curl/docs/INSTALL
2014-12-05 14:54:25 +01:00

1106 lines
42 KiB
Plaintext

_ _ ____ _
___| | | | _ \| |
/ __| | | | |_) | |
| (__| |_| | _ <| |___
\___|\___/|_| \_\_____|
How To Compile
Installing Binary Packages
==========================
Lots of people download binary distributions of curl and libcurl. This
document does not describe how to install curl or libcurl using such a
binary package. This document describes how to compile, build and install
curl and libcurl from source code.
Building from git
=================
If you get your code off a git repository, see the GIT-INFO file in the
root directory for specific instructions on how to proceed.
UNIX
====
A normal unix installation is made in three or four steps (after you've
unpacked the source archive):
./configure
make
make test (optional)
make install
You probably need to be root when doing the last command.
If you have checked out the sources from the git repository, read the
GIT-INFO on how to proceed.
Get a full listing of all available configure options by invoking it like:
./configure --help
If you want to install curl in a different file hierarchy than /usr/local,
you need to specify that already when running configure:
./configure --prefix=/path/to/curl/tree
If you happen to have write permission in that directory, you can do 'make
install' without being root. An example of this would be to make a local
install in your own home directory:
./configure --prefix=$HOME
make
make install
The configure script always tries to find a working SSL library unless
explicitly told not to. If you have OpenSSL installed in the default search
path for your compiler/linker, you don't need to do anything special. If
you have OpenSSL installed in /usr/local/ssl, you can run configure like:
./configure --with-ssl
If you have OpenSSL installed somewhere else (for example, /opt/OpenSSL)
and you have pkg-config installed, set the pkg-config path first, like this:
env PKG_CONFIG_PATH=/opt/OpenSSL/lib/pkgconfig ./configure --with-ssl
Without pkg-config installed, use this:
./configure --with-ssl=/opt/OpenSSL
If you insist on forcing a build without SSL support, even though you may
have OpenSSL installed in your system, you can run configure like this:
./configure --without-ssl
If you have OpenSSL installed, but with the libraries in one place and the
header files somewhere else, you have to set the LDFLAGS and CPPFLAGS
environment variables prior to running configure. Something like this
should work:
(with the Bourne shell and its clones):
CPPFLAGS="-I/path/to/ssl/include" LDFLAGS="-L/path/to/ssl/lib" \
./configure
(with csh, tcsh and their clones):
env CPPFLAGS="-I/path/to/ssl/include" LDFLAGS="-L/path/to/ssl/lib" \
./configure
If you have shared SSL libs installed in a directory where your run-time
linker doesn't find them (which usually causes configure failures), you can
provide the -R option to ld on some operating systems to set a hard-coded
path to the run-time linker:
env LDFLAGS=-R/usr/local/ssl/lib ./configure --with-ssl
MORE OPTIONS
------------
To force configure to use the standard cc compiler if both cc and gcc are
present, run configure like
CC=cc ./configure
or
env CC=cc ./configure
To force a static library compile, disable the shared library creation
by running configure like:
./configure --disable-shared
To tell the configure script to skip searching for thread-safe functions,
add an option like:
./configure --disable-thread
If you're a curl developer and use gcc, you might want to enable more
debug options with the --enable-debug option.
curl can be built to use a whole range of libraries to provide various
useful services, and configure will try to auto-detect a decent
default. But if you want to alter it, you can select how to deal with
each individual library.
To build with GnuTLS for SSL/TLS, use both --without-ssl and
--with-gnutls.
To build with Cyassl for SSL/TLS, use both --without-ssl and
--with-cyassl.
To build with NSS for SSL/TLS, use both --without-ssl and --with-nss.
To build with PolarSSL for SSL/TLS, use both --without-ssl and
--with-polarssl.
To build with axTLS for SSL/TLS, use both --without-ssl and --with-axtls.
To build with GSS-API support, use --with-gssapi and have the MIT Kerberos
or Heimdal packages installed.
To get support for SCP and SFTP, build with --with-libssh2 and have
libssh2 0.16 or later installed.
To get Metalink support, build with --with-libmetalink and have the
libmetalink packages installed.
SPECIAL CASES
-------------
Some versions of uClibc require configuring with CPPFLAGS=-D_GNU_SOURCE=1
to get correct large file support.
The Open Watcom C compiler on Linux requires configuring with the variables:
./configure CC=owcc AR="$WATCOM/binl/wlib" AR_FLAGS=-q \
RANLIB=/bin/true STRIP="$WATCOM/binl/wstrip" CFLAGS=-Wextra
Win32
=====
Building Windows DLLs and C run-time (CRT) linkage issues
---------------------------------------------------------
As a general rule, building a DLL with static CRT linkage is highly
discouraged, and intermixing CRTs in the same app is something to
avoid at any cost.
Reading and comprehension of Microsoft Knowledge Base articles
KB94248 and KB140584 is a must for any Windows developer. Especially
important is full understanding if you are not going to follow the
advice given above.
KB94248 - How To Use the C Run-Time
http://support.microsoft.com/kb/94248/en-us
KB140584 - How to link with the correct C Run-Time (CRT) library
http://support.microsoft.com/kb/140584/en-us
KB190799 - Potential Errors Passing CRT Objects Across DLL Boundaries
http://msdn.microsoft.com/en-us/library/ms235460
If your app is misbehaving in some strange way, or it is suffering
from memory corruption, before asking for further help, please try
first to rebuild every single library your app uses as well as your
app using the debug multithreaded dynamic C runtime.
If you get linkage errors read section 5.7 of the FAQ document.
MingW32
-------
Make sure that MinGW32's bin dir is in the search path, for example:
set PATH=c:\mingw32\bin;%PATH%
then run 'mingw32-make mingw32' in the root dir. There are other
make targets available to build libcurl with more features, use:
'mingw32-make mingw32-zlib' to build with Zlib support;
'mingw32-make mingw32-ssl-zlib' to build with SSL and Zlib enabled;
'mingw32-make mingw32-ssh2-ssl-zlib' to build with SSH2, SSL, Zlib;
'mingw32-make mingw32-ssh2-ssl-sspi-zlib' to build with SSH2, SSL, Zlib
and SSPI support.
If you have any problems linking libraries or finding header files, be sure
to verify that the provided "Makefile.m32" files use the proper paths, and
adjust as necessary. It is also possible to override these paths with
environment variables, for example:
set ZLIB_PATH=c:\zlib-1.2.8
set OPENSSL_PATH=c:\openssl-0.9.8zc
set LIBSSH2_PATH=c:\libssh2-1.4.3
ATTENTION: if you want to build with libssh2 support you have to use latest
version 0.17 - previous versions will NOT work with 7.17.0 and later!
Use 'mingw32-make mingw32-ssh2-ssl-zlib' to build with SSH2 and SSL enabled.
It is now also possible to build with other LDAP SDKs than MS LDAP;
currently it is possible to build with native Win32 OpenLDAP, or with the
Novell CLDAP SDK. If you want to use these you need to set these vars:
set LDAP_SDK=c:\openldap
set USE_LDAP_OPENLDAP=1
or for using the Novell SDK:
set USE_LDAP_NOVELL=1
If you want to enable LDAPS support then set LDAPS=1.
- optional MingW32-built OpenLDAP SDK available from:
http://www.gknw.net/mirror/openldap/
- optional recent Novell CLDAP SDK available from:
http://developer.novell.com/ndk/cldap.htm
Cygwin
------
Almost identical to the unix installation. Run the configure script in the
curl root with 'sh configure'. Make sure you have the sh executable in
/bin/ or you'll see the configure fail toward the end.
Run 'make'
Dev-Cpp
-------
See the separate INSTALL.devcpp file for details.
MSVC 6 caveats
--------------
If you use MSVC 6 it is required that you use the February 2003 edition of
the 'Platform SDK' which can be downloaded from:
http://www.microsoft.com/en-us/download/details.aspx?id=12261
Building any software with MSVC 6 without having PSDK installed is just
asking for trouble down the road once you have released it, you might notice
the problems in the first corner or ten miles ahead, depending mostly on your
choice of static vs dynamic runtime and third party libraries. Anyone using
software built in such way will at some point regret having done so.
If the compiler has been updated with the installation of a service pack as
those mentioned in http://support.microsoft.com/kb/194022 the compiler can be
safely used to read source code, translate and make it object code.
But, even with the service packs mentioned above installed, the resulting
software generated in such an environment will be using outdated system
header files and libraries with bugs and security issues which have already
been addressed and fixed long time ago.
So, building curl and libcurl with MSVC 6 without PSDK is absolutely
discouraged for the benefit of anyone using software built in such
environment. And it will not be supported in any way, as we could just
be hunting bugs which have already been fixed way back in 2003.
When building with MSVC 6 we attempt to detect if PSDK is not being used,
and if this is the case the build process will fail hard with an error
message stating that the February 2003 PSDK is required. This is done to
protect the unsuspecting and avoid PEBKAC issues.
Additionally it might happen that a die hard MSVC hacker still wants to
build curl and libcurl with MSVC 6 without PSDK installed, even knowing
that this is a highly discouraged and unsupported build environment. In
this case the brave of heart will be able to build in such an environment
with the requisite of defining preprocessor symbol ALLOW_MSVC6_WITHOUT_PSDK
in lib/config-win32.h and knowing that LDAP and IPv6 support will be missing.
MSVC from command line
----------------------
Run the 'vcvars32.bat' file to get a proper environment. The
vcvars32.bat file is part of the Microsoft development environment and
you may find it in 'C:\Program Files\Microsoft Visual Studio\vc98\bin'
provided that you installed Visual C/C++ 6 in the default directory.
Then run 'nmake vc' in curl's root directory.
If you want to compile with zlib support, you will need to build
zlib (http://www.gzip.org/zlib/) as well. Please read the zlib
documentation on how to compile zlib. Define the ZLIB_PATH environment
variable to the location of zlib.h and zlib.lib, for example:
set ZLIB_PATH=c:\zlib-1.2.8
Then run 'nmake vc-zlib' in curl's root directory.
If you want to compile with SSL support you need the OpenSSL package.
Please read the OpenSSL documentation on how to compile and install
the OpenSSL libraries. The build process of OpenSSL generates the
libeay32.dll and ssleay32.dll files in the out32dll subdirectory in
the OpenSSL home directory. OpenSSL static libraries (libeay32.lib,
ssleay32.lib, RSAglue.lib) are created in the out32 subdirectory.
Before running nmake define the OPENSSL_PATH environment variable with
the root/base directory of OpenSSL, for example:
set OPENSSL_PATH=c:\openssl-0.9.8zc
Then run 'nmake vc-ssl' or 'nmake vc-ssl-dll' in curl's root
directory. 'nmake vc-ssl' will create a libcurl static and dynamic
libraries in the lib subdirectory, as well as a statically linked
version of curl.exe in the src subdirectory. This statically linked
version is a standalone executable not requiring any DLL at
runtime. This make method requires that you have the static OpenSSL
libraries available in OpenSSL's out32 subdirectory.
'nmake vc-ssl-dll' creates the libcurl dynamic library and
links curl.exe against libcurl and OpenSSL dynamically.
This executable requires libcurl.dll and the OpenSSL DLLs
at runtime.
Run 'nmake vc-ssl-zlib' to build with both ssl and zlib support.
MSVC IDE
--------
A fairly comprehensive set of Visual Studio project files are available for
v6.0 through v12.0 and are located in the projects folder to allow proper
building of both the libcurl library as well as the curl tool.
For more information about these projects and building via Visual Studio
please see the README file located in the projects folder.
Borland C++ compiler
--------------------
Ensure that your build environment is properly set up to use the compiler
and associated tools. PATH environment variable must include the path to
bin subdirectory of your compiler installation, eg: c:\Borland\BCC55\bin
It is advisable to set environment variable BCCDIR to the base path of
the compiler installation.
set BCCDIR=c:\Borland\BCC55
In order to build a plain vanilla version of curl and libcurl run the
following command from curl's root directory:
make borland
To build curl and libcurl with zlib and OpenSSL support set environment
variables ZLIB_PATH and OPENSSL_PATH to the base subdirectories of the
already built zlib and OpenSSL libraries and from curl's root directory
run command:
make borland-ssl-zlib
libcurl library will be built in 'lib' subdirectory while curl tool
is built in 'src' subdirectory. In order to use libcurl library it is
advisable to modify compiler's configuration file bcc32.cfg located
in c:\Borland\BCC55\bin to reflect the location of libraries include
paths for example the '-I' line could result in something like:
-I"c:\Borland\BCC55\include;c:\curl\include;c:\openssl\inc32"
bcc3.cfg '-L' line could also be modified to reflect the location of
of libcurl library resulting for example:
-L"c:\Borland\BCC55\lib;c:\curl\lib;c:\openssl\out32"
In order to build sample program 'simple.c' from the docs\examples
subdirectory run following command from mentioned subdirectory:
bcc32 simple.c libcurl.lib cw32mt.lib
In order to build sample program simplessl.c an SSL enabled libcurl
is required, as well as the OpenSSL libeay32.lib and ssleay32.lib
libraries.
OTHER MSVC IDEs
---------------
If you use VC++, Borland or similar compilers. Include all lib source
files in a static lib "project" (all .c and .h files that is).
(you should name it libcurl or similar)
Make the sources in the src/ drawer be a "win32 console application"
project. Name it curl.
Disabling Specific Protocols in Win32 builds
--------------------------------------------
The configure utility, unfortunately, is not available for the Windows
environment, therefore, you cannot use the various disable-protocol
options of the configure utility on this platform.
However, you can use the following defines to disable specific
protocols:
HTTP_ONLY disables all protocols except HTTP
CURL_DISABLE_FTP disables FTP
CURL_DISABLE_LDAP disables LDAP
CURL_DISABLE_TELNET disables TELNET
CURL_DISABLE_DICT disables DICT
CURL_DISABLE_FILE disables FILE
CURL_DISABLE_TFTP disables TFTP
CURL_DISABLE_HTTP disables HTTP
CURL_DISABLE_IMAP disables IMAP
CURL_DISABLE_POP3 disables POP3
CURL_DISABLE_SMTP disables SMTP
If you want to set any of these defines you have the following options:
- Modify lib/config-win32.h
- Modify lib/curl_setup.h
- Modify lib/Makefile.vc6
- Modify the "Preprocessor Definitions" in the libcurl project
Note: The pre-processor settings can be found using the Visual Studio IDE
under "Project -> Settings -> C/C++ -> General" in VC6 and "Project ->
Properties -> Configuration Properties -> C/C++ -> Preprocessor" in later
versions.
Using BSD-style lwIP instead of Winsock TCP/IP stack in Win32 builds
--------------------------------------------------------------------
In order to compile libcurl and curl using BSD-style lwIP TCP/IP stack
it is necessary to make definition of preprocessor symbol USE_LWIPSOCK
visible to libcurl and curl compilation processes. To set this definition
you have the following alternatives:
- Modify lib/config-win32.h and src/config-win32.h
- Modify lib/Makefile.vc6
- Modify the "Preprocessor Definitions" in the libcurl project
Note: The pre-processor settings can be found using the Visual Studio IDE
under "Project -> Settings -> C/C++ -> General" in VC6 and "Project ->
Properties -> Configuration Properties -> C/C++ -> Preprocessor" in later
versions.
Once that libcurl has been built with BSD-style lwIP TCP/IP stack support,
in order to use it with your program it is mandatory that your program
includes lwIP header file <lwip/opt.h> (or another lwIP header that includes
this) before including any libcurl header. Your program does not need the
USE_LWIPSOCK preprocessor definition which is for libcurl internals only.
Compilation has been verified with lwIP 1.4.0 and contrib-1.4.0 from:
http://download.savannah.gnu.org/releases/lwip/lwip-1.4.0.zip
http://download.savannah.gnu.org/releases/lwip/contrib-1.4.0.zip
This BSD-style lwIP TCP/IP stack support must be considered experimental
given that it has been verified that lwIP 1.4.0 still needs some polish,
and libcurl might yet need some additional adjustment, caveat emptor.
Important static libcurl usage note
-----------------------------------
When building an application that uses the static libcurl library, you must
add '-DCURL_STATICLIB' to your CFLAGS. Otherwise the linker will look for
dynamic import symbols.
Apple iOS and Mac OS X
======================
On recent Apple operating systems, curl can be built to use Apple's
SSL/TLS implementation, Secure Transport, instead of OpenSSL. To build with
Secure Transport for SSL/TLS, use the configure option --with-darwinssl. (It
is not necessary to use the option --without-ssl.) This feature requires iOS
5.0 or later, or OS X 10.5 ("Leopard") or later.
When Secure Transport is in use, the curl options --cacert and --capath and
their libcurl equivalents, will be ignored, because Secure Transport uses
the certificates stored in the Keychain to evaluate whether or not to trust
the server. This, of course, includes the root certificates that ship with
the OS. The --cert and --engine options, and their libcurl equivalents, are
currently unimplemented in curl with Secure Transport.
For OS X users: In OS X 10.8 ("Mountain Lion"), Apple made a major
overhaul to the Secure Transport API that, among other things, added
support for the newer TLS 1.1 and 1.2 protocols. To get curl to support
TLS 1.1 and 1.2, you must build curl on Mountain Lion or later, or by
using the equivalent SDK. If you set the MACOSX_DEPLOYMENT_TARGET
environmental variable to an earlier version of OS X prior to building curl,
then curl will use the new Secure Transport API on Mountain Lion and later,
and fall back on the older API when the same curl binary is executed on
older cats. For example, running these commands in curl's directory in the
shell will build the code such that it will run on cats as old as OS X 10.6
("Snow Leopard") (using bash):
export MACOSX_DEPLOYMENT_TARGET="10.6"
./configure --with-darwinssl
make
IBM OS/2
========
Building under OS/2 is not much different from building under unix.
You need:
- emx 0.9d
- GNU make
- GNU patch
- ksh
- GNU bison
- GNU file utilities
- GNU sed
- autoconf 2.13
If you want to build with OpenSSL or OpenLDAP support, you'll need to
download those libraries, too. Dirk Ohme has done some work to port SSL
libraries under OS/2, but it looks like he doesn't care about emx. You'll
find his patches on: http://come.to/Dirk_Ohme
If during the linking you get an error about _errno being an undefined
symbol referenced from the text segment, you need to add -D__ST_MT_ERRNO__
in your definitions.
If everything seems to work fine but there's no curl.exe, you need to add
-Zexe to your linker flags.
If you're getting huge binaries, probably your makefiles have the -g in
CFLAGS.
VMS
===
(The VMS section is in whole contributed by the friendly Nico Baggus)
Curl seems to work with FTP & HTTP other protocols are not tested. (the
perl http/ftp testing server supplied as testing too cannot work on VMS
because vms has no concept of fork(). [ I tried to give it a whack, but
that's of no use.
SSL stuff has not been ported.
Telnet has about the same issues as for Win32. When the changes for Win32
are clear maybe they'll work for VMS too. The basic problem is that select
ONLY works for sockets.
Marked instances of fopen/[f]stat that might become a problem, especially
for non stream files. In this regard, the files opened for writing will be
created stream/lf and will thus be safe. Just keep in mind that non-binary
read/wring from/to files will have a records size limit of 32767 bytes
imposed.
Stat to get the size of the files is again only safe for stream files &
fixed record files without implied CC.
-- My guess is that only allowing access to stream files is the quickest
way to get around the most issues. Therefore all files need to to be
checked to be sure they will be stream/lf before processing them. This is
the easiest way out, I know. The reason for this is that code that needs to
report the filesize will become a pain in the ass otherwise.
Exit status.... Well we needed something done here,
VMS has a structured exist status:
| 3 | 2 | 1 | 0|
|1098|765432109876|5432109876543|210|
+----+------------+-------------+---+
|Ctrl| Facility | Error code |sev|
+----+------------+-------------+---+
With the Ctrl-bits an application can tell if part or the whole message has
already been printed from the program, DCL doesn't need to print it again.
Facility - basically the program ID. A code assigned to the program
the name can be fetched from external or internal message libraries
Error code - the err codes assigned by the application
Sev. - severity: Even = error, off = non error
0 = Warning
1 = Success
2 = Error
3 = Information
4 = Fatal
<5-7> reserved.
This all presents itself with:
%<FACILITY>-<Sev>-<Errorname>, <Error message>
See also the src/curlmsg.msg file, it has the source for the messages In
src/main.c a section is devoted to message status values, the globalvalues
create symbols with certain values, referenced from a compiled message
file. Have all exit function use a exit status derived from a translation
table with the compiled message codes.
This was all compiled with:
Compaq C V6.2-003 on OpenVMS Alpha V7.1-1H2
So far for porting notes as of:
13-jul-2001
N. Baggus
QNX
===
(This section was graciously brought to us by David Bentham)
As QNX is targeted for resource constrained environments, the QNX headers
set conservative limits. This includes the FD_SETSIZE macro, set by default
to 32. Socket descriptors returned within the CURL library may exceed this,
resulting in memory faults/SIGSEGV crashes when passed into select(..)
calls using fd_set macros.
A good all-round solution to this is to override the default when building
libcurl, by overriding CFLAGS during configure, example
# configure CFLAGS='-DFD_SETSIZE=64 -g -O2'
RISC OS
=======
The library can be cross-compiled using gccsdk as follows:
CC=riscos-gcc AR=riscos-ar RANLIB='riscos-ar -s' ./configure \
--host=arm-riscos-aof --without-random --disable-shared
make
where riscos-gcc and riscos-ar are links to the gccsdk tools.
You can then link your program with curl/lib/.libs/libcurl.a
AmigaOS
=======
(This section was graciously brought to us by Diego Casorran)
To build cURL/libcurl on AmigaOS just type 'make amiga' ...
What you need is: (not tested with others versions)
GeekGadgets / gcc 2.95.3 (http://www.geekgadgets.org/)
AmiTCP SDK v4.3 (http://www.aminet.net/comm/tcp/AmiTCP-SDK-4.3.lha)
Native Developer Kit (http://www.amiga.com/3.9/download/NDK3.9.lha)
As no ixemul.library is required you will be able to build it for
WarpOS/PowerPC (not tested by me), as well a MorphOS version should be
possible with no problems.
To enable SSL support, you need a OpenSSL native version (without ixemul),
you can find a precompiled package at http://amiga.sourceforge.net/OpenSSL/
NetWare
=======
To compile curl.nlm / libcurl.nlm you need:
- either any gcc / nlmconv, or CodeWarrior 7 PDK 4 or later.
- gnu make and awk running on the platform you compile on;
native Win32 versions can be downloaded from:
http://www.gknw.net/development/prgtools/
- recent Novell LibC SDK available from:
http://developer.novell.com/ndk/libc.htm
- or recent Novell CLib SDK available from:
http://developer.novell.com/ndk/clib.htm
- optional recent Novell CLDAP SDK available from:
http://developer.novell.com/ndk/cldap.htm
- optional zlib sources (static or dynamic linking with zlib.imp);
sources with NetWare Makefile can be obtained from:
http://www.gknw.net/mirror/zlib/
- optional OpenSSL sources (version 0.9.8 or later build with BSD sockets);
you can find precompiled packages at:
http://www.gknw.net/development/ossl/netware/
for CLIB-based builds OpenSSL 0.9.8h or later is required - earlier versions
don't support building with CLIB BSD sockets.
- optional SSH2 sources (version 0.17 or later);
Set a search path to your compiler, linker and tools; on Linux make
sure that the var OSTYPE contains the string 'linux'; set the var
NDKBASE to point to the base of your Novell NDK; and then type
'make netware' from the top source directory; other targets available
are 'netware-ssl', 'netware-ssl-zlib', 'netware-zlib' and 'netware-ares';
if you need other combinations you can control the build with the
environment variables WITH_SSL, WITH_ZLIB, WITH_ARES, WITH_SSH2, and
ENABLE_IPV6; you can set LINK_STATIC=1 to link curl.nlm statically.
By default LDAP support is enabled, however currently you will need a patch
in order to use the CLDAP NDK with BSD sockets (Novell Bug 300237):
http://www.gknw.net/test/curl/cldap_ndk/ldap_ndk.diff
I found on some Linux systems (RH9) that OS detection didn't work although
a 'set | grep OSTYPE' shows the var present and set; I simply overwrote it
with 'OSTYPE=linux-rh9-gnu' and the detection in the Makefile worked...
Any help in testing appreciated!
Builds automatically created 8 times a day from current git are here:
http://www.gknw.net/mirror/curl/autobuilds/
the status of these builds can be viewed at the autobuild table:
http://curl.haxx.se/dev/builds.html
eCos
====
curl does not use the eCos build system, so you must first build eCos
separately, then link curl to the resulting eCos library. Here's a sample
configure line to do so on an x86 Linux box targeting x86:
GCCLIB=`gcc -print-libgcc-file-name` && \
CFLAGS="-D__ECOS=1 -nostdinc -I$ECOS_INSTALL/include \
-I`dirname $GCCLIB`/include" \
LDFLAGS="-nostdlib -Wl,--gc-sections -Wl,-static \
-L$ECOS_INSTALL/lib -Ttarget.ld -ltarget" \
./configure --host=i386 --disable-shared \
--without-ssl --without-zlib --disable-manual --disable-ldap
In most cases, eCos users will be using libcurl from within a custom
embedded application. Using the standard 'curl' executable from
within eCos means facing the limitation of the standard eCos C
startup code which does not allow passing arguments in main(). To
run 'curl' from eCos and have it do something useful, you will need
to either modify the eCos startup code to pass in some arguments, or
modify the curl application itself to retrieve its arguments from
some location set by the bootloader or hard-code them.
Something like the following patch could be used to hard-code some
arguments. The MTAB_ENTRY line mounts a RAM disk as the root filesystem
(without mounting some kind of filesystem, eCos errors out all file
operations which curl does not take to well). The next section synthesizes
some command-line arguments for curl to use, in this case to direct curl
to read further arguments from a file. It then creates that file on the
RAM disk and places within it a URL to download: a file: URL that
just happens to point to the configuration file itself. The results
of running curl in this way is the contents of the configuration file
printed to the console.
--- src/main.c 19 Jul 2006 19:09:56 -0000 1.363
+++ src/main.c 24 Jul 2006 21:37:23 -0000
@@ -4286,11 +4286,31 @@
}
+#ifdef __ECOS
+#include <cyg/fileio/fileio.h>
+MTAB_ENTRY( testfs_mte1,
+ "/",
+ "ramfs",
+ "",
+ 0);
+#endif
int main(int argc, char *argv[])
{
int res;
struct Configurable config;
+#ifdef __ECOS
+ char *args[] = {"ecos-curl", "-K", "curlconf.txt"};
+ FILE *f;
+ argc = sizeof(args)/sizeof(args[0]);
+ argv = args;
+
+ f = fopen("curlconf.txt", "w");
+ if (f) {
+ fprintf(f, "--url file:curlconf.txt");
+ fclose(f);
+ }
+#endif
memset(&config, 0, sizeof(struct Configurable));
config.errors = stderr; /* default errors to stderr */
Minix
=====
curl can be compiled on Minix 3 using gcc or ACK (starting with
ver. 3.1.3). Ensure that GNU gawk and bash are both installed and
available in the PATH.
ACK
---
Increase the heap sizes of the compiler with the command:
binsizes xxl
then configure and compile curl with:
./configure CC=cc LD=cc AR=/usr/bin/aal GREP=grep \
CPPFLAGS='-D_POSIX_SOURCE=1 -I/usr/local/include'
make
chmem =256000 src/curl
GCC
---
Make sure gcc is in your PATH with the command:
export PATH=/usr/gnu/bin:$PATH
then configure and compile curl with:
./configure CC=gcc AR=/usr/gnu/bin/gar GREP=grep
make
chmem =256000 src/curl
Symbian OS
==========
The Symbian OS port uses the Symbian build system to compile. From the
packages/Symbian/group/ directory, run:
bldmake bldfiles
abld build
to compile and install curl and libcurl using SBSv1. If your Symbian
SDK doesn't include support for P.I.P.S., you will need to contact
your SDK vendor to obtain that first.
VxWorks
========
Build for VxWorks is performed using cross compilation.
That means you build on Windows machine using VxWorks tools and
run the built image on the VxWorks device.
To build libcurl for VxWorks you need:
- CYGWIN (free, http://cygwin.com/)
- Wind River Workbench (commercial)
If you have CYGWIN and Workbench installed on you machine
follow after next steps:
1. Open the Command Prompt window and change directory ('cd')
to the libcurl 'lib' folder.
2. Add CYGWIN 'bin' folder to the PATH environment variable.
For example, type 'set PATH=C:/embedded/cygwin/bin;%PATH%'.
3. Adjust environment variables defined in 'Environment' section
of the Makefile.vxworks file to point to your software folders.
4. Build the libcurl by typing 'make -f ./Makefile.vxworks'
As a result the libcurl.a library should be created in the 'lib' folder.
To clean the build results type 'make -f ./Makefile.vxworks clean'.
Android
=======
Method using the static makefile:
- see the build notes in the packages/Android/Android.mk file.
Method using a configure cross-compile (tested with Android NDK r7c, r8):
- prepare the toolchain of the Android NDK for standalone use; this can
be done by invoking the script:
./build/tools/make-standalone-toolchain.sh
which creates a usual cross-compile toolchain. Lets assume that you put
this toolchain below /opt then invoke configure with something like:
export PATH=/opt/arm-linux-androideabi-4.4.3/bin:$PATH
./configure --host=arm-linux-androideabi [more configure options]
make
- if you want to compile directly from our GIT repo you might run into
this issue with older automake stuff:
checking host system type...
Invalid configuration `arm-linux-androideabi':
system `androideabi' not recognized
configure: error: /bin/sh ./config.sub arm-linux-androideabi failed
this issue can be fixed with using more recent versions of config.sub
and config.guess which can be obtained here:
http://git.savannah.gnu.org/gitweb/?p=config.git;a=tree
you need to replace your system-own versions which usually can be
found in your automake folder:
find /usr -name config.sub
Wrapper for pkg-config:
- In order to make proper use of pkg-config so that configure is able to
find all dependencies you should create a wrapper script for pkg-config;
file /opt/arm-linux-androideabi-4.4.3/bin/arm-linux-androideabi-pkg-config:
#!/bin/sh
SYSROOT=$(dirname ${0%/*})/sysroot
export PKG_CONFIG_DIR=
export PKG_CONFIG_LIBDIR=${SYSROOT}/usr/local/lib/pkgconfig:${SYSROOT}/usr/share/pkgconfig
export PKG_CONFIG_SYSROOT_DIR=${SYSROOT}
exec pkg-config "$@"
also create a copy or symlink with name arm-unknown-linux-androideabi-pkg-config.
CROSS COMPILE
=============
(This section was graciously brought to us by Jim Duey, with additions by
Dan Fandrich)
Download and unpack the cURL package.
'cd' to the new directory. (e.g. cd curl-7.12.3)
Set environment variables to point to the cross-compile toolchain and call
configure with any options you need. Be sure and specify the '--host' and
'--build' parameters at configuration time. The following script is an
example of cross-compiling for the IBM 405GP PowerPC processor using the
toolchain from MonteVista for Hardhat Linux.
(begin script)
#! /bin/sh
export PATH=$PATH:/opt/hardhat/devkit/ppc/405/bin
export CPPFLAGS="-I/opt/hardhat/devkit/ppc/405/target/usr/include"
export AR=ppc_405-ar
export AS=ppc_405-as
export LD=ppc_405-ld
export RANLIB=ppc_405-ranlib
export CC=ppc_405-gcc
export NM=ppc_405-nm
./configure --target=powerpc-hardhat-linux \
--host=powerpc-hardhat-linux \
--build=i586-pc-linux-gnu \
--prefix=/opt/hardhat/devkit/ppc/405/target/usr/local \
--exec-prefix=/usr/local
(end script)
You may also need to provide a parameter like '--with-random=/dev/urandom'
to configure as it cannot detect the presence of a random number
generating device for a target system. The '--prefix' parameter
specifies where cURL will be installed. If 'configure' completes
successfully, do 'make' and 'make install' as usual.
In some cases, you may be able to simplify the above commands to as
little as:
./configure --host=ARCH-OS
REDUCING SIZE
=============
There are a number of configure options that can be used to reduce the
size of libcurl for embedded applications where binary size is an
important factor. First, be sure to set the CFLAGS variable when
configuring with any relevant compiler optimization flags to reduce the
size of the binary. For gcc, this would mean at minimum the -Os option,
and potentially the -march=X and -mdynamic-no-pic options as well, e.g.
./configure CFLAGS='-Os' ...
Note that newer compilers often produce smaller code than older versions
due to improved optimization.
Be sure to specify as many --disable- and --without- flags on the configure
command-line as you can to disable all the libcurl features that you
know your application is not going to need. Besides specifying the
--disable-PROTOCOL flags for all the types of URLs your application
will not use, here are some other flags that can reduce the size of the
library:
--disable-ares (disables support for the C-ARES DNS library)
--disable-cookies (disables support for HTTP cookies)
--disable-crypto-auth (disables HTTP cryptographic authentication)
--disable-ipv6 (disables support for IPv6)
--disable-manual (disables support for the built-in documentation)
--disable-proxy (disables support for HTTP and SOCKS proxies)
--disable-verbose (eliminates debugging strings and error code strings)
--enable-hidden-symbols (eliminates unneeded symbols in the shared library)
--without-libidn (disables support for the libidn DNS library)
--without-librtmp (disables support for RTMP)
--without-ssl (disables support for SSL/TLS)
--without-zlib (disables support for on-the-fly decompression)
The GNU compiler and linker have a number of options that can reduce the
size of the libcurl dynamic libraries on some platforms even further.
Specify them by providing appropriate CFLAGS and LDFLAGS variables on the
configure command-line, e.g.
CFLAGS="-Os -ffunction-sections -fdata-sections \
-fno-unwind-tables -fno-asynchronous-unwind-tables" \
LDFLAGS="-Wl,-s -Wl,-Bsymbolic -Wl,--gc-sections"
Be sure also to strip debugging symbols from your binaries after
compiling using 'strip' (or the appropriate variant if cross-compiling).
If space is really tight, you may be able to remove some unneeded
sections of the shared library using the -R option to objcopy (e.g. the
.comment section).
Using these techniques it is possible to create a basic HTTP-only shared
libcurl library for i386 Linux platforms that is only 114 KiB in size, and
an FTP-only library that is 115 KiB in size (as of libcurl version 7.35.0,
using gcc 4.8.2).
You may find that statically linking libcurl to your application will
result in a lower total size than dynamically linking.
Note that the curl test harness can detect the use of some, but not all, of
the --disable statements suggested above. Use will cause tests relying on
those features to fail. The test harness can be manually forced to skip
the relevant tests by specifying certain key words on the runtests.pl
command line. Following is a list of appropriate key words:
--disable-cookies !cookies
--disable-manual !--manual
--disable-proxy !HTTP\ proxy !proxytunnel !SOCKS4 !SOCKS5
PORTS
=====
This is a probably incomplete list of known hardware and operating systems
that curl has been compiled for. If you know a system curl compiles and
runs on, that isn't listed, please let us know!
- Alpha DEC OSF 4
- Alpha Digital UNIX v3.2
- Alpha FreeBSD 4.1, 4.5
- Alpha Linux 2.2, 2.4
- Alpha NetBSD 1.5.2
- Alpha OpenBSD 3.0
- Alpha OpenVMS V7.1-1H2
- Alpha Tru64 v5.0 5.1
- AVR32 Linux
- ARM Android 1.5, 2.1, 2.3, 3.2, 4.x
- ARM INTEGRITY
- ARM iOS
- Cell Linux
- Cell Cell OS
- HP-PA HP-UX 9.X 10.X 11.X
- HP-PA Linux
- HP3000 MPE/iX
- MicroBlaze uClinux
- MIPS IRIX 6.2, 6.5
- MIPS Linux
- OS/400
- Pocket PC/Win CE 3.0
- Power AIX 3.2.5, 4.2, 4.3.1, 4.3.2, 5.1, 5.2
- PowerPC Darwin 1.0
- PowerPC INTEGRITY
- PowerPC Linux
- PowerPC Mac OS 9
- PowerPC Mac OS X
- SH4 Linux 2.6.X
- SH4 OS21
- SINIX-Z v5
- Sparc Linux
- Sparc Solaris 2.4, 2.5, 2.5.1, 2.6, 7, 8, 9, 10
- Sparc SunOS 4.1.X
- StrongARM (and other ARM) RISC OS 3.1, 4.02
- StrongARM/ARM7/ARM9 Linux 2.4, 2.6
- StrongARM NetBSD 1.4.1
- Symbian OS (P.I.P.S.) 9.x
- TPF
- Ultrix 4.3a
- UNICOS 9.0
- i386 BeOS
- i386 DOS
- i386 eCos 1.3.1
- i386 Esix 4.1
- i386 FreeBSD
- i386 HURD
- i386 Haiku OS
- i386 Linux 1.3, 2.0, 2.2, 2.3, 2.4, 2.6
- i386 Mac OS X
- i386 MINIX 3.1
- i386 NetBSD
- i386 Novell NetWare
- i386 OS/2
- i386 OpenBSD
- i386 QNX 6
- i386 SCO unix
- i386 Solaris 2.7
- i386 Windows 95, 98, ME, NT, 2000, XP, 2003
- i486 ncr-sysv4.3.03 (NCR MP-RAS)
- ia64 Linux 2.3.99
- m68k AmigaOS 3
- m68k Linux
- m68k uClinux
- m68k OpenBSD
- m88k dg-dgux5.4R3.00
- s390 Linux
- x86_64 Linux
- XScale/PXA250 Linux 2.4
- Nios II uClinux
Useful URLs
===========
axTLS http://axtls.sourceforge.net/
c-ares http://c-ares.haxx.se/
GNU GSS http://www.gnu.org/software/gss/
GnuTLS http://www.gnu.org/software/gnutls/
Heimdal http://www.pdc.kth.se/heimdal/
libidn http://www.gnu.org/software/libidn/
libmetalink https://launchpad.net/libmetalink/
libssh2 http://www.libssh2.org/
MIT Kerberos http://web.mit.edu/kerberos/www/dist/
NSS http://www.mozilla.org/projects/security/pki/nss/
OpenLDAP http://www.openldap.org/
OpenSSL http://www.openssl.org/
PolarSSL http://polarssl.org/
yassl http://www.yassl.com/
Zlib http://www.zlib.net/
MingW http://www.mingw.org/
MinGW-w64 http://mingw-w64.sourceforge.net/
OpenWatcom http://www.openwatcom.org/