MIME-Version: 1.0
References: <20211012082708.121931-1-iivanov@suse.de> <YWVKAk4h5bsUA3b6@light.dominikbrodowski.net>
 <YaivhAV8LouB0zGV@light.dominikbrodowski.net>
In-Reply-To: <YaivhAV8LouB0zGV@light.dominikbrodowski.net>
From:   "Jason A. Donenfeld" <Jason@zx2c4.com>
Date:   Thu, 2 Dec 2021 11:55:10 -0500
Message-ID: <CAHmME9qxBeBzfKCjzfAFX9ZWAGKv1TKCQw3x22d_DmJtaAewLw@mail.gmail.com>
Subject: Re: [PATCH v3, resend] random: fix crash on multiple early calls to add_bootloader_randomness()
To:     linux@dominikbrodowski.net
Cc:     "Theodore Ts'o" <tytso@mit.edu>,
        "Ivan T. Ivanov" <iivanov@suse.de>,
        Ard Biesheuvel <ardb@kernel.org>, linux-efi@vger.kernel.org,
        LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Hi Dominik,

Thanks for the patch. One trivial nit and one question:

On Thu, Dec 2, 2021 at 6:35 AM Dominik Brodowski
<linux@dominikbrodowski.net> wrote:
> +       /* We cannot do much with the input pool until it is set up in
> +        * rand_initalize(); therefore just mix into the crng state.

I think you meant "rand_initialize()" here (missing 'i').

> If the added entropy suffices to increase crng_init to 1, future calls
> to add_bootloader_randomness() or add_hwgenerator_randomness() used to
> progress to credit_entropy_bits(). However, if the input pool is not yet
> properly set up, the cmpxchg call within that function can lead to an
> infinite recursion.

I see what this patch does with crng_global_init_time, and that seems
probably sensible, but I didn't understand this part of the reasoning
in the commit message; I might just be a bit slow here. Where's the
recursion exactly? Or even an infinite loop?

As far as I can tell, that portion of credit_entropy_bits() breaks down as:

retry:
        entropy_count = orig = READ_ONCE(r->entropy_count);
   [ ... do some arithmetic on entropy_count ... ]
        if (cmpxchg(&r->entropy_count, orig, entropy_count) != orig)
                goto retry;

Why would this be infinite? Why wouldn't the cmpxchg eventually
converge to a stable value? I don't see any call that modifies
r->entropy_count or orig from inside that block. Is there some other
super-spinny concurrent operation?

Thanks,
Jason