linux-kernel - Re: BUG: KCSAN: data-race in add_device

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: <CAHmME9phcxQ8gLpBLVwyJszPdnpxOdeTgcDYzno-TLdV15HZcw@mail.gmail.com>
Date:   Mon, 7 Feb 2022 23:20:47 +0100
From:   "Jason A. Donenfeld" <Jason@...c4.com>
To:     Jann Horn <jannh@...gle.com>
Cc:     pmenzel@...gen.mpg.de, "Theodore Y. Ts'o" <tytso@....edu>,
        LKML <linux-kernel@...r.kernel.org>,
        Dominik Brodowski <linux@...inikbrodowski.net>
Subject: Re: BUG: KCSAN: data-race in add_device_randomness+0x20d/0x290

Narrowing this a bit by looking at the disassembly from as similar as
a config as I could infer, there are three calls to _tsan_read1():

1. ... = lfsr;
2. ... = dest_buf[i % CHACHA_KEY_SIZE];
3. ... ^= src_buf[i % len] ...

1. lfsr, as you mentioned, appears protected by primary_crng.lock.
2. dest_buf is &primary_crng.state, which /should/ be protected by
primary_crng.lock. The only place where it's not is rand_initialize(),
but this is an early boot function that should complete well in
advance of where we are.
3. That would be tsk->se.sum_exec_runtime...