Re: [PATCH 1/3] drm/gem: Add callback for when handle count goes to 0

Re: [PATCH 1/3] drm/gem: Add callback for when handle count goes to 0

[Thorsten Leemhuis]

On 8/31/26 20:33, Alex Deucher wrote:
> From: Mukul Joshi <mukul.joshi@amd.com>
> 
> Add an optional callback for driver-specific cleanup when the GEM
> handle of an object is freed. This will be used by AMDGPU to enable
> freeing of memory exported to other nodes in a UALink pod once all
> user mode references are gone.
> 
> The callback is called outside the object_name_lock and before
> releasing the reference count on the GEM object

This showed up in -next yesterday and afaics broke build on arm64 and
x86_64 for me with the following error from rust:

"""
>> error[E0063]: missing field `handle_free` in initializer of `drm_gem_object_funcs`
>>    --> rust/kernel/drm/gem/mod.rs:265:58
>>     |
>> 265 |     const OBJECT_FUNCS: bindings::drm_gem_object_funcs = bindings::drm_gem_object_funcs {
>>     |                                                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ missing `handle_free`
>> 
>> error: aborting due to 1 previous error
>> 
>> For more information about this error, try `rustc --explain E0063`.
>> make[2]: *** [rust/Makefile:799: rust/kernel.o] Error 1
>> make[1]: *** [/builddir/build/BUILD/kernel-7.3.0-build/kernel-next-20260904/linux-7.3.0-0.0.next.20260904.221.vanilla.fc43.aarch64/Makefile:1442: prepare] Error 2
>> make: *** [Makefile:256: __sub-make] Error 2
"""
Full log:
https://download.copr.fedorainfracloud.org/results/@kernel-vanilla/next/fedora-43-aarch64/10951622-next-next-all/builder-live.log.gz

Reverting this change and 2/3 from this set fixed the problem for me.

Ciao, Thorsten
> Suggested-by: Christian König <christian.koenig@amd.com>
> Signed-off-by: Mukul Joshi <mukul.joshi@amd.com>
> Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c |  9 +++++++++
>  drivers/gpu/drm/drm_gem.c               |  5 ++++-
>  include/drm/drm_gem.h                   | 11 +++++++++++
>  3 files changed, 24 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
> index f754a4a3a1c22..0d579517c03ce 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
> @@ -386,6 +386,14 @@ static int amdgpu_gem_object_mmap(struct drm_gem_object *obj, struct vm_area_str
>  	return drm_gem_ttm_mmap(obj, vma);
>  }
>  
> +static void amdgpu_gem_object_handle_free(struct drm_gem_object *gobj)
> +{
> +	struct amdgpu_bo *aobj = gem_to_amdgpu_bo(gobj);
> +
> +	amdgpu_ualink_revoke_exported_memory(aobj);
> +
> +}
> +
>  const struct drm_gem_object_funcs amdgpu_gem_object_funcs = {
>  	.free = amdgpu_gem_object_free,
>  	.open = amdgpu_gem_object_open,
> @@ -395,6 +403,7 @@ const struct drm_gem_object_funcs amdgpu_gem_object_funcs = {
>  	.vunmap = drm_gem_ttm_vunmap,
>  	.mmap = amdgpu_gem_object_mmap,
>  	.vm_ops = &amdgpu_gem_vm_ops,
> +	.handle_free = amdgpu_gem_object_handle_free
>  };
>  
>  static bool amdgpu_gem_are_domains_valid(u32 domains)
> diff --git a/drivers/gpu/drm/drm_gem.c b/drivers/gpu/drm/drm_gem.c
> index e3ed684ddcf29..6a86bd2a0343e 100644
> --- a/drivers/gpu/drm/drm_gem.c
> +++ b/drivers/gpu/drm/drm_gem.c
> @@ -354,8 +354,11 @@ void drm_gem_object_handle_put_unlocked(struct drm_gem_object *obj)
>  	}
>  	mutex_unlock(&dev->object_name_lock);
>  
> -	if (final)
> +	if (final) {
> +		if (obj->funcs->handle_free)
> +			obj->funcs->handle_free(obj);
>  		drm_gem_object_put(obj);
> +	}
>  }
>  
>  /*
> diff --git a/include/drm/drm_gem.h b/include/drm/drm_gem.h
> index 8a704f6a65c15..95d8ae6f85df7 100644
> --- a/include/drm/drm_gem.h
> +++ b/include/drm/drm_gem.h
> @@ -227,6 +227,17 @@ struct drm_gem_object_funcs {
>  	 */
>  	size_t (*rss)(struct drm_gem_object *obj);
>  
> +	/**
> +	 * @handle_free:
> +	 *
> +	 * This callback is called when the GEM handle count goes down to 0.
> +	 * It is currently used by AMDGPU driver to release their exported BO
> +	 * handles.
> +	 *
> +	 * This callback is optional.
> +	 */
> +	void (*handle_free)(struct drm_gem_object *obj);
> +
>  	/**
>  	 * @vm_ops:
>  	 *

Re: [PATCH 1/3] drm/gem: Add callback for when handle count goes to 0

[Manuel Ebner]

On Sat, 2026-09-05 at 07:36 +0200, Thorsten Leemhuis wrote:
> On 8/31/26 20:33, Alex Deucher wrote:
> > From: Mukul Joshi <mukul.joshi@amd.com>
> > ...
> ...
> 
> > > error[E0063]: missing field `handle_free` in initializer of `drm_gem_object_funcs`
> > >    --> rust/kernel/drm/gem/mod.rs:265:58
> > >     |
> > > 265 |     const OBJECT_FUNCS: bindings::drm_gem_object_funcs = bindings::drm_gem_object_funcs {
> > >     |                                                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ missing `handle_free`
> > > 
> > > error: aborting due to 1 previous error
> > > 
> > > For more information about this error, try `rustc --explain E0063`.
> > > make[2]: *** [rust/Makefile:799: rust/kernel.o] Error 1
> > > make[1]: *** [/builddir/build/BUILD/kernel-7.3.0-build/kernel-next-20260904/linux-7.3.0-0.0.next.20260904.221.vanilla.fc43.aarch64/Makefile:1442: prepare] Error 2
> > > make: *** [Makefile:256: __sub-make] Error 2
> """
> Full log:
> https://download.copr.fedorainfracloud.org/results/@kernel-vanilla/next/fedora-43-aarch64/10951622-next-next-all/builder-live.log.gz
> 
> Reverting this change and 2/3 from this set fixed the problem for me.

Verified.

Thanks
 Manuel
> 
> Ciao, Thorsten
> > ...

Re: [PATCH 1/3] drm/gem: Add callback for when handle count goes to 0

[Klara Modin]

On 2026-09-05 11:31:41 +0200, Manuel Ebner wrote:
> On Sat, 2026-09-05 at 07:36 +0200, Thorsten Leemhuis wrote:
> > On 8/31/26 20:33, Alex Deucher wrote:
> > > From: Mukul Joshi <mukul.joshi@amd.com>
> > > ...
> > ...
> > 
> > > > error[E0063]: missing field `handle_free` in initializer of `drm_gem_object_funcs`
> > > >    --> rust/kernel/drm/gem/mod.rs:265:58
> > > >     |
> > > > 265 |     const OBJECT_FUNCS: bindings::drm_gem_object_funcs = bindings::drm_gem_object_funcs {
> > > >     |                                                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ missing `handle_free`
> > > > 
> > > > error: aborting due to 1 previous error
> > > > 
> > > > For more information about this error, try `rustc --explain E0063`.
> > > > make[2]: *** [rust/Makefile:799: rust/kernel.o] Error 1
> > > > make[1]: *** [/builddir/build/BUILD/kernel-7.3.0-build/kernel-next-20260904/linux-7.3.0-0.0.next.20260904.221.vanilla.fc43.aarch64/Makefile:1442: prepare] Error 2
> > > > make: *** [Makefile:256: __sub-make] Error 2
> > """
> > Full log:
> > https://download.copr.fedorainfracloud.org/results/@kernel-vanilla/next/fedora-43-aarch64/10951622-next-next-all/builder-live.log.gz
> > 
> > Reverting this change and 2/3 from this set fixed the problem for me.

That leaves only the documentation without the feature so might as well
drop 3/3 too in that case?

Otherwise, the following also fixes it for me:

diff --git a/rust/kernel/drm/gem/mod.rs b/rust/kernel/drm/gem/mod.rs
index e1ebad77ebe2..847e0ee38863 100644
--- a/rust/kernel/drm/gem/mod.rs
+++ b/rust/kernel/drm/gem/mod.rs
@@ -278,6 +278,7 @@ impl<T: DriverObject, Ctx: DeviceContext> Object<T, Ctx> {
         vm_ops: core::ptr::null_mut(),
         evict: None,
         rss: None,
+        handle_free: None,
     };
 
     /// Returns the `Device` that owns this GEM object.

> 
> Verified.
> 
> Thanks
>  Manuel
> > 
> > Ciao, Thorsten
> > > ...

Regards,
Klara Modin

Re: [PATCH 1/3] drm/gem: Add callback for when handle count goes to 0

[Gary Guo]

On Sat Sep 5, 2026 at 11:25 PM BST, Klara Modin wrote:
> On 2026-09-05 11:31:41 +0200, Manuel Ebner wrote:
>> On Sat, 2026-09-05 at 07:36 +0200, Thorsten Leemhuis wrote:
>> > On 8/31/26 20:33, Alex Deucher wrote:
>> > > From: Mukul Joshi <mukul.joshi@amd.com>
>> > > ...
>> > ...
>> > 
>> > > > error[E0063]: missing field `handle_free` in initializer of `drm_gem_object_funcs`
>> > > >    --> rust/kernel/drm/gem/mod.rs:265:58
>> > > >     |
>> > > > 265 |     const OBJECT_FUNCS: bindings::drm_gem_object_funcs = bindings::drm_gem_object_funcs {
>> > > >     |                                                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ missing `handle_free`
>> > > > 
>> > > > error: aborting due to 1 previous error
>> > > > 
>> > > > For more information about this error, try `rustc --explain E0063`.
>> > > > make[2]: *** [rust/Makefile:799: rust/kernel.o] Error 1
>> > > > make[1]: *** [/builddir/build/BUILD/kernel-7.3.0-build/kernel-next-20260904/linux-7.3.0-0.0.next.20260904.221.vanilla.fc43.aarch64/Makefile:1442: prepare] Error 2
>> > > > make: *** [Makefile:256: __sub-make] Error 2
>> > """
>> > Full log:
>> > https://download.copr.fedorainfracloud.org/results/@kernel-vanilla/next/fedora-43-aarch64/10951622-next-next-all/builder-live.log.gz
>> > 
>> > Reverting this change and 2/3 from this set fixed the problem for me.
>
> That leaves only the documentation without the feature so might as well
> drop 3/3 too in that case?
>
> Otherwise, the following also fixes it for me:
>
> diff --git a/rust/kernel/drm/gem/mod.rs b/rust/kernel/drm/gem/mod.rs
> index e1ebad77ebe2..847e0ee38863 100644
> --- a/rust/kernel/drm/gem/mod.rs
> +++ b/rust/kernel/drm/gem/mod.rs
> @@ -278,6 +278,7 @@ impl<T: DriverObject, Ctx: DeviceContext> Object<T, Ctx> {
>          vm_ops: core::ptr::null_mut(),
>          evict: None,
>          rss: None,
> +        handle_free: None,

IMO this should use

    ..pin_init::zeroed()

to set all other fields as zero rather than explicitly listing all of them.

Best,
Gary

>      };
>  
>      /// Returns the `Device` that owns this GEM object.
>
>> 
>> Verified.
>> 
>> Thanks
>>  Manuel
>> > 
>> > Ciao, Thorsten
>> > > ...
>
> Regards,
> Klara Modin

Re: [PATCH 1/3] drm/gem: Add callback for when handle count goes to 0

[Thorsten Leemhuis]

On 9/6/26 00:25, Klara Modin wrote:
> On 2026-09-05 11:31:41 +0200, Manuel Ebner wrote:
>> On Sat, 2026-09-05 at 07:36 +0200, Thorsten Leemhuis wrote:
>>> On 8/31/26 20:33, Alex Deucher wrote:
>>>> From: Mukul Joshi <mukul.joshi@amd.com>
>
>>>>> error[E0063]: missing field `handle_free` in initializer of `drm_gem_object_funcs`
>>>>>    --> rust/kernel/drm/gem/mod.rs:265:58
>>>>>     |
>>>>> 265 |     const OBJECT_FUNCS: bindings::drm_gem_object_funcs = bindings::drm_gem_object_funcs {
>>>>>     |                                                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ missing `handle_free`
>>>>>
>>>>> error: aborting due to 1 previous error
>>>>>
>>>>> For more information about this error, try `rustc --explain E0063`.
>>>>> make[2]: *** [rust/Makefile:799: rust/kernel.o] Error 1
>>>>> make[1]: *** [/builddir/build/BUILD/kernel-7.3.0-build/kernel-next-20260904/linux-7.3.0-0.0.next.20260904.221.vanilla.fc43.aarch64/Makefile:1442: prepare] Error 2
>>>>> make: *** [Makefile:256: __sub-make] Error 2
>>> Full log:
>>> https://download.copr.fedorainfracloud.org/results/@kernel-vanilla/next/fedora-43-aarch64/10951622-next-next-all/builder-live.log.gz
>>>
>>> Reverting this change and 2/3 from this set fixed the problem for me.
> 
> That leaves only the documentation without the feature so might as well
> drop 3/3 too in that case?

My intention was just to confirm that these patches are the problem, not
suggest this as a solution, so I didn't bother about the docs. :-D
> Otherwise, the following also fixes it for me:

Thx, for me, too.

Ciao, Thorsten
> diff --git a/rust/kernel/drm/gem/mod.rs b/rust/kernel/drm/gem/mod.rs
> index e1ebad77ebe2..847e0ee38863 100644
> --- a/rust/kernel/drm/gem/mod.rs
> +++ b/rust/kernel/drm/gem/mod.rs
> @@ -278,6 +278,7 @@ impl<T: DriverObject, Ctx: DeviceContext> Object<T, Ctx> {
>          vm_ops: core::ptr::null_mut(),
>          evict: None,
>          rss: None,
> +        handle_free: None,
>      };
>  
>      /// Returns the `Device` that owns this GEM object.

Re: [PATCH 1/3] drm/gem: Add callback for when handle count goes to 0

[Alex Deucher]

On Mon, Sep 7, 2026 at 3:25 AM Klara Modin <klarasmodin@gmail.com> wrote:
>
> On 2026-09-05 11:31:41 +0200, Manuel Ebner wrote:
> > On Sat, 2026-09-05 at 07:36 +0200, Thorsten Leemhuis wrote:
> > > On 8/31/26 20:33, Alex Deucher wrote:
> > > > From: Mukul Joshi <mukul.joshi@amd.com>
> > > > ...
> > > ...
> > >
> > > > > error[E0063]: missing field `handle_free` in initializer of `drm_gem_object_funcs`
> > > > >    --> rust/kernel/drm/gem/mod.rs:265:58
> > > > >     |
> > > > > 265 |     const OBJECT_FUNCS: bindings::drm_gem_object_funcs = bindings::drm_gem_object_funcs {
> > > > >     |                                                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ missing `handle_free`
> > > > >
> > > > > error: aborting due to 1 previous error
> > > > >
> > > > > For more information about this error, try `rustc --explain E0063`.
> > > > > make[2]: *** [rust/Makefile:799: rust/kernel.o] Error 1
> > > > > make[1]: *** [/builddir/build/BUILD/kernel-7.3.0-build/kernel-next-20260904/linux-7.3.0-0.0.next.20260904.221.vanilla.fc43.aarch64/Makefile:1442: prepare] Error 2
> > > > > make: *** [Makefile:256: __sub-make] Error 2
> > > """
> > > Full log:
> > > https://download.copr.fedorainfracloud.org/results/@kernel-vanilla/next/fedora-43-aarch64/10951622-next-next-all/builder-live.log.gz
> > >
> > > Reverting this change and 2/3 from this set fixed the problem for me.
>
> That leaves only the documentation without the feature so might as well
> drop 3/3 too in that case?
>
> Otherwise, the following also fixes it for me:
>
> diff --git a/rust/kernel/drm/gem/mod.rs b/rust/kernel/drm/gem/mod.rs
> index e1ebad77ebe2..847e0ee38863 100644
> --- a/rust/kernel/drm/gem/mod.rs
> +++ b/rust/kernel/drm/gem/mod.rs
> @@ -278,6 +278,7 @@ impl<T: DriverObject, Ctx: DeviceContext> Object<T, Ctx> {
>          vm_ops: core::ptr::null_mut(),
>          evict: None,
>          rss: None,
> +        handle_free: None,
>      };
>
>      /// Returns the `Device` that owns this GEM object.
>

Can you send this as a proper patch?

Thanks,

Alex

> >
> > Verified.
> >
> > Thanks
> >  Manuel
> > >
> > > Ciao, Thorsten
> > > > ...
>
> Regards,
> Klara Modin

Re: [PATCH 1/3] drm/gem: Add callback for when handle count goes to 0

[Klara Modin]

On 2026-09-08 12:03:57 -0400, Alex Deucher wrote:
> On Mon, Sep 7, 2026 at 3:25 AM Klara Modin <klarasmodin@gmail.com> wrote:
> >
> > On 2026-09-05 11:31:41 +0200, Manuel Ebner wrote:
> > > On Sat, 2026-09-05 at 07:36 +0200, Thorsten Leemhuis wrote:
> > > > On 8/31/26 20:33, Alex Deucher wrote:
> > > > > From: Mukul Joshi <mukul.joshi@amd.com>
> > > > > ...
> > > > ...
> > > >
> > > > > > error[E0063]: missing field `handle_free` in initializer of `drm_gem_object_funcs`
> > > > > >    --> rust/kernel/drm/gem/mod.rs:265:58
> > > > > >     |
> > > > > > 265 |     const OBJECT_FUNCS: bindings::drm_gem_object_funcs = bindings::drm_gem_object_funcs {
> > > > > >     |                                                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ missing `handle_free`
> > > > > >
> > > > > > error: aborting due to 1 previous error
> > > > > >
> > > > > > For more information about this error, try `rustc --explain E0063`.
> > > > > > make[2]: *** [rust/Makefile:799: rust/kernel.o] Error 1
> > > > > > make[1]: *** [/builddir/build/BUILD/kernel-7.3.0-build/kernel-next-20260904/linux-7.3.0-0.0.next.20260904.221.vanilla.fc43.aarch64/Makefile:1442: prepare] Error 2
> > > > > > make: *** [Makefile:256: __sub-make] Error 2
> > > > """
> > > > Full log:
> > > > https://download.copr.fedorainfracloud.org/results/@kernel-vanilla/next/fedora-43-aarch64/10951622-next-next-all/builder-live.log.gz
> > > >
> > > > Reverting this change and 2/3 from this set fixed the problem for me.
> >
> > That leaves only the documentation without the feature so might as well
> > drop 3/3 too in that case?
> >
> > Otherwise, the following also fixes it for me:
> >
> > diff --git a/rust/kernel/drm/gem/mod.rs b/rust/kernel/drm/gem/mod.rs
> > index e1ebad77ebe2..847e0ee38863 100644
> > --- a/rust/kernel/drm/gem/mod.rs
> > +++ b/rust/kernel/drm/gem/mod.rs
> > @@ -278,6 +278,7 @@ impl<T: DriverObject, Ctx: DeviceContext> Object<T, Ctx> {
> >          vm_ops: core::ptr::null_mut(),
> >          evict: None,
> >          rss: None,
> > +        handle_free: None,
> >      };
> >
> >      /// Returns the `Device` that owns this GEM object.
> >
> 
> Can you send this as a proper patch?

Here: https://lore.kernel.org/lkml/20260908175427.47207-1-klarasmodin@gmail.com

I liked Gary's suggestion so went with that instead.

> 
> Thanks,
> 
> Alex

Regards,
Klara Modin

> 
> > >
> > > Verified.
> > >
> > > Thanks
> > >  Manuel
> > > >
> > > > Ciao, Thorsten
> > > > > ...
> >
> > Regards,
> > Klara Modin