On 5/2/25 1:13 PM, Nicolin Chen wrote:
On Fri, May 02, 2025 at 11:27:03AM +0100, Shameer Kolothum wrote:
@@ -43,6 +43,7 @@
#include "hw/acpi/generic_event_device.h"
#include "hw/acpi/tpm.h"
#include "hw/acpi/hmat.h"
+#include "hw/arm/smmuv3.h"
#include "hw/pci/pcie_host.h"
#include "hw/pci/pci.h"
#include "hw/pci/pci_bus.h"
@@ -266,6 +267,75 @@ static int iort_idmap_compare(gconstpointer a,
gconstpointer b)
return idmap_a->input_base - idmap_b->input_base;
}
+struct SMMUv3Device {
+ int irq;
+ hwaddr base;
+ GArray *smmu_idmaps;
+ size_t offset;
+};
+typedef struct SMMUv3Device SMMUv3Device;
"SMMUv3Device" sounds too general, like coming from smmuv3.h :-/
Given this describes SMMUv3's IORT node, I still think we should
name it something like "IortSMMUv3Node" or so.
+1.. the more generic name had me thinking it was broader than IORT..
the IORT-related naming is an improvement.
+static int smmuv3_dev_idmap_compare(gconstpointer a, gconstpointer b)
+{
+ SMMUv3Device *sdev_a = (SMMUv3Device *)a;
+ SMMUv3Device *sdev_b = (SMMUv3Device *)b;
+ AcpiIortIdMapping *map_a = &g_array_index(sdev_a->smmu_idmaps,
+ AcpiIortIdMapping, 0);
+ AcpiIortIdMapping *map_b = &g_array_index(sdev_b->smmu_idmaps,
+ AcpiIortIdMapping, 0);
+ return map_a->input_base - map_b->input_base;
+}
+
+static void
+get_smmuv3_legacy_dev(VirtMachineState *vms, GArray * smmuv3_devices)
GArray *smmuv3_devices
Or maybe align with the non-legacy path, i.e. "sdev_blob"? Or the
other way around.
Otherwise, lgtm
Reviewed-by: Nicolin Chen <nicol...@nvidia.com>
