[Qemu-devel] [PATCH v6 7/7] vfio: Add guest side IOMMU support

[Alexey Kardashevskiy]

From: David Gibson <address@hidden>

This patch uses the new IOMMU notifiers to allow VFIO pass through devices
to work with guest side IOMMUs, as long as the host-side VFIO iommu has
sufficient capability and granularity to match the guest side. This works
by tracking all map and unmap operations on the guest IOMMU using the
notifiers, and mirroring them into VFIO.

There are a number of FIXMEs, and the scheme involves rather more notifier
structures than I'd like, but it should make for a reasonable proof of
concept.

Signed-off-by: David Gibson <address@hidden>
Signed-off-by: Alexey Kardashevskiy <address@hidden>

---
Changes:
v6:
* changed comments about FIXMEs

v4:
* fixed list objects naming
* vfio_listener_region_add() reworked to call memory_region_ref() from one
place only, it is also easier to review the changes
* fixes boundary check not to fail on sections == 2^64 bytes,
the "vfio: Fix debug output for int128 values" patch is required;
this obsoletes the "[PATCH v3 0/3] vfio: fixes for better support
for 128 bit memory section sizes" patch proposal
---
 hw/misc/vfio.c | 139 ++++++++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 134 insertions(+), 5 deletions(-)

diff --git a/hw/misc/vfio.c b/hw/misc/vfio.c
index 3e3e000..eb728ad 100644
--- a/hw/misc/vfio.c
+++ b/hw/misc/vfio.c
@@ -160,10 +160,18 @@ typedef struct VFIOContainer {
         };
         void (*release)(struct VFIOContainer *);
     } iommu_data;
+    QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
     QLIST_HEAD(, VFIOGroup) group_list;
     QLIST_ENTRY(VFIOContainer) next;
 } VFIOContainer;
 
+typedef struct VFIOGuestIOMMU {
+    VFIOContainer *container;
+    MemoryRegion *iommu;
+    Notifier n;
+    QLIST_ENTRY(VFIOGuestIOMMU) giommu_next;
+} VFIOGuestIOMMU;
+
 /* Cache of MSI-X setup plus extra mmap and memory region for split BAR map */
 typedef struct VFIOMSIXInfo {
     uint8_t table_bar;
@@ -2239,7 +2247,8 @@ static int vfio_dma_map(VFIOContainer *container, hwaddr 
iova,
 
 static bool vfio_listener_skipped_section(MemoryRegionSection *section)
 {
-    return !memory_region_is_ram(section->mr) ||
+    return (!memory_region_is_ram(section->mr) &&
+            !memory_region_is_iommu(section->mr)) ||
            /*
             * Sizing an enabled 64-bit BAR can cause spurious mappings to
             * addresses in the upper part of the 64-bit address space.  These
@@ -2249,6 +2258,65 @@ static bool 
vfio_listener_skipped_section(MemoryRegionSection *section)
            section->offset_within_address_space & (1ULL << 63);
 }
 
+static void vfio_iommu_map_notify(Notifier *n, void *data)
+{
+    VFIOGuestIOMMU *giommu = container_of(n, VFIOGuestIOMMU, n);
+    VFIOContainer *container = giommu->container;
+    IOMMUTLBEntry *iotlb = data;
+    MemoryRegion *mr;
+    hwaddr xlat;
+    hwaddr len = iotlb->addr_mask + 1;
+    void *vaddr;
+    int ret;
+
+    DPRINTF("iommu map @ %"HWADDR_PRIx" - %"HWADDR_PRIx"\n",
+            iotlb->iova, iotlb->iova + iotlb->addr_mask);
+
+    /*
+     * The IOMMU TLB entry we have just covers translation through
+     * this IOMMU to its immediate target.  We need to translate
+     * it the rest of the way through to memory.
+     */
+    mr = address_space_translate(&address_space_memory,
+                                 iotlb->translated_addr,
+                                 &xlat, &len, iotlb->perm & IOMMU_WO);
+    if (!memory_region_is_ram(mr)) {
+        DPRINTF("iommu map to non memory area %"HWADDR_PRIx"\n",
+                xlat);
+        return;
+    }
+    /*
+     * Translation truncates length to the IOMMU page size,
+     * check that it did not truncate too much.
+     */
+    if (len & iotlb->addr_mask) {
+        DPRINTF("iommu has granularity incompatible with target AS\n");
+        return;
+    }
+
+    if (iotlb->perm != IOMMU_NONE) {
+        vaddr = memory_region_get_ram_ptr(mr) + xlat;
+
+        ret = vfio_dma_map(container, iotlb->iova,
+                           iotlb->addr_mask + 1, vaddr,
+                           !(iotlb->perm & IOMMU_WO) || mr->readonly);
+        if (ret) {
+            error_report("vfio_dma_map(%p, 0x%"HWADDR_PRIx", "
+                         "0x%"HWADDR_PRIx", %p) = %d (%m)",
+                         container, iotlb->iova,
+                         iotlb->addr_mask + 1, vaddr, ret);
+        }
+    } else {
+        ret = vfio_dma_unmap(container, iotlb->iova, iotlb->addr_mask + 1);
+        if (ret) {
+            error_report("vfio_dma_unmap(%p, 0x%"HWADDR_PRIx", "
+                         "0x%"HWADDR_PRIx") = %d (%m)",
+                         container, iotlb->iova,
+                         iotlb->addr_mask + 1, ret);
+        }
+    }
+}
+
 static void vfio_listener_region_add(MemoryListener *listener,
                                      MemoryRegionSection *section)
 {
@@ -2259,8 +2327,6 @@ static void vfio_listener_region_add(MemoryListener 
*listener,
     void *vaddr;
     int ret;
 
-    assert(!memory_region_is_iommu(section->mr));
-
     if (vfio_listener_skipped_section(section)) {
         DPRINTF("SKIPPING region_add %"HWADDR_PRIx" - %"PRIx64"\n",
                 section->offset_within_address_space,
@@ -2284,15 +2350,57 @@ static void vfio_listener_region_add(MemoryListener 
*listener,
         return;
     }
 
+    memory_region_ref(section->mr);
+
+    if (memory_region_is_iommu(section->mr)) {
+        VFIOGuestIOMMU *giommu;
+
+        DPRINTF("region_add [iommu] %"HWADDR_PRIx" - %"HWADDR_PRIx"\n",
+                iova, int128_get64(int128_sub(llend, int128_one())));
+        /*
+         * FIXME: We should do some checking to see if the
+         * capabilities of the host VFIO IOMMU are adequate to model
+         * the guest IOMMU
+         *
+         * 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).
+         */
+        /*
+         * This assumes that the guest IOMMU is empty of
+         * mappings at this point.
+         *
+         * One way of doing this is:
+         * 1. Avoid sharing IOMMUs between emulated devices or different
+         * IOMMU groups.
+         * 2. Implement VFIO_IOMMU_ENABLE in the host kernel to fail if
+         * there are some mappings in IOMMU.
+         *
+         * VFIO on SPAPR does that. Other IOMMU models may do that different,
+         * they must make sure there are no existing mappings or
+         * loop through existing mappings to map them into VFIO.
+         */
+        giommu = g_malloc0(sizeof(*giommu));
+        giommu->iommu = section->mr;
+        giommu->container = container;
+        giommu->n.notify = vfio_iommu_map_notify;
+        QLIST_INSERT_HEAD(&container->giommu_list, giommu, giommu_next);
+        memory_region_register_iommu_notifier(giommu->iommu, &giommu->n);
+
+        return;
+    }
+
+    /* Here we assume that memory_region_is_ram(section->mr)==true */
+
     end = int128_get64(llend);
     vaddr = memory_region_get_ram_ptr(section->mr) +
             section->offset_within_region +
             (iova - section->offset_within_address_space);
 
-    DPRINTF("region_add %"HWADDR_PRIx" - %"HWADDR_PRIx" [%p]\n",
+    DPRINTF("region_add [ram] %"HWADDR_PRIx" - %"HWADDR_PRIx" [%p]\n",
             iova, end - 1, vaddr);
 
-    memory_region_ref(section->mr);
     ret = vfio_dma_map(container, iova, end - iova, vaddr, section->readonly);
     if (ret) {
         error_report("vfio_dma_map(%p, 0x%"HWADDR_PRIx", "
@@ -2336,6 +2444,27 @@ static void vfio_listener_region_del(MemoryListener 
*listener,
         return;
     }
 
+    if (memory_region_is_iommu(section->mr)) {
+        VFIOGuestIOMMU *giommu;
+
+        QLIST_FOREACH(giommu, &container->giommu_list, giommu_next) {
+            if (giommu->iommu == section->mr) {
+                memory_region_unregister_iommu_notifier(&giommu->n);
+                QLIST_REMOVE(giommu, giommu_next);
+                g_free(giommu);
+                break;
+            }
+        }
+
+        /*
+         * FIXME: We assume the one big unmap below is adequate to
+         * remove any individual page mappings in the IOMMU which
+         * might have been copied into VFIO. This works for a page table
+         * based IOMMU where a big unmap flattens a large range of IO-PTEs.
+         * That may not be true for all IOMMU types.
+         */
+    }
+
     iova = TARGET_PAGE_ALIGN(section->offset_within_address_space);
     end = (section->offset_within_address_space + int128_get64(section->size)) 
&
           TARGET_PAGE_MASK;
-- 
1.9.rc0