Fix warnings in blktrace

These two patches fix wrong usage of lock primitives with CONFIG_PREEMPT_RT in the blktrace code.

This version fixes a missing piece in the blktrace patch.

The patches apply to the PREEMPT_RT tree.

Wander Lairson Costa (2):
block: Avoid sleeping function called from invalid context bug
blktrace: switch trace spinlock to a raw spinlock

 block/blk-cgroup.c      |  4 ++--
 kernel/trace/blktrace.c | 18 +++++++++---------
 2 files changed, 11 insertions(+), 11 deletions(-)

2.27.0

This was caught during QA test:

 BUG: sleeping function called from invalid context at kernel/locking/rtmutex.c:942
 in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 243401, name: sed
 INFO: lockdep is turned off.
 Preemption disabled at:
 [<ffffffff89b26268>] blk_cgroup_bio_start+0x28/0xd0

 CPU: 2 PID: 243401 Comm: sed Kdump: loaded Not tainted 4.18.0-353.rt7.138.el8.x86_64+debug #1
 Hardware name: HP ProLiant DL380 Gen9, BIOS P89 05/06/2015
 Call Trace:
  dump_stack+0x5c/0x80
  ___might_sleep.cold.89+0xf5/0x109
  rt_spin_lock+0x3e/0xd0
  ? __blk_add_trace+0x428/0x4b0
  __blk_add_trace+0x428/0x4b0
  blk_add_trace_bio+0x16e/0x1c0
  generic_make_request_checks+0x7e8/0x8c0
  generic_make_request+0x3c/0x420
  ? membarrier_private_expedited+0xd0/0x2b0
  ? lock_release+0x1ca/0x450
  ? submit_bio+0x3c/0x160
  ? _raw_spin_unlock_irqrestore+0x3c/0x80
  submit_bio+0x3c/0x160
  ? rt_mutex_futex_unlock+0x66/0xa0
  iomap_submit_ioend.isra.36+0x4a/0x70
  xfs_vm_writepages+0x65/0x90 [xfs]
  do_writepages+0x41/0xe0
  ? rt_mutex_futex_unlock+0x66/0xa0
  __filemap_fdatawrite_range+0xce/0x110
  xfs_release+0x11c/0x160 [xfs]
  __fput+0xd5/0x270
  task_work_run+0xa1/0xd0
  exit_to_usermode_loop+0x14d/0x160
  do_syscall_64+0x23b/0x240
  entry_SYSCALL_64_after_hwframe+0x6a/0xdf

We replace the get/put_cpu() call by get/put_cpu_light to avoid this bug.

---
 block/blk-cgroup.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

index 663aabfeba18..0a532bb3003c 100644
@@ -1911,7 +1911,7 @@ void blk_cgroup_bio_start(struct bio *bio)
 	struct blkg_iostat_set *bis;
 	unsigned long flags;
 
-	cpu = get_cpu();
+	cpu = get_cpu_light();
 	bis = per_cpu_ptr(bio->bi_blkg->iostat_cpu, cpu);
 	flags = u64_stats_update_begin_irqsave(&bis->sync);
 

@@ -1928,7 +1928,7 @@ void blk_cgroup_bio_start(struct bio *bio)
 	u64_stats_update_end_irqrestore(&bis->sync, flags);
 	if (cgroup_subsys_on_dfl(io_cgrp_subsys))
 		cgroup_rstat_updated(bio->bi_blkg->blkcg->css.cgroup, cpu);
-	put_cpu();
+	put_cpu_light();
 }
 
 static int __init blkcg_init(void)

2.27.0

You're right. I double-checked and noticed my tree didn't have the call to u64_stats_update_begin_irqsave. Only patch 2 is necessary indeed.

TRACE_EVENT disables preemption before calling the callback. Because of that blktrace triggers the following bug under PREEMPT_RT:

 BUG: sleeping function called from invalid context at kernel/locking/spinlock_rt.c:35
 in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 119, name: kworker/u2:2
 5 locks held by kworker/u2:2/119:
  #0: ffff8c2e4a88f538 ((wq_completion)xfs-cil/dm-0){+.+.}-{0:0}, at: process_one_work+0x200/0x450
  #1: ffffab3840ac7e68 ((work_completion)(&cil->xc_push_work)){+.+.}-{0:0}, at: process_one_work+0x200/0x450
  #2: ffff8c2e4a887128 (&cil->xc_ctx_lock){++++}-{3:3}, at: xlog_cil_push_work+0xb7/0x670 [xfs]
  #3: ffffffffa6a63780 (rcu_read_lock){....}-{1:2}, at: blk_add_trace_bio+0x0/0x1f0
  #4: ffffffffa6610620 (running_trace_lock){+.+.}-{2:2}, at: __blk_add_trace+0x3ef/0x480
 Preemption disabled at:
 [<ffffffffa4d35c05>] migrate_enable+0x45/0x140
 CPU: 0 PID: 119 Comm: kworker/u2:2 Kdump: loaded Not tainted 5.14.0-25.rt21.25.light.el9.x86_64+debug #1
 Hardware name: Red Hat KVM, BIOS 0.5.1 01/01/2011
 Workqueue: xfs-cil/dm-0 xlog_cil_push_work [xfs]
 Call Trace:
  ? migrate_enable+0x45/0x140
  dump_stack_lvl+0x57/0x7d
  ___might_sleep.cold+0xe3/0xf7
  rt_spin_lock+0x3a/0xd0
  ? __blk_add_trace+0x3ef/0x480
  __blk_add_trace+0x3ef/0x480
  blk_add_trace_bio+0x18d/0x1f0
  trace_block_bio_queue+0xb5/0x150
  submit_bio_checks+0x1f0/0x520
  ? sched_clock_cpu+0xb/0x100
  submit_bio_noacct+0x30/0x1d0
  ? bio_associate_blkg+0x66/0x190
  xlog_cil_push_work+0x1b6/0x670 [xfs]
  ? register_lock_class+0x43/0x4f0
  ? xfs_swap_extents+0x5f0/0x5f0 [xfs]
  process_one_work+0x275/0x450
  ? process_one_work+0x200/0x450
  worker_thread+0x55/0x3c0
  ? process_one_work+0x450/0x450
  kthread+0x188/0x1a0
  ? set_kthread_struct+0x40/0x40
  ret_from_fork+0x22/0x30

To avoid this bug, we switch the trace lock to a raw spinlock.

---
 kernel/trace/blktrace.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

index 1183c88634aa..a86e022f7155 100644
@@ -34,7 +34,7 @@ static struct trace_array *blk_tr;
 static bool blk_tracer_enabled __read_mostly;
 
 static LIST_HEAD(running_trace_list);
-static __cacheline_aligned_in_smp DEFINE_SPINLOCK(running_trace_lock);
+static __cacheline_aligned_in_smp DEFINE_RAW_SPINLOCK(running_trace_lock);
 
 /* Select an alternative, minimalistic output than the original one */
 #define TRACE_BLK_OPT_CLASSIC	0x1

@@ -121,12 +121,12 @@ static void trace_note_tsk(struct task_struct *tsk)
 	struct blk_trace *bt;
 
 	tsk->btrace_seq = blktrace_seq;
-	spin_lock_irqsave(&running_trace_lock, flags);
+	raw_spin_lock_irqsave(&running_trace_lock, flags);
 	list_for_each_entry(bt, &running_trace_list, running_list) {
 		trace_note(bt, tsk->pid, BLK_TN_PROCESS, tsk->comm,
 			   sizeof(tsk->comm), 0);
 	}
-	spin_unlock_irqrestore(&running_trace_lock, flags);
+	raw_spin_unlock_irqrestore(&running_trace_lock, flags);
 }
 
 static void trace_note_time(struct blk_trace *bt)

@@ -666,9 +666,9 @@ static int __blk_trace_startstop(struct request_queue *q, int start)
 			blktrace_seq++;
 			smp_mb();
 			bt->trace_state = Blktrace_running;
-			spin_lock_irq(&running_trace_lock);
+			raw_spin_lock_irq(&running_trace_lock);
 			list_add(&bt->running_list, &running_trace_list);
-			spin_unlock_irq(&running_trace_lock);
+			raw_spin_unlock_irq(&running_trace_lock);
 
 			trace_note_time(bt);
 			ret = 0;

@@ -676,9 +676,9 @@ static int __blk_trace_startstop(struct request_queue *q, int start)
 	} else {
 		if (bt->trace_state == Blktrace_running) {
 			bt->trace_state = Blktrace_stopped;
-			spin_lock_irq(&running_trace_lock);
+			raw_spin_lock_irq(&running_trace_lock);
 			list_del_init(&bt->running_list);
-			spin_unlock_irq(&running_trace_lock);
+			raw_spin_unlock_irq(&running_trace_lock);
 			relay_flush(bt->rchan);
 			ret = 0;
 		}

@@ -1608,9 +1608,9 @@ static int blk_trace_remove_queue(struct request_queue *q)
 
 	if (bt->trace_state == Blktrace_running) {
 		bt->trace_state = Blktrace_stopped;
-		spin_lock_irq(&running_trace_lock);
+		raw_spin_lock_irq(&running_trace_lock);
 		list_del_init(&bt->running_list);
-		spin_unlock_irq(&running_trace_lock);
+		raw_spin_unlock_irq(&running_trace_lock);
 		relay_flush(bt->rchan);
 	}
 

2.27.0

On 2021-12-13 09:37:37 [-0300], Wander Lairson Costa wrote:
…

Steven, do you think this is a good idea?

Sebastian

blktrace is actually maintained by Jens.

-- Steve

Looks like there are two patches, and none were CC'ed linux-block. Please resend.

Jens Axboe

Ok, I am going to do it, thanks.

I've seen it, need to think about it.

Sebastian