2022-05-17 00:09:41

by Eugene Syromiatnikov

Subject: [PATCH bpf 1/4] bpf_trace: check size for overflow in bpf_kprobe_multi_link_attach

Check that size would not overflow before calculation (and return
-EOVERFLOW if it will), to prevent potential out-of-bounds write
with the following copy_from_user. Add the same check
to kprobe_multi_resolve_syms in case it will be called from elsewhere
in the future.

Fixes: 0dcac272540613d4 ("bpf: Add multi kprobe link")
Signed-off-by: Eugene Syromiatnikov <[email protected]>
---
kernel/trace/bpf_trace.c | 4 ++++
1 file changed, 4 insertions(+)

diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
index d8553f4..e90c4ce7 100644
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -2358,6 +2358,8 @@ kprobe_multi_resolve_syms(const void __user *usyms, u32 cnt,
unsigned int i;
char *func;

+ if (check_mul_overflow(cnt, sizeof(*syms), &size))
+ return -EOVERFLOW;
size = cnt * sizeof(*syms);
syms = kvzalloc(size, GFP_KERNEL);
if (!syms)
@@ -2429,6 +2431,8 @@ int bpf_kprobe_multi_link_attach(const union bpf_attr *attr, struct bpf_prog *pr
if (!cnt)
return -EINVAL;

+ if (check_mul_overflow(cnt, (u32)sizeof(*addrs), &size))
+ return -EOVERFLOW;
size = cnt * sizeof(*addrs);
addrs = kvmalloc(size, GFP_KERNEL);
if (!addrs)
--
2.1.4

2022-05-17 01:18:00

by Eugene Syromiatnikov

Subject: Re: [PATCH bpf 1/4] bpf_trace: check size for overflow in bpf_kprobe_multi_link_attach

On Mon, May 16, 2022 at 11:34:45PM +0200, Jiri Olsa wrote:
> On Mon, May 16, 2022 at 08:27:08PM +0200, Eugene Syromiatnikov wrote:
> > + if (check_mul_overflow(cnt, sizeof(*syms), &size))
> > + return -EOVERFLOW;
>
> there was an update already:
>
> 0236fec57a15 bpf: Resolve symbols with ftrace_lookup_symbols for kprobe multi link
>
> so this won't apply anymore, could you please rebase on top of the latest bpf-next/master?

The issue that this specific check has to go in 4.18, as it covers
possible out-of-bounds write, I'm not sure how to handle it, have
a branch where it is merged manually?

2022-05-17 13:29:06

by Alexei Starovoitov

Subject: Re: [PATCH bpf 1/4] bpf_trace: check size for overflow in bpf_kprobe_multi_link_attach

On Tue, May 17, 2022 at 12:49:34AM +0200, Eugene Syromiatnikov wrote:
> On Mon, May 16, 2022 at 11:34:45PM +0200, Jiri Olsa wrote:
> > On Mon, May 16, 2022 at 08:27:08PM +0200, Eugene Syromiatnikov wrote:
> > > + if (check_mul_overflow(cnt, sizeof(*syms), &size))
> > > + return -EOVERFLOW;
> >
> > there was an update already:
> >
> > 0236fec57a15 bpf: Resolve symbols with ftrace_lookup_symbols for kprobe multi link
> >
> > so this won't apply anymore, could you please rebase on top of the latest bpf-next/master?
>
> The issue that this specific check has to go in 4.18, as it covers
> possible out-of-bounds write, I'm not sure how to handle it, have
> a branch where it is merged manually?

As Jiri said, please use bpf-next.

2022-05-17 19:58:47

by Jiri Olsa

Subject: Re: [PATCH bpf 1/4] bpf_trace: check size for overflow in bpf_kprobe_multi_link_attach

On Mon, May 16, 2022 at 08:27:08PM +0200, Eugene Syromiatnikov wrote:
> Check that size would not overflow before calculation (and return
> -EOVERFLOW if it will), to prevent potential out-of-bounds write
> with the following copy_from_user. Add the same check
> to kprobe_multi_resolve_syms in case it will be called from elsewhere
> in the future.
>
> Fixes: 0dcac272540613d4 ("bpf: Add multi kprobe link")
> Signed-off-by: Eugene Syromiatnikov <[email protected]>
> ---
> kernel/trace/bpf_trace.c | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/kernel/trace/bpf_trace.c b/kernel/trace/bpf_trace.c
> index d8553f4..e90c4ce7 100644
> --- a/kernel/trace/bpf_trace.c
> +++ b/kernel/trace/bpf_trace.c
> @@ -2358,6 +2358,8 @@ kprobe_multi_resolve_syms(const void __user *usyms, u32 cnt,
> unsigned int i;
> char *func;
>
> + if (check_mul_overflow(cnt, sizeof(*syms), &size))
> + return -EOVERFLOW;

there was an update already:

0236fec57a15 bpf: Resolve symbols with ftrace_lookup_symbols for kprobe multi link

so this won't apply anymore, could you please rebase on top of the latest bpf-next/master?

thanks,
jirka

> size = cnt * sizeof(*syms);
> syms = kvzalloc(size, GFP_KERNEL);
> if (!syms)
> @@ -2429,6 +2431,8 @@ int bpf_kprobe_multi_link_attach(const union bpf_attr *attr, struct bpf_prog *pr
> if (!cnt)
> return -EINVAL;
>
> + if (check_mul_overflow(cnt, (u32)sizeof(*addrs), &size))
> + return -EOVERFLOW;
> size = cnt * sizeof(*addrs);
> addrs = kvmalloc(size, GFP_KERNEL);
> if (!addrs)
> --
> 2.1.4
>