kernel_optimize_test/kernel/locking/spinlock_debug.c
Marco Elver 1a365e8223 locking/spinlock/debug: Fix various data races
This fixes various data races in spinlock_debug. By testing with KCSAN,
it is observable that the console gets spammed with data races reports,
suggesting these are extremely frequent.

Example data race report:

  read to 0xffff8ab24f403c48 of 4 bytes by task 221 on cpu 2:
   debug_spin_lock_before kernel/locking/spinlock_debug.c:85 [inline]
   do_raw_spin_lock+0x9b/0x210 kernel/locking/spinlock_debug.c:112
   __raw_spin_lock include/linux/spinlock_api_smp.h:143 [inline]
   _raw_spin_lock+0x39/0x40 kernel/locking/spinlock.c:151
   spin_lock include/linux/spinlock.h:338 [inline]
   get_partial_node.isra.0.part.0+0x32/0x2f0 mm/slub.c:1873
   get_partial_node mm/slub.c:1870 [inline]
  <snip>

  write to 0xffff8ab24f403c48 of 4 bytes by task 167 on cpu 3:
   debug_spin_unlock kernel/locking/spinlock_debug.c:103 [inline]
   do_raw_spin_unlock+0xc9/0x1a0 kernel/locking/spinlock_debug.c:138
   __raw_spin_unlock_irqrestore include/linux/spinlock_api_smp.h:159 [inline]
   _raw_spin_unlock_irqrestore+0x2d/0x50 kernel/locking/spinlock.c:191
   spin_unlock_irqrestore include/linux/spinlock.h:393 [inline]
   free_debug_processing+0x1b3/0x210 mm/slub.c:1214
   __slab_free+0x292/0x400 mm/slub.c:2864
  <snip>

As a side-effect, with KCSAN, this eventually locks up the console, most
likely due to deadlock, e.g. .. -> printk lock -> spinlock_debug ->
KCSAN detects data race -> kcsan_print_report() -> printk lock ->
deadlock.

This fix will 1) avoid the data races, and 2) allow using lock debugging
together with KCSAN.

Reported-by: Qian Cai <cai@lca.pw>
Signed-off-by: Marco Elver <elver@google.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Paul E. McKenney <paulmck@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will.deacon@arm.com>
Link: https://lkml.kernel.org/r/20191120155715.28089-1-elver@google.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2019-11-29 08:03:27 +01:00

231 lines
5.7 KiB
C

/*
* Copyright 2005, Red Hat, Inc., Ingo Molnar
* Released under the General Public License (GPL).
*
* This file contains the spinlock/rwlock implementations for
* DEBUG_SPINLOCK.
*/
#include <linux/spinlock.h>
#include <linux/nmi.h>
#include <linux/interrupt.h>
#include <linux/debug_locks.h>
#include <linux/delay.h>
#include <linux/export.h>
void __raw_spin_lock_init(raw_spinlock_t *lock, const char *name,
struct lock_class_key *key)
{
#ifdef CONFIG_DEBUG_LOCK_ALLOC
/*
* Make sure we are not reinitializing a held lock:
*/
debug_check_no_locks_freed((void *)lock, sizeof(*lock));
lockdep_init_map(&lock->dep_map, name, key, 0);
#endif
lock->raw_lock = (arch_spinlock_t)__ARCH_SPIN_LOCK_UNLOCKED;
lock->magic = SPINLOCK_MAGIC;
lock->owner = SPINLOCK_OWNER_INIT;
lock->owner_cpu = -1;
}
EXPORT_SYMBOL(__raw_spin_lock_init);
void __rwlock_init(rwlock_t *lock, const char *name,
struct lock_class_key *key)
{
#ifdef CONFIG_DEBUG_LOCK_ALLOC
/*
* Make sure we are not reinitializing a held lock:
*/
debug_check_no_locks_freed((void *)lock, sizeof(*lock));
lockdep_init_map(&lock->dep_map, name, key, 0);
#endif
lock->raw_lock = (arch_rwlock_t) __ARCH_RW_LOCK_UNLOCKED;
lock->magic = RWLOCK_MAGIC;
lock->owner = SPINLOCK_OWNER_INIT;
lock->owner_cpu = -1;
}
EXPORT_SYMBOL(__rwlock_init);
static void spin_dump(raw_spinlock_t *lock, const char *msg)
{
struct task_struct *owner = READ_ONCE(lock->owner);
if (owner == SPINLOCK_OWNER_INIT)
owner = NULL;
printk(KERN_EMERG "BUG: spinlock %s on CPU#%d, %s/%d\n",
msg, raw_smp_processor_id(),
current->comm, task_pid_nr(current));
printk(KERN_EMERG " lock: %pS, .magic: %08x, .owner: %s/%d, "
".owner_cpu: %d\n",
lock, READ_ONCE(lock->magic),
owner ? owner->comm : "<none>",
owner ? task_pid_nr(owner) : -1,
READ_ONCE(lock->owner_cpu));
dump_stack();
}
static void spin_bug(raw_spinlock_t *lock, const char *msg)
{
if (!debug_locks_off())
return;
spin_dump(lock, msg);
}
#define SPIN_BUG_ON(cond, lock, msg) if (unlikely(cond)) spin_bug(lock, msg)
static inline void
debug_spin_lock_before(raw_spinlock_t *lock)
{
SPIN_BUG_ON(READ_ONCE(lock->magic) != SPINLOCK_MAGIC, lock, "bad magic");
SPIN_BUG_ON(READ_ONCE(lock->owner) == current, lock, "recursion");
SPIN_BUG_ON(READ_ONCE(lock->owner_cpu) == raw_smp_processor_id(),
lock, "cpu recursion");
}
static inline void debug_spin_lock_after(raw_spinlock_t *lock)
{
WRITE_ONCE(lock->owner_cpu, raw_smp_processor_id());
WRITE_ONCE(lock->owner, current);
}
static inline void debug_spin_unlock(raw_spinlock_t *lock)
{
SPIN_BUG_ON(lock->magic != SPINLOCK_MAGIC, lock, "bad magic");
SPIN_BUG_ON(!raw_spin_is_locked(lock), lock, "already unlocked");
SPIN_BUG_ON(lock->owner != current, lock, "wrong owner");
SPIN_BUG_ON(lock->owner_cpu != raw_smp_processor_id(),
lock, "wrong CPU");
WRITE_ONCE(lock->owner, SPINLOCK_OWNER_INIT);
WRITE_ONCE(lock->owner_cpu, -1);
}
/*
* We are now relying on the NMI watchdog to detect lockup instead of doing
* the detection here with an unfair lock which can cause problem of its own.
*/
void do_raw_spin_lock(raw_spinlock_t *lock)
{
debug_spin_lock_before(lock);
arch_spin_lock(&lock->raw_lock);
mmiowb_spin_lock();
debug_spin_lock_after(lock);
}
int do_raw_spin_trylock(raw_spinlock_t *lock)
{
int ret = arch_spin_trylock(&lock->raw_lock);
if (ret) {
mmiowb_spin_lock();
debug_spin_lock_after(lock);
}
#ifndef CONFIG_SMP
/*
* Must not happen on UP:
*/
SPIN_BUG_ON(!ret, lock, "trylock failure on UP");
#endif
return ret;
}
void do_raw_spin_unlock(raw_spinlock_t *lock)
{
mmiowb_spin_unlock();
debug_spin_unlock(lock);
arch_spin_unlock(&lock->raw_lock);
}
static void rwlock_bug(rwlock_t *lock, const char *msg)
{
if (!debug_locks_off())
return;
printk(KERN_EMERG "BUG: rwlock %s on CPU#%d, %s/%d, %p\n",
msg, raw_smp_processor_id(), current->comm,
task_pid_nr(current), lock);
dump_stack();
}
#define RWLOCK_BUG_ON(cond, lock, msg) if (unlikely(cond)) rwlock_bug(lock, msg)
void do_raw_read_lock(rwlock_t *lock)
{
RWLOCK_BUG_ON(lock->magic != RWLOCK_MAGIC, lock, "bad magic");
arch_read_lock(&lock->raw_lock);
}
int do_raw_read_trylock(rwlock_t *lock)
{
int ret = arch_read_trylock(&lock->raw_lock);
#ifndef CONFIG_SMP
/*
* Must not happen on UP:
*/
RWLOCK_BUG_ON(!ret, lock, "trylock failure on UP");
#endif
return ret;
}
void do_raw_read_unlock(rwlock_t *lock)
{
RWLOCK_BUG_ON(lock->magic != RWLOCK_MAGIC, lock, "bad magic");
arch_read_unlock(&lock->raw_lock);
}
static inline void debug_write_lock_before(rwlock_t *lock)
{
RWLOCK_BUG_ON(lock->magic != RWLOCK_MAGIC, lock, "bad magic");
RWLOCK_BUG_ON(lock->owner == current, lock, "recursion");
RWLOCK_BUG_ON(lock->owner_cpu == raw_smp_processor_id(),
lock, "cpu recursion");
}
static inline void debug_write_lock_after(rwlock_t *lock)
{
WRITE_ONCE(lock->owner_cpu, raw_smp_processor_id());
WRITE_ONCE(lock->owner, current);
}
static inline void debug_write_unlock(rwlock_t *lock)
{
RWLOCK_BUG_ON(lock->magic != RWLOCK_MAGIC, lock, "bad magic");
RWLOCK_BUG_ON(lock->owner != current, lock, "wrong owner");
RWLOCK_BUG_ON(lock->owner_cpu != raw_smp_processor_id(),
lock, "wrong CPU");
WRITE_ONCE(lock->owner, SPINLOCK_OWNER_INIT);
WRITE_ONCE(lock->owner_cpu, -1);
}
void do_raw_write_lock(rwlock_t *lock)
{
debug_write_lock_before(lock);
arch_write_lock(&lock->raw_lock);
debug_write_lock_after(lock);
}
int do_raw_write_trylock(rwlock_t *lock)
{
int ret = arch_write_trylock(&lock->raw_lock);
if (ret)
debug_write_lock_after(lock);
#ifndef CONFIG_SMP
/*
* Must not happen on UP:
*/
RWLOCK_BUG_ON(!ret, lock, "trylock failure on UP");
#endif
return ret;
}
void do_raw_write_unlock(rwlock_t *lock)
{
debug_write_unlock(lock);
arch_write_unlock(&lock->raw_lock);
}