Messages in this thread Patch in this message |  | | From | Guenter Roeck <> | Subject | [PATCH] locking: Fix runtime warning in ww mutex selftest | Date | Tue, 2 Oct 2018 14:48:49 -0700 |
| |
If CONFIG_WW_MUTEX_SELFTEST is enabled, booting an image in an arm64 virtual machine results in the following traceback if 8 CPUs are enabled.
DEBUG_LOCKS_WARN_ON(__owner_task(owner) != current) WARNING: CPU: 2 PID: 537 at kernel/locking/mutex.c:1033 __mutex_unlock_slowpath+0x1a8/0x2e0 Modules linked in: CPU: 2 PID: 537 Comm: kworker/u16:11 Not tainted 4.18.12-rc1-00133-g74ba23ae47e5 #1 Hardware name: linux,dummy-virt (DT) Workqueue: test-ww_mutex test_cycle_work pstate: 40000005 (nZcv daif -PAN -UAO) pc : __mutex_unlock_slowpath+0x1a8/0x2e0 lr : __mutex_unlock_slowpath+0x1a8/0x2e0 sp : ffff00000acbbc50 x29: ffff00000acbbc50 x28: 0000000000000000 x27: 0000000000000000 x26: ffff000009fc1d80 x25: ffff000009f86558 x24: ffff000008f8d410 x23: ffff000009f86000 x22: ffff00000926f000 x21: ffff00000acbbcb8 x20: ffff80001be208c8 x19: 0000000000000000 x18: ffffffffffffffff x17: 0000000000000000 x16: ffff80001c2a0000 x15: ffff00000926f808 x14: ffff000089f85d3f x13: ffff000009f85d4d x12: ffff00000928a000 x11: 0000000005f5e0ff x10: ffff00000acbb8a0 x9 : 0000000000000000 x8 : ffff00000926f808 x7 : ffff000008149180 x6 : 0000000000000000 x5 : 0000000000000000 x4 : 0000000000000000 x3 : ffffffffffffffff x2 : ffff00000928ab40 x1 : 6172e063a21fe300 x0 : 0000000000000000 Call trace: __mutex_unlock_slowpath+0x1a8/0x2e0 ww_mutex_unlock+0x48/0xa0 test_cycle_work+0xf8/0x200 process_one_work+0x258/0x410 worker_thread+0x40/0x458 kthread+0x128/0x130 ret_from_fork+0x10/0x18 irq event stamp: 1305 _raw_spin_unlock_irq+0x2c/0x60 __schedule+0xbc/0x780 __do_softirq+0x1e8/0x260 irq_exit+0x144/0x150
If requesting b_mutex fails with -EDEADLK, the error variable is reassigned to the return value from calling ww_mutex_lock on a_mutex again. If this call fails, a_mutex is not locked. It is, however, unconditionally unlocked subsequently, causing the reported warning. Fix the problem by using two error variables.
With this change, the selftest still fails as follows.
cyclic deadlock not resolved, ret[7/8] = -35
However, the traceback is gone.
Fixes: d1b42b800e5d0 ("locking/ww_mutex: Add kselftests for resolving ww_mutex cyclic deadlocks") Cc: Will Deacon <will.deacon@arm.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ingo Molnar <mingo@kernel.org> Signed-off-by: Guenter Roeck <linux@roeck-us.net> --- kernel/locking/test-ww_mutex.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-)
diff --git a/kernel/locking/test-ww_mutex.c b/kernel/locking/test-ww_mutex.c index 0be047dbd897..65a3b7e55b9f 100644 --- a/kernel/locking/test-ww_mutex.c +++ b/kernel/locking/test-ww_mutex.c @@ -260,7 +260,7 @@ static void test_cycle_work(struct work_struct *work) { struct test_cycle *cycle = container_of(work, typeof(*cycle), work); struct ww_acquire_ctx ctx; - int err; + int err, erra = 0; ww_acquire_init(&ctx, &ww_class); ww_mutex_lock(&cycle->a_mutex, &ctx); @@ -270,17 +270,19 @@ static void test_cycle_work(struct work_struct *work) err = ww_mutex_lock(cycle->b_mutex, &ctx); if (err == -EDEADLK) { + err = 0; ww_mutex_unlock(&cycle->a_mutex); ww_mutex_lock_slow(cycle->b_mutex, &ctx); - err = ww_mutex_lock(&cycle->a_mutex, &ctx); + erra = ww_mutex_lock(&cycle->a_mutex, &ctx); } if (!err) ww_mutex_unlock(cycle->b_mutex); - ww_mutex_unlock(&cycle->a_mutex); + if (!erra) + ww_mutex_unlock(&cycle->a_mutex); ww_acquire_fini(&ctx); - cycle->result = err; + cycle->result = err ?: erra; } static int __test_cycle(unsigned int nthreads) -- 2.7.4
|  |