[PATCH 1/3] cpuset: zero malloc - fix for old cpusets

[PATCH 1/3] cpuset: zero malloc - fix for old cpusets

[Paul Jackson]

From: Paul Jackson <pj@sgi.com>

First of three -- this one goes before the container patches,
and should be sent to Linus for 2.6.22.

The cpuset code to present a list of tasks using a cpuset to
user space could write to an array that it had kmalloc'd,
after a kmalloc request of zero size.

The problem was that the code didn't check for writes past
the allocated end of the array until -after- the first write.

This is a race condition that is likely rare -- it would only
show up if a cpuset went from being empty to having a task
in it, during the brief time between the allocation and the
first write.

Prior to roughly 2.6.22 kernels, this was also a benign
problem, because a zero kmalloc returned a few usable bytes
anyway, and no harm was done with the bogus write.

With the 2.6.22 kernel changes to make issue a warning if
code tries to write to the location returned from a zero
size allocation, this problem is no longer benign.  This
cpuset code would occassionally trigger that warning.

The fix is trivial -- check before storing into the array,
not after, whether the array is big enough to hold the store.

Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: "Serge E. Hallyn" <serue@us.ibm.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: Dave Hansen <haveblue@us.ibm.com>
Cc: Herbert Poetzl <herbert@13thfloor.at>
Cc: Kirill Korotaev <dev@openvz.org>
Cc: Paul Menage <menage@google.com>
Cc: Srivatsa Vaddagiri <vatsa@in.ibm.com>
Cc: Christoph Lameter <clameter@sgi.com>

Signed-off-by: Paul Jackson <pj@sgi.com>

---

Andrew - this fix collides with the container changes, and now
that you have sent Christoph's patch to Linus to warn on writes
to zero-sized allocations, this fix should go to Linus too.

So it comes in three parts: fix the old, undo that, fix the new.

This patch should be applied first, before the container patches,
and sent to Linus for 2.6.22.

 kernel/cpuset.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

--- 2.6.22-rc4-mm2.orig/kernel/cpuset.c	2007-06-08 14:44:58.899036450 -0700
+++ 2.6.22-rc4-mm2/kernel/cpuset.c	2007-06-08 14:47:00.996907442 -0700
@@ -1620,9 +1620,9 @@ static int pid_array_load(pid_t *pidarra
 
 	do_each_thread(g, p) {
 		if (p->cpuset == cs) {
-			pidarray[n++] = p->pid;
 			if (unlikely(n == npids))
 				goto array_full;
+			pidarray[n++] = p->pid;
 		}
 	} while_each_thread(g, p);
 

-- 
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <pj@sgi.com> 1.650.933.1373

[PATCH 2/3] cpuset: zero malloc - revert the old cpuset fix

[Paul Jackson]

From: Paul Jackson <pj@sgi.com>

Second of three -- This one undoes the first one.  Apply before
the container patch that changes cpusets to use containers,
in order to avoid a patch conflict.

The cpuset code to present a list of tasks using a cpuset to
user space could write to an array that it had kmalloc'd,
after a kmalloc request of zero size.

The problem was that the code didn't check for writes past
the allocated end of the array until -after- the first write.

This is a race condition that is likely rare -- it would only
show up if a cpuset went from being empty to having a task
in it, during the brief time between the allocation and the
first write.

Prior to roughly 2.6.22 kernels, this was also a benign
problem, because a zero kmalloc returned a few usable bytes
anyway, and no harm was done with the bogus write.

With the 2.6.22 kernel changes to make issue a warning if
code tries to write to the location returned from a zero
size allocation, this problem is no longer benign.  This
cpuset code would occassionally trigger that warning.

The fix is trivial -- check before storing into the array,
not after, whether the array is big enough to hold the store.

Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: "Serge E. Hallyn" <serue@us.ibm.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: Dave Hansen <haveblue@us.ibm.com>
Cc: Herbert Poetzl <herbert@13thfloor.at>
Cc: Kirill Korotaev <dev@openvz.org>
Cc: Paul Menage <menage@google.com>
Cc: Srivatsa Vaddagiri <vatsa@in.ibm.com>
Cc: Christoph Lameter <clameter@sgi.com>

Signed-off-by: Paul Jackson <pj@sgi.com>

---

Andrew,

This is the second of three in the set - it reverses the
first one, to avoid colliding with the container patches.

Apply somewhere before, or entirely merge into:
    containersv10-share-css_group-arrays-between-tasks-with-same-container-memberships.patch


 kernel/cpuset.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

--- 2.6.22-rc4-mm2.orig/kernel/cpuset.c	2007-06-08 14:57:59.091074320 -0700
+++ 2.6.22-rc4-mm2/kernel/cpuset.c	2007-06-08 15:34:24.336172342 -0700
@@ -1620,9 +1620,9 @@ static int pid_array_load(pid_t *pidarra
 
 	do_each_thread(g, p) {
 		if (p->cpuset == cs) {
+			pidarray[n++] = p->pid;
 			if (unlikely(n == npids))
 				goto array_full;
-			pidarray[n++] = p->pid;
 		}
 	} while_each_thread(g, p);
 

-- 
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <pj@sgi.com> 1.650.933.1373

[PATCH 3/3] cpuset: zero malloc - fix for new containers

[Paul Jackson]

From: Paul Jackson <pj@sgi.com>

Third of three -- This applies after the container patches,
fixing a problem that earlier patches also fixed in the older
cpuset code.

The container code to present a list of tasks using a container
to user space could write to an array that it had kmalloc'd,
after a kmalloc request of zero size.

The problem was that the code didn't check for writes past the
allocated end of the array until -after- the first write.

This is a race condition that is likely rare -- it would only show
up if a container went from being empty to having a task in it,
during the brief time between the allocation and the first write.

Prior to roughly 2.6.22 kernels, this was also a benign problem,
because a zero kmalloc returned a few usable bytes anyway,
and no harm was done with the bogus write.

With the 2.6.22 kernel changes to make issue a warning if
code tries to write to the location returned from a zero size
allocation, this problem is no longer benign.  This container code
would occassionally trigger that warning.

The fix is trivial -- check before storing into the array,
not after, whether the array is big enough to hold the store.

Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: "Serge E. Hallyn" <serue@us.ibm.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: Dave Hansen <haveblue@us.ibm.com>
Cc: Herbert Poetzl <herbert@13thfloor.at>
Cc: Kirill Korotaev <dev@openvz.org>
Cc: Paul Menage <menage@google.com>
Cc: Srivatsa Vaddagiri <vatsa@in.ibm.com>
Cc: Christoph Lameter <clameter@sgi.com>

Signed-off-by: Paul Jackson <pj@sgi.com>

---

Andrew - this is the third of three similar patches.  This one
redoes the fix for the container code.

Please apply somewhere after:
  containersv10-share-css_group-arrays-between-tasks-with-same-container-memberships.patch

 kernel/container.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

--- 2.6.22-rc4-mm2.orig/kernel/container.c	2007-06-08 15:36:14.997835546 -0700
+++ 2.6.22-rc4-mm2/kernel/container.c	2007-06-08 15:37:09.606655972 -0700
@@ -1451,9 +1451,9 @@ static int pid_array_load(pid_t *pidarra
 	struct task_struct *tsk;
 	container_iter_start(cont, &it);
 	while ((tsk = container_iter_next(cont, &it))) {
-		pidarray[n++] = pid_nr(task_pid(tsk));
 		if (unlikely(n == npids))
 			break;
+		pidarray[n++] = pid_nr(task_pid(tsk));
 	}
 	container_iter_end(cont, &it);
 	return n;

-- 
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <pj@sgi.com> 1.650.933.1373