linux-kernel - [PATCH 3.16 355/410] aio: kill the misleading rcu read locks in ioctx_add_table() and kill

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <lsq.1528380321.458161244@decadent.org.uk>
Date:   Thu, 07 Jun 2018 15:05:21 +0100
From:   Ben Hutchings <ben@...adent.org.uk>
To:     linux-kernel@...r.kernel.org, stable@...r.kernel.org
CC:     akpm@...ux-foundation.org, "Oleg Nesterov" <oleg@...hat.com>,
        "Benjamin LaHaise" <bcrl@...ck.org>
Subject: [PATCH 3.16 355/410] aio: kill the misleading rcu read locks in
 ioctx_add_table() and kill_ioctx()

3.16.57-rc1 review patch.  If anyone has any objections, please let me know.

------------------

From: Oleg Nesterov <oleg@...hat.com>

commit 855ef0dec7271ff7be7381feaaf3f4aed80bd503 upstream.

ioctx_add_table() is the writer, it does not need rcu_read_lock() to
protect ->ioctx_table. It relies on mm->ioctx_lock and rcu locks just
add the confusion.

And it doesn't need rcu_dereference() by the same reason, it must see
any updates previously done under the same ->ioctx_lock. We could use
rcu_dereference_protected() but the patch uses rcu_dereference_raw(),
the function is simple enough.

The same for kill_ioctx(), although it does not update the pointer.

Signed-off-by: Oleg Nesterov <oleg@...hat.com>
Signed-off-by: Benjamin LaHaise <bcrl@...ck.org>
Signed-off-by: Ben Hutchings <ben@...adent.org.uk>
---
 fs/aio.c | 14 +++-----------
 1 file changed, 3 insertions(+), 11 deletions(-)

--- a/fs/aio.c
+++ b/fs/aio.c
@@ -563,8 +563,7 @@ static int ioctx_add_table(struct kioctx
 	struct aio_ring *ring;
 
 	spin_lock(&mm->ioctx_lock);
-	rcu_read_lock();
-	table = rcu_dereference(mm->ioctx_table);
+	table = rcu_dereference_raw(mm->ioctx_table);
 
 	while (1) {
 		if (table)
@@ -572,7 +571,6 @@ static int ioctx_add_table(struct kioctx
 				if (!table->table[i]) {
 					ctx->id = i;
 					table->table[i] = ctx;
-					rcu_read_unlock();
 					spin_unlock(&mm->ioctx_lock);
 
 					/* While kioctx setup is in progress,
@@ -586,8 +584,6 @@ static int ioctx_add_table(struct kioctx
 				}
 
 		new_nr = (table ? table->nr : 1) * 4;
-
-		rcu_read_unlock();
 		spin_unlock(&mm->ioctx_lock);
 
 		table = kzalloc(sizeof(*table) + sizeof(struct kioctx *) *
@@ -598,8 +594,7 @@ static int ioctx_add_table(struct kioctx
 		table->nr = new_nr;
 
 		spin_lock(&mm->ioctx_lock);
-		rcu_read_lock();
-		old = rcu_dereference(mm->ioctx_table);
+		old = rcu_dereference_raw(mm->ioctx_table);
 
 		if (!old) {
 			rcu_assign_pointer(mm->ioctx_table, table);
@@ -749,12 +744,9 @@ static int kill_ioctx(struct mm_struct *
 
 
 	spin_lock(&mm->ioctx_lock);
-	rcu_read_lock();
-	table = rcu_dereference(mm->ioctx_table);
-
+	table = rcu_dereference_raw(mm->ioctx_table);
 	WARN_ON(ctx != table->table[ctx->id]);
 	table->table[ctx->id] = NULL;
-	rcu_read_unlock();
 	spin_unlock(&mm->ioctx_lock);
 
 	/* percpu_ref_kill() will do the necessary call_rcu() */