| Message ID | 20221201183406.1203621-2-davemarchevsky@fb.com (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 78b037bd402df8eca0f45ef003c6d0ab25a26ecc |
| Delegated to: | BPF |
| Headers | show |
| Series | [v2,bpf-next,1/2] bpf: Fix release_on_unlock release logic for multiple refs | expand |
On 12/1/22 10:34 AM, Dave Marchevsky wrote: > Modify list_push_pop_multiple to alloc and insert nodes 2-at-a-time. > Without the previous patch's fix, this block of code: > > bpf_spin_lock(lock); > bpf_list_push_front(head, &f[i]->node); > bpf_list_push_front(head, &f[i + 1]->node); > bpf_spin_unlock(lock); > > would fail check_reference_leak check as release_on_unlock logic would miss > a ref that should've been released. > > Signed-off-by: Dave Marchevsky <davemarchevsky@fb.com> > cc: Kumar Kartikeya Dwivedi <memxor@gmail.com> Acked-by: Yonghong Song <yhs@fb.com>
diff --git a/tools/testing/selftests/bpf/progs/linked_list.c b/tools/testing/selftests/bpf/progs/linked_list.c index 2c7b615c6d41..4ad88da5cda2 100644 --- a/tools/testing/selftests/bpf/progs/linked_list.c +++ b/tools/testing/selftests/bpf/progs/linked_list.c @@ -99,13 +99,28 @@ int list_push_pop_multiple(struct bpf_spin_lock *lock, struct bpf_list_head *hea struct foo *f[8], *pf; int i; - for (i = 0; i < ARRAY_SIZE(f); i++) { + /* Loop following this check adds nodes 2-at-a-time in order to + * validate multiple release_on_unlock release logic + */ + if (ARRAY_SIZE(f) % 2) + return 10; + + for (i = 0; i < ARRAY_SIZE(f); i += 2) { f[i] = bpf_obj_new(typeof(**f)); if (!f[i]) return 2; f[i]->data = i; + + f[i + 1] = bpf_obj_new(typeof(**f)); + if (!f[i + 1]) { + bpf_obj_drop(f[i]); + return 9; + } + f[i + 1]->data = i + 1; + bpf_spin_lock(lock); bpf_list_push_front(head, &f[i]->node); + bpf_list_push_front(head, &f[i + 1]->node); bpf_spin_unlock(lock); }
Modify list_push_pop_multiple to alloc and insert nodes 2-at-a-time. Without the previous patch's fix, this block of code: bpf_spin_lock(lock); bpf_list_push_front(head, &f[i]->node); bpf_list_push_front(head, &f[i + 1]->node); bpf_spin_unlock(lock); would fail check_reference_leak check as release_on_unlock logic would miss a ref that should've been released. Signed-off-by: Dave Marchevsky <davemarchevsky@fb.com> cc: Kumar Kartikeya Dwivedi <memxor@gmail.com> --- tools/testing/selftests/bpf/progs/linked_list.c | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-)