netdev - Re: [PATCH v2] bpf: btf: Fix a missing-check bug

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 24 Oct 2018 18:22:46 +0000
From:   Martin Lau <kafai@...com>
To:     Wenwen Wang <wang6495@....edu>
CC:     Kangjie Lu <kjlu@....edu>, Alexei Starovoitov <ast@...nel.org>,
        "Daniel Borkmann" <daniel@...earbox.net>,
        "open list:BPF (Safe dynamic programs and tools)" 
        <netdev@...r.kernel.org>,
        "open list:BPF (Safe dynamic programs and tools)" 
        <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2] bpf: btf: Fix a missing-check bug

On Wed, Oct 24, 2018 at 05:26:23PM +0000, Martin Lau wrote:
> On Wed, Oct 24, 2018 at 08:00:19AM -0500, Wenwen Wang wrote:
> > In btf_parse(), the header of the user-space btf data 'btf_data' is firstly
> > parsed and verified through btf_parse_hdr(). In btf_parse_hdr(), the header
> > is copied from user-space 'btf_data' to kernel-space 'btf->hdr' and then
> > verified. If no error happens during the verification process, the whole
> > data of 'btf_data', including the header, is then copied to 'data' in
> > btf_parse(). It is obvious that the header is copied twice here. More
> > importantly, no check is enforced after the second copy to make sure the
> > headers obtained in these two copies are same. Given that 'btf_data'
> > resides in the user space, a malicious user can race to modify the header
> > between these two copies. By doing so, the user can inject inconsistent
> > data, which can cause undefined behavior of the kernel and introduce
> > potential security risk.
btw, I am working on a patch that copies the btf_data before parsing/verifying
the header.  That should avoid this from happening but that will
require a bit more code churns for the bpf branch.

> > 
> > To avoid the above issue, this patch copies the parsed header from
> > 'btf->hdr' to 'data'. The remaining part in 'data' is still copied from the
> > user-space 'btf_data'.
> LGTM.
> 
> Acked-by: Martin KaFai Lau <kafai@...com>
> 
> > 
> > Signed-off-by: Wenwen Wang <wang6495@....edu>
> > ---
> >  kernel/bpf/btf.c | 11 ++++++++++-
> >  1 file changed, 10 insertions(+), 1 deletion(-)
> > 
> > diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
> > index 378cef7..b52a834a 100644
> > --- a/kernel/bpf/btf.c
> > +++ b/kernel/bpf/btf.c
> > @@ -2152,6 +2152,7 @@ static struct btf *btf_parse(void __user *btf_data, u32 btf_data_size,
> >  	struct btf_verifier_env *env = NULL;
> >  	struct bpf_verifier_log *log;
> >  	struct btf *btf = NULL;
> > +	u32 hdr_len;
> >  	u8 *data;
> >  	int err;
> >  
> > @@ -2200,7 +2201,15 @@ static struct btf *btf_parse(void __user *btf_data, u32 btf_data_size,
> >  	btf->data_size = btf_data_size;
> >  	btf->nohdr_data = btf->data + btf->hdr.hdr_len;
> >  
> > -	if (copy_from_user(data, btf_data, btf_data_size)) {
> > +	/*
> > +	 * The header at btf_data could be modified by a malicious user
> > +	 * after it is parsed. So we copy the parsed header here. The
> > +	 * remaining part is still copied from btf_data.
> > +	 */
> > +	hdr_len = min_t(u32, btf->hdr.hdr_len, sizeof(btf->hdr));
> > +	memcpy(data, &btf->hdr, hdr_len);
> > +	if (copy_from_user(data + hdr_len, (u8 __user *)btf_data + hdr_len,
> > +				btf_data_size - hdr_len)) {
> >  		err = -EFAULT;
> >  		goto errout;
> >  	}
> > -- 
> > 2.7.4
> >