netdev - [PATCH bpf v3 3/9] bpf: enable access to ax register also from verifier rewrite

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]

Message-Id: <20190102235835.3311-4-daniel@iogearbox.net>
Date:   Thu,  3 Jan 2019 00:58:29 +0100
From:   Daniel Borkmann <daniel@...earbox.net>
To:     ast@...nel.org
Cc:     jannh@...gle.com, davem@...emloft.net,
        jakub.kicinski@...ronome.com, netdev@...r.kernel.org,
        Daniel Borkmann <daniel@...earbox.net>
Subject: [PATCH bpf v3 3/9] bpf: enable access to ax register also from verifier rewrite

Right now we are using BPF ax register in JIT for constant blinding as
well as in interpreter as temporary variable. Verifier will not be able
to use it simply because its use will get overridden from the former in
bpf_jit_blind_insn(). However, it can be made to work in that blinding
will be skipped if there is prior use in either source or destination
register on the instruction. Taking constraints of ax into account, the
verifier is then open to use it in rewrites under some constraints. Note,
ax register already has mappings in every eBPF JIT.

Signed-off-by: Daniel Borkmann <daniel@...earbox.net>
Acked-by: Alexei Starovoitov <ast@...nel.org>
---
 include/linux/filter.h |  7 +------
 kernel/bpf/core.c      | 20 ++++++++++++++++++++
 2 files changed, 21 insertions(+), 6 deletions(-)

diff --git a/include/linux/filter.h b/include/linux/filter.h
index 84a6a98..ad106d8 100644
--- a/include/linux/filter.h
+++ b/include/linux/filter.h
@@ -53,12 +53,7 @@ struct sock_reuseport;
 #define BPF_REG_D	BPF_REG_8	/* data, callee-saved */
 #define BPF_REG_H	BPF_REG_9	/* hlen, callee-saved */
 
-/* Kernel hidden auxiliary/helper register for hardening step.
- * Only used by eBPF JITs. It's nothing more than a temporary
- * register that JITs use internally, only that here it's part
- * of eBPF instructions that have been rewritten for blinding
- * constants. See JIT pre-step in bpf_jit_blind_constants().
- */
+/* Kernel hidden auxiliary/helper register. */
 #define BPF_REG_AX		MAX_BPF_REG
 #define MAX_BPF_EXT_REG		(MAX_BPF_REG + 1)
 #define MAX_BPF_JIT_REG		MAX_BPF_EXT_REG
diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c
index a34312a..f908b93 100644
--- a/kernel/bpf/core.c
+++ b/kernel/bpf/core.c
@@ -858,6 +858,26 @@ static int bpf_jit_blind_insn(const struct bpf_insn *from,
 	BUILD_BUG_ON(BPF_REG_AX  + 1 != MAX_BPF_JIT_REG);
 	BUILD_BUG_ON(MAX_BPF_REG + 1 != MAX_BPF_JIT_REG);
 
+	/* Constraints on AX register:
+	 *
+	 * AX register is inaccessible from user space. It is mapped in
+	 * all JITs, and used here for constant blinding rewrites. It is
+	 * typically "stateless" meaning its contents are only valid within
+	 * the executed instruction, but not across several instructions.
+	 * There are a few exceptions however which are further detailed
+	 * below.
+	 *
+	 * Constant blinding is only used by JITs, not in the interpreter.
+	 * The interpreter uses AX in some occasions as a local temporary
+	 * register e.g. in DIV or MOD instructions.
+	 *
+	 * In restricted circumstances, the verifier can also use the AX
+	 * register for rewrites as long as they do not interfere with
+	 * the above cases!
+	 */
+	if (from->dst_reg == BPF_REG_AX || from->src_reg == BPF_REG_AX)
+		goto out;
+
 	if (from->imm == 0 &&
 	    (from->code == (BPF_ALU   | BPF_MOV | BPF_K) ||
 	     from->code == (BPF_ALU64 | BPF_MOV | BPF_K))) {
-- 
2.9.5