netdev - [PATCH bpf 8/9] tools/bpftool: generate data section struct with conservative alignment

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 for Android: free password hash cracker in your pocket

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

Message-ID: <20200813071722.2213397-9-andriin@fb.com>
Date:   Thu, 13 Aug 2020 00:17:21 -0700
From:   Andrii Nakryiko <andriin@...com>
To:     <bpf@...r.kernel.org>, <netdev@...r.kernel.org>, <ast@...com>,
        <daniel@...earbox.net>
CC:     <andrii.nakryiko@...il.com>, <kernel-team@...com>,
        Andrii Nakryiko <andriin@...com>
Subject: [PATCH bpf 8/9] tools/bpftool: generate data section struct with conservative alignment

The comment in the code describes this in good details. Generate such a memory
layout that would work both on 32-bit and 64-bit architectures for user-space.

Signed-off-by: Andrii Nakryiko <andriin@...com>
---
 tools/bpf/bpftool/gen.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/tools/bpf/bpftool/gen.c b/tools/bpf/bpftool/gen.c
index 8a4c2b3b0cd6..17507fba9eb2 100644
--- a/tools/bpf/bpftool/gen.c
+++ b/tools/bpf/bpftool/gen.c
@@ -143,6 +143,20 @@ static int codegen_datasec_def(struct bpf_object *obj,
 			      var_name, align);
 			return -EINVAL;
 		}
+		/* Assume 32-bit architectures when generating data section
+		 * struct memory layout. Given bpftool can't know which target
+		 * host architecture it's emitting skeleton for, we need to be
+		 * conservative and assume 32-bit one to ensure enough padding
+		 * bytes are generated for pointer and long types. This will
+		 * still work correctly for 64-bit architectures, because in
+		 * the worst case we'll generate unnecessary padding field,
+		 * which on 64-bit architectures is not strictly necessary and
+		 * would be handled by natural 8-byte alignment. But it still
+		 * will be a correct memory layout, based on recorded offsets
+		 * in BTF.
+		 */
+		if (align > 4)
+			align = 4;
 
 		align_off = (off + align - 1) / align * align;
 		if (align_off != need_off) {
-- 
2.24.1