bpf: remove explicit handling of 0 for arg2 in bpf_probe_read
authorGianluca Borello <g.borello@gmail.com>
Wed, 22 Nov 2017 18:32:54 +0000 (18:32 +0000)
committerDaniel Borkmann <daniel@iogearbox.net>
Wed, 22 Nov 2017 20:40:54 +0000 (21:40 +0100)
Commit 9c019e2bc4b2 ("bpf: change helper bpf_probe_read arg2 type to
ARG_CONST_SIZE_OR_ZERO") changed arg2 type to ARG_CONST_SIZE_OR_ZERO to
simplify writing bpf programs by taking advantage of the new semantics
introduced for ARG_CONST_SIZE_OR_ZERO which allows <!NULL, 0> arguments.

In order to prevent the helper from actually passing a NULL pointer to
probe_kernel_read, which can happen when <NULL, 0> is passed to the helper,
the commit also introduced an explicit check against size == 0.

After the recent introduction of the ARG_PTR_TO_MEM_OR_NULL type,
bpf_probe_read can not receive a pair of <NULL, 0> arguments anymore, thus
the check is not needed anymore and can be removed, since probe_kernel_read
can correctly handle a <!NULL, 0> call. This also fixes the semantics of
the helper before it gets officially released and bpf programs start
relying on this check.

Fixes: 9c019e2bc4b2 ("bpf: change helper bpf_probe_read arg2 type to ARG_CONST_SIZE_OR_ZERO")
Signed-off-by: Gianluca Borello <g.borello@gmail.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
kernel/trace/bpf_trace.c

index a5580c6708668033313e9c5185f2122d364ab72f..728909f7951ca3e0299f4fb1c67c24f6026934c8 100644 (file)
@@ -78,16 +78,12 @@ EXPORT_SYMBOL_GPL(trace_call_bpf);
 
 BPF_CALL_3(bpf_probe_read, void *, dst, u32, size, const void *, unsafe_ptr)
 {
-       int ret = 0;
-
-       if (unlikely(size == 0))
-               goto out;
+       int ret;
 
        ret = probe_kernel_read(dst, unsafe_ptr, size);
        if (unlikely(ret < 0))
                memset(dst, 0, size);
 
- out:
        return ret;
 }