openssl/NOTES.WIN
Andy Polyakov 97a479c6f8 NOTES.WIN: classify targets to "native" and "hosted" and restructure.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5647)
2018-03-18 13:29:44 +01:00

165 lines
6.8 KiB
Plaintext

NOTES FOR THE WINDOWS PLATFORMS
===============================
Windows targets can be classified as "native", ones that use Windows API
directly, and "hosted" which rely on POSIX-compatible layer. "Native"
targets are VC-* (where "VC" stems from abbreviating Microsoft Visual C
compiler) and mingw[64]. "Hosted" platforms are Cygwin and MSYS[2]. Even
though the latter is not directly supported by OpenSSL Team, it's #1
popular choice for building MinGW targets. In the nutshell MinGW builds
are always cross-compiled. On Linux and Cygwin they look exactly as such
and require --cross-compile-prefix option. While on MSYS[2] it's solved
rather by placing gcc that produces "MinGW binary" code 1st on $PATH.
This is customarily source of confusion. "Hosted" applications "live" in
emulated file system name space with POSIX-y root, mount points, /dev
and even /proc. Confusion is intensified by the fact that MSYS2 shell
(or rather emulated execve(2) call) examines the binary it's about to
start, and if it's found *not* to be linked with MSYS2 POSIX-y thing,
command line arguments that look like file names get translated from
emulated name space to "native". For example '/c/some/where' becomes
'c:\some\where', '/dev/null' - 'nul'. This creates an illusion that
there is no difference between MSYS2 shell and "MinGW binary", but
there is. Just keep in mind that "MinGW binary" "experiences" Windows
system in exactly same way as one produced by VC, and in its essence
is indistinguishable from the latter. (Which by the way is why
it's referred to in quotes here, as "MinGW binary", it's just as
"native" as it can get.)
Visual C++ builds, a.k.a. VC-*
==============================
Requirement details
-------------------
In addition to the requirements and instructions listed in INSTALL,
these are required as well:
- Perl. We recommend ActiveState Perl, available from
https://www.activestate.com/ActivePerl. Another viable alternative
appears to be Strawberry Perl, http://strawberryperl.com.
You also need the perl module Text::Template, available on CPAN.
Please read NOTES.PERL for more information.
- Microsoft Visual C compiler. Since we can't test them all, there is
unavoidable uncertainty about which versions are supported. Latest
version along with couple of previous are certainly supported. On
the other hand oldest one is known not to work. Everything between
falls into best-effort category.
- Netwide Assembler, a.k.a. NASM, available from http://www.nasm.us,
is required if you intend to utilize assembler modules. Note that NASM
is the only supported assembler. The Microsoft provided assembler is NOT
supported.
Installation directories
------------------------
The default installation directories are derived from environment
variables.
For VC-WIN32, the following defaults are use:
PREFIX: %ProgramFiles(86)%\OpenSSL
OPENSSLDIR: %CommonProgramFiles(86)%\SSL
For VC-WIN64, the following defaults are use:
PREFIX: %ProgramW6432%\OpenSSL
OPENSSLDIR: %CommonProgramW6432%\SSL
Should those environment variables not exist (on a pure Win32
installation for examples), these fallbacks are used:
PREFIX: %ProgramFiles%\OpenSSL
OPENSSLDIR: %CommonProgramFiles%\SSL
ALSO NOTE that those directories are usually write protected, even if
your account is in the Administrators group. To work around that,
start the command prompt by right-clicking on it and choosing "Run as
Administrator" before running 'nmake install'. The other solution
is, of course, to choose a different set of directories by using
--prefix and --openssldir when configuring.
mingw and mingw64
=================
* MSYS2 shell and development environment installation:
Download MSYS2 from https://msys2.github.io/ and follow installation
instructions. Once up and running install even make, perl, (git if
needed,) mingw-w64-i686-gcc and/or mingw-w64-x86_64-gcc. You should
have corresponding MinGW items on your start menu, use *them*, not
generic MSYS2. As implied in opening note, difference between them
is which compiler is found 1st on $PATH. At this point ./config
should recognize correct target, roll as if it was Unix...
* It is also possible to build mingw[64] on Linux or Cygwin by
configuring with corresponding --cross-compile-prefix= option. For
example
./Configure mingw --cross-compile-prefix=i686-w64-mingw32- ...
or
./Configure mingw64 --cross-compile-prefix=x86_64-w64-mingw32- ...
This naturally implies that you've installed corresponding add-on
packages.
Linking your application
========================
This section applies to all "native" builds.
If you link with static OpenSSL libraries then you're expected to
additionally link your application with WS2_32.LIB, GDI32.LIB,
ADVAPI32.LIB, CRYPT32.LIB and USER32.LIB. Those developing
non-interactive service applications might feel concerned about
linking with GDI32.LIB and USER32.LIB, as they are justly associated
with interactive desktop, which is not available to service
processes. The toolkit is designed to detect in which context it's
currently executed, GUI, console app or service, and act accordingly,
namely whether or not to actually make GUI calls. Additionally those
who wish to /DELAYLOAD:GDI32.DLL and /DELAYLOAD:USER32.DLL and
actually keep them off service process should consider implementing
and exporting from .exe image in question own _OPENSSL_isservice not
relying on USER32.DLL. E.g., on Windows Vista and later you could:
__declspec(dllexport) __cdecl BOOL _OPENSSL_isservice(void)
{ DWORD sess;
if (ProcessIdToSessionId(GetCurrentProcessId(),&sess))
return sess==0;
return FALSE;
}
If you link with OpenSSL .DLLs, then you're expected to include into
your application code small "shim" snippet, which provides glue between
OpenSSL BIO layer and your compiler run-time. See the OPENSSL_Applink
manual page for further details.
Cygwin, "hosted" environment
============================
Cygwin implements a Posix/Unix runtime system (cygwin1.dll) on top of the
Windows subsystem and provides a bash shell and GNU tools environment.
Consequently, a make of OpenSSL with Cygwin is virtually identical to the
Unix procedure.
To build OpenSSL using Cygwin, you need to:
* Install Cygwin (see https://cygwin.com/)
* Install Cygwin Perl and ensure it is in the path. Recall that
as least 5.10.0 is required.
* Run the Cygwin bash shell
Apart from that, follow the Unix instructions in INSTALL.
NOTE: "make test" and normal file operations may fail in directories
mounted as text (i.e. mount -t c:\somewhere /home) due to Cygwin
stripping of carriage returns. To avoid this ensure that a binary
mount is used, e.g. mount -b c:\somewhere /home.