kernel_optimize_test/include/linux/iocontext.h
Sahitya Tummala 30a2da7b7e block: Fix use-after-free issue accessing struct io_cq
There is a potential race between ioc_release_fn() and
ioc_clear_queue() as shown below, due to which below kernel
crash is observed. It also can result into use-after-free
issue.

context#1:				context#2:
ioc_release_fn()			__ioc_clear_queue() gets the same icq
->spin_lock(&ioc->lock);		->spin_lock(&ioc->lock);
->ioc_destroy_icq(icq);
  ->list_del_init(&icq->q_node);
  ->call_rcu(&icq->__rcu_head,
  	icq_free_icq_rcu);
->spin_unlock(&ioc->lock);
					->ioc_destroy_icq(icq);
					  ->hlist_del_init(&icq->ioc_node);
					  This results into below crash as this memory
					  is now used by icq->__rcu_head in context#1.
					  There is a chance that icq could be free'd
					  as well.

22150.386550:   <6> Unable to handle kernel write to read-only memory
at virtual address ffffffaa8d31ca50
...
Call trace:
22150.607350:   <2>  ioc_destroy_icq+0x44/0x110
22150.611202:   <2>  ioc_clear_queue+0xac/0x148
22150.615056:   <2>  blk_cleanup_queue+0x11c/0x1a0
22150.619174:   <2>  __scsi_remove_device+0xdc/0x128
22150.623465:   <2>  scsi_forget_host+0x2c/0x78
22150.627315:   <2>  scsi_remove_host+0x7c/0x2a0
22150.631257:   <2>  usb_stor_disconnect+0x74/0xc8
22150.635371:   <2>  usb_unbind_interface+0xc8/0x278
22150.639665:   <2>  device_release_driver_internal+0x198/0x250
22150.644897:   <2>  device_release_driver+0x24/0x30
22150.649176:   <2>  bus_remove_device+0xec/0x140
22150.653204:   <2>  device_del+0x270/0x460
22150.656712:   <2>  usb_disable_device+0x120/0x390
22150.660918:   <2>  usb_disconnect+0xf4/0x2e0
22150.664684:   <2>  hub_event+0xd70/0x17e8
22150.668197:   <2>  process_one_work+0x210/0x480
22150.672222:   <2>  worker_thread+0x32c/0x4c8

Fix this by adding a new ICQ_DESTROYED flag in ioc_destroy_icq() to
indicate this icq is once marked as destroyed. Also, ensure
__ioc_clear_queue() is accessing icq within rcu_read_lock/unlock so
that icq doesn't get free'd up while it is still using it.

Signed-off-by: Sahitya Tummala <stummala@codeaurora.org>
Co-developed-by: Pradeep P V K <ppvk@codeaurora.org>
Signed-off-by: Pradeep P V K <ppvk@codeaurora.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-03-12 07:07:38 -06:00

160 lines
4.8 KiB
C

/* SPDX-License-Identifier: GPL-2.0 */
#ifndef IOCONTEXT_H
#define IOCONTEXT_H
#include <linux/radix-tree.h>
#include <linux/rcupdate.h>
#include <linux/workqueue.h>
enum {
ICQ_EXITED = 1 << 2,
ICQ_DESTROYED = 1 << 3,
};
/*
* An io_cq (icq) is association between an io_context (ioc) and a
* request_queue (q). This is used by elevators which need to track
* information per ioc - q pair.
*
* Elevator can request use of icq by setting elevator_type->icq_size and
* ->icq_align. Both size and align must be larger than that of struct
* io_cq and elevator can use the tail area for private information. The
* recommended way to do this is defining a struct which contains io_cq as
* the first member followed by private members and using its size and
* align. For example,
*
* struct snail_io_cq {
* struct io_cq icq;
* int poke_snail;
* int feed_snail;
* };
*
* struct elevator_type snail_elv_type {
* .ops = { ... },
* .icq_size = sizeof(struct snail_io_cq),
* .icq_align = __alignof__(struct snail_io_cq),
* ...
* };
*
* If icq_size is set, block core will manage icq's. All requests will
* have its ->elv.icq field set before elevator_ops->elevator_set_req_fn()
* is called and be holding a reference to the associated io_context.
*
* Whenever a new icq is created, elevator_ops->elevator_init_icq_fn() is
* called and, on destruction, ->elevator_exit_icq_fn(). Both functions
* are called with both the associated io_context and queue locks held.
*
* Elevator is allowed to lookup icq using ioc_lookup_icq() while holding
* queue lock but the returned icq is valid only until the queue lock is
* released. Elevators can not and should not try to create or destroy
* icq's.
*
* As icq's are linked from both ioc and q, the locking rules are a bit
* complex.
*
* - ioc lock nests inside q lock.
*
* - ioc->icq_list and icq->ioc_node are protected by ioc lock.
* q->icq_list and icq->q_node by q lock.
*
* - ioc->icq_tree and ioc->icq_hint are protected by ioc lock, while icq
* itself is protected by q lock. However, both the indexes and icq
* itself are also RCU managed and lookup can be performed holding only
* the q lock.
*
* - icq's are not reference counted. They are destroyed when either the
* ioc or q goes away. Each request with icq set holds an extra
* reference to ioc to ensure it stays until the request is completed.
*
* - Linking and unlinking icq's are performed while holding both ioc and q
* locks. Due to the lock ordering, q exit is simple but ioc exit
* requires reverse-order double lock dance.
*/
struct io_cq {
struct request_queue *q;
struct io_context *ioc;
/*
* q_node and ioc_node link io_cq through icq_list of q and ioc
* respectively. Both fields are unused once ioc_exit_icq() is
* called and shared with __rcu_icq_cache and __rcu_head which are
* used for RCU free of io_cq.
*/
union {
struct list_head q_node;
struct kmem_cache *__rcu_icq_cache;
};
union {
struct hlist_node ioc_node;
struct rcu_head __rcu_head;
};
unsigned int flags;
};
/*
* I/O subsystem state of the associated processes. It is refcounted
* and kmalloc'ed. These could be shared between processes.
*/
struct io_context {
atomic_long_t refcount;
atomic_t active_ref;
atomic_t nr_tasks;
/* all the fields below are protected by this lock */
spinlock_t lock;
unsigned short ioprio;
/*
* For request batching
*/
int nr_batch_requests; /* Number of requests left in the batch */
unsigned long last_waited; /* Time last woken after wait for request */
struct radix_tree_root icq_tree;
struct io_cq __rcu *icq_hint;
struct hlist_head icq_list;
struct work_struct release_work;
};
/**
* get_io_context_active - get active reference on ioc
* @ioc: ioc of interest
*
* Only iocs with active reference can issue new IOs. This function
* acquires an active reference on @ioc. The caller must already have an
* active reference on @ioc.
*/
static inline void get_io_context_active(struct io_context *ioc)
{
WARN_ON_ONCE(atomic_long_read(&ioc->refcount) <= 0);
WARN_ON_ONCE(atomic_read(&ioc->active_ref) <= 0);
atomic_long_inc(&ioc->refcount);
atomic_inc(&ioc->active_ref);
}
static inline void ioc_task_link(struct io_context *ioc)
{
get_io_context_active(ioc);
WARN_ON_ONCE(atomic_read(&ioc->nr_tasks) <= 0);
atomic_inc(&ioc->nr_tasks);
}
struct task_struct;
#ifdef CONFIG_BLOCK
void put_io_context(struct io_context *ioc);
void put_io_context_active(struct io_context *ioc);
void exit_io_context(struct task_struct *task);
struct io_context *get_task_io_context(struct task_struct *task,
gfp_t gfp_flags, int node);
#else
struct io_context;
static inline void put_io_context(struct io_context *ioc) { }
static inline void exit_io_context(struct task_struct *task) { }
#endif
#endif