[PATCH] find_task_by_pid() needs tasklist_lock
Andrew Morton [Sat, 25 Mar 2006 11:07:48 +0000 (03:07 -0800)]
A couple of places are forgetting to take it.

The kswapd case is probably unimportant.  keventd_create_kthread() was racy.

The whole thing is a bit flakey: you start a kernel thread, get its pid from
kernel_thread() then look up its task_struct.

a) It assumes that pid recycling takes a "long" time.

b) We get a task_struct but no reference was taken on it.  The owner of the
   kswapd and kthread task_struct*'s must assume that the new thread won't
   exit unexpectedly.  Because if it does, they're left holding dead memory
   and any attempt to control or stop that task will crash.

Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

kernel/kthread.c
mm/vmscan.c

index 6a53738..c5f3c66 100644 (file)
@@ -115,7 +115,9 @@ static void keventd_create_kthread(void *_create)
                create->result = ERR_PTR(pid);
        } else {
                wait_for_completion(&create->started);
+               read_lock(&tasklist_lock);
                create->result = find_task_by_pid(pid);
+               read_unlock(&tasklist_lock);
        }
        complete(&create->done);
 }
index fd572bb..78865c8 100644 (file)
@@ -1356,7 +1356,9 @@ static int __init kswapd_init(void)
 
                pid = kernel_thread(kswapd, pgdat, CLONE_KERNEL);
                BUG_ON(pid < 0);
+               read_lock(&tasklist_lock);
                pgdat->kswapd = find_task_by_pid(pid);
+               read_unlock(&tasklist_lock);
        }
        total_memory = nr_free_pagecache_pages();
        hotcpu_notifier(cpu_callback, 0);