libbpf: Do not close un-owned FD 0 on errors
authorDaniel Xu <dxu@dxuuu.xyz>
Wed, 28 Jul 2021 23:09:21 +0000 (16:09 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 18 Aug 2021 07:06:55 +0000 (09:06 +0200)
[ Upstream commit c34c338a40e4f3b6f80889cd17fd9281784d1c32 ]

Before this patch, btf_new() was liable to close an arbitrary FD 0 if
BTF parsing failed. This was because:

* btf->fd was initialized to 0 through the calloc()
* btf__free() (in the `done` label) closed any FDs >= 0
* btf->fd is left at 0 if parsing fails

This issue was discovered on a system using libbpf v0.3 (without
BTF_KIND_FLOAT support) but with a kernel that had BTF_KIND_FLOAT types
in BTF. Thus, parsing fails.

While this patch technically doesn't fix any issues b/c upstream libbpf
has BTF_KIND_FLOAT support, it'll help prevent issues in the future if
more BTF types are added. It also allow the fix to be backported to
older libbpf's.

Fixes: 3289959b97ca ("libbpf: Support BTF loading and raw data output in both endianness")
Signed-off-by: Daniel Xu <dxu@dxuuu.xyz>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong Song <yhs@fb.com>
Link: https://lore.kernel.org/bpf/5969bb991adedb03c6ae93e051fd2a00d293cf25.1627513670.git.dxu@dxuuu.xyz
Signed-off-by: Sasha Levin <sashal@kernel.org>
tools/lib/bpf/btf.c

index d57e13a13798e2182eb589dee48056d2ccff675f..1d9e5b35524cf5455e223b48e65f92f7b6aa8554 100644 (file)
@@ -805,6 +805,7 @@ static struct btf *btf_new(const void *data, __u32 size, struct btf *base_btf)
        btf->nr_types = 0;
        btf->start_id = 1;
        btf->start_str_off = 0;
+       btf->fd = -1;
 
        if (base_btf) {
                btf->base_btf = base_btf;
@@ -833,8 +834,6 @@ static struct btf *btf_new(const void *data, __u32 size, struct btf *base_btf)
        if (err)
                goto done;
 
-       btf->fd = -1;
-
 done:
        if (err) {
                btf__free(btf);