=%d)"
+vfio_iommufd_bind_device(int iommufd, const char *name, int devfd, int devid) " [iommufd=%d] Successfully bound device %s (fd=%d): output devid=%d"
+vfio_iommufd_fail_attach_existing_hwpt(const char *msg) " %s"
+vfio_iommufd_attach_device(int iommufd, const char *name, int devfd, int ioasid, int hwptid) " [iommufd=%d] Successfully attached device %s (%d) to ioasid=%d: output hwptd=%d"
+vfio_iommufd_detach_device(int iommufd, const char *name, int ioasid) " [iommufd=%d] Detached %s from ioasid=%d"
+vfio_iommufd_alloc_ioas(int iommufd, int ioas_id) " [iommufd=%d] new IOMMUFD container with ioasid=%d"
+vfio_iommufd_device_info(char *name, int devfd, int num_irqs, int num_regions, int flags) " %s (%d) num_irqs=%d num_regions=%d flags=%d"
+vfio_iommufd_fail_attach_existing_container(const char *msg) " %s"
+vfio_iommufd_container_reset(char *name) " Successfully reset %s"
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 20/27] vfio/container: Bypass EEH if iommufd backend
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (18 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 19/27] vfio/iommufd: Implement the iommufd backend Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 21/27] vfio/pci: Adapt vfio pci hot reset support with iommufd BE Zhenzhong Duan
` (6 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
IBM EEH is only supported by legacy backend currently, bypass it
for IOMMUFD backend.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
hw/vfio/container.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index c86accdb38..dd9534afab 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -1047,6 +1047,8 @@ static VFIOLegacyContainer *vfio_eeh_as_container(AddressSpace *as)
{
VFIOAddressSpace *space = vfio_get_address_space(as);
VFIOContainer *bcontainer = NULL;
+ const VFIOIOMMUBackendOpsClass *ops = VFIO_IOMMU_BACKEND_OPS_CLASS(
+ object_class_by_name(TYPE_VFIO_IOMMU_BACKEND_LEGACY_OPS));
if (QLIST_EMPTY(&space->containers)) {
/* No containers to act on */
@@ -1055,7 +1057,7 @@ static VFIOLegacyContainer *vfio_eeh_as_container(AddressSpace *as)
bcontainer = QLIST_FIRST(&space->containers);
- if (QLIST_NEXT(bcontainer, next)) {
+ if (QLIST_NEXT(bcontainer, next) || bcontainer->ops != ops) {
/*
* We don't yet have logic to synchronize EEH state across
* multiple containers
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 21/27] vfio/pci: Adapt vfio pci hot reset support with iommufd BE
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (19 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 20/27] vfio/container: Bypass EEH if " Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 22/27] vfio/pci: Allow the selection of a given iommu backend Zhenzhong Duan
` (5 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
As pci hot reset path need to reference pci specific functions
and data structures, adding container level callback functions
for legacy and iommufd BE and referencing those pci specific
func/data is no better than implementing reset support with
iommufd BE directly in pci.c
This way we can also share the common bus reset and system reset
path for both BEs.
A help function vfio_pci_get_pci_hot_reset_info() is extracted out
for usage by both BEs.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
hw/vfio/pci.c | 212 +++++++++++++++++++++++++++++++++++++++----
hw/vfio/trace-events | 1 +
2 files changed, 196 insertions(+), 17 deletions(-)
diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
index b27011cee7..24fc047423 100644
--- a/hw/vfio/pci.c
+++ b/hw/vfio/pci.c
@@ -42,6 +42,7 @@
#include "qapi/error.h"
#include "migration/blocker.h"
#include "migration/qemu-file.h"
+#include "linux/iommufd.h"
#define TYPE_VFIO_PCI_NOHOTPLUG "vfio-pci-nohotplug"
@@ -2445,22 +2446,13 @@ static bool vfio_pci_host_match(PCIHostDeviceAddress *addr, const char *name)
return (strcmp(tmp, name) == 0);
}
-static int vfio_pci_hot_reset(VFIOPCIDevice *vdev, bool single)
+static int vfio_pci_get_pci_hot_reset_info(VFIOPCIDevice *vdev,
+ struct vfio_pci_hot_reset_info **info_p)
{
- VFIOGroup *group;
struct vfio_pci_hot_reset_info *info;
- struct vfio_pci_dependent_device *devices;
- struct vfio_pci_hot_reset *reset;
- int32_t *fds;
- int ret, i, count;
- bool multi = false;
+ int ret, count;
- trace_vfio_pci_hot_reset(vdev->vbasedev.name, single ? "one" : "multi");
-
- if (!single) {
- vfio_pci_pre_reset(vdev);
- }
- vdev->vbasedev.needs_reset = false;
+ assert(info_p && !*info_p);
info = g_malloc0(sizeof(*info));
info->argsz = sizeof(*info);
@@ -2468,24 +2460,53 @@ static int vfio_pci_hot_reset(VFIOPCIDevice *vdev, bool single)
ret = ioctl(vdev->vbasedev.fd, VFIO_DEVICE_GET_PCI_HOT_RESET_INFO, info);
if (ret && errno != ENOSPC) {
ret = -errno;
+ g_free(info);
if (!vdev->has_pm_reset) {
error_report("vfio: Cannot reset device %s, "
"no available reset mechanism.", vdev->vbasedev.name);
}
- goto out_single;
+ return ret;
}
count = info->count;
- info = g_realloc(info, sizeof(*info) + (count * sizeof(*devices)));
- info->argsz = sizeof(*info) + (count * sizeof(*devices));
- devices = &info->devices[0];
+ info = g_realloc(info, sizeof(*info) + (count * sizeof(info->devices[0])));
+ info->argsz = sizeof(*info) + (count * sizeof(info->devices[0]));
ret = ioctl(vdev->vbasedev.fd, VFIO_DEVICE_GET_PCI_HOT_RESET_INFO, info);
if (ret) {
ret = -errno;
+ g_free(info);
error_report("vfio: hot reset info failed: %m");
+ return ret;
+ }
+
+ *info_p = info;
+ return 0;
+}
+
+static int vfio_pci_hot_reset_legacy(VFIOPCIDevice *vdev, bool single)
+{
+ VFIOGroup *group;
+ struct vfio_pci_hot_reset_info *info = NULL;
+ struct vfio_pci_dependent_device *devices;
+ struct vfio_pci_hot_reset *reset;
+ int32_t *fds;
+ int ret, i, count;
+ bool multi = false;
+
+ trace_vfio_pci_hot_reset(vdev->vbasedev.name, single ? "one" : "multi");
+
+ if (!single) {
+ vfio_pci_pre_reset(vdev);
+ }
+ vdev->vbasedev.needs_reset = false;
+
+ ret = vfio_pci_get_pci_hot_reset_info(vdev, &info);
+
+ if (ret) {
goto out_single;
}
+ devices = &info->devices[0];
trace_vfio_pci_hot_reset_has_dep_devices(vdev->vbasedev.name);
@@ -2627,6 +2648,163 @@ out_single:
return ret;
}
+#ifdef CONFIG_IOMMUFD
+static VFIODevice *vfio_pci_find_by_iommufd_devid(__u32 devid)
+{
+ VFIODevice *vbasedev_iter;
+ VFIOIOMMUBackendOpsClass *ops = VFIO_IOMMU_BACKEND_OPS_CLASS(
+ object_class_by_name(TYPE_VFIO_IOMMU_BACKEND_IOMMUFD_OPS));
+
+ QLIST_FOREACH(vbasedev_iter, &vfio_device_list, global_next) {
+ if (vbasedev_iter->bcontainer->ops != ops) {
+ continue;
+ }
+ if (devid == vbasedev_iter->devid) {
+ return vbasedev_iter;
+ }
+ }
+ return NULL;
+}
+
+static int vfio_pci_hot_reset_iommufd(VFIOPCIDevice *vdev, bool single)
+{
+ struct vfio_pci_hot_reset_info *info = NULL;
+ struct vfio_pci_dependent_device *devices;
+ struct vfio_pci_hot_reset *reset;
+ int ret, i;
+ bool multi = false;
+
+ trace_vfio_pci_hot_reset(vdev->vbasedev.name, single ? "one" : "multi");
+
+ if (!single) {
+ vfio_pci_pre_reset(vdev);
+ }
+ vdev->vbasedev.needs_reset = false;
+
+ ret = vfio_pci_get_pci_hot_reset_info(vdev, &info);
+
+ if (ret) {
+ goto out_single;
+ }
+
+ assert(info->flags & VFIO_PCI_HOT_RESET_FLAG_DEV_ID);
+
+ devices = &info->devices[0];
+
+ if (!(info->flags & VFIO_PCI_HOT_RESET_FLAG_DEV_ID_OWNED)) {
+ if (!vdev->has_pm_reset) {
+ for (i = 0; i < info->count; i++) {
+ if (devices[i].devid == VFIO_PCI_DEVID_NOT_OWNED) {
+ error_report("vfio: Cannot reset device %s, "
+ "depends on device %04x:%02x:%02x.%x "
+ "which is not owned.",
+ vdev->vbasedev.name, devices[i].segment,
+ devices[i].bus, PCI_SLOT(devices[i].devfn),
+ PCI_FUNC(devices[i].devfn));
+ }
+ }
+ }
+ ret = -EPERM;
+ goto out_single;
+ }
+
+ trace_vfio_pci_hot_reset_has_dep_devices(vdev->vbasedev.name);
+
+ for (i = 0; i < info->count; i++) {
+ VFIOPCIDevice *tmp;
+ VFIODevice *vbasedev_iter;
+
+ trace_vfio_pci_hot_reset_dep_devices_iommufd(devices[i].segment,
+ devices[i].bus,
+ PCI_SLOT(devices[i].devfn),
+ PCI_FUNC(devices[i].devfn),
+ devices[i].devid);
+
+ /*
+ * If a VFIO cdev device is resettable, all the dependent devices
+ * are either bound to same iommufd or within same iommu_groups as
+ * one of the iommufd bound devices.
+ */
+ assert(devices[i].devid != VFIO_PCI_DEVID_NOT_OWNED);
+
+ if (devices[i].devid == vdev->vbasedev.devid ||
+ devices[i].devid == VFIO_PCI_DEVID_OWNED) {
+ continue;
+ }
+
+ vbasedev_iter = vfio_pci_find_by_iommufd_devid(devices[i].devid);
+ if (!vbasedev_iter || !vbasedev_iter->dev->realized ||
+ vbasedev_iter->type != VFIO_DEVICE_TYPE_PCI) {
+ continue;
+ }
+ tmp = container_of(vbasedev_iter, VFIOPCIDevice, vbasedev);
+ if (single) {
+ ret = -EINVAL;
+ goto out_single;
+ }
+ vfio_pci_pre_reset(tmp);
+ tmp->vbasedev.needs_reset = false;
+ multi = true;
+ }
+
+ if (!single && !multi) {
+ ret = -EINVAL;
+ goto out_single;
+ }
+
+ /* Use zero length array for hot reset with iommufd backend */
+ reset = g_malloc0(sizeof(*reset));
+ reset->argsz = sizeof(*reset);
+
+ /* Bus reset! */
+ ret = ioctl(vdev->vbasedev.fd, VFIO_DEVICE_PCI_HOT_RESET, reset);
+ g_free(reset);
+
+ trace_vfio_pci_hot_reset_result(vdev->vbasedev.name,
+ ret ? strerror(errno) : "Success");
+
+ /* Re-enable INTx on affected devices */
+ for (i = 0; i < info->count; i++) {
+ VFIOPCIDevice *tmp;
+ VFIODevice *vbasedev_iter;
+
+ if (devices[i].devid == vdev->vbasedev.devid ||
+ devices[i].devid == VFIO_PCI_DEVID_OWNED) {
+ continue;
+ }
+
+ vbasedev_iter = vfio_pci_find_by_iommufd_devid(devices[i].devid);
+ if (!vbasedev_iter || !vbasedev_iter->dev->realized ||
+ vbasedev_iter->type != VFIO_DEVICE_TYPE_PCI) {
+ continue;
+ }
+ tmp = container_of(vbasedev_iter, VFIOPCIDevice, vbasedev);
+ vfio_pci_post_reset(tmp);
+ }
+out_single:
+ if (!single) {
+ vfio_pci_post_reset(vdev);
+ }
+ g_free(info);
+
+ return ret;
+}
+#endif
+
+static int vfio_pci_hot_reset(VFIOPCIDevice *vdev, bool single)
+{
+#ifdef CONFIG_IOMMUFD
+ if (vdev->vbasedev.iommufd) {
+ return vfio_pci_hot_reset_iommufd(vdev, single);
+ } else
+#endif
+ {
+ return vfio_pci_hot_reset_legacy(vdev, single);
+ }
+}
+
+
+
/*
* We want to differentiate hot reset of multiple in-use devices vs hot reset
* of a single in-use device. VFIO_DEVICE_RESET will already handle the case
diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
index 9b180cf77c..71c5840636 100644
--- a/hw/vfio/trace-events
+++ b/hw/vfio/trace-events
@@ -34,6 +34,7 @@ vfio_check_af_flr(const char *name) "%s Supports FLR via AF cap"
vfio_pci_hot_reset(const char *name, const char *type) " (%s) %s"
vfio_pci_hot_reset_has_dep_devices(const char *name) "%s: hot reset dependent devices:"
vfio_pci_hot_reset_dep_devices(int domain, int bus, int slot, int function, int group_id) "\t%04x:%02x:%02x.%x group %d"
+vfio_pci_hot_reset_dep_devices_iommufd(int domain, int bus, int slot, int function, int dev_id) "\t%04x:%02x:%02x.%x devid %d"
vfio_pci_hot_reset_result(const char *name, const char *result) "%s hot reset: %s"
vfio_populate_device_config(const char *name, unsigned long size, unsigned long offset, unsigned long flags) "Device %s config:\n size: 0x%lx, offset: 0x%lx, flags: 0x%lx"
vfio_populate_device_get_irq_info_failure(const char *errstr) "VFIO_DEVICE_GET_IRQ_INFO failure: %s"
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 22/27] vfio/pci: Allow the selection of a given iommu backend
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (20 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 21/27] vfio/pci: Adapt vfio pci hot reset support with iommufd BE Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 23/27] vfio/pci: Make vfio cdev pre-openable by passing a file handle Zhenzhong Duan
` (4 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
Now we support two types of iommu backends, let's add the capability
to select one of them. This depends on whether an iommufd object has
been linked with the vfio-pci device:
if the user wants to use the legacy backend, it shall not
link the vfio-pci device with any iommufd object:
-device vfio-pci,host=0000:02:00.0
This is called the legacy mode/backend.
If the user wants to use the iommufd backend (/dev/iommu) it
shall pass an iommufd object id in the vfio-pci device options:
-object iommufd,id=iommufd0
-device vfio-pci,host=0000:02:00.0,iommufd=iommufd0
Suggested-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-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>
---
hw/vfio/pci.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
index 24fc047423..15e1b771b0 100644
--- a/hw/vfio/pci.c
+++ b/hw/vfio/pci.c
@@ -43,6 +43,7 @@
#include "migration/blocker.h"
#include "migration/qemu-file.h"
#include "linux/iommufd.h"
+#include "sysemu/iommufd.h"
#define TYPE_VFIO_PCI_NOHOTPLUG "vfio-pci-nohotplug"
@@ -3700,6 +3701,10 @@ static Property vfio_pci_dev_properties[] = {
* DEFINE_PROP_STRING("vfiofd", VFIOPCIDevice, vfiofd_name),
* DEFINE_PROP_STRING("vfiogroupfd, VFIOPCIDevice, vfiogroupfd_name),
*/
+#ifdef CONFIG_IOMMUFD
+ DEFINE_PROP_LINK("iommufd", VFIOPCIDevice, vbasedev.iommufd,
+ TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
+#endif
DEFINE_PROP_END_OF_LIST(),
};
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 23/27] vfio/pci: Make vfio cdev pre-openable by passing a file handle
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (21 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 22/27] vfio/pci: Allow the selection of a given iommu backend Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 24/27] vfio: Allow the selection of a given iommu backend for platform ap and ccw Zhenzhong Duan
` (3 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
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 a helper function vfio_device_get_name() to check fd and get
device name, it 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>
---
include/hw/vfio/vfio-common.h | 1 +
hw/vfio/helpers.c | 33 +++++++++++++++++++++++++++++
hw/vfio/iommufd.c | 12 +++++++----
hw/vfio/pci.c | 40 ++++++++++++++++++++++++-----------
4 files changed, 70 insertions(+), 16 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index e72f5962ee..e6804baa6d 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -244,6 +244,7 @@ struct vfio_info_cap_header *
vfio_get_device_info_cap(struct vfio_device_info *info, uint16_t id);
struct vfio_info_cap_header *
vfio_get_cap(void *ptr, uint32_t cap_offset, uint16_t id);
+int vfio_device_get_name(VFIODevice *vbasedev, Error **errp);
#endif
extern const MemoryListener vfio_prereg_listener;
diff --git a/hw/vfio/helpers.c b/hw/vfio/helpers.c
index 7e5da21b31..70c65cf71d 100644
--- a/hw/vfio/helpers.c
+++ b/hw/vfio/helpers.c
@@ -610,3 +610,36 @@ bool vfio_has_region_cap(VFIODevice *vbasedev, int region, uint16_t cap_type)
return ret;
}
+
+int vfio_device_get_name(VFIODevice *vbasedev, Error **errp)
+{
+ struct stat st;
+
+ if (vbasedev->fd < 0) {
+ if (stat(vbasedev->sysfsdev, &st) < 0) {
+ error_setg_errno(errp, errno, "no such host device");
+ error_prepend(errp, VFIO_MSG_PREFIX, vbasedev->sysfsdev);
+ return -errno;
+ }
+ /* User may specify a name, e.g: VFIO platform device */
+ if (!vbasedev->name) {
+ vbasedev->name = g_path_get_basename(vbasedev->sysfsdev);
+ }
+ }
+#ifdef CONFIG_IOMMUFD
+ else {
+ if (!vbasedev->iommufd) {
+ error_setg(errp, "Use FD passing only with iommufd backend");
+ return -EINVAL;
+ }
+ /*
+ * Give a name with fd so any function printing out vbasedev->name
+ * will not break.
+ */
+ if (!vbasedev->name) {
+ vbasedev->name = g_strdup_printf("VFIO_FD%d", vbasedev->fd);
+ }
+ }
+#endif
+ return 0;
+}
diff --git a/hw/vfio/iommufd.c b/hw/vfio/iommufd.c
index ee8c4620b6..aabc1d1024 100644
--- a/hw/vfio/iommufd.c
+++ b/hw/vfio/iommufd.c
@@ -361,11 +361,15 @@ static int iommufd_attach_device(char *name, VFIODevice *vbasedev,
uint32_t ioas_id;
Error *err = NULL;
- devfd = vfio_get_devicefd(vbasedev->sysfsdev, errp);
- if (devfd < 0) {
- return devfd;
+ if (vbasedev->fd < 0) {
+ devfd = vfio_get_devicefd(vbasedev->sysfsdev, errp);
+ if (devfd < 0) {
+ return devfd;
+ }
+ vbasedev->fd = devfd;
+ } else {
+ devfd = vbasedev->fd;
}
- vbasedev->fd = devfd;
ret = iommufd_connect_and_bind(vbasedev, errp);
if (ret) {
diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
index 15e1b771b0..edb787d3d1 100644
--- a/hw/vfio/pci.c
+++ b/hw/vfio/pci.c
@@ -44,6 +44,7 @@
#include "migration/qemu-file.h"
#include "linux/iommufd.h"
#include "sysemu/iommufd.h"
+#include "monitor/monitor.h"
#define TYPE_VFIO_PCI_NOHOTPLUG "vfio-pci-nohotplug"
@@ -3257,18 +3258,23 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
VFIODevice *vbasedev = &vdev->vbasedev;
char *tmp, *subsys;
Error *err = NULL;
- struct stat st;
int i, ret;
bool is_mdev;
char uuid[UUID_FMT_LEN];
char *name;
- if (!vbasedev->sysfsdev) {
+ if (vbasedev->fd < 0 && !vbasedev->sysfsdev) {
if (!(~vdev->host.domain || ~vdev->host.bus ||
~vdev->host.slot || ~vdev->host.function)) {
error_setg(errp, "No provided host device");
+#ifdef CONFIG_IOMMUFD
+ error_append_hint(errp, "Use -device vfio-pci,host=DDDD:BB:DD.F, "
+ "-device vfio-pci,sysfsdev=PATH_TO_DEVICE "
+ "or -device vfio-pci,fd=DEVICE_FD\n");
+#else
error_append_hint(errp, "Use -device vfio-pci,host=DDDD:BB:DD.F "
"or -device vfio-pci,sysfsdev=PATH_TO_DEVICE\n");
+#endif
return;
}
vbasedev->sysfsdev =
@@ -3277,13 +3283,9 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
vdev->host.slot, vdev->host.function);
}
- if (stat(vbasedev->sysfsdev, &st) < 0) {
- error_setg_errno(errp, errno, "no such host device");
- error_prepend(errp, VFIO_MSG_PREFIX, vbasedev->sysfsdev);
+ if (vfio_device_get_name(vbasedev, errp)) {
return;
}
-
- vbasedev->name = g_path_get_basename(vbasedev->sysfsdev);
vbasedev->ops = &vfio_pci_ops;
vbasedev->type = VFIO_DEVICE_TYPE_PCI;
vbasedev->dev = DEVICE(vdev);
@@ -3643,6 +3645,7 @@ static void vfio_instance_init(Object *obj)
vdev->host.bus = ~0U;
vdev->host.slot = ~0U;
vdev->host.function = ~0U;
+ vdev->vbasedev.fd = -1;
vdev->nv_gpudirect_clique = 0xFF;
@@ -3696,11 +3699,6 @@ static Property vfio_pci_dev_properties[] = {
qdev_prop_nv_gpudirect_clique, uint8_t),
DEFINE_PROP_OFF_AUTO_PCIBAR("x-msix-relocation", VFIOPCIDevice, msix_relo,
OFF_AUTOPCIBAR_OFF),
- /*
- * TODO - support passed fds... is this necessary?
- * DEFINE_PROP_STRING("vfiofd", VFIOPCIDevice, vfiofd_name),
- * DEFINE_PROP_STRING("vfiogroupfd, VFIOPCIDevice, vfiogroupfd_name),
- */
#ifdef CONFIG_IOMMUFD
DEFINE_PROP_LINK("iommufd", VFIOPCIDevice, vbasedev.iommufd,
TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
@@ -3708,6 +3706,21 @@ static Property vfio_pci_dev_properties[] = {
DEFINE_PROP_END_OF_LIST(),
};
+#ifdef CONFIG_IOMMUFD
+static void vfio_pci_set_fd(Object *obj, const char *str, Error **errp)
+{
+ VFIOPCIDevice *vdev = VFIO_PCI(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;
+ }
+ vdev->vbasedev.fd = fd;
+}
+#endif
+
static void vfio_pci_dev_class_init(ObjectClass *klass, void *data)
{
DeviceClass *dc = DEVICE_CLASS(klass);
@@ -3715,6 +3728,9 @@ static void vfio_pci_dev_class_init(ObjectClass *klass, void *data)
dc->reset = vfio_pci_reset;
device_class_set_props(dc, vfio_pci_dev_properties);
+#ifdef CONFIG_IOMMUFD
+ object_class_property_add_str(klass, "fd", NULL, vfio_pci_set_fd);
+#endif
dc->desc = "VFIO-based PCI device assignment";
set_bit(DEVICE_CATEGORY_MISC, dc->categories);
pdc->realize = vfio_realize;
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 24/27] vfio: Allow the selection of a given iommu backend for platform ap and ccw
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (22 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 23/27] vfio/pci: Make vfio cdev pre-openable by passing a file handle Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 25/27] vfio/platform: Make vfio cdev pre-openable by passing a file handle Zhenzhong Duan
` (2 subsequent siblings)
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Tony Krowiak, Halil Pasic, Jason Herne,
Thomas Huth, Eric Farman, Matthew Rosato, open list:vfio-ap
Previously we added support to select iommu backend for vfio pci
device. Now we added others, E.g: platform, ap and ccw.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-platform.h | 1 +
hw/vfio/ap.c | 5 +++++
hw/vfio/ccw.c | 5 +++++
hw/vfio/platform.c | 4 ++++
4 files changed, 15 insertions(+)
diff --git a/include/hw/vfio/vfio-platform.h b/include/hw/vfio/vfio-platform.h
index c414c3dffc..f57f4276f2 100644
--- a/include/hw/vfio/vfio-platform.h
+++ b/include/hw/vfio/vfio-platform.h
@@ -18,6 +18,7 @@
#include "hw/sysbus.h"
#include "hw/vfio/vfio-common.h"
+#include "sysemu/iommufd.h"
#include "qemu/event_notifier.h"
#include "qemu/queue.h"
#include "qom/object.h"
diff --git a/hw/vfio/ap.c b/hw/vfio/ap.c
index 5f257bffb9..1f8e88aeb3 100644
--- a/hw/vfio/ap.c
+++ b/hw/vfio/ap.c
@@ -16,6 +16,7 @@
#include "qapi/error.h"
#include "hw/vfio/vfio.h"
#include "hw/vfio/vfio-common.h"
+#include "sysemu/iommufd.h"
#include "hw/s390x/ap-device.h"
#include "qemu/error-report.h"
#include "qemu/event_notifier.h"
@@ -205,6 +206,10 @@ static void vfio_ap_unrealize(DeviceState *dev)
static Property vfio_ap_properties[] = {
DEFINE_PROP_STRING("sysfsdev", VFIOAPDevice, vdev.sysfsdev),
+#ifdef CONFIG_IOMMUFD
+ DEFINE_PROP_LINK("iommufd", VFIOAPDevice, vdev.iommufd,
+ TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
+#endif
DEFINE_PROP_END_OF_LIST(),
};
diff --git a/hw/vfio/ccw.c b/hw/vfio/ccw.c
index 6623ae237b..c7f8e70783 100644
--- a/hw/vfio/ccw.c
+++ b/hw/vfio/ccw.c
@@ -22,6 +22,7 @@
#include "qapi/error.h"
#include "hw/vfio/vfio.h"
#include "hw/vfio/vfio-common.h"
+#include "sysemu/iommufd.h"
#include "hw/s390x/s390-ccw.h"
#include "hw/s390x/vfio-ccw.h"
#include "hw/qdev-properties.h"
@@ -678,6 +679,10 @@ static void vfio_ccw_unrealize(DeviceState *dev)
static Property vfio_ccw_properties[] = {
DEFINE_PROP_STRING("sysfsdev", VFIOCCWDevice, vdev.sysfsdev),
DEFINE_PROP_BOOL("force-orb-pfch", VFIOCCWDevice, force_orb_pfch, false),
+#ifdef CONFIG_IOMMUFD
+ DEFINE_PROP_LINK("iommufd", VFIOCCWDevice, vdev.iommufd,
+ TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
+#endif
DEFINE_PROP_END_OF_LIST(),
};
diff --git a/hw/vfio/platform.c b/hw/vfio/platform.c
index 8e3d4ac458..a1c25e0337 100644
--- a/hw/vfio/platform.c
+++ b/hw/vfio/platform.c
@@ -649,6 +649,10 @@ static Property vfio_platform_dev_properties[] = {
DEFINE_PROP_UINT32("mmap-timeout-ms", VFIOPlatformDevice,
mmap_timeout, 1100),
DEFINE_PROP_BOOL("x-irqfd", VFIOPlatformDevice, irqfd_allowed, true),
+#ifdef CONFIG_IOMMUFD
+ DEFINE_PROP_LINK("iommufd", VFIOPlatformDevice, vbasedev.iommufd,
+ TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
+#endif
DEFINE_PROP_END_OF_LIST(),
};
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 25/27] vfio/platform: Make vfio cdev pre-openable by passing a file handle
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (23 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 24/27] vfio: Allow the selection of a given iommu backend for platform ap and ccw Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 26/27] vfio/ap: " Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 27/27] vfio/ccw: " Zhenzhong Duan
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
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.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
hw/vfio/platform.c | 41 +++++++++++++++++++++++++++++++++--------
1 file changed, 33 insertions(+), 8 deletions(-)
diff --git a/hw/vfio/platform.c b/hw/vfio/platform.c
index a1c25e0337..aa0b2b9583 100644
--- a/hw/vfio/platform.c
+++ b/hw/vfio/platform.c
@@ -35,6 +35,7 @@
#include "hw/platform-bus.h"
#include "hw/qdev-properties.h"
#include "sysemu/kvm.h"
+#include "monitor/monitor.h"
/*
* Functions used whatever the injection method
@@ -529,14 +530,13 @@ static VFIODeviceOps vfio_platform_ops = {
*/
static int vfio_base_device_init(VFIODevice *vbasedev, Error **errp)
{
- struct stat st;
int ret;
- /* @sysfsdev takes precedence over @host */
- if (vbasedev->sysfsdev) {
+ /* @fd takes precedence over @sysfsdev which takes precedence over @host */
+ if (vbasedev->fd < 0 && vbasedev->sysfsdev) {
g_free(vbasedev->name);
vbasedev->name = g_path_get_basename(vbasedev->sysfsdev);
- } else {
+ } else if (vbasedev->fd < 0) {
if (!vbasedev->name || strchr(vbasedev->name, '/')) {
error_setg(errp, "wrong host device name");
return -EINVAL;
@@ -546,10 +546,9 @@ static int vfio_base_device_init(VFIODevice *vbasedev, Error **errp)
vbasedev->name);
}
- if (stat(vbasedev->sysfsdev, &st) < 0) {
- error_setg_errno(errp, errno,
- "failed to get the sysfs host device file status");
- return -errno;
+ ret = vfio_device_get_name(vbasedev, errp);
+ if (ret) {
+ return ret;
}
ret = vfio_attach_device(vbasedev->name, vbasedev,
@@ -656,6 +655,28 @@ static Property vfio_platform_dev_properties[] = {
DEFINE_PROP_END_OF_LIST(),
};
+static void vfio_platform_instance_init(Object *obj)
+{
+ VFIOPlatformDevice *vdev = VFIO_PLATFORM_DEVICE(obj);
+
+ vdev->vbasedev.fd = -1;
+}
+
+#ifdef CONFIG_IOMMUFD
+static void vfio_platform_set_fd(Object *obj, const char *str, Error **errp)
+{
+ VFIOPlatformDevice *vdev = VFIO_PLATFORM_DEVICE(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;
+ }
+ vdev->vbasedev.fd = fd;
+}
+#endif
+
static void vfio_platform_class_init(ObjectClass *klass, void *data)
{
DeviceClass *dc = DEVICE_CLASS(klass);
@@ -663,6 +684,9 @@ static void vfio_platform_class_init(ObjectClass *klass, void *data)
dc->realize = vfio_platform_realize;
device_class_set_props(dc, vfio_platform_dev_properties);
+#ifdef CONFIG_IOMMUFD
+ object_class_property_add_str(klass, "fd", NULL, vfio_platform_set_fd);
+#endif
dc->vmsd = &vfio_platform_vmstate;
dc->desc = "VFIO-based platform device assignment";
sbc->connect_irq_notifier = vfio_start_irqfd_injection;
@@ -675,6 +699,7 @@ static const TypeInfo vfio_platform_dev_info = {
.name = TYPE_VFIO_PLATFORM,
.parent = TYPE_SYS_BUS_DEVICE,
.instance_size = sizeof(VFIOPlatformDevice),
+ .instance_init = vfio_platform_instance_init,
.class_init = vfio_platform_class_init,
.class_size = sizeof(VFIOPlatformDeviceClass),
};
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 26/27] vfio/ap: Make vfio cdev pre-openable by passing a file handle
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (24 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 25/27] vfio/platform: Make vfio cdev pre-openable by passing a file handle Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 27/27] vfio/ccw: " Zhenzhong Duan
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Thomas Huth, Tony Krowiak, Halil Pasic,
Jason Herne, open list:S390 general arch...
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.
Opportunisticly, remove some unnecessory double-cast.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
hw/vfio/ap.c | 32 +++++++++++++++++++++++++++++++-
1 file changed, 31 insertions(+), 1 deletion(-)
diff --git a/hw/vfio/ap.c b/hw/vfio/ap.c
index 1f8e88aeb3..a34cae31a2 100644
--- a/hw/vfio/ap.c
+++ b/hw/vfio/ap.c
@@ -30,6 +30,7 @@
#include "hw/s390x/ap-bridge.h"
#include "exec/address-spaces.h"
#include "qom/object.h"
+#include "monitor/monitor.h"
#define TYPE_VFIO_AP_DEVICE "vfio-ap"
@@ -160,7 +161,10 @@ static void vfio_ap_realize(DeviceState *dev, Error **errp)
VFIOAPDevice *vapdev = VFIO_AP_DEVICE(dev);
VFIODevice *vbasedev = &vapdev->vdev;
- vbasedev->name = g_path_get_basename(vbasedev->sysfsdev);
+ if (vfio_device_get_name(vbasedev, errp)) {
+ return;
+ }
+
vbasedev->ops = &vfio_ap_ops;
vbasedev->type = VFIO_DEVICE_TYPE_AP;
vbasedev->dev = dev;
@@ -230,11 +234,36 @@ static const VMStateDescription vfio_ap_vmstate = {
.unmigratable = 1,
};
+static void vfio_ap_instance_init(Object *obj)
+{
+ VFIOAPDevice *vapdev = VFIO_AP_DEVICE(obj);
+
+ vapdev->vdev.fd = -1;
+}
+
+#ifdef CONFIG_IOMMUFD
+static void vfio_ap_set_fd(Object *obj, const char *str, Error **errp)
+{
+ VFIOAPDevice *vapdev = VFIO_AP_DEVICE(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;
+ }
+ vapdev->vdev.fd = fd;
+}
+#endif
+
static void vfio_ap_class_init(ObjectClass *klass, void *data)
{
DeviceClass *dc = DEVICE_CLASS(klass);
device_class_set_props(dc, vfio_ap_properties);
+#ifdef CONFIG_IOMMUFD
+ object_class_property_add_str(klass, "fd", NULL, vfio_ap_set_fd);
+#endif
dc->vmsd = &vfio_ap_vmstate;
dc->desc = "VFIO-based AP device assignment";
set_bit(DEVICE_CATEGORY_MISC, dc->categories);
@@ -249,6 +278,7 @@ static const TypeInfo vfio_ap_info = {
.name = TYPE_VFIO_AP_DEVICE,
.parent = TYPE_AP_DEVICE,
.instance_size = sizeof(VFIOAPDevice),
+ .instance_init = vfio_ap_instance_init,
.class_init = vfio_ap_class_init,
};
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* [PATCH v2 27/27] vfio/ccw: Make vfio cdev pre-openable by passing a file handle
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
` (25 preceding siblings ...)
2023-10-16 8:32 ` [PATCH v2 26/27] vfio/ap: " Zhenzhong Duan
@ 2023-10-16 8:32 ` Zhenzhong Duan
26 siblings, 0 replies; 43+ messages in thread
From: Zhenzhong Duan @ 2023-10-16 8:32 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Thomas Huth, Eric Farman, Matthew Rosato,
open list:S390 general arch...
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.
Opportunisticly, remove a redundant definition of TYPE_VFIO_CCW.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
hw/vfio/ccw.c | 34 +++++++++++++++++++++++++++++++---
1 file changed, 31 insertions(+), 3 deletions(-)
diff --git a/hw/vfio/ccw.c b/hw/vfio/ccw.c
index c7f8e70783..f151652bc2 100644
--- a/hw/vfio/ccw.c
+++ b/hw/vfio/ccw.c
@@ -31,6 +31,7 @@
#include "qemu/error-report.h"
#include "qemu/main-loop.h"
#include "qemu/module.h"
+#include "monitor/monitor.h"
struct VFIOCCWDevice {
S390CCWDevice cdev;
@@ -590,11 +591,12 @@ static void vfio_ccw_realize(DeviceState *dev, Error **errp)
}
}
+ if (vfio_device_get_name(vbasedev, errp)) {
+ return;
+ }
+
vbasedev->ops = &vfio_ccw_ops;
vbasedev->type = VFIO_DEVICE_TYPE_CCW;
- vbasedev->name = g_strdup_printf("%x.%x.%04x", vcdev->cdev.hostid.cssid,
- vcdev->cdev.hostid.ssid,
- vcdev->cdev.hostid.devid);
vbasedev->dev = dev;
/*
@@ -691,12 +693,37 @@ static const VMStateDescription vfio_ccw_vmstate = {
.unmigratable = 1,
};
+static void vfio_ccw_instance_init(Object *obj)
+{
+ VFIOCCWDevice *vcdev = VFIO_CCW(obj);
+
+ vcdev->vdev.fd = -1;
+}
+
+#ifdef CONFIG_IOMMUFD
+static void vfio_ccw_set_fd(Object *obj, const char *str, Error **errp)
+{
+ VFIOCCWDevice *vcdev = VFIO_CCW(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;
+ }
+ vcdev->vdev.fd = fd;
+}
+#endif
+
static void vfio_ccw_class_init(ObjectClass *klass, void *data)
{
DeviceClass *dc = DEVICE_CLASS(klass);
S390CCWDeviceClass *cdc = S390_CCW_DEVICE_CLASS(klass);
device_class_set_props(dc, vfio_ccw_properties);
+#ifdef CONFIG_IOMMUFD
+ object_class_property_add_str(klass, "fd", NULL, vfio_ccw_set_fd);
+#endif
dc->vmsd = &vfio_ccw_vmstate;
dc->desc = "VFIO-based subchannel assignment";
set_bit(DEVICE_CATEGORY_MISC, dc->categories);
@@ -714,6 +741,7 @@ static const TypeInfo vfio_ccw_info = {
.name = TYPE_VFIO_CCW,
.parent = TYPE_S390_CCW,
.instance_size = sizeof(VFIOCCWDevice),
+ .instance_init = vfio_ccw_instance_init,
.class_init = vfio_ccw_class_init,
};
--
2.34.1
^ permalink raw reply related [flat|nested] 43+ messages in thread
* Re: [PATCH v2 17/27] backends/iommufd: Introduce the iommufd object
2023-10-16 8:32 ` [PATCH v2 17/27] backends/iommufd: Introduce the iommufd object Zhenzhong Duan
@ 2023-10-16 10:00 ` Markus Armbruster
2023-10-17 8:27 ` Duan, Zhenzhong
0 siblings, 1 reply; 43+ messages in thread
From: Markus Armbruster @ 2023-10-16 10:00 UTC (permalink / raw)
To: Zhenzhong Duan
Cc: qemu-devel, alex.williamson, clg, jgg, nicolinc, joao.m.martins,
eric.auger, peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun,
chao.p.peng, Paolo Bonzini, Daniel P. Berrangé,
Eduardo Habkost, Eric Blake
Zhenzhong Duan <zhenzhong.duan@intel.com> writes:
> From: Eric Auger <eric.auger@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.williamson@redhat.com>
> Signed-off-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>
> ---
> MAINTAINERS | 7 +
> qapi/qom.json | 18 ++-
> include/sysemu/iommufd.h | 46 +++++++
> backends/iommufd-stub.c | 59 +++++++++
> backends/iommufd.c | 268 +++++++++++++++++++++++++++++++++++++++
> backends/Kconfig | 4 +
> backends/meson.build | 5 +
> backends/trace-events | 12 ++
> qemu-options.hx | 13 ++
> 9 files changed, 431 insertions(+), 1 deletion(-)
> 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 9e7dec4a58..a7cdeb7825 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -2081,6 +2081,13 @@ 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>
> +S: Supported
> +F: backends/iommufd.c
> +F: include/sysemu/iommufd.h
> +
> vhost
> M: Michael S. Tsirkin <mst@redhat.com>
> S: Supported
> diff --git a/qapi/qom.json b/qapi/qom.json
> index c53ef978ff..3f964e57f5 100644
> --- a/qapi/qom.json
> +++ b/qapi/qom.json
> @@ -794,6 +794,18 @@
> { 'struct': 'VfioUserServerProperties',
> 'data': { 'socket': 'SocketAddress', 'device': 'str' } }
>
> +##
> +# @IOMMUFDProperties:
> +#
> +# Properties for IOMMUFDbackend objects.
This is the only occurence of "IOMMUFDbackend". Do you mean iommufd?
> +#
> +# fd: file descriptor name
Awfully terse. Like many existing FOOProperties. Would it make sense
to document what objects of type iommufd are good for, and how they use
@fd? Or is it already documented elsewhere?
> +#
> +# Since: 7.2
Since: 8.2.
> +##
> +{ 'struct': 'IOMMUFDProperties',
> + 'data': { '*fd': 'str' } }
> +
> ##
> # @RngProperties:
> #
> @@ -948,6 +960,8 @@
> 'qtest',
> 'rng-builtin',
> 'rng-egd',
> + { 'name': 'iommufd',
> + 'if': 'CONFIG_IOMMUFD' },
> { 'name': 'rng-random',
> 'if': 'CONFIG_POSIX' },
> 'secret',
Please keep the members sorted.
> @@ -1029,7 +1043,9 @@
> 'tls-creds-x509': 'TlsCredsX509Properties',
> 'tls-cipher-suites': 'TlsCredsProperties',
> 'x-remote-object': 'RemoteObjectProperties',
> - 'x-vfio-user-server': 'VfioUserServerProperties'
> + 'x-vfio-user-server': 'VfioUserServerProperties',
> + 'iommufd': { 'type': 'IOMMUFDProperties',
> + 'if': 'CONFIG_IOMMUFD' }
Likewise.
> } }
>
> ##
[...]
^ permalink raw reply [flat|nested] 43+ messages in thread
* RE: [PATCH v2 17/27] backends/iommufd: Introduce the iommufd object
2023-10-16 10:00 ` Markus Armbruster
@ 2023-10-17 8:27 ` Duan, Zhenzhong
0 siblings, 0 replies; 43+ messages in thread
From: Duan, Zhenzhong @ 2023-10-17 8:27 UTC (permalink / raw)
To: Markus Armbruster
Cc: qemu-devel, alex.williamson, clg, jgg, nicolinc, Martins, Joao,
eric.auger, peterx, jasowang, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Paolo Bonzini, Daniel P.Berrangé,
Eduardo Habkost, Eric Blake
Hi Markus,
>-----Original Message-----
>From: Markus Armbruster <armbru@redhat.com>
>Sent: Monday, October 16, 2023 6:01 PM
>Subject: Re: [PATCH v2 17/27] backends/iommufd: Introduce the iommufd object
>
>Zhenzhong Duan <zhenzhong.duan@intel.com> writes:
>
>> From: Eric Auger <eric.auger@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.williamson@redhat.com>
>> Signed-off-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>
>> ---
>> MAINTAINERS | 7 +
>> qapi/qom.json | 18 ++-
>> include/sysemu/iommufd.h | 46 +++++++
>> backends/iommufd-stub.c | 59 +++++++++
>> backends/iommufd.c | 268 +++++++++++++++++++++++++++++++++++++++
>> backends/Kconfig | 4 +
>> backends/meson.build | 5 +
>> backends/trace-events | 12 ++
>> qemu-options.hx | 13 ++
>> 9 files changed, 431 insertions(+), 1 deletion(-)
>> 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 9e7dec4a58..a7cdeb7825 100644
>> --- a/MAINTAINERS
>> +++ b/MAINTAINERS
>> @@ -2081,6 +2081,13 @@ 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>
>> +S: Supported
>> +F: backends/iommufd.c
>> +F: include/sysemu/iommufd.h
>> +
>> vhost
>> M: Michael S. Tsirkin <mst@redhat.com>
>> S: Supported
>> diff --git a/qapi/qom.json b/qapi/qom.json
>> index c53ef978ff..3f964e57f5 100644
>> --- a/qapi/qom.json
>> +++ b/qapi/qom.json
>> @@ -794,6 +794,18 @@
>> { 'struct': 'VfioUserServerProperties',
>> 'data': { 'socket': 'SocketAddress', 'device': 'str' } }
>>
>> +##
>> +# @IOMMUFDProperties:
>> +#
>> +# Properties for IOMMUFDbackend objects.
>
>This is the only occurence of "IOMMUFDbackend". Do you mean iommufd?
Yes, will fix.
>
>> +#
>> +# fd: file descriptor name
>
>Awfully terse. Like many existing FOOProperties. Would it make sense
>to document what objects of type iommufd are good for, and how they use
>@fd? Or is it already documented elsewhere?
Will add more comments here, thanks for point out.
>
>> +#
>> +# Since: 7.2
>
>Since: 8.2.
Yes.
>
>> +##
>> +{ 'struct': 'IOMMUFDProperties',
>> + 'data': { '*fd': 'str' } }
>> +
>> ##
>> # @RngProperties:
>> #
>> @@ -948,6 +960,8 @@
>> 'qtest',
>> 'rng-builtin',
>> 'rng-egd',
>> + { 'name': 'iommufd',
>> + 'if': 'CONFIG_IOMMUFD' },
>> { 'name': 'rng-random',
>> 'if': 'CONFIG_POSIX' },
>> 'secret',
>
>Please keep the members sorted.
Will fix.
>
>> @@ -1029,7 +1043,9 @@
>> 'tls-creds-x509': 'TlsCredsX509Properties',
>> 'tls-cipher-suites': 'TlsCredsProperties',
>> 'x-remote-object': 'RemoteObjectProperties',
>> - 'x-vfio-user-server': 'VfioUserServerProperties'
>> + 'x-vfio-user-server': 'VfioUserServerProperties',
>> + 'iommufd': { 'type': 'IOMMUFDProperties',
>> + 'if': 'CONFIG_IOMMUFD' }
>
>Likewise.
Will fix.
Thanks
Zhenzhong
>
>> } }
>>
>> ##
>
>[...]
^ permalink raw reply [flat|nested] 43+ messages in thread
* Re: [PATCH v2 01/27] vfio: Rename VFIOContainer into VFIOLegacyContainer
2023-10-16 8:31 ` [PATCH v2 01/27] vfio: Rename VFIOContainer into VFIOLegacyContainer Zhenzhong Duan
@ 2023-10-17 15:50 ` Cédric Le Goater
2023-10-18 2:33 ` Duan, Zhenzhong
0 siblings, 1 reply; 43+ messages in thread
From: Cédric Le Goater @ 2023-10-17 15:50 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Nicholas Piggin, Daniel Henrique Barboza, David Gibson,
Harsh Prateek Bora, open list:sPAPR (pseries)
Hello,
On 10/16/23 10:31, Zhenzhong Duan wrote:
> From: Eric Auger <eric.auger@redhat.com>
>
> In the prospect to introduce a base object for the VFIOContainer
> and derive into the existing legacy container and the iommufd
> based container, let's rename the existing one into
> VFIOLegacyContainer. This is just an incremental step to ease
> the migration. Soon there won't be any reference to the legacy
> container in the common.c code. Only the container.c should
> handle the VFIOLegacyContainer object.
So I think I would have kept the current name as it is, named the
new abtract QOM object VFIOContainerBase and then, derived from
VFIOContainerBase, the current implementation VFIOContainer and
the new one VFIOContainerIOMMUFD.
I just starting the review, don't mind too much my comments until
PATCH 1-15 are covered.
Thanks,
C.
>
> No functional change intended.
>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---
> include/hw/vfio/vfio-common.h | 46 ++++++++++++-------------
> hw/vfio/common.c | 63 ++++++++++++++++++++---------------
> hw/vfio/container.c | 45 +++++++++++++------------
> hw/vfio/spapr.c | 12 +++----
> 4 files changed, 89 insertions(+), 77 deletions(-)
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index 7780b9073a..34648e518e 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -74,13 +74,13 @@ typedef struct VFIOMigration {
>
> typedef struct VFIOAddressSpace {
> AddressSpace *as;
> - QLIST_HEAD(, VFIOContainer) containers;
> + QLIST_HEAD(, VFIOLegacyContainer) containers;
> QLIST_ENTRY(VFIOAddressSpace) list;
> } VFIOAddressSpace;
>
> struct VFIOGroup;
>
> -typedef struct VFIOContainer {
> +typedef struct VFIOLegacyContainer {
> VFIOAddressSpace *space;
> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
> MemoryListener listener;
> @@ -97,12 +97,12 @@ typedef struct VFIOContainer {
> QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
> QLIST_HEAD(, VFIOGroup) group_list;
> QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
> - QLIST_ENTRY(VFIOContainer) next;
> + QLIST_ENTRY(VFIOLegacyContainer) next;
> QLIST_HEAD(, VFIODevice) device_list;
> -} VFIOContainer;
> +} VFIOLegacyContainer;
>
> typedef struct VFIOGuestIOMMU {
> - VFIOContainer *container;
> + VFIOLegacyContainer *container;
> IOMMUMemoryRegion *iommu_mr;
> hwaddr iommu_offset;
> IOMMUNotifier n;
> @@ -110,7 +110,7 @@ typedef struct VFIOGuestIOMMU {
> } VFIOGuestIOMMU;
>
> typedef struct VFIORamDiscardListener {
> - VFIOContainer *container;
> + VFIOLegacyContainer *container;
> MemoryRegion *mr;
> hwaddr offset_within_address_space;
> hwaddr size;
> @@ -133,7 +133,7 @@ typedef struct VFIODevice {
> QLIST_ENTRY(VFIODevice) container_next;
> QLIST_ENTRY(VFIODevice) global_next;
> struct VFIOGroup *group;
> - VFIOContainer *container;
> + VFIOLegacyContainer *container;
> char *sysfsdev;
> char *name;
> DeviceState *dev;
> @@ -167,7 +167,7 @@ struct VFIODeviceOps {
> typedef struct VFIOGroup {
> int fd;
> int groupid;
> - VFIOContainer *container;
> + VFIOLegacyContainer *container;
> QLIST_HEAD(, VFIODevice) device_list;
> QLIST_ENTRY(VFIOGroup) next;
> QLIST_ENTRY(VFIOGroup) container_next;
> @@ -206,28 +206,28 @@ typedef struct {
> hwaddr pages;
> } VFIOBitmap;
>
> -void vfio_host_win_add(VFIOContainer *container,
> +void vfio_host_win_add(VFIOLegacyContainer *container,
> hwaddr min_iova, hwaddr max_iova,
> uint64_t iova_pgsizes);
> -int vfio_host_win_del(VFIOContainer *container, hwaddr min_iova,
> +int vfio_host_win_del(VFIOLegacyContainer *container, hwaddr min_iova,
> hwaddr max_iova);
> VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
> void vfio_put_address_space(VFIOAddressSpace *space);
> -bool vfio_devices_all_running_and_saving(VFIOContainer *container);
> +bool vfio_devices_all_running_and_saving(VFIOLegacyContainer *container);
>
> /* container->fd */
> -int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
> +int vfio_dma_unmap(VFIOLegacyContainer *container, hwaddr iova,
> ram_addr_t size, IOMMUTLBEntry *iotlb);
> -int vfio_dma_map(VFIOContainer *container, hwaddr iova,
> +int vfio_dma_map(VFIOLegacyContainer *container, hwaddr iova,
> ram_addr_t size, void *vaddr, bool readonly);
> -int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start);
> -int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
> +int vfio_set_dirty_page_tracking(VFIOLegacyContainer *container, bool start);
> +int vfio_query_dirty_bitmap(VFIOLegacyContainer *container, VFIOBitmap *vbmap,
> hwaddr iova, hwaddr size);
>
> -int vfio_container_add_section_window(VFIOContainer *container,
> +int vfio_container_add_section_window(VFIOLegacyContainer *container,
> MemoryRegionSection *section,
> Error **errp);
> -void vfio_container_del_section_window(VFIOContainer *container,
> +void vfio_container_del_section_window(VFIOLegacyContainer *container,
> MemoryRegionSection *section);
>
> void vfio_disable_irqindex(VFIODevice *vbasedev, int index);
> @@ -290,21 +290,21 @@ vfio_get_cap(void *ptr, uint32_t cap_offset, uint16_t id);
> #endif
> extern const MemoryListener vfio_prereg_listener;
>
> -int vfio_spapr_create_window(VFIOContainer *container,
> +int vfio_spapr_create_window(VFIOLegacyContainer *container,
> MemoryRegionSection *section,
> hwaddr *pgsize);
> -int vfio_spapr_remove_window(VFIOContainer *container,
> +int vfio_spapr_remove_window(VFIOLegacyContainer *container,
> hwaddr offset_within_address_space);
>
> bool vfio_migration_realize(VFIODevice *vbasedev, Error **errp);
> void vfio_migration_exit(VFIODevice *vbasedev);
>
> int vfio_bitmap_alloc(VFIOBitmap *vbmap, hwaddr size);
> -bool vfio_devices_all_running_and_mig_active(VFIOContainer *container);
> -bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container);
> -int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
> +bool vfio_devices_all_running_and_mig_active(VFIOLegacyContainer *container);
> +bool vfio_devices_all_device_dirty_tracking(VFIOLegacyContainer *container);
> +int vfio_devices_query_dirty_bitmap(VFIOLegacyContainer *container,
> VFIOBitmap *vbmap, hwaddr iova,
> hwaddr size);
> -int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
> +int vfio_get_dirty_bitmap(VFIOLegacyContainer *container, uint64_t iova,
> uint64_t size, ram_addr_t ram_addr);
> #endif /* HW_VFIO_VFIO_COMMON_H */
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index 5ff5acf1d8..b51ef3a15a 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -184,7 +184,7 @@ bool vfio_device_state_is_precopy(VFIODevice *vbasedev)
> migration->device_state == VFIO_DEVICE_STATE_PRE_COPY_P2P;
> }
>
> -static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
> +static bool vfio_devices_all_dirty_tracking(VFIOLegacyContainer *container)
> {
> VFIODevice *vbasedev;
> MigrationState *ms = migrate_get_current();
> @@ -210,7 +210,7 @@ static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
> return true;
> }
>
> -bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container)
> +bool vfio_devices_all_device_dirty_tracking(VFIOLegacyContainer *container)
> {
> VFIODevice *vbasedev;
>
> @@ -227,7 +227,7 @@ bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container)
> * Check if all VFIO devices are running and migration is active, which is
> * essentially equivalent to the migration being in pre-copy phase.
> */
> -bool vfio_devices_all_running_and_mig_active(VFIOContainer *container)
> +bool vfio_devices_all_running_and_mig_active(VFIOLegacyContainer *container)
> {
> VFIODevice *vbasedev;
>
> @@ -252,7 +252,7 @@ bool vfio_devices_all_running_and_mig_active(VFIOContainer *container)
> return true;
> }
>
> -void vfio_host_win_add(VFIOContainer *container, hwaddr min_iova,
> +void vfio_host_win_add(VFIOLegacyContainer *container, hwaddr min_iova,
> hwaddr max_iova, uint64_t iova_pgsizes)
> {
> VFIOHostDMAWindow *hostwin;
> @@ -274,7 +274,7 @@ void vfio_host_win_add(VFIOContainer *container, hwaddr min_iova,
> QLIST_INSERT_HEAD(&container->hostwin_list, hostwin, hostwin_next);
> }
>
> -int vfio_host_win_del(VFIOContainer *container,
> +int vfio_host_win_del(VFIOLegacyContainer *container,
> hwaddr min_iova, hwaddr max_iova)
> {
> VFIOHostDMAWindow *hostwin;
> @@ -337,7 +337,7 @@ static bool vfio_get_xlat_addr(IOMMUTLBEntry *iotlb, void **vaddr,
> static void vfio_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
> {
> VFIOGuestIOMMU *giommu = container_of(n, VFIOGuestIOMMU, n);
> - VFIOContainer *container = giommu->container;
> + VFIOLegacyContainer *container = giommu->container;
> hwaddr iova = iotlb->iova + giommu->iommu_offset;
> void *vaddr;
> int ret;
> @@ -441,7 +441,7 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
> return 0;
> }
>
> -static void vfio_register_ram_discard_listener(VFIOContainer *container,
> +static void vfio_register_ram_discard_listener(VFIOLegacyContainer *container,
> MemoryRegionSection *section)
> {
> RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
> @@ -515,7 +515,7 @@ static void vfio_register_ram_discard_listener(VFIOContainer *container,
> }
> }
>
> -static void vfio_unregister_ram_discard_listener(VFIOContainer *container,
> +static void vfio_unregister_ram_discard_listener(VFIOLegacyContainer *container,
> MemoryRegionSection *section)
> {
> RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
> @@ -538,7 +538,7 @@ static void vfio_unregister_ram_discard_listener(VFIOContainer *container,
> g_free(vrdl);
> }
>
> -static VFIOHostDMAWindow *vfio_find_hostwin(VFIOContainer *container,
> +static VFIOHostDMAWindow *vfio_find_hostwin(VFIOLegacyContainer *container,
> hwaddr iova, hwaddr end)
> {
> VFIOHostDMAWindow *hostwin;
> @@ -599,7 +599,7 @@ static bool vfio_listener_valid_section(MemoryRegionSection *section,
> return true;
> }
>
> -static bool vfio_get_section_iova_range(VFIOContainer *container,
> +static bool vfio_get_section_iova_range(VFIOLegacyContainer *container,
> MemoryRegionSection *section,
> hwaddr *out_iova, hwaddr *out_end,
> Int128 *out_llend)
> @@ -627,7 +627,9 @@ static bool vfio_get_section_iova_range(VFIOContainer *container,
> static void vfio_listener_region_add(MemoryListener *listener,
> MemoryRegionSection *section)
> {
> - VFIOContainer *container = container_of(listener, VFIOContainer, listener);
> + VFIOLegacyContainer *container = container_of(listener,
> + VFIOLegacyContainer,
> + listener);
> hwaddr iova, end;
> Int128 llend, llsize;
> void *vaddr;
> @@ -788,7 +790,9 @@ fail:
> static void vfio_listener_region_del(MemoryListener *listener,
> MemoryRegionSection *section)
> {
> - VFIOContainer *container = container_of(listener, VFIOContainer, listener);
> + VFIOLegacyContainer *container = container_of(listener,
> + VFIOLegacyContainer,
> + listener);
> hwaddr iova, end;
> Int128 llend, llsize;
> int ret;
> @@ -881,13 +885,13 @@ typedef struct VFIODirtyRanges {
> } VFIODirtyRanges;
>
> typedef struct VFIODirtyRangesListener {
> - VFIOContainer *container;
> + VFIOLegacyContainer *container;
> VFIODirtyRanges ranges;
> MemoryListener listener;
> } VFIODirtyRangesListener;
>
> static bool vfio_section_is_vfio_pci(MemoryRegionSection *section,
> - VFIOContainer *container)
> + VFIOLegacyContainer *container)
> {
> VFIOPCIDevice *pcidev;
> VFIODevice *vbasedev;
> @@ -966,7 +970,7 @@ static const MemoryListener vfio_dirty_tracking_listener = {
> .region_add = vfio_dirty_tracking_update,
> };
>
> -static void vfio_dirty_tracking_init(VFIOContainer *container,
> +static void vfio_dirty_tracking_init(VFIOLegacyContainer *container,
> VFIODirtyRanges *ranges)
> {
> VFIODirtyRangesListener dirty;
> @@ -991,7 +995,7 @@ static void vfio_dirty_tracking_init(VFIOContainer *container,
> memory_listener_unregister(&dirty.listener);
> }
>
> -static void vfio_devices_dma_logging_stop(VFIOContainer *container)
> +static void vfio_devices_dma_logging_stop(VFIOLegacyContainer *container)
> {
> uint64_t buf[DIV_ROUND_UP(sizeof(struct vfio_device_feature),
> sizeof(uint64_t))] = {};
> @@ -1016,7 +1020,7 @@ static void vfio_devices_dma_logging_stop(VFIOContainer *container)
> }
>
> static struct vfio_device_feature *
> -vfio_device_feature_dma_logging_start_create(VFIOContainer *container,
> +vfio_device_feature_dma_logging_start_create(VFIOLegacyContainer *container,
> VFIODirtyRanges *tracking)
> {
> struct vfio_device_feature *feature;
> @@ -1089,7 +1093,7 @@ static void vfio_device_feature_dma_logging_start_destroy(
> g_free(feature);
> }
>
> -static int vfio_devices_dma_logging_start(VFIOContainer *container)
> +static int vfio_devices_dma_logging_start(VFIOLegacyContainer *container)
> {
> struct vfio_device_feature *feature;
> VFIODirtyRanges ranges;
> @@ -1130,7 +1134,9 @@ out:
>
> static void vfio_listener_log_global_start(MemoryListener *listener)
> {
> - VFIOContainer *container = container_of(listener, VFIOContainer, listener);
> + VFIOLegacyContainer *container = container_of(listener,
> + VFIOLegacyContainer,
> + listener);
> int ret;
>
> if (vfio_devices_all_device_dirty_tracking(container)) {
> @@ -1148,7 +1154,9 @@ static void vfio_listener_log_global_start(MemoryListener *listener)
>
> static void vfio_listener_log_global_stop(MemoryListener *listener)
> {
> - VFIOContainer *container = container_of(listener, VFIOContainer, listener);
> + VFIOLegacyContainer *container = container_of(listener,
> + VFIOLegacyContainer,
> + listener);
> int ret = 0;
>
> if (vfio_devices_all_device_dirty_tracking(container)) {
> @@ -1190,7 +1198,7 @@ static int vfio_device_dma_logging_report(VFIODevice *vbasedev, hwaddr iova,
> return 0;
> }
>
> -int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
> +int vfio_devices_query_dirty_bitmap(VFIOLegacyContainer *container,
> VFIOBitmap *vbmap, hwaddr iova,
> hwaddr size)
> {
> @@ -1213,7 +1221,7 @@ int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
> return 0;
> }
>
> -int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
> +int vfio_get_dirty_bitmap(VFIOLegacyContainer *container, uint64_t iova,
> uint64_t size, ram_addr_t ram_addr)
> {
> bool all_device_dirty_tracking =
> @@ -1265,7 +1273,7 @@ static void vfio_iommu_map_dirty_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
> vfio_giommu_dirty_notifier *gdn = container_of(n,
> vfio_giommu_dirty_notifier, n);
> VFIOGuestIOMMU *giommu = gdn->giommu;
> - VFIOContainer *container = giommu->container;
> + VFIOLegacyContainer *container = giommu->container;
> hwaddr iova = iotlb->iova + giommu->iommu_offset;
> ram_addr_t translated_addr;
> int ret = -EINVAL;
> @@ -1313,7 +1321,8 @@ static int vfio_ram_discard_get_dirty_bitmap(MemoryRegionSection *section,
> return vfio_get_dirty_bitmap(vrdl->container, iova, size, ram_addr);
> }
>
> -static int vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainer *container,
> +static int
> +vfio_sync_ram_discard_listener_dirty_bitmap(VFIOLegacyContainer *container,
> MemoryRegionSection *section)
> {
> RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
> @@ -1340,7 +1349,7 @@ static int vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainer *container,
> &vrdl);
> }
>
> -static int vfio_sync_dirty_bitmap(VFIOContainer *container,
> +static int vfio_sync_dirty_bitmap(VFIOLegacyContainer *container,
> MemoryRegionSection *section)
> {
> ram_addr_t ram_addr;
> @@ -1386,7 +1395,9 @@ static int vfio_sync_dirty_bitmap(VFIOContainer *container,
> static void vfio_listener_log_sync(MemoryListener *listener,
> MemoryRegionSection *section)
> {
> - VFIOContainer *container = container_of(listener, VFIOContainer, listener);
> + VFIOLegacyContainer *container = container_of(listener,
> + VFIOLegacyContainer,
> + listener);
> int ret;
>
> if (vfio_listener_skipped_section(section)) {
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index adc467210f..8fde302ae9 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -42,7 +42,8 @@
> VFIOGroupList vfio_group_list =
> QLIST_HEAD_INITIALIZER(vfio_group_list);
>
> -static int vfio_ram_block_discard_disable(VFIOContainer *container, bool state)
> +static int vfio_ram_block_discard_disable(VFIOLegacyContainer *container,
> + bool state)
> {
> switch (container->iommu_type) {
> case VFIO_TYPE1v2_IOMMU:
> @@ -65,7 +66,7 @@ static int vfio_ram_block_discard_disable(VFIOContainer *container, bool state)
> }
> }
>
> -static int vfio_dma_unmap_bitmap(VFIOContainer *container,
> +static int vfio_dma_unmap_bitmap(VFIOLegacyContainer *container,
> hwaddr iova, ram_addr_t size,
> IOMMUTLBEntry *iotlb)
> {
> @@ -120,7 +121,7 @@ unmap_exit:
> /*
> * DMA - Mapping and unmapping for the "type1" IOMMU interface used on x86
> */
> -int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
> +int vfio_dma_unmap(VFIOLegacyContainer *container, hwaddr iova,
> ram_addr_t size, IOMMUTLBEntry *iotlb)
> {
> struct vfio_iommu_type1_dma_unmap unmap = {
> @@ -175,7 +176,7 @@ int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
> return 0;
> }
>
> -int vfio_dma_map(VFIOContainer *container, hwaddr iova,
> +int vfio_dma_map(VFIOLegacyContainer *container, hwaddr iova,
> ram_addr_t size, void *vaddr, bool readonly)
> {
> struct vfio_iommu_type1_dma_map map = {
> @@ -205,7 +206,7 @@ int vfio_dma_map(VFIOContainer *container, hwaddr iova,
> return -errno;
> }
>
> -int vfio_container_add_section_window(VFIOContainer *container,
> +int vfio_container_add_section_window(VFIOLegacyContainer *container,
> MemoryRegionSection *section,
> Error **errp)
> {
> @@ -273,7 +274,7 @@ int vfio_container_add_section_window(VFIOContainer *container,
> return 0;
> }
>
> -void vfio_container_del_section_window(VFIOContainer *container,
> +void vfio_container_del_section_window(VFIOLegacyContainer *container,
> MemoryRegionSection *section)
> {
> if (container->iommu_type != VFIO_SPAPR_TCE_v2_IOMMU) {
> @@ -291,7 +292,7 @@ void vfio_container_del_section_window(VFIOContainer *container,
> }
> }
>
> -int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start)
> +int vfio_set_dirty_page_tracking(VFIOLegacyContainer *container, bool start)
> {
> int ret;
> struct vfio_iommu_type1_dirty_bitmap dirty = {
> @@ -318,7 +319,7 @@ int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start)
> return ret;
> }
>
> -int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
> +int vfio_query_dirty_bitmap(VFIOLegacyContainer *container, VFIOBitmap *vbmap,
> hwaddr iova, hwaddr size)
> {
> struct vfio_iommu_type1_dirty_bitmap *dbitmap;
> @@ -355,7 +356,7 @@ int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
> return ret;
> }
>
> -static void vfio_listener_release(VFIOContainer *container)
> +static void vfio_listener_release(VFIOLegacyContainer *container)
> {
> memory_listener_unregister(&container->listener);
> if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
> @@ -415,7 +416,7 @@ static void vfio_kvm_device_del_group(VFIOGroup *group)
> /*
> * vfio_get_iommu_type - selects the richest iommu_type (v2 first)
> */
> -static int vfio_get_iommu_type(VFIOContainer *container,
> +static int vfio_get_iommu_type(VFIOLegacyContainer *container,
> Error **errp)
> {
> int iommu_types[] = { VFIO_TYPE1v2_IOMMU, VFIO_TYPE1_IOMMU,
> @@ -431,7 +432,7 @@ static int vfio_get_iommu_type(VFIOContainer *container,
> return -EINVAL;
> }
>
> -static int vfio_init_container(VFIOContainer *container, int group_fd,
> +static int vfio_init_container(VFIOLegacyContainer *container, int group_fd,
> Error **errp)
> {
> int iommu_type, ret;
> @@ -466,7 +467,7 @@ static int vfio_init_container(VFIOContainer *container, int group_fd,
> return 0;
> }
>
> -static int vfio_get_iommu_info(VFIOContainer *container,
> +static int vfio_get_iommu_info(VFIOLegacyContainer *container,
> struct vfio_iommu_type1_info **info)
> {
>
> @@ -510,7 +511,7 @@ vfio_get_iommu_info_cap(struct vfio_iommu_type1_info *info, uint16_t id)
> return NULL;
> }
>
> -static void vfio_get_iommu_info_migration(VFIOContainer *container,
> +static void vfio_get_iommu_info_migration(VFIOLegacyContainer *container,
> struct vfio_iommu_type1_info *info)
> {
> struct vfio_info_cap_header *hdr;
> @@ -538,7 +539,7 @@ static void vfio_get_iommu_info_migration(VFIOContainer *container,
> static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> Error **errp)
> {
> - VFIOContainer *container;
> + VFIOLegacyContainer *container;
> int ret, fd;
> VFIOAddressSpace *space;
>
> @@ -778,7 +779,7 @@ put_space_exit:
>
> static void vfio_disconnect_container(VFIOGroup *group)
> {
> - VFIOContainer *container = group->container;
> + VFIOLegacyContainer *container = group->container;
>
> QLIST_REMOVE(group, container_next);
> group->container = NULL;
> @@ -978,7 +979,7 @@ static void vfio_put_base_device(VFIODevice *vbasedev)
> /*
> * Interfaces for IBM EEH (Enhanced Error Handling)
> */
> -static bool vfio_eeh_container_ok(VFIOContainer *container)
> +static bool vfio_eeh_container_ok(VFIOLegacyContainer *container)
> {
> /*
> * As of 2016-03-04 (linux-4.5) the host kernel EEH/VFIO
> @@ -1006,7 +1007,7 @@ static bool vfio_eeh_container_ok(VFIOContainer *container)
> return true;
> }
>
> -static int vfio_eeh_container_op(VFIOContainer *container, uint32_t op)
> +static int vfio_eeh_container_op(VFIOLegacyContainer *container, uint32_t op)
> {
> struct vfio_eeh_pe_op pe_op = {
> .argsz = sizeof(pe_op),
> @@ -1029,10 +1030,10 @@ static int vfio_eeh_container_op(VFIOContainer *container, uint32_t op)
> return ret;
> }
>
> -static VFIOContainer *vfio_eeh_as_container(AddressSpace *as)
> +static VFIOLegacyContainer *vfio_eeh_as_container(AddressSpace *as)
> {
> VFIOAddressSpace *space = vfio_get_address_space(as);
> - VFIOContainer *container = NULL;
> + VFIOLegacyContainer *container = NULL;
>
> if (QLIST_EMPTY(&space->containers)) {
> /* No containers to act on */
> @@ -1057,14 +1058,14 @@ out:
>
> bool vfio_eeh_as_ok(AddressSpace *as)
> {
> - VFIOContainer *container = vfio_eeh_as_container(as);
> + VFIOLegacyContainer *container = vfio_eeh_as_container(as);
>
> return (container != NULL) && vfio_eeh_container_ok(container);
> }
>
> int vfio_eeh_as_op(AddressSpace *as, uint32_t op)
> {
> - VFIOContainer *container = vfio_eeh_as_container(as);
> + VFIOLegacyContainer *container = vfio_eeh_as_container(as);
>
> if (!container) {
> return -ENODEV;
> @@ -1109,7 +1110,7 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
> int groupid = vfio_device_groupid(vbasedev, errp);
> VFIODevice *vbasedev_iter;
> VFIOGroup *group;
> - VFIOContainer *container;
> + VFIOLegacyContainer *container;
> int ret;
>
> if (groupid < 0) {
> diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
> index 9ec1e95f6d..683252c506 100644
> --- a/hw/vfio/spapr.c
> +++ b/hw/vfio/spapr.c
> @@ -39,8 +39,8 @@ static void *vfio_prereg_gpa_to_vaddr(MemoryRegionSection *section, hwaddr gpa)
> static void vfio_prereg_listener_region_add(MemoryListener *listener,
> MemoryRegionSection *section)
> {
> - VFIOContainer *container = container_of(listener, VFIOContainer,
> - prereg_listener);
> + VFIOLegacyContainer *container = container_of(listener, VFIOLegacyContainer,
> + prereg_listener);
> const hwaddr gpa = section->offset_within_address_space;
> hwaddr end;
> int ret;
> @@ -97,8 +97,8 @@ static void vfio_prereg_listener_region_add(MemoryListener *listener,
> static void vfio_prereg_listener_region_del(MemoryListener *listener,
> MemoryRegionSection *section)
> {
> - VFIOContainer *container = container_of(listener, VFIOContainer,
> - prereg_listener);
> + VFIOLegacyContainer *container = container_of(listener, VFIOLegacyContainer,
> + prereg_listener);
> const hwaddr gpa = section->offset_within_address_space;
> hwaddr end;
> int ret;
> @@ -141,7 +141,7 @@ const MemoryListener vfio_prereg_listener = {
> .region_del = vfio_prereg_listener_region_del,
> };
>
> -int vfio_spapr_create_window(VFIOContainer *container,
> +int vfio_spapr_create_window(VFIOLegacyContainer *container,
> MemoryRegionSection *section,
> hwaddr *pgsize)
> {
> @@ -233,7 +233,7 @@ int vfio_spapr_create_window(VFIOContainer *container,
> return 0;
> }
>
> -int vfio_spapr_remove_window(VFIOContainer *container,
> +int vfio_spapr_remove_window(VFIOLegacyContainer *container,
> hwaddr offset_within_address_space)
> {
> struct vfio_iommu_spapr_tce_remove remove = {
^ permalink raw reply [flat|nested] 43+ messages in thread
* Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-16 8:31 ` [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface Zhenzhong Duan
@ 2023-10-17 15:51 ` Cédric Le Goater
2023-10-18 2:41 ` Duan, Zhenzhong
0 siblings, 1 reply; 43+ messages in thread
From: Cédric Le Goater @ 2023-10-17 15:51 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun
On 10/16/23 10:31, Zhenzhong Duan wrote:
> From: Eric Auger <eric.auger@redhat.com>
>
> Introduce a dumb VFIOContainer base object and its targetted interface.
> This is willingly not a QOM object because we don't want it to be
> visible from the user interface. The VFIOContainer will be smoothly
> populated in subsequent patches as well as interfaces.
>
> No fucntional change intended.
>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---
> include/hw/vfio/vfio-common.h | 8 +--
> include/hw/vfio/vfio-container-base.h | 82 +++++++++++++++++++++++++++
> 2 files changed, 84 insertions(+), 6 deletions(-)
> create mode 100644 include/hw/vfio/vfio-container-base.h
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index 34648e518e..9651cf921c 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -30,6 +30,7 @@
> #include <linux/vfio.h>
> #endif
> #include "sysemu/sysemu.h"
> +#include "hw/vfio/vfio-container-base.h"
>
> #define VFIO_MSG_PREFIX "vfio %s: "
>
> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
> struct VFIOGroup;
>
> typedef struct VFIOLegacyContainer {
> + VFIOContainer bcontainer;
That's the parent class, right ?
> VFIOAddressSpace *space;
> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
> MemoryListener listener;
> @@ -200,12 +202,6 @@ typedef struct VFIODisplay {
> } dmabuf;
> } VFIODisplay;
>
> -typedef struct {
> - unsigned long *bitmap;
> - hwaddr size;
> - hwaddr pages;
> -} VFIOBitmap;
> -
> void vfio_host_win_add(VFIOLegacyContainer *container,
> hwaddr min_iova, hwaddr max_iova,
> uint64_t iova_pgsizes);
> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
> new file mode 100644
> index 0000000000..afc8543d22
> --- /dev/null
> +++ b/include/hw/vfio/vfio-container-base.h
> @@ -0,0 +1,82 @@
> +/*
> + * VFIO BASE CONTAINER
> + *
> + * Copyright (C) 2023 Intel Corporation.
> + * Copyright Red Hat, Inc. 2023
> + *
> + * Authors: Yi Liu <yi.l.liu@intel.com>
> + * Eric Auger <eric.auger@redhat.com>
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> +
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> +
> + * You should have received a copy of the GNU General Public License along
> + * with this program; if not, see <http://www.gnu.org/licenses/>.
> + */
> +
> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
> +
> +#include "exec/memory.h"
> +#ifndef CONFIG_USER_ONLY
> +#include "exec/hwaddr.h"
> +#endif
> +
> +typedef struct VFIOContainer VFIOContainer;
> +typedef struct VFIODevice VFIODevice;
> +typedef struct VFIOIOMMUBackendOpsClass VFIOIOMMUBackendOpsClass;
> +
> +typedef struct {
> + unsigned long *bitmap;
> + hwaddr size;
> + hwaddr pages;
> +} VFIOBitmap;
> +
> +/*
> + * This is the base object for vfio container backends
> + */
> +struct VFIOContainer {
> + VFIOIOMMUBackendOpsClass *ops;
This is unexpected.
I thought that an abstract QOM model for VFIOContainer was going
to be introduced with a VFIOContainerClass with the ops below
(VFIOIOMMUBackendOpsClass).
Then, we would call :
VFIOContainerClass *vcc = VFIO_CONTAINER_GET_CLASS(container);
to get the specific implementation for the current container.
I don't understand the VFIOIOMMUBackendOpsClass pointer and
TYPE_VFIO_IOMMU_BACKEND_OPS. It seems redundant.
Thanks,
C.
> +};
> +
> +#define TYPE_VFIO_IOMMU_BACKEND_OPS "vfio-iommu-backend-ops"
> +
> +DECLARE_CLASS_CHECKERS(VFIOIOMMUBackendOpsClass,
> + VFIO_IOMMU_BACKEND_OPS, TYPE_VFIO_IOMMU_BACKEND_OPS)
> +
> +struct VFIOIOMMUBackendOpsClass {
> + /*< private >*/
> + ObjectClass parent_class;
> +
> + /*< public >*/
> + /* required */
> + int (*dma_map)(VFIOContainer *bcontainer,
> + hwaddr iova, ram_addr_t size,
> + void *vaddr, bool readonly);
> + int (*dma_unmap)(VFIOContainer *bcontainer,
> + hwaddr iova, ram_addr_t size,
> + IOMMUTLBEntry *iotlb);
> + int (*attach_device)(char *name, VFIODevice *vbasedev,
> + AddressSpace *as, Error **errp);
> + void (*detach_device)(VFIODevice *vbasedev);
> + /* migration feature */
> + int (*set_dirty_page_tracking)(VFIOContainer *bcontainer, bool start);
> + int (*query_dirty_bitmap)(VFIOContainer *bcontainer, VFIOBitmap *vbmap,
> + hwaddr iova, hwaddr size);
> +
> + /* SPAPR specific */
> + int (*add_window)(VFIOContainer *bcontainer,
> + MemoryRegionSection *section,
> + Error **errp);
> + void (*del_window)(VFIOContainer *bcontainer,
> + MemoryRegionSection *section);
> +};
> +
> +#endif /* HW_VFIO_VFIO_BASE_CONTAINER_H */
^ permalink raw reply [flat|nested] 43+ messages in thread
* RE: [PATCH v2 01/27] vfio: Rename VFIOContainer into VFIOLegacyContainer
2023-10-17 15:50 ` Cédric Le Goater
@ 2023-10-18 2:33 ` Duan, Zhenzhong
0 siblings, 0 replies; 43+ messages in thread
From: Duan, Zhenzhong @ 2023-10-18 2:33 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel
Cc: alex.williamson, jgg, nicolinc, Martins, Joao, eric.auger,
peterx, jasowang, Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng,
Chao P, Yi Sun, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Tuesday, October 17, 2023 11:51 PM
>Subject: Re: [PATCH v2 01/27] vfio: Rename VFIOContainer into
>VFIOLegacyContainer
>
>Hello,
>
>On 10/16/23 10:31, Zhenzhong Duan wrote:
>> From: Eric Auger <eric.auger@redhat.com>
>>
>> In the prospect to introduce a base object for the VFIOContainer
>> and derive into the existing legacy container and the iommufd
>> based container, let's rename the existing one into
>> VFIOLegacyContainer. This is just an incremental step to ease
>> the migration. Soon there won't be any reference to the legacy
>> container in the common.c code. Only the container.c should
>> handle the VFIOLegacyContainer object.
>
>So I think I would have kept the current name as it is, named the
>new abtract QOM object VFIOContainerBase and then, derived from
>VFIOContainerBase, the current implementation VFIOContainer and
>the new one VFIOContainerIOMMUFD.
OK, will do.
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 43+ messages in thread
* RE: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-17 15:51 ` Cédric Le Goater
@ 2023-10-18 2:41 ` Duan, Zhenzhong
2023-10-18 8:04 ` Cédric Le Goater
0 siblings, 1 reply; 43+ messages in thread
From: Duan, Zhenzhong @ 2023-10-18 2:41 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel
Cc: alex.williamson, jgg, nicolinc, Martins, Joao, eric.auger,
peterx, jasowang, Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng,
Chao P, Yi Sun
Hi Cédric,
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Tuesday, October 17, 2023 11:51 PM
>Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and
>targetted interface
>
>On 10/16/23 10:31, Zhenzhong Duan wrote:
>> From: Eric Auger <eric.auger@redhat.com>
>>
>> Introduce a dumb VFIOContainer base object and its targetted interface.
>> This is willingly not a QOM object because we don't want it to be
>> visible from the user interface. The VFIOContainer will be smoothly
>> populated in subsequent patches as well as interfaces.
>>
>> No fucntional change intended.
>>
>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>> ---
>> include/hw/vfio/vfio-common.h | 8 +--
>> include/hw/vfio/vfio-container-base.h | 82 +++++++++++++++++++++++++++
>> 2 files changed, 84 insertions(+), 6 deletions(-)
>> create mode 100644 include/hw/vfio/vfio-container-base.h
>>
>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
>> index 34648e518e..9651cf921c 100644
>> --- a/include/hw/vfio/vfio-common.h
>> +++ b/include/hw/vfio/vfio-common.h
>> @@ -30,6 +30,7 @@
>> #include <linux/vfio.h>
>> #endif
>> #include "sysemu/sysemu.h"
>> +#include "hw/vfio/vfio-container-base.h"
>>
>> #define VFIO_MSG_PREFIX "vfio %s: "
>>
>> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
>> struct VFIOGroup;
>>
>> typedef struct VFIOLegacyContainer {
>> + VFIOContainer bcontainer;
>
>That's the parent class, right ?
Right.
>
>> VFIOAddressSpace *space;
>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>> MemoryListener listener;
>> @@ -200,12 +202,6 @@ typedef struct VFIODisplay {
>> } dmabuf;
>> } VFIODisplay;
>>
>> -typedef struct {
>> - unsigned long *bitmap;
>> - hwaddr size;
>> - hwaddr pages;
>> -} VFIOBitmap;
>> -
>> void vfio_host_win_add(VFIOLegacyContainer *container,
>> hwaddr min_iova, hwaddr max_iova,
>> uint64_t iova_pgsizes);
>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>container-base.h
>> new file mode 100644
>> index 0000000000..afc8543d22
>> --- /dev/null
>> +++ b/include/hw/vfio/vfio-container-base.h
>> @@ -0,0 +1,82 @@
>> +/*
>> + * VFIO BASE CONTAINER
>> + *
>> + * Copyright (C) 2023 Intel Corporation.
>> + * Copyright Red Hat, Inc. 2023
>> + *
>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>> + * Eric Auger <eric.auger@redhat.com>
>> + *
>> + * This program is free software; you can redistribute it and/or modify
>> + * it under the terms of the GNU General Public License as published by
>> + * the Free Software Foundation; either version 2 of the License, or
>> + * (at your option) any later version.
>> +
>> + * This program is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>> + * GNU General Public License for more details.
>> +
>> + * You should have received a copy of the GNU General Public License along
>> + * with this program; if not, see <http://www.gnu.org/licenses/>.
>> + */
>> +
>> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
>> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
>> +
>> +#include "exec/memory.h"
>> +#ifndef CONFIG_USER_ONLY
>> +#include "exec/hwaddr.h"
>> +#endif
>> +
>> +typedef struct VFIOContainer VFIOContainer;
>> +typedef struct VFIODevice VFIODevice;
>> +typedef struct VFIOIOMMUBackendOpsClass VFIOIOMMUBackendOpsClass;
>> +
>> +typedef struct {
>> + unsigned long *bitmap;
>> + hwaddr size;
>> + hwaddr pages;
>> +} VFIOBitmap;
>> +
>> +/*
>> + * This is the base object for vfio container backends
>> + */
>> +struct VFIOContainer {
>> + VFIOIOMMUBackendOpsClass *ops;
>
>This is unexpected.
>
>I thought that an abstract QOM model for VFIOContainer was going
>to be introduced with a VFIOContainerClass with the ops below
>(VFIOIOMMUBackendOpsClass).
>
>Then, we would call :
>
> VFIOContainerClass *vcc = VFIO_CONTAINER_GET_CLASS(container);
>
>to get the specific implementation for the current container.
>
>I don't understand the VFIOIOMMUBackendOpsClass pointer and
>TYPE_VFIO_IOMMU_BACKEND_OPS. It seems redundant.
The original implementation was abstract QOM model. But it wasn't accepted,
see https://lore.kernel.org/all/YmuFv2s5TPuw7K%2Fu@yekko/ for details.
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 43+ messages in thread
* Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-18 2:41 ` Duan, Zhenzhong
@ 2023-10-18 8:04 ` Cédric Le Goater
2023-10-19 2:29 ` Duan, Zhenzhong
0 siblings, 1 reply; 43+ messages in thread
From: Cédric Le Goater @ 2023-10-18 8:04 UTC (permalink / raw)
To: Duan, Zhenzhong, qemu-devel
Cc: alex.williamson, jgg, nicolinc, Martins, Joao, eric.auger,
peterx, jasowang, Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng,
Chao P, Yi Sun
On 10/18/23 04:41, Duan, Zhenzhong wrote:
> Hi Cédric,
>
>> -----Original Message-----
>> From: Cédric Le Goater <clg@redhat.com>
>> Sent: Tuesday, October 17, 2023 11:51 PM
>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and
>> targetted interface
>>
>> On 10/16/23 10:31, Zhenzhong Duan wrote:
>>> From: Eric Auger <eric.auger@redhat.com>
>>>
>>> Introduce a dumb VFIOContainer base object and its targetted interface.
>>> This is willingly not a QOM object because we don't want it to be
>>> visible from the user interface. The VFIOContainer will be smoothly
>>> populated in subsequent patches as well as interfaces.
>>>
>>> No fucntional change intended.
>>>
>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>> ---
>>> include/hw/vfio/vfio-common.h | 8 +--
>>> include/hw/vfio/vfio-container-base.h | 82 +++++++++++++++++++++++++++
>>> 2 files changed, 84 insertions(+), 6 deletions(-)
>>> create mode 100644 include/hw/vfio/vfio-container-base.h
>>>
>>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
>>> index 34648e518e..9651cf921c 100644
>>> --- a/include/hw/vfio/vfio-common.h
>>> +++ b/include/hw/vfio/vfio-common.h
>>> @@ -30,6 +30,7 @@
>>> #include <linux/vfio.h>
>>> #endif
>>> #include "sysemu/sysemu.h"
>>> +#include "hw/vfio/vfio-container-base.h"
>>>
>>> #define VFIO_MSG_PREFIX "vfio %s: "
>>>
>>> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
>>> struct VFIOGroup;
>>>
>>> typedef struct VFIOLegacyContainer {
>>> + VFIOContainer bcontainer;
>>
>> That's the parent class, right ?
>
> Right.
>
>>
>>> VFIOAddressSpace *space;
>>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>>> MemoryListener listener;
>>> @@ -200,12 +202,6 @@ typedef struct VFIODisplay {
>>> } dmabuf;
>>> } VFIODisplay;
>>>
>>> -typedef struct {
>>> - unsigned long *bitmap;
>>> - hwaddr size;
>>> - hwaddr pages;
>>> -} VFIOBitmap;
>>> -
>>> void vfio_host_win_add(VFIOLegacyContainer *container,
>>> hwaddr min_iova, hwaddr max_iova,
>>> uint64_t iova_pgsizes);
>>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>> container-base.h
>>> new file mode 100644
>>> index 0000000000..afc8543d22
>>> --- /dev/null
>>> +++ b/include/hw/vfio/vfio-container-base.h
>>> @@ -0,0 +1,82 @@
>>> +/*
>>> + * VFIO BASE CONTAINER
>>> + *
>>> + * Copyright (C) 2023 Intel Corporation.
>>> + * Copyright Red Hat, Inc. 2023
>>> + *
>>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>>> + * Eric Auger <eric.auger@redhat.com>
>>> + *
>>> + * This program is free software; you can redistribute it and/or modify
>>> + * it under the terms of the GNU General Public License as published by
>>> + * the Free Software Foundation; either version 2 of the License, or
>>> + * (at your option) any later version.
>>> +
>>> + * This program is distributed in the hope that it will be useful,
>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>>> + * GNU General Public License for more details.
>>> +
>>> + * You should have received a copy of the GNU General Public License along
>>> + * with this program; if not, see <http://www.gnu.org/licenses/>.
>>> + */
>>> +
>>> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
>>> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
>>> +
>>> +#include "exec/memory.h"
>>> +#ifndef CONFIG_USER_ONLY
>>> +#include "exec/hwaddr.h"
>>> +#endif
>>> +
>>> +typedef struct VFIOContainer VFIOContainer;
>>> +typedef struct VFIODevice VFIODevice;
>>> +typedef struct VFIOIOMMUBackendOpsClass VFIOIOMMUBackendOpsClass;
>>> +
>>> +typedef struct {
>>> + unsigned long *bitmap;
>>> + hwaddr size;
>>> + hwaddr pages;
>>> +} VFIOBitmap;
>>> +
>>> +/*
>>> + * This is the base object for vfio container backends
>>> + */
>>> +struct VFIOContainer {
>>> + VFIOIOMMUBackendOpsClass *ops;
>>
>> This is unexpected.
>>
>> I thought that an abstract QOM model for VFIOContainer was going
>> to be introduced with a VFIOContainerClass with the ops below
>> (VFIOIOMMUBackendOpsClass).
>>
>> Then, we would call :
>>
>> VFIOContainerClass *vcc = VFIO_CONTAINER_GET_CLASS(container);
>>
>> to get the specific implementation for the current container.
>>
>> I don't understand the VFIOIOMMUBackendOpsClass pointer and
>> TYPE_VFIO_IOMMU_BACKEND_OPS. It seems redundant.
>
> The original implementation was abstract QOM model. But it wasn't accepted,
> see https://lore.kernel.org/all/YmuFv2s5TPuw7K%2Fu@yekko/ for details.
I see the idea was challenged, not rejected. I need to dig in further and this
will take time.
Thanks,
C.
^ permalink raw reply [flat|nested] 43+ messages in thread
* RE: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-18 8:04 ` Cédric Le Goater
@ 2023-10-19 2:29 ` Duan, Zhenzhong
2023-10-19 12:17 ` Cédric Le Goater
0 siblings, 1 reply; 43+ messages in thread
From: Duan, Zhenzhong @ 2023-10-19 2:29 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel
Cc: alex.williamson, jgg, nicolinc, Martins, Joao, eric.auger,
peterx, jasowang, Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng,
Chao P, Yi Sun, David Gibson
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Wednesday, October 18, 2023 4:04 PM
>Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and
>targetted interface
>
>On 10/18/23 04:41, Duan, Zhenzhong wrote:
>> Hi Cédric,
>>
>>> -----Original Message-----
>>> From: Cédric Le Goater <clg@redhat.com>
>>> Sent: Tuesday, October 17, 2023 11:51 PM
>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer
>and
>>> targetted interface
>>>
>>> On 10/16/23 10:31, Zhenzhong Duan wrote:
>>>> From: Eric Auger <eric.auger@redhat.com>
>>>>
>>>> Introduce a dumb VFIOContainer base object and its targetted interface.
>>>> This is willingly not a QOM object because we don't want it to be
>>>> visible from the user interface. The VFIOContainer will be smoothly
>>>> populated in subsequent patches as well as interfaces.
>>>>
>>>> No fucntional change intended.
>>>>
>>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>>> ---
>>>> include/hw/vfio/vfio-common.h | 8 +--
>>>> include/hw/vfio/vfio-container-base.h | 82
>+++++++++++++++++++++++++++
>>>> 2 files changed, 84 insertions(+), 6 deletions(-)
>>>> create mode 100644 include/hw/vfio/vfio-container-base.h
>>>>
>>>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-
>common.h
>>>> index 34648e518e..9651cf921c 100644
>>>> --- a/include/hw/vfio/vfio-common.h
>>>> +++ b/include/hw/vfio/vfio-common.h
>>>> @@ -30,6 +30,7 @@
>>>> #include <linux/vfio.h>
>>>> #endif
>>>> #include "sysemu/sysemu.h"
>>>> +#include "hw/vfio/vfio-container-base.h"
>>>>
>>>> #define VFIO_MSG_PREFIX "vfio %s: "
>>>>
>>>> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
>>>> struct VFIOGroup;
>>>>
>>>> typedef struct VFIOLegacyContainer {
>>>> + VFIOContainer bcontainer;
>>>
>>> That's the parent class, right ?
>>
>> Right.
>>
>>>
>>>> VFIOAddressSpace *space;
>>>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>>>> MemoryListener listener;
>>>> @@ -200,12 +202,6 @@ typedef struct VFIODisplay {
>>>> } dmabuf;
>>>> } VFIODisplay;
>>>>
>>>> -typedef struct {
>>>> - unsigned long *bitmap;
>>>> - hwaddr size;
>>>> - hwaddr pages;
>>>> -} VFIOBitmap;
>>>> -
>>>> void vfio_host_win_add(VFIOLegacyContainer *container,
>>>> hwaddr min_iova, hwaddr max_iova,
>>>> uint64_t iova_pgsizes);
>>>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>>> container-base.h
>>>> new file mode 100644
>>>> index 0000000000..afc8543d22
>>>> --- /dev/null
>>>> +++ b/include/hw/vfio/vfio-container-base.h
>>>> @@ -0,0 +1,82 @@
>>>> +/*
>>>> + * VFIO BASE CONTAINER
>>>> + *
>>>> + * Copyright (C) 2023 Intel Corporation.
>>>> + * Copyright Red Hat, Inc. 2023
>>>> + *
>>>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>>>> + * Eric Auger <eric.auger@redhat.com>
>>>> + *
>>>> + * This program is free software; you can redistribute it and/or modify
>>>> + * it under the terms of the GNU General Public License as published by
>>>> + * the Free Software Foundation; either version 2 of the License, or
>>>> + * (at your option) any later version.
>>>> +
>>>> + * This program is distributed in the hope that it will be useful,
>>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>>>> + * GNU General Public License for more details.
>>>> +
>>>> + * You should have received a copy of the GNU General Public License along
>>>> + * with this program; if not, see <http://www.gnu.org/licenses/>.
>>>> + */
>>>> +
>>>> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
>>>> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
>>>> +
>>>> +#include "exec/memory.h"
>>>> +#ifndef CONFIG_USER_ONLY
>>>> +#include "exec/hwaddr.h"
>>>> +#endif
>>>> +
>>>> +typedef struct VFIOContainer VFIOContainer;
>>>> +typedef struct VFIODevice VFIODevice;
>>>> +typedef struct VFIOIOMMUBackendOpsClass VFIOIOMMUBackendOpsClass;
>>>> +
>>>> +typedef struct {
>>>> + unsigned long *bitmap;
>>>> + hwaddr size;
>>>> + hwaddr pages;
>>>> +} VFIOBitmap;
>>>> +
>>>> +/*
>>>> + * This is the base object for vfio container backends
>>>> + */
>>>> +struct VFIOContainer {
>>>> + VFIOIOMMUBackendOpsClass *ops;
>>>
>>> This is unexpected.
>>>
>>> I thought that an abstract QOM model for VFIOContainer was going
>>> to be introduced with a VFIOContainerClass with the ops below
>>> (VFIOIOMMUBackendOpsClass).
>>>
>>> Then, we would call :
>>>
>>> VFIOContainerClass *vcc = VFIO_CONTAINER_GET_CLASS(container);
>>>
>>> to get the specific implementation for the current container.
>>>
>>> I don't understand the VFIOIOMMUBackendOpsClass pointer and
>>> TYPE_VFIO_IOMMU_BACKEND_OPS. It seems redundant.
>>
>> The original implementation was abstract QOM model. But it wasn't accepted,
>> see https://lore.kernel.org/all/YmuFv2s5TPuw7K%2Fu@yekko/ for details.
>
>I see the idea was challenged, not rejected. I need to dig in further and this
>will take time.
Thanks for help looking into it.
+David, Hi David, I'm digging into your concern of using QOM to abstract base
container and legacy VFIOContainer:
"The QOM class of things is visible to the user/config layer via QMP (and sometimes command line).
It doesn't necessarily correspond to guest visible differences, but it often does."
AIUI, while it's true when the QOM type includes TYPE_USER_CREATABLE interface,
otherwise, user can't create an object of this type. Only difference is user will see
"object type '%s' isn't supported by object-add" instead of "invalid object type: %s".
Is your expectation to not permit user to create this object or only want user
to see "invalid object type: %s".
If you mean the first, then I think QOM could be suitable if we don't include
TYPE_USER_CREATABLE interface?
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 43+ messages in thread
* Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-19 2:29 ` Duan, Zhenzhong
@ 2023-10-19 12:17 ` Cédric Le Goater
2023-10-20 5:48 ` Duan, Zhenzhong
0 siblings, 1 reply; 43+ messages in thread
From: Cédric Le Goater @ 2023-10-19 12:17 UTC (permalink / raw)
To: Duan, Zhenzhong, qemu-devel
Cc: alex.williamson, jgg, nicolinc, Martins, Joao, eric.auger,
peterx, jasowang, Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng,
Chao P, Yi Sun, David Gibson
On 10/19/23 04:29, Duan, Zhenzhong wrote:
>> -----Original Message-----
>> From: Cédric Le Goater <clg@redhat.com>
>> Sent: Wednesday, October 18, 2023 4:04 PM
>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and
>> targetted interface
>>
>> On 10/18/23 04:41, Duan, Zhenzhong wrote:
>>> Hi Cédric,
>>>
>>>> -----Original Message-----
>>>> From: Cédric Le Goater <clg@redhat.com>
>>>> Sent: Tuesday, October 17, 2023 11:51 PM
>>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer
>> and
>>>> targetted interface
>>>>
>>>> On 10/16/23 10:31, Zhenzhong Duan wrote:
>>>>> From: Eric Auger <eric.auger@redhat.com>
>>>>>
>>>>> Introduce a dumb VFIOContainer base object and its targetted interface.
>>>>> This is willingly not a QOM object because we don't want it to be
>>>>> visible from the user interface. The VFIOContainer will be smoothly
>>>>> populated in subsequent patches as well as interfaces.
>>>>>
>>>>> No fucntional change intended.
>>>>>
>>>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>>>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>>>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>>>> ---
>>>>> include/hw/vfio/vfio-common.h | 8 +--
>>>>> include/hw/vfio/vfio-container-base.h | 82
>> +++++++++++++++++++++++++++
>>>>> 2 files changed, 84 insertions(+), 6 deletions(-)
>>>>> create mode 100644 include/hw/vfio/vfio-container-base.h
>>>>>
>>>>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-
>> common.h
>>>>> index 34648e518e..9651cf921c 100644
>>>>> --- a/include/hw/vfio/vfio-common.h
>>>>> +++ b/include/hw/vfio/vfio-common.h
>>>>> @@ -30,6 +30,7 @@
>>>>> #include <linux/vfio.h>
>>>>> #endif
>>>>> #include "sysemu/sysemu.h"
>>>>> +#include "hw/vfio/vfio-container-base.h"
>>>>>
>>>>> #define VFIO_MSG_PREFIX "vfio %s: "
>>>>>
>>>>> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
>>>>> struct VFIOGroup;
>>>>>
>>>>> typedef struct VFIOLegacyContainer {
>>>>> + VFIOContainer bcontainer;
>>>>
>>>> That's the parent class, right ?
>>>
>>> Right.
>>>
>>>>
>>>>> VFIOAddressSpace *space;
>>>>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>>>>> MemoryListener listener;
>>>>> @@ -200,12 +202,6 @@ typedef struct VFIODisplay {
>>>>> } dmabuf;
>>>>> } VFIODisplay;
>>>>>
>>>>> -typedef struct {
>>>>> - unsigned long *bitmap;
>>>>> - hwaddr size;
>>>>> - hwaddr pages;
>>>>> -} VFIOBitmap;
>>>>> -
>>>>> void vfio_host_win_add(VFIOLegacyContainer *container,
>>>>> hwaddr min_iova, hwaddr max_iova,
>>>>> uint64_t iova_pgsizes);
>>>>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>>>> container-base.h
>>>>> new file mode 100644
>>>>> index 0000000000..afc8543d22
>>>>> --- /dev/null
>>>>> +++ b/include/hw/vfio/vfio-container-base.h
>>>>> @@ -0,0 +1,82 @@
>>>>> +/*
>>>>> + * VFIO BASE CONTAINER
>>>>> + *
>>>>> + * Copyright (C) 2023 Intel Corporation.
>>>>> + * Copyright Red Hat, Inc. 2023
>>>>> + *
>>>>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>>>>> + * Eric Auger <eric.auger@redhat.com>
>>>>> + *
>>>>> + * This program is free software; you can redistribute it and/or modify
>>>>> + * it under the terms of the GNU General Public License as published by
>>>>> + * the Free Software Foundation; either version 2 of the License, or
>>>>> + * (at your option) any later version.
>>>>> +
>>>>> + * This program is distributed in the hope that it will be useful,
>>>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>>>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>>>>> + * GNU General Public License for more details.
>>>>> +
>>>>> + * You should have received a copy of the GNU General Public License along
>>>>> + * with this program; if not, see <http://www.gnu.org/licenses/>.
>>>>> + */
>>>>> +
>>>>> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
>>>>> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
>>>>> +
>>>>> +#include "exec/memory.h"
>>>>> +#ifndef CONFIG_USER_ONLY
>>>>> +#include "exec/hwaddr.h"
>>>>> +#endif
>>>>> +
>>>>> +typedef struct VFIOContainer VFIOContainer;
>>>>> +typedef struct VFIODevice VFIODevice;
>>>>> +typedef struct VFIOIOMMUBackendOpsClass VFIOIOMMUBackendOpsClass;
>>>>> +
>>>>> +typedef struct {
>>>>> + unsigned long *bitmap;
>>>>> + hwaddr size;
>>>>> + hwaddr pages;
>>>>> +} VFIOBitmap;
>>>>> +
>>>>> +/*
>>>>> + * This is the base object for vfio container backends
>>>>> + */
>>>>> +struct VFIOContainer {
>>>>> + VFIOIOMMUBackendOpsClass *ops;
>>>>
>>>> This is unexpected.
>>>>
>>>> I thought that an abstract QOM model for VFIOContainer was going
>>>> to be introduced with a VFIOContainerClass with the ops below
>>>> (VFIOIOMMUBackendOpsClass).
>>>>
>>>> Then, we would call :
>>>>
>>>> VFIOContainerClass *vcc = VFIO_CONTAINER_GET_CLASS(container);
>>>>
>>>> to get the specific implementation for the current container.
>>>>
>>>> I don't understand the VFIOIOMMUBackendOpsClass pointer and
>>>> TYPE_VFIO_IOMMU_BACKEND_OPS. It seems redundant.
>>>
>>> The original implementation was abstract QOM model. But it wasn't accepted,
>>> see https://lore.kernel.org/all/YmuFv2s5TPuw7K%2Fu@yekko/ for details.
>>
>> I see the idea was challenged, not rejected. I need to dig in further and this
>> will take time.
> Thanks for help looking into it.
>
> +David, Hi David, I'm digging into your concern of using QOM to abstract base
> container and legacy VFIOContainer:
> "The QOM class of things is visible to the user/config layer via QMP (and sometimes command line).
> It doesn't necessarily correspond to guest visible differences, but it often does."
>
> AIUI, while it's true when the QOM type includes TYPE_USER_CREATABLE interface,
> otherwise, user can't create an object of this type. Only difference is user will see
> "object type '%s' isn't supported by object-add" instead of "invalid object type: %s".
>
> Is your expectation to not permit user to create this object or only want user
> to see "invalid object type: %s".
> If you mean the first, then I think QOM could be suitable if we don't include
> TYPE_USER_CREATABLE interface?
I was imagining some kind of QOM hierarchy under the vfio device
with various QOM interfaces (similar to the ops) to define the
possible IOMMU backends. The fact that we use the IOMMUFD object
from the command line made it more plausible. I might be mistaking.
Anyhow, the series looks pretty good. There are other aspect to
check, like are all this iommu ops well suited for the need ?
Let's stress the models in parallel of the reviews. If we could get
some of it for 8.2 that'd be nice. It's on top of my list now.
Thanks,
C.
> Thanks
> Zhenzhong
^ permalink raw reply [flat|nested] 43+ messages in thread
* RE: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-19 12:17 ` Cédric Le Goater
@ 2023-10-20 5:48 ` Duan, Zhenzhong
2023-10-20 8:19 ` Eric Auger
0 siblings, 1 reply; 43+ messages in thread
From: Duan, Zhenzhong @ 2023-10-20 5:48 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel
Cc: alex.williamson, jgg, nicolinc, Martins, Joao, eric.auger,
peterx, jasowang, Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng,
Chao P, Yi Sun, David Gibson
Hi Cédric,
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Thursday, October 19, 2023 8:18 PM
>Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and
>targetted interface
>
>On 10/19/23 04:29, Duan, Zhenzhong wrote:
>>> -----Original Message-----
>>> From: Cédric Le Goater <clg@redhat.com>
>>> Sent: Wednesday, October 18, 2023 4:04 PM
>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer
>and
>>> targetted interface
>>>
>>> On 10/18/23 04:41, Duan, Zhenzhong wrote:
>>>> Hi Cédric,
>>>>
>>>>> -----Original Message-----
>>>>> From: Cédric Le Goater <clg@redhat.com>
>>>>> Sent: Tuesday, October 17, 2023 11:51 PM
>>>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer
>>> and
>>>>> targetted interface
>>>>>
>>>>> On 10/16/23 10:31, Zhenzhong Duan wrote:
>>>>>> From: Eric Auger <eric.auger@redhat.com>
>>>>>>
>>>>>> Introduce a dumb VFIOContainer base object and its targetted interface.
>>>>>> This is willingly not a QOM object because we don't want it to be
>>>>>> visible from the user interface. The VFIOContainer will be smoothly
>>>>>> populated in subsequent patches as well as interfaces.
>>>>>>
>>>>>> No fucntional change intended.
>>>>>>
>>>>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>>>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>>>>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>>>>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>>>>> ---
>>>>>> include/hw/vfio/vfio-common.h | 8 +--
>>>>>> include/hw/vfio/vfio-container-base.h | 82
>>> +++++++++++++++++++++++++++
>>>>>> 2 files changed, 84 insertions(+), 6 deletions(-)
>>>>>> create mode 100644 include/hw/vfio/vfio-container-base.h
>>>>>>
>>>>>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-
>>> common.h
>>>>>> index 34648e518e..9651cf921c 100644
>>>>>> --- a/include/hw/vfio/vfio-common.h
>>>>>> +++ b/include/hw/vfio/vfio-common.h
>>>>>> @@ -30,6 +30,7 @@
>>>>>> #include <linux/vfio.h>
>>>>>> #endif
>>>>>> #include "sysemu/sysemu.h"
>>>>>> +#include "hw/vfio/vfio-container-base.h"
>>>>>>
>>>>>> #define VFIO_MSG_PREFIX "vfio %s: "
>>>>>>
>>>>>> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
>>>>>> struct VFIOGroup;
>>>>>>
>>>>>> typedef struct VFIOLegacyContainer {
>>>>>> + VFIOContainer bcontainer;
>>>>>
>>>>> That's the parent class, right ?
>>>>
>>>> Right.
>>>>
>>>>>
>>>>>> VFIOAddressSpace *space;
>>>>>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>>>>>> MemoryListener listener;
>>>>>> @@ -200,12 +202,6 @@ typedef struct VFIODisplay {
>>>>>> } dmabuf;
>>>>>> } VFIODisplay;
>>>>>>
>>>>>> -typedef struct {
>>>>>> - unsigned long *bitmap;
>>>>>> - hwaddr size;
>>>>>> - hwaddr pages;
>>>>>> -} VFIOBitmap;
>>>>>> -
>>>>>> void vfio_host_win_add(VFIOLegacyContainer *container,
>>>>>> hwaddr min_iova, hwaddr max_iova,
>>>>>> uint64_t iova_pgsizes);
>>>>>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>>>>> container-base.h
>>>>>> new file mode 100644
>>>>>> index 0000000000..afc8543d22
>>>>>> --- /dev/null
>>>>>> +++ b/include/hw/vfio/vfio-container-base.h
>>>>>> @@ -0,0 +1,82 @@
>>>>>> +/*
>>>>>> + * VFIO BASE CONTAINER
>>>>>> + *
>>>>>> + * Copyright (C) 2023 Intel Corporation.
>>>>>> + * Copyright Red Hat, Inc. 2023
>>>>>> + *
>>>>>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>>>>>> + * Eric Auger <eric.auger@redhat.com>
>>>>>> + *
>>>>>> + * This program is free software; you can redistribute it and/or modify
>>>>>> + * it under the terms of the GNU General Public License as published by
>>>>>> + * the Free Software Foundation; either version 2 of the License, or
>>>>>> + * (at your option) any later version.
>>>>>> +
>>>>>> + * This program is distributed in the hope that it will be useful,
>>>>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>>>>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>>>>>> + * GNU General Public License for more details.
>>>>>> +
>>>>>> + * You should have received a copy of the GNU General Public License
>along
>>>>>> + * with this program; if not, see <http://www.gnu.org/licenses/>.
>>>>>> + */
>>>>>> +
>>>>>> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
>>>>>> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
>>>>>> +
>>>>>> +#include "exec/memory.h"
>>>>>> +#ifndef CONFIG_USER_ONLY
>>>>>> +#include "exec/hwaddr.h"
>>>>>> +#endif
>>>>>> +
>>>>>> +typedef struct VFIOContainer VFIOContainer;
>>>>>> +typedef struct VFIODevice VFIODevice;
>>>>>> +typedef struct VFIOIOMMUBackendOpsClass
>VFIOIOMMUBackendOpsClass;
>>>>>> +
>>>>>> +typedef struct {
>>>>>> + unsigned long *bitmap;
>>>>>> + hwaddr size;
>>>>>> + hwaddr pages;
>>>>>> +} VFIOBitmap;
>>>>>> +
>>>>>> +/*
>>>>>> + * This is the base object for vfio container backends
>>>>>> + */
>>>>>> +struct VFIOContainer {
>>>>>> + VFIOIOMMUBackendOpsClass *ops;
>>>>>
>>>>> This is unexpected.
>>>>>
>>>>> I thought that an abstract QOM model for VFIOContainer was going
>>>>> to be introduced with a VFIOContainerClass with the ops below
>>>>> (VFIOIOMMUBackendOpsClass).
>>>>>
>>>>> Then, we would call :
>>>>>
>>>>> VFIOContainerClass *vcc = VFIO_CONTAINER_GET_CLASS(container);
>>>>>
>>>>> to get the specific implementation for the current container.
>>>>>
>>>>> I don't understand the VFIOIOMMUBackendOpsClass pointer and
>>>>> TYPE_VFIO_IOMMU_BACKEND_OPS. It seems redundant.
>>>>
>>>> The original implementation was abstract QOM model. But it wasn't
>accepted,
>>>> see https://lore.kernel.org/all/YmuFv2s5TPuw7K%2Fu@yekko/ for details.
>>>
>>> I see the idea was challenged, not rejected. I need to dig in further and this
>>> will take time.
>> Thanks for help looking into it.
>>
>> +David, Hi David, I'm digging into your concern of using QOM to abstract base
>> container and legacy VFIOContainer:
>> "The QOM class of things is visible to the user/config layer via QMP (and
>sometimes command line).
>> It doesn't necessarily correspond to guest visible differences, but it often does."
>>
>> AIUI, while it's true when the QOM type includes TYPE_USER_CREATABLE
>interface,
>> otherwise, user can't create an object of this type. Only difference is user will
>see
>> "object type '%s' isn't supported by object-add" instead of "invalid object
>type: %s".
>>
>> Is your expectation to not permit user to create this object or only want user
>> to see "invalid object type: %s".
>> If you mean the first, then I think QOM could be suitable if we don't include
>> TYPE_USER_CREATABLE interface?
>
>I was imagining some kind of QOM hierarchy under the vfio device
>with various QOM interfaces (similar to the ops) to define the
>possible IOMMU backends. The fact that we use the IOMMUFD object
>from the command line made it more plausible. I might be mistaking.
Got your point.
This way we introduce a new QOM type "vfio-pci-iommufd" for iommufd support,
and vfio-pci keep same for legacy backend, e.g:
#qemu -object iommufd,id=iommufd0 \
-device vfio-pci-iommufd,iommufd=iommufd0,id=vfio0... \
-device vfio-pci,id=vfio1...
Below is a draft for vfio-pci based on your imagination. not pasted here the change
for platform/ap/ccw vfio device which should be similar.
Not clear if this fully match your imagination.
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 5345986993..54a6ce4d73 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -124,7 +124,7 @@ DECLARE_CLASS_CHECKERS(VFIOIOMMUBackendOpsClass,
struct VFIOIOMMUBackendOpsClass {
/*< private >*/
- ObjectClass parent_class;
+ InterfaceClass parent_class;
/*< public >*/
/* required */
diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
index edb787d3d1..829deddc7d 100644
--- a/hw/vfio/pci.c
+++ b/hw/vfio/pci.c
@@ -3725,6 +3725,11 @@ static void vfio_pci_dev_class_init(ObjectClass *klass, void *data)
{
DeviceClass *dc = DEVICE_CLASS(klass);
PCIDeviceClass *pdc = PCI_DEVICE_CLASS(klass);
+ IOMMU_Backend_Ops_Class *be_ops = IOMMU_BACKEND_OPS_CLASS(klass);
+
+ be_ops->dma_map = vfio_legacy_dma_map;
+ be_ops->dma_unmap = vfio_legacy_dma_unmap;
+ ...
dc->reset = vfio_pci_reset;
device_class_set_props(dc, vfio_pci_dev_properties);
@@ -3749,10 +3754,40 @@ static const TypeInfo vfio_pci_dev_info = {
.interfaces = (InterfaceInfo[]) {
{ INTERFACE_PCIE_DEVICE },
{ INTERFACE_CONVENTIONAL_PCI_DEVICE },
+ { INTERFACE_IOMMU_BACKEND_OPS },
{ }
},
};
+static Property vfio_pci_dev_iommufd_properties[] = {
+#ifdef CONFIG_IOMMUFD
+ DEFINE_PROP_LINK("iommufd", VFIOPCIDevice, vbasedev.iommufd,
+ TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
+#endif
+};
+
+static void vfio_pci_iommufd_dev_class_init(ObjectClass *klass, void *data)
+{
+ DeviceClass *dc = DEVICE_CLASS(klass);
+ IOMMU_Backend_Ops_Class *be_ops = IOMMU_BACKEND_OPS_CLASS(klass);
+
+ device_class_set_props(dc, vfio_pci_dev_iommufd_properties);
+
+ be_ops->dma_map = iommufd_map;
+ be_ops->dma_unmap = iommufd_unmap;
+ ...
+ /* Unimplemented ops */
+ be_ops->set_dirty_page_tracking = NULL;
+ ...
+}
+
+static const TypeInfo vfio_pci_iommufd_dev_info = {
+ .name = TYPE_VFIO_PCI_IOMMUFD,
+ .parent = TYPE_VFIO_PCI,
+ .instance_size = sizeof(VFIOPCIDevice),
+ .class_init = vfio_pci_iommufd_dev_class_init,
+};
+
static Property vfio_pci_dev_nohotplug_properties[] = {
DEFINE_PROP_BOOL("ramfb", VFIOPCIDevice, enable_ramfb, false),
DEFINE_PROP_ON_OFF_AUTO("x-ramfb-migrate", VFIOPCIDevice, ramfb_migrate,
@@ -3770,13 +3805,20 @@ static void vfio_pci_nohotplug_dev_class_init(ObjectClass *klass, void *data)
static const TypeInfo vfio_pci_nohotplug_dev_info = {
.name = TYPE_VFIO_PCI_NOHOTPLUG,
- .parent = TYPE_VFIO_PCI,
+ .parent = TYPE_VFIO_PCI_IOMMUFD,
.instance_size = sizeof(VFIOPCIDevice),
.class_init = vfio_pci_nohotplug_dev_class_init,
};
static void register_vfio_pci_dev_type(void)
{
+ static const TypeInfo iommu_be_ops_interface_info = {
+ .name = TYPE_IOMMU_BACKEND_OPS,
+ .parent = TYPE_INTERFACE,
+ .class_size = sizeof(VFIOIOMMUBackendOpsClass),
+ };
+
+ type_register_static(&iommu_be_ops_interface_info);
type_register_static(&vfio_pci_dev_info);
type_register_static(&vfio_pci_nohotplug_dev_info);
}
Thanks
Zhenzhong
>
>Anyhow, the series looks pretty good. There are other aspect to
>check, like are all this iommu ops well suited for the need ?
>Let's stress the models in parallel of the reviews. If we could get
>some of it for 8.2 that'd be nice. It's on top of my list now.
>
>Thanks,
>
>C.
>
>
>> Thanks
>> Zhenzhong
^ permalink raw reply related [flat|nested] 43+ messages in thread
* Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-20 5:48 ` Duan, Zhenzhong
@ 2023-10-20 8:19 ` Eric Auger
2023-10-20 8:28 ` Duan, Zhenzhong
2023-10-23 15:28 ` Cédric Le Goater
0 siblings, 2 replies; 43+ messages in thread
From: Eric Auger @ 2023-10-20 8:19 UTC (permalink / raw)
To: Duan, Zhenzhong, Cédric Le Goater, qemu-devel
Cc: alex.williamson, jgg, nicolinc, Martins, Joao, peterx, jasowang,
Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng, Chao P, Yi Sun,
David Gibson
Hi,
On 10/20/23 07:48, Duan, Zhenzhong wrote:
> Hi Cédric,
>
>> -----Original Message-----
>> From: Cédric Le Goater <clg@redhat.com>
>> Sent: Thursday, October 19, 2023 8:18 PM
>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and
>> targetted interface
>>
>> On 10/19/23 04:29, Duan, Zhenzhong wrote:
>>>> -----Original Message-----
>>>> From: Cédric Le Goater <clg@redhat.com>
>>>> Sent: Wednesday, October 18, 2023 4:04 PM
>>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer
>> and
>>>> targetted interface
>>>>
>>>> On 10/18/23 04:41, Duan, Zhenzhong wrote:
>>>>> Hi Cédric,
>>>>>
>>>>>> -----Original Message-----
>>>>>> From: Cédric Le Goater <clg@redhat.com>
>>>>>> Sent: Tuesday, October 17, 2023 11:51 PM
>>>>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer
>>>> and
>>>>>> targetted interface
>>>>>>
>>>>>> On 10/16/23 10:31, Zhenzhong Duan wrote:
>>>>>>> From: Eric Auger <eric.auger@redhat.com>
>>>>>>>
>>>>>>> Introduce a dumb VFIOContainer base object and its targetted interface.
>>>>>>> This is willingly not a QOM object because we don't want it to be
>>>>>>> visible from the user interface. The VFIOContainer will be smoothly
>>>>>>> populated in subsequent patches as well as interfaces.
>>>>>>>
>>>>>>> No fucntional change intended.
>>>>>>>
>>>>>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>>>>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>>>>>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>>>>>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>>>>>> ---
>>>>>>> include/hw/vfio/vfio-common.h | 8 +--
>>>>>>> include/hw/vfio/vfio-container-base.h | 82
>>>> +++++++++++++++++++++++++++
>>>>>>> 2 files changed, 84 insertions(+), 6 deletions(-)
>>>>>>> create mode 100644 include/hw/vfio/vfio-container-base.h
>>>>>>>
>>>>>>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-
>>>> common.h
>>>>>>> index 34648e518e..9651cf921c 100644
>>>>>>> --- a/include/hw/vfio/vfio-common.h
>>>>>>> +++ b/include/hw/vfio/vfio-common.h
>>>>>>> @@ -30,6 +30,7 @@
>>>>>>> #include <linux/vfio.h>
>>>>>>> #endif
>>>>>>> #include "sysemu/sysemu.h"
>>>>>>> +#include "hw/vfio/vfio-container-base.h"
>>>>>>>
>>>>>>> #define VFIO_MSG_PREFIX "vfio %s: "
>>>>>>>
>>>>>>> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
>>>>>>> struct VFIOGroup;
>>>>>>>
>>>>>>> typedef struct VFIOLegacyContainer {
>>>>>>> + VFIOContainer bcontainer;
>>>>>> That's the parent class, right ?
>>>>> Right.
>>>>>
>>>>>>> VFIOAddressSpace *space;
>>>>>>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>>>>>>> MemoryListener listener;
>>>>>>> @@ -200,12 +202,6 @@ typedef struct VFIODisplay {
>>>>>>> } dmabuf;
>>>>>>> } VFIODisplay;
>>>>>>>
>>>>>>> -typedef struct {
>>>>>>> - unsigned long *bitmap;
>>>>>>> - hwaddr size;
>>>>>>> - hwaddr pages;
>>>>>>> -} VFIOBitmap;
>>>>>>> -
>>>>>>> void vfio_host_win_add(VFIOLegacyContainer *container,
>>>>>>> hwaddr min_iova, hwaddr max_iova,
>>>>>>> uint64_t iova_pgsizes);
>>>>>>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>>>>>> container-base.h
>>>>>>> new file mode 100644
>>>>>>> index 0000000000..afc8543d22
>>>>>>> --- /dev/null
>>>>>>> +++ b/include/hw/vfio/vfio-container-base.h
>>>>>>> @@ -0,0 +1,82 @@
>>>>>>> +/*
>>>>>>> + * VFIO BASE CONTAINER
>>>>>>> + *
>>>>>>> + * Copyright (C) 2023 Intel Corporation.
>>>>>>> + * Copyright Red Hat, Inc. 2023
>>>>>>> + *
>>>>>>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>>>>>>> + * Eric Auger <eric.auger@redhat.com>
>>>>>>> + *
>>>>>>> + * This program is free software; you can redistribute it and/or modify
>>>>>>> + * it under the terms of the GNU General Public License as published by
>>>>>>> + * the Free Software Foundation; either version 2 of the License, or
>>>>>>> + * (at your option) any later version.
>>>>>>> +
>>>>>>> + * This program is distributed in the hope that it will be useful,
>>>>>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>>>>>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>>>>>>> + * GNU General Public License for more details.
>>>>>>> +
>>>>>>> + * You should have received a copy of the GNU General Public License
>> along
>>>>>>> + * with this program; if not, see <http://www.gnu.org/licenses/>.
>>>>>>> + */
>>>>>>> +
>>>>>>> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
>>>>>>> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
>>>>>>> +
>>>>>>> +#include "exec/memory.h"
>>>>>>> +#ifndef CONFIG_USER_ONLY
>>>>>>> +#include "exec/hwaddr.h"
>>>>>>> +#endif
>>>>>>> +
>>>>>>> +typedef struct VFIOContainer VFIOContainer;
>>>>>>> +typedef struct VFIODevice VFIODevice;
>>>>>>> +typedef struct VFIOIOMMUBackendOpsClass
>> VFIOIOMMUBackendOpsClass;
>>>>>>> +
>>>>>>> +typedef struct {
>>>>>>> + unsigned long *bitmap;
>>>>>>> + hwaddr size;
>>>>>>> + hwaddr pages;
>>>>>>> +} VFIOBitmap;
>>>>>>> +
>>>>>>> +/*
>>>>>>> + * This is the base object for vfio container backends
>>>>>>> + */
>>>>>>> +struct VFIOContainer {
>>>>>>> + VFIOIOMMUBackendOpsClass *ops;
>>>>>> This is unexpected.
>>>>>>
>>>>>> I thought that an abstract QOM model for VFIOContainer was going
>>>>>> to be introduced with a VFIOContainerClass with the ops below
>>>>>> (VFIOIOMMUBackendOpsClass).
>>>>>>
>>>>>> Then, we would call :
>>>>>>
>>>>>> VFIOContainerClass *vcc = VFIO_CONTAINER_GET_CLASS(container);
>>>>>>
>>>>>> to get the specific implementation for the current container.
>>>>>>
>>>>>> I don't understand the VFIOIOMMUBackendOpsClass pointer and
>>>>>> TYPE_VFIO_IOMMU_BACKEND_OPS. It seems redundant.
>>>>> The original implementation was abstract QOM model. But it wasn't
>> accepted,
>>>>> see https://lore.kernel.org/all/YmuFv2s5TPuw7K%2Fu@yekko/ for details.
>>>> I see the idea was challenged, not rejected. I need to dig in further and this
>>>> will take time.
>>> Thanks for help looking into it.
>>>
>>> +David, Hi David, I'm digging into your concern of using QOM to abstract base
>>> container and legacy VFIOContainer:
>>> "The QOM class of things is visible to the user/config layer via QMP (and
>> sometimes command line).
>>> It doesn't necessarily correspond to guest visible differences, but it often does."
>>>
>>> AIUI, while it's true when the QOM type includes TYPE_USER_CREATABLE
>> interface,
>>> otherwise, user can't create an object of this type. Only difference is user will
>> see
>>> "object type '%s' isn't supported by object-add" instead of "invalid object
>> type: %s".
>>> Is your expectation to not permit user to create this object or only want user
>>> to see "invalid object type: %s".
>>> If you mean the first, then I think QOM could be suitable if we don't include
>>> TYPE_USER_CREATABLE interface?
>> I was imagining some kind of QOM hierarchy under the vfio device
>> with various QOM interfaces (similar to the ops) to define the
>> possible IOMMU backends. The fact that we use the IOMMUFD object
> >from the command line made it more plausible. I might be mistaking.
>
> Got your point.
> This way we introduce a new QOM type "vfio-pci-iommufd" for iommufd support,
> and vfio-pci keep same for legacy backend, e.g:
>
> #qemu -object iommufd,id=iommufd0 \
> -device vfio-pci-iommufd,iommufd=iommufd0,id=vfio0... \
> -device vfio-pci,id=vfio1...
you would need to do that for all types for vfio devices, ap, ccw,
platform. Looks heavy to me. Why would we need to use a different
vfio-pci-* device while we could switch the iommu backend according to
the "iommufd" prop presence. The initial discussion was about QOMyfying
the container instead.
Thanks
Eric
>
> Below is a draft for vfio-pci based on your imagination. not pasted here the change
> for platform/ap/ccw vfio device which should be similar.
> Not clear if this fully match your imagination.
>
> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
> index 5345986993..54a6ce4d73 100644
> --- a/include/hw/vfio/vfio-container-base.h
> +++ b/include/hw/vfio/vfio-container-base.h
> @@ -124,7 +124,7 @@ DECLARE_CLASS_CHECKERS(VFIOIOMMUBackendOpsClass,
>
> struct VFIOIOMMUBackendOpsClass {
> /*< private >*/
> - ObjectClass parent_class;
> + InterfaceClass parent_class;
>
> /*< public >*/
> /* required */
> diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
> index edb787d3d1..829deddc7d 100644
> --- a/hw/vfio/pci.c
> +++ b/hw/vfio/pci.c
> @@ -3725,6 +3725,11 @@ static void vfio_pci_dev_class_init(ObjectClass *klass, void *data)
> {
> DeviceClass *dc = DEVICE_CLASS(klass);
> PCIDeviceClass *pdc = PCI_DEVICE_CLASS(klass);
> + IOMMU_Backend_Ops_Class *be_ops = IOMMU_BACKEND_OPS_CLASS(klass);
> +
> + be_ops->dma_map = vfio_legacy_dma_map;
> + be_ops->dma_unmap = vfio_legacy_dma_unmap;
> + ...
>
> dc->reset = vfio_pci_reset;
> device_class_set_props(dc, vfio_pci_dev_properties);
> @@ -3749,10 +3754,40 @@ static const TypeInfo vfio_pci_dev_info = {
> .interfaces = (InterfaceInfo[]) {
> { INTERFACE_PCIE_DEVICE },
> { INTERFACE_CONVENTIONAL_PCI_DEVICE },
> + { INTERFACE_IOMMU_BACKEND_OPS },
> { }
> },
> };
>
> +static Property vfio_pci_dev_iommufd_properties[] = {
> +#ifdef CONFIG_IOMMUFD
> + DEFINE_PROP_LINK("iommufd", VFIOPCIDevice, vbasedev.iommufd,
> + TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
> +#endif
> +};
> +
> +static void vfio_pci_iommufd_dev_class_init(ObjectClass *klass, void *data)
> +{
> + DeviceClass *dc = DEVICE_CLASS(klass);
> + IOMMU_Backend_Ops_Class *be_ops = IOMMU_BACKEND_OPS_CLASS(klass);
> +
> + device_class_set_props(dc, vfio_pci_dev_iommufd_properties);
> +
> + be_ops->dma_map = iommufd_map;
> + be_ops->dma_unmap = iommufd_unmap;
> + ...
> + /* Unimplemented ops */
> + be_ops->set_dirty_page_tracking = NULL;
> + ...
> +}
> +
> +static const TypeInfo vfio_pci_iommufd_dev_info = {
> + .name = TYPE_VFIO_PCI_IOMMUFD,
> + .parent = TYPE_VFIO_PCI,
> + .instance_size = sizeof(VFIOPCIDevice),
> + .class_init = vfio_pci_iommufd_dev_class_init,
> +};
> +
> static Property vfio_pci_dev_nohotplug_properties[] = {
> DEFINE_PROP_BOOL("ramfb", VFIOPCIDevice, enable_ramfb, false),
> DEFINE_PROP_ON_OFF_AUTO("x-ramfb-migrate", VFIOPCIDevice, ramfb_migrate,
> @@ -3770,13 +3805,20 @@ static void vfio_pci_nohotplug_dev_class_init(ObjectClass *klass, void *data)
>
> static const TypeInfo vfio_pci_nohotplug_dev_info = {
> .name = TYPE_VFIO_PCI_NOHOTPLUG,
> - .parent = TYPE_VFIO_PCI,
> + .parent = TYPE_VFIO_PCI_IOMMUFD,
> .instance_size = sizeof(VFIOPCIDevice),
> .class_init = vfio_pci_nohotplug_dev_class_init,
> };
>
> static void register_vfio_pci_dev_type(void)
> {
> + static const TypeInfo iommu_be_ops_interface_info = {
> + .name = TYPE_IOMMU_BACKEND_OPS,
> + .parent = TYPE_INTERFACE,
> + .class_size = sizeof(VFIOIOMMUBackendOpsClass),
> + };
> +
> + type_register_static(&iommu_be_ops_interface_info);
> type_register_static(&vfio_pci_dev_info);
> type_register_static(&vfio_pci_nohotplug_dev_info);
> }
>
> Thanks
> Zhenzhong
>
>> Anyhow, the series looks pretty good. There are other aspect to
>> check, like are all this iommu ops well suited for the need ?
>> Let's stress the models in parallel of the reviews. If we could get
>> some of it for 8.2 that'd be nice. It's on top of my list now.
>>
>> Thanks,
>>
>> C.
>>
>>
>>> Thanks
>>> Zhenzhong
^ permalink raw reply [flat|nested] 43+ messages in thread
* RE: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-20 8:19 ` Eric Auger
@ 2023-10-20 8:28 ` Duan, Zhenzhong
2023-10-23 15:28 ` Cédric Le Goater
1 sibling, 0 replies; 43+ messages in thread
From: Duan, Zhenzhong @ 2023-10-20 8:28 UTC (permalink / raw)
To: eric.auger, Cédric Le Goater, qemu-devel
Cc: alex.williamson, jgg, nicolinc, Martins, Joao, peterx, jasowang,
Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng, Chao P, Yi Sun,
David Gibson
>-----Original Message-----
>From: Eric Auger <eric.auger@redhat.com>
>Sent: Friday, October 20, 2023 4:19 PM
>Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and
>targetted interface
>
>Hi,
>On 10/20/23 07:48, Duan, Zhenzhong wrote:
>> Hi Cédric,
>>
>>> -----Original Message-----
>>> From: Cédric Le Goater <clg@redhat.com>
>>> Sent: Thursday, October 19, 2023 8:18 PM
>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer
>and
>>> targetted interface
>>>
>>> On 10/19/23 04:29, Duan, Zhenzhong wrote:
>>>>> -----Original Message-----
>>>>> From: Cédric Le Goater <clg@redhat.com>
>>>>> Sent: Wednesday, October 18, 2023 4:04 PM
>>>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer
>>> and
>>>>> targetted interface
>>>>>
>>>>> On 10/18/23 04:41, Duan, Zhenzhong wrote:
>>>>>> Hi Cédric,
>>>>>>
>>>>>>> -----Original Message-----
>>>>>>> From: Cédric Le Goater <clg@redhat.com>
>>>>>>> Sent: Tuesday, October 17, 2023 11:51 PM
>>>>>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for
>VFIOContainer
>>>>> and
>>>>>>> targetted interface
>>>>>>>
>>>>>>> On 10/16/23 10:31, Zhenzhong Duan wrote:
>>>>>>>> From: Eric Auger <eric.auger@redhat.com>
>>>>>>>>
>>>>>>>> Introduce a dumb VFIOContainer base object and its targetted interface.
>>>>>>>> This is willingly not a QOM object because we don't want it to be
>>>>>>>> visible from the user interface. The VFIOContainer will be smoothly
>>>>>>>> populated in subsequent patches as well as interfaces.
>>>>>>>>
>>>>>>>> No fucntional change intended.
>>>>>>>>
>>>>>>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>>>>>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>>>>>>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>>>>>>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>>>>>>> ---
>>>>>>>> include/hw/vfio/vfio-common.h | 8 +--
>>>>>>>> include/hw/vfio/vfio-container-base.h | 82
>>>>> +++++++++++++++++++++++++++
>>>>>>>> 2 files changed, 84 insertions(+), 6 deletions(-)
>>>>>>>> create mode 100644 include/hw/vfio/vfio-container-base.h
>>>>>>>>
>>>>>>>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-
>>>>> common.h
>>>>>>>> index 34648e518e..9651cf921c 100644
>>>>>>>> --- a/include/hw/vfio/vfio-common.h
>>>>>>>> +++ b/include/hw/vfio/vfio-common.h
>>>>>>>> @@ -30,6 +30,7 @@
>>>>>>>> #include <linux/vfio.h>
>>>>>>>> #endif
>>>>>>>> #include "sysemu/sysemu.h"
>>>>>>>> +#include "hw/vfio/vfio-container-base.h"
>>>>>>>>
>>>>>>>> #define VFIO_MSG_PREFIX "vfio %s: "
>>>>>>>>
>>>>>>>> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
>>>>>>>> struct VFIOGroup;
>>>>>>>>
>>>>>>>> typedef struct VFIOLegacyContainer {
>>>>>>>> + VFIOContainer bcontainer;
>>>>>>> That's the parent class, right ?
>>>>>> Right.
>>>>>>
>>>>>>>> VFIOAddressSpace *space;
>>>>>>>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>>>>>>>> MemoryListener listener;
>>>>>>>> @@ -200,12 +202,6 @@ typedef struct VFIODisplay {
>>>>>>>> } dmabuf;
>>>>>>>> } VFIODisplay;
>>>>>>>>
>>>>>>>> -typedef struct {
>>>>>>>> - unsigned long *bitmap;
>>>>>>>> - hwaddr size;
>>>>>>>> - hwaddr pages;
>>>>>>>> -} VFIOBitmap;
>>>>>>>> -
>>>>>>>> void vfio_host_win_add(VFIOLegacyContainer *container,
>>>>>>>> hwaddr min_iova, hwaddr max_iova,
>>>>>>>> uint64_t iova_pgsizes);
>>>>>>>> diff --git a/include/hw/vfio/vfio-container-base.h
>b/include/hw/vfio/vfio-
>>>>>>> container-base.h
>>>>>>>> new file mode 100644
>>>>>>>> index 0000000000..afc8543d22
>>>>>>>> --- /dev/null
>>>>>>>> +++ b/include/hw/vfio/vfio-container-base.h
>>>>>>>> @@ -0,0 +1,82 @@
>>>>>>>> +/*
>>>>>>>> + * VFIO BASE CONTAINER
>>>>>>>> + *
>>>>>>>> + * Copyright (C) 2023 Intel Corporation.
>>>>>>>> + * Copyright Red Hat, Inc. 2023
>>>>>>>> + *
>>>>>>>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>>>>>>>> + * Eric Auger <eric.auger@redhat.com>
>>>>>>>> + *
>>>>>>>> + * This program is free software; you can redistribute it and/or modify
>>>>>>>> + * it under the terms of the GNU General Public License as published by
>>>>>>>> + * the Free Software Foundation; either version 2 of the License, or
>>>>>>>> + * (at your option) any later version.
>>>>>>>> +
>>>>>>>> + * This program is distributed in the hope that it will be useful,
>>>>>>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty
>of
>>>>>>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See
>the
>>>>>>>> + * GNU General Public License for more details.
>>>>>>>> +
>>>>>>>> + * You should have received a copy of the GNU General Public License
>>> along
>>>>>>>> + * with this program; if not, see <http://www.gnu.org/licenses/>.
>>>>>>>> + */
>>>>>>>> +
>>>>>>>> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
>>>>>>>> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
>>>>>>>> +
>>>>>>>> +#include "exec/memory.h"
>>>>>>>> +#ifndef CONFIG_USER_ONLY
>>>>>>>> +#include "exec/hwaddr.h"
>>>>>>>> +#endif
>>>>>>>> +
>>>>>>>> +typedef struct VFIOContainer VFIOContainer;
>>>>>>>> +typedef struct VFIODevice VFIODevice;
>>>>>>>> +typedef struct VFIOIOMMUBackendOpsClass
>>> VFIOIOMMUBackendOpsClass;
>>>>>>>> +
>>>>>>>> +typedef struct {
>>>>>>>> + unsigned long *bitmap;
>>>>>>>> + hwaddr size;
>>>>>>>> + hwaddr pages;
>>>>>>>> +} VFIOBitmap;
>>>>>>>> +
>>>>>>>> +/*
>>>>>>>> + * This is the base object for vfio container backends
>>>>>>>> + */
>>>>>>>> +struct VFIOContainer {
>>>>>>>> + VFIOIOMMUBackendOpsClass *ops;
>>>>>>> This is unexpected.
>>>>>>>
>>>>>>> I thought that an abstract QOM model for VFIOContainer was going
>>>>>>> to be introduced with a VFIOContainerClass with the ops below
>>>>>>> (VFIOIOMMUBackendOpsClass).
>>>>>>>
>>>>>>> Then, we would call :
>>>>>>>
>>>>>>> VFIOContainerClass *vcc = VFIO_CONTAINER_GET_CLASS(container);
>>>>>>>
>>>>>>> to get the specific implementation for the current container.
>>>>>>>
>>>>>>> I don't understand the VFIOIOMMUBackendOpsClass pointer and
>>>>>>> TYPE_VFIO_IOMMU_BACKEND_OPS. It seems redundant.
>>>>>> The original implementation was abstract QOM model. But it wasn't
>>> accepted,
>>>>>> see https://lore.kernel.org/all/YmuFv2s5TPuw7K%2Fu@yekko/ for details.
>>>>> I see the idea was challenged, not rejected. I need to dig in further and this
>>>>> will take time.
>>>> Thanks for help looking into it.
>>>>
>>>> +David, Hi David, I'm digging into your concern of using QOM to abstract
>base
>>>> container and legacy VFIOContainer:
>>>> "The QOM class of things is visible to the user/config layer via QMP (and
>>> sometimes command line).
>>>> It doesn't necessarily correspond to guest visible differences, but it often
>does."
>>>>
>>>> AIUI, while it's true when the QOM type includes TYPE_USER_CREATABLE
>>> interface,
>>>> otherwise, user can't create an object of this type. Only difference is user will
>>> see
>>>> "object type '%s' isn't supported by object-add" instead of "invalid object
>>> type: %s".
>>>> Is your expectation to not permit user to create this object or only want user
>>>> to see "invalid object type: %s".
>>>> If you mean the first, then I think QOM could be suitable if we don't include
>>>> TYPE_USER_CREATABLE interface?
>>> I was imagining some kind of QOM hierarchy under the vfio device
>>> with various QOM interfaces (similar to the ops) to define the
>>> possible IOMMU backends. The fact that we use the IOMMUFD object
>> >from the command line made it more plausible. I might be mistaking.
>>
>> Got your point.
>> This way we introduce a new QOM type "vfio-pci-iommufd" for iommufd
>support,
>> and vfio-pci keep same for legacy backend, e.g:
>>
>> #qemu -object iommufd,id=iommufd0 \
>> -device vfio-pci-iommufd,iommufd=iommufd0,id=vfio0... \
>> -device vfio-pci,id=vfio1...
>you would need to do that for all types for vfio devices, ap, ccw,
>platform. Looks heavy to me.
Yes, this is a try out following Cédric's imagination, pasted here just for discuss.
>Why would we need to use a different
>vfio-pci-* device while we could switch the iommu backend according to
>the "iommufd" prop presence. The initial discussion was about QOMyfying
>the container instead.
We indeed are switching backend according to "iommufd" prop though not
QOMtfying the container in this series.
Thanks
Zhenzhong
>
>Thanks
>
>Eric
>
>>
>> Below is a draft for vfio-pci based on your imagination. not pasted here the
>change
>> for platform/ap/ccw vfio device which should be similar.
>> Not clear if this fully match your imagination.
>>
>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>container-base.h
>> index 5345986993..54a6ce4d73 100644
>> --- a/include/hw/vfio/vfio-container-base.h
>> +++ b/include/hw/vfio/vfio-container-base.h
>> @@ -124,7 +124,7 @@
>DECLARE_CLASS_CHECKERS(VFIOIOMMUBackendOpsClass,
>>
>> struct VFIOIOMMUBackendOpsClass {
>> /*< private >*/
>> - ObjectClass parent_class;
>> + InterfaceClass parent_class;
>>
>> /*< public >*/
>> /* required */
>> diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
>> index edb787d3d1..829deddc7d 100644
>> --- a/hw/vfio/pci.c
>> +++ b/hw/vfio/pci.c
>> @@ -3725,6 +3725,11 @@ static void vfio_pci_dev_class_init(ObjectClass
>*klass, void *data)
>> {
>> DeviceClass *dc = DEVICE_CLASS(klass);
>> PCIDeviceClass *pdc = PCI_DEVICE_CLASS(klass);
>> + IOMMU_Backend_Ops_Class *be_ops =
>IOMMU_BACKEND_OPS_CLASS(klass);
>> +
>> + be_ops->dma_map = vfio_legacy_dma_map;
>> + be_ops->dma_unmap = vfio_legacy_dma_unmap;
>> + ...
>>
>> dc->reset = vfio_pci_reset;
>> device_class_set_props(dc, vfio_pci_dev_properties);
>> @@ -3749,10 +3754,40 @@ static const TypeInfo vfio_pci_dev_info = {
>> .interfaces = (InterfaceInfo[]) {
>> { INTERFACE_PCIE_DEVICE },
>> { INTERFACE_CONVENTIONAL_PCI_DEVICE },
>> + { INTERFACE_IOMMU_BACKEND_OPS },
>> { }
>> },
>> };
>>
>> +static Property vfio_pci_dev_iommufd_properties[] = {
>> +#ifdef CONFIG_IOMMUFD
>> + DEFINE_PROP_LINK("iommufd", VFIOPCIDevice, vbasedev.iommufd,
>> + TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
>> +#endif
>> +};
>> +
>> +static void vfio_pci_iommufd_dev_class_init(ObjectClass *klass, void *data)
>> +{
>> + DeviceClass *dc = DEVICE_CLASS(klass);
>> + IOMMU_Backend_Ops_Class *be_ops =
>IOMMU_BACKEND_OPS_CLASS(klass);
>> +
>> + device_class_set_props(dc, vfio_pci_dev_iommufd_properties);
>> +
>> + be_ops->dma_map = iommufd_map;
>> + be_ops->dma_unmap = iommufd_unmap;
>> + ...
>> + /* Unimplemented ops */
>> + be_ops->set_dirty_page_tracking = NULL;
>> + ...
>> +}
>> +
>> +static const TypeInfo vfio_pci_iommufd_dev_info = {
>> + .name = TYPE_VFIO_PCI_IOMMUFD,
>> + .parent = TYPE_VFIO_PCI,
>> + .instance_size = sizeof(VFIOPCIDevice),
>> + .class_init = vfio_pci_iommufd_dev_class_init,
>> +};
>> +
>> static Property vfio_pci_dev_nohotplug_properties[] = {
>> DEFINE_PROP_BOOL("ramfb", VFIOPCIDevice, enable_ramfb, false),
>> DEFINE_PROP_ON_OFF_AUTO("x-ramfb-migrate", VFIOPCIDevice,
>ramfb_migrate,
>> @@ -3770,13 +3805,20 @@ static void
>vfio_pci_nohotplug_dev_class_init(ObjectClass *klass, void *data)
>>
>> static const TypeInfo vfio_pci_nohotplug_dev_info = {
>> .name = TYPE_VFIO_PCI_NOHOTPLUG,
>> - .parent = TYPE_VFIO_PCI,
>> + .parent = TYPE_VFIO_PCI_IOMMUFD,
>> .instance_size = sizeof(VFIOPCIDevice),
>> .class_init = vfio_pci_nohotplug_dev_class_init,
>> };
>>
>> static void register_vfio_pci_dev_type(void)
>> {
>> + static const TypeInfo iommu_be_ops_interface_info = {
>> + .name = TYPE_IOMMU_BACKEND_OPS,
>> + .parent = TYPE_INTERFACE,
>> + .class_size = sizeof(VFIOIOMMUBackendOpsClass),
>> + };
>> +
>> + type_register_static(&iommu_be_ops_interface_info);
>> type_register_static(&vfio_pci_dev_info);
>> type_register_static(&vfio_pci_nohotplug_dev_info);
>> }
>>
>> Thanks
>> Zhenzhong
>>
>>> Anyhow, the series looks pretty good. There are other aspect to
>>> check, like are all this iommu ops well suited for the need ?
>>> Let's stress the models in parallel of the reviews. If we could get
>>> some of it for 8.2 that'd be nice. It's on top of my list now.
>>>
>>> Thanks,
>>>
>>> C.
>>>
>>>
>>>> Thanks
>>>> Zhenzhong
^ permalink raw reply [flat|nested] 43+ messages in thread
* Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-20 8:19 ` Eric Auger
2023-10-20 8:28 ` Duan, Zhenzhong
@ 2023-10-23 15:28 ` Cédric Le Goater
2023-10-24 6:03 ` Duan, Zhenzhong
1 sibling, 1 reply; 43+ messages in thread
From: Cédric Le Goater @ 2023-10-23 15:28 UTC (permalink / raw)
To: eric.auger, Duan, Zhenzhong, qemu-devel
Cc: alex.williamson, jgg, nicolinc, Martins, Joao, peterx, jasowang,
Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng, Chao P, Yi Sun,
David Gibson
On 10/20/23 10:19, Eric Auger wrote:
> Hi,
> On 10/20/23 07:48, Duan, Zhenzhong wrote:
>> Hi Cédric,
>>
>>> -----Original Message-----
>>> From: Cédric Le Goater <clg@redhat.com>
>>> Sent: Thursday, October 19, 2023 8:18 PM
>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and
>>> targetted interface
>>>
>>> On 10/19/23 04:29, Duan, Zhenzhong wrote:
>>>>> -----Original Message-----
>>>>> From: Cédric Le Goater <clg@redhat.com>
>>>>> Sent: Wednesday, October 18, 2023 4:04 PM
>>>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer
>>> and
>>>>> targetted interface
>>>>>
>>>>> On 10/18/23 04:41, Duan, Zhenzhong wrote:
>>>>>> Hi Cédric,
>>>>>>
>>>>>>> -----Original Message-----
>>>>>>> From: Cédric Le Goater <clg@redhat.com>
>>>>>>> Sent: Tuesday, October 17, 2023 11:51 PM
>>>>>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer
>>>>> and
>>>>>>> targetted interface
>>>>>>>
>>>>>>> On 10/16/23 10:31, Zhenzhong Duan wrote:
>>>>>>>> From: Eric Auger <eric.auger@redhat.com>
>>>>>>>>
>>>>>>>> Introduce a dumb VFIOContainer base object and its targetted interface.
>>>>>>>> This is willingly not a QOM object because we don't want it to be
>>>>>>>> visible from the user interface. The VFIOContainer will be smoothly
>>>>>>>> populated in subsequent patches as well as interfaces.
>>>>>>>>
>>>>>>>> No fucntional change intended.
>>>>>>>>
>>>>>>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>>>>>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>>>>>>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>>>>>>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>>>>>>> ---
>>>>>>>> include/hw/vfio/vfio-common.h | 8 +--
>>>>>>>> include/hw/vfio/vfio-container-base.h | 82
>>>>> +++++++++++++++++++++++++++
>>>>>>>> 2 files changed, 84 insertions(+), 6 deletions(-)
>>>>>>>> create mode 100644 include/hw/vfio/vfio-container-base.h
>>>>>>>>
>>>>>>>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-
>>>>> common.h
>>>>>>>> index 34648e518e..9651cf921c 100644
>>>>>>>> --- a/include/hw/vfio/vfio-common.h
>>>>>>>> +++ b/include/hw/vfio/vfio-common.h
>>>>>>>> @@ -30,6 +30,7 @@
>>>>>>>> #include <linux/vfio.h>
>>>>>>>> #endif
>>>>>>>> #include "sysemu/sysemu.h"
>>>>>>>> +#include "hw/vfio/vfio-container-base.h"
>>>>>>>>
>>>>>>>> #define VFIO_MSG_PREFIX "vfio %s: "
>>>>>>>>
>>>>>>>> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
>>>>>>>> struct VFIOGroup;
>>>>>>>>
>>>>>>>> typedef struct VFIOLegacyContainer {
>>>>>>>> + VFIOContainer bcontainer;
>>>>>>> That's the parent class, right ?
>>>>>> Right.
>>>>>>
>>>>>>>> VFIOAddressSpace *space;
>>>>>>>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>>>>>>>> MemoryListener listener;
>>>>>>>> @@ -200,12 +202,6 @@ typedef struct VFIODisplay {
>>>>>>>> } dmabuf;
>>>>>>>> } VFIODisplay;
>>>>>>>>
>>>>>>>> -typedef struct {
>>>>>>>> - unsigned long *bitmap;
>>>>>>>> - hwaddr size;
>>>>>>>> - hwaddr pages;
>>>>>>>> -} VFIOBitmap;
>>>>>>>> -
>>>>>>>> void vfio_host_win_add(VFIOLegacyContainer *container,
>>>>>>>> hwaddr min_iova, hwaddr max_iova,
>>>>>>>> uint64_t iova_pgsizes);
>>>>>>>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>>>>>>> container-base.h
>>>>>>>> new file mode 100644
>>>>>>>> index 0000000000..afc8543d22
>>>>>>>> --- /dev/null
>>>>>>>> +++ b/include/hw/vfio/vfio-container-base.h
>>>>>>>> @@ -0,0 +1,82 @@
>>>>>>>> +/*
>>>>>>>> + * VFIO BASE CONTAINER
>>>>>>>> + *
>>>>>>>> + * Copyright (C) 2023 Intel Corporation.
>>>>>>>> + * Copyright Red Hat, Inc. 2023
>>>>>>>> + *
>>>>>>>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>>>>>>>> + * Eric Auger <eric.auger@redhat.com>
>>>>>>>> + *
>>>>>>>> + * This program is free software; you can redistribute it and/or modify
>>>>>>>> + * it under the terms of the GNU General Public License as published by
>>>>>>>> + * the Free Software Foundation; either version 2 of the License, or
>>>>>>>> + * (at your option) any later version.
>>>>>>>> +
>>>>>>>> + * This program is distributed in the hope that it will be useful,
>>>>>>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>>>>>>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>>>>>>>> + * GNU General Public License for more details.
>>>>>>>> +
>>>>>>>> + * You should have received a copy of the GNU General Public License
>>> along
>>>>>>>> + * with this program; if not, see <http://www.gnu.org/licenses/>.
>>>>>>>> + */
>>>>>>>> +
>>>>>>>> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
>>>>>>>> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
>>>>>>>> +
>>>>>>>> +#include "exec/memory.h"
>>>>>>>> +#ifndef CONFIG_USER_ONLY
>>>>>>>> +#include "exec/hwaddr.h"
>>>>>>>> +#endif
>>>>>>>> +
>>>>>>>> +typedef struct VFIOContainer VFIOContainer;
>>>>>>>> +typedef struct VFIODevice VFIODevice;
>>>>>>>> +typedef struct VFIOIOMMUBackendOpsClass
>>> VFIOIOMMUBackendOpsClass;
>>>>>>>> +
>>>>>>>> +typedef struct {
>>>>>>>> + unsigned long *bitmap;
>>>>>>>> + hwaddr size;
>>>>>>>> + hwaddr pages;
>>>>>>>> +} VFIOBitmap;
>>>>>>>> +
>>>>>>>> +/*
>>>>>>>> + * This is the base object for vfio container backends
>>>>>>>> + */
>>>>>>>> +struct VFIOContainer {
>>>>>>>> + VFIOIOMMUBackendOpsClass *ops;
>>>>>>> This is unexpected.
>>>>>>>
>>>>>>> I thought that an abstract QOM model for VFIOContainer was going
>>>>>>> to be introduced with a VFIOContainerClass with the ops below
>>>>>>> (VFIOIOMMUBackendOpsClass).
>>>>>>>
>>>>>>> Then, we would call :
>>>>>>>
>>>>>>> VFIOContainerClass *vcc = VFIO_CONTAINER_GET_CLASS(container);
>>>>>>>
>>>>>>> to get the specific implementation for the current container.
>>>>>>>
>>>>>>> I don't understand the VFIOIOMMUBackendOpsClass pointer and
>>>>>>> TYPE_VFIO_IOMMU_BACKEND_OPS. It seems redundant.
>>>>>> The original implementation was abstract QOM model. But it wasn't
>>> accepted,
>>>>>> see https://lore.kernel.org/all/YmuFv2s5TPuw7K%2Fu@yekko/ for details.
>>>>> I see the idea was challenged, not rejected. I need to dig in further and this
>>>>> will take time.
>>>> Thanks for help looking into it.
>>>>
>>>> +David, Hi David, I'm digging into your concern of using QOM to abstract base
>>>> container and legacy VFIOContainer:
>>>> "The QOM class of things is visible to the user/config layer via QMP (and
>>> sometimes command line).
>>>> It doesn't necessarily correspond to guest visible differences, but it often does."
>>>>
>>>> AIUI, while it's true when the QOM type includes TYPE_USER_CREATABLE
>>> interface,
>>>> otherwise, user can't create an object of this type. Only difference is user will
>>> see
>>>> "object type '%s' isn't supported by object-add" instead of "invalid object
>>> type: %s".
>>>> Is your expectation to not permit user to create this object or only want user
>>>> to see "invalid object type: %s".
>>>> If you mean the first, then I think QOM could be suitable if we don't include
>>>> TYPE_USER_CREATABLE interface?
>>> I was imagining some kind of QOM hierarchy under the vfio device
>>> with various QOM interfaces (similar to the ops) to define the
>>> possible IOMMU backends. The fact that we use the IOMMUFD object
>> >from the command line made it more plausible. I might be mistaking.
>>
>> Got your point.
>> This way we introduce a new QOM type "vfio-pci-iommufd" for iommufd support,
>> and vfio-pci keep same for legacy backend, e.g:
>>
>> #qemu -object iommufd,id=iommufd0 \
>> -device vfio-pci-iommufd,iommufd=iommufd0,id=vfio0... \
>> -device vfio-pci,id=vfio1...
> you would need to do that for all types for vfio devices, ap, ccw,
> platform. Looks heavy to me. Why would we need to use a different
> vfio-pci-* device while we could switch the iommu backend according to
> the "iommufd" prop presence. The initial discussion was about QOMyfying
> the container instead.
yes.
I took a closer look at the first part which adds the backend ops,
including patch 19 adding the iommufd backend, not saying that I have
identified all the dark corners.
A QOM-like design would have introduced a VFIOLegacyContainer,
inheriting from VFIOContainer (same for VFIOIOMMUFDContainer) with a
VFIOContainerClass to implement the specific backend ops.
VFIOspaprContainer would have made sense also.
But QOM doesn't seem well adapted for the current needs. So let's try
a simpler approach. It seems that VFIOIOMMUBackendOpsClass is
useless. IMO, it could be a callbacks structure like we have for
memory regions initialized with vfio_container_init(). This would
remove some noise around the QOM typeinfo definitions.
'struct vfio_iommu_ops' reads/sounds like a good name.
Can we try that in a v3 ? It should not be such an earthquake.
spapr has some singularities which would be good to isolate in a
vfio_iommu_spapr_ops to remove all the VFIO_SPAPR_TCE_*_IOMMU code in
container.c. vfio_legacy_{add,del}_section_window are SPAPR specific.
FYI, I did some adjustements bc of the recent introduction of iova_ranges
in my branch :
https://github.com/legoater/qemu/commits/vfio-8.2
Thanks,
C.
^ permalink raw reply [flat|nested] 43+ messages in thread
* RE: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-23 15:28 ` Cédric Le Goater
@ 2023-10-24 6:03 ` Duan, Zhenzhong
2023-10-24 6:51 ` Cédric Le Goater
0 siblings, 1 reply; 43+ messages in thread
From: Duan, Zhenzhong @ 2023-10-24 6:03 UTC (permalink / raw)
To: Cédric Le Goater, eric.auger, qemu-devel
Cc: alex.williamson, jgg, nicolinc, Martins, Joao, peterx, jasowang,
Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng, Chao P, Yi Sun,
David Gibson
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Monday, October 23, 2023 11:29 PM
><david@gibson.dropbear.id.au>
>Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and
>targetted interface
>
>On 10/20/23 10:19, Eric Auger wrote:
>> Hi,
>> On 10/20/23 07:48, Duan, Zhenzhong wrote:
>>> Hi Cédric,
>>>
>>>> -----Original Message-----
>>>> From: Cédric Le Goater <clg@redhat.com>
>>>> Sent: Thursday, October 19, 2023 8:18 PM
>>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer
>and
>>>> targetted interface
>>>>
>>>> On 10/19/23 04:29, Duan, Zhenzhong wrote:
>>>>>> -----Original Message-----
>>>>>> From: Cédric Le Goater <clg@redhat.com>
>>>>>> Sent: Wednesday, October 18, 2023 4:04 PM
>>>>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for
>VFIOContainer
>>>> and
>>>>>> targetted interface
>>>>>>
>>>>>> On 10/18/23 04:41, Duan, Zhenzhong wrote:
>>>>>>> Hi Cédric,
>>>>>>>
>>>>>>>> -----Original Message-----
>>>>>>>> From: Cédric Le Goater <clg@redhat.com>
>>>>>>>> Sent: Tuesday, October 17, 2023 11:51 PM
>>>>>>>> Subject: Re: [PATCH v2 02/27] vfio: Introduce base object for
>VFIOContainer
>>>>>> and
>>>>>>>> targetted interface
>>>>>>>>
>>>>>>>> On 10/16/23 10:31, Zhenzhong Duan wrote:
>>>>>>>>> From: Eric Auger <eric.auger@redhat.com>
>>>>>>>>>
>>>>>>>>> Introduce a dumb VFIOContainer base object and its targetted
>interface.
>>>>>>>>> This is willingly not a QOM object because we don't want it to be
>>>>>>>>> visible from the user interface. The VFIOContainer will be smoothly
>>>>>>>>> populated in subsequent patches as well as interfaces.
>>>>>>>>>
>>>>>>>>> No fucntional change intended.
>>>>>>>>>
>>>>>>>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>>>>>>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>>>>>>>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>>>>>>>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>>>>>>>> ---
>>>>>>>>> include/hw/vfio/vfio-common.h | 8 +--
>>>>>>>>> include/hw/vfio/vfio-container-base.h | 82
>>>>>> +++++++++++++++++++++++++++
>>>>>>>>> 2 files changed, 84 insertions(+), 6 deletions(-)
>>>>>>>>> create mode 100644 include/hw/vfio/vfio-container-base.h
>>>>>>>>>
>>>>>>>>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-
>>>>>> common.h
>>>>>>>>> index 34648e518e..9651cf921c 100644
>>>>>>>>> --- a/include/hw/vfio/vfio-common.h
>>>>>>>>> +++ b/include/hw/vfio/vfio-common.h
>>>>>>>>> @@ -30,6 +30,7 @@
>>>>>>>>> #include <linux/vfio.h>
>>>>>>>>> #endif
>>>>>>>>> #include "sysemu/sysemu.h"
>>>>>>>>> +#include "hw/vfio/vfio-container-base.h"
>>>>>>>>>
>>>>>>>>> #define VFIO_MSG_PREFIX "vfio %s: "
>>>>>>>>>
>>>>>>>>> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
>>>>>>>>> struct VFIOGroup;
>>>>>>>>>
>>>>>>>>> typedef struct VFIOLegacyContainer {
>>>>>>>>> + VFIOContainer bcontainer;
>>>>>>>> That's the parent class, right ?
>>>>>>> Right.
>>>>>>>
>>>>>>>>> VFIOAddressSpace *space;
>>>>>>>>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>>>>>>>>> MemoryListener listener;
>>>>>>>>> @@ -200,12 +202,6 @@ typedef struct VFIODisplay {
>>>>>>>>> } dmabuf;
>>>>>>>>> } VFIODisplay;
>>>>>>>>>
>>>>>>>>> -typedef struct {
>>>>>>>>> - unsigned long *bitmap;
>>>>>>>>> - hwaddr size;
>>>>>>>>> - hwaddr pages;
>>>>>>>>> -} VFIOBitmap;
>>>>>>>>> -
>>>>>>>>> void vfio_host_win_add(VFIOLegacyContainer *container,
>>>>>>>>> hwaddr min_iova, hwaddr max_iova,
>>>>>>>>> uint64_t iova_pgsizes);
>>>>>>>>> diff --git a/include/hw/vfio/vfio-container-base.h
>b/include/hw/vfio/vfio-
>>>>>>>> container-base.h
>>>>>>>>> new file mode 100644
>>>>>>>>> index 0000000000..afc8543d22
>>>>>>>>> --- /dev/null
>>>>>>>>> +++ b/include/hw/vfio/vfio-container-base.h
>>>>>>>>> @@ -0,0 +1,82 @@
>>>>>>>>> +/*
>>>>>>>>> + * VFIO BASE CONTAINER
>>>>>>>>> + *
>>>>>>>>> + * Copyright (C) 2023 Intel Corporation.
>>>>>>>>> + * Copyright Red Hat, Inc. 2023
>>>>>>>>> + *
>>>>>>>>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>>>>>>>>> + * Eric Auger <eric.auger@redhat.com>
>>>>>>>>> + *
>>>>>>>>> + * This program is free software; you can redistribute it and/or modify
>>>>>>>>> + * it under the terms of the GNU General Public License as published
>by
>>>>>>>>> + * the Free Software Foundation; either version 2 of the License, or
>>>>>>>>> + * (at your option) any later version.
>>>>>>>>> +
>>>>>>>>> + * This program is distributed in the hope that it will be useful,
>>>>>>>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty
>of
>>>>>>>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See
>the
>>>>>>>>> + * GNU General Public License for more details.
>>>>>>>>> +
>>>>>>>>> + * You should have received a copy of the GNU General Public License
>>>> along
>>>>>>>>> + * with this program; if not, see <http://www.gnu.org/licenses/>.
>>>>>>>>> + */
>>>>>>>>> +
>>>>>>>>> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
>>>>>>>>> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
>>>>>>>>> +
>>>>>>>>> +#include "exec/memory.h"
>>>>>>>>> +#ifndef CONFIG_USER_ONLY
>>>>>>>>> +#include "exec/hwaddr.h"
>>>>>>>>> +#endif
>>>>>>>>> +
>>>>>>>>> +typedef struct VFIOContainer VFIOContainer;
>>>>>>>>> +typedef struct VFIODevice VFIODevice;
>>>>>>>>> +typedef struct VFIOIOMMUBackendOpsClass
>>>> VFIOIOMMUBackendOpsClass;
>>>>>>>>> +
>>>>>>>>> +typedef struct {
>>>>>>>>> + unsigned long *bitmap;
>>>>>>>>> + hwaddr size;
>>>>>>>>> + hwaddr pages;
>>>>>>>>> +} VFIOBitmap;
>>>>>>>>> +
>>>>>>>>> +/*
>>>>>>>>> + * This is the base object for vfio container backends
>>>>>>>>> + */
>>>>>>>>> +struct VFIOContainer {
>>>>>>>>> + VFIOIOMMUBackendOpsClass *ops;
>>>>>>>> This is unexpected.
>>>>>>>>
>>>>>>>> I thought that an abstract QOM model for VFIOContainer was going
>>>>>>>> to be introduced with a VFIOContainerClass with the ops below
>>>>>>>> (VFIOIOMMUBackendOpsClass).
>>>>>>>>
>>>>>>>> Then, we would call :
>>>>>>>>
>>>>>>>> VFIOContainerClass *vcc = VFIO_CONTAINER_GET_CLASS(container);
>>>>>>>>
>>>>>>>> to get the specific implementation for the current container.
>>>>>>>>
>>>>>>>> I don't understand the VFIOIOMMUBackendOpsClass pointer and
>>>>>>>> TYPE_VFIO_IOMMU_BACKEND_OPS. It seems redundant.
>>>>>>> The original implementation was abstract QOM model. But it wasn't
>>>> accepted,
>>>>>>> see https://lore.kernel.org/all/YmuFv2s5TPuw7K%2Fu@yekko/ for
>details.
>>>>>> I see the idea was challenged, not rejected. I need to dig in further and this
>>>>>> will take time.
>>>>> Thanks for help looking into it.
>>>>>
>>>>> +David, Hi David, I'm digging into your concern of using QOM to abstract
>base
>>>>> container and legacy VFIOContainer:
>>>>> "The QOM class of things is visible to the user/config layer via QMP (and
>>>> sometimes command line).
>>>>> It doesn't necessarily correspond to guest visible differences, but it often
>does."
>>>>>
>>>>> AIUI, while it's true when the QOM type includes TYPE_USER_CREATABLE
>>>> interface,
>>>>> otherwise, user can't create an object of this type. Only difference is user
>will
>>>> see
>>>>> "object type '%s' isn't supported by object-add" instead of "invalid object
>>>> type: %s".
>>>>> Is your expectation to not permit user to create this object or only want
>user
>>>>> to see "invalid object type: %s".
>>>>> If you mean the first, then I think QOM could be suitable if we don't include
>>>>> TYPE_USER_CREATABLE interface?
>>>> I was imagining some kind of QOM hierarchy under the vfio device
>>>> with various QOM interfaces (similar to the ops) to define the
>>>> possible IOMMU backends. The fact that we use the IOMMUFD object
>>> >from the command line made it more plausible. I might be mistaking.
>>>
>>> Got your point.
>>> This way we introduce a new QOM type "vfio-pci-iommufd" for iommufd
>support,
>>> and vfio-pci keep same for legacy backend, e.g:
>>>
>>> #qemu -object iommufd,id=iommufd0 \
>>> -device vfio-pci-iommufd,iommufd=iommufd0,id=vfio0... \
>>> -device vfio-pci,id=vfio1...
>> you would need to do that for all types for vfio devices, ap, ccw,
>> platform. Looks heavy to me. Why would we need to use a different
>> vfio-pci-* device while we could switch the iommu backend according to
>> the "iommufd" prop presence. The initial discussion was about QOMyfying
>> the container instead.
>
>yes.
>
>I took a closer look at the first part which adds the backend ops,
>including patch 19 adding the iommufd backend, not saying that I have
>identified all the dark corners.
>
>A QOM-like design would have introduced a VFIOLegacyContainer,
>inheriting from VFIOContainer (same for VFIOIOMMUFDContainer) with a
>VFIOContainerClass to implement the specific backend ops.
>VFIOspaprContainer would have made sense also.
>
>But QOM doesn't seem well adapted for the current needs. So let's try
>a simpler approach. It seems that VFIOIOMMUBackendOpsClass is
>useless. IMO, it could be a callbacks structure like we have for
>memory regions initialized with vfio_container_init(). This would
>remove some noise around the QOM typeinfo definitions.
Yes, good suggestion, will do.
>
>'struct vfio_iommu_ops' reads/sounds like a good name.
>
>Can we try that in a v3 ? It should not be such an earthquake.
Sure.
>
>spapr has some singularities which would be good to isolate in a
>vfio_iommu_spapr_ops to remove all the VFIO_SPAPR_TCE_*_IOMMU code in
>container.c. vfio_legacy_{add,del}_section_window are SPAPR specific.
Yes, let me try it.
Thanks
Zhenzhong
>
>FYI, I did some adjustements bc of the recent introduction of iova_ranges
>in my branch :
>
> https://github.com/legoater/qemu/commits/vfio-8.2
>
>Thanks,
>
>C.
>
^ permalink raw reply [flat|nested] 43+ messages in thread
* Re: [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-24 6:03 ` Duan, Zhenzhong
@ 2023-10-24 6:51 ` Cédric Le Goater
0 siblings, 0 replies; 43+ messages in thread
From: Cédric Le Goater @ 2023-10-24 6:51 UTC (permalink / raw)
To: Duan, Zhenzhong, eric.auger, qemu-devel
Cc: alex.williamson, jgg, nicolinc, Martins, Joao, peterx, jasowang,
Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng, Chao P, Yi Sun,
David Gibson
Hello Zhenzhong,
>>> you would need to do that for all types for vfio devices, ap, ccw,
>>> platform. Looks heavy to me. Why would we need to use a different
>>> vfio-pci-* device while we could switch the iommu backend according to
>>> the "iommufd" prop presence. The initial discussion was about QOMyfying
>>> the container instead.
>>
>> yes.
>>
>> I took a closer look at the first part which adds the backend ops,
>> including patch 19 adding the iommufd backend, not saying that I have
>> identified all the dark corners.
>>
>> A QOM-like design would have introduced a VFIOLegacyContainer,
>> inheriting from VFIOContainer (same for VFIOIOMMUFDContainer) with a
>> VFIOContainerClass to implement the specific backend ops.
>> VFIOspaprContainer would have made sense also.
>>
>> But QOM doesn't seem well adapted for the current needs. So let's try
>> a simpler approach. It seems that VFIOIOMMUBackendOpsClass is
>> useless. IMO, it could be a callbacks structure like we have for
>> memory regions initialized with vfio_container_init(). This would
>> remove some noise around the QOM typeinfo definitions.
>
> Yes, good suggestion, will do.
>
Thanks,
>>
>> 'struct vfio_iommu_ops' reads/sounds like a good name.
This should be read as VFIOIOMMUOps :)
>> Can we try that in a v3 ? It should not be such an earthquake.
>
> Sure.
>
>>
>> spapr has some singularities which would be good to isolate in a
>> vfio_iommu_spapr_ops to remove all the VFIO_SPAPR_TCE_*_IOMMU code in
>> container.c. vfio_legacy_{add,del}_section_window are SPAPR specific.
>
> Yes, let me try it.
There is a large amount of spapr code in container.c. If it could
be moved under its own backend (and in hw/ppc/spapr_pci_vfio.c),
we would have two backends : spapr and legacy. We could possibly
merge this first part in 8.2.
I have access to a PPC P9 system. I can make sure we are not breaking
support.
C.
> Thanks
> Zhenzhong
>>
>> FYI, I did some adjustements bc of the recent introduction of iova_ranges
>> in my branch :
>>
>> https://github.com/legoater/qemu/commits/vfio-8.2
>>
>> Thanks,
>>
>> C.
>>
>
^ permalink raw reply [flat|nested] 43+ messages in thread
end of thread, other threads:[~2023-10-24 6:53 UTC | newest]
Thread overview: 43+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-10-16 8:31 [PATCH v2 00/27] vfio: Adopt iommufd Zhenzhong Duan
2023-10-16 8:31 ` [PATCH v2 01/27] vfio: Rename VFIOContainer into VFIOLegacyContainer Zhenzhong Duan
2023-10-17 15:50 ` Cédric Le Goater
2023-10-18 2:33 ` Duan, Zhenzhong
2023-10-16 8:31 ` [PATCH v2 02/27] vfio: Introduce base object for VFIOContainer and targetted interface Zhenzhong Duan
2023-10-17 15:51 ` Cédric Le Goater
2023-10-18 2:41 ` Duan, Zhenzhong
2023-10-18 8:04 ` Cédric Le Goater
2023-10-19 2:29 ` Duan, Zhenzhong
2023-10-19 12:17 ` Cédric Le Goater
2023-10-20 5:48 ` Duan, Zhenzhong
2023-10-20 8:19 ` Eric Auger
2023-10-20 8:28 ` Duan, Zhenzhong
2023-10-23 15:28 ` Cédric Le Goater
2023-10-24 6:03 ` Duan, Zhenzhong
2023-10-24 6:51 ` Cédric Le Goater
2023-10-16 8:31 ` [PATCH v2 03/27] VFIO/container: Introduce dummy VFIOContainerClass implementation Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 04/27] vfio/container: Switch to dma_map|unmap API Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 05/27] vfio/common: Move giommu_list in base container Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 06/27] vfio/container: Move space field to " Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 07/27] vfio/container: switch to IOMMU BE add/del_section_window Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 08/27] vfio/container: Move hostwin_list in base container Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 09/27] vfio/container: Switch to IOMMU BE set_dirty_page_tracking/query_dirty_bitmap API Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 10/27] vfio/container: Move per container device list in base container Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 11/27] vfio/container: Convert functions to " Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 12/27] vfio/container: Move vrdl_list, pgsizes and dma_max_mappings " Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 13/27] vfio/container: Move listener " Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 14/27] vfio/container: Move dirty_pgsizes and max_dirty_bitmap_size " Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 15/27] vfio/container: Implement attach/detach_device Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 16/27] Add iommufd configure option Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 17/27] backends/iommufd: Introduce the iommufd object Zhenzhong Duan
2023-10-16 10:00 ` Markus Armbruster
2023-10-17 8:27 ` Duan, Zhenzhong
2023-10-16 8:32 ` [PATCH v2 18/27] util/char_dev: Add open_cdev() Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 19/27] vfio/iommufd: Implement the iommufd backend Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 20/27] vfio/container: Bypass EEH if " Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 21/27] vfio/pci: Adapt vfio pci hot reset support with iommufd BE Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 22/27] vfio/pci: Allow the selection of a given iommu backend Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 23/27] vfio/pci: Make vfio cdev pre-openable by passing a file handle Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 24/27] vfio: Allow the selection of a given iommu backend for platform ap and ccw Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 25/27] vfio/platform: Make vfio cdev pre-openable by passing a file handle Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 26/27] vfio/ap: " Zhenzhong Duan
2023-10-16 8:32 ` [PATCH v2 27/27] vfio/ccw: " Zhenzhong Duan
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.