1========================== 2Exception Handling in LLVM 3========================== 4 5.. contents:: 6 :local: 7 8Introduction 9============ 10 11This document is the central repository for all information pertaining to 12exception handling in LLVM. It describes the format that LLVM exception 13handling information takes, which is useful for those interested in creating 14front-ends or dealing directly with the information. Further, this document 15provides specific examples of what exception handling information is used for in 16C and C++. 17 18Itanium ABI Zero-cost Exception Handling 19---------------------------------------- 20 21Exception handling for most programming languages is designed to recover from 22conditions that rarely occur during general use of an application. To that end, 23exception handling should not interfere with the main flow of an application's 24algorithm by performing checkpointing tasks, such as saving the current pc or 25register state. 26 27The Itanium ABI Exception Handling Specification defines a methodology for 28providing outlying data in the form of exception tables without inlining 29speculative exception handling code in the flow of an application's main 30algorithm. Thus, the specification is said to add "zero-cost" to the normal 31execution of an application. 32 33A more complete description of the Itanium ABI exception handling runtime 34support of can be found at `Itanium C++ ABI: Exception Handling 35<http://mentorembedded.github.com/cxx-abi/abi-eh.html>`_. A description of the 36exception frame format can be found at `Exception Frames 37<http://refspecs.linuxfoundation.org/LSB_3.0.0/LSB-Core-generic/LSB-Core-generic/ehframechpt.html>`_, 38with details of the DWARF 4 specification at `DWARF 4 Standard 39<http://dwarfstd.org/Dwarf4Std.php>`_. A description for the C++ exception 40table formats can be found at `Exception Handling Tables 41<http://mentorembedded.github.com/cxx-abi/exceptions.pdf>`_. 42 43Setjmp/Longjmp Exception Handling 44--------------------------------- 45 46Setjmp/Longjmp (SJLJ) based exception handling uses LLVM intrinsics 47`llvm.eh.sjlj.setjmp`_ and `llvm.eh.sjlj.longjmp`_ to handle control flow for 48exception handling. 49 50For each function which does exception processing --- be it ``try``/``catch`` 51blocks or cleanups --- that function registers itself on a global frame 52list. When exceptions are unwinding, the runtime uses this list to identify 53which functions need processing. 54 55Landing pad selection is encoded in the call site entry of the function 56context. The runtime returns to the function via `llvm.eh.sjlj.longjmp`_, where 57a switch table transfers control to the appropriate landing pad based on the 58index stored in the function context. 59 60In contrast to DWARF exception handling, which encodes exception regions and 61frame information in out-of-line tables, SJLJ exception handling builds and 62removes the unwind frame context at runtime. This results in faster exception 63handling at the expense of slower execution when no exceptions are thrown. As 64exceptions are, by their nature, intended for uncommon code paths, DWARF 65exception handling is generally preferred to SJLJ. 66 67Overview 68-------- 69 70When an exception is thrown in LLVM code, the runtime does its best to find a 71handler suited to processing the circumstance. 72 73The runtime first attempts to find an *exception frame* corresponding to the 74function where the exception was thrown. If the programming language supports 75exception handling (e.g. C++), the exception frame contains a reference to an 76exception table describing how to process the exception. If the language does 77not support exception handling (e.g. C), or if the exception needs to be 78forwarded to a prior activation, the exception frame contains information about 79how to unwind the current activation and restore the state of the prior 80activation. This process is repeated until the exception is handled. If the 81exception is not handled and no activations remain, then the application is 82terminated with an appropriate error message. 83 84Because different programming languages have different behaviors when handling 85exceptions, the exception handling ABI provides a mechanism for 86supplying *personalities*. An exception handling personality is defined by 87way of a *personality function* (e.g. ``__gxx_personality_v0`` in C++), 88which receives the context of the exception, an *exception structure* 89containing the exception object type and value, and a reference to the exception 90table for the current function. The personality function for the current 91compile unit is specified in a *common exception frame*. 92 93The organization of an exception table is language dependent. For C++, an 94exception table is organized as a series of code ranges defining what to do if 95an exception occurs in that range. Typically, the information associated with a 96range defines which types of exception objects (using C++ *type info*) that are 97handled in that range, and an associated action that should take place. Actions 98typically pass control to a *landing pad*. 99 100A landing pad corresponds roughly to the code found in the ``catch`` portion of 101a ``try``/``catch`` sequence. When execution resumes at a landing pad, it 102receives an *exception structure* and a *selector value* corresponding to the 103*type* of exception thrown. The selector is then used to determine which *catch* 104should actually process the exception. 105 106LLVM Code Generation 107==================== 108 109From a C++ developer's perspective, exceptions are defined in terms of the 110``throw`` and ``try``/``catch`` statements. In this section we will describe the 111implementation of LLVM exception handling in terms of C++ examples. 112 113Throw 114----- 115 116Languages that support exception handling typically provide a ``throw`` 117operation to initiate the exception process. Internally, a ``throw`` operation 118breaks down into two steps. 119 120#. A request is made to allocate exception space for an exception structure. 121 This structure needs to survive beyond the current activation. This structure 122 will contain the type and value of the object being thrown. 123 124#. A call is made to the runtime to raise the exception, passing the exception 125 structure as an argument. 126 127In C++, the allocation of the exception structure is done by the 128``__cxa_allocate_exception`` runtime function. The exception raising is handled 129by ``__cxa_throw``. The type of the exception is represented using a C++ RTTI 130structure. 131 132Try/Catch 133--------- 134 135A call within the scope of a *try* statement can potentially raise an 136exception. In those circumstances, the LLVM C++ front-end replaces the call with 137an ``invoke`` instruction. Unlike a call, the ``invoke`` has two potential 138continuation points: 139 140#. where to continue when the call succeeds as per normal, and 141 142#. where to continue if the call raises an exception, either by a throw or the 143 unwinding of a throw 144 145The term used to define a the place where an ``invoke`` continues after an 146exception is called a *landing pad*. LLVM landing pads are conceptually 147alternative function entry points where an exception structure reference and a 148type info index are passed in as arguments. The landing pad saves the exception 149structure reference and then proceeds to select the catch block that corresponds 150to the type info of the exception object. 151 152The LLVM :ref:`i_landingpad` is used to convey information about the landing 153pad to the back end. For C++, the ``landingpad`` instruction returns a pointer 154and integer pair corresponding to the pointer to the *exception structure* and 155the *selector value* respectively. 156 157The ``landingpad`` instruction takes a reference to the personality function to 158be used for this ``try``/``catch`` sequence. The remainder of the instruction is 159a list of *cleanup*, *catch*, and *filter* clauses. The exception is tested 160against the clauses sequentially from first to last. The selector value is a 161positive number if the exception matched a type info, a negative number if it 162matched a filter, and zero if it matched a cleanup. If nothing is matched, the 163behavior of the program is `undefined`_. If a type info matched, then the 164selector value is the index of the type info in the exception table, which can 165be obtained using the `llvm.eh.typeid.for`_ intrinsic. 166 167Once the landing pad has the type info selector, the code branches to the code 168for the first catch. The catch then checks the value of the type info selector 169against the index of type info for that catch. Since the type info index is not 170known until all the type infos have been gathered in the backend, the catch code 171must call the `llvm.eh.typeid.for`_ intrinsic to determine the index for a given 172type info. If the catch fails to match the selector then control is passed on to 173the next catch. 174 175Finally, the entry and exit of catch code is bracketed with calls to 176``__cxa_begin_catch`` and ``__cxa_end_catch``. 177 178* ``__cxa_begin_catch`` takes an exception structure reference as an argument 179 and returns the value of the exception object. 180 181* ``__cxa_end_catch`` takes no arguments. This function: 182 183 #. Locates the most recently caught exception and decrements its handler 184 count, 185 186 #. Removes the exception from the *caught* stack if the handler count goes to 187 zero, and 188 189 #. Destroys the exception if the handler count goes to zero and the exception 190 was not re-thrown by throw. 191 192 .. note:: 193 194 a rethrow from within the catch may replace this call with a 195 ``__cxa_rethrow``. 196 197Cleanups 198-------- 199 200A cleanup is extra code which needs to be run as part of unwinding a scope. C++ 201destructors are a typical example, but other languages and language extensions 202provide a variety of different kinds of cleanups. In general, a landing pad may 203need to run arbitrary amounts of cleanup code before actually entering a catch 204block. To indicate the presence of cleanups, a :ref:`i_landingpad` should have 205a *cleanup* clause. Otherwise, the unwinder will not stop at the landing pad if 206there are no catches or filters that require it to. 207 208.. note:: 209 210 Do not allow a new exception to propagate out of the execution of a 211 cleanup. This can corrupt the internal state of the unwinder. Different 212 languages describe different high-level semantics for these situations: for 213 example, C++ requires that the process be terminated, whereas Ada cancels both 214 exceptions and throws a third. 215 216When all cleanups are finished, if the exception is not handled by the current 217function, resume unwinding by calling the `resume 218instruction <LangRef.html#i_resume>`_, passing in the result of the 219``landingpad`` instruction for the original landing pad. 220 221Throw Filters 222------------- 223 224C++ allows the specification of which exception types may be thrown from a 225function. To represent this, a top level landing pad may exist to filter out 226invalid types. To express this in LLVM code the :ref:`i_landingpad` will have a 227filter clause. The clause consists of an array of type infos. 228``landingpad`` will return a negative value 229if the exception does not match any of the type infos. If no match is found then 230a call to ``__cxa_call_unexpected`` should be made, otherwise 231``_Unwind_Resume``. Each of these functions requires a reference to the 232exception structure. Note that the most general form of a ``landingpad`` 233instruction can have any number of catch, cleanup, and filter clauses (though 234having more than one cleanup is pointless). The LLVM C++ front-end can generate 235such ``landingpad`` instructions due to inlining creating nested exception 236handling scopes. 237 238.. _undefined: 239 240Restrictions 241------------ 242 243The unwinder delegates the decision of whether to stop in a call frame to that 244call frame's language-specific personality function. Not all unwinders guarantee 245that they will stop to perform cleanups. For example, the GNU C++ unwinder 246doesn't do so unless the exception is actually caught somewhere further up the 247stack. 248 249In order for inlining to behave correctly, landing pads must be prepared to 250handle selector results that they did not originally advertise. Suppose that a 251function catches exceptions of type ``A``, and it's inlined into a function that 252catches exceptions of type ``B``. The inliner will update the ``landingpad`` 253instruction for the inlined landing pad to include the fact that ``B`` is also 254caught. If that landing pad assumes that it will only be entered to catch an 255``A``, it's in for a rude awakening. Consequently, landing pads must test for 256the selector results they understand and then resume exception propagation with 257the `resume instruction <LangRef.html#i_resume>`_ if none of the conditions 258match. 259 260Exception Handling Intrinsics 261============================= 262 263In addition to the ``landingpad`` and ``resume`` instructions, LLVM uses several 264intrinsic functions (name prefixed with ``llvm.eh``) to provide exception 265handling information at various points in generated code. 266 267.. _llvm.eh.typeid.for: 268 269``llvm.eh.typeid.for`` 270---------------------- 271 272.. code-block:: llvm 273 274 i32 @llvm.eh.typeid.for(i8* %type_info) 275 276 277This intrinsic returns the type info index in the exception table of the current 278function. This value can be used to compare against the result of 279``landingpad`` instruction. The single argument is a reference to a type info. 280 281.. _llvm.eh.sjlj.setjmp: 282 283``llvm.eh.sjlj.setjmp`` 284----------------------- 285 286.. code-block:: llvm 287 288 i32 @llvm.eh.sjlj.setjmp(i8* %setjmp_buf) 289 290For SJLJ based exception handling, this intrinsic forces register saving for the 291current function and stores the address of the following instruction for use as 292a destination address by `llvm.eh.sjlj.longjmp`_. The buffer format and the 293overall functioning of this intrinsic is compatible with the GCC 294``__builtin_setjmp`` implementation allowing code built with the clang and GCC 295to interoperate. 296 297The single parameter is a pointer to a five word buffer in which the calling 298context is saved. The front end places the frame pointer in the first word, and 299the target implementation of this intrinsic should place the destination address 300for a `llvm.eh.sjlj.longjmp`_ in the second word. The following three words are 301available for use in a target-specific manner. 302 303.. _llvm.eh.sjlj.longjmp: 304 305``llvm.eh.sjlj.longjmp`` 306------------------------ 307 308.. code-block:: llvm 309 310 void @llvm.eh.sjlj.longjmp(i8* %setjmp_buf) 311 312For SJLJ based exception handling, the ``llvm.eh.sjlj.longjmp`` intrinsic is 313used to implement ``__builtin_longjmp()``. The single parameter is a pointer to 314a buffer populated by `llvm.eh.sjlj.setjmp`_. The frame pointer and stack 315pointer are restored from the buffer, then control is transferred to the 316destination address. 317 318``llvm.eh.sjlj.lsda`` 319--------------------- 320 321.. code-block:: llvm 322 323 i8* @llvm.eh.sjlj.lsda() 324 325For SJLJ based exception handling, the ``llvm.eh.sjlj.lsda`` intrinsic returns 326the address of the Language Specific Data Area (LSDA) for the current 327function. The SJLJ front-end code stores this address in the exception handling 328function context for use by the runtime. 329 330``llvm.eh.sjlj.callsite`` 331------------------------- 332 333.. code-block:: llvm 334 335 void @llvm.eh.sjlj.callsite(i32 %call_site_num) 336 337For SJLJ based exception handling, the ``llvm.eh.sjlj.callsite`` intrinsic 338identifies the callsite value associated with the following ``invoke`` 339instruction. This is used to ensure that landing pad entries in the LSDA are 340generated in matching order. 341 342Asm Table Formats 343================= 344 345There are two tables that are used by the exception handling runtime to 346determine which actions should be taken when an exception is thrown. 347 348Exception Handling Frame 349------------------------ 350 351An exception handling frame ``eh_frame`` is very similar to the unwind frame 352used by DWARF debug info. The frame contains all the information necessary to 353tear down the current frame and restore the state of the prior frame. There is 354an exception handling frame for each function in a compile unit, plus a common 355exception handling frame that defines information common to all functions in the 356unit. 357 358Exception Tables 359---------------- 360 361An exception table contains information about what actions to take when an 362exception is thrown in a particular part of a function's code. There is one 363exception table per function, except leaf functions and functions that have 364calls only to non-throwing functions. They do not need an exception table. 365