aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorJan Kara <jack@suse.cz>2020-02-06 15:28:12 +0100
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>2020-03-05 16:43:52 +0100
commit6f9cff84dde800b4d9eab071810fbe284686601e (patch)
tree71371e687691843338017ae8f209490650250ffa
parentcad53d5e2e23d26aab0357040dc6e3393c3ba7f5 (diff)
downloadlinux-yocto-6f9cff84dde800b4d9eab071810fbe284686601e.tar.gz
linux-yocto-6f9cff84dde800b4d9eab071810fbe284686601e.tar.bz2
linux-yocto-6f9cff84dde800b4d9eab071810fbe284686601e.zip
blktrace: Protect q->blk_trace with RCU
[ Upstream commit c780e86dd48ef6467a1146cf7d0fe1e05a635039 ] KASAN is reporting that __blk_add_trace() has a use-after-free issue when accessing q->blk_trace. Indeed the switching of block tracing (and thus eventual freeing of q->blk_trace) is completely unsynchronized with the currently running tracing and thus it can happen that the blk_trace structure is being freed just while __blk_add_trace() works on it. Protect accesses to q->blk_trace by RCU during tracing and make sure we wait for the end of RCU grace period when shutting down tracing. Luckily that is rare enough event that we can afford that. Note that postponing the freeing of blk_trace to an RCU callback should better be avoided as it could have unexpected user visible side-effects as debugfs files would be still existing for a short while block tracing has been shut down. Link: https://bugzilla.kernel.org/show_bug.cgi?id=205711 CC: stable@vger.kernel.org Reviewed-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com> Reviewed-by: Ming Lei <ming.lei@redhat.com> Tested-by: Ming Lei <ming.lei@redhat.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> Reported-by: Tristan Madani <tristmd@gmail.com> Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Sasha Levin <sashal@kernel.org>
-rw-r--r--include/linux/blkdev.h2
-rw-r--r--include/linux/blktrace_api.h18
-rw-r--r--kernel/trace/blktrace.c114
3 files changed, 97 insertions, 37 deletions
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 51ccb4b8770a..bff1def62eed 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -531,7 +531,7 @@ struct request_queue {
unsigned int sg_reserved_size;
int node;
#ifdef CONFIG_BLK_DEV_IO_TRACE
- struct blk_trace *blk_trace;
+ struct blk_trace __rcu *blk_trace;
struct mutex blk_trace_mutex;
#endif
/*
diff --git a/include/linux/blktrace_api.h b/include/linux/blktrace_api.h
index 7bb2d8de9f30..3b6ff5902edc 100644
--- a/include/linux/blktrace_api.h
+++ b/include/linux/blktrace_api.h
@@ -51,9 +51,13 @@ void __trace_note_message(struct blk_trace *, struct blkcg *blkcg, const char *f
**/
#define blk_add_cgroup_trace_msg(q, cg, fmt, ...) \
do { \
- struct blk_trace *bt = (q)->blk_trace; \
+ struct blk_trace *bt; \
+ \
+ rcu_read_lock(); \
+ bt = rcu_dereference((q)->blk_trace); \
if (unlikely(bt)) \
__trace_note_message(bt, cg, fmt, ##__VA_ARGS__);\
+ rcu_read_unlock(); \
} while (0)
#define blk_add_trace_msg(q, fmt, ...) \
blk_add_cgroup_trace_msg(q, NULL, fmt, ##__VA_ARGS__)
@@ -61,10 +65,14 @@ void __trace_note_message(struct blk_trace *, struct blkcg *blkcg, const char *f
static inline bool blk_trace_note_message_enabled(struct request_queue *q)
{
- struct blk_trace *bt = q->blk_trace;
- if (likely(!bt))
- return false;
- return bt->act_mask & BLK_TC_NOTIFY;
+ struct blk_trace *bt;
+ bool ret;
+
+ rcu_read_lock();
+ bt = rcu_dereference(q->blk_trace);
+ ret = bt && (bt->act_mask & BLK_TC_NOTIFY);
+ rcu_read_unlock();
+ return ret;
}
extern void blk_add_driver_data(struct request_queue *q, struct request *rq,
diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 2d6e93ab0478..4b2ad374167b 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -336,6 +336,7 @@ static void put_probe_ref(void)
static void blk_trace_cleanup(struct blk_trace *bt)
{
+ synchronize_rcu();
blk_trace_free(bt);
put_probe_ref();
}
@@ -630,8 +631,10 @@ static int compat_blk_trace_setup(struct request_queue *q, char *name,
static int __blk_trace_startstop(struct request_queue *q, int start)
{
int ret;
- struct blk_trace *bt = q->blk_trace;
+ struct blk_trace *bt;
+ bt = rcu_dereference_protected(q->blk_trace,
+ lockdep_is_held(&q->blk_trace_mutex));
if (bt == NULL)
return -EINVAL;
@@ -741,8 +744,8 @@ int blk_trace_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg)
void blk_trace_shutdown(struct request_queue *q)
{
mutex_lock(&q->blk_trace_mutex);
-
- if (q->blk_trace) {
+ if (rcu_dereference_protected(q->blk_trace,
+ lockdep_is_held(&q->blk_trace_mutex))) {
__blk_trace_startstop(q, 0);
__blk_trace_remove(q);
}
@@ -754,8 +757,10 @@ void blk_trace_shutdown(struct request_queue *q)
static union kernfs_node_id *
blk_trace_bio_get_cgid(struct request_queue *q, struct bio *bio)
{
- struct blk_trace *bt = q->blk_trace;
+ struct blk_trace *bt;
+ /* We don't use the 'bt' value here except as an optimization... */
+ bt = rcu_dereference_protected(q->blk_trace, 1);
if (!bt || !(blk_tracer_flags.val & TRACE_BLK_OPT_CGROUP))
return NULL;
@@ -800,10 +805,14 @@ static void blk_add_trace_rq(struct request *rq, int error,
unsigned int nr_bytes, u32 what,
union kernfs_node_id *cgid)
{
- struct blk_trace *bt = rq->q->blk_trace;
+ struct blk_trace *bt;
- if (likely(!bt))
+ rcu_read_lock();
+ bt = rcu_dereference(rq->q->blk_trace);
+ if (likely(!bt)) {
+ rcu_read_unlock();
return;
+ }
if (blk_rq_is_passthrough(rq))
what |= BLK_TC_ACT(BLK_TC_PC);
@@ -812,6 +821,7 @@ static void blk_add_trace_rq(struct request *rq, int error,
__blk_add_trace(bt, blk_rq_trace_sector(rq), nr_bytes, req_op(rq),
rq->cmd_flags, what, error, 0, NULL, cgid);
+ rcu_read_unlock();
}
static void blk_add_trace_rq_insert(void *ignore,
@@ -857,14 +867,19 @@ static void blk_add_trace_rq_complete(void *ignore, struct request *rq,
static void blk_add_trace_bio(struct request_queue *q, struct bio *bio,
u32 what, int error)
{
- struct blk_trace *bt = q->blk_trace;
+ struct blk_trace *bt;
- if (likely(!bt))
+ rcu_read_lock();
+ bt = rcu_dereference(q->blk_trace);
+ if (likely(!bt)) {
+ rcu_read_unlock();
return;
+ }
__blk_add_trace(bt, bio->bi_iter.bi_sector, bio->bi_iter.bi_size,
bio_op(bio), bio->bi_opf, what, error, 0, NULL,
blk_trace_bio_get_cgid(q, bio));
+ rcu_read_unlock();
}
static void blk_add_trace_bio_bounce(void *ignore,
@@ -909,11 +924,14 @@ static void blk_add_trace_getrq(void *ignore,
if (bio)
blk_add_trace_bio(q, bio, BLK_TA_GETRQ, 0);
else {
- struct blk_trace *bt = q->blk_trace;
+ struct blk_trace *bt;
+ rcu_read_lock();
+ bt = rcu_dereference(q->blk_trace);
if (bt)
__blk_add_trace(bt, 0, 0, rw, 0, BLK_TA_GETRQ, 0, 0,
NULL, NULL);
+ rcu_read_unlock();
}
}
@@ -925,27 +943,35 @@ static void blk_add_trace_sleeprq(void *ignore,
if (bio)
blk_add_trace_bio(q, bio, BLK_TA_SLEEPRQ, 0);
else {
- struct blk_trace *bt = q->blk_trace;
+ struct blk_trace *bt;
+ rcu_read_lock();
+ bt = rcu_dereference(q->blk_trace);
if (bt)
__blk_add_trace(bt, 0, 0, rw, 0, BLK_TA_SLEEPRQ,
0, 0, NULL, NULL);
+ rcu_read_unlock();
}
}
static void blk_add_trace_plug(void *ignore, struct request_queue *q)
{
- struct blk_trace *bt = q->blk_trace;
+ struct blk_trace *bt;
+ rcu_read_lock();
+ bt = rcu_dereference(q->blk_trace);
if (bt)
__blk_add_trace(bt, 0, 0, 0, 0, BLK_TA_PLUG, 0, 0, NULL, NULL);
+ rcu_read_unlock();
}
static void blk_add_trace_unplug(void *ignore, struct request_queue *q,
unsigned int depth, bool explicit)
{
- struct blk_trace *bt = q->blk_trace;
+ struct blk_trace *bt;
+ rcu_read_lock();
+ bt = rcu_dereference(q->blk_trace);
if (bt) {
__be64 rpdu = cpu_to_be64(depth);
u32 what;
@@ -957,14 +983,17 @@ static void blk_add_trace_unplug(void *ignore, struct request_queue *q,
__blk_add_trace(bt, 0, 0, 0, 0, what, 0, sizeof(rpdu), &rpdu, NULL);
}
+ rcu_read_unlock();
}
static void blk_add_trace_split(void *ignore,
struct request_queue *q, struct bio *bio,
unsigned int pdu)
{
- struct blk_trace *bt = q->blk_trace;
+ struct blk_trace *bt;
+ rcu_read_lock();
+ bt = rcu_dereference(q->blk_trace);
if (bt) {
__be64 rpdu = cpu_to_be64(pdu);
@@ -973,6 +1002,7 @@ static void blk_add_trace_split(void *ignore,
BLK_TA_SPLIT, bio->bi_status, sizeof(rpdu),
&rpdu, blk_trace_bio_get_cgid(q, bio));
}
+ rcu_read_unlock();
}
/**
@@ -992,11 +1022,15 @@ static void blk_add_trace_bio_remap(void *ignore,
struct request_queue *q, struct bio *bio,
dev_t dev, sector_t from)
{
- struct blk_trace *bt = q->blk_trace;
+ struct blk_trace *bt;
struct blk_io_trace_remap r;
- if (likely(!bt))
+ rcu_read_lock();
+ bt = rcu_dereference(q->blk_trace);
+ if (likely(!bt)) {
+ rcu_read_unlock();
return;
+ }
r.device_from = cpu_to_be32(dev);
r.device_to = cpu_to_be32(bio_dev(bio));
@@ -1005,6 +1039,7 @@ static void blk_add_trace_bio_remap(void *ignore,
__blk_add_trace(bt, bio->bi_iter.bi_sector, bio->bi_iter.bi_size,
bio_op(bio), bio->bi_opf, BLK_TA_REMAP, bio->bi_status,
sizeof(r), &r, blk_trace_bio_get_cgid(q, bio));
+ rcu_read_unlock();
}
/**
@@ -1025,11 +1060,15 @@ static void blk_add_trace_rq_remap(void *ignore,
struct request *rq, dev_t dev,
sector_t from)
{
- struct blk_trace *bt = q->blk_trace;
+ struct blk_trace *bt;
struct blk_io_trace_remap r;
- if (likely(!bt))
+ rcu_read_lock();
+ bt = rcu_dereference(q->blk_trace);
+ if (likely(!bt)) {
+ rcu_read_unlock();
return;
+ }
r.device_from = cpu_to_be32(dev);
r.device_to = cpu_to_be32(disk_devt(rq->rq_disk));
@@ -1038,6 +1077,7 @@ static void blk_add_trace_rq_remap(void *ignore,
__blk_add_trace(bt, blk_rq_pos(rq), blk_rq_bytes(rq),
rq_data_dir(rq), 0, BLK_TA_REMAP, 0,
sizeof(r), &r, blk_trace_request_get_cgid(q, rq));
+ rcu_read_unlock();
}
/**
@@ -1055,14 +1095,19 @@ void blk_add_driver_data(struct request_queue *q,
struct request *rq,
void *data, size_t len)
{
- struct blk_trace *bt = q->blk_trace;
+ struct blk_trace *bt;
- if (likely(!bt))
+ rcu_read_lock();
+ bt = rcu_dereference(q->blk_trace);
+ if (likely(!bt)) {
+ rcu_read_unlock();
return;
+ }
__blk_add_trace(bt, blk_rq_trace_sector(rq), blk_rq_bytes(rq), 0, 0,
BLK_TA_DRV_DATA, 0, len, data,
blk_trace_request_get_cgid(q, rq));
+ rcu_read_unlock();
}
EXPORT_SYMBOL_GPL(blk_add_driver_data);
@@ -1589,6 +1634,7 @@ static int blk_trace_remove_queue(struct request_queue *q)
return -EINVAL;
put_probe_ref();
+ synchronize_rcu();
blk_trace_free(bt);
return 0;
}
@@ -1750,6 +1796,7 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
struct hd_struct *p = dev_to_part(dev);
struct request_queue *q;
struct block_device *bdev;
+ struct blk_trace *bt;
ssize_t ret = -ENXIO;
bdev = bdget(part_devt(p));
@@ -1762,21 +1809,23 @@ static ssize_t sysfs_blk_trace_attr_show(struct device *dev,
mutex_lock(&q->blk_trace_mutex);
+ bt = rcu_dereference_protected(q->blk_trace,
+ lockdep_is_held(&q->blk_trace_mutex));
if (attr == &dev_attr_enable) {
- ret = sprintf(buf, "%u\n", !!q->blk_trace);
+ ret = sprintf(buf, "%u\n", !!bt);
goto out_unlock_bdev;
}
- if (q->blk_trace == NULL)
+ if (bt == NULL)
ret = sprintf(buf, "disabled\n");
else if (attr == &dev_attr_act_mask)
- ret = blk_trace_mask2str(buf, q->blk_trace->act_mask);
+ ret = blk_trace_mask2str(buf, bt->act_mask);
else if (attr == &dev_attr_pid)
- ret = sprintf(buf, "%u\n", q->blk_trace->pid);
+ ret = sprintf(buf, "%u\n", bt->pid);
else if (attr == &dev_attr_start_lba)
- ret = sprintf(buf, "%llu\n", q->blk_trace->start_lba);
+ ret = sprintf(buf, "%llu\n", bt->start_lba);
else if (attr == &dev_attr_end_lba)
- ret = sprintf(buf, "%llu\n", q->blk_trace->end_lba);
+ ret = sprintf(buf, "%llu\n", bt->end_lba);
out_unlock_bdev:
mutex_unlock(&q->blk_trace_mutex);
@@ -1793,6 +1842,7 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
struct block_device *bdev;
struct request_queue *q;
struct hd_struct *p;
+ struct blk_trace *bt;
u64 value;
ssize_t ret = -EINVAL;
@@ -1823,8 +1873,10 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
mutex_lock(&q->blk_trace_mutex);
+ bt = rcu_dereference_protected(q->blk_trace,
+ lockdep_is_held(&q->blk_trace_mutex));
if (attr == &dev_attr_enable) {
- if (!!value == !!q->blk_trace) {
+ if (!!value == !!bt) {
ret = 0;
goto out_unlock_bdev;
}
@@ -1836,18 +1888,18 @@ static ssize_t sysfs_blk_trace_attr_store(struct device *dev,
}
ret = 0;
- if (q->blk_trace == NULL)
+ if (bt == NULL)
ret = blk_trace_setup_queue(q, bdev);
if (ret == 0) {
if (attr == &dev_attr_act_mask)
- q->blk_trace->act_mask = value;
+ bt->act_mask = value;
else if (attr == &dev_attr_pid)
- q->blk_trace->pid = value;
+ bt->pid = value;
else if (attr == &dev_attr_start_lba)
- q->blk_trace->start_lba = value;
+ bt->start_lba = value;
else if (attr == &dev_attr_end_lba)
- q->blk_trace->end_lba = value;
+ bt->end_lba = value;
}
out_unlock_bdev: