bpf: bpf_iter_task_next: use __next_thread() rather than next_thread()
authorOleg Nesterov <oleg@redhat.com>
Tue, 14 Nov 2023 16:32:37 +0000 (17:32 +0100)
committerAlexei Starovoitov <ast@kernel.org>
Sun, 19 Nov 2023 19:43:44 +0000 (11:43 -0800)
commit5a34f9dabd9aa567e2d37e1aa27a67f80acfaa1c
treec070c12042571603f0516ae4a95be695cb98dc00
parent2d1618054f25e11c44d189dbff4a60342a4cfb4b
bpf: bpf_iter_task_next: use __next_thread() rather than next_thread()

Lockless use of next_thread() should be avoided, kernel/bpf/task_iter.c
is the last user and the usage is wrong.

bpf_iter_task_next() can loop forever, "kit->pos == kit->task" can never
happen if kit->pos execs. Change this code to use __next_thread().

With or without this change the usage of kit->pos/task and next_task()
doesn't look nice, see the next patch.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20231114163237.GA897@redhat.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
kernel/bpf/task_iter.c