public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Oliver Xymoron <oxymoron@waste.org>
To: Robert Love <rml@tech9.net>
Cc: Linus Torvalds <torvalds@transmeta.com>,
	linux-kernel <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] (0/4) Entropy accounting fixes
Date: Sun, 18 Aug 2002 01:48:46 -0500	[thread overview]
Message-ID: <20020818064846.GQ21643@waste.org> (raw)
In-Reply-To: <1029652262.898.12.camel@phantasy>

On Sun, Aug 18, 2002 at 02:31:02AM -0400, Robert Love wrote:
> On Sun, 2002-08-18 at 00:01, Linus Torvalds wrote:
> 
> > So I think that if we just made the code be much less trusting (say, 
> > consider the TSC information per interrupt to give only a single bit of 
> > entropy, for example), and coupled that with making network devices always 
> > be considered sources of entropy, we'd have a reasonable balance. 
> 
> I think that sounds good.
> 
> I have a patch which I can send - it needs to be rediffed I suspect -
> that has each network device feed the entropy pool. (Actually, it
> creates a new flag, SA_NET_RANDOM, that defines to SA_SAMPLE_RANDOM or 0
> depending on a configure setting.  If you want it unconditional, that is
> just as easy though).

I think I have a compromise that'll make everyone happy here. I've
added a tunable called trust_pct in /proc/sys/kernel/random. Set it to
100% and untrusted sources will always add a bit of entropy. Defaults
to zero, which should be fine for everyone who's got a head.

Then we can add SA_SAMPLE_RANDOM for network devices unconditionally.

Untested, applies on top of my previous patches:

diff -ur a/drivers/char/random.c b/drivers/char/random.c
--- a/drivers/char/random.c	2002-08-17 20:54:02.000000000 -0500
+++ b/drivers/char/random.c	2002-08-18 01:38:58.000000000 -0500
@@ -724,6 +724,7 @@
  */
 static int benford[16]={0,0,0,1,2,3,4,5,5,6,7,7,8,9,9,10};
 static int last_ctxt=0;
+static int trust_break=50, trust_pct=0, trust_min=0, trust_max=100;
 
 void add_timing_entropy(void *src, unsigned datum)
 {
@@ -764,6 +765,18 @@
 		delta>>=es->shift;
 		bits=benford[int_log2_16bits(delta & 0xffff)];
 	}
+
+	/* Throw in an untrusted bit as entropy trust_pct% of the time */
+	if(trust_pct && !bits)
+	{
+		trust_break+=trust_pct;
+		if(trust_break>100)
+		{
+			bits=1;
+			trust_break-=100;
+		}
+	}
+
 	batch_entropy_store(datum^time, bits);
 }
 
@@ -1779,6 +1792,10 @@
 	{RANDOM_UUID, "uuid",
 	 NULL, 16, 0444, NULL,
 	 &proc_do_uuid, &uuid_strategy},
+	{RANDOM_TRUST_PCT, "trust_pct",
+	 &trust_pct, sizeof(int), 0644, NULL,
+	 &proc_dointvec_minmax, &sysctl_intvec, 0,
+	 &trust_min, &trust_max},
 	{0}
 };
 
diff -ur a/include/linux/sysctl.h b/include/linux/sysctl.h
--- a/include/linux/sysctl.h	2002-08-17 00:30:00.000000000 -0500
+++ b/include/linux/sysctl.h	2002-08-18 01:37:54.000000000 -0500
@@ -182,7 +182,8 @@
 	RANDOM_READ_THRESH=3,
 	RANDOM_WRITE_THRESH=4,
 	RANDOM_BOOT_ID=5,
-	RANDOM_UUID=6
+	RANDOM_UUID=6,
+	RANDOM_TRUST_PCT=7
 };
 
 /* /proc/sys/bus/isa */


-- 
 "Love the dolphins," she advised him. "Write by W.A.S.T.E.." 

  reply	other threads:[~2002-08-18  6:44 UTC|newest]

Thread overview: 86+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2002-08-18  2:15 [PATCH] (0/4) Entropy accounting fixes Oliver Xymoron
2002-08-18  2:23 ` [PATCH] (1/4) " Oliver Xymoron
2002-08-18  2:26   ` [PATCH] (2/4) Update input drivers Oliver Xymoron
2002-08-18  2:29     ` [PATCH] (3/4) SA_RANDOM user fixup Oliver Xymoron
2002-08-18  2:32       ` [PATCH] (4/4) entropy batching update Oliver Xymoron
2002-08-18  2:30 ` [PATCH] (0/4) Entropy accounting fixes Linus Torvalds
2002-08-18  2:59   ` Oliver Xymoron
2002-08-18  3:08     ` Linus Torvalds
2002-08-18  3:25       ` Linus Torvalds
2002-08-18  4:42         ` Oliver Xymoron
2002-08-18  4:53           ` Linus Torvalds
2002-08-18  5:05             ` Dmitri
2002-08-18  6:18               ` Oliver Xymoron
2002-08-22  3:33             ` David Wagner
2002-08-18 10:30         ` Alan Cox
2002-08-18 15:08           ` Oliver Xymoron
2002-08-18 17:31           ` Jonathan Lundell
2002-08-22  3:27         ` David Wagner
2002-08-18  4:30       ` Oliver Xymoron
2002-08-21  8:44       ` Rogier Wolff
2002-08-21 12:47         ` Oliver Xymoron
2002-08-18  5:28     ` Andreas Dilger
2002-08-18  5:53       ` Oliver Xymoron
2002-08-22  3:25   ` David Wagner
2002-08-18  3:05 ` Linus Torvalds
2002-08-18  3:51   ` Robert Love
2002-08-18  4:01     ` Linus Torvalds
2002-08-18  5:38       ` Oliver Xymoron
2002-08-19  4:21         ` Theodore Ts'o
2002-08-19 10:15           ` Marco Colombo
2002-08-19 10:25             ` Oliver Neukum
2002-08-19 11:03               ` Marco Colombo
2002-08-19 14:22                 ` Oliver Neukum
2002-08-19 15:21                   ` Marco Colombo
2002-08-19 16:29                     ` Oliver Neukum
2002-08-19 12:39           ` Oliver Xymoron
2002-08-18  6:31       ` Robert Love
2002-08-18  6:48         ` Oliver Xymoron [this message]
2002-08-18  4:06     ` dean gaudet
2002-08-18  4:44       ` Oliver Xymoron
2002-08-18  7:31       ` Bernd Eckenfels
2002-08-18  9:48         ` Ralf Baechle
2002-08-20 12:51           ` Bernd Eckenfels
2002-08-18 16:58         ` Robert Love
2002-08-18 10:25       ` Alan Cox
2002-08-19 10:47         ` Marco Colombo
2002-08-19 12:29           ` Alan Cox
2002-08-19 12:56             ` Marco Colombo
2002-09-08  3:43             ` D. Hugh Redelmeier
2002-09-08 18:03               ` David Wagner
2002-09-09 16:53                 ` Oliver Xymoron
2002-09-09 16:58                   ` David Wagner
2002-09-09 19:47                     ` Oliver Xymoron
2002-09-09 23:22                       ` David Wagner
2002-09-16 22:51                       ` dean gaudet
2002-09-17  1:18                         ` Oliver Xymoron
2002-09-09 18:54                   ` Kent Borg
2002-09-09 19:57                     ` Oliver Xymoron
2002-09-09 20:11                       ` Kent Borg
2002-08-18  4:57     ` Oliver Xymoron
2002-08-18  4:28   ` Oliver Xymoron
2002-08-18  4:51     ` Linus Torvalds
2002-08-18  5:24       ` Oliver Xymoron
2002-08-18 16:59         ` Linus Torvalds
2001-11-02 10:34           ` Pavel Machek
2002-08-23 20:16             ` Linus Torvalds
2002-08-18 17:03           ` Robert Love
2002-08-18 17:31           ` Oliver Xymoron
2002-08-18 16:54     ` Robert Love
2002-08-18 17:18       ` Oliver Xymoron
2002-08-18 17:20         ` Robert Love
2002-08-19  5:43 ` Theodore Ts'o
2001-11-02 10:05   ` Pavel Machek
2002-08-19  6:06   ` *Challenge* Finding a solution (When kernel boots it does not display any system info) louie miranda
2002-08-19  7:30     ` Gilad Ben-Yossef
2002-08-19  7:30     ` Ryan Cumming
2002-08-20  0:55       ` louie miranda
2002-08-19 13:52   ` [PATCH] (0/4) Entropy accounting fixes Oliver Xymoron
2002-08-20  8:59     ` Tommi Kyntola
2002-08-20 13:21       ` Oliver Xymoron
2002-08-20 16:19         ` Tommi Kyntola
2002-08-20 17:22           ` Oliver Xymoron
2002-09-08  3:51             ` D. Hugh Redelmeier
2002-09-08  4:31               ` Oliver Xymoron
  -- strict thread matches above, loose matches on Subject: below --
2002-08-18  4:57 David Brownell
2002-08-18  6:02 ` Oliver Xymoron

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20020818064846.GQ21643@waste.org \
    --to=oxymoron@waste.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=rml@tech9.net \
    --cc=torvalds@transmeta.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox