clear-pkgs-linux-iot-lts2018/0547-vhm-init-client-kthread_exit-true.patch

From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Li Zhijian <lizhijian@cn.fujitsu.com>
Date: Wed, 22 Aug 2018 16:12:20 +0000
Subject: [PATCH] vhm: init client->kthread_exit true

Previously, there is a deadlock at below case
- acrn-dm create gvt instance successfully
- acrn-dm open uos image failed(wrong image path), the acrn-dm does some cleanup,
  like destroy gvt instance
then acrn-dm stucks.

when destroying gvt instance, it waits client->kthread_exit to be true
while client->kthread_exit is set to be 0 at initializing and the thread
is not created/started actually.

V3: add vhm_create_kthread as a condition (Yakui)

Signed-off-by: Li Zhijian <lizhijian@cn.fujitsu.com>
Reviewed-by: Jason Chen CJ <jason.cj.chen@intel.com>
Reviewed-by: Zhao Yakui <yakui.zhao@intel.com>
---
 drivers/vhm/vhm_ioreq.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/vhm/vhm_ioreq.c b/drivers/vhm/vhm_ioreq.c
index da61069b8d01..0bcb0e053947 100644
--- a/drivers/vhm/vhm_ioreq.c
+++ b/drivers/vhm/vhm_ioreq.c
@@ -149,6 +149,7 @@ static int alloc_client(void)
 	if (!client)
 		return -ENOMEM;
 	client->id = i;
+	client->kthread_exit = true;
 	clients[i] = client;
 
 	return i;
@@ -266,7 +267,7 @@ static void acrn_ioreq_destroy_client_pervm(struct ioreq_client *client,
 	/* the client thread will mark kthread_exit flag as true before exit,
 	 * so wait for it exited.
 	 */
-	while (!client->kthread_exit)
+	while (client->vhm_create_kthread && !client->kthread_exit)
 		msleep(10);
 
 	spin_lock_irqsave(&client->range_lock, flags);
@@ -533,7 +534,9 @@ int acrn_ioreq_attach_client(int client_id, bool check_kthread_stop)
 					"for client %s\n", client->name);
 			return -ENOMEM;
 		}
+		client->kthread_exit = false;
 	} else {
+		client->kthread_exit = false;
 		might_sleep();
 
 		if (check_kthread_stop) {
-- 
https://clearlinux.org