locking/osq_lock: mark an intentional data race
PATCH 1/1 locking/osq_lock: mark an intentional data race
node->next could be accessed concurrently as reported by KCSAN,
BUG: KCSAN: data-race in osq_lock / osq_unlock
write (marked) to 0xffff8bb2f1abbe40 of 8 bytes by task 1138 on cpu 44:
osq_lock+0x149/0x340 kernel/locking/osq_lock.c:143 __mutex_lock+0x277/0xd20 kernel/locking/mutex.c:657 mutex_lock_nested+0x31/0x40 kernfs_iop_getattr+0x58/0x90 vfs_getattr_nosec+0x11a/0x170 vfs_statx_fd+0x54/0x90 __do_sys_newfstat+0x40/0x90 __x64_sys_newfstat+0x3a/0x50 do_syscall_64+0x91/0xb47 entry_SYSCALL_64_after_hwframe+0x49/0xbe
read to 0xffff8bb2f1abbe40 of 8 bytes by task 1150 on cpu 29:
osq_unlock+0xee/0x170 kernel/locking/osq_lock.c:78 __mutex_lock+0xb68/0xd20 kernel/locking/mutex.c:686 mutex_lock_nested+0x31/0x40 kernfs_iop_getattr+0x58/0x90 vfs_getattr_nosec+0x11a/0x170 vfs_statx_fd+0x54/0x90 __do_sys_newfstat+0x40/0x90 __x64_sys_newfstat+0x3a/0x50 do_syscall_64+0x91/0xb47 entry_SYSCALL_64_after_hwframe+0x49/0xbe
This is a false positive. Since even if that load is shattered the code will function correctly -- it checks for any !0 value, any byte composite that is !0 is sufficient. Hence, mark it as an intentional data race using the data_race() macro.
Signed-off-by Qian Cai <[email protected]>
--- kernel/locking/osq_lock.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/locking/osq_lock.c +1 -1
@@ -75,7 +75,7 @@ osq_wait_next(struct optimistic_spin_queue *lock,* wait for either @lock to point to us, through its Step-B, or * wait for a new @node->next from its Step-C. */ -if (node->next) { +if (data_race (node->next)) { next =xchg (&node->next,NULL );if (next)break ;
Signature · 1 line
2.21.0 (Apple Git-122.2)