OSDN Git Service

bpf: Initialize storage pointers to NULL to prevent freeing garbage pointer
authorAndrii Nakryiko <andriin@fb.com>
Mon, 9 Mar 2020 22:27:55 +0000 (15:27 -0700)
committerAlexei Starovoitov <ast@kernel.org>
Tue, 10 Mar 2020 02:56:48 +0000 (19:56 -0700)
Local storage array isn't initialized, so if cgroup storage allocation fails
for BPF_CGROUP_STORAGE_SHARED, error handling code will attempt to free
uninitialized pointer for BPF_CGROUP_STORAGE_PERCPU storage type. Avoid this
by always initializing storage pointers to NULLs.

Fixes: 8bad74f9840f ("bpf: extend cgroup bpf core to allow multiple cgroup storage types")
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20200309222756.1018737-1-andriin@fb.com
kernel/bpf/cgroup.c

index 9a500fa..b2bc4c3 100644 (file)
@@ -302,8 +302,8 @@ int __cgroup_bpf_attach(struct cgroup *cgrp, struct bpf_prog *prog,
        u32 saved_flags = (flags & (BPF_F_ALLOW_OVERRIDE | BPF_F_ALLOW_MULTI));
        struct list_head *progs = &cgrp->bpf.progs[type];
        struct bpf_prog *old_prog = NULL;
-       struct bpf_cgroup_storage *storage[MAX_BPF_CGROUP_STORAGE_TYPE],
-               *old_storage[MAX_BPF_CGROUP_STORAGE_TYPE] = {NULL};
+       struct bpf_cgroup_storage *storage[MAX_BPF_CGROUP_STORAGE_TYPE] = {};
+       struct bpf_cgroup_storage *old_storage[MAX_BPF_CGROUP_STORAGE_TYPE] = {};
        struct bpf_prog_list *pl, *replace_pl = NULL;
        enum bpf_cgroup_storage_type stype;
        int err;