On Wed, Dec 11, 2024 at 12:41:43AM +0800, Cindy Lu wrote:
Restored the previous functions kthread_wakeup and kthread_stop.
nit: "Add back the previously removed"
Also add 2 new function pointer.
"Also add 2 new function pointers to wakeup and stop the workers."
The function vhost_worker_create
Will initializes this pointer based on the value of inherit_owner.
nit: s/Will/will
Signed-off-by: Cindy Lu <lulu@xxxxxxxxxx>
---
drivers/vhost/vhost.c | 84 +++++++++++++++++++++++++++++++++++--------
drivers/vhost/vhost.h | 3 ++
2 files changed, 73 insertions(+), 14 deletions(-)
diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index 812dfd218bc2..0175bbf4d8b3 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -721,14 +721,38 @@ static void vhost_workers_free(struct vhost_dev *dev)
xa_destroy(&dev->worker_xa);
}
+static int vhost_task_wakeup_fn(struct vhost_worker *worker)
+{
+ vhost_task_wake(worker->vtsk);
+ return 0;
+}
+
+static int vhost_kthread_wakeup_fn(struct vhost_worker *worker)
+{
+ return wake_up_process(worker->kthread_task);
+}
+
+static int vhost_task_stop_fn(struct vhost_worker *worker)
+{
+ vhost_task_stop(worker->vtsk);
+ return 0;
+}
+
+static int vhost_kthread_stop_fn(struct vhost_worker *worker)
+{
+ return kthread_stop(worker->kthread_task);
+}
+
static struct vhost_worker *vhost_worker_create(struct vhost_dev *dev)
{
struct vhost_worker *worker;
- struct vhost_task *vtsk;
+ struct vhost_task *vtsk = NULL;
+ struct task_struct *task = NULL;
char name[TASK_COMM_LEN];
int ret;
u32 id;
+ /* Allocate resources for the worker */
worker = kzalloc(sizeof(*worker), GFP_KERNEL_ACCOUNT);
if (!worker)
return NULL;
@@ -736,27 +760,59 @@ static struct vhost_worker
*vhost_worker_create(struct vhost_dev *dev)
worker->dev = dev;
snprintf(name, sizeof(name), "vhost-%d", current->pid);
- vtsk = vhost_task_create(vhost_run_work_list, vhost_worker_killed,
- worker, name);
- if (!vtsk)
- goto free_worker;
-
mutex_init(&worker->mutex);
init_llist_head(&worker->work_list);
worker->kcov_handle = kcov_common_handle();
- worker->vtsk = vtsk;
- vhost_task_start(vtsk);
+ if (dev->inherit_owner) {
+ /*
+ * If inherit_owner is true we use vhost_tasks to create
+ * the worker so all settings/limits like cgroups, NPROC,
+ * scheduler, etc are inherited from the owner. If false,
+ * we use kthreads and only attach to the same cgroups
+ * as the owner for compat with older kernels.
+ */
+ vtsk = vhost_task_create(vhost_run_work_list,
+ vhost_worker_killed, worker, name);
+ if (!vtsk)
+ goto free_worker;
+
+ worker->vtsk = vtsk;
+ worker->task_wakeup = vhost_task_wakeup_fn;
+ worker->task_stop = vhost_task_stop_fn;
+
+ vhost_task_start(vtsk);
+ ret = xa_alloc(&dev->worker_xa, &id, worker, xa_limit_32b,
+ GFP_KERNEL);
+ if (ret < 0)
+ goto stop_worker;
+ } else {
+ /* Create and start a kernel thread */
I would move here the comment included in the previous branch:
"If false we use kthreads and only attach to... "
Or move the entire comment block before the if.
+ task = kthread_create(vhost_run_work_kthread_list, worker,
+ "vhost-%d", current->pid);
+ if (IS_ERR(task)) {
+ ret = PTR_ERR(task);
+ goto free_worker;
+ }
+ worker->kthread_task = task;
+ worker->task_wakeup = vhost_kthread_wakeup_fn;
+ worker->task_stop = vhost_kthread_stop_fn;
- ret = xa_alloc(&dev->worker_xa, &id, worker, xa_limit_32b, GFP_KERNEL);
- if (ret < 0)
- goto stop_worker;
- worker->id = id;
+ wake_up_process(task);
+ ret = xa_alloc(&dev->worker_xa, &id, worker, xa_limit_32b,
+ GFP_KERNEL);
+ if (ret < 0)
+ goto stop_worker;
- return worker;
+ ret = vhost_attach_task_to_cgroups(worker);
+ if (ret)
+ goto stop_worker;
+ }
+ worker->id = id;
+ return worker;
stop_worker:
- vhost_task_stop(vtsk);
+ worker->task_stop(worker);
free_worker:
kfree(worker);
return NULL;
diff --git a/drivers/vhost/vhost.h b/drivers/vhost/vhost.h
index c650c4506c70..a7dc6e168753 100644
--- a/drivers/vhost/vhost.h
+++ b/drivers/vhost/vhost.h
@@ -27,6 +27,7 @@ struct vhost_work {
};
struct vhost_worker {
+ struct task_struct *kthread_task;
struct vhost_task *vtsk;
struct vhost_dev *dev;
/* Used to serialize device wide flushing with worker swapping. */
@@ -36,6 +37,8 @@ struct vhost_worker {
u32 id;
int attachment_cnt;
bool killed;
+ int (*task_wakeup)(struct vhost_worker *worker);
+ int (*task_stop)(struct vhost_worker *worker);
We never read the return values of these functions, so either mark them
both void or check their return value.
What about renaming in worker_wakeup and worker_stop?
Or even better since they are part of vhost_worker, just wakeup and
stop.
Thanks,
Stefano
};
/* Poll a file (eventfd or socket) */
--
2.45.0
