@@ -275,6 +275,18 @@ static void vfio_host_win_add(VFIOContainer *container,
QLIST_INSERT_HEAD(&container->hostwin_list, hostwin, hostwin_next);
}
+static int vfio_host_win_del(VFIOContainer *container, hwaddr min_iova)
+{
+ VFIOHostDMAWindow *hostwin = vfio_host_win_lookup(container, min_iova, 1);
+
+ if (!hostwin) {
+ return -1;
+ }
+ QLIST_REMOVE(hostwin, hostwin_next);
+
+ return 0;
+}
+
static bool vfio_listener_skipped_section(MemoryRegionSection *section)
{
return (!memory_region_is_ram(section->mr) &&
@@ -388,6 +400,30 @@ static void vfio_listener_region_add(MemoryListener *listener,
}
end = int128_get64(int128_sub(llend, int128_one()));
+ if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
+ VFIOHostDMAWindow *hostwin;
+ hwaddr pgsize = 0;
+
+ /* For now intersections are not allowed, we may relax this later */
+ QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
+ if (ranges_overlap(hostwin->min_iova,
+ hostwin->max_iova - hostwin->min_iova + 1,
+ section->offset_within_address_space,
+ int128_get64(section->size))) {
+ goto fail;
+ }
+ }
+
+ ret = vfio_spapr_create_window(container, section, &pgsize);
+ if (ret) {
+ goto fail;
+ }
+
+ vfio_host_win_add(container, section->offset_within_address_space,
+ section->offset_within_address_space +
+ int128_get64(section->size) - 1, pgsize);
+ }
+
if (!vfio_host_win_lookup(container, iova, end)) {
error_report("vfio: IOMMU container %p can't map guest IOVA region"
" 0x%"HWADDR_PRIx"..0x%"HWADDR_PRIx,
@@ -523,6 +559,18 @@ static void vfio_listener_region_del(MemoryListener *listener,
"0x%"HWADDR_PRIx") = %d (%m)",
container, iova, int128_get64(llsize), ret);
}
+
+ if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
+ vfio_spapr_remove_window(container,
+ section->offset_within_address_space);
+ if (vfio_host_win_del(container,
+ section->offset_within_address_space) < 0) {
+ hw_error("%s: Cannot delete missing window at %"HWADDR_PRIx,
+ __func__, section->offset_within_address_space);
+ }
+
+ trace_vfio_spapr_remove_window(section->offset_within_address_space);
+ }
}
static const MemoryListener vfio_memory_listener = {
@@ -960,11 +1008,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as)
}
}
- /*
- * This only considers the host IOMMU's 32-bit window. At
- * some point we need to add support for the optional 64-bit
- * window and dynamic windows
- */
info.argsz = sizeof(info);
ret = ioctl(fd, VFIO_IOMMU_SPAPR_TCE_GET_INFO, &info);
if (ret) {
@@ -973,11 +1016,24 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as)
goto listener_release_exit;
}
- /* The default table uses 4K pages */
- vfio_host_win_add(container, info.dma32_window_start,
- info.dma32_window_start +
- info.dma32_window_size - 1,
- 0x1000);
+ if (v2) {
+ /*
+ * There is a default window in just created container.
+ * To make region_add/del simpler, we better remove this
+ * window now and let those iommu_listener callbacks
+ * create/remove them when needed.
+ */
+ ret = vfio_spapr_remove_window(container, info.dma32_window_start);
+ if (ret) {
+ goto free_container_exit;
+ }
+ } else {
+ /* The default table uses 4K pages */
+ vfio_host_win_add(container, info.dma32_window_start,
+ info.dma32_window_start +
+ info.dma32_window_size - 1,
+ 0x1000);
+ }
} else {
error_report("vfio: No available IOMMU models");
ret = -EINVAL;
@@ -135,3 +135,73 @@ const MemoryListener vfio_prereg_listener = {
.region_add = vfio_prereg_listener_region_add,
.region_del = vfio_prereg_listener_region_del,
};
+
+int vfio_spapr_create_window(VFIOContainer *container,
+ MemoryRegionSection *section,
+ hwaddr *pgsize)
+{
+ int ret;
+ unsigned pagesizes = memory_region_iommu_get_page_sizes(section->mr);
+ unsigned pagesize = (hwaddr)1 << ctz64(pagesizes);
+ unsigned entries, pages;
+ struct vfio_iommu_spapr_tce_create create = { .argsz = sizeof(create) };
+
+ /*
+ * FIXME: For VFIO iommu types which have KVM acceleration to
+ * avoid bouncing all map/unmaps through qemu this way, this
+ * would be the right place to wire that up (tell the KVM
+ * device emulation the VFIO iommu handles to use).
+ */
+ create.window_size = int128_get64(section->size);
+ create.page_shift = ctz64(pagesize);
+ /*
+ * SPAPR host supports multilevel TCE tables, there is some
+ * heuristic to decide how many levels we want for our table:
+ * 0..64 = 1; 65..4096 = 2; 4097..262144 = 3; 262145.. = 4
+ */
+ entries = create.window_size >> create.page_shift;
+ pages = MAX((entries * sizeof(uint64_t)) / getpagesize(), 1);
+ pages = MAX(pow2ceil(pages) - 1, 1); /* Round up */
+ create.levels = ctz64(pages) / 6 + 1;
+
+ ret = ioctl(container->fd, VFIO_IOMMU_SPAPR_TCE_CREATE, &create);
+ if (ret) {
+ error_report("Failed to create a window, ret = %d (%m)", ret);
+ return -errno;
+ }
+
+ if (create.start_addr != section->offset_within_address_space) {
+ vfio_spapr_remove_window(container, create.start_addr);
+
+ error_report("Host doesn't support DMA window at %"HWADDR_PRIx", must be %"PRIx64,
+ section->offset_within_address_space,
+ create.start_addr);
+ ioctl(container->fd, VFIO_IOMMU_SPAPR_TCE_REMOVE, &remove);
+ return -EINVAL;
+ }
+ trace_vfio_spapr_create_window(create.page_shift,
+ create.window_size,
+ create.start_addr);
+ *pgsize = pagesize;
+
+ return 0;
+}
+
+int vfio_spapr_remove_window(VFIOContainer *container,
+ hwaddr offset_within_address_space)
+{
+ struct vfio_iommu_spapr_tce_remove remove = {
+ .argsz = sizeof(remove),
+ .start_addr = offset_within_address_space,
+ };
+ int ret;
+
+ ret = ioctl(container->fd, VFIO_IOMMU_SPAPR_TCE_REMOVE, &remove);
+ if (ret) {
+ error_report("Failed to remove window at %"PRIx64,
+ remove.start_addr);
+ return -errno;
+ }
+
+ return 0;
+}
@@ -167,4 +167,10 @@ int vfio_get_dev_region_info(VFIODevice *vbasedev, uint32_t type,
#endif
extern const MemoryListener vfio_prereg_listener;
+int vfio_spapr_create_window(VFIOContainer *container,
+ MemoryRegionSection *section,
+ hwaddr *pgsize);
+int vfio_spapr_remove_window(VFIOContainer *container,
+ hwaddr offset_within_address_space);
+
#endif /* !HW_VFIO_VFIO_COMMON_H */
@@ -1768,6 +1768,8 @@ vfio_region_sparse_mmap_entry(int i, unsigned long start, unsigned long end) "sp
vfio_get_dev_region(const char *name, int index, uint32_t type, uint32_t subtype) "%s index %d, %08x/%0x8"
vfio_ram_register(uint64_t va, uint64_t size, int ret) "va=%"PRIx64" size=%"PRIx64" ret=%d"
vfio_ram_unregister(uint64_t va, uint64_t size, int ret) "va=%"PRIx64" size=%"PRIx64" ret=%d"
+vfio_spapr_create_window(int ps, uint64_t ws, uint64_t off) "pageshift=0x%x winsize=0x%"PRIx64" offset=0x%"PRIx64
+vfio_spapr_remove_window(uint64_t off) "offset=%"PRIx64
# hw/vfio/platform.c
vfio_platform_base_device_init(char *name, int groupid) "%s belongs to group #%d"
New VFIO_SPAPR_TCE_v2_IOMMU type supports dynamic DMA window management. This adds ability to VFIO common code to dynamically allocate/remove DMA windows in the host kernel when new VFIO container is added/removed. This adds VFIO_IOMMU_SPAPR_TCE_CREATE ioctl to vfio_listener_region_add and adds just created IOMMU into the host IOMMU list; the opposite action is taken in vfio_listener_region_del. When creating a new window, this uses heuristic to decide on the TCE table levels number. This should cause no guest visible change in behavior. Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru> --- Changes: v17: * moved spapr window create/remove helpers to separate file * added hw_error() if vfio_host_win_del() failed v16: * used memory_region_iommu_get_page_sizes() in vfio_listener_region_add() * enforced no intersections between windows v14: * new to the series --- hw/vfio/common.c | 76 +++++++++++++++++++++++++++++++++++++------ hw/vfio/spapr.c | 70 +++++++++++++++++++++++++++++++++++++++ include/hw/vfio/vfio-common.h | 6 ++++ trace-events | 2 ++ 4 files changed, 144 insertions(+), 10 deletions(-)