1. Patchew
  2. linux
  3. [PATCH 0/7] don't abuse task_struct.group_leader
  4. View patch

[PATCH 4/7] drm/amd: kill the outdated "Only the pthreads threading model is supported" checks

Oleg Nesterov posted 7 patches 1 week, 4 days ago
[PATCH 4/7] drm/amd: kill the outdated "Only the pthreads threading model is supported" checks
Posted by Oleg Nesterov 1 week, 4 days ago 
Nowaday task->group_leader->mm != task->mm is only possible if
a) task is not a group leader and b) task->group_leader->mm == NULL
because task->group_leader has already exited using sys_exit().

I don't think that drm/amd tries to detect/nack this case.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c   |  3 ---
 drivers/gpu/drm/amd/amdkfd/kfd_process.c | 10 ----------
 2 files changed, 13 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index a0f8ba382b9e..e44f158a11f0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -2551,9 +2551,6 @@ void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
 	vm->task_info->task.pid = current->pid;
 	get_task_comm(vm->task_info->task.comm, current);
 
-	if (current->group_leader->mm != current->mm)
-		return;
-
 	vm->task_info->tgid = current->tgid;
 	get_task_comm(vm->task_info->process_name, current->group_leader);
 }
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
index a085faac9fe1..f8ef18a3aa71 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
@@ -833,12 +833,6 @@ struct kfd_process *kfd_create_process(struct task_struct *thread)
 	if (!(thread->mm && mmget_not_zero(thread->mm)))
 		return ERR_PTR(-EINVAL);
 
-	/* Only the pthreads threading model is supported. */
-	if (thread->group_leader->mm != thread->mm) {
-		mmput(thread->mm);
-		return ERR_PTR(-EINVAL);
-	}
-
 	/* If the process just called exec(3), it is possible that the
 	 * cleanup of the kfd_process (following the release of the mm
 	 * of the old process image) is still in the cleanup work queue.
@@ -918,10 +912,6 @@ struct kfd_process *kfd_get_process(const struct task_struct *thread)
 	if (!thread->mm)
 		return ERR_PTR(-EINVAL);
 
-	/* Only the pthreads threading model is supported. */
-	if (thread->group_leader->mm != thread->mm)
-		return ERR_PTR(-EINVAL);
-
 	process = find_process(thread, false);
 	if (!process)
 		return ERR_PTR(-EINVAL);
-- 
2.52.0
Re: [PATCH 4/7] drm/amd: kill the outdated "Only the pthreads threading model is supported" checks
Posted by Felix Kuehling 1 week, 2 days ago 
On 2025-12-07 07:39, Oleg Nesterov wrote:
> Nowaday task->group_leader->mm != task->mm is only possible if
> a) task is not a group leader and b) task->group_leader->mm == NULL
> because task->group_leader has already exited using sys_exit().
>
> I don't think that drm/amd tries to detect/nack this case.
>
> Signed-off-by: Oleg Nesterov <oleg@redhat.com>

Acked-by: Felix Kuehling <felix.kuehling@amd.com>


> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c   |  3 ---
>   drivers/gpu/drm/amd/amdkfd/kfd_process.c | 10 ----------
>   2 files changed, 13 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> index a0f8ba382b9e..e44f158a11f0 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> @@ -2551,9 +2551,6 @@ void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
>   	vm->task_info->task.pid = current->pid;
>   	get_task_comm(vm->task_info->task.comm, current);
>   
> -	if (current->group_leader->mm != current->mm)
> -		return;
> -
>   	vm->task_info->tgid = current->tgid;
>   	get_task_comm(vm->task_info->process_name, current->group_leader);
>   }
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
> index a085faac9fe1..f8ef18a3aa71 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
> @@ -833,12 +833,6 @@ struct kfd_process *kfd_create_process(struct task_struct *thread)
>   	if (!(thread->mm && mmget_not_zero(thread->mm)))
>   		return ERR_PTR(-EINVAL);
>   
> -	/* Only the pthreads threading model is supported. */
> -	if (thread->group_leader->mm != thread->mm) {
> -		mmput(thread->mm);
> -		return ERR_PTR(-EINVAL);
> -	}
> -
>   	/* If the process just called exec(3), it is possible that the
>   	 * cleanup of the kfd_process (following the release of the mm
>   	 * of the old process image) is still in the cleanup work queue.
> @@ -918,10 +912,6 @@ struct kfd_process *kfd_get_process(const struct task_struct *thread)
>   	if (!thread->mm)
>   		return ERR_PTR(-EINVAL);
>   
> -	/* Only the pthreads threading model is supported. */
> -	if (thread->group_leader->mm != thread->mm)
> -		return ERR_PTR(-EINVAL);
> -
>   	process = find_process(thread, false);
>   	if (!process)
>   		return ERR_PTR(-EINVAL);
Re: [PATCH 4/7] drm/amd: kill the outdated "Only the pthreads threading model is supported" checks
Posted by Christian König 1 week, 4 days ago 
On 12/7/25 13:39, Oleg Nesterov wrote:
> Nowaday task->group_leader->mm != task->mm is only possible if
> a) task is not a group leader and b) task->group_leader->mm == NULL
> because task->group_leader has already exited using sys_exit().

Just for my understanding: That is because CLONE_THREAD can only be specified together with CLONE_SIGHAND and CLONE_VM, correct?

> 
> I don't think that drm/amd tries to detect/nack this case.
> 
> Signed-off-by: Oleg Nesterov <oleg@redhat.com>

Seems to make sense of hand.

Reviewed-by: Christian König <christian.koenig@amd.com>

Should we pick that one up or do you want to merge it upstream somehow?

Regards,
Christian.

> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c   |  3 ---
>  drivers/gpu/drm/amd/amdkfd/kfd_process.c | 10 ----------
>  2 files changed, 13 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> index a0f8ba382b9e..e44f158a11f0 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> @@ -2551,9 +2551,6 @@ void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
>  	vm->task_info->task.pid = current->pid;
>  	get_task_comm(vm->task_info->task.comm, current);
>  
> -	if (current->group_leader->mm != current->mm)
> -		return;
> -
>  	vm->task_info->tgid = current->tgid;
>  	get_task_comm(vm->task_info->process_name, current->group_leader);
>  }
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
> index a085faac9fe1..f8ef18a3aa71 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
> @@ -833,12 +833,6 @@ struct kfd_process *kfd_create_process(struct task_struct *thread)
>  	if (!(thread->mm && mmget_not_zero(thread->mm)))
>  		return ERR_PTR(-EINVAL);
>  
> -	/* Only the pthreads threading model is supported. */
> -	if (thread->group_leader->mm != thread->mm) {
> -		mmput(thread->mm);
> -		return ERR_PTR(-EINVAL);
> -	}
> -
>  	/* If the process just called exec(3), it is possible that the
>  	 * cleanup of the kfd_process (following the release of the mm
>  	 * of the old process image) is still in the cleanup work queue.
> @@ -918,10 +912,6 @@ struct kfd_process *kfd_get_process(const struct task_struct *thread)
>  	if (!thread->mm)
>  		return ERR_PTR(-EINVAL);
>  
> -	/* Only the pthreads threading model is supported. */
> -	if (thread->group_leader->mm != thread->mm)
> -		return ERR_PTR(-EINVAL);
> -
>  	process = find_process(thread, false);
>  	if (!process)
>  		return ERR_PTR(-EINVAL);
Re: [PATCH 4/7] drm/amd: kill the outdated "Only the pthreads threading model is supported" checks
Posted by Oleg Nesterov 1 week, 4 days ago 
On 12/08, Christian König wrote:
>
> On 12/7/25 13:39, Oleg Nesterov wrote:
> > Nowaday task->group_leader->mm != task->mm is only possible if
> > a) task is not a group leader and b) task->group_leader->mm == NULL
> > because task->group_leader has already exited using sys_exit().
>
> Just for my understanding: That is because CLONE_THREAD can only be
> specified together with CLONE_SIGHAND and CLONE_VM, correct?

Yes, copy copy_process() does

	if ((clone_flags & CLONE_THREAD) && !(clone_flags & CLONE_SIGHAND))
		return ERR_PTR(-EINVAL);

	if ((clone_flags & CLONE_SIGHAND) && !(clone_flags & CLONE_VM))
		return ERR_PTR(-EINVAL);

	...

	if (clone_flags & CLONE_THREAD) {
		p->group_leader = current->group_leader;
		p->tgid = current->tgid;
	} else {
		p->group_leader = p;
		p->tgid = p->pid;
	}
> Reviewed-by: Christian König <christian.koenig@amd.com>

Thanks!

> Should we pick that one up or do you want to merge it upstream somehow?

If you don't object, I would like to route this series via -mm tree.

See 0/7, I am going to send more (simple) tree-wide changes which depend
on this series.

Oleg.

> Regards,
> Christian.
>
> > ---
> >  drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c   |  3 ---
> >  drivers/gpu/drm/amd/amdkfd/kfd_process.c | 10 ----------
> >  2 files changed, 13 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> > index a0f8ba382b9e..e44f158a11f0 100644
> > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> > @@ -2551,9 +2551,6 @@ void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
> >  	vm->task_info->task.pid = current->pid;
> >  	get_task_comm(vm->task_info->task.comm, current);
> >
> > -	if (current->group_leader->mm != current->mm)
> > -		return;
> > -
> >  	vm->task_info->tgid = current->tgid;
> >  	get_task_comm(vm->task_info->process_name, current->group_leader);
> >  }
> > diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
> > index a085faac9fe1..f8ef18a3aa71 100644
> > --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c
> > +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
> > @@ -833,12 +833,6 @@ struct kfd_process *kfd_create_process(struct task_struct *thread)
> >  	if (!(thread->mm && mmget_not_zero(thread->mm)))
> >  		return ERR_PTR(-EINVAL);
> >
> > -	/* Only the pthreads threading model is supported. */
> > -	if (thread->group_leader->mm != thread->mm) {
> > -		mmput(thread->mm);
> > -		return ERR_PTR(-EINVAL);
> > -	}
> > -
> >  	/* If the process just called exec(3), it is possible that the
> >  	 * cleanup of the kfd_process (following the release of the mm
> >  	 * of the old process image) is still in the cleanup work queue.
> > @@ -918,10 +912,6 @@ struct kfd_process *kfd_get_process(const struct task_struct *thread)
> >  	if (!thread->mm)
> >  		return ERR_PTR(-EINVAL);
> >
> > -	/* Only the pthreads threading model is supported. */
> > -	if (thread->group_leader->mm != thread->mm)
> > -		return ERR_PTR(-EINVAL);
> > -
> >  	process = find_process(thread, false);
> >  	if (!process)
> >  		return ERR_PTR(-EINVAL);
>

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.