Skip to contentSkip to conversation

don't abuse task_struct.group_leader

PATCH In review Oleg Nesterov 3 participants 11 messages Jan 25 – Jan 26

Subsystems (10)
lore
Share & tools
lore
#0 lore raw

v2 cover don't abuse task_struct.group_leader

Andrew, can you take these simple cleanups?

The patches do not depend on each other, this series just removes the usage of ->group_leader when it is "obviously unnecessary".

I am going to move ->group_leader from task_struct to signal_struct or at least add the new task_group_leader() helper. So I will send more tree-wide changes on top of this series.

Link to V1: https://lore.kernel.org/all/[email protected]/

V2 doesn't differ, I only added the acks I got (thanks!).

7/7 was not reviewed, but looks really trivial.

Oleg.
---

 drivers/android/binder.c                         |  9 ++++-----
 drivers/android/binder_alloc.c                   |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c           |  5 +----
 drivers/gpu/drm/amd/amdkfd/kfd_process.c         | 10 ----------
 drivers/gpu/drm/panfrost/panfrost_gem.c          |  2 +-
 drivers/gpu/drm/panthor/panthor_gem.c            |  2 +-
 drivers/infiniband/core/umem_odp.c               |  4 ++--
 net/core/netclassid_cgroup.c                     |  2 +-
 9 files changed, 12 insertions(+), 26 deletions(-)
#1 lore raw

PATCH v2 1/7 android/binder: don't abuse current->group_leader

Cleanup and preparation to simplify the next changes.

- Use current->tgid instead of current->group_leader->pid

- Use the value returned by get_task_struct() to initialize proc->tsk

---
 drivers/android/binder.c       | 7 +++----
 drivers/android/binder_alloc.c | 2 +-
 2 files changed, 4 insertions(+), 5 deletions(-)
drivers/android/binder.c +3 -4
index 535fc881c8da..dea701daabb0 100644
@@ -6046,7 +6046,7 @@ static int binder_open(struct inode *nodp, struct file *filp)
 	bool existing_pid = false;
 
 	binder_debug(BINDER_DEBUG_OPEN_CLOSE, "%s: %d:%d\n", __func__,
-		     current->group_leader->pid, current->pid);
+		     current->tgid, current->pid);
 
 	proc = kzalloc(sizeof(*proc), GFP_KERNEL);
 	if (proc == NULL)
@@ -6055,8 +6055,8 @@ static int binder_open(struct inode *nodp, struct file *filp)
 	dbitmap_init(&proc->dmap);
 	spin_lock_init(&proc->inner_lock);
 	spin_lock_init(&proc->outer_lock);
-	get_task_struct(current->group_leader);
-	proc->tsk = current->group_leader;
+	proc->tsk = get_task_struct(current->group_leader);
+	proc->pid = current->tgid;
 	proc->cred = get_cred(filp->f_cred);
 	INIT_LIST_HEAD(&proc->todo);
 	init_waitqueue_head(&proc->freeze_wait);
@@ -6075,7 +6075,6 @@ static int binder_open(struct inode *nodp, struct file *filp)
 	binder_alloc_init(&proc->alloc);
 
 	binder_stats_created(BINDER_STAT_PROC);
-	proc->pid = current->group_leader->pid;
 	INIT_LIST_HEAD(&proc->delivered_death);
 	INIT_LIST_HEAD(&proc->delivered_freeze);
 	INIT_LIST_HEAD(&proc->waiting_threads);
drivers/android/binder_alloc.c +1 -1
index 979c96b74cad..145ed5f14cdb 100644
@@ -1233,7 +1233,7 @@ static struct shrinker *binder_shrinker;
 VISIBLE_IF_KUNIT void __binder_alloc_init(struct binder_alloc *alloc,
 					  struct list_lru *freelist)
 {
-	alloc->pid = current->group_leader->pid;
+	alloc->pid = current->tgid;
 	alloc->mm = current->mm;
 	mmgrab(alloc->mm);
 	mutex_init(&alloc->mutex);
Signature · 1 line
2.52.0
#2 lore raw

PATCH v2 2/7 android/binder: use same_thread_group(proc->tsk, current) in binder_mmap()

In reply to Oleg Nesterov

With or without this change the checked condition can be falsely true if proc->tsk execs, but this is fine: binder_alloc_mmap_handler() checks vma->vm_mm == alloc->mm.

---
 drivers/android/binder.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/android/binder.c +1 -1
index dea701daabb0..b3b73303f84d 100644
@@ -6015,7 +6015,7 @@ static int binder_mmap(struct file *filp, struct vm_area_struct *vma)
 {
 	struct binder_proc *proc = filp->private_data;
 
-	if (proc->tsk != current->group_leader)
+	if (!same_thread_group(proc->tsk, current))
 		return -EINVAL;
 
 	binder_debug(BINDER_DEBUG_OPEN_CLOSE,
Signature · 1 line
2.52.0
#3 lore raw

PATCH v2 3/7 drm/amdgpu: don't abuse current->group_leader

In reply to Oleg Nesterov

Cleanup and preparation to simplify the next changes.

- Use current->tgid instead of current->group_leader->pid

- Use get_task_pid(current, PIDTYPE_TGID) instead of get_task_pid(current->group_leader, PIDTYPE_PID)

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c           | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +1 -1
index b1c24c8fa686..df22b54ba346 100644
@@ -1421,7 +1421,7 @@ static int init_kfd_vm(struct amdgpu_vm *vm, void **process_info,
 			goto create_evict_fence_fail;
 		}
 
-		info->pid = get_task_pid(current->group_leader, PIDTYPE_PID);
+		info->pid = get_task_pid(current, PIDTYPE_TGID);
 		INIT_DELAYED_WORK(&info->restore_userptr_work,
 				  amdgpu_amdkfd_restore_userptr_worker);
 
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +1 -1
index a67285118c37..a0f8ba382b9e 100644
@@ -2554,7 +2554,7 @@ void amdgpu_vm_set_task_info(struct amdgpu_vm *vm)
 	if (current->group_leader->mm != current->mm)
 		return;
 
-	vm->task_info->tgid = current->group_leader->pid;
+	vm->task_info->tgid = current->tgid;
 	get_task_comm(vm->task_info->process_name, current->group_leader);
 }
 
Signature · 1 line
2.52.0
#4 lore raw

PATCH v2 4/7 drm/amd: kill the outdated "Only the pthreads threading model is supported" checks

In reply to Oleg Nesterov

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.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c   |  3 ---
 drivers/gpu/drm/amd/amdkfd/kfd_process.c | 10 ----------
 2 files changed, 13 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +0 -3
index a0f8ba382b9e..e44f158a11f0 100644
@@ -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);
 }
drivers/gpu/drm/amd/amdkfd/kfd_process.c +0 -10
index a085faac9fe1..f8ef18a3aa71 100644
@@ -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);
Signature · 1 line
2.52.0
#5 lore raw

PATCH v2 5/7 drm/pan*: don't abuse current->group_leader

In reply to Oleg Nesterov

Cleanup and preparation to simplify the next changes.

Use current->tgid instead of current->group_leader->pid.

---
 drivers/gpu/drm/panfrost/panfrost_gem.c | 2 +-
 drivers/gpu/drm/panthor/panthor_gem.c   | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/panfrost/panfrost_gem.c +1 -1
index 8041b65c6609..1ff1f2c8b726 100644
@@ -17,7 +17,7 @@
 static void panfrost_gem_debugfs_bo_add(struct panfrost_device *pfdev,
 					struct panfrost_gem_object *bo)
 {
-	bo->debugfs.creator.tgid = current->group_leader->pid;
+	bo->debugfs.creator.tgid = current->tgid;
 	get_task_comm(bo->debugfs.creator.process_name, current->group_leader);
 
 	mutex_lock(&pfdev->debugfs.gems_lock);
drivers/gpu/drm/panthor/panthor_gem.c +1 -1
index fbde78db270a..29cc57efc4b9 100644
@@ -27,7 +27,7 @@ static void panthor_gem_debugfs_bo_add(struct panthor_gem_object *bo)
 	struct panthor_device *ptdev = container_of(bo->base.base.dev,
 						    struct panthor_device, base);
 
-	bo->debugfs.creator.tgid = current->group_leader->pid;
+	bo->debugfs.creator.tgid = current->tgid;
 	get_task_comm(bo->debugfs.creator.process_name, current->group_leader);
 
 	mutex_lock(&ptdev->gems.lock);
Signature · 1 line
2.52.0
#6 lore raw

PATCH v2 6/7 RDMA/umem: don't abuse current->group_leader

In reply to Oleg Nesterov

Cleanup and preparation to simplify the next changes.

Use current->tgid instead of current->group_leader->pid.

---
 drivers/infiniband/core/umem_odp.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/infiniband/core/umem_odp.c +2 -2
index 572a91a62a7b..32267258a19c 100644
@@ -149,7 +149,7 @@ struct ib_umem_odp *ib_umem_odp_alloc_implicit(struct ib_device *device,
 	umem->owning_mm = current->mm;
 	umem_odp->page_shift = PAGE_SHIFT;
 
-	umem_odp->tgid = get_task_pid(current->group_leader, PIDTYPE_PID);
+	umem_odp->tgid = get_task_pid(current, PIDTYPE_TGID);
 	ib_init_umem_implicit_odp(umem_odp);
 	return umem_odp;
 }
@@ -258,7 +258,7 @@ struct ib_umem_odp *ib_umem_odp_get(struct ib_device *device,
 		umem_odp->page_shift = HPAGE_SHIFT;
 #endif
 
-	umem_odp->tgid = get_task_pid(current->group_leader, PIDTYPE_PID);
+	umem_odp->tgid = get_task_pid(current, PIDTYPE_TGID);
 	ret = ib_init_umem_odp(umem_odp, ops);
 	if (ret)
 		goto err_put_pid;
Signature · 1 line
2.52.0
#7 lore raw

PATCH v2 7/7 netclassid: use thread_group_leader(p) in update_classid_task()

In reply to Oleg Nesterov

Cleanup and preparation to simplify the next changes.

---
 net/core/netclassid_cgroup.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/core/netclassid_cgroup.c +1 -1
index dff66d8fb325..db9a5354f9de 100644
@@ -93,7 +93,7 @@ static void update_classid_task(struct task_struct *p, u32 classid)
 	/* Only update the leader task, when many threads in this task,
 	 * so it can avoid the useless traversal.
 	 */
-	if (p != p->group_leader)
+	if (!thread_group_leader(p))
 		return;
 
 	do {
Signature · 1 line
2.52.0
maintainer Acked-by on v2 7/7
#8 lore raw

Re: [PATCH v2 7/7] netclassid: use thread_group_leader(p) in update_classid_task()

Quoted 1 line · On Sun, 25 Jan 2026 17:08:00 +0100 Oleg Nesterov wrote:

Cleanup and preparation to simplify the next changes.

#9 lore raw

Re: [PATCH v2 7/7] netclassid: use thread_group_leader(p) in update_classid_task()

In reply to Oleg Nesterov
Quoted 2 lines · On Sun, 25 Jan 2026 17:08:00 +0100 Oleg Nesterov <[email protected]> wrote:

Cleanup and preparation to simplify the next changes.

"next changes". This is the final patch?

Quoted 13 lines
net/core/netclassid_cgroup.c +1 -1
@@ -93,7 +93,7 @@ static void update_classid_task(struct task_struct *p, u32 classid)
 	/* Only update the leader task, when many threads in this task,
 	 * so it can avoid the useless traversal.
 	 */
-	if (p != p->group_leader)
+	if (!thread_group_leader(p))
 		return;
 
 	do {
Signature · 1 line
2.52.0
#10 lore raw

Re: [PATCH v2 7/7] netclassid: use thread_group_leader(p) in update_classid_task()

Quoted 6 lines · On Mon, 26 Jan 2026 12:50:48 -0800 Andrew Morton <[email protected]>…
Quoted 2 lines · On Sun, 25 Jan 2026 17:08:00 +0100 Oleg Nesterov <[email protected]> wrote:

Cleanup and preparation to simplify the next changes.

"next changes". This is the final patch?

Ah, sorry, [0/n] explains. I'll make that "planned future changes".