[PATCH 0/5] Trivial cleanup

[PATCH 0/5] Trivial cleanup

[Bernhard Beschow]

This series contains some low-hanging fruit fixes I came up with when
investigating Rust for QEMU, especially QOM object lifecycle vs. the Drop trait.

I noticed that no finalize methods were called in Rust sysbus devices when QEMU
quits, and neither are its C variants. What is probably missing in the first
place is a

  object_unref(object_get_root());

at the end of qemu_cleanup() in runstate.c (see TODO there). Adding it before
this series won't clean up machines since their refcount won't reach zero. With
this series applied this additional line will cause various issues when quitting
QEMU, allowing the problem to be investigated. Fixing it is beyond the scope of
this series (and my time budget).

Testing done:
* `make check`
* Run imx8mp-evk with a Buildroot image

Bernhard Beschow (5):
  hw/arm/fsl-imx8mp: Do not create redundant unimplemented devices
  hw/arm/fsl-imx8mp: Fix parent of ocram memory region
  hw/arm/imx8mp-evk: Fix reference count of SoC object
  system/vl: Fix reference count of machine object
  Revert "sysbus: add irq_routing_notifier"

 include/hw/core/sysbus.h | 1 -
 hw/arm/fsl-imx8mp.c      | 3 ++-
 hw/arm/imx8mp-evk.c      | 4 ++--
 hw/core/sysbus.c         | 6 ------
 system/vl.c              | 1 +
 5 files changed, 5 insertions(+), 10 deletions(-)

-- 
2.53.0

[PATCH 1/5] hw/arm/fsl-imx8mp: Do not create redundant unimplemented devices

[Bernhard Beschow]

The GPT devices are implemented. No need to create redundant
unimplemented devices.

Fixes: f8b26121762c ("hw/arm/fsl-imx8mp: Implement general purpose
timers")
cc: Gaurav Sharma <gaurav.sharma_7@nxp.com>
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
---
 hw/arm/fsl-imx8mp.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/hw/arm/fsl-imx8mp.c b/hw/arm/fsl-imx8mp.c
index 79f9142777..0119a18416 100644
--- a/hw/arm/fsl-imx8mp.c
+++ b/hw/arm/fsl-imx8mp.c
@@ -688,6 +688,7 @@ static void fsl_imx8mp_realize(DeviceState *dev, Error **errp)
         case FSL_IMX8MP_GIC_DIST:
         case FSL_IMX8MP_GIC_REDIST:
         case FSL_IMX8MP_GPIO1 ... FSL_IMX8MP_GPIO5:
+        case FSL_IMX8MP_GPT1 ... FSL_IMX8MP_GPT6:
         case FSL_IMX8MP_ECSPI1 ... FSL_IMX8MP_ECSPI3:
         case FSL_IMX8MP_ENET1:
         case FSL_IMX8MP_I2C1 ... FSL_IMX8MP_I2C6:
-- 
2.53.0

[PATCH 2/5] hw/arm/fsl-imx8mp: Fix parent of ocram memory region

[Bernhard Beschow]

Rather than having a NULL parent, let the containing SoC object be the
parent. This cleans up the QOM composition tree a bit.

Fixes: 1aaf3478684f ("hw/arm/fsl-imx8mp: Add on-chip RAM")
cc: Gaurav Sharma <gaurav.sharma_7@nxp.com>
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
---
 hw/arm/fsl-imx8mp.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/hw/arm/fsl-imx8mp.c b/hw/arm/fsl-imx8mp.c
index 0119a18416..839c925e4a 100644
--- a/hw/arm/fsl-imx8mp.c
+++ b/hw/arm/fsl-imx8mp.c
@@ -671,7 +671,7 @@ static void fsl_imx8mp_realize(DeviceState *dev, Error **errp)
                     fsl_imx8mp_memmap[FSL_IMX8MP_PCIE_PHY1].addr);
 
     /* On-Chip RAM */
-    if (!memory_region_init_ram(&s->ocram, NULL, "imx8mp.ocram",
+    if (!memory_region_init_ram(&s->ocram, OBJECT(dev), "imx8mp.ocram",
                                 fsl_imx8mp_memmap[FSL_IMX8MP_OCRAM].size,
                                 errp)) {
         return;
-- 
2.53.0

[PATCH 3/5] hw/arm/imx8mp-evk: Fix reference count of SoC object

[Bernhard Beschow]

object_new() leaves a reference count of one, and object_property_add_child()
then leaves it at two. This would leak the SoC object if the machine object was
destroyed since the machine is the SoC's sole owner. Fix this by converting to
object_new_with_props() which gets the parenting and reference count right in
one go.

Fixes: a4eefc69b237 ("hw/arm: Add i.MX 8M Plus EVK board")
cc: Gaurav Sharma <gaurav.sharma_7@nxp.com>
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
---
 hw/arm/imx8mp-evk.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/hw/arm/imx8mp-evk.c b/hw/arm/imx8mp-evk.c
index b84ac91a17..cfc3038c36 100644
--- a/hw/arm/imx8mp-evk.c
+++ b/hw/arm/imx8mp-evk.c
@@ -74,8 +74,8 @@ static void imx8mp_evk_init(MachineState *machine)
         .modify_dtb = imx8mp_evk_modify_dtb,
     };
 
-    s = FSL_IMX8MP(object_new(TYPE_FSL_IMX8MP));
-    object_property_add_child(OBJECT(machine), "soc", OBJECT(s));
+    s = FSL_IMX8MP(object_new_with_props(TYPE_FSL_IMX8MP, OBJECT(machine),
+                                         "soc", &error_fatal, NULL));
     object_property_set_uint(OBJECT(s), "fec1-phy-num", 1, &error_fatal);
     sysbus_realize_and_unref(SYS_BUS_DEVICE(s), &error_fatal);
 
-- 
2.53.0

[PATCH 4/5] system/vl: Fix reference count of machine object

[Bernhard Beschow]

object_new_with_class() and object_property_add_child() leave a reference count
of two while the QOM root is the sole owner. Fix the reference count to be one
by calling object_unref() once.

Signed-off-by: Bernhard Beschow <shentey@gmail.com>
---
 system/vl.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/system/vl.c b/system/vl.c
index 38d7b849e0..91b390dafd 100644
--- a/system/vl.c
+++ b/system/vl.c
@@ -2197,6 +2197,7 @@ static void qemu_create_machine(QDict *qdict)
     current_machine = MACHINE(object_new_with_class(OBJECT_CLASS(machine_class)));
     object_property_add_child(object_get_root(), "machine",
                               OBJECT(current_machine));
+    object_unref(current_machine);
     qemu_create_machine_containers(OBJECT(current_machine));
     object_property_add_child(machine_get_container("unattached"),
                               "sysbus", OBJECT(sysbus_get_default()));
-- 
2.53.0

[PATCH 5/5] Revert "sysbus: add irq_routing_notifier"

[Bernhard Beschow]

The callback has been introduced in commit 715ca691daca ("sysbus: add
irq_routing_notifier") for use in VFIO platform. Meanwhile, VFIO
platform has been removed via commit 762c85543948 ("vfio: Remove 'vfio-
platform'") which was its only user. Remove this unused code.

This reverts commit 715ca691daca081108b33306faa6fa102f0df8d8.

cc: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
---
 include/hw/core/sysbus.h | 1 -
 hw/core/sysbus.c         | 6 ------
 2 files changed, 7 deletions(-)

diff --git a/include/hw/core/sysbus.h b/include/hw/core/sysbus.h
index c0d18d9e00..f3c4259d29 100644
--- a/include/hw/core/sysbus.h
+++ b/include/hw/core/sysbus.h
@@ -50,7 +50,6 @@ struct SysBusDeviceClass {
      * omitted then. (This is not considered a fatal error.)
      */
     char *(*explicit_ofw_unit_address)(const SysBusDevice *dev);
-    void (*connect_irq_notifier)(SysBusDevice *dev, qemu_irq irq);
 };
 
 struct SysBusDevice {
diff --git a/hw/core/sysbus.c b/hw/core/sysbus.c
index 3adf2f2faf..3e1160ee92 100644
--- a/hw/core/sysbus.c
+++ b/hw/core/sysbus.c
@@ -104,13 +104,7 @@ qemu_irq sysbus_get_connected_irq(const SysBusDevice *dev, int n)
 
 void sysbus_connect_irq(SysBusDevice *dev, int n, qemu_irq irq)
 {
-    SysBusDeviceClass *sbd = SYS_BUS_DEVICE_GET_CLASS(dev);
-
     qdev_connect_gpio_out_named(DEVICE(dev), SYSBUS_DEVICE_GPIO_IRQ, n, irq);
-
-    if (sbd->connect_irq_notifier) {
-        sbd->connect_irq_notifier(dev, irq);
-    }
 }
 
 /* Check whether an MMIO region exists */
-- 
2.53.0

Re: [PATCH 0/5] Trivial cleanup

[Bernhard Beschow]

Am 8. März 2026 20:35:11 UTC schrieb Bernhard Beschow <shentey@gmail.com>:

+Markus

>This series contains some low-hanging fruit fixes I came up with when
>investigating Rust for QEMU, especially QOM object lifecycle vs. the Drop trait.
>
>I noticed that no finalize methods were called in Rust sysbus devices when QEMU
>quits, and neither are its C variants. What is probably missing in the first
>place is a
>
>  object_unref(object_get_root());
>
>at the end of qemu_cleanup() in runstate.c (see TODO there). Adding it before
>this series won't clean up machines since their refcount won't reach zero. With
>this series applied this additional line will cause various issues when quitting
>QEMU, allowing the problem to be investigated. Fixing it is beyond the scope of
>this series (and my time budget).
>
>Testing done:
>* `make check`
>* Run imx8mp-evk with a Buildroot image
>
>Bernhard Beschow (5):
>  hw/arm/fsl-imx8mp: Do not create redundant unimplemented devices
>  hw/arm/fsl-imx8mp: Fix parent of ocram memory region
>  hw/arm/imx8mp-evk: Fix reference count of SoC object
>  system/vl: Fix reference count of machine object
>  Revert "sysbus: add irq_routing_notifier"
>
> include/hw/core/sysbus.h | 1 -
> hw/arm/fsl-imx8mp.c      | 3 ++-
> hw/arm/imx8mp-evk.c      | 4 ++--
> hw/core/sysbus.c         | 6 ------
> system/vl.c              | 1 +
> 5 files changed, 5 insertions(+), 10 deletions(-)
>

Re: [PATCH 5/5] Revert "sysbus: add irq_routing_notifier"

[Cédric Le Goater]

On 3/8/26 21:35, Bernhard Beschow wrote:
> The callback has been introduced in commit 715ca691daca ("sysbus: add
> irq_routing_notifier") for use in VFIO platform. Meanwhile, VFIO
> platform has been removed via commit 762c85543948 ("vfio: Remove 'vfio-
> platform'") which was its only user. Remove this unused code.
> 
> This reverts commit 715ca691daca081108b33306faa6fa102f0df8d8.
> 
> cc: Cédric Le Goater <clg@redhat.com>
> Signed-off-by: Bernhard Beschow <shentey@gmail.com>
> ---
>   include/hw/core/sysbus.h | 1 -
>   hw/core/sysbus.c         | 6 ------
>   2 files changed, 7 deletions(-)
> 
> diff --git a/include/hw/core/sysbus.h b/include/hw/core/sysbus.h
> index c0d18d9e00..f3c4259d29 100644
> --- a/include/hw/core/sysbus.h
> +++ b/include/hw/core/sysbus.h
> @@ -50,7 +50,6 @@ struct SysBusDeviceClass {
>        * omitted then. (This is not considered a fatal error.)
>        */
>       char *(*explicit_ofw_unit_address)(const SysBusDevice *dev);
> -    void (*connect_irq_notifier)(SysBusDevice *dev, qemu_irq irq);
>   };
>   
>   struct SysBusDevice {
> diff --git a/hw/core/sysbus.c b/hw/core/sysbus.c
> index 3adf2f2faf..3e1160ee92 100644
> --- a/hw/core/sysbus.c
> +++ b/hw/core/sysbus.c
> @@ -104,13 +104,7 @@ qemu_irq sysbus_get_connected_irq(const SysBusDevice *dev, int n)
>   
>   void sysbus_connect_irq(SysBusDevice *dev, int n, qemu_irq irq)
>   {
> -    SysBusDeviceClass *sbd = SYS_BUS_DEVICE_GET_CLASS(dev);
> -
>       qdev_connect_gpio_out_named(DEVICE(dev), SYSBUS_DEVICE_GPIO_IRQ, n, irq);
> -
> -    if (sbd->connect_irq_notifier) {
> -        sbd->connect_irq_notifier(dev, irq);
> -    }
>   }
>   
>   /* Check whether an MMIO region exists */

Reviewed-by: Cédric Le Goater <clg@redhat.com>

Thanks,

C.

Re: [PATCH 2/5] hw/arm/fsl-imx8mp: Fix parent of ocram memory region

[Markus Armbruster]

Bernhard Beschow <shentey@gmail.com> writes:

> Rather than having a NULL parent, let the containing SoC object be the
> parent. This cleans up the QOM composition tree a bit.
>
> Fixes: 1aaf3478684f ("hw/arm/fsl-imx8mp: Add on-chip RAM")
> cc: Gaurav Sharma <gaurav.sharma_7@nxp.com>
> Signed-off-by: Bernhard Beschow <shentey@gmail.com>
> ---
>  hw/arm/fsl-imx8mp.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/hw/arm/fsl-imx8mp.c b/hw/arm/fsl-imx8mp.c
> index 0119a18416..839c925e4a 100644
> --- a/hw/arm/fsl-imx8mp.c
> +++ b/hw/arm/fsl-imx8mp.c
> @@ -671,7 +671,7 @@ static void fsl_imx8mp_realize(DeviceState *dev, Error **errp)
>                      fsl_imx8mp_memmap[FSL_IMX8MP_PCIE_PHY1].addr);
>  
>      /* On-Chip RAM */
> -    if (!memory_region_init_ram(&s->ocram, NULL, "imx8mp.ocram",
> +    if (!memory_region_init_ram(&s->ocram, OBJECT(dev), "imx8mp.ocram",
>                                  fsl_imx8mp_memmap[FSL_IMX8MP_OCRAM].size,
>                                  errp)) {
>          return;

This moves /machine/unattached/imx8mp.ocram[0] to
/machine/soc/imx8mp.ocram[0], as expected.

Reviewed-by: Markus Armbruster <armbru@redhat.com>

Aside: not sure why [0].  Might just be something memory regions do.

Re: [PATCH 2/5] hw/arm/fsl-imx8mp: Fix parent of ocram memory region

[Peter Maydell]

On Sun, 8 Mar 2026 at 20:35, Bernhard Beschow <shentey@gmail.com> wrote:
>
> Rather than having a NULL parent, let the containing SoC object be the
> parent. This cleans up the QOM composition tree a bit.
>
> Fixes:  ("hw/arm/fsl-imx8mp: Add on-chip RAM")
> cc: Gaurav Sharma <gaurav.sharma_7@nxp.com>
> Signed-off-by: Bernhard Beschow <shentey@gmail.com>
> ---
>  hw/arm/fsl-imx8mp.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/hw/arm/fsl-imx8mp.c b/hw/arm/fsl-imx8mp.c
> index 0119a18416..839c925e4a 100644
> --- a/hw/arm/fsl-imx8mp.c
> +++ b/hw/arm/fsl-imx8mp.c
> @@ -671,7 +671,7 @@ static void fsl_imx8mp_realize(DeviceState *dev, Error **errp)
>                      fsl_imx8mp_memmap[FSL_IMX8MP_PCIE_PHY1].addr);
>
>      /* On-Chip RAM */
> -    if (!memory_region_init_ram(&s->ocram, NULL, "imx8mp.ocram",
> +    if (!memory_region_init_ram(&s->ocram, OBJECT(dev), "imx8mp.ocram",
>                                  fsl_imx8mp_memmap[FSL_IMX8MP_OCRAM].size,
>                                  errp)) {
>          return;


This is a migration break, because qemu_ram_set_idstr()
includes the qdev path of the owning device when it
creates the name of the ramblock for migrating it.
But since we haven't released anything with this change in
it and this board isn't versioned either, that's not a problem.

-- PMM