[Qemu-devel] [RFC PATCH 7/7] vfio/pci: Find and expose Intel IGD OpRegion

[Alex Williamson]

This is provided via a device specific region, look for it on Intel
VGA class devices, initialize it, and tie it to the config space
register at 0xFC.  Note that on bare metal it seems that 0xFC only
points to the memory reserved by the BIOS for the OpRegion, in the
model used here, programming the 0xFC register makes the host OpRegion
pages appear in the VM address space at that address.

Register 0x5C is the base of the stolen memory region (BDSM).
Emulating this register and setting it to zero prevents the driver
from trying to blindly use this address, which might be in-use RAM in
the VM.  This avoids DMAR faults, or potentially VM memory corruption,
but we likely need a better solution, this is just a hack.  It also
avoids framebuffer corruption.

Signed-off-by: Alex Williamson <address@hidden>
---
 hw/vfio/common.c              |    2 +
 hw/vfio/pci.c                 |   74 +++++++++++++++++++++++++++++++++++++++++
 hw/vfio/pci.h                 |    1 +
 include/hw/vfio/vfio-common.h |    2 +
 4 files changed, 78 insertions(+), 1 deletion(-)

diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 879a657..c201bee 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -493,7 +493,7 @@ static void vfio_listener_release(VFIOContainer *container)
     memory_listener_unregister(&container->listener);
 }
 
-static struct vfio_info_cap_header *
+struct vfio_info_cap_header *
 vfio_get_region_info_cap(struct vfio_region_info *info, uint16_t id)
 {
     struct vfio_info_cap_header *hdr;
diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
index 8e20781..cf1e0c5 100644
--- a/hw/vfio/pci.c
+++ b/hw/vfio/pci.c
@@ -34,6 +34,7 @@
 #include "trace.h"
 
 #define MSIX_CAP_LENGTH 12
+#define IGD_OPREGION 0xFC
 
 static void vfio_disable_interrupts(VFIOPCIDevice *vdev);
 static void vfio_mmap_set_enabled(VFIOPCIDevice *vdev, bool enabled);
@@ -1108,6 +1109,25 @@ void vfio_pci_write_config(PCIDevice *pdev,
         } else if (was_enabled && !is_enabled) {
             vfio_msix_disable(vdev);
         }
+    } else if (vdev->igd_opregion &&
+               ranges_overlap(addr, len, IGD_OPREGION, 4)) {
+        uint32_t orig, cur;
+
+        orig = pci_get_long(pdev->config + IGD_OPREGION);
+        pci_default_write_config(pdev, addr, val, len);
+        cur = pci_get_long(pdev->config + IGD_OPREGION);
+
+        if (cur != orig) {
+            if (orig) {
+                memory_region_del_subregion(get_system_memory(),
+                                            vdev->igd_opregion->mem);
+            }
+
+            if (cur) {
+                memory_region_add_subregion(get_system_memory(),
+                                            cur, vdev->igd_opregion->mem);
+            }
+        }
     } else {
         /* Write everything to QEMU to keep emulated bits correct */
         pci_default_write_config(pdev, addr, val, len);
@@ -1459,6 +1479,10 @@ static void vfio_bars_exit(VFIOPCIDevice *vdev)
         pci_unregister_vga(&vdev->pdev);
         vfio_vga_quirk_exit(vdev);
     }
+
+    if (vdev->igd_opregion) {
+        vfio_region_exit(vdev->igd_opregion);
+    }
 }
 
 static void vfio_bars_finalize(VFIOPCIDevice *vdev)
@@ -1477,6 +1501,11 @@ static void vfio_bars_finalize(VFIOPCIDevice *vdev)
         }
         g_free(vdev->vga);
     }
+
+    if (vdev->igd_opregion) {
+        vfio_region_finalize(vdev->igd_opregion);
+        g_free(vdev->igd_opregion);
+    }
 }
 
 /*
@@ -2123,6 +2152,45 @@ static int vfio_populate_device(VFIOPCIDevice *vdev)
         QLIST_INIT(&vdev->vga->region[QEMU_PCI_VGA_IO_HI].quirks);
     }
 
+    if (vbasedev->num_regions > VFIO_PCI_NUM_REGIONS) {
+        for (i = VFIO_PCI_NUM_REGIONS; i < vbasedev->num_regions; i++) {
+            struct vfio_info_cap_header *hdr;
+            struct vfio_region_info_cap_type *type;
+
+            ret = vfio_get_region_info(vbasedev, i, &reg_info);
+            if (ret) {
+                continue;
+            }
+
+            hdr = vfio_get_region_info_cap(reg_info, 
VFIO_REGION_INFO_CAP_TYPE);
+            if (!hdr) {
+                g_free(reg_info);
+                continue;
+            }
+
+            type = container_of(hdr, struct vfio_region_info_cap_type, header);
+            if (type->type ==
+                (VFIO_REGION_TYPE_PCI_VENDOR_TYPE | PCI_VENDOR_ID_INTEL) &&
+                type->subtype == VFIO_REGION_SUBTYPE_INTEL_IGD_OPREGION) {
+                char *name;
+
+                vdev->igd_opregion = g_new0(VFIORegion, 1);
+                name = g_strdup_printf("%s IGD OpRegion", vbasedev->name);
+
+                ret = vfio_region_setup(OBJECT(vdev), vbasedev,
+                                        vdev->igd_opregion, i, name);
+                g_free(name);
+
+                if (ret) {
+                    error_report("vfio: Error setting up IGD OpRegion\n");
+                    goto error;
+                }
+            }
+
+            g_free(reg_info);
+        }
+    }
+
     irq_info.index = VFIO_PCI_ERR_IRQ_INDEX;
 
     ret = ioctl(vdev->vbasedev.fd, VFIO_DEVICE_GET_IRQ_INFO, &irq_info);
@@ -2525,6 +2593,12 @@ static int vfio_initfn(PCIDevice *pdev)
                vdev->msi_cap_size);
     }
 
+    if (vdev->igd_opregion) {
+        pci_set_long(vdev->pdev.config + IGD_OPREGION, 0);
+        pci_set_long(vdev->pdev.wmask + IGD_OPREGION, ~0);
+        vfio_add_emulated_long(vdev, 0x5C, 0, ~0);
+    }
+
     if (vfio_pci_read_config(&vdev->pdev, PCI_INTERRUPT_PIN, 1)) {
         vdev->intx.mmap_timer = timer_new_ms(QEMU_CLOCK_VIRTUAL,
                                                   vfio_intx_mmap_enable, vdev);
diff --git a/hw/vfio/pci.h b/hw/vfio/pci.h
index b8a7189..c2d2cdf 100644
--- a/hw/vfio/pci.h
+++ b/hw/vfio/pci.h
@@ -115,6 +115,7 @@ typedef struct VFIOPCIDevice {
     int interrupt; /* Current interrupt type */
     VFIOBAR bars[PCI_NUM_REGIONS - 1]; /* No ROM */
     VFIOVGA *vga; /* 0xa0000, 0x3b0, 0x3c0 */
+    VFIORegion *igd_opregion;
     PCIHostDeviceAddress host;
     EventNotifier err_notifier;
     EventNotifier req_notifier;
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 594905a..45ba6a3 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -146,6 +146,8 @@ int vfio_get_device(VFIOGroup *group, const char *name,
                     VFIODevice *vbasedev);
 int vfio_get_region_info(VFIODevice *vbasedev, int index,
                          struct vfio_region_info **info);
+struct vfio_info_cap_header *
+    vfio_get_region_info_cap(struct vfio_region_info *info, uint16_t id);
 
 extern const MemoryRegionOps vfio_region_ops;
 extern QLIST_HEAD(vfio_group_head, VFIOGroup) vfio_group_list;