Retro68/gcc/libstdc++-v3/doc/html/manual/using_exceptions.html
2017-10-07 02:16:47 +02:00

313 lines
19 KiB
HTML
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Exceptions</title><meta name="generator" content="DocBook XSL-NS Stylesheets V1.78.1" /><meta name="keywords" content="C++, exception, error, exception neutrality, exception safety, exception propagation, -fno-exceptions" /><meta name="keywords" content="ISO C++, library" /><meta name="keywords" content="ISO C++, runtime, library" /><link rel="home" href="../index.html" title="The GNU C++ Library" /><link rel="up" href="using.html" title="Chapter 3. Using" /><link rel="prev" href="using_concurrency.html" title="Concurrency" /><link rel="next" href="debug.html" title="Debugging Support" /></head><body><div class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="3" align="center">Exceptions</th></tr><tr><td width="20%" align="left"><a accesskey="p" href="using_concurrency.html">Prev</a> </td><th width="60%" align="center">Chapter 3. Using</th><td width="20%" align="right"> <a accesskey="n" href="debug.html">Next</a></td></tr></table><hr /></div><div class="section"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="manual.intro.using.exceptions"></a>Exceptions</h2></div></div></div><p>
The C++ language provides language support for stack unwinding
with <code class="literal">try</code> and <code class="literal">catch</code> blocks and
the <code class="literal">throw</code> keyword.
</p><p>
These are very powerful constructs, and require some thought when
applied to the standard library in order to yield components that work
efficiently while cleaning up resources when unexpectedly killed via
exceptional circumstances.
</p><p>
Two general topics of discussion follow:
exception neutrality and exception safety.
</p><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="intro.using.exception.safety"></a>Exception Safety</h3></div></div></div><p>
What is exception-safe code?
</p><p>
Will define this as reasonable and well-defined behavior by classes
and functions from the standard library when used by user-defined
classes and functions that are themselves exception safe.
</p><p>
Please note that using exceptions in combination with templates
imposes an additional requirement for exception
safety. Instantiating types are required to have destructors that
do no throw.
</p><p>
Using the layered approach from Abrahams, can classify library
components as providing set levels of safety. These will be called
exception guarantees, and can be divided into three categories.
</p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
One. Don't throw.
</p><p>
As specified in 23.2.1 general container requirements. Applicable
to container and string classes.
</p><p>
Member
functions <code class="function">erase</code>, <code class="function">pop_back</code>, <code class="function">pop_front</code>, <code class="function">swap</code>, <code class="function">clear</code>. And <span class="type">iterator</span>
copy constructor and assignment operator.
</p></li><li class="listitem"><p>
Two. Don't leak resources when exceptions are thrown. This is
also referred to as the <span class="quote"><span class="quote">basic</span></span> exception safety guarantee.
</p><p>
This applicable throughout the standard library.
</p></li><li class="listitem"><p>
Three. Commit-or-rollback semantics. This is
referred to as <span class="quote"><span class="quote">strong</span></span> exception safety guarantee.
</p><p>
As specified in 23.2.1 general container requirements. Applicable
to container and string classes.
</p><p>
Member functions <code class="function">insert</code> of a single
element, <code class="function">push_back</code>, <code class="function">push_front</code>,
and <code class="function">rehash</code>.
</p></li></ul></div></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="intro.using.exception.propagating"></a>Exception Neutrality</h3></div></div></div><p>
Simply put, once thrown an exception object should continue in
flight unless handled explicitly. In practice, this means
propagating exceptions should not be swallowed in
gratuitous <code class="literal">catch(...)</code> blocks. Instead,
matching <code class="literal">try</code> and <code class="literal">catch</code>
blocks should have specific catch handlers and allow un-handed
exception objects to propagate. If a
terminating <code class="literal">catch(...)</code> blocks exist then it
should end with a <code class="literal">throw</code> to re-throw the current
exception.
</p><p>
Why do this?
</p><p>
By allowing exception objects to propagate, a more flexible
approach to error handling is made possible (although not
required.) Instead of dealing with an error immediately, one can
allow the exception to propagate up until sufficient context is
available and the choice of exiting or retrying can be made in an
informed manner.
</p><p>
Unfortunately, this tends to be more of a guideline than a strict
rule as applied to the standard library. As such, the following is
a list of known problem areas where exceptions are not propagated.
</p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p>
Input/Output
</p><p>
The destructor <code class="function">ios_base::Init::~Init()</code>
swallows all exceptions from <code class="function">flush</code> called on
all open streams at termination.
</p><p>
All formatted input in <code class="classname">basic_istream</code> or
formatted output in <code class="classname">basic_ostream</code> can be
configured to swallow exceptions
when <code class="function">exceptions</code> is set to
ignore <span class="type">ios_base::badbit</span>.
</p><p>
Functions that have been registered
with <code class="function">ios_base::register_callback</code> swallow all
exceptions when called as part of a callback event.
</p><p>
When closing the underlying
file, <code class="function">basic_filebuf::close</code> will swallow
(non-cancellation) exceptions thrown and return <code class="literal">NULL</code>.
</p></li><li class="listitem"><p>
Thread
</p><p>
The constructors of <code class="classname">thread</code> that take a
callable function argument swallow all exceptions resulting from
executing the function argument.
</p></li></ul></div></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="intro.using.exception.no"></a>Doing without</h3></div></div></div><p>
C++ is a language that strives to be as efficient as is possible
in delivering features. As such, considerable care is used by both
language implementer and designers to make sure unused features
not impose hidden or unexpected costs. The GNU system tries to be
as flexible and as configurable as possible. So, it should come as
no surprise that GNU C++ provides an optional language extension,
spelled <code class="literal">-fno-exceptions</code>, as a way to excise the
implicitly generated magic necessary to
support <code class="literal">try</code> and <code class="literal">catch</code> blocks
and thrown objects. (Language support
for <code class="literal">-fno-exceptions</code> is documented in the GNU
GCC <a class="link" href="http://gcc.gnu.org/onlinedocs/gcc/Code-Gen-Options.html#Code-Gen-Options" target="_top">manual</a>.)
</p><p>Before detailing the library support
for <code class="literal">-fno-exceptions</code>, first a passing note on
the things lost when this flag is used: it will break exceptions
trying to pass through code compiled
with <code class="literal">-fno-exceptions</code> whether or not that code
has any <code class="literal">try</code> or <code class="literal">catch</code>
constructs. If you might have some code that throws, you shouldn't
use <code class="literal">-fno-exceptions</code>. If you have some code that
uses <code class="literal">try</code> or <code class="literal">catch</code>, you
shouldn't use <code class="literal">-fno-exceptions</code>.
</p><p>
And what it to be gained, tinkering in the back alleys with a
language like this? Exception handling overhead can be measured
in the size of the executable binary, and varies with the
capabilities of the underlying operating system and specific
configuration of the C++ compiler. On recent hardware with GNU
system software of the same age, the combined code and data size
overhead for enabling exception handling is around 7%. Of course,
if code size is of singular concern than using the appropriate
optimizer setting with exception handling enabled
(ie, <code class="literal">-Os -fexceptions</code>) may save up to twice
that, and preserve error checking.
</p><p>
So. Hell bent, we race down the slippery track, knowing the brakes
are a little soft and that the right front wheel has a tendency to
wobble at speed. Go on: detail the standard library support
for <code class="literal">-fno-exceptions</code>.
</p><p>
In sum, valid C++ code with exception handling is transformed into
a dialect without exception handling. In detailed steps: all use
of the C++
keywords <code class="literal">try</code>, <code class="literal">catch</code>,
and <code class="literal">throw</code> in the standard library have been
permanently replaced with the pre-processor controlled equivalents
spelled <code class="literal">__try</code>, <code class="literal">__catch</code>,
and <code class="literal">__throw_exception_again</code>. They are defined
as follows.
</p><pre class="programlisting">
#if __cpp_exceptions
# define __try try
# define __catch(X) catch(X)
# define __throw_exception_again throw
#else
# define __try if (true)
# define __catch(X) if (false)
# define __throw_exception_again
#endif
</pre><p>
In addition, for every object derived from
class <code class="classname">exception</code>, there exists a corresponding
function with C language linkage. An example:
</p><pre class="programlisting">
#if __cpp_exceptions
void __throw_bad_exception(void)
{ throw bad_exception(); }
#else
void __throw_bad_exception(void)
{ abort(); }
#endif
</pre><p>
The last language feature needing to be transformed
by <code class="literal">-fno-exceptions</code> is treatment of exception
specifications on member functions. Fortunately, the compiler deals
with this by ignoring exception specifications and so no alternate
source markup is needed.
</p><p>
By using this combination of language re-specification by the
compiler, and the pre-processor tricks and the functional
indirection layer for thrown exception objects by the library,
libstdc++ files can be compiled
with <code class="literal">-fno-exceptions</code>.
</p><p>
User code that uses C++ keywords
like <code class="literal">throw</code>, <code class="literal">try</code>,
and <code class="literal">catch</code> will produce errors even if the user
code has included libstdc++ headers and is using constructs
like <code class="classname">basic_iostream</code>. Even though the standard
library has been transformed, user code may need modification. User
code that attempts or expects to do error checking on standard
library components compiled with exception handling disabled should
be evaluated and potentially made conditional.
</p><p>
Some issues remain with this approach (see bugzilla entry
25191). Code paths are not equivalent, in
particular <code class="literal">catch</code> blocks are not evaluated. Also
problematic are <code class="literal">throw</code> expressions expecting a
user-defined throw handler. Known problem areas in the standard
library include using an instance
of <code class="classname">basic_istream</code>
with <code class="function">exceptions</code> set to specific
<span class="type">ios_base::iostate</span> conditions, or
cascading <code class="literal">catch</code> blocks that dispatch error
handling or recovery efforts based on the type of exception object
thrown.
</p><p>
Oh, and by the way: none of this hackery is at all
special. (Although perhaps well-deserving of a raised eyebrow.)
Support continues to evolve and may change in the future. Similar
and even additional techniques are used in other C++ libraries and
compilers.
</p><p>
C++ hackers with a bent for language and control-flow purity have
been successfully consoled by grizzled C veterans lamenting the
substitution of the C language keyword
<code class="literal">const</code> with the uglified
doppelganger <code class="literal">__const</code>.
</p></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="intro.using.exception.compat"></a>Compatibility</h3></div></div></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="using.exception.compat.c"></a>With <code class="literal">C</code></h4></div></div></div><p>
C language code that is expecting to interoperate with C++ should be
compiled with <code class="literal">-fexceptions</code>. This will make
debugging a C language function called as part of C++-induced stack
unwinding possible.
</p><p>
In particular, unwinding into a frame with no exception handling
data will cause a runtime abort. If the unwinder runs out of unwind
info before it finds a handler, <code class="function">std::terminate()</code>
is called.
</p><p>
Please note that most development environments should take care of
getting these details right. For GNU systems, all appropriate parts
of the GNU C library are already compiled
with <code class="literal">-fexceptions</code>.
</p></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="using.exception.compat.posix"></a>With <code class="literal">POSIX</code> thread cancellation</h4></div></div></div><p>
GNU systems re-use some of the exception handling mechanisms to
track control flow for <code class="literal">POSIX</code> thread cancellation.
</p><p>
Cancellation points are functions defined by POSIX as worthy of
special treatment. The standard library may use some of these
functions to implement parts of the ISO C++ standard or depend on
them for extensions.
</p><p>
Of note:
</p><p>
<code class="function">nanosleep</code>,
<code class="function">read</code>, <code class="function">write</code>, <code class="function">open</code>, <code class="function">close</code>,
and <code class="function">wait</code>.
</p><p>
The parts of libstdc++ that use C library functions marked as
cancellation points should take pains to be exception neutral.
Failing this, <code class="literal">catch</code> blocks have been augmented to
show that the POSIX cancellation object is in flight.
</p><p>
This augmentation adds a <code class="literal">catch</code> block
for <code class="classname">__cxxabiv1::__forced_unwind</code>, which is the
object representing the POSIX cancellation object. Like so:
</p><pre class="programlisting">
catch(const __cxxabiv1::__forced_unwind&amp;)
{
this-&gt;_M_setstate(ios_base::badbit);
throw;
}
catch(...)
{ this-&gt;_M_setstate(ios_base::badbit); }
</pre></div></div><div class="bibliography"><div class="titlepage"><div><div><h3 class="title"><a id="using.exceptions.biblio"></a>Bibliography</h3></div></div></div><div class="biblioentry"><a id="id-1.3.3.4.9.9.2"></a><p><span class="title"><em>
<a class="link" href="http://www.opengroup.org/austin/" target="_top">
System Interface Definitions, Issue 7 (IEEE Std. 1003.1-2008)
</a>
</em>. </span><span class="pagenums">
2.9.5 Thread Cancellation
. </span><span class="copyright">Copyright © 2008
The Open Group/The Institute of Electrical and Electronics
Engineers, Inc.
. </span></p></div><div class="biblioentry"><a id="id-1.3.3.4.9.9.3"></a><p><span class="title"><em>
<a class="link" href="http://www.boost.org/community/error_handling.html" target="_top">
Error and Exception Handling
</a>
</em>. </span><span class="author"><span class="firstname">David</span> <span class="surname">Abrahams </span>. </span><span class="publisher"><span class="publishername">
Boost
. </span></span></p></div><div class="biblioentry"><a id="id-1.3.3.4.9.9.4"></a><p><span class="title"><em>
<a class="link" href="http://www.boost.org/community/exception_safety.html" target="_top">
Exception-Safety in Generic Components
</a>
</em>. </span><span class="author"><span class="firstname">David</span> <span class="surname">Abrahams</span>. </span><span class="publisher"><span class="publishername">
Boost
. </span></span></p></div><div class="biblioentry"><a id="id-1.3.3.4.9.9.5"></a><p><span class="title"><em>
<a class="link" href="http://www.open-std.org/jtc1/sc22/wg21/docs/papers/1997/N1077.pdf" target="_top">
Standard Library Exception Policy
</a>
</em>. </span><span class="author"><span class="firstname">Matt</span> <span class="surname">Austern</span>. </span><span class="publisher"><span class="publishername">
WG21 N1077
. </span></span></p></div><div class="biblioentry"><a id="id-1.3.3.4.9.9.6"></a><p><span class="title"><em>
<a class="link" href="http://gcc.gnu.org/ml/gcc-patches/2001-03/msg00661.html" target="_top">
ia64 c++ abi exception handling
</a>
</em>. </span><span class="author"><span class="firstname">Richard</span> <span class="surname">Henderson</span>. </span><span class="publisher"><span class="publishername">
GNU
. </span></span></p></div><div class="biblioentry"><a id="id-1.3.3.4.9.9.7"></a><p><span class="title"><em>
<a class="link" href="http://www.stroustrup.com/3rd_safe.pdf" target="_top">
Appendix E: Standard-Library Exception Safety
</a>
</em>. </span><span class="author"><span class="firstname">Bjarne</span> <span class="surname">Stroustrup</span>. </span></p></div><div class="biblioentry"><a id="id-1.3.3.4.9.9.8"></a><p><span class="citetitle"><em class="citetitle">
Exceptional C++
</em>. </span><span class="pagenums">
Exception-Safety Issues and Techniques
. </span><span class="author"><span class="firstname">Herb</span> <span class="surname">Sutter</span>. </span></p></div><div class="biblioentry"><a id="id-1.3.3.4.9.9.9"></a><p><span class="title"><em>
<a class="link" href="http://gcc.gnu.org/PR25191" target="_top">
GCC Bug 25191: exception_defines.h #defines try/catch
</a>
</em>. </span></p></div></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="using_concurrency.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="using.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="debug.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">Concurrency </td><td width="20%" align="center"><a accesskey="h" href="../index.html">Home</a></td><td width="40%" align="right" valign="top"> Debugging Support</td></tr></table></div></body></html>