Fix Network Name Resolution
Summary
The GNU C Library getaddrinfo
API is the core name resolution API for C library. The API interprets and implements the requirements of several RFCs and harmonizes the requirements as needed by the project. Unfortunately some of the cases of name resolution are difficult implement and up for interpretation depending on the RFC. This has lead to sometimes different behaviour between OS implementations. In some cases the getaddrinfo()
function in glibc return the wrong results.
This feature aims to fix a number of name resolution bugs in glibc that prevent applications from fully using name resolution functions without workarounds. The fixes should bring glibc closer to other OSs and their interpretations of the RFCs.
Owner
- Name Carlos O'Donell
- Email: carlos at redhat.com
- Name: Pavel Šimerda
- Email: psimerda at redhat.com
- Name: David Jaša
- Email: djasa at redhat.com
- Name Tore Anderson
- Email: tore at fud.no
Current status
- Targeted release: Fedora 20
- Last updated: 2013-03-14
- Percentage of completion: 10%
Detailed description
Currently the getaddrinfo()
function doesn't work as it was designed. Many of its features are buggy and cannot be used without extensive workarounds. Many software packages are using getaddrinfo()
with such workarounds. Many can trigger its failures. And many packages that don't use getaddrinfo()
will be ported in the near future.
Rationale
We are submitting this bug fixing effort as a Feature because:
- It is a high-impact change that will (positively) affect almost all networking software
- Developers will be able to use
getaddrinfo()
without ugly workarounds for new code - We are going to publish guidelines for proper
getaddrinfo()
usage - Documentation for
getaddrinfo()
bugs will be available - Possible workarounds will be offered for backward compatibility
- Comments and errata will be sent to standards organizations
- We want to receive critical response during the whole process
- It will be part of the next networking testweek
Problem statement
The behavior of getaddrinfo()
is often nonstandard, undocumented, surprising, or just plain wrong. We already identified a number of problems. The most prominent examples are here.
getaddrinfo()
may return duplicate or even wrong addresses from /etc/hostsgetaddrinfo()
with NULL servname may return duplicate addressesgetaddrinfo()
with AI_PASSIVE may still return address list not suitable forbind()
getaddrinfo()
with AI_ADDRCONFIG may fail to translate literal addressesgetaddrinfo()
with AI_ADDRCONFIG may fail to resolve /etc/hosts addressesgetaddrinfo()
with AI_ADDRCONFIG may send unwanted AAAA queriesgetaddrinfo()
has a bad choice of default flags
Whether or not the problematic actually occurs depends on /etc/hosts
configuration, /etc/resolv.conf
configuration, getaddrinfo()
input parameters, runtime kernel network interface configuration, and more. While testing the known bugs or reading the source code, more and more bugs are discovered.
Bug reports related to getaddrinfo()
can be found upstream:
http://sourceware.org/bugzilla/buglist.cgi?quicksearch=getaddrinfo
Affected software
The above problems affect software that wants to use getaddrinfo()
to:
- Get parameters for
connect()
orsendto()
to start communicating - Get parameters for
bind()
to listen on specific addresses - Build IP address based accesslists
- Perform name resolution for other purposes
Although it would be nice to also test and fix all software in Fedora using getaddrinfo()
, that is not feasible. Therefore we are going to concentrate on checking and fixing the GNU C library, checking and fixing the most important toolkits dealing with networking, and documenting a set of guidelines for daemons and application software.
Fedora bugs related to dualstack networking including name resolution problems should be added to the following tracker bug:
http://bugzilla.redhat.com/show_bug.cgi?id=883152
Benefit to Fedora
Most Fedora installations are being used in networking environments. Many of them are laptop installations that connect to various networks or even work offline. As name resolution is a critical part of networking experience, Fedora will benefit from being able offer reliability and avoid surprising and buggy behaviour.
Bugs addressed by this feature page are especially tricky because an ordinary user is usually not able to properly report them, not to say analyze them. Especially bugs affecting plain old IPv4 services should be fixed and avoided in the future.
And there's a bonus. As all the documentation will be on Fedora Wiki, attracting attention of other members of the community. Together with other network-related efforts, Fedora could be seen as the leader in linux networking.
Scope
Main focus:
getaddrinfo()
- Name service switch
- DNS module
- hosts module
Secondary libraries:
- GLib networking API
- Qt networking API
Other output:
- Guidelines for daemons and applications
- Backward compatibility guidelines
- Documentation and hints for testing
- Comments and/or errata for standards organizations
How to test
- Reproducers will be always documented in the (usually upstream) bug reports
- The best way to test
getaddrinfo()
is with Python's socket module - Qt and Glib can be also tested in Python
- Some test programs in C could also be helpful
- Lot of tests can be done with actual software using
getaddrinfo()
or one of the libraries - Everyday use can help, too
Please note that we are going to test this feature as part of the next Fedora Network Test Week.
User experience
- Users will be less likely to see confusing and hard-to-reproduce errors
- Some configurations will start working that didn't work before
- Name resolution bugs in applications will be resolved more quickly
- Developers will be able to concentrate on their real work
Dependencies
Almost all development will be done via contributions upstream. Each library will have to be either updated or the fixes backported. Networking APIs are fairly independent from the rest of the libraries.
Libraries that may need updating or fixing:
- GNU C Library
- GLib
- Qt
Contingency plan
In case the fixes are not ready in time, all packages would continue to use workarounds that make them work with the buggy versions of glibc and/or other libraries.h
If, on the other hand, there are packages negatively affected by the changes, we will offer help with fixing the package and/or revert the changes for the next Fedora release. We do not expect any negative impact, though, and we are analysing any potential problems in advance.
We plan to publish a list of packages that might be directly affected by our modifications.
Documentation
Currently, new documentation for name resolution is being drafted on the Fedora Wiki:
Existing documentation:
- POSIX.1-2008
- RFC 3493 (informational)
- RFC 2553 (obsolete informational)
- Current getaddrinfo manpage: mostly like RFC 3493
- Ulrich Drepper: Userlevel IPv6 Programming Introduction
Release notes
Name resolution for network host and service names now works as expected and in line with the standards and user needs. Daemons and applications will now resolve names correctly without extensive workarounds. Both IPv4 and IPv6 addresses are fully supported.