linux-kernel - Re: [PATCHv4] random: Make /dev/random wait for input

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: <VI1PR0702MB38409DE458C09B046CDED5DEE47C0@VI1PR0702MB3840.eurprd07.prod.outlook.com>
Date:   Tue, 19 Feb 2019 07:16:28 +0000
From:   Bernd Edlinger <bernd.edlinger@...mail.de>
To:     "Theodore Y. Ts'o" <tytso@....edu>, Arnd Bergmann <arnd@...db.de>,
        "Greg Kroah-Hartman" <gregkh@...uxfoundation.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCHv4] random: Make /dev/random wait for input_pool
 initialized

> @@ -1826,7 +1830,9 @@ _random_read(int nonblock, char __user *buf, size_t nbytes)
> 
>         nbytes = min_t(size_t, nbytes, SEC_XFER_SIZE);
>         while (1) {
> -               n = extract_entropy_user(&blocking_pool, buf, nbytes);
> +               n = input_pool.initialized
> +                       ? extract_entropy_user(&blocking_pool, buf, nbytes)

Aehm, sorry, now I see this creates a race condition with this code here, since
this the crng_reseed here also tries to read from the input_pool,
but input_pool.initialized is already true:

                if (crng_init < 2 && entropy_bits >= 128) {
                        crng_reseed(&primary_crng, r);
                        entropy_bits = r->entropy_count >> ENTROPY_SHIFT;


I was able to get a system in this behavior by running 3 instances of
#include <stdio.h>
#include <unistd.h>
#include <fcntl.h>

int main()
{
  int f = open("/dev/random", O_NDELAY);
  if (f<0) return 1;
  for(;;)
  {
    unsigned char buf[16];
    int x = read(f, buf, sizeof(buf));
    if (x>=0)
    {
      int i;

      printf("read %d bytes: ", x);
      for (i=0; i<x; i++) printf("%02x ", buf[i]);
      printf("\n");
    }
  }
}

and it managed to steal the entropy away,
before the crng_reseed was able to run.

So I think I will have to change this condition to:
> +               n = input_pool.initialized && crng_ready()
> +                       ? extract_entropy_user(&blocking_pool, buf, nbytes)


Thanks (for your patience :-)
Bernd.