Skip to contentSkip to conversation

hw/sysbus/platform-bus: Introduce TYPE_DYNAMIC_SYS_BUS_DEVICE

PATCH In review Philippe Mathieu-Daudé 9 participants 41 messages Jan 25, 2025 – Feb 10, 2025

lore
Share & tools
lore
#0 lore raw

cover hw/sysbus/platform-bus: Introduce TYPE_DYNAMIC_SYS_BUS_DEVICE

Some SysBus devices can optionally be dynamically plugged onto the sysbus-platform-bus (then virtual guests are aware of mmio mapping and IRQs via device tree / ACPI rules).

This series makes these devices explicit by having them implement the DYNAMIC_SYS_BUS_DEVICE class, which only sets 'user_creatable' flag to %true but makes the codebase a bit clearer (IMHO, at least now we can grep for DYNAMIC_SYS_BUS_DEVICE).

Philippe Mathieu-Daudé (9):
hw/sysbus: Use sizeof(BusState) in main_system_bus_create() hw/sysbus: Declare QOM types using DEFINE_TYPES() macro hw/sysbus: Introduce TYPE_DYNAMIC_SYS_BUS_DEVICE
hw/vfio: Have VFIO_PLATFORM devices inherit from
DYNAMIC_SYS_BUS_DEVICE
hw/display: Have RAMFB device inherit from DYNAMIC_SYS_BUS_DEVICE hw/i386: Have X86_IOMMU devices inherit from DYNAMIC_SYS_BUS_DEVICE hw/net: Have eTSEC device inherit from DYNAMIC_SYS_BUS_DEVICE hw/tpm: Have TPM TIS sysbus device inherit from DYNAMIC_SYS_BUS_DEVICE hw/xen: Have legacy Xen backend inherit from DYNAMIC_SYS_BUS_DEVICE

 include/hw/sysbus.h           |  2 ++
 include/hw/xen/xen_pvdev.h    |  3 +-
 hw/core/sysbus.c              | 53 ++++++++++++++++++++---------------
 hw/display/ramfb-standalone.c |  3 +-
 hw/i386/amd_iommu.c           |  2 --
 hw/i386/intel_iommu.c         |  2 --
 hw/i386/x86-iommu.c           |  2 +-
 hw/net/fsl_etsec/etsec.c      |  4 +--
 hw/tpm/tpm_tis_sysbus.c       |  3 +-
 hw/vfio/amd-xgbe.c            |  2 --
 hw/vfio/calxeda-xgmac.c       |  2 --
 hw/vfio/platform.c            |  4 +--
 hw/xen/xen-legacy-backend.c   |  7 ++---
 13 files changed, 42 insertions(+), 47 deletions(-)
Signature · 1 line
2.47.1
#1 lore raw

PATCH 1/9 hw/sysbus: Use sizeof(BusState) in main_system_bus_create()

Rather than using the obscure system_bus_info.instance_size, directly use sizeof(BusState).

---
 hw/core/sysbus.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
hw/core/sysbus.c +2 -2
index 9355849ff0a..f713bbfe04f 100644
@@ -323,8 +323,8 @@ static void main_system_bus_create(void)
      * assign main_system_bus before qbus_init()
      * in order to make "if (bus != sysbus_get_default())" work
      */
-    main_system_bus = g_malloc0(system_bus_info.instance_size);
-    qbus_init(main_system_bus, system_bus_info.instance_size,
+    main_system_bus = g_new0(BusState, 1);
+    qbus_init(main_system_bus, sizeof(BusState),
               TYPE_SYSTEM_BUS, NULL, "main-system-bus");
     OBJECT(main_system_bus)->free = g_free;
 }
Signature · 1 line
2.47.1
Reviewed-by on v1 1/9
#3 lore raw

Re: [PATCH 1/9] hw/sysbus: Use sizeof(BusState) in main_system_bus_create()

In reply to Philippe Mathieu-Daudé
Quoted 29 lines · Philippe Mathieu-Daudé

Caution: External email. Do not open attachments or click links, unless this email comes from a known sender and you know the content is safe.

Rather than using the obscure system_bus_info.instance_size, directly use sizeof(BusState).

---
  hw/core/sysbus.c | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)
hw/core/sysbus.c +2 -3
index 9355849ff0a..f713bbfe04f 100644
@@ -323,8 +323,8 @@ static void main_system_bus_create(void)
       * assign main_system_bus before qbus_init()
       * in order to make "if (bus != sysbus_get_default())" work
       */
-    main_system_bus = g_malloc0(system_bus_info.instance_size);
-    qbus_init(main_system_bus, system_bus_info.instance_size,
+    main_system_bus = g_new0(BusState, 1);
+    qbus_init(main_system_bus, sizeof(BusState),
                TYPE_SYSTEM_BUS, NULL, "main-system-bus");
      OBJECT(main_system_bus)->free = g_free;
  }
--

2.47.1

#4 lore raw

PATCH 2/9 hw/sysbus: Declare QOM types using DEFINE_TYPES() macro

In reply to Philippe Mathieu-Daudé

When multiple QOM types are registered in the same file, it is simpler to use the the DEFINE_TYPES() macro. In particular because type array declared with such macro are easier to review.

---
 hw/core/sysbus.c | 39 +++++++++++++++++----------------------
 1 file changed, 17 insertions(+), 22 deletions(-)
hw/core/sysbus.c +17 -22
index f713bbfe04f..306f98406c0 100644
@@ -80,13 +80,6 @@ static void system_bus_class_init(ObjectClass *klass, void *data)
     k->get_fw_dev_path = sysbus_get_fw_dev_path;
 }
 
-static const TypeInfo system_bus_info = {
-    .name = TYPE_SYSTEM_BUS,
-    .parent = TYPE_BUS,
-    .instance_size = sizeof(BusState),
-    .class_init = system_bus_class_init,
-};
-
 /* Check whether an IRQ source exists */
 bool sysbus_has_irq(SysBusDevice *dev, int n)
 {
@@ -306,15 +299,6 @@ static void sysbus_device_class_init(ObjectClass *klass, void *data)
     k->user_creatable = false;
 }
 
-static const TypeInfo sysbus_device_type_info = {
-    .name = TYPE_SYS_BUS_DEVICE,
-    .parent = TYPE_DEVICE,
-    .instance_size = sizeof(SysBusDevice),
-    .abstract = true,
-    .class_size = sizeof(SysBusDeviceClass),
-    .class_init = sysbus_device_class_init,
-};
-
 static BusState *main_system_bus;
 
 static void main_system_bus_create(void)
@@ -337,10 +321,21 @@ BusState *sysbus_get_default(void)
     return main_system_bus;
 }
 
-static void sysbus_register_types(void)
-{
-    type_register_static(&system_bus_info);
-    type_register_static(&sysbus_device_type_info);
-}
+static const TypeInfo sysbus_types[] = {
+    {
+        .name           = TYPE_SYSTEM_BUS,
+        .parent         = TYPE_BUS,
+        .instance_size  = sizeof(BusState),
+        .class_init     = system_bus_class_init,
+    },
+    {
+        .name           = TYPE_SYS_BUS_DEVICE,
+        .parent         = TYPE_DEVICE,
+        .instance_size  = sizeof(SysBusDevice),
+        .abstract       = true,
+        .class_size     = sizeof(SysBusDeviceClass),
+        .class_init     = sysbus_device_class_init,
+    },
+};
 
-type_init(sysbus_register_types)
+DEFINE_TYPES(sysbus_types)
Signature · 1 line
2.47.1
Reviewed-by on v1 2/9
#5 lore raw

Re: [PATCH 2/9] hw/sysbus: Declare QOM types using DEFINE_TYPES() macro

Quoted 9 lines · Philippe Mathieu-Daudé

When multiple QOM types are registered in the same file, it is simpler to use the the DEFINE_TYPES() macro. In particular because type array declared with such macro are easier to review.

---
  hw/core/sysbus.c | 39 +++++++++++++++++----------------------
  1 file changed, 17 insertions(+), 22 deletions(-)

r~

Reviewed-by on v1 2/9
#6 lore raw

Re: [PATCH 2/9] hw/sysbus: Declare QOM types using DEFINE_TYPES() macro

In reply to Philippe Mathieu-Daudé
Quoted 78 lines · Philippe Mathieu-Daudé

Caution: External email. Do not open attachments or click links, unless this email comes from a known sender and you know the content is safe.

When multiple QOM types are registered in the same file, it is simpler to use the the DEFINE_TYPES() macro. In particular because type array declared with such macro are easier to review.

---
  hw/core/sysbus.c | 39 +++++++++++++++++----------------------
  1 file changed, 17 insertions(+), 22 deletions(-)
hw/core/sysbus.c +17 -23
index f713bbfe04f..306f98406c0 100644
@@ -80,13 +80,6 @@ static void system_bus_class_init(ObjectClass *klass, void *data)
      k->get_fw_dev_path = sysbus_get_fw_dev_path;
  }
 
-static const TypeInfo system_bus_info = {
-    .name = TYPE_SYSTEM_BUS,
-    .parent = TYPE_BUS,
-    .instance_size = sizeof(BusState),
-    .class_init = system_bus_class_init,
-};
-
  /* Check whether an IRQ source exists */
  bool sysbus_has_irq(SysBusDevice *dev, int n)
  {
@@ -306,15 +299,6 @@ static void sysbus_device_class_init(ObjectClass *klass, void *data)
      k->user_creatable = false;
  }
 
-static const TypeInfo sysbus_device_type_info = {
-    .name = TYPE_SYS_BUS_DEVICE,
-    .parent = TYPE_DEVICE,
-    .instance_size = sizeof(SysBusDevice),
-    .abstract = true,
-    .class_size = sizeof(SysBusDeviceClass),
-    .class_init = sysbus_device_class_init,
-};
-
  static BusState *main_system_bus;
 
  static void main_system_bus_create(void)
@@ -337,10 +321,21 @@ BusState *sysbus_get_default(void)
      return main_system_bus;
  }
 
-static void sysbus_register_types(void)
-{
-    type_register_static(&system_bus_info);
-    type_register_static(&sysbus_device_type_info);
-}
+static const TypeInfo sysbus_types[] = {
+    {
+        .name           = TYPE_SYSTEM_BUS,
+        .parent         = TYPE_BUS,
+        .instance_size  = sizeof(BusState),
+        .class_init     = system_bus_class_init,
+    },
+    {
+        .name           = TYPE_SYS_BUS_DEVICE,
+        .parent         = TYPE_DEVICE,
+        .instance_size  = sizeof(SysBusDevice),
+        .abstract       = true,
+        .class_size     = sizeof(SysBusDeviceClass),
+        .class_init     = sysbus_device_class_init,
+    },
+};
 
-type_init(sysbus_register_types)
+DEFINE_TYPES(sysbus_types)
--

2.47.1

Reviewed-by on v1 2/9
#7 lore raw

Re: [PATCH 2/9] hw/sysbus: Declare QOM types using DEFINE_TYPES() macro

In reply to Philippe Mathieu-Daudé

Am 25. Januar 2025 18:13:36 UTC schrieb "Philippe Mathieu-Daudé" <[email protected]>:

Quoted 62 lines

When multiple QOM types are registered in the same file, it is simpler to use the the DEFINE_TYPES() macro. In particular because type array declared with such macro are easier to review.

---
hw/core/sysbus.c | 39 +++++++++++++++++----------------------

1 file changed, 17 insertions(+), 22 deletions(-)
hw/core/sysbus.c +9 -21
index f713bbfe04f..306f98406c0 100644
@@ -80,13 +80,6 @@ static void system_bus_class_init(ObjectClass *klass, void *data)
    k->get_fw_dev_path = sysbus_get_fw_dev_path;
 }
 
-static const TypeInfo system_bus_info = {
-    .name = TYPE_SYSTEM_BUS,
-    .parent = TYPE_BUS,
-    .instance_size = sizeof(BusState),
-    .class_init = system_bus_class_init,
-};
-
 /* Check whether an IRQ source exists */
 bool sysbus_has_irq(SysBusDevice *dev, int n)
 {
@@ -306,15 +299,6 @@ static void sysbus_device_class_init(ObjectClass *klass, void *data)
    k->user_creatable = false;
 }
 
-static const TypeInfo sysbus_device_type_info = {
-    .name = TYPE_SYS_BUS_DEVICE,
-    .parent = TYPE_DEVICE,
-    .instance_size = sizeof(SysBusDevice),
-    .abstract = true,
-    .class_size = sizeof(SysBusDeviceClass),
-    .class_init = sysbus_device_class_init,
-};
-
 static BusState *main_system_bus;
 
 static void main_system_bus_create(void)
@@ -337,10 +321,21 @@ BusState *sysbus_get_default(void)
    return main_system_bus;
 }
 
-static void sysbus_register_types(void)
-{
-    type_register_static(&system_bus_info);
-    type_register_static(&sysbus_device_type_info);
-}
+static const TypeInfo sysbus_types[] = {
+    {
+        .name           = TYPE_SYSTEM_BUS,
+        .parent         = TYPE_BUS,
+        .instance_size  = sizeof(BusState),
+        .class_init     = system_bus_class_init,
+    },
+    {
+        .name           = TYPE_SYS_BUS_DEVICE,

hw/core/sysbus.c BusState *sysbus_get_default(void) · line 341

Quoted 10 lines
hw/core/sysbus.c +8 -1
+        .parent         = TYPE_DEVICE,
+        .instance_size  = sizeof(SysBusDevice),
+        .abstract       = true,
+        .class_size     = sizeof(SysBusDeviceClass),
+        .class_init     = sysbus_device_class_init,
+    },
+};
 
-type_init(sysbus_register_types)
+DEFINE_TYPES(sysbus_types)

Can now omit the "qom/module.h" include. With that changed:

Reviewed-by Bernhard Beschow
#8 lore raw

PATCH 3/9 hw/sysbus: Introduce TYPE_DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé

Some TYPE_SYS_BUS_DEVICEs can be optionally dynamically plugged on the TYPE_PLATFORM_BUS_DEVICE.
Rather than sometimes noting that with comment around the 'user_creatable = true' line in each DeviceRealize handler, introduce an abstract TYPE_DYNAMIC_SYS_BUS_DEVICE class.

---
 include/hw/sysbus.h |  2 ++
 hw/core/sysbus.c    | 14 ++++++++++++++
 2 files changed, 16 insertions(+)
include/hw/sysbus.h +2 -0
index c9b1e0e90e3..81bbda10d37 100644
@@ -19,6 +19,8 @@ DECLARE_INSTANCE_CHECKER(BusState, SYSTEM_BUS,
 OBJECT_DECLARE_TYPE(SysBusDevice, SysBusDeviceClass,
                     SYS_BUS_DEVICE)
 
+#define TYPE_DYNAMIC_SYS_BUS_DEVICE "dynamic-sysbus-device"
+
 /**
  * SysBusDeviceClass:
  *
hw/core/sysbus.c +14 -0
index 306f98406c0..e8d03fd28d9 100644
@@ -321,6 +321,14 @@ BusState *sysbus_get_default(void)
     return main_system_bus;
 }
 
+static void dynamic_sysbus_device_class_init(ObjectClass *klass, void *data)
+{
+    DeviceClass *k = DEVICE_CLASS(klass);
+
+    k->user_creatable = true;
+    k->hotpluggable = false;
+}
+
 static const TypeInfo sysbus_types[] = {
     {
         .name           = TYPE_SYSTEM_BUS,
@@ -336,6 +344,12 @@ static const TypeInfo sysbus_types[] = {
         .class_size     = sizeof(SysBusDeviceClass),
         .class_init     = sysbus_device_class_init,
     },
+    {
+        .name           = TYPE_DYNAMIC_SYS_BUS_DEVICE,
+        .parent         = TYPE_SYS_BUS_DEVICE,
+        .class_init     = dynamic_sysbus_device_class_init,
+        .abstract       = true,
+    }
 };
 
 DEFINE_TYPES(sysbus_types)
Signature · 1 line
2.47.1
Reviewed-by on v1 3/9
#9 lore raw

Re: [PATCH 3/9] hw/sysbus: Introduce TYPE_DYNAMIC_SYS_BUS_DEVICE

Quoted 7 lines · Philippe Mathieu-Daudé

Some TYPE_SYS_BUS_DEVICEs can be optionally dynamically plugged on the TYPE_PLATFORM_BUS_DEVICE.
Rather than sometimes noting that with comment around the 'user_creatable = true' line in each DeviceRealize handler, introduce an abstract TYPE_DYNAMIC_SYS_BUS_DEVICE class.

Quoted 5 lines
---
  include/hw/sysbus.h |  2 ++
  hw/core/sysbus.c    | 14 ++++++++++++++
  2 files changed, 16 insertions(+)

r~

Reviewed-by on v1 3/9
#10 lore raw

Re: [PATCH 3/9] hw/sysbus: Introduce TYPE_DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé
Quoted 64 lines · Philippe Mathieu-Daudé

Caution: External email. Do not open attachments or click links, unless this email comes from a known sender and you know the content is safe.

Some TYPE_SYS_BUS_DEVICEs can be optionally dynamically plugged on the TYPE_PLATFORM_BUS_DEVICE.
Rather than sometimes noting that with comment around the 'user_creatable = true' line in each DeviceRealize handler, introduce an abstract TYPE_DYNAMIC_SYS_BUS_DEVICE class.

---
  include/hw/sysbus.h |  2 ++
  hw/core/sysbus.c    | 14 ++++++++++++++
  2 files changed, 16 insertions(+)
include/hw/sysbus.h +2 -0
index c9b1e0e90e3..81bbda10d37 100644
@@ -19,6 +19,8 @@ DECLARE_INSTANCE_CHECKER(BusState, SYSTEM_BUS,
  OBJECT_DECLARE_TYPE(SysBusDevice, SysBusDeviceClass,
                      SYS_BUS_DEVICE)
 
+#define TYPE_DYNAMIC_SYS_BUS_DEVICE "dynamic-sysbus-device"
+
  /**
   * SysBusDeviceClass:
   *
hw/core/sysbus.c +14 -1
index 306f98406c0..e8d03fd28d9 100644
@@ -321,6 +321,14 @@ BusState *sysbus_get_default(void)
      return main_system_bus;
  }
 
+static void dynamic_sysbus_device_class_init(ObjectClass *klass, void *data)
+{
+    DeviceClass *k = DEVICE_CLASS(klass);
+
+    k->user_creatable = true;
+    k->hotpluggable = false;
+}
+
  static const TypeInfo sysbus_types[] = {
      {
          .name           = TYPE_SYSTEM_BUS,
@@ -336,6 +344,12 @@ static const TypeInfo sysbus_types[] = {
          .class_size     = sizeof(SysBusDeviceClass),
          .class_init     = sysbus_device_class_init,
      },
+    {
+        .name           = TYPE_DYNAMIC_SYS_BUS_DEVICE,
+        .parent         = TYPE_SYS_BUS_DEVICE,
+        .class_init     = dynamic_sysbus_device_class_init,
+        .abstract       = true,
+    }
  };
 
  DEFINE_TYPES(sysbus_types)
--

2.47.1

#11 lore raw

PATCH 4/9 hw/vfio: Have VFIO_PLATFORM devices inherit from DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé

Do not explain why VFIO_PLATFORM devices are user_creatable, have them inherit TYPE_DYNAMIC_SYS_BUS_DEVICE, to explicit they can optionally be plugged on TYPE_PLATFORM_BUS_DEVICE.

---
 hw/vfio/amd-xgbe.c      | 2 --
 hw/vfio/calxeda-xgmac.c | 2 --
 hw/vfio/platform.c      | 4 +---
 3 files changed, 1 insertion(+), 7 deletions(-)
hw/vfio/amd-xgbe.c +0 -2
index 96bd608b8dd..aaa96903db0 100644
@@ -41,8 +41,6 @@ static void vfio_amd_xgbe_class_init(ObjectClass *klass, void *data)
                                     &vcxc->parent_realize);
     dc->desc = "VFIO AMD XGBE";
     dc->vmsd = &vfio_platform_amd_xgbe_vmstate;
-    /* Supported by TYPE_VIRT_MACHINE */
-    dc->user_creatable = true;
 }
 
 static const TypeInfo vfio_amd_xgbe_dev_info = {
hw/vfio/calxeda-xgmac.c +0 -2
index 87c382e7361..b016d42b496 100644
@@ -41,8 +41,6 @@ static void vfio_calxeda_xgmac_class_init(ObjectClass *klass, void *data)
                                     &vcxc->parent_realize);
     dc->desc = "VFIO Calxeda XGMAC";
     dc->vmsd = &vfio_platform_calxeda_xgmac_vmstate;
-    /* Supported by TYPE_VIRT_MACHINE */
-    dc->user_creatable = true;
 }
 
 static const TypeInfo vfio_calxeda_xgmac_dev_info = {
hw/vfio/platform.c +1 -3
index 1070a2113a1..f491f4dc954 100644
@@ -672,13 +672,11 @@ static void vfio_platform_class_init(ObjectClass *klass, void *data)
     dc->desc = "VFIO-based platform device assignment";
     sbc->connect_irq_notifier = vfio_start_irqfd_injection;
     set_bit(DEVICE_CATEGORY_MISC, dc->categories);
-    /* Supported by TYPE_VIRT_MACHINE */
-    dc->user_creatable = true;
 }
 
 static const TypeInfo vfio_platform_dev_info = {
     .name = TYPE_VFIO_PLATFORM,
-    .parent = TYPE_SYS_BUS_DEVICE,
+    .parent = TYPE_DYNAMIC_SYS_BUS_DEVICE,
     .instance_size = sizeof(VFIOPlatformDevice),
     .instance_init = vfio_platform_instance_init,
     .class_init = vfio_platform_class_init,
Signature · 1 line
2.47.1
Reviewed-by on v1 4/9
#12 lore raw

Re: [PATCH 4/9] hw/vfio: Have VFIO_PLATFORM devices inherit from DYNAMIC_SYS_BUS_DEVICE

Quoted 3 lines · Philippe Mathieu-Daudé

Do not explain why VFIO_PLATFORM devices are user_creatable, have them inherit TYPE_DYNAMIC_SYS_BUS_DEVICE, to explicit they can optionally be plugged on TYPE_PLATFORM_BUS_DEVICE.

to make it explicit that they can be

r~

#13 lore raw

PATCH 5/9 hw/display: Have RAMFB device inherit from DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé

Because the RAM FB device can be optionally plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE.

---
 hw/display/ramfb-standalone.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
hw/display/ramfb-standalone.c +1 -2
index 6c35028965d..1be106b57f2 100644
@@ -72,13 +72,12 @@ static void ramfb_class_initfn(ObjectClass *klass, void *data)
     dc->vmsd = &ramfb_dev_vmstate;
     dc->realize = ramfb_realizefn;
     dc->desc = "ram framebuffer standalone device";
-    dc->user_creatable = true;
     device_class_set_props(dc, ramfb_properties);
 }
 
 static const TypeInfo ramfb_info = {
     .name          = TYPE_RAMFB_DEVICE,
-    .parent        = TYPE_SYS_BUS_DEVICE,
+    .parent        = TYPE_DYNAMIC_SYS_BUS_DEVICE,
     .instance_size = sizeof(RAMFBStandaloneState),
     .class_init    = ramfb_class_initfn,
 };
Signature · 1 line
2.47.1
Reviewed-by on v1 5/9
#14 lore raw

Re: [PATCH 5/9] hw/display: Have RAMFB device inherit from DYNAMIC_SYS_BUS_DEVICE

Quoted 7 lines · Philippe Mathieu-Daudé

Because the RAM FB device can be optionally plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE.

---
  hw/display/ramfb-standalone.c | 3 +--
  1 file changed, 1 insertion(+), 2 deletions(-)

r~

Reviewed-by on v1 5/9
#15 lore raw

Re: [PATCH 5/9] hw/display: Have RAMFB device inherit from DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé
Quoted 33 lines · Philippe Mathieu-Daudé

Caution: External email. Do not open attachments or click links, unless this email comes from a known sender and you know the content is safe.

Because the RAM FB device can be optionally plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE.

---
  hw/display/ramfb-standalone.c | 3 +--
  1 file changed, 1 insertion(+), 2 deletions(-)
hw/display/ramfb-standalone.c +1 -3
index 6c35028965d..1be106b57f2 100644
@@ -72,13 +72,12 @@ static void ramfb_class_initfn(ObjectClass *klass, void *data)
      dc->vmsd = &ramfb_dev_vmstate;
      dc->realize = ramfb_realizefn;
      dc->desc = "ram framebuffer standalone device";
-    dc->user_creatable = true;
      device_class_set_props(dc, ramfb_properties);
  }
 
  static const TypeInfo ramfb_info = {
      .name          = TYPE_RAMFB_DEVICE,
-    .parent        = TYPE_SYS_BUS_DEVICE,
+    .parent        = TYPE_DYNAMIC_SYS_BUS_DEVICE,
      .instance_size = sizeof(RAMFBStandaloneState),
      .class_init    = ramfb_class_initfn,
  };
--

2.47.1

#16 lore raw

PATCH 6/9 hw/i386: Have X86_IOMMU devices inherit from DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé

Do not explain why _X86_IOMMU devices are user_creatable, have them inherit TYPE_DYNAMIC_SYS_BUS_DEVICE, to explicit they can optionally be plugged on TYPE_PLATFORM_BUS_DEVICE.

---
 hw/i386/amd_iommu.c   | 2 --
 hw/i386/intel_iommu.c | 2 --
 hw/i386/x86-iommu.c   | 2 +-
 3 files changed, 1 insertion(+), 5 deletions(-)
hw/i386/amd_iommu.c +0 -2
index 6b13ce894b1..e8e084c7cf8 100644
@@ -1687,8 +1687,6 @@ static void amdvi_sysbus_class_init(ObjectClass *klass, void *data)
     dc->hotpluggable = false;
     dc_class->realize = amdvi_sysbus_realize;
     dc_class->int_remap = amdvi_int_remap;
-    /* Supported by the pc-q35-* machine types */
-    dc->user_creatable = true;
     set_bit(DEVICE_CATEGORY_MISC, dc->categories);
     dc->desc = "AMD IOMMU (AMD-Vi) DMA Remapping device";
     device_class_set_props(dc, amdvi_properties);
hw/i386/intel_iommu.c +0 -2
index f366c223d0e..7fde0603bfe 100644
@@ -4871,8 +4871,6 @@ static void vtd_class_init(ObjectClass *klass, void *data)
     dc->hotpluggable = false;
     x86_class->realize = vtd_realize;
     x86_class->int_remap = vtd_int_remap;
-    /* Supported by the pc-q35-* machine types */
-    dc->user_creatable = true;
     set_bit(DEVICE_CATEGORY_MISC, dc->categories);
     dc->desc = "Intel IOMMU (VT-d) DMA Remapping device";
 }
hw/i386/x86-iommu.c +1 -1
index fed34b2fcfa..5cdd165af0d 100644
@@ -146,7 +146,7 @@ bool x86_iommu_ir_supported(X86IOMMUState *s)
 
 static const TypeInfo x86_iommu_info = {
     .name          = TYPE_X86_IOMMU_DEVICE,
-    .parent        = TYPE_SYS_BUS_DEVICE,
+    .parent        = TYPE_DYNAMIC_SYS_BUS_DEVICE,
     .instance_size = sizeof(X86IOMMUState),
     .class_init    = x86_iommu_class_init,
     .class_size    = sizeof(X86IOMMUClass),
Signature · 1 line
2.47.1
Reviewed-by on v1 6/9
#17 lore raw

Re: [PATCH 6/9] hw/i386: Have X86_IOMMU devices inherit from DYNAMIC_SYS_BUS_DEVICE

Quoted 10 lines · Philippe Mathieu-Daudé

Do not explain why _X86_IOMMU devices are user_creatable, have them inherit TYPE_DYNAMIC_SYS_BUS_DEVICE, to explicit they can optionally be plugged on TYPE_PLATFORM_BUS_DEVICE.

---
  hw/i386/amd_iommu.c   | 2 --
  hw/i386/intel_iommu.c | 2 --
  hw/i386/x86-iommu.c   | 2 +-
  3 files changed, 1 insertion(+), 5 deletions(-)

r~

Reviewed-by on v1 6/9
#18 lore raw

Re: [PATCH 6/9] hw/i386: Have X86_IOMMU devices inherit from DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé

Thanks phil

Quoted 56 lines · Philippe Mathieu-Daudé

Caution: External email. Do not open attachments or click links, unless this email comes from a known sender and you know the content is safe.

Do not explain why _X86_IOMMU devices are user_creatable, have them inherit TYPE_DYNAMIC_SYS_BUS_DEVICE, to explicit they can optionally be plugged on TYPE_PLATFORM_BUS_DEVICE.

---
  hw/i386/amd_iommu.c   | 2 --
  hw/i386/intel_iommu.c | 2 --
  hw/i386/x86-iommu.c   | 2 +-
  3 files changed, 1 insertion(+), 5 deletions(-)
hw/i386/amd_iommu.c +0 -2
index 6b13ce894b1..e8e084c7cf8 100644
@@ -1687,8 +1687,6 @@ static void amdvi_sysbus_class_init(ObjectClass *klass, void *data)
      dc->hotpluggable = false;
      dc_class->realize = amdvi_sysbus_realize;
      dc_class->int_remap = amdvi_int_remap;
-    /* Supported by the pc-q35-* machine types */
-    dc->user_creatable = true;
      set_bit(DEVICE_CATEGORY_MISC, dc->categories);
      dc->desc = "AMD IOMMU (AMD-Vi) DMA Remapping device";
      device_class_set_props(dc, amdvi_properties);
hw/i386/intel_iommu.c +0 -2
index f366c223d0e..7fde0603bfe 100644
@@ -4871,8 +4871,6 @@ static void vtd_class_init(ObjectClass *klass, void *data)
      dc->hotpluggable = false;
      x86_class->realize = vtd_realize;
      x86_class->int_remap = vtd_int_remap;
-    /* Supported by the pc-q35-* machine types */
-    dc->user_creatable = true;
      set_bit(DEVICE_CATEGORY_MISC, dc->categories);
      dc->desc = "Intel IOMMU (VT-d) DMA Remapping device";
  }
hw/i386/x86-iommu.c +1 -2
index fed34b2fcfa..5cdd165af0d 100644
@@ -146,7 +146,7 @@ bool x86_iommu_ir_supported(X86IOMMUState *s)
 
  static const TypeInfo x86_iommu_info = {
      .name          = TYPE_X86_IOMMU_DEVICE,
-    .parent        = TYPE_SYS_BUS_DEVICE,
+    .parent        = TYPE_DYNAMIC_SYS_BUS_DEVICE,
      .instance_size = sizeof(X86IOMMUState),
      .class_init    = x86_iommu_class_init,
      .class_size    = sizeof(X86IOMMUClass),
--

2.47.1

#19 lore raw

PATCH 7/9 hw/net: Have eTSEC device inherit from DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé

Because the network eTSEC device can be optionally plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE.

---
 hw/net/fsl_etsec/etsec.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
hw/net/fsl_etsec/etsec.c +1 -3
index 781b9003954..3ce4fa2662d 100644
@@ -425,14 +425,12 @@ static void etsec_class_init(ObjectClass *klass, void *data)
     dc->realize = etsec_realize;
     device_class_set_legacy_reset(dc, etsec_reset);
     device_class_set_props(dc, etsec_properties);
-    /* Supported by ppce500 machine */
-    dc->user_creatable = true;
 }
 
 static const TypeInfo etsec_types[] = {
     {
         .name          = TYPE_ETSEC_COMMON,
-        .parent        = TYPE_SYS_BUS_DEVICE,
+        .parent        = TYPE_DYNAMIC_SYS_BUS_DEVICE,
         .instance_size = sizeof(eTSEC),
         .class_init    = etsec_class_init,
         .instance_init = etsec_instance_init,
Signature · 1 line
2.47.1
Reviewed-by on v1 7/9
#20 lore raw

Re: [PATCH 7/9] hw/net: Have eTSEC device inherit from DYNAMIC_SYS_BUS_DEVICE

Quoted 7 lines · Philippe Mathieu-Daudé

Because the network eTSEC device can be optionally plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE.

---
  hw/net/fsl_etsec/etsec.c | 4 +---
  1 file changed, 1 insertion(+), 3 deletions(-)

r~

Reviewed-by on v1 7/9
#21 lore raw

Re: [PATCH 7/9] hw/net: Have eTSEC device inherit from DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé
Quoted 34 lines · Philippe Mathieu-Daudé

Caution: External email. Do not open attachments or click links, unless this email comes from a known sender and you know the content is safe.

Because the network eTSEC device can be optionally plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE.

---
  hw/net/fsl_etsec/etsec.c | 4 +---
  1 file changed, 1 insertion(+), 3 deletions(-)
hw/net/fsl_etsec/etsec.c +1 -4
index 781b9003954..3ce4fa2662d 100644
@@ -425,14 +425,12 @@ static void etsec_class_init(ObjectClass *klass, void *data)
      dc->realize = etsec_realize;
      device_class_set_legacy_reset(dc, etsec_reset);
      device_class_set_props(dc, etsec_properties);
-    /* Supported by ppce500 machine */
-    dc->user_creatable = true;
  }
 
  static const TypeInfo etsec_types[] = {
      {
          .name          = TYPE_ETSEC_COMMON,
-        .parent        = TYPE_SYS_BUS_DEVICE,
+        .parent        = TYPE_DYNAMIC_SYS_BUS_DEVICE,
          .instance_size = sizeof(eTSEC),
          .class_init    = etsec_class_init,
          .instance_init = etsec_instance_init,
--

2.47.1

Tested-by Acked-by on v1 7/9
#22 lore raw

Re: [PATCH 7/9] hw/net: Have eTSEC device inherit from DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé

Am 25. Januar 2025 18:13:41 UTC schrieb "Philippe Mathieu-Daudé" <[email protected]>:

Quoted 4 lines

Because the network eTSEC device can be optionally plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE.

Quoted 24 lines

---
hw/net/fsl_etsec/etsec.c | 4 +---

1 file changed, 1 insertion(+), 3 deletions(-)
hw/net/fsl_etsec/etsec.c +1 -3
index 781b9003954..3ce4fa2662d 100644
@@ -425,14 +425,12 @@ static void etsec_class_init(ObjectClass *klass, void *data)
    dc->realize = etsec_realize;
    device_class_set_legacy_reset(dc, etsec_reset);
    device_class_set_props(dc, etsec_properties);
-    /* Supported by ppce500 machine */
-    dc->user_creatable = true;
 }
 
 static const TypeInfo etsec_types[] = {
    {
        .name          = TYPE_ETSEC_COMMON,
-        .parent        = TYPE_SYS_BUS_DEVICE,
+        .parent        = TYPE_DYNAMIC_SYS_BUS_DEVICE,
        .instance_size = sizeof(eTSEC),
        .class_init    = etsec_class_init,
        .instance_init = etsec_instance_init,
#23 lore raw

PATCH 8/9 hw/tpm: Have TPM TIS sysbus device inherit from DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé

Because the TPM TIS sysbus device can be optionally plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE.

---
 hw/tpm/tpm_tis_sysbus.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
hw/tpm/tpm_tis_sysbus.c +1 -2
index ee0bfe9538e..4f187690a28 100644
@@ -133,7 +133,6 @@ static void tpm_tis_sysbus_class_init(ObjectClass *klass, void *data)
     dc->vmsd  = &vmstate_tpm_tis_sysbus;
     tc->model = TPM_MODEL_TPM_TIS;
     dc->realize = tpm_tis_sysbus_realizefn;
-    dc->user_creatable = true;
     device_class_set_legacy_reset(dc, tpm_tis_sysbus_reset);
     tc->request_completed = tpm_tis_sysbus_request_completed;
     tc->get_version = tpm_tis_sysbus_get_tpm_version;
@@ -142,7 +141,7 @@ static void tpm_tis_sysbus_class_init(ObjectClass *klass, void *data)
 
 static const TypeInfo tpm_tis_sysbus_info = {
     .name = TYPE_TPM_TIS_SYSBUS,
-    .parent = TYPE_SYS_BUS_DEVICE,
+    .parent = TYPE_DYNAMIC_SYS_BUS_DEVICE,
     .instance_size = sizeof(TPMStateSysBus),
     .instance_init = tpm_tis_sysbus_initfn,
     .class_init  = tpm_tis_sysbus_class_init,
Signature · 1 line
2.47.1
Reviewed-by on v1 8/9
#24 lore raw

Re: [PATCH 8/9] hw/tpm: Have TPM TIS sysbus device inherit from DYNAMIC_SYS_BUS_DEVICE

Quoted 7 lines · Philippe Mathieu-Daudé

Because the TPM TIS sysbus device can be optionally plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE.

---
  hw/tpm/tpm_tis_sysbus.c | 3 +--
  1 file changed, 1 insertion(+), 2 deletions(-)

r~

Reviewed-by on v1 8/9
#25 lore raw

Re: [PATCH 8/9] hw/tpm: Have TPM TIS sysbus device inherit from DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé
Quoted 35 lines · Philippe Mathieu-Daudé

Caution: External email. Do not open attachments or click links, unless this email comes from a known sender and you know the content is safe.

Because the TPM TIS sysbus device can be optionally plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE.

---
  hw/tpm/tpm_tis_sysbus.c | 3 +--
  1 file changed, 1 insertion(+), 2 deletions(-)
hw/tpm/tpm_tis_sysbus.c +1 -3
index ee0bfe9538e..4f187690a28 100644
@@ -133,7 +133,6 @@ static void tpm_tis_sysbus_class_init(ObjectClass *klass, void *data)
      dc->vmsd  = &vmstate_tpm_tis_sysbus;
      tc->model = TPM_MODEL_TPM_TIS;
      dc->realize = tpm_tis_sysbus_realizefn;
-    dc->user_creatable = true;
      device_class_set_legacy_reset(dc, tpm_tis_sysbus_reset);
      tc->request_completed = tpm_tis_sysbus_request_completed;
      tc->get_version = tpm_tis_sysbus_get_tpm_version;
@@ -142,7 +141,7 @@ static void tpm_tis_sysbus_class_init(ObjectClass *klass, void *data)
 
  static const TypeInfo tpm_tis_sysbus_info = {
      .name = TYPE_TPM_TIS_SYSBUS,
-    .parent = TYPE_SYS_BUS_DEVICE,
+    .parent = TYPE_DYNAMIC_SYS_BUS_DEVICE,
      .instance_size = sizeof(TPMStateSysBus),
      .instance_init = tpm_tis_sysbus_initfn,
      .class_init  = tpm_tis_sysbus_class_init,
--

2.47.1

Reviewed-by on v1 8/9
#26 lore raw

Re: [PATCH 8/9] hw/tpm: Have TPM TIS sysbus device inherit from DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé
Quoted 4 lines · Philippe Mathieu-Daudé

Because the TPM TIS sysbus device can be optionally plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE.

Quoted 25 lines
---
  hw/tpm/tpm_tis_sysbus.c | 3 +--
  1 file changed, 1 insertion(+), 2 deletions(-)
hw/tpm/tpm_tis_sysbus.c +1 -2
index ee0bfe9538e..4f187690a28 100644
@@ -133,7 +133,6 @@ static void tpm_tis_sysbus_class_init(ObjectClass *klass, void *data)
      dc->vmsd  = &vmstate_tpm_tis_sysbus;
      tc->model = TPM_MODEL_TPM_TIS;
      dc->realize = tpm_tis_sysbus_realizefn;
-    dc->user_creatable = true;
      device_class_set_legacy_reset(dc, tpm_tis_sysbus_reset);
      tc->request_completed = tpm_tis_sysbus_request_completed;
      tc->get_version = tpm_tis_sysbus_get_tpm_version;
@@ -142,7 +141,7 @@ static void tpm_tis_sysbus_class_init(ObjectClass *klass, void *data)
 
  static const TypeInfo tpm_tis_sysbus_info = {
      .name = TYPE_TPM_TIS_SYSBUS,
-    .parent = TYPE_SYS_BUS_DEVICE,
+    .parent = TYPE_DYNAMIC_SYS_BUS_DEVICE,
      .instance_size = sizeof(TPMStateSysBus),
      .instance_init = tpm_tis_sysbus_initfn,
      .class_init  = tpm_tis_sysbus_class_init,
#27 lore raw

RFC PATCH 9/9 hw/xen: Have legacy Xen backend inherit from DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé

Because the legacy Xen backend devices can optionally be plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE. Remove the implicit TYPE_XENSYSDEV instance_size.

Untested, but I'm surprised the legacy devices work because they had a broken instance size (QDev instead of Sysbus...), so accesses of XenLegacyDevice fields were overwritting sysbus ones.

---
 include/hw/xen/xen_pvdev.h  | 3 ++-
 hw/xen/xen-legacy-backend.c | 7 ++-----
 2 files changed, 4 insertions(+), 6 deletions(-)
include/hw/xen/xen_pvdev.h +2 -1
index 0c984440476..48950dc2b57 100644
@@ -32,7 +32,8 @@ struct XenDevOps {
 };
 
 struct XenLegacyDevice {
-    DeviceState        qdev;
+    SysBusDevice parent_obj;
+
     const char         *type;
     int                dom;
     int                dev;
hw/xen/xen-legacy-backend.c +2 -5
index 118c571b3a7..4d079e35d83 100644
@@ -640,16 +640,14 @@ static void xendev_class_init(ObjectClass *klass, void *data)
     DeviceClass *dc = DEVICE_CLASS(klass);
 
     set_bit(DEVICE_CATEGORY_MISC, dc->categories);
-    /* xen-backend devices can be plugged/unplugged dynamically */
-    dc->user_creatable = true;
     dc->bus_type = TYPE_XENSYSBUS;
 }
 
 static const TypeInfo xendev_type_info = {
     .name          = TYPE_XENBACKEND,
-    .parent        = TYPE_DEVICE,
+    .parent        = TYPE_DYNAMIC_SYS_BUS_DEVICE,
     .class_init    = xendev_class_init,
-    .instance_size = sizeof(struct XenLegacyDevice),
+    .instance_size = sizeof(XenLegacyDevice),
 };
 
 static void xen_sysbus_class_init(ObjectClass *klass, void *data)
@@ -672,7 +670,6 @@ static const TypeInfo xensysbus_info = {
 static const TypeInfo xensysdev_info = {
     .name          = TYPE_XENSYSDEV,
     .parent        = TYPE_SYS_BUS_DEVICE,
-    .instance_size = sizeof(SysBusDevice),
 };
 
 static void xenbe_register_types(void)
Signature · 1 line
2.47.1
#28 lore raw

PATCH 1/1 hw/*/xen*: Prefer QOM cast for XenLegacyDevice

Makes the code less sensitive regarding changes in the class hierarchy which will be performed in the next patch.

---
 hw/usb/xen-usb.c            | 6 +++---
 hw/xen/xen-legacy-backend.c | 2 +-
 hw/xen/xen_pvdev.c          | 2 +-
 3 files changed, 5 insertions(+), 5 deletions(-)
hw/usb/xen-usb.c +3 -3
index 13901625c0..3da30efc44 100644
@@ -755,10 +755,10 @@ static void usbback_portid_add(struct usbback_info *usbif, unsigned port,
 
     qdict = qdict_new();
     qdict_put_str(qdict, "driver", "usb-host");
-    tmp = g_strdup_printf("%s.0", usbif->xendev.qdev.id);
+    tmp = g_strdup_printf("%s.0", DEVICE(&usbif->xendev)->id);
     qdict_put_str(qdict, "bus", tmp);
     g_free(tmp);
-    tmp = g_strdup_printf("%s-%u", usbif->xendev.qdev.id, port);
+    tmp = g_strdup_printf("%s-%u", DEVICE(&usbif->xendev)->id, port);
     qdict_put_str(qdict, "id", tmp);
     g_free(tmp);
     qdict_put_int(qdict, "port", port);
@@ -1022,7 +1022,7 @@ static void usbback_alloc(struct XenLegacyDevice *xendev)
     usbif = container_of(xendev, struct usbback_info, xendev);
 
     usb_bus_new(&usbif->bus, sizeof(usbif->bus), &xen_usb_bus_ops,
-                DEVICE(&xendev->qdev));
+                DEVICE(xendev));
     for (i = 0; i < USBBACK_MAXPORTS; i++) {
         p = &(usbif->ports[i].port);
         usb_register_port(&usbif->bus, p, usbif, i, &xen_usb_port_ops,
hw/xen/xen-legacy-backend.c +1 -1
index 118c571b3a..ca2fe0e6b3 100644
@@ -163,7 +163,7 @@ static struct XenLegacyDevice *xen_be_get_xendev(const char *type, int dom,
 
     /* init new xendev */
     xendev = g_malloc0(ops->size);
-    object_initialize(&xendev->qdev, ops->size, TYPE_XENBACKEND);
+    object_initialize(xendev, ops->size, TYPE_XENBACKEND);
     OBJECT(xendev)->free = g_free;
     qdev_set_id(DEVICE(xendev), g_strdup_printf("xen-%s-%d", type, dev),
                 &error_fatal);
hw/xen/xen_pvdev.c +1 -1
index c9143ba259..fe95b62d13 100644
@@ -273,7 +273,7 @@ void xen_pv_del_xendev(struct XenLegacyDevice *xendev)
 
     QTAILQ_REMOVE(&xendevs, xendev, next);
 
-    qdev_unplug(&xendev->qdev, NULL);
+    qdev_unplug(DEVICE(xendev), NULL);
 }
 
 void xen_pv_insert_xendev(struct XenLegacyDevice *xendev)
Signature · 1 line
2.48.1
author Reviewed-by on v1 1/1
#29 lore raw

Re: [PATCH] hw/*/xen*: Prefer QOM cast for XenLegacyDevice

Quoted 9 lines · Bernhard Beschow

Makes the code less sensitive regarding changes in the class hierarchy which will be performed in the next patch.

---
  hw/usb/xen-usb.c            | 6 +++---
  hw/xen/xen-legacy-backend.c | 2 +-
  hw/xen/xen_pvdev.c          | 2 +-
  3 files changed, 5 insertions(+), 5 deletions(-)
#30 lore raw

Re: [RFC PATCH 9/9] hw/xen: Have legacy Xen backend inherit from DYNAMIC_SYS_BUS_DEVICE

In reply to Philippe Mathieu-Daudé

Am 25. Januar 2025 18:13:43 UTC schrieb "Philippe Mathieu-Daudé" <[email protected]>:

Quoted 19 lines

Because the legacy Xen backend devices can optionally be plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE. Remove the implicit TYPE_XENSYSDEV instance_size.

Untested, but I'm surprised the legacy devices work because they had a broken instance size (QDev instead of Sysbus...), so accesses of XenLegacyDevice fields were overwritting sysbus ones.

---
include/hw/xen/xen_pvdev.h | 3 ++-
hw/xen/xen-legacy-backend.c | 7 ++-----

2 files changed, 4 insertions(+), 6 deletions(-)
include/hw/xen/xen_pvdev.h +0 -0
index 0c984440476..48950dc2b57 100644

include/hw/xen/xen_pvdev.h struct XenDevOps { · line 35

Quoted 5 lines
include/hw/xen/xen_pvdev.h +1 -1
@@ -32,7 +32,8 @@ struct XenDevOps {
 };
 
 struct XenLegacyDevice {
-    DeviceState        qdev;
+    SysBusDevice parent_obj;

This then needs sysbus.h rather than qdev-core.h include.

Moreover, the patch in the reply needs to be inserted into the series before this patch.

Both are needed for the patch to compile.

Best regards,
Bernhard

Quoted 35 lines
include/hw/xen/xen_pvdev.h +1 -0
+
    const char         *type;
    int                dom;
    int                dev;
hw/xen/xen-legacy-backend.c +2 -5
index 118c571b3a7..4d079e35d83 100644
@@ -640,16 +640,14 @@ static void xendev_class_init(ObjectClass *klass, void *data)
    DeviceClass *dc = DEVICE_CLASS(klass);
 
    set_bit(DEVICE_CATEGORY_MISC, dc->categories);
-    /* xen-backend devices can be plugged/unplugged dynamically */
-    dc->user_creatable = true;
    dc->bus_type = TYPE_XENSYSBUS;
 }
 
 static const TypeInfo xendev_type_info = {
    .name          = TYPE_XENBACKEND,
-    .parent        = TYPE_DEVICE,
+    .parent        = TYPE_DYNAMIC_SYS_BUS_DEVICE,
    .class_init    = xendev_class_init,
-    .instance_size = sizeof(struct XenLegacyDevice),
+    .instance_size = sizeof(XenLegacyDevice),
 };
 
 static void xen_sysbus_class_init(ObjectClass *klass, void *data)
@@ -672,7 +670,6 @@ static const TypeInfo xensysbus_info = {
 static const TypeInfo xensysdev_info = {
    .name          = TYPE_XENSYSDEV,
    .parent        = TYPE_SYS_BUS_DEVICE,
-    .instance_size = sizeof(SysBusDevice),
 };
 
 static void xenbe_register_types(void)
author Tested-by Bernhard Beschow Acked-by Bernhard Beschow
#31 lore raw

Re: [RFC PATCH 9/9] hw/xen: Have legacy Xen backend inherit from DYNAMIC_SYS_BUS_DEVICE

Hi Bernhard,

Quoted 28 lines · Bernhard Beschow

Am 25. Januar 2025 18:13:43 UTC schrieb "Philippe Mathieu-Daudé" <[email protected]>:

Quoted 19 lines

Because the legacy Xen backend devices can optionally be plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE. Remove the implicit TYPE_XENSYSDEV instance_size.

Untested, but I'm surprised the legacy devices work because they had a broken instance size (QDev instead of Sysbus...), so accesses of XenLegacyDevice fields were overwritting sysbus ones.

---
include/hw/xen/xen_pvdev.h | 3 ++-
hw/xen/xen-legacy-backend.c | 7 ++-----

2 files changed, 4 insertions(+), 6 deletions(-)
include/hw/xen/xen_pvdev.h +0 -0
index 0c984440476..48950dc2b57 100644

include/hw/xen/xen_pvdev.h struct XenDevOps { · line 35

Quoted 5 lines
include/hw/xen/xen_pvdev.h +1 -1
@@ -32,7 +32,8 @@ struct XenDevOps {
 };
 
 struct XenLegacyDevice {
-    DeviceState        qdev;
+    SysBusDevice parent_obj;

This then needs sysbus.h rather than qdev-core.h include.

Quoted 3 lines

Moreover, the patch in the reply needs to be inserted into the series before this patch.

Both are needed for the patch to compile.

Per your reply on patch #7, might I include your

(or R-b)

squashing:

-- >8 --

include/hw/xen/xen_pvdev.h +1 -1
index 48950dc2b57..629bec90d09 100644
@@ -1,7 +1,7 @@
 #ifndef QEMU_HW_XEN_PVDEV_H
 #define QEMU_HW_XEN_PVDEV_H
 
-#include "hw/qdev-core.h"
+#include "hw/sysbus.h"
 #include "hw/xen/xen_backend_ops.h"
 
 /* ------------------------------------------------------------- */

---

?

Reviewed-by
#32 lore raw

Re: [RFC PATCH 9/9] hw/xen: Have legacy Xen backend inherit from DYNAMIC_SYS_BUS_DEVICE

Am 4. Februar 2025 21:25:46 UTC schrieb "Philippe Mathieu-Daudé" <[email protected]>:

Quoted 33 lines

Hi Bernhard,

Quoted 30 lines · Bernhard Beschow

Am 25. Januar 2025 18:13:43 UTC schrieb "Philippe Mathieu-Daudé" <[email protected]>:

Quoted 19 lines

Because the legacy Xen backend devices can optionally be plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE. Remove the implicit TYPE_XENSYSDEV instance_size.

Untested, but I'm surprised the legacy devices work because they had a broken instance size (QDev instead of Sysbus...), so accesses of XenLegacyDevice fields were overwritting sysbus ones.

---
include/hw/xen/xen_pvdev.h | 3 ++-
hw/xen/xen-legacy-backend.c | 7 ++-----

2 files changed, 4 insertions(+), 6 deletions(-)
include/hw/xen/xen_pvdev.h +0 -0
index 0c984440476..48950dc2b57 100644

include/hw/xen/xen_pvdev.h struct XenDevOps { · line 35

Quoted 5 lines
include/hw/xen/xen_pvdev.h +1 -1
@@ -32,7 +32,8 @@ struct XenDevOps {
 };
 
 struct XenLegacyDevice {
-    DeviceState        qdev;
+    SysBusDevice parent_obj;

This then needs sysbus.h rather than qdev-core.h include.

Moreover, the patch in the reply needs to be inserted into the series before this patch.

Quoted 7 lines
Quoted 1 line

Both are needed for the patch to compile.

Per your reply on patch #7, might I include your

(or R-b)

I only did a compile test and I'm not a Xen maintainer, so I guess above tags don't apply. Right?

Quoted 19 lines

squashing:

-- >8 --

include/hw/xen/xen_pvdev.h +1 -2
index 48950dc2b57..629bec90d09 100644
@@ -1,7 +1,7 @@
 #ifndef QEMU_HW_XEN_PVDEV_H
 #define QEMU_HW_XEN_PVDEV_H
 
-#include "hw/qdev-core.h"
+#include "hw/sysbus.h"
 #include "hw/xen/xen_backend_ops.h"
 
 /* ------------------------------------------------------------- */
---
Quoted 1 line

?

With the squash applied:

#33 lore raw

Re: [RFC PATCH 9/9] hw/xen: Have legacy Xen backend inherit from DYNAMIC_SYS_BUS_DEVICE

Quoted 38 lines · Bernhard Beschow

Am 4. Februar 2025 21:25:46 UTC schrieb "Philippe Mathieu-Daudé" <[email protected]>:

Quoted 35 lines

Hi Bernhard,

Quoted 31 lines · Bernhard Beschow

Am 25. Januar 2025 18:13:43 UTC schrieb "Philippe Mathieu-Daudé" <[email protected]>:

Quoted 19 lines

Because the legacy Xen backend devices can optionally be plugged on the TYPE_PLATFORM_BUS_DEVICE, have it inherit TYPE_DYNAMIC_SYS_BUS_DEVICE. Remove the implicit TYPE_XENSYSDEV instance_size.

Untested, but I'm surprised the legacy devices work because they had a broken instance size (QDev instead of Sysbus...), so accesses of XenLegacyDevice fields were overwritting sysbus ones.

---
include/hw/xen/xen_pvdev.h | 3 ++-
hw/xen/xen-legacy-backend.c | 7 ++-----

2 files changed, 4 insertions(+), 6 deletions(-)
include/hw/xen/xen_pvdev.h +0 -0
index 0c984440476..48950dc2b57 100644

include/hw/xen/xen_pvdev.h struct XenDevOps { · line 35

Quoted 5 lines
include/hw/xen/xen_pvdev.h +1 -1
@@ -32,7 +32,8 @@ struct XenDevOps {
 };
 
 struct XenLegacyDevice {
-    DeviceState        qdev;
+    SysBusDevice parent_obj;

This then needs sysbus.h rather than qdev-core.h include.

Moreover, the patch in the reply needs to be inserted into the series before this patch.

Both are needed for the patch to compile.

Quoted 7 lines
Quoted 5 lines

Per your reply on patch #7, might I include your

(or R-b)

I only did a compile test and I'm not a Xen maintainer, so I guess above tags don't apply. Right?

Indeed, A-b is preferable for maintainers, but its meaning depends.

Xen maintainers have been Cc'ed for 2 weeks. If they report a problem, we can still revert.

Quoted 18 lines
Quoted 18 lines

squashing:

-- >8 --

include/hw/xen/xen_pvdev.h +1 -1
index 48950dc2b57..629bec90d09 100644
@@ -1,7 +1,7 @@
 #ifndef QEMU_HW_XEN_PVDEV_H
 #define QEMU_HW_XEN_PVDEV_H
 
-#include "hw/qdev-core.h"
+#include "hw/sysbus.h"
 #include "hw/xen/xen_backend_ops.h"
 
Quoted 7 lines
Quoted 4 lines
include/hw/xen/xen_pvdev.h +0 -1
 /* ------------------------------------------------------------- */
---

?

With the squash applied:

Thanks!

Reviewed-by on the v1 series
#34 lore raw
In reply to Philippe Mathieu-Daudé
Quoted 8 lines · Philippe Mathieu-Daudé

Some SysBus devices can optionally be dynamically plugged onto the sysbus-platform-bus (then virtual guests are aware of mmio mapping and IRQs via device tree / ACPI rules).

This series makes these devices explicit by having them implement the DYNAMIC_SYS_BUS_DEVICE class, which only sets 'user_creatable' flag to %true but makes the codebase a bit clearer (IMHO, at least now we can grep for DYNAMIC_SYS_BUS_DEVICE).

I love it. Thank you! This is clearly much more readable than the magical hint swizzling I did :).

Alex

#35 lore raw
In reply to Philippe Mathieu-Daudé
Quoted 3 lines · Philippe Mathieu-Daudé

Some SysBus devices can optionally be dynamically plugged onto the sysbus-platform-bus (then virtual guests are aware of mmio mapping and IRQs via device tree / ACPI rules).

Do we have some sane way to have user-pluggable sysbus devices on arm?

I've played around with that a bit, with the uefi variable service I'm working on. Specifically I'd prefer to *not* have a patch wiring things up in machine type code like this ...

https://lore.kernel.org/qemu-devel/[email protected]/

... and just use 'qemu -device uefi-vars-sysbus' instead.

Something like AcpiDevAmlIfClass but for device tree seems to not exist though. Also apparently AcpiDevAmlIfClass is not used.

take care,
Gerd

#36 lore raw
Quoted 7 lines · Gerd Hoffmann
Quoted 3 lines · Philippe Mathieu-Daudé

Some SysBus devices can optionally be dynamically plugged onto the sysbus-platform-bus (then virtual guests are aware of mmio mapping and IRQs via device tree / ACPI rules).

Do we have some sane way to have user-pluggable sysbus devices on arm?

The answer in a general sense is "no, because user pluggable sysbus is a weird idea". "sysbus" means "it's wired into a specific bit of the memory map and to specific IRQs, and whoever does that needs to know what IRQs and bits of memory are usable, and the guest OS needs to know it's there". "user-pluggable" means "it's all automatic and the guest can just do some kind of probing for what is or isn't present". All the platform bus stuff is a nasty mess that's working around the things people want to plug in not being clean devices on probeable buses :-( And the platform bus is only supported on the "virt" board, because that's the only one where QEMU is generating its own dtb or ACPI tables where it can tell the guest "hey, there's some device here".

-- PMM

#37 lore raw
Quoted 9 lines · Peter Maydell
Quoted 7 lines · Gerd Hoffmann
Quoted 3 lines · Philippe Mathieu-Daudé

Some SysBus devices can optionally be dynamically plugged onto the sysbus-platform-bus (then virtual guests are aware of mmio mapping and IRQs via device tree / ACPI rules).

Do we have some sane way to have user-pluggable sysbus devices on arm?

Quoted 13 lines

The answer in a general sense is "no, because user pluggable sysbus is a weird idea". "sysbus" means "it's wired into a specific bit of the memory map and to specific IRQs, and whoever does that needs to know what IRQs and bits of memory are usable, and the guest OS needs to know it's there". "user-pluggable" means "it's all automatic and the guest can just do some kind of probing for what is or isn't present". All the platform bus stuff is a nasty mess that's working around the things people want to plug in not being clean devices on probeable buses :-( And the platform bus is only supported on the "virt" board, because that's the only one where QEMU is generating its own dtb or ACPI tables where it can tell the guest "hey, there's some device here".

There are some SoCs that have memory mapped devices but different versions in the same family have different devices. Either older ones missing some devices or have less USB or network ports while newer SoCs have more of those or they have PCIe instead of PCI. Modelling these could use pluggable sysbus devices so one could add the devices needed for a SoC version without having to write or modify a board code. I think Bernhard's attempt to try creating e500 SoCs from a device tree goes in that direction too. We could also model this by having a SoC that can instantiate devices based on some properties but maybe pluggable devices could be more generic for this. The issue seems to be how to tell the board or SoC where to map it and what IRQ to connect it as this is done by the board and not the device so properties on the device to set these does not really help unless the board can somehow query it and instantiate the devices based on that. Otherwise whatever handles the -device option to create the device would need knowledge about the board. (E.g. the e500 devices are mapped in the CCSR memory region so one can't just use system address space for them.)

Regards,
BALATON Zoltan

#38 lore raw
Quoted 24 lines · BALATON Zoltan
Quoted 9 lines · Peter Maydell
Quoted 7 lines · Gerd Hoffmann
Quoted 3 lines · Philippe Mathieu-Daudé

Some SysBus devices can optionally be dynamically plugged onto the sysbus-platform-bus (then virtual guests are aware of mmio mapping and IRQs via device tree / ACPI rules).

Do we have some sane way to have user-pluggable sysbus devices on arm?

Quoted 13 lines

The answer in a general sense is "no, because user pluggable sysbus is a weird idea". "sysbus" means "it's wired into a specific bit of the memory map and to specific IRQs, and whoever does that needs to know what IRQs and bits of memory are usable, and the guest OS needs to know it's there". "user-pluggable" means "it's all automatic and the guest can just do some kind of probing for what is or isn't present". All the platform bus stuff is a nasty mess that's working around the things people want to plug in not being clean devices on probeable buses :-( And the platform bus is only supported on the "virt" board, because that's the only one where QEMU is generating its own dtb or ACPI tables where it can tell the guest "hey, there's some device here".

Quoted 1 line

There are some SoCs that have memory mapped devices but different versions in the same family have different devices. Either older ones missing some devices or have less USB or network ports while newer SoCs have more of those or they have PCIe instead of PCI. Modelling these could use pluggable sysbus devices so one could add the devices needed for a SoC version without having to write or modify a board code. I think Bernhard's attempt to try creating e500 SoCs from a device tree goes in that direction too. We could also model this by having a SoC that can instantiate devices based on some properties but maybe pluggable devices could be more generic for this. The issue seems to be how to tell the board or SoC where to map it and what IRQ to connect it as this is done by the board and not the device so properties on the device to set these does not really help unless the board can somehow query it and instantiate the devices based on that. Otherwise whatever handles the -device option to create the device would need knowledge about the board. (E.g. the e500 devices are mapped in the CCSR memory region so one can't just use system address space for them.)

IIRC Bernard's series takes a DTB as input and create the machine matching this DTB.

As Peter explained, sysbus-platform-bus fits TYPE_DYNAMIC_SYS_BUS_DEVICE in free slots, then generates the corresponding ACPI/DTB.

What you describe seems closer to the QEMU Dynamic Machine project, following Damien's idea:
https://lore.kernel.org/qemu-devel/[email protected]/
We are not quite there yet...

#39 lore raw

Am 28. Januar 2025 15:10:18 UTC schrieb "Philippe Mathieu-Daudé" <[email protected]>:

Quoted 25 lines · BALATON Zoltan
Quoted 24 lines · Peter Maydell
Quoted 22 lines · Gerd Hoffmann
Quoted 7 lines · Philippe Mathieu-Daudé
Quoted 3 lines

Some SysBus devices can optionally be dynamically plugged onto the sysbus-platform-bus (then virtual guests are aware of mmio mapping and IRQs via device tree / ACPI rules).

Do we have some sane way to have user-pluggable sysbus devices on arm?

The answer in a general sense is "no, because user pluggable sysbus is a weird idea". "sysbus" means "it's wired into a specific bit of the memory map and to specific IRQs, and whoever does that needs to know what IRQs and bits of memory are usable, and the guest OS needs to know it's there". "user-pluggable" means "it's all automatic and the guest can just do some kind of probing for what is or isn't present". All the platform bus stuff is a nasty mess that's working around the things people want to plug in not being clean devices on probeable buses :-( And the platform bus is only supported on the "virt" board, because that's the only one where QEMU is generating its own dtb or ACPI tables where it can tell the guest "hey, there's some device here".

Quoted 4 lines
Quoted 1 line

There are some SoCs that have memory mapped devices but different versions in the same family have different devices. Either older ones missing some devices or have less USB or network ports while newer SoCs have more of those or they have PCIe instead of PCI. Modelling these could use pluggable sysbus devices so one could add the devices needed for a SoC version without having to write or modify a board code. I think Bernhard's attempt to try creating e500 SoCs from a device tree goes in that direction too. We could also model this by having a SoC that can instantiate devices based on some properties but maybe pluggable devices could be more generic for this. The issue seems to be how to tell the board or SoC where to map it and what IRQ to connect it as this is done by the board and not the device so properties on the device to set these does not really help unless the board can somehow query it and instantiate the devices based on that. Otherwise whatever handles the -device option to create the device would need knowledge about the board. (E.g. the e500 devices are mapped in the CCSR memory region so one can't just use system address space for them.)

IIRC Bernard's series takes a DTB as input and create the machine matching this DTB.

That's correct. It's still on my todo list to send an RFC. I first wanted to gain some experience implementing a machine in the classic way which I've now done by means of the imx8mp-evk series. Once I clean up the e500-fdt branch I'd send an RFC.

Best regards,
Bernhard

Quoted 8 lines

As Peter explained, sysbus-platform-bus fits TYPE_DYNAMIC_SYS_BUS_DEVICE in free slots, then generates the corresponding ACPI/DTB.

What you describe seems closer to the QEMU Dynamic Machine project, following Damien's idea:
https://lore.kernel.org/qemu-devel/[email protected]/
We are not quite there yet...

#40 lore raw
In reply to Philippe Mathieu-Daudé
Quoted 11 lines · Philippe Mathieu-Daudé

Philippe Mathieu-Daudé (9):
hw/sysbus: Use sizeof(BusState) in main_system_bus_create() hw/sysbus: Declare QOM types using DEFINE_TYPES() macro hw/sysbus: Introduce TYPE_DYNAMIC_SYS_BUS_DEVICE
hw/vfio: Have VFIO_PLATFORM devices inherit from
DYNAMIC_SYS_BUS_DEVICE
hw/display: Have RAMFB device inherit from DYNAMIC_SYS_BUS_DEVICE hw/i386: Have X86_IOMMU devices inherit from DYNAMIC_SYS_BUS_DEVICE hw/net: Have eTSEC device inherit from DYNAMIC_SYS_BUS_DEVICE hw/tpm: Have TPM TIS sysbus device inherit from DYNAMIC_SYS_BUS_DEVICE hw/xen: Have legacy Xen backend inherit from DYNAMIC_SYS_BUS_DEVICE

Series queued (including Bernhard's patch), thanks.