=head1 DESCRIPTION
-The B<llc> command compiles LLVM bytecode into assembly language for a
+The B<llc> command compiles LLVM source inputs into assembly language for a
specified architecture. The assembly language output can then be passed through
-a native assembler and linker to generate native code.
+a native assembler and linker to generate a native executable.
-The choice of architecture for the output assembly code is determined as
-follows, by attempting to satisfy each of the following rules in turn (first
-one wins):
-
-=over
-
-=item *
-
-If the user has specified an architecture with the -m option, use that
-architecture.
-
-=item *
-
-Examine the input LLVM bytecode file: if it is little endian and has a
-pointer size of 32 bits, select the Intel IA-32 architecture. If it is big
-endian and has a pointer size of 64 bits, select the SparcV9 architecture.
-
-=item *
-
-If B<llc> was compiled on an architecture for which it can generate code, select
-the architecture upon which B<llc> was compiled.
-
-=item *
-
-Exit with an error message telling the user to specify the output
-architecture explicitly.
-
-=back
+The choice of architecture for the output assembly code is automatically
+determined from the input file, unless the B<-march> option is used to override
+the default.
=head1 OPTIONS
-If I<filename> is - or omitted, B<llc> reads LLVM bytecode from standard input.
-Otherwise, it will read LLVM bytecode from I<filename>.
+If I<filename> is - or omitted, B<llc> reads from standard input. Otherwise, it
+will from I<filename>. Inputs can be in either the LLVM assembly language
+format (.ll) or the LLVM bitcode format (.bc).
If the B<-o> option is omitted, then B<llc> will send its output to standard
output if the input is from standard input. If the B<-o> option specifies -,
Other B<llc> options are as follows:
+=head2 End-user Options
+
=over
-=item B<-f>
+=item B<-help>
+
+Print a summary of command line options.
+
+=item B<-O>=I<uint>
+
+Generate code at different optimization levels. These correspond to the I<-O0>,
+I<-O1>, I<-O2>, I<-O3>, and I<-O4> optimization levels used by B<llvm-gcc> and
+B<clang>.
-Overwrite output files. By default, B<llc> will refuse to overwrite
-an output file which already exists.
+=item B<-mtriple>=I<target triple>
+
+Override the target triple specified in the input file with the specified
+string.
=item B<-march>=I<arch>
-Specify the architecture for which to generate assembly. Valid
-architectures are:
+Specify the architecture for which to generate assembly, overriding the target
+encoded in the input file. See the output of B<llc -help> for a list of
+valid architectures. By default this is inferred from the target triple or
+autodetected to the current architecture.
-=over
+=item B<-mcpu>=I<cpuname>
-=item I<x86>
+Specify a specific chip in the current architecture to generate code for.
+By default this is inferred from the target triple and autodetected to
+the current architecture. For a list of available CPUs, use:
+B<llvm-as E<lt> /dev/null | llc -march=xyz -mcpu=help>
-Intel IA-32 (Pentium and above)
+=item B<-mattr>=I<a1,+a2,-a3,...>
-=item I<sparcv9>
+Override or control specific attributes of the target, such as whether SIMD
+operations are enabled or not. The default set of attributes is set by the
+current CPU. For a list of available attributes, use:
+B<llvm-as E<lt> /dev/null | llc -march=xyz -mattr=help>
-64-bit SPARC V9
+=item B<--disable-fp-elim>
-=item I<c>
+Disable frame pointer elimination optimization.
-Emit C code, not assembly
+=item B<--disable-excess-fp-precision>
-=back
+Disable optimizations that may produce excess precision for floating point.
+Note that this option can dramatically slow down code on some systems
+(e.g. X86).
-=item B<-enable-correct-eh-support>
+=item B<--enable-unsafe-fp-math>
-Instruct the B<-lowerinvoke> pass to insert code for correct exception handling
-support. This is expensive and is by default omitted for efficiency.
+Enable optimizations that make unsafe assumptions about IEEE math (e.g. that
+addition is associative) or may not work for all input ranges. These
+optimizations allow the code generator to make use of some instructions which
+would otherwise not be usable (such as fsin on X86).
-=item B<-help>
+=item B<--enable-correct-eh-support>
-Print a summary of command line options.
+Instruct the B<lowerinvoke> pass to insert code for correct exception handling
+support. This is expensive and is by default omitted for efficiency.
-=item B<-stats>
+=item B<--stats>
Print statistics recorded by code-generation passes.
-=item B<-time-passes>
+=item B<--time-passes>
Record the amount of time needed for each pass and print a report to standard
error.
-=back
+=item B<--load>=F<dso_path>
-=head2 Intel IA-32-specific Options
+Dynamically load F<dso_path> (a path to a dynamically shared object) that
+implements an LLVM target. This will permit the target name to be used with the
+B<-march> option so that code can be generated for that target.
-=over
-
-=item B<--disable-fp-elim>
-
-Disable frame pointer elimination optimization.
+=back
-=item B<--disable-pattern-isel>
+=head2 Tuning/Configuration Options
-Use the 'simple' X86 instruction selector (the default).
+=over
=item B<--print-machineinstrs>
-Print generated machine code.
+Print generated machine code between compilation phases (useful for debugging).
=item B<--regalloc>=I<allocator>
=item I<linearscan>
-Linear scan global register allocator (experimental)
+Linear scan global register allocator
+
+=item I<iterativescan>
+
+Iterative scan global register allocator
=back
=back
-=head2 SPARCV9-specific Options
+=head2 Intel IA-32-specific Options
=over
-=item B<--disable-peephole>
-
-Disable peephole optimization pass.
-
-=item B<--disable-sched>
-
-Disable local scheduling pass.
-
-=item B<--disable-strip>
-
-The Sparc backend embeds the LLVM bytecode into the assembly output. This
-option requests that symbol names be retained; by default, they are stripped out.
-
-=item B<--enable-maps>
+=item B<--x86-asm-syntax=att|intel>
-Emit LLVM-to-machine code mapping information into the assembly output.
+Specify whether to emit assembly code in AT&T syntax (the default) or intel
+syntax.
=back
=head1 AUTHORS
-Maintained by the LLVM Team (L<http://llvm.cs.uiuc.edu>).
+Maintained by the LLVM Team (L<http://llvm.org>).
=cut