netfilter-devel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jan Engelhardt <jengelh@medozas.de>
To: kaber@trash.net
Cc: netfilter-devel@vger.kernel.org
Subject: [PATCH 09/13] libxt_statistic: increase precision on create and dump
Date: Mon,  9 May 2011 11:37:07 +0200	[thread overview]
Message-ID: <1304933832-16412-10-git-send-email-jengelh@medozas.de> (raw)
In-Reply-To: <1304933832-16412-1-git-send-email-jengelh@medozas.de>

Currently, libxt_statistic only dumps the probability with a
granularity of 1/1000000. Assuming only stuffed packets with 1440
bytes payload, this would match approximately every 1.341 GB, which is
pretty low for a high-volume router. Trying to match any larger
interval than that (e.g. 2 GB) will cause libxt_statistic to output
"--probability 0.000000", and when restored, will cause it to never
match again.

Bump the dump precision to what xt_statistic can really do, and adjust
the manpage to include a word about it.

Furthermore, employ explicit rounding when reading the argument from
the command line, because the previous implicit conversion would use
truncation, which is not very exact.

Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
---
 extensions/libxt_statistic.c   |    7 ++++---
 extensions/libxt_statistic.man |    7 +++----
 2 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/extensions/libxt_statistic.c b/extensions/libxt_statistic.c
index bce83fa..f13cdba 100644
--- a/extensions/libxt_statistic.c
+++ b/extensions/libxt_statistic.c
@@ -1,3 +1,4 @@
+#include <math.h>
 #include <stdbool.h>
 #include <stdio.h>
 #include <netdb.h>
@@ -62,11 +63,11 @@ statistic_parse(int c, char **argv, int invert, unsigned int *flags,
 	case '2':
 		if (*flags & 0x2)
 			xtables_error(PARAMETER_PROBLEM, "double --probability");
-		prob = atof(optarg);
+		prob = strtod(optarg, NULL);
 		if (prob < 0 || prob > 1)
 			xtables_error(PARAMETER_PROBLEM,
 				   "--probability must be between 0 and 1");
-		info->u.random.probability = 0x80000000 * prob;
+		info->u.random.probability = lround(0x80000000 * prob);
 		*flags |= 0x2;
 		break;
 	case '3':
@@ -127,7 +128,7 @@ static void print_match(const struct xt_statistic_info *info, char *prefix)
 {
 	switch (info->mode) {
 	case XT_STATISTIC_MODE_RANDOM:
-		printf(" %smode random%s %sprobability %f", prefix,
+		printf(" %smode random%s %sprobability %.11f", prefix,
 		       (info->flags & XT_STATISTIC_INVERT) ? " !" : "",
 		       prefix,
 		       1.0 * info->u.random.probability / 0x80000000);
diff --git a/extensions/libxt_statistic.man b/extensions/libxt_statistic.man
index 4947daf..47182bf 100644
--- a/extensions/libxt_statistic.man
+++ b/extensions/libxt_statistic.man
@@ -12,10 +12,9 @@ and
 .B nth. 
 .TP
 [\fB!\fP] \fB\-\-probability\fP \fIp\fP
-Set the probability from 0 to 1 for a packet to be randomly
-matched. It works only with the
-.B random
-mode.
+Set the probability for a packet to be randomly matched. It only works with the
+\fBrandom\fP mode. \fIp\fP must be within 0.0 and 1.0. The supported
+granularity is in 1/2147483648th increments.
 .TP
 [\fB!\fP] \fB\-\-every\fP \fIn\fP
 Match one packet every nth packet. It works only with the
-- 
1.7.1


  parent reply	other threads:[~2011-05-09  9:37 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-05-09  9:36 guided option parser, run 5 Jan Engelhardt
2011-05-09  9:36 ` [PATCH 01/13] libxt_tos: add inversion support back again Jan Engelhardt
2011-05-09  9:37 ` [PATCH 02/13] libxtables: fix assignment in wrong offset (XTTYPE_UINT*RC) Jan Engelhardt
2011-05-09  9:37 ` [PATCH 03/13] libxt_u32: add missing call to xtables_option_parse Jan Engelhardt
2011-05-09  9:37 ` [PATCH 04/13] extensions: remove bogus use of XT_GETOPT_TABLEEND Jan Engelhardt
2011-05-09  9:37 ` [PATCH 05/13] libxt_owner: remove ifdef IPT_COMM_OWNER Jan Engelhardt
2011-05-09  9:37 ` [PATCH 06/13] libxtables: output name of extension on rev detect failure Jan Engelhardt
2011-05-09  9:37 ` [PATCH 07/13] extensions: const annotations Jan Engelhardt
2011-05-09  9:37 ` [PATCH 08/13] libxt_statistic: streamline and document possible placement of negation Jan Engelhardt
2011-05-09  9:37 ` Jan Engelhardt [this message]
2011-05-09  9:37 ` [PATCH 10/13] libxtables: XTTYPE_DOUBLE support Jan Engelhardt
2011-05-09  9:37 ` [PATCH 11/13] libxt_statistic: use guided option parser Jan Engelhardt
2011-05-09  9:37 ` [PATCH 12/13] libxt_IDLETIMER: " Jan Engelhardt
2011-05-09  9:37 ` [PATCH 13/13] libxt_NFLOG: " Jan Engelhardt
2011-05-09 18:24 ` guided option parser, run 5 Patrick McHardy

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=1304933832-16412-10-git-send-email-jengelh@medozas.de \
    --to=jengelh@medozas.de \
    --cc=kaber@trash.net \
    --cc=netfilter-devel@vger.kernel.org \
    /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;
as well as URLs for NNTP newsgroup(s).