This document contains the release notes for the LLVM compiler
-infrastructure, release 1.5. Here we describe the status of LLVM, including any
+infrastructure, release 1.7. Here we describe the status of LLVM, including any
known problems and major improvements from the previous release. The most
up-to-date version of this document can be found on the LLVM 1.5 web site. If you are
+href="http://llvm.org/releases/">LLVM releases web site. If you are
not reading this on the LLVM web pages, you should probably go there because
this document may be updated after the release.
@@ -48,7 +48,7 @@ list is a good place to send them.
Note that if you are reading this file from CVS or the main LLVM web page,
this document applies to the next release, not the current one. To see
the release notes for the current or previous releases, see the releases page.
This is the sixth public release of the LLVM Compiler Infrastructure.
-
-
At this time, LLVM is known to correctly compile a wide range of C and C++
-programs, including the SPEC CPU95 & 2000 suite. It includes bug fixes for
-those problems found since the 1.4 release and a large number of new features
-and enhancements, described below.
+
This is the eighth public release of the LLVM Compiler Infrastructure. This
+release incorporates a large number of enhancements and new features,
+including vector support (Intel SSE and Altivec), a new GCC4.0-based
+C/C++ front-end, Objective C/C++ support, inline assembly support, and many
+other big features.
+
-This release includes new native code generators for Alpha, IA-64, and SPARC-V8 (32-bit
-SPARC). These code generators are still beta quality, but are progressing
-rapidly.
+
+
LLVM 1.7 includes a brand new llvm-gcc, based on GCC 4.0.1. This version
+of llvm-gcc solves many serious long-standing problems with llvm-gcc, including
+all of those blocked by the llvm-gcc 4 meta
+bug. In addition, llvm-gcc4 implements support for many new features,
+including GCC inline assembly, generic vector support, SSE and Altivec
+intrinsics, and several new GCC attributes. Finally, llvm-gcc4 is
+significantly faster than llvm-gcc3, respects -O options, its -c/-S options
+correspond to GCC's (they emit native code), supports Objective C/C++, and
+it has debugging support well underway.
+
+
If you can use it, llvm-gcc4 offers significant new functionality, and we
+hope that it will replace llvm-gcc3 completely in a future release.
+Unfortunately, it does not currently support C++ exception handling at all, and
+it only works on Apple Mac OS/X machines with X86 or PowerPC processors.
This release includes a new framework
-for building instruction selectors, which has long been the hardest part of
-building a new LLVM target. This framework handles a lot of the mundane (but
-easy to get wrong) details of writing the instruction selector, such as
-generating efficient code for getelementptr instructions, promoting
-small integer types to larger types (e.g. for RISC targets with one size of
-integer registers), expanding 64-bit integer operations for 32-bit hosts, etc.
-Currently, the X86, PowerPC, Alpha, and IA-64 backends use this framework. The
-SPARC backends will be migrated when time permits.
-
+
+
The LLVM IR and llvm-gcc4 front-end now fully support arbitrary GCC inline assembly. The LLVM X86 and PowerPC
+code generators have initial support for it,
+being able to compile basic statements, but are missing some features. Please
+report any inline asm statements that crash the compiler or that are miscompiled
+as bugs.
LLVM 1.5 adds supports for custom and
-target-specific calling conventions. Traditionally, the LLVM code
-generators match the native C calling conventions for a target. This is
-important for compatibility, but is not very flexible. This release allows
-custom calling conventions to be established for functions, and defines three
-target-independent conventions (C call, fast call, and cold call) which may be
-supported by code generators. When possible, the LLVM optimizer promotes C
-functions to use the "fastcc" convention, allowing the use of more efficient
-calling sequences (e.g., parameters are passed in registers in the X86 target).
-
-
Targets may now also define target-specific calling conventions, allowing
-LLVM to fully support calling convention altering options (e.g. GCC's
--mregparm flag) and well-defined target conventions (e.g. stdcall and
-fastcall on X86).
+
LLVM 1.7 includes a new, fully functional, SPARC backend built in the
+target-independent code generator. This SPARC backend includes support for
+SPARC V8 and SPARC V9 subtargets (controlling whether V9 features can be used),
+and targets the 32-bit SPARC ABI.
+
+
The LLVM 1.7 release is the last release that will include the LLVM "SparcV9"
+backend, which was the very first LLVM native code generator. It will
+be removed in LLVM 1.8, being replaced with the new SPARC backend.
The release now includes support for proper tail calls, as
-required to implement languages like Scheme. Tail calls make use of two
-features: custom calling conventions (described above), which allow the code
-generator to emit code for the caller to deallocate its own stack when it
-returns. The second feature is a flag on the call
-instruction, which indicates that the callee does not access the callers
-stack frame (indicating that it is acceptable to deallocate the caller stack
-before invoking the callee). LLVM proper tail calls run on the system stack (as
-do normal calls), supports indirect tail calls, tail calls with arbitrary
-numbers of arguments, tail calls where the callee requires more argument space
-than the caller, etc. The only case not supported are varargs calls, but that
-could be added if desired.
-
-
In order for a front-end to get guaranteed tail call, it must mark functions
-as "fastcc", mark calls with the 'tail' marker, and follow the call with a
-return of the called value (or void). The optimizer and code generator attempt
-to handle more general cases, but the simple case will always work if the code
-generator supports tail calls. Here is a simple example:
+
LLVM now includes significantly extended support for SIMD vectors in its
+core instruction set. It now includes three new instructions for manipulating
+vectors: extractelement,
+insertelement, and
+shufflevector. Further,
+many bugs in vector handling have been fixed, and vectors are now supported by
+the target-independent code generator. For example, if a vector operation is
+not supported by a particular target, it will be correctly broken down and
+executed as scalar operations.
-
- fastcc int %bar(int %X, int(double, int)* %FP) { ; fastcc
- %Y = tail call fastcc int %FP(double 0.0, int %X) ; tail, fastcc
- ret int %Y
- }
-
+
Because llvm-gcc3 does not support GCC generic vectors or vector intrinsics,
+llvm-gcc4 must be used.
+
-
In LLVM 1.5, the X86 code generator is the only target that has been enhanced
-to support proper tail calls (other targets will be enhanced in future).
-Further, because this support was added very close to the release, it is
-disabled by default. Pass -enable-x86-fastcc to llc to enable it (this
-will be enabled by default in the next release). The example above compiles to:
-
-
- bar:
- sub ESP, 8 # Callee uses more space than the caller
- mov ECX, DWORD PTR [ESP + 8] # Get the old return address
- mov DWORD PTR [ESP + 4], 0 # First half of 0.0
- mov DWORD PTR [ESP + 8], 0 # Second half of 0.0
- mov DWORD PTR [ESP], ECX # Put the return address where it belongs
- jmp EDX # Tail call "FP"
-
-With fastcc on X86, the first two integer arguments are passed in EAX/EDX, the
-callee pops its arguments off the stack, and the argument area is always a
-multiple of 8 bytes in size.
+
+
+
The LLVM X86 backend now supports Intel SSE 1, 2, and 3, and now uses scalar
+SSE operations to implement scalar floating point math when the target supports
+SSE1 (for floats) or SSE2 (for doubles). Vector SSE instructions are generated
+by llvm-gcc4 when the generic vector mechanism or specific SSE intrinsics are
+used.
+
+
+
The LLVM PowerPC backend now supports the Altivec instruction set, including
+both GCC -maltivec and -faltivec modes. Altivec instructions are generated
+by llvm-gcc4 when the generic vector mechanism or specific Altivec intrinsics
+are used.
The Loop Unswitching pass (-loop-unswitch) has had several bugs
+ fixed, has several new features, and is enabled by default in llvmgcc3
+ now.
+
The Loop Strength Reduction pass (-loop-reduce) is now enabled for
+ the X86 and Alpha backends.
+
The Instruction Combining pass (-instcombine) now includes a
+ framework and implementation for simplifying code based on whether computed
+ bits are demanded or not.
+
The Scalar Replacement of Aggregates pass (-scalarrepl) can now
+ promote simple unions to registers.
+
The Reassociation pass (-reassociate) can now
+ factor expressions, e.g. turning "A*A+A*B" into "A*(A+B)".
The -globalopt pass now promotes non-address-taken static globals that are
-only accessed in main to SSA registers.
-
-
The new -simplify-libcalls pass improves code generated for well-known
-library calls. The pass optimizes calls to many of the string, memory, and
-standard I/O functions (e.g. replace the calls with simpler/faster calls) when
-possible, given information known statically about the arguments to the call.
-
+
+
LLVM has a new prepass (before register allocation) list scheduler, which
+ supports bottom-up and top-down scheduling, pluggable priority functions and
+ pluggable hazard recognizers. The X86 backend uses this to reduce register
+ pressure and RISC targets schedule based on operation latency.
+
The tblgen-based target description framework introduced in LLVM 1.6 has
+ several new features, useful for targets that can fold loads and stores into
+ operations, and features that make the .td files more expressive.
+
The instruction selector is significantly faster in 1.7 than in 1.6.
+
The X86, Alpha and Itanium backends use new DAG-DAG instruction selectors,
+ making them easier to maintain and generate slightly better code.
+
The X86 backend now supports generation of Scalar SSE code for scalar FP
+ expressions. LLVM provides significantly better performance with Scalar SSE
+ instructions than it does with the Intel floating point stack
+ instructions.
+
The Itanium backend now has a bundling pass, which improves performance
+ by ~10% and reduces code size (previously it unconditionally inserted a stop
+ bit after every instruction).
+
+
-
Loops with trip counts based on array pointer comparisons (e.g. "for (i
-= 0; &A[i] != &A[100]; ++i) ...") are optimized better than before,
-which primarily helps iterator-intensive C++ codes.
The code generator now uses information about takes advantage of commutative
-two-address instructions when performing register allocation.
+
+
+
The Mac OS/X PowerPC and X86 backends now have initial support for
+ Darwin DWARF
+ debugging information, however, debug info generation has been disabled for
+ the 1.7 release in llvmgcc4.
+
LLVM includes the new
+ llvm-config utility, which makes it easier to build and link programs
+ against the LLVM libraries when not using the LLVM makefiles.
+
LLVM now supports first class global ctor/dtor initialization lists, no
+ longer forcing targets to use "__main".
+
LLVM supports assigning globals and functions to a particular section
+ in the result executable using the GCC section attribute.
The LLVM intrinsics used to be overloaded based on type: for example,
+ llvm.ctpop could work with any
+ integer datatype. They are now separated into different intrinsics with
+ suffixes to denote their argument type (e.g. llvm.ctpop.i32)). Old
+ LLVM .ll and .bc files that use these intrinsics will continue to work with
+ new LLVM versions (they are transparently upgraded by the parsers), but will
+ cause a warning to be emitted.
+
The llvm.readport, llvm.writeport, llvm.readio,
+ and llvm.writeio intrinsics have been removed. The first two
+ were ever only supported by the X86 backend, the last two were never
+ correctly supported by any target, and none were accessible through the
+ C front-end. Inline assembly support can now be used to
+ implement these operations.
+
The llvm-db tool had basic support for stepping through code, which
+ used the JIT. This code has been removed, and DWARF emission support added
+ instead. llvm-db still exists in CVS if someone wanted to write a
+ ptrace backend for it.
Intel and AMD machines running Red Hat Linux and FreeBSD (and probably
- other unix-like systems).
+
Intel and AMD machines running Red Hat Linux, Fedora Core and FreeBSD
+ (and probably other unix-like systems).
Sun UltraSPARC workstations running Solaris 8.
Intel and AMD machines running on Win32 with the Cygwin libraries (limited
support is available for native builds with Visual C++).
-
PowerPC-based Mac OS X systems, running 10.2 and above.
+
PowerPC and X86-based Mac OS X systems, running 10.2 and above.
Alpha-based machines running Debian GNU/Linux.
Itanium-based machines running Linux and HP-UX.
@@ -330,7 +316,7 @@ portability patches and reports of successful builds or error messages.
This section contains all known problems with the LLVM system, listed by
component. As new problems are discovered, they will be added to these
sections. If you run into a problem, please check the LLVM bug database and submit a bug if
+href="http://llvm.org/bugs/">LLVM bug database and submit a bug if
there isn't already one.
@@ -349,17 +335,29 @@ useful to some people. In particular, if you would like to work on one of these
components, please contact us on the llvmdev list.
-
The following passes are incomplete or buggy, and may be removed in future
- releases: -cee, -branch-combine, -instloops, -paths, -pre
-
The llvm-db tool is in a very early stage of development, but can
- be used to step through programs and inspect the stack.
-
The "iterative scan" register allocator (enabled with
- -regalloc=iterativescan) is not stable.
-
The SparcV8, Alpha, and IA64 code generators are experimental.
+
The -cee pass is known to be buggy, and may be removed in in a
+ future release.
+
The IA64 code generator is experimental.
+
The Alpha JIT is experimental.
+
"-filetype=asm" (the default) is the only supported value for the
+ -filetype llc option.
In the JIT, dlsym() on a symbol compiled by the JIT will not
work.
-
The JIT does not use mutexes to protect its internal data structures. As
- such, execution of a threaded program could cause these data structures to be
- corrupted.
-
@@ -390,8 +380,15 @@ components, please contact us on the llvmdev list.
Bugs
+
+
+llvm-gcc3 has many significant problems that are fixed by llvm-gcc4. See
+ those blocked on the llvm-gcc4 meta bug.
+Two major ones include:
+
-
C99 Variable sized arrays do not release stack memory when they go out of
+
With llvm-gcc3,
+ C99 variable sized arrays do not release stack memory when they go out of
scope. Thus, the following program may run out of stack space:
for (i = 0; i != 1000000; ++i) {
@@ -400,8 +397,8 @@ components, please contact us on the llvmdev list.
}
The C++ front-end inherits all problems afflicting the C
front-end.
-
IA-64 specific: The C++ front-end does not use IA64 ABI compliant layout of v-tables.
-In particular, it just stores function pointers instead of function
-descriptors in the vtable. This bug prevents mixing C++ code compiled with
-LLVM with C++ objects compiled by other C++ compilers.
-
@@ -592,11 +581,6 @@ LLVM with C++ objects compiled by other C++ compilers.
-
The C++ front-end is based on a pre-release of the GCC 3.4 C++ parser. This
-parser is significantly more standards compliant (and picky) than prior GCC
-versions. For more information, see the C++ section of the GCC 3.4 release notes.
-
Destructors for local objects are not always run when a longjmp is
performed. In particular, destructors for objects in the longjmping
function and in the setjmp receiver function may not be run.
@@ -620,27 +604,42 @@ href="http://gcc.gnu.org/gcc-3.4/changes.html">GCC 3.4 release notes.
The C back-end produces code that violates the ANSI C Type-Based Alias
+Analysis rules. As such, special options may be necessary to compile the code
+(for example, GCC requires the -fno-strict-aliasing option). This
+problem probably cannot be fixed.
The C backend does not correctly implement the llvm.stacksave or
+llvm.stackrestore
+intrinsics. This means that some code compiled by it can run out of stack
+space if they depend on these (e.g. C99 varargs).
The C back-end produces code that violates the ANSI C Type-Based Alias
-Analysis rules. As such, special options may be necessary to compile the code
-(for example, GCC requires the -fno-strict-aliasing option). This
-problem probably cannot be fixed.
On 21164s, some rare FP arithmatic sequences which may trap do not have the appropriate nops inserted to ensure restartability.
+
C++ programs are likely to fail on IA64, as calls to setjmp are
+made where the argument is not 16-byte aligned, as required on IA64. (Strictly
+speaking this is not a bug in the IA64 back-end; it will also be encountered
+when building C++ programs using the C back-end.)
+
+
The C++ front-end does not use IA64
+ABI compliant layout of v-tables. In particular, it just stores function
+pointers instead of function descriptors in the vtable. This bug prevents
+mixing C++ code compiled with LLVM with C++ objects compiled by other C++
+compilers.
-
Vararg functions are not supported.
+
There are a few ABI violations which will lead to problems when mixing LLVM
+output with code built with other compilers, particularly for floating-point
+programs.
-
Due to the vararg problems, C++ exceptions do not work. Small changes are required to the CFE (which break correctness in the exception handler) to compile the exception handling library (and thus the C++ standard library).
+
Defining vararg functions is not supported (but calling them is ok).
The SPARC backend only supports the 32-bit SPARC ABI (-m32), it does not
+ support the 64-bit SPARC ABI (-m64).
+
+
+
-
C++ programs are likely to fail on IA64, as calls to setjmp are
-made where the argument is not 16-byte aligned, as required on IA64. (Strictly
-speaking this is not a bug in the IA64 back-end; it will also be encountered
-when building C++ programs using the C back-end.)
-
There are a few ABI violations which will lead to problems
-when mixing LLVM output with code built with other compilers,
-particularly for C++ and floating-point programs.