newlib-cygwin/winsup/doc/faq-programming.xml

1106 lines
41 KiB
XML

<?xml version="1.0" encoding='UTF-8'?>
<!DOCTYPE article PUBLIC "-//OASIS//DTD DocBook V4.5//EN"
"http://www.oasis-open.org/docbook/xml/4.5/docbookx.dtd">
<!-- faq-programming.xml -->
<qandadiv id="faq.programming">
<title>Programming Questions</title>
<qandaentry id="faq.programming.packages">
<question><para>How do I contribute a package?</para></question>
<answer>
<para>If you are willing to be a package maintainer, great! We urgently need
volunteers to prepare and maintain packages, because the priority of the
Cygwin Team is Cygwin itself.
</para>
<para>The Cygwin Package Contributor's Guide at
<ulink url="https://cygwin.com/packages.html"/> details everything you need to know
about Cygwin packaging.
</para>
<para>For questions about package maintenance, use the cygwin-apps mailing
list (start at <ulink url="https://cygwin.com/lists.html"/>) <emphasis>after</emphasis>
searching and browsing the cygwin-apps list archives, of course. Be
sure to look at the <emphasis>Submitting a package</emphasis> checklist at
<ulink url="https://cygwin.com/packaging-contributors-guide.html#submitting"/>
before sending an ITP (Intent To Package) email to cygwin-apps.
</para>
<para>You should also announce your intentions to the general cygwin list, in
case others were thinking the same thing.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.contribute">
<question><para>How do I contribute to Cygwin?</para></question>
<answer>
<para>If you want to contribute to Cygwin itself, see
<ulink url="https://cygwin.com/contrib.html"/>.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.huge-executables">
<question><para>Why are compiled executables so huge?!?</para></question>
<answer>
<para>By default, gcc compiles in all symbols. You'll also find that gcc
creates large executables on UNIX.
</para>
<para>If that bothers you, just use the 'strip' program, part of the binutils
package. Or compile with the <literal>-s</literal> option to gcc.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.64bitporting">
<question><para>What do I have to look out for when porting applications to 64 bit Cygwin?</para></question>
<answer>
<para>The Cygwin x86_64 toolchain is using the
<ulink url="http://en.wikipedia.org/wiki/LLP64#64-bit_data_models">LP64</ulink>
data model. That means, in contrast to Windows, which uses an
<ulink url="http://en.wikipedia.org/wiki/LLP64#64-bit_data_models">LLP64</ulink>
data model, sizeof(long) != sizeof(int), just as on Linux.</para>
<para>For comparison:</para>
<screen>
Cygwin Windows Cygwin
Linux x86_64 Linux
Windows x86_64
i686
sizeof(int) 4 4 4
sizeof(long) 4 4 8
sizeof(size_t) 4 8 8
sizeof(void*) 4 8 8
</screen>
<para>This difference can result in interesting problems, especially when
using Windows API functions using pointers to Windows
datatypes like LONG, ULONG, DWORD. Given that Windows is LLP64, all of
the aforementioned types are 4 byte in size, on 32 as well as on 64 bit
Windows, while `long' on 64 bit Cygwin is 8 bytes.</para>
<para>Take the example ReadFile:</para>
<screen>
ReadFile (HANDLE, LPVOID, DWORD, LPDWORD, LPOVERLAPPED);
</screen>
<para>In the 32 bit Cygwin and Mingw-w64 environments, as well as in the 64 bit
Mingw-w64 environment, it is no problem to substitute DWORD with unsigned
long:</para>
<screen>
unsigned long number_of_bytes_read;
[...]
ReadFile (fhdl, buf, buflen, &amp;number_of_bytes_read, NULL);
</screen>
<para>However, in 64 bit Cygwin, using LP64, number_of_bytes_read is 8 bytes
in size. But since ReadFile expects a pointer to a 4 byte type, the function
will only change the lower 4 bytes of number_of_bytes_read on return, while
the content of the upper 4 bytes stays undefined.</para>
<para>Here are a few <emphasis>donts</emphasis> which should help porting
applications from the known ILP32 data model of 32 bit Cygwin, to the LP64
data model of 64 bit Cygwin. Note that these are not Cygwin-only problems.
Many Linux applications suffered the same somewhat liberal handling of
datatypes when the AMD64 CPU was new.</para>
<itemizedlist mark="bullet">
<listitem><para>
<emphasis>Don't</emphasis> mix up int and long in printf/scanf. This:
<screen>
int i; long l;
printf ("%d %ld\n", l, i);
</screen>
may not print what you think it should. Enable the gcc options -Wformat or
-Wall, which warn about type mismatches in printf/scanf functions.
<note>Using -Wall (optionally with -Werror to drive the point home) makes a
lot of sense in general, not only when porting code to a new platform.</note>
</para></listitem>
<listitem><para>
<emphasis>Don't</emphasis> mix int and long pointers.
<screen>
long *long_ptr = (long *) &amp;my_int; /* Uh oh! */
*long_ptr = 42;
</screen>
The assignment will write 8 bytes to the address of my_int. Since my_int
is only 4 bytes, <emphasis>something else</emphasis> gets randomly overwritten.
Finding this kind of bug is very hard, because you will often see a problem
which has no immediate connection to the actual bug.
</para></listitem>
<listitem><para>
<emphasis>Don't</emphasis> mix int and pointers at all! This will
<emphasis>not</emphasis> work as expected anymore:
<screen>
void *ptr;
printf ("Pointer value is %x\n", ptr);
</screen>
%x denotes an int argument. The value printed by printf is a 4 byte value,
so on x86_64 the printed pointer value is missing its upper 4 bytes; the output
is very likely wrong. Use %p instead, which portable across architectures:
<screen>
void *ptr;
printf ("Pointer value is %p\n", ptr);
</screen>
</para></listitem>
<listitem><para>
Along the same lines <emphasis>don't</emphasis> use the type int in
pointer arithmetic. Don't cast pointers to int, don't cast pointer
differences to int, and don't store pointer differences in an int type.
Use the types <literal>intptr_t</literal>, <literal>uintptr_t</literal>
and <literal>ptrdiff_t</literal> instead, they are designed for performing
architecture-independent pointer arithmetic.
</para></listitem>
<listitem><para>
<emphasis>Don't</emphasis> make blind assumptions about the size of a POSIX
type. For instance, <literal>time_t</literal> is 8 bytes on 64 bit Cygwin,
while it is (still, at the time of writing this) 4 bytes on 32 bit Cygwin,
since time_t is based on the type long.
</para></listitem>
<listitem><para>
<emphasis>Don't</emphasis> use functions returning pointers without declaration.
For instance
<screen>
printf ("Error message is: %s\n", strerror (errno));
</screen>
This code will <emphasis>crash</emphasis>, unless you included
<filename>string.h</filename>. The implicit rule in C is that an undeclared
function is of type int. But int is 4 byte and pointers are 8 byte, so the
string pointer given to printf is missing the upper 4 bytes.
</para></listitem>
<listitem><para>
<emphasis>Don't</emphasis> use C base types together with Windows API functions.
Keep in mind that DWORD, LONG, ULONG are <emphasis>not</emphasis> the same
as long and unsigned long. Try to use only Windows datatypes in conjunction
with Windows API function calls to avoid type problems. See the above
ReadFile example. Windows functions in printf calls should be treated
carefully as well. This code is common for 32 bit code, but probably prints
the wrong value on 64 bit:
<screen>
printf ("Error message is: %lu\n", GetLastError ());
</screen>
Using gcc's -Wformat option would warn about this. Casting to the requested
base type helps in this case:
<screen>
printf ("Error message is: %lu\n", (unsigned long) GetLastError ());
</screen>
</para></listitem>
<listitem><para>
<emphasis>Don't</emphasis> mix Windows datatypes with POSIX type-specific
MIN/MAX values.
<screen>
unsigned long l_max = ULONG_MAX; /* That's right. */
ULONG w32_biggest = ULONG_MAX; /* Hey, wait! What? */
ULONG w32_biggest = UINT_MAX; /* Ok, but borderline. */
</screen>
Again, keep in mind that ULONG (or DWORD) is <emphasis>not</emphasis> unsigned
long but rather unsigned int on 64 bit.
</para></listitem>
</itemizedlist>
</answer></qandaentry>
<qandaentry id="faq.programming.64bitporting-fail">
<question><para>My project doesn't build at all on 64 bit Cygwin. What's up?</para></question>
<answer>
<para>Typically reasons for that are:</para>
<itemizedlist mark="bullet">
<listitem><para><literal>__CYGWIN32__</literal> is not defined in the
64 bit toolchain. This may hit a few projects which are around since before
Y2K. Check your project for occurences of <literal>__CYGWIN32__</literal>
and change them to <literal>__CYGWIN__</literal>, which is defined in the
Cygwin toolchain since 1998, to get the same Cygwin-specific code changes done.
</para></listitem>
<listitem><para>The project maintainers took it for granted that Cygwin is
running only on i686 CPUs and the code is making this assumption blindly.
You have to check the code for such assumptions and fix them.
</para></listitem>
<listitem><para>The project is using autotools, the
<filename>config.sub</filename> and <filename>config.guess</filename> files
are hopelessly outdated and don't recognize
<literal>x86_64-{pc,unknown}-cygwin</literal> as valid target. Update the
project configury (cygport will do this by default) and try again.
</para></listitem>
<listitem><para>The project uses Windows functions on Cygwin and it's suffering
from the problems described in the preceeding FAQ entry.
</para></listitem>
</itemizedlist>
<para>In all of this cases, please make sure to fix that upstream, or send
your patches to the upstream maintainers, so the problems get fixed for the
future.</para>
</answer></qandaentry>
<qandaentry id="faq.programming.64bitporting-cygwin64">
<question><para>Why is __CYGWIN64__ not defined for 64 bit?</para></question>
<answer>
<para>There is no <literal>__CYGWIN64__</literal> because we would like to
have a unified way to handle Cygwin code in portable projects. Using
<literal>__CYGWIN32__</literal> and <literal>__CYGWIN64__</literal> only
complicates the code for no good reason. Along the same lines you won't
find predefined macros <literal>__linux32__</literal> and
<literal>__linux64__</literal> on Linux.</para>
<para>If you really have to differ between 32 and 64 bit in some way, you have
three choices.</para>
<itemizedlist mark="bullet">
<listitem><para>If your code depends on the CPU architecture, use the
predefined compiler definition for the architecture, like this:</para>
<screen>
#ifdef __CYGWIN__
# ifdef __x86_64__ /* Alternatively __x86_64, __amd64__, __amd64 */
/* Code specific for AMD64 CPU */
# elif __X86__
/* Code specific for ix86 CPUs */
# else
# error Unsupported Architecture
# endif
#endif
</screen></listitem>
<listitem><para>If your code depends on differences in the data model, you
should consider to use the <literal>__LP64__</literal> definition
instead:</para>
<screen>
#ifdef __CYGWIN__
# ifdef __LP64__ /* Alternatively _LP64 */
/* Code specific for 64 bit CPUs */
# else
/* Code specific for 32 bit CPUs */
# endif
#endif
</screen></listitem>
<listitem><para>If your code uses Windows functions, and some of the
functionality is 64 bit Windows-specific, use <literal>_WIN64</literal>,
which is defined on 64 bit Cygwin, as soon as you include
<filename>windows.h</filename>. This should only be used in the most
desperate of occasions, though, and <emphasis>only</emphasis> if it's
really about a difference in Windows API functionality!</para>
<screen>
#ifdef __CYGWIN__
# ifdef _WIN64
/* Code specific for 64 bit Windows */
# else
/* Code specific for 32 bit Windows */
# endif
#endif
</screen></listitem>
</itemizedlist>
</answer></qandaentry>
<qandaentry id="faq.programming.glibc">
<question><para>Where is glibc?</para></question>
<answer>
<para>Cygwin does not provide glibc. It uses newlib instead, which provides
much (but not all) of the same functionality. Porting glibc to Cygwin
would be difficult.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.objective-c">
<question><para>Where is Objective C?</para></question>
<answer>
<para>Support for compiling Objective C is available in the <literal>gcc-objc</literal>
package; resulting binaries will depend on the <literal>libobjc2</literal>
package at runtime.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.make-execvp">
<question><para>Why does my make fail on Cygwin with an execvp error? </para></question>
<answer>
<para>Beware of using non-portable shell features in your Makefiles (see tips
at <xref linkend="faq.using.shell-scripts"></xref>).
</para>
<para>Errors of <literal>make: execvp: /bin/sh: Illegal Argument</literal> or
<literal>make: execvp: /bin/sh: Argument list too long</literal> are often
caused by the command-line being to long for the Windows execution model.
To circumvent this, mount the path of the executable using the -X switch
to enable cygexec for all executables in that folder; you will also need
to exclude non-cygwin executables with the -x switch. Enabling cygexec
causes cygwin executables to talk directly to one another, which increases
the command-line limit. To enable cygexec for <literal>/bin</literal> and
<literal>/usr/bin</literal>, you can add or change these entries in /etc/fstab:
</para>
<screen>
C:/cygwin/bin /bin ntfs binary,cygexec 0 0
C:/cygwin/bin /usr/bin ntfs binary,cygexec 0 0
</screen>
<para>If you have added other non-Cygwin programs to a path you want to mount
cygexec, you can find them with a script like this:
</para>
<screen>
#!/bin/sh
cd /bin; for f in `find . -type f -name '*.exe'`; do
cygcheck $f | (fgrep -qi cygwin1.dll || echo $f)
done
</screen>
<para>
See <ulink url="https://cygwin.com/cygwin-ug-net/using.html#mount-table"/>
for more information on using mount.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.ipc">
<question><para>How can I use IPC, or why do I get a <literal>Bad system call</literal>
error?</para></question>
<answer>
<para>
Try running cygserver. Read
<ulink url="https://cygwin.com/cygwin-ug-net/using-cygserver.html"/>. If you're
trying to use PostgreSQL, also read
<literal>/usr/share/doc/Cygwin/postgresql-*.README</literal>.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.winmain">
<question><para>Why the undefined reference to <literal>WinMain@16</literal>?</para></question>
<answer>
<para>If you're using <literal>gcc</literal>, try adding an empty main() function to one
of your sources. Or, perhaps you have <literal>-lm</literal> too early in the
link command line. It should be at the end:
</para>
<screen>
bash$ gcc hello.c -lm
bash$ ./a.exe
Hello World!
</screen>
<para>works, but
</para>
<screen>
bash$ gcc -lm hello.c
/c/TEMP/ccjLEGlU.o(.text+0x10):hello.c: multiple definition of `main'
/usr/lib/libm.a(libcmain.o)(.text+0x0):libcmain.c: first defined here
/usr/lib/libm.a(libcmain.o)(.text+0x6a):libcmain.c: undefined reference to `WinMain@16'
collect2: ld returned 1 exit status
</screen>
<para>If you're using GCJ, you need to pass a "--main" flag:
</para>
<screen>
gcj --main=Hello Hello.java
</screen>
</answer></qandaentry>
<qandaentry id="faq.programming.win32-api">
<question><para>How do I use Windows API calls?</para></question>
<answer>
<para>Cygwin tools require that you explicitly link the import libraries
for whatever Windows API functions that you are going to use, with the exception
of kernel32, which is linked automatically (because the startup and/or
built-in code uses it).
</para>
<para>For example, to use graphics functions (GDI) you must link
with gdi32 like this:
</para>
<para>gcc -o foo.exe foo.o bar.o -lgdi32
</para>
<para>or (compiling and linking in one step):
</para>
<para>gcc -o foo.exe foo.c bar.c -lgdi32
</para>
<para>The regular setup allows you to use the option -mwindows on the
command line to include a set of the basic libraries (and also
make your program a GUI program instead of a console program),
including user32, gdi32 and comdlg32.
</para>
<para>It is a good idea to put import libraries last on your link line,
or at least after all the object files and static libraries that reference them.
</para>
<note><para>There are a few restrictions for calls to the Windows API.
For details, see the User's Guide section
<ulink url="https://cygwin.com/cygwin-ug-net/setup-env.html#setup-env-win32">Restricted Win32 environment</ulink>,
as well as the User's Guide section
<ulink url="https://cygwin.com/cygwin-ug-net/using.html#pathnames-win32-api">Using the Win32 file API in Cygwin applications</ulink>.</para></note>
</answer></qandaentry>
<qandaentry id="faq.programming.win32-no-cygwin">
<question><para>How do I compile a Windows executable that doesn't use Cygwin?</para></question>
<answer>
<para>The compilers provided by the <literal>mingw64-i686-gcc</literal> and
<literal>mingw64-x86_64-gcc</literal> packages link against standard Microsoft
DLLs instead of Cygwin. This is desirable for native Windows programs that
don't need a UNIX emulation layer.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.static-linking">
<question><para>Can I build a Cygwin program that does not require cygwin1.dll at runtime?</para></question>
<answer>
<para>No. If your program uses the Cygwin API, then your executable cannot
run without cygwin1.dll. In particular, it is not possible to
statically link with a Cygwin library to obtain an independent,
self-contained executable.
</para>
<para>If this is an issue because you intend to distribute your Cygwin
application, then you had better read and understand
<ulink url="https://cygwin.com/licensing.html"/>, which explains the
licensing options.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.msvcrt-and-cygwin">
<question><para>Can I link with both MSVCRT*.DLL and cygwin1.dll?</para></question>
<answer>
<para>No, you must use one or the other, they are mutually exclusive.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.no-console-window">
<question><para>How do I make the console window go away?</para></question>
<answer>
<para>The default during compilation is to produce a console application.
It you are writing a GUI program, you should either compile with
-mwindows as explained above, or add the string
"-Wl,--subsystem,windows" to the GCC command line.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.make-spaces">
<question><para>Why does make complain about a "missing separator"?</para></question>
<answer>
<para>This problem usually occurs as a result of someone editing a Makefile
with a text editor that replaces tab characters with spaces. Command
lines must start with tabs. This is not specific to Cygwin.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.win32-headers">
<question><para>Why can't we redistribute Microsoft's Windows API headers?</para></question>
<answer>
<para>Subsection 2.d.f of the `Microsoft Open Tools License agreement' looks
like it says that one may not "permit further redistribution of the
Redistributables to their end users". We take this to mean that we can
give them to you, but you can't give them to anyone else, which is
something that we can't agree to. Fortunately, we
have our own Windows API headers which are pretty complete.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.msvs-mingw">
<question><para>How do I use <literal>cygwin1.dll</literal> with Visual Studio or Mingw-w64?</para></question>
<answer>
<para>If you want to load the DLL dynamically, read
<literal>winsup/cygwin/how-cygtls-works.txt</literal> and the sample code in
<literal>winsup/testsuite/cygload</literal> to understand how this works.
The short version is:
</para>
<orderedlist><listitem><para>Make sure you have 4K of scratch space at the bottom of your stack.
</para></listitem>
<listitem><para>Invoke <literal>cygwin_dll_init()</literal>:
<screen>
HMODULE h = LoadLibrary("cygwin1.dll");
void (*init)() = GetProcAddress(h, "cygwin_dll_init");
init();
</screen>
</para></listitem>
</orderedlist>
<para>If you want to link statically from Visual Studio, to my knowledge
none of the Cygwin developers have done this, but we have this report
from the mailing list that it can be done this way:
</para>
<orderedlist><listitem><para>Use the impdef program to generate a .def file for the cygwin1.dll
(if you build the cygwin dll from source, you will already have a def
file)
</para>
<screen>
impdef cygwin1.dll &gt; cygwin1.def
</screen>
</listitem>
<listitem><para>Use the MS VS linker (lib) to generate an import library
</para>
<screen>
lib /def=cygwin1.def /out=cygwin1.lib
</screen>
</listitem>
<listitem><para>Create a file "my_crt0.c" with the following contents
</para>
<screen>
#include &lt;sys/cygwin.h&gt;
#include &lt;stdlib.h&gt;
typedef int (*MainFunc) (int argc, char *argv[], char **env);
void
my_crt0 (MainFunc f)
{
cygwin_crt0(f);
}
</screen>
</listitem>
<listitem><para>Use gcc in a Cygwin prompt to build my_crt0.c into a DLL
(e.g. my_crt0.dll). Follow steps 1 and 2 to generate .def and
.lib files for the DLL.
</para>
</listitem>
<listitem><para>Download crt0.c from the cygwin website and include it in
your sources. Modify it to call my_crt0() instead of
cygwin_crt0().
</para>
</listitem>
<listitem><para>Build your object files using the MS VC compiler cl.
</para>
</listitem>
<listitem><para>Link your object files, cygwin1.lib, and my_crt0.lib (or
whatever you called it) into the executable.
</para></listitem>
</orderedlist>
<para>Note that if you are using any other Cygwin based libraries
that you will probably need to build them as DLLs using gcc and
then generate import libraries for the MS VC linker.
</para>
<para>Thanks to Alastair Growcott (alastair dot growcott at bakbone dot co
dot uk) for this tip.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.linking-lib">
<question><para>How do I link against a <literal>.lib</literal> file?</para></question>
<answer>
<para>If your <literal>.lib</literal> file is a normal static or import library with
C-callable entry points, you can list <literal>foo.lib</literal> as an object file for
gcc/g++, just like any <literal>*.o</literal> file. Otherwise, here are some steps:
</para>
<orderedlist><listitem><para>Build a C file with a function table. Put all functions you intend
to use in that table. This forces the linker to include all the object
files from the .lib. Maybe there is an option to force LINK.EXE to
include an object file.
</para></listitem>
<listitem><para>Build a dummy 'LibMain'.
</para></listitem>
<listitem><para>Build a .def with all the exports you need.
</para></listitem>
<listitem><para>Link with your .lib using link.exe.
</para></listitem>
</orderedlist>
<para>or
</para>
<orderedlist><listitem><para>Extract all the object files from the .lib using LIB.EXE.
</para></listitem>
<listitem><para>Build a dummy C file referencing all the functions you need, either
with a direct call or through an initialized function pointer.
</para></listitem>
<listitem><para>Build a dummy LibMain.
</para></listitem>
<listitem><para>Link all the objects with this file+LibMain.
</para></listitem>
<listitem><para>Write a .def.
</para></listitem>
<listitem><para>Link.
</para></listitem>
</orderedlist>
<para>You can use these methods to use MSVC (and many other runtime libs)
with Cygwin development tools.
</para>
<para>Note that this is a lot of work (half a day or so), but much less than
rewriting the runtime library in question from specs...
</para>
<para>Thanks to Jacob Navia (root at jacob dot remcomp dot fr) for this explanation.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.building-cygwin" xreflabel="Build Cygwin">
<question><para>How do I build Cygwin on my own?</para></question>
<answer>
<para>First, you need to make sure you have the necessary build tools
installed; you at least need <literal>gcc-g++</literal>,
<literal>make</literal>, <literal>automake</literal>,
<literal>autoconf</literal>, <literal>git</literal>, <literal>perl</literal>,
<literal>cocom</literal> and <literal>patch</literal>.
</para>
<para>
Additionally, building the <code>dumper</code> utility requires
<literal>gettext-devel</literal>, <literal>libiconv-devel</literal>, <literal>libzstd-devel</literal> and
<literal>zlib-devel</literal>. Building this program can be disabled with the
<literal>--disable-dumper</literal> option to <literal>configure</literal>.
</para>
<para>
Building for 64-bit Cygwin also requires
<literal>mingw64-x86_64-gcc-g++</literal> and
<literal>mingw64-x86_64-zlib</literal>.
</para>
<!-- If you want to run the tests, <literal>dejagnu</literal> is also required. -->
<para>
Building the documentation also requires the <literal>dblatex</literal>,
<literal>docbook2X</literal>, <literal>docbook-xml45</literal>,
<literal>docbook-xsl</literal>, and <literal>xmlto</literal> packages. Building
the documentation can be disabled with the <literal>--disable-doc</literal>
option to <literal>configure</literal>.
</para>
<para>Next, check out the Cygwin sources from the
<ulink url="https://cygwin.com/git.html">Cygwin GIT source repository</ulink>).
This is the <emphasis>preferred method</emphasis> for acquiring the sources.
Otherwise, if you are trying to duplicate a cygwin release then you should
download the corresponding source package
(<literal>cygwin-x.y.z-n-src.tar.bz2</literal>). </para>
<para>You <emphasis>must</emphasis> build cygwin in a separate directory from
the source, so create something like a <literal>build/</literal> directory.
Assuming you checked out the source to
<literal>/oss/src/newlib-cygwin/</literal>, and you want to install to the
temporary location <literal>/oss/install/</literal>, these are the required
steps to build Cygwin:
</para>
<screen>
$ mkdir -p /oss/src/newlib-cygwin/build # create build dir
$ mkdir -p /oss/install # create install dir
$ cd /oss/src/newlib-cygwin/winsup # chdir into Cygwin source dir and...
$ ./autogen.sh # create config files
$ cd /oss/src/newlib-cygwin/build # chdir into build dir
$ # create makefiles...
$ /oss/src/newlib-cygwin/configure --prefix=/oss/install
$ make # build Cygwin
$ make install # install Cygwin into install dir
</screen>
<para>
If the build worked, you can install everything you like into the currently
running system, <emphasis>except</emphasis> the Cygwin DLL
<command>cygwin1.dll</command> itself. For installing the DLL, close down
all Cygwin programs (including bash windows, any servers like
<command>cygserver</command>, etc.), save your old dll, and copy the new dll
to the correct place. Then, for first testing, start up a Cygwin program
from the Windows command prompt and see what happens.
</para>
<para>If you get a lengthy error messages like
<literal>"user shared memory version mismatch detected"</literal>, it's
very likely a Cygwin process still running using the old DLL. Kill it
in Windows' Task Manager or <command>taskkill</command> and try again.
If it's still not working, and if you're sure there's no older Cygwin
process still running, it's probably a bug you introduced with your changes.
From here on, you're on your own or discuss problems on the
<ulink url="https://cygwin.com/lists.html">Cygwin-developers mailing list</ulink>.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.debugging-cygwin">
<question><para>I may have found a bug in Cygwin, how can I debug it (the symbols in gdb look funny)?</para></question>
<answer>
<para>Debugging symbols are stripped from distibuted Cygwin binaries, so to
debug with <command>gdb</command> you will need to install the
<package>cygwin-debuginfo</package> package to obtain the debug symbols for
<filename>cygwin1.dll</filename>
</para>
<para>
If your bug causes an exception inside <filename>cygwin1.dll</filename> you will
need to use the <command>gdb</command> command <userinput>set cygwin-exceptions
on</userinput> to tell <command>gdb</command> to stop on exceptions inside the
Cygwin DLL (by default they are ignored, as they may be generated during normal
operation e.g. when checking a pointer is valid)
</para>
<para>
It is also a good idea to use the latest code in case the bug has been fixed,
so we recommend trying the latest cygwin test release (see
<xref linkend="faq.setup.testrels"></xref>) or building the DLL from git.
</para>
<para>To build a debugging version of the Cygwin DLL, you will need to follow
the instructions at <xref linkend="faq.programming.building-cygwin"></xref>.
</para>
<para>
You can also contact the mailing list for pointers (a simple test case that
demonstrates the bug is always welcome).
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.compiling-unsupported">
<question><para>How can I compile Cygwin for an unsupported platform (PowerPC, Alpha, ARM, Itanium)?</para></question>
<answer>
<para>Unfortunately, this will be difficult. Exception handling and signals
support semantics and args have been designed for x86_64 so you would need
to write specific support for your platform. We don't know of any other
incompatibilities. Please send us patches if you do this work!
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.adjusting-heap">
<question><para>How can I adjust the heap/stack size of an application?</para></question>
<answer>
<para>If you need to change the maximum amount of memory available to Cygwin, see
<ulink url="https://cygwin.com/cygwin-ug-net/setup-maxmem.html"/>. Otherwise,
just pass heap/stack linker arguments to gcc. To create foo.exe with
a heap size of 200MB and a stack size of 8MB, you would invoke
gcc as:
</para>
<para><literal>gcc -Wl,--heap,200000000,--stack,8000000 -o foo foo.c</literal>
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.dll-cygcheck">
<question><para>How can I find out which DLLs are needed by an executable?</para></question>
<answer>
<para><literal>objdump -p</literal> provides this information, but is rather verbose.
</para>
<para><literal>cygcheck</literal> will do this much more concisely, and operates
recursively, provided the command is in your path.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.dll-building">
<question><para>How do I build a DLL?</para></question>
<answer>
<para>There's documentation that explains the process in the Cygwin User's
Guide here: <ulink url="https://cygwin.com/cygwin-ug-net/dll.html"/>.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.breakpoint">
<question><para>How can I set a breakpoint at mainCRTStartup?</para></question>
<answer>
<para>
Set a breakpoint in <command>gdb</command> with <command>b *0x401000</command>
(for i686), or <command>b *0x100401000</command> (for x86_64).
</para>
<para>
This entrypoint address can be computed as the sum of the ImageBase and
AddressOfEntryPoint values given by <command>objdump -p</command>.
</para>
<para>
Note that the DllMain entrypoints for linked DLLs will have been executed
before this breakpoint is hit.
</para>
<para>
(It may be necessary to use the <command>gdb</command> command <command>set
disable-randomization on</command> to turn off ASLR for the debuggee to
prevent the base address getting randomized.)
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.debug">
<question><para>How can I debug what's going on?</para></question>
<answer>
<para>You can debug your application using <literal>gdb</literal>. Make sure you
compile it with the -g flag! If your application calls functions in
MS DLLs, gdb will complain about not being able to load debug information
for them when you run your program. This is normal since these DLLs
don't contain debugging information (and even if they did, that debug
info would not be compatible with gdb).
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.system-trace">
<question><para>Can I use a system trace mechanism instead?</para></question>
<answer>
<para>Yes. You can use the <literal>strace.exe</literal> utility to run other cygwin
programs with various debug and trace messages enabled. For information
on using <literal>strace</literal>, see the Cygwin User's Guide.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.gdb-signals">
<question><para>How does gdb handle signals?</para></question>
<answer>
<para>
gdb maps known Windows exceptions to signals such as SIGSEGV, SIGFPE, SIGTRAP,
SIGINT and SIGILL. Other Windows exceptions are passed on to the handler (if
any), and reported as an unknown signal if an unhandled (second chance)
exception occurs.
</para>
<para>
There is also an experimental feature to notify gdb of purely Cygwin signals
like SIGABRT, SIGHUP or SIGUSR1. This currently has some known problems, for
example, single-stepping from these signals may not work as expected.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.linker">
<question><para>The linker complains that it can't find something.</para></question>
<answer>
<para>A common error is to put the library on the command line before
the thing that needs things from it.
</para>
<para>This is wrong <literal>gcc -lstdc++ hello.cc</literal>.
This is right <literal>gcc hello.cc -lstdc++</literal>.
</para>
<para>
The first command above (usually) works on Linux, because:
<itemizedlist mark="bullet">
<listitem>A DT_NEEDED tag for libstdc++ is added when the library name is seen.</listitem>
<listitem>The executable has unresolved symbols, which can be found in libstdc++.</listitem>
<listitem>When executed, the ELF loader resolves those symbols.</listitem>
</itemizedlist>
</para>
<para>
Note that this won't work if the linker flags <literal>--as-needed</literal>
or <literal>--no-undefined</literal> are used, or if the library being linked
with is a static library.
</para>
<para>
PE/COFF executables work very differently, and the dynamic library which
provides a symbol must be fully resolved <emphasis>at link time</emphasis>
(so the library which provides a symbol must follow a reference to it).
</para>
<para>
See point 3 in <xref linkend="faq.programming.unix-gui"></xref> for more
discussion of how this affects plugins.
</para>
<para>
This also has consequences for how weak symbols are resolved. See <ulink
url="https://cygwin.com/ml/cygwin/2010-04/msg00281.html"></ulink> for more
discussion of that.
</para>
</answer>
</qandaentry>
<qandaentry id="faq.programming.stat64">
<question><para>Why do I get an error using <literal>struct stat64</literal>?</para></question>
<answer>
<para><literal>struct stat64</literal> is not used in Cygwin, just
use <literal>struct stat</literal>. It's 64 bit aware.</para>
</answer></qandaentry>
<qandaentry id="faq.programming.libc">
<question><para>Can you make DLLs that are linked against libc ?</para></question>
<answer>
<para>Yes.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.malloc-h">
<question><para>Where is malloc.h?</para></question>
<answer>
<para>It exists, but you should rather include stdlib.h instead of malloc.h.
stdlib.h is POSIX standard for defining malloc and friends, malloc.h is
definitely non-standard.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.own-malloc">
<question><para>Can I use my own malloc?</para></question>
<answer>
<para>If you define a function called <literal>malloc</literal> in your own code, and link
with the DLL, the DLL <emphasis>will</emphasis> call your <literal>malloc</literal>. Needless to
say, you will run into serious problems if your malloc is buggy.
</para>
<para>If you run any programs from the DOS command prompt, rather than from in
bash, the DLL will try and expand the wildcards on the command line.
This process uses <literal>malloc</literal> <emphasis>before</emphasis> your main line is started.
If you have written your own <literal>malloc</literal> to need some initialization
to occur after <literal>main</literal> is called, then this will surely break.
</para>
<para>Moreover, there is an outstanding issue with <literal>_malloc_r</literal> in
<literal>newlib</literal>. This re-entrant version of <literal>malloc</literal> will be called
directly from within <literal>newlib</literal>, by-passing your custom version, and
is probably incompatible with it. But it may not be possible to replace
<literal>_malloc_r</literal> too, because <literal>cygwin1.dll</literal> does not export it and
Cygwin does not expect your program to replace it. This is really a
newlib issue, but we are open to suggestions on how to deal with it.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.msvc-gcc-objects">
<question><para>Can I mix objects compiled with msvc++ and gcc?</para></question>
<answer>
<para>Yes, but only if you are combining C object files. MSVC C++ uses a
different mangling scheme than GNU C++, so you will have difficulties
combining C++ objects.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.gdb-msvc">
<question><para>Can I use the gdb debugger to debug programs built by VC++?</para></question>
<answer>
<para>No, not for full (high level source language) debugging.
The Microsoft compilers generate a different type of debugging
symbol information, which gdb does not understand.
</para>
<para>However, the low-level (assembly-type) symbols generated by
Microsoft compilers are coff, which gdb DOES understand.
Therefore you should at least be able to see all of your
global symbols; you just won't have any information about
data types, line numbers, local variables etc.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.make-scripts">
<question><para>Shell scripts aren't running properly from my makefiles?</para></question>
<answer>
<para>If your scripts are in the current directory, you must have <literal>.</literal>
(dot) in your $PATH. (It is not normally there by default.) Better yet,
add /bin/sh in front of each and every shell script invoked in your Makefiles.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.preprocessor">
<question><para>What preprocessor macros do I need to know about?</para></question>
<answer>
<para>gcc for Cygwin defines __CYGWIN__ when building for a Cygwin
environment.
</para>
<para>Microsoft defines the preprocessor symbol _WIN32 in their Windows
development environment.
</para>
<para>In gcc for Cygwin, _WIN32 is only defined when you use the -mwin32
gcc command line options. This is because Cygwin is supposed to be a
POSIX emulation environment in the first place and defining _WIN32 confuses
some programs which think that they have to make special concessions for
a Windows environment which Cygwin handles automatically.
</para>
<para>Check out the predefined symbols in detail by running, for example
</para>
<screen>
$ gcc -dM -E -xc /dev/null &gt;gcc.txt
$ gcc -mwin32 -dM -E -xc /dev/null &gt;gcc-mwin32.txt
</screen>
<para>Then use the diff and grep utilities to check what the difference is.
</para>
</answer></qandaentry>
<qandaentry id="faq.programming.unix-gui">
<question><para>How should I port my Unix GUI to Windows?</para></question>
<answer>
<para>Like other Unix-like platforms, the Cygwin distribution includes many of
the common GUI toolkits, including X11, X Athena widgets, Motif, Tk, GTK+,
and Qt. Many programs which rely on these toolkits will work with little, if
any, porting work if they are otherwise portable. However, there are a few
things to look out for:</para>
<orderedlist>
<listitem><para>Some packages written for both Windows and X11 incorrectly
treat Cygwin as a Windows platform rather than a Unix variant. Mixing Cygwin's
Unix APIs with Windows' GDI is best avoided; rather, remove these assumptions
so that Cygwin is treated like other X11 platforms.</para></listitem>
<listitem><para>GTK+ programs which use <literal>gtk_builder_connect_signals()</literal>
or <literal>glade_xml_signal_autoconnect()</literal> need to be able to
<literal>dlopen()</literal> themselves. In order for this to work, the program
must be linked with the <literal>-Wl,--export-all-symbols</literal> linker flag.
This can be added to LDFLAGS manually, or handled automatically with the
<literal>-export-dynamic</literal> libtool flag (requires libtool 2.2.8) or
by adding <literal>gmodule-export-2.0</literal> to the pkg-config modules used
to build the package.</para></listitem>
<listitem><para>Programs which include their own loadable modules (plugins)
often must have its modules linked against the symbols in the program. The
most portable solution is for such programs to provide all its symbols (except
for <literal>main()</literal>) in a shared library, against which the plugins
can be linked. Otherwise, the symbols from the executable itself must be
exported.</para>
<para>If the package uses the CMake build system, this can be done by adding
<literal>ENABLE_EXPORTS TRUE</literal> to the executable's <literal>set_target_properties</literal>
command, then adding the executable's target name to the <literal>target_link_libraries</literal>
command for the plugins.</para>
<para>For other build systems, the following steps are required:</para>
<orderedlist>
<listitem><para>The executable must be built before its plugins.</para></listitem>
<listitem><para>Symbols must be exported from the executable with a
<literal>-Wl,--export-all-symbols,--out-implib,libfoo.exe.a</literal>
linker flag, where <literal>foo</literal> represents the name of the
executable.</para></listitem>
<listitem><para>The plugins must be linked with a <literal>-Wl,/path/to/libfoo.exe.a</literal>
linker flag.</para></listitem>
</orderedlist></listitem></orderedlist>
</answer></qandaentry>
</qandadiv>