https://github.com/cokebar/lede-dnsmasq-fastlookup

dnsmasq fork with fast ipset/server/address lookup
https://github.com/cokebar/lede-dnsmasq-fastlookup

dnsmasq lede

Last synced: 5 months ago
JSON representation

dnsmasq fork with fast ipset/server/address lookup

• Host: GitHub
• URL: https://github.com/cokebar/lede-dnsmasq-fastlookup
• Owner: cokebar
• License: gpl-2.0
• Created: 2017-03-06T17:09:17.000Z (over 8 years ago)
• Default Branch: master
• Last Pushed: 2017-04-08T16:09:28.000Z (over 8 years ago)
• Last Synced: 2025-04-08T20:48:07.865Z (6 months ago)
• Topics: dnsmasq, lede
• Language: Shell
• Size: 32.2 KB
• Stars: 8
• Watchers: 3
• Forks: 6
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
DNSMASQ fork for improving --ipsets, --server, and --address performance

========================================================================

__This is the LEDE version for https://github.com/infinet/dnsmasq/tree/fastlookup-v2.77test4__

__Dnsmasq version: 2.77test4; Tested on LEDE snapshot r3644__

Dnsmasq matches domain names for --ipsets, --server, and --address options by

iterates over linked list. It is good enough for general use, but slows down as

the domain names to be matched grows.

Here introduce a modified dnsmasq for fast lookup ipset/server/address options.

The match time is constant regardless the size of rules.

                       root

                        |

             +---------------------+

            com                   org

             |                     |

    +------------------+     +-------------+

    yahoo google twitter   debian       freebsd

      |      |               |             |

     www    mail          +---------+     www

                          cn jp uk us

                          |

                         ftp

The lookup steps over domain name hierarchy top-down. All labels are stored in

open addressing hash tables. Sub-level labels that belong to different parent

nodes are stored separately. e.g. yahoo, google, and twitter are in one hash

table, while debian and freebsd are in another.

The hash table size is power of 2, two hash functions are used to compute hash

bucket. For locating a particular label from hash table, two hash values are

compared first, only if they are match, should the more expensive string

comparison be used to confirm the search.

Example usage

-------------

Use DNS blackhole to block malware site:

    address=/example.com/

Force google DNS as upstream server for special domain:

    server=/example.com/8.8.8.8

Add all IPs of a paticular domain to IPSET:

    ipset=/example.com/example-ipset