Attachment 308287 Details for Bug 448934 – The patch itself

[patch] The patch itself

bz435280_tasklist_lock.patch (text/plain), 1.46 KB, created by Vitaly Mayatskikh on 2008-06-03 20:03:04 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Vitaly Mayatskikh

Created: 2008-06-03 20:03:04 UTC

Size: 1.46 KB

patch

obsolete

>diff -up linux-2.6.9/kernel/sys.c.orig linux-2.6.9/kernel/sys.c
>--- linux-2.6.9/kernel/sys.c.orig	2008-05-30 18:35:40.000000000 +0200
>+++ linux-2.6.9/kernel/sys.c	2008-06-03 15:22:40.000000000 +0200
>@@ -975,7 +975,11 @@ asmlinkage long sys_times(struct tms __u
> 		struct task_struct *t;
> 		unsigned long utime, stime, cutime, cstime;
> 
>-		spin_lock_irq(&tsk->proc_lock);
>+		/* We need to hold both proc_lock and sighand->siglock
>+		   to aware modification of threads list by dying thread
>+		   and, thus, to keep consistence of next_thread.
>+		*/
>+		spin_lock(&tsk->proc_lock);
> 		spin_lock_irq(&tsk->sighand->siglock);
> 		utime = tsk->signal->utime;
> 		stime = tsk->signal->stime;
>@@ -986,19 +990,10 @@ asmlinkage long sys_times(struct tms __u
> 			t = next_thread(t);
> 		} while (t != tsk);
> 
>-		/*
>-		 * While we have tasklist_lock read-locked, no dying thread
>-		 * can be updating current->signal->[us]time.  Instead,
>-		 * we got their counts included in the live thread loop.
>-		 * However, another thread can come in right now and
>-		 * do a wait call that updates current->signal->c[us]time.
>-		 * To make sure we always see that pair updated atomically,
>-		 * we take the siglock around fetching them.
>-		 */
> 		cutime = tsk->signal->cutime;
> 		cstime = tsk->signal->cstime;
> 		spin_unlock_irq(&tsk->sighand->siglock);
>-		spin_unlock_irq(&tsk->proc_lock);
>+		spin_unlock(&tsk->proc_lock);
> 
> 		tmp.tms_utime = jiffies_to_clock_t(utime);
> 		tmp.tms_stime = jiffies_to_clock_t(stime);

Actions: View | Diff

Attachments on bug 448934: 308287