In the CI, we're seeing tests failing due to excessive would_deadlock
transaction restarts - the tracepoint now includes the lock cycle that
occured.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>