[PATCH RFC 07/10] vfio/iommufd: Add HWPT_GET_DIRTY_IOVA support

[Joao Martins <joao.m.martins@oracle.com>]

ioctl(iommufd, IOMMU_HWPT_GET_DIRTY_IOVA, arg) is the UAPI
that fetches the bitmap that tells what was dirty in an IOVA
range.

A single bitmap is allocated and used across all the hw_pagetables
sharing an IOAS which is then used in log_sync() to set Qemu
global bitmaps.

There's no point of even attempting to fetch these bitmaps,
should the iommu tracker fail to start in a previous call
to HWPT_SET_DIRTY.

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
---
 hw/iommufd/iommufd.c         | 24 ++++++++++++++++++++
 hw/iommufd/trace-events      |  1 +
 hw/vfio/iommufd.c            | 44 ++++++++++++++++++++++++++++++++++++
 include/hw/iommufd/iommufd.h |  2 ++
 4 files changed, 71 insertions(+)

diff --git a/hw/iommufd/iommufd.c b/hw/iommufd/iommufd.c
index e5aff5deaf14..bc870b5e9b2f 100644
--- a/hw/iommufd/iommufd.c
+++ b/hw/iommufd/iommufd.c
@@ -219,6 +219,30 @@ int iommufd_set_dirty_tracking(int iommufd, uint32_t hwpt_id, bool start)
     return !ret ? 0 : -errno;
 }
 
+int iommufd_get_dirty_iova(int iommufd, uint32_t hwpt_id, uint64_t iova,
+                           ram_addr_t size, uint64_t page_size, uint64_t *data)
+{
+    int ret;
+    struct iommu_hwpt_get_dirty_iova get_dirty_iova = {
+        .size = sizeof(get_dirty_iova),
+        .hwpt_id = hwpt_id,
+        .bitmap = {
+            .iova = iova, .length = size,
+            .page_size = page_size, .data = (__u64 *)data,
+        },
+    };
+
+    ret = ioctl(iommufd, IOMMU_HWPT_GET_DIRTY_IOVA, &get_dirty_iova);
+    trace_iommufd_get_dirty_iova(iommufd, hwpt_id, iova, size, page_size, ret);
+    if (ret) {
+        error_report("IOMMU_HWPT_GET_DIRTY_IOVA (iova: 0x%"PRIx64
+                     " size: 0x%"PRIx64") failed: %s", iova,
+                     size, strerror(errno));
+    }
+
+    return !ret ? 0 : -errno;
+}
+
 static void iommufd_register_types(void)
 {
     qemu_mutex_init(&iommufd_lock);
diff --git a/hw/iommufd/trace-events b/hw/iommufd/trace-events
index d3c2b5a0ab95..9fe2cc60c6fe 100644
--- a/hw/iommufd/trace-events
+++ b/hw/iommufd/trace-events
@@ -10,3 +10,4 @@ iommufd_unmap_dma(int iommufd, uint32_t ioas, uint64_t iova, uint64_t size, int
 iommufd_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_copy_dma(int iommufd, uint32_t src_ioas, uint32_t dst_ioas, uint64_t iova, uint64_t size, bool readonly, int ret) " iommufd=%d src_ioas=%d dst_ioas=%d iova=0x%"PRIx64" size=0x%"PRIx64" readonly=%d (%d)"
 iommufd_set_dirty(int iommufd, uint32_t hwpt_id, bool start, int ret) " iommufd=%d hwpt=%d enable=%d (%d)"
+iommufd_get_dirty_iova(int iommufd, uint32_t hwpt_id, uint64_t iova, uint64_t size, uint64_t page_size, int ret) " iommufd=%d hwpt=%d iova=0x%"PRIx64" size=0x%"PRIx64" page_size=0x%"PRIx64" (%d)"
diff --git a/hw/vfio/iommufd.c b/hw/vfio/iommufd.c
index 8146407feedd..6c12239a40ab 100644
--- a/hw/vfio/iommufd.c
+++ b/hw/vfio/iommufd.c
@@ -33,6 +33,7 @@
 #include "hw/qdev-core.h"
 #include "sysemu/reset.h"
 #include "qemu/cutils.h"
+#include "exec/ram_addr.h"
 #include "migration/migration.h"
 
 static bool iommufd_check_extension(VFIOContainer *bcontainer,
@@ -102,6 +103,48 @@ static void iommufd_set_dirty_page_tracking(VFIOContainer *bcontainer,
     bcontainer->dirty_pages_supported = start;
 }
 
+static int iommufd_get_dirty_bitmap(VFIOContainer *bcontainer, uint64_t iova,
+                                    uint64_t size, ram_addr_t ram_addr)
+{
+    VFIOIOMMUFDContainer *container = container_of(bcontainer,
+                                                   VFIOIOMMUFDContainer, obj);
+    int ret;
+    VFIOIOASHwpt *hwpt;
+    unsigned long *data, page_size, bitmap_size, pages;
+
+    if (!bcontainer->dirty_pages_supported) {
+        return 0;
+    }
+
+    page_size = qemu_real_host_page_size;
+    pages = REAL_HOST_PAGE_ALIGN(size) / qemu_real_host_page_size;
+    bitmap_size = ROUND_UP(pages, sizeof(__u64) * BITS_PER_BYTE) /
+                                         BITS_PER_BYTE;
+    data = g_try_malloc0(bitmap_size);
+    if (!data) {
+        ret = -ENOMEM;
+        goto err_out;
+    }
+
+    QLIST_FOREACH(hwpt, &container->hwpt_list, next) {
+        ret = iommufd_get_dirty_iova(container->iommufd, hwpt->hwpt_id,
+                                     iova, size, page_size, data);
+        if (ret) {
+            goto err_out;
+        }
+    }
+
+    cpu_physical_memory_set_dirty_lebitmap(data, ram_addr, pages);
+
+    trace_vfio_get_dirty_bitmap(container->iommufd, iova, size, bitmap_size,
+                                ram_addr);
+
+err_out:
+    g_free(data);
+
+    return ret;
+}
+
 static int vfio_get_devicefd(const char *sysfs_path, Error **errp)
 {
     long int vfio_id = -1, ret = -ENOTTY;
@@ -611,6 +654,7 @@ static void vfio_iommufd_class_init(ObjectClass *klass,
     vccs->reset = vfio_iommufd_container_reset;
     vccs->devices_all_dirty_tracking = vfio_iommufd_devices_all_dirty_tracking;
     vccs->set_dirty_page_tracking = iommufd_set_dirty_page_tracking;
+    vccs->get_dirty_bitmap = iommufd_get_dirty_bitmap;
 }
 
 static const TypeInfo vfio_iommufd_info = {
diff --git a/include/hw/iommufd/iommufd.h b/include/hw/iommufd/iommufd.h
index 61fd83771099..9b467e57723b 100644
--- a/include/hw/iommufd/iommufd.h
+++ b/include/hw/iommufd/iommufd.h
@@ -34,5 +34,7 @@ int iommufd_map_dma(int iommufd, uint32_t ioas, hwaddr iova,
 int iommufd_copy_dma(int iommufd, uint32_t src_ioas, uint32_t dst_ioas,
                      hwaddr iova, ram_addr_t size, bool readonly);
 int iommufd_set_dirty_tracking(int iommufd, uint32_t hwpt_id, bool start);
+int iommufd_get_dirty_iova(int iommufd, uint32_t hwpt_id, uint64_t iova,
+                           ram_addr_t size, uint64_t page_size, uint64_t *data);
 bool iommufd_supported(void);
 #endif /* HW_IOMMUFD_IOMMUFD_H */
-- 
2.17.2