Received-SPF: pass (google.com: domain of linux-kernel+bounces-89730-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Message-ID: <tencent_DC4C9767C786D2D2FDC64F099FAEFDEEC106@qq.com>
From: Edward Adam Davis <eadavis@qq.com>
To: syzbot+b91eb2ed18f599dd3c31@syzkaller.appspotmail.com
Cc: bcrl@kvack.org,
	brauner@kernel.org,
	bvanassche@acm.org,
	jack@suse.cz,
	linux-aio@kvack.org,
	linux-fsdevel@vger.kernel.org,
	linux-kernel@vger.kernel.org,
	syzkaller-bugs@googlegroups.com,
	viro@zeniv.linux.org.uk
Subject: [PATCH] fs/aio: fix uaf in sys_io_cancel
Date: Sun,  3 Mar 2024 20:21:00 +0800
In-Reply-To: <0000000000006945730612bc9173@google.com>
References: <0000000000006945730612bc9173@google.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit

The aio poll work aio_poll_complete_work() need to be synchronized with syscall
io_cancel(). Otherwise, when poll work executes first, syscall may access the
released aio_kiocb object.

Fixes: 54cbc058d86b ("fs/aio: Make io_cancel() generate completions again")
Reported-and-tested-by: syzbot+b91eb2ed18f599dd3c31@syzkaller.appspotmail.com
Signed-off-by: Edward Adam Davis <eadavis@qq.com>
---
 fs/aio.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/fs/aio.c b/fs/aio.c
index 28223f511931..0fed22ed9eb8 100644
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -1762,9 +1762,8 @@ static void aio_poll_complete_work(struct work_struct *work)
 	} /* else, POLLFREE has freed the waitqueue, so we must complete */
 	list_del_init(&iocb->ki_list);
 	iocb->ki_res.res = mangle_poll(mask);
-	spin_unlock_irq(&ctx->ctx_lock);
-
 	iocb_put(iocb);
+	spin_unlock_irq(&ctx->ctx_lock);
 }
 
 /* assumes we are called with irqs disabled */
@@ -2198,7 +2197,6 @@ SYSCALL_DEFINE3(io_cancel, aio_context_t, ctx_id, struct iocb __user *, iocb,
 			break;
 		}
 	}
-	spin_unlock_irq(&ctx->ctx_lock);
 
 	/*
 	 * The result argument is no longer used - the io_event is always
@@ -2206,6 +2204,7 @@ SYSCALL_DEFINE3(io_cancel, aio_context_t, ctx_id, struct iocb __user *, iocb,
 	 */
 	if (ret == 0 && kiocb->rw.ki_flags & IOCB_AIO_RW)
 		aio_complete_rw(&kiocb->rw, -EINTR);
+	spin_unlock_irq(&ctx->ctx_lock);
 
 	percpu_ref_put(&ctx->users);
 
-- 
2.43.0