[PATCH] Cpuset: cant mmput inside tasklist_lock fix

[Paul Jackson]

mmput() inside tasklist_lock() can deadlock (thanks, Andrew).

To fix, move the range check before getting the mm, so the mm
doesn't have to be mmput if the check failed.

Also the BUG() was excessively cruel (panics) for this case.
Even though "it can't happen", still it is easy to carry
on and keep the kernel healthy, at the penalty of not placing
memory exactly where the user expected.

Signed-off-by: Paul Jackson <pj@xxxxxxx>

---

 kernel/cpuset.c |   14 +++++---------
 1 files changed, 5 insertions(+), 9 deletions(-)

--- 2.6.15-rc3-mm1.orig/kernel/cpuset.c	2005-12-10 01:06:00.717770568 -0800
+++ 2.6.15-rc3-mm1/kernel/cpuset.c	2005-12-10 01:29:23.604065478 -0800
@@ -881,20 +881,16 @@ static int update_nodemask(struct cpuset
 	do_each_thread(g, p) {
 		struct mm_struct *mm;
 
+		if (n >= ntasks) {
+			printk(KERN_WARNING
+				"Cpuset mempolicy rebind incomplete.\n");
+			continue;
+		}
 		if (p->cpuset != cs)
 			continue;
 		mm = get_task_mm(p);
 		if (!mm)
 			continue;
-		if (n >= ntasks) {
-			if (printk_ratelimit()) {
-				printk (KERN_ERR
-					"Cpuset mempolicy rebind failed.\n");
-				BUG();
-			}
-			mmput(mm);
-			continue;
-		}
 		mmarray[n++] = mm;
 	} while_each_thread(g, p);
 	write_unlock_irq(&tasklist_lock);

-- 
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <pj@xxxxxxx> 1.650.933.1373
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/