vfio/common: Move vfio_host_win_add/del into spapr.c

Only spapr supports a customed host window list, other vfio driver
assume 64bit host window. So remove the check in listener callback
and move vfio_host_win_add/del into spapr.c and make it static.

With the check removed, we still need to do the same check for
VFIO_SPAPR_TCE_IOMMU which allows a single host window range
[dma32_window_start, dma32_window_size). Move vfio_find_hostwin
into spapr.c and do same check in vfio_container_add_section_window
instead.

When mapping a ram device section, if it's unaligned with
hostwin->iova_pgsizes, this mapping is bypassed. With hostwin
moved into spapr, we changed to check container->pgsizes.

Suggested-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
This commit is contained in:
Zhenzhong Duan 2023-11-02 15:12:26 +08:00 committed by Cédric Le Goater
parent a17879f0e2
commit a2347c60a8
4 changed files with 85 additions and 89 deletions

View file

@ -207,11 +207,6 @@ typedef struct {
hwaddr pages;
} VFIOBitmap;
void vfio_host_win_add(VFIOContainer *container,
hwaddr min_iova, hwaddr max_iova,
uint64_t iova_pgsizes);
int vfio_host_win_del(VFIOContainer *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);