From: Sasha Levin <sashal@kernel.org>
To: linux-kernel@vger.kernel.org, stable@vger.kernel.org
Cc: Christoph Hellwig <hch@lst.de>, Jens Axboe <axboe@kernel.dk>,
Sasha Levin <sashal@kernel.org>,
rostedt@goodmis.org, mingo@redhat.com,
linux-block@vger.kernel.org
Subject: [PATCH AUTOSEL 5.18 21/53] block: serialize all debugfs operations using q->debugfs_mutex
Date: Mon, 27 Jun 2022 22:18:07 -0400 [thread overview]
Message-ID: <20220628021839.594423-21-sashal@kernel.org> (raw)
In-Reply-To: <20220628021839.594423-1-sashal@kernel.org>
From: Christoph Hellwig <hch@lst.de>
[ Upstream commit 5cf9c91ba927119fc6606b938b1895bb2459d3bc ]
Various places like I/O schedulers or the QOS infrastructure try to
register debugfs files on demans, which can race with creating and
removing the main queue debugfs directory. Use the existing
debugfs_mutex to serialize all debugfs operations that rely on
q->debugfs_dir or the directories hanging off it.
To make the teardown code a little simpler declare all debugfs dentry
pointers and not just the main one uncoditionally in blkdev.h.
Move debugfs_mutex next to the dentries that it protects and document
what it is used for.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220614074827.458955-3-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
block/blk-mq-debugfs.c | 25 ++++++++++++++++++++-----
block/blk-mq-debugfs.h | 5 -----
block/blk-mq-sched.c | 11 +++++++++++
block/blk-rq-qos.c | 2 ++
block/blk-rq-qos.h | 7 ++++++-
block/blk-sysfs.c | 20 +++++++++-----------
include/linux/blkdev.h | 8 ++++----
kernel/trace/blktrace.c | 3 ---
8 files changed, 52 insertions(+), 29 deletions(-)
diff --git a/block/blk-mq-debugfs.c b/block/blk-mq-debugfs.c
index aa0349e9f083..54f292a1d278 100644
--- a/block/blk-mq-debugfs.c
+++ b/block/blk-mq-debugfs.c
@@ -713,11 +713,6 @@ void blk_mq_debugfs_register(struct request_queue *q)
}
}
-void blk_mq_debugfs_unregister(struct request_queue *q)
-{
- q->sched_debugfs_dir = NULL;
-}
-
static void blk_mq_debugfs_register_ctx(struct blk_mq_hw_ctx *hctx,
struct blk_mq_ctx *ctx)
{
@@ -748,6 +743,8 @@ void blk_mq_debugfs_register_hctx(struct request_queue *q,
void blk_mq_debugfs_unregister_hctx(struct blk_mq_hw_ctx *hctx)
{
+ if (!hctx->queue->debugfs_dir)
+ return;
debugfs_remove_recursive(hctx->debugfs_dir);
hctx->sched_debugfs_dir = NULL;
hctx->debugfs_dir = NULL;
@@ -775,6 +772,8 @@ void blk_mq_debugfs_register_sched(struct request_queue *q)
{
struct elevator_type *e = q->elevator->type;
+ lockdep_assert_held(&q->debugfs_mutex);
+
/*
* If the parent directory has not been created yet, return, we will be
* called again later on and the directory/files will be created then.
@@ -792,6 +791,8 @@ void blk_mq_debugfs_register_sched(struct request_queue *q)
void blk_mq_debugfs_unregister_sched(struct request_queue *q)
{
+ lockdep_assert_held(&q->debugfs_mutex);
+
debugfs_remove_recursive(q->sched_debugfs_dir);
q->sched_debugfs_dir = NULL;
}
@@ -813,6 +814,10 @@ static const char *rq_qos_id_to_name(enum rq_qos_id id)
void blk_mq_debugfs_unregister_rqos(struct rq_qos *rqos)
{
+ lockdep_assert_held(&rqos->q->debugfs_mutex);
+
+ if (!rqos->q->debugfs_dir)
+ return;
debugfs_remove_recursive(rqos->debugfs_dir);
rqos->debugfs_dir = NULL;
}
@@ -822,6 +827,8 @@ void blk_mq_debugfs_register_rqos(struct rq_qos *rqos)
struct request_queue *q = rqos->q;
const char *dir_name = rq_qos_id_to_name(rqos->id);
+ lockdep_assert_held(&q->debugfs_mutex);
+
if (rqos->debugfs_dir || !rqos->ops->debugfs_attrs)
return;
@@ -837,6 +844,8 @@ void blk_mq_debugfs_register_rqos(struct rq_qos *rqos)
void blk_mq_debugfs_unregister_queue_rqos(struct request_queue *q)
{
+ lockdep_assert_held(&q->debugfs_mutex);
+
debugfs_remove_recursive(q->rqos_debugfs_dir);
q->rqos_debugfs_dir = NULL;
}
@@ -846,6 +855,8 @@ void blk_mq_debugfs_register_sched_hctx(struct request_queue *q,
{
struct elevator_type *e = q->elevator->type;
+ lockdep_assert_held(&q->debugfs_mutex);
+
/*
* If the parent debugfs directory has not been created yet, return;
* We will be called again later on with appropriate parent debugfs
@@ -865,6 +876,10 @@ void blk_mq_debugfs_register_sched_hctx(struct request_queue *q,
void blk_mq_debugfs_unregister_sched_hctx(struct blk_mq_hw_ctx *hctx)
{
+ lockdep_assert_held(&hctx->queue->debugfs_mutex);
+
+ if (!hctx->queue->debugfs_dir)
+ return;
debugfs_remove_recursive(hctx->sched_debugfs_dir);
hctx->sched_debugfs_dir = NULL;
}
diff --git a/block/blk-mq-debugfs.h b/block/blk-mq-debugfs.h
index 69918f4170d6..771d45832878 100644
--- a/block/blk-mq-debugfs.h
+++ b/block/blk-mq-debugfs.h
@@ -21,7 +21,6 @@ int __blk_mq_debugfs_rq_show(struct seq_file *m, struct request *rq);
int blk_mq_debugfs_rq_show(struct seq_file *m, void *v);
void blk_mq_debugfs_register(struct request_queue *q);
-void blk_mq_debugfs_unregister(struct request_queue *q);
void blk_mq_debugfs_register_hctx(struct request_queue *q,
struct blk_mq_hw_ctx *hctx);
void blk_mq_debugfs_unregister_hctx(struct blk_mq_hw_ctx *hctx);
@@ -42,10 +41,6 @@ static inline void blk_mq_debugfs_register(struct request_queue *q)
{
}
-static inline void blk_mq_debugfs_unregister(struct request_queue *q)
-{
-}
-
static inline void blk_mq_debugfs_register_hctx(struct request_queue *q,
struct blk_mq_hw_ctx *hctx)
{
diff --git a/block/blk-mq-sched.c b/block/blk-mq-sched.c
index 9e56a69422b6..e84bec39fd3a 100644
--- a/block/blk-mq-sched.c
+++ b/block/blk-mq-sched.c
@@ -593,7 +593,9 @@ int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e)
if (ret)
goto err_free_map_and_rqs;
+ mutex_lock(&q->debugfs_mutex);
blk_mq_debugfs_register_sched(q);
+ mutex_unlock(&q->debugfs_mutex);
queue_for_each_hw_ctx(q, hctx, i) {
if (e->ops.init_hctx) {
@@ -606,7 +608,9 @@ int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e)
return ret;
}
}
+ mutex_lock(&q->debugfs_mutex);
blk_mq_debugfs_register_sched_hctx(q, hctx);
+ mutex_unlock(&q->debugfs_mutex);
}
return 0;
@@ -647,14 +651,21 @@ void blk_mq_exit_sched(struct request_queue *q, struct elevator_queue *e)
unsigned int flags = 0;
queue_for_each_hw_ctx(q, hctx, i) {
+ mutex_lock(&q->debugfs_mutex);
blk_mq_debugfs_unregister_sched_hctx(hctx);
+ mutex_unlock(&q->debugfs_mutex);
+
if (e->type->ops.exit_hctx && hctx->sched_data) {
e->type->ops.exit_hctx(hctx, i);
hctx->sched_data = NULL;
}
flags = hctx->flags;
}
+
+ mutex_lock(&q->debugfs_mutex);
blk_mq_debugfs_unregister_sched(q);
+ mutex_unlock(&q->debugfs_mutex);
+
if (e->type->ops.exit_sched)
e->type->ops.exit_sched(e);
blk_mq_sched_tags_teardown(q, flags);
diff --git a/block/blk-rq-qos.c b/block/blk-rq-qos.c
index e83af7bc7591..249a6f05dd3b 100644
--- a/block/blk-rq-qos.c
+++ b/block/blk-rq-qos.c
@@ -294,7 +294,9 @@ void rq_qos_wait(struct rq_wait *rqw, void *private_data,
void rq_qos_exit(struct request_queue *q)
{
+ mutex_lock(&q->debugfs_mutex);
blk_mq_debugfs_unregister_queue_rqos(q);
+ mutex_unlock(&q->debugfs_mutex);
while (q->rq_qos) {
struct rq_qos *rqos = q->rq_qos;
diff --git a/block/blk-rq-qos.h b/block/blk-rq-qos.h
index 68267007da1c..0e46052b018a 100644
--- a/block/blk-rq-qos.h
+++ b/block/blk-rq-qos.h
@@ -104,8 +104,11 @@ static inline void rq_qos_add(struct request_queue *q, struct rq_qos *rqos)
blk_mq_unfreeze_queue(q);
- if (rqos->ops->debugfs_attrs)
+ if (rqos->ops->debugfs_attrs) {
+ mutex_lock(&q->debugfs_mutex);
blk_mq_debugfs_register_rqos(rqos);
+ mutex_unlock(&q->debugfs_mutex);
+ }
}
static inline void rq_qos_del(struct request_queue *q, struct rq_qos *rqos)
@@ -129,7 +132,9 @@ static inline void rq_qos_del(struct request_queue *q, struct rq_qos *rqos)
blk_mq_unfreeze_queue(q);
+ mutex_lock(&q->debugfs_mutex);
blk_mq_debugfs_unregister_rqos(rqos);
+ mutex_unlock(&q->debugfs_mutex);
}
typedef bool (acquire_inflight_cb_t)(struct rq_wait *rqw, void *private_data);
diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index 88bd41d4cb59..6e4801b217a7 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -779,14 +779,13 @@ static void blk_release_queue(struct kobject *kobj)
if (queue_is_mq(q))
blk_mq_release(q);
- blk_trace_shutdown(q);
mutex_lock(&q->debugfs_mutex);
+ blk_trace_shutdown(q);
debugfs_remove_recursive(q->debugfs_dir);
+ q->debugfs_dir = NULL;
+ q->sched_debugfs_dir = NULL;
mutex_unlock(&q->debugfs_mutex);
- if (queue_is_mq(q))
- blk_mq_debugfs_unregister(q);
-
bioset_exit(&q->bio_split);
if (blk_queue_has_srcu(q))
@@ -836,17 +835,16 @@ int blk_register_queue(struct gendisk *disk)
goto unlock;
}
+ if (queue_is_mq(q))
+ __blk_mq_register_dev(dev, q);
+ mutex_lock(&q->sysfs_lock);
+
mutex_lock(&q->debugfs_mutex);
q->debugfs_dir = debugfs_create_dir(kobject_name(q->kobj.parent),
blk_debugfs_root);
- mutex_unlock(&q->debugfs_mutex);
-
- if (queue_is_mq(q)) {
- __blk_mq_register_dev(dev, q);
+ if (queue_is_mq(q))
blk_mq_debugfs_register(q);
- }
-
- mutex_lock(&q->sysfs_lock);
+ mutex_unlock(&q->debugfs_mutex);
ret = disk_register_independent_access_ranges(disk, NULL);
if (ret)
diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 108e3d114bfc..cc6b24a5098f 100644
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -466,7 +466,6 @@ struct request_queue {
#endif /* CONFIG_BLK_DEV_ZONED */
int node;
- struct mutex debugfs_mutex;
#ifdef CONFIG_BLK_DEV_IO_TRACE
struct blk_trace __rcu *blk_trace;
#endif
@@ -510,11 +509,12 @@ struct request_queue {
struct bio_set bio_split;
struct dentry *debugfs_dir;
-
-#ifdef CONFIG_BLK_DEBUG_FS
struct dentry *sched_debugfs_dir;
struct dentry *rqos_debugfs_dir;
-#endif
+ /*
+ * Serializes all debugfs metadata operations using the above dentries.
+ */
+ struct mutex debugfs_mutex;
bool mq_sysfs_init_done;
diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 4d5629196d01..134a09ae63fc 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -770,14 +770,11 @@ int blk_trace_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg)
**/
void blk_trace_shutdown(struct request_queue *q)
{
- mutex_lock(&q->debugfs_mutex);
if (rcu_dereference_protected(q->blk_trace,
lockdep_is_held(&q->debugfs_mutex))) {
__blk_trace_startstop(q, 0);
__blk_trace_remove(q);
}
-
- mutex_unlock(&q->debugfs_mutex);
}
#ifdef CONFIG_BLK_CGROUP
--
2.35.1
next prev parent reply other threads:[~2022-06-28 2:20 UTC|newest]
Thread overview: 54+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-06-28 2:17 [PATCH AUTOSEL 5.18 01/53] spi: spi-cadence: Fix SPI CS gets toggling sporadically Sasha Levin
2022-06-28 2:17 ` [PATCH AUTOSEL 5.18 02/53] spi: cadence: Detect transmit FIFO depth Sasha Levin
2022-06-28 2:17 ` [PATCH AUTOSEL 5.18 03/53] spi: spi-mem: Fix spi_mem_poll_status() Sasha Levin
2022-06-28 2:17 ` [PATCH AUTOSEL 5.18 04/53] arm64: s32g2: Pass unit name to soc node Sasha Levin
2022-06-28 2:17 ` [PATCH AUTOSEL 5.18 05/53] regulator: qcom_smd: correct MP5496 ranges Sasha Levin
2022-06-28 2:17 ` [PATCH AUTOSEL 5.18 06/53] ALSA: usb-audio: US16x08: Move overflow check before array access Sasha Levin
2022-06-28 2:17 ` [PATCH AUTOSEL 5.18 07/53] bus: bt1-apb: Don't print error on -EPROBE_DEFER Sasha Levin
2022-06-28 2:17 ` [PATCH AUTOSEL 5.18 08/53] bus: bt1-axi: " Sasha Levin
2022-06-28 2:17 ` [PATCH AUTOSEL 5.18 09/53] selftests: Fix clang cross compilation Sasha Levin
2022-06-28 2:17 ` [PATCH AUTOSEL 5.18 10/53] drm/vc4: plane: Prevent async update if we don't have a dlist Sasha Levin
2022-06-28 2:17 ` [PATCH AUTOSEL 5.18 11/53] drm/vc4: crtc: Use an union to store the page flip callback Sasha Levin
2022-06-28 2:17 ` [PATCH AUTOSEL 5.18 12/53] drm/vc4: crtc: Move the BO handling out of common page-flip callback Sasha Levin
2022-06-28 2:17 ` [PATCH AUTOSEL 5.18 13/53] selftests: vm: Fix resource leak when return error Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 14/53] scsi: ufs: Simplify ufshcd_clear_cmd() Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 15/53] scsi: ufs: Support clearing multiple commands at once Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 16/53] scsi: ufs: Fix a race between the interrupt handler and the reset handler Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 17/53] selftests/bpf: Shuffle cookies symbols in kprobe multi test Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 18/53] ALSA: x86: intel_hdmi_audio: enable pm_runtime and set autosuspend delay Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 19/53] ALSA: x86: intel_hdmi_audio: use pm_runtime_resume_and_get() Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 20/53] hamradio: 6pack: fix array-index-out-of-bounds in decode_std_command() Sasha Levin
2022-06-28 2:18 ` Sasha Levin [this message]
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 22/53] block: remove per-disk debugfs files in blk_unregister_queue Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 23/53] block: freeze the queue earlier in del_gendisk Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 24/53] drivers/net/ethernet/neterion/vxge: Fix a use-after-free bug in vxge-main.c Sasha Levin
2022-06-28 4:38 ` Jakub Kicinski
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 25/53] powerpc/prom_init: Fix build failure with GCC_PLUGIN_STRUCTLEAK_BYREF_ALL and KASAN Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 26/53] libperf evsel: Open shouldn't leak fd on failure Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 27/53] drm: panel-orientation-quirks: Add quirk for Aya Neo Next Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 28/53] iio: freq: admv1014: Fix warning about dubious x & !y and improve readability Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 29/53] spi: rockchip: Unmask IRQ at the final to avoid preemption Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 30/53] video: fbdev: skeletonfb: Fix syntax errors in comments Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 31/53] video: fbdev: intelfb: Use aperture size from pci_resource_len Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 32/53] io_uring: mark reissue requests with REQ_F_PARTIAL_IO Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 33/53] video: fbdev: pxa3xx-gcu: Fix integer overflow in pxa3xx_gcu_write Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 34/53] video: fbdev: simplefb: Check before clk_put() not needed Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 35/53] btrfs: add missing inode updates on each iteration when replacing extents Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 36/53] btrfs: do not BUG_ON() on failure to migrate space " Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 37/53] io_uring: fix merge error in checking send/recv addr2 flags Sasha Levin
2022-06-28 13:12 ` Jens Axboe
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 38/53] drm/xen: Add missing VM_DONTEXPAND flag in mmap callback Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 39/53] arch: mips: generic: Add missing of_node_put() in board-ranchu.c Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 40/53] mips: mti-malta: Fix refcount leak in malta-time.c Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 41/53] mips: ralink: Fix refcount leak in of.c Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 42/53] mips: lantiq: falcon: Fix refcount leak bug in sysctrl Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 43/53] mips: lantiq: xway: " Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 44/53] mips/pic32/pic32mzda: Fix refcount leak bugs Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 45/53] mips: dts: ingenic: Add TCU clock to x1000/x1830 tcu device node Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 46/53] mips: lantiq: Add missing of_node_put() in irq.c Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 47/53] drm/sun4i: Add DMA mask and segment size Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 48/53] drm/sun4i: Return if frontend is not present Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 49/53] hinic: Replace memcpy() with direct assignment Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 50/53] drm/amdgpu: Adjust logic around GTT size (v3) Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 51/53] nvme: add a bogus subsystem NQN quirk for Micron MTFDKBA2T0TFH Sasha Levin
2022-06-28 2:18 ` [PATCH AUTOSEL 5.18 52/53] gpio: grgpio: Fix device removing Sasha Levin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20220628021839.594423-21-sashal@kernel.org \
--to=sashal@kernel.org \
--cc=axboe@kernel.dk \
--cc=hch@lst.de \
--cc=linux-block@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=rostedt@goodmis.org \
--cc=stable@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®