After Paul's recent improvement to syzkaller to improve coverage for
bpf_mprog and tcx, it hit a splat that the program limit was surpassed.
What happened is that the maximum number of progs got added, followed
by another prog add request which adds with BPF_F_BEFORE flag relative
to the last program in the array. The idx >= bpf_mprog_max() check in
bpf_mprog_attach() still passes because the index is below the maximum
but the maximum will be surpassed. We need to add a check upfront for
insertions to catch this situation.
Fixes: 053c8e1f235d ("bpf: Add generic attach/detach/query API for multi-progs")
Reported-by: syzbot+baa44e3dbbe48e05c1ad@xxxxxxxxxxxxxxxxxxxxxxxxx
Reported-by: syzbot+b97d20ed568ce0951a06@xxxxxxxxxxxxxxxxxxxxxxxxx
Reported-by: syzbot+2558ca3567a77b7af4e3@xxxxxxxxxxxxxxxxxxxxxxxxx
Co-developed-by: Nikolay Aleksandrov <razor@xxxxxxxxxxxxx>
Signed-off-by: Nikolay Aleksandrov <razor@xxxxxxxxxxxxx>
Signed-off-by: Daniel Borkmann <daniel@xxxxxxxxxxxxx>
Tested-by: syzbot+baa44e3dbbe48e05c1ad@xxxxxxxxxxxxxxxxxxxxxxxxx
Tested-by: syzbot+b97d20ed568ce0951a06@xxxxxxxxxxxxxxxxxxxxxxxxx
Link: https://github.com/google/syzkaller/pull/4207
---
kernel/bpf/mprog.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/kernel/bpf/mprog.c b/kernel/bpf/mprog.c
index 32d2c4829eb8..007d98c799e2 100644
--- a/kernel/bpf/mprog.c
+++ b/kernel/bpf/mprog.c
@@ -253,6 +253,9 @@ int bpf_mprog_attach(struct bpf_mprog_entry *entry,
goto out;
}
idx = tidx;
+ } else if (bpf_mprog_total(entry) == bpf_mprog_max()) {
+ ret = -ERANGE;
+ goto out;
}
if (flags & BPF_F_BEFORE) {
tidx = bpf_mprog_pos_before(entry, &rtuple);
--
2.34.1
