| C User's Guide |    
|
| C User's Guide |
|
| A P P E N D I X A |
|
C Compiler Options |
This chapter describes the C compiler options. Take note that the C compiler recognizes by default some of the constructs of the 1999 ISO/IEC C standard. Specifically, the supported features are detailed in Supported Features of C99. Use the -xc99=%none command if you want to limit the compiler to the 1990 ISO/IEC C standard.
If you are porting a K&R C program to ISO C, make special note of the section on compatibility flags, Section A.3.59, -X[c|a|t|s]. Using them makes the transition to ISO C easier. Also refer to the discussion on the transition in Chapter 7.
The syntax of the cc command is:
The C compiler accepts a list of C source files and object files contained in the list of files specified by filenames. The resulting executable code is placed in a.out, unless the -o option is used. In this case, the code is placed in the file named by the -o option.
Use the C compiler to compile and link any combination of the following:
After linking, the C compiler places the linked files, now in executable code, into a file named a.out, or into the file specified by the -o option.
See option -YP, dir to change the default directories used for finding libraries. dir is a colon-separated path list. The default library search order for cc is:
cc uses getopt to parse command-line options. Options are treated as a single letter or a single letter followed by an argument. See getopt(3c).
In this section, the compiler options are grouped by function to provide an easy reference. The details are in the sections of the following pages. The following table summarizes the cc compiler options by functionality. Some flags serve more than one purpose and appear more than once.
This section describes the cc options, arranged alphabetically. These descriptions are also available in the man page, cc(1). Use the cc -flags option for a one-line summary of these descriptions.
Options noted as being unique to one or more platforms are accepted without error and ignored on all other platforms. For an explanation of the typographic notations used with the options and arguments, refer to Typographic Conventions.
Turns on verbose mode, showing how command options expand. Shows each component as it is invoked.
Shows each component as it would be invoked, but does not actually execute it. Also shows how command options would expand.
Associates name as a predicate with the specified tokens as if by a #assert preprocessing directive. Preassertions:
These preassertions are not valid in -Xc mode.
Specifies whether bindings of libraries for linking are static or dynamic, indicating whether libraries are non-shared or shared, respectively.
-Bdynamic causes the link editor to look for files named libx.so and then for files named libx.a when given the -lx option.
-Bstatic causes the link editor to look only for files named libx.a. This option may be specified multiple times on the command line as a toggle. This option and its argument are passed to ld(1).
Prevents the C preprocessor from removing comments, except those on the preprocessing directive lines.
Directs cc to suppress linking with ld(1) and to produce a .o file for each source file. You can explicitly name a single object file using the -o option. When the compiler produces object code for each .i or .c input file, it always creates an object (.o) file in the current working directory. If you suppress the linking step, you also suppress the removal of the object files.
Associates name with the specified tokens as if by a #define preprocessing directive. If no =tokens is specified, the token 1 is supplied.
Predefinitions (not valid in -Xc mode):
The following predefinitions are valid in all modes.
The following is predefined in -Xa and -Xt modes only:
The compiler also predefines the object-like macro __PRAGMA_REDEFINE_EXTNAME, to indicate the pragma will be recognized.
-dy specifies dynamic linking, which is the default, in the link editor.
-dn specifies static linking in the link editor.
This option and its arguments are passed to ld(1).
-dalign is equivalent to -xmemalign=8s. See Section A.3.102, -xmemalign=ab.
Runs the source file through the preprocessor only and sends the output to stdout. The preprocessor is built directly into the compiler, except in -Xs mode, where
/usr/ccs/lib/cpp is invoked. Includes the preprocessor line numbering information. See also the -P option.
Use this option if you want to prefix the string "error:" to the beginning of error messages so they are more easily distinguishable from warning messages. The prefix is also attached to warnings that are converted to errors by -errwarn.
If you do not use this option, the compiler sets it to -errfmt=no%error. If you use specify -errfmt, but do not supply a value, the compiler sets it to -errfmt=error.
This command suppresses C compiler warning messages and has no effect on error messages.
t is a comma-separated list that consists of one or more of the following: tag, no%tag, %all, %none. Order is important; for example, %all,no%tag suppresses all warning messages except tag. The following table lists the -erroff values:
|
Suppresses the warning message specified by this tag. You can display the tag for a message by using the -errtags=yes option. |
|
The default is -erroff=%none. Specifying -erroff is equivalent to specifying -erroff=%all.
Only warning messages from the C compiler front-end that display a tag when the -errtags option is used can be suppressed with the -erroff option. You can achieve finer control over error message suppression. See Section 2.8.5, error_messages.
Use this option to control how much detail is in the error message produced by the compiler when it discovers a type mismatch. This option is particularly useful when the compiler discovers a type mismatch that involves a large aggregate.
i can be one of the following:
If you do not use -errshort, the compiler sets the option to -errshort=full. If you specify -errshort, but do not provide a value, the compiler sets the option to -errshort=tags.
This option does not accumulate, it accepts the last value specified on the command line.
Displays the message tag for each warning message of the C compiler front-end that can be suppressed with the -erroff option or made a fatal error with the -errwarn option. Messages from the C compiler driver and other components of the C compilation system do not have error tags, and cannot be suppressed with -erroff and made fatal with -errwarn.
a can be either yes or no. The default is -errtags=no. Specifying -errtags is equivalent to specifying -errtags=yes.
Use -errwarn to cause the C compiler to exit with a failure status for the given warning messages.
t is a comma-separated list that consists of one or more of the following: tag, no%tag, %all, %none. Order is important; for example %all,no%tag causes cc to exit with a fatal status if any warning except tag is issued.
The warning messages generated by the C compiler change from release to release as the compiler error checking improves and features are added. Code that compiles using -errwarn=%all without error may not compile without error in the next release of the compiler.
Only warning messages from the C compiler front-end that display a tag when the -errtags option is used can be specified with the -errwarn option to cause the C compiler to exit with a failure status.
The following table details the -errwarn values:
The default is -errwarn=%none. If you specify -errwarn alone, it is equivalent to -errwarn=%all.
Selects a set of baseline options for optimizing benchmark applications. These optimizations may alter the behavior of programs from that defined by the ISO C and IEEE standards. Modules compiled with -fast must also be linked with -fast.
-fast is a macro option that can be effectively used as a starting point for tuning an executable for maximum runtime performance. -fast is a macro that can change from one release of the compiler to the next and expands to options that are target platform specific. We suggest that you use the -# option to examine the expansion of -fast, and incorporate the appropriate options of -fast into the ongoing process of tunning the executable.
The -fast option is unsuitable for programs intended to run on a different target than the compilation machine. In such cases, follow -fast with the appropriate -xtarget option. For example:
For C modules that depend on exception handling specified by SUID, follow -fast by -xnolibmil:
With -xlibmil, exceptions cannot be noted by setting errno or calling matherr(3m).
The -fast option is unsuitable for programs that require strict conformance to the IEEE 754 Standard.
The following table lists the set of options selected by -fast across platforms.
The optimizations performed by these options may alter the behavior of programs from that defined by the ISO C and IEEE standards. See the description of the specific option for details.
-fast acts like a macro expansion on the command line. Therefore, you can override the optimization level and code generation option aspects by following -fast with the desired optimization level or code generation option. Compiling with the -fast -xO4 pair is like compiling with the -xO2 -xO4 pair. The latter specification takes precedence.
In previous releases, the -fast macro option included -fnonstd; now it includes -fns instead.
-fast also defines the macro __MATHERR_ERRNO_DONTCARE. This macro causes math.h to assert performance-related pragmas such as the following for some math routines prototyped in <math.h>:
If your code relies on the return value of errno in exceptional cases as documented in the matherr(3M) man page, you must turn off the macro by issuing the -U__MATHERR_ERRNO_DONTCARE macro after the -fast option.
You can usually improve performance for most programs with this option.
Do not use this option for programs that depend on IEEE standard exception handling; you can get different numerical results, premature program termination, or unexpected SIGFPE signals.
See -# and -### for details of how you can see the expansion of macro options.
Reports K&R-style function definitions and declarations.
Prints a brief summary of each available compiler option.
Causes nonstandard initialization of floating-point arithmetic hardware. In addition, the -fnonstd option causes hardware traps to be enabled for floating-point overflow, division by zero, and invalid operations exceptions. These are converted into SIGFPE signals; if the program has no SIGFPE handler, it terminates with a memory dump.
By default, IEEE 754 floating-point arithmetic is nonstop, and underflows are gradual. (See Section 2.4, Floating Point, Nonstandard Mode for a further explanation.)
(SPARC) Synonym for -fns -ftrap=common.
(SPARC) Turns on the SPARC nonstandard floating-point mode.
The default is -fns=no, the SPARC standard floating-point mode. -fns is the same as -fns=yes.
Optional use of =yes or =no provides a way of toggling the -fns flag following some other macro flag that includes -fns, such as -fast. This flag enables the nonstandard floating point mode when a program begins execution. By default, the non-standard floating point mode will not be enabled automatically.
On some SPARC systems, the nonstandard floating point mode disables "gradual underflow," causing tiny results to be flushed to zero rather than producing subnormal numbers. It also causes subnormal operands to be replaced silently by zero. On those SPARC systems that do not support gradual underflow and subnormal numbers in hardware, use of this option can significantly improve the performance of some programs.
When nonstandard mode is enabled, floating point arithmetic may produce results that do not conform to the requirements of the IEEE 754 standard. See the Numerical Computation Guide for more information.
This option is effective only on SPARC systems and only if used when compiling the main program. On x86 systems, the option is ignored.
(x86) -fprecision={single, double, extended}
Initializes the rounding-precision mode bits in the Floating-point Control Word to single (24 bits), double (53 bits), or extended (64 bits), respectively. The default floating-point rounding-precision mode is extended.
Note that on Intel, only the precision, not exponent, range is affected by the setting of floating-point rounding precision mode.
Sets the IEEE 754 rounding mode that is established at runtime during the program initialization.
r must be one of: nearest, tozero, negative, positive.
The default is -fround=nearest.
The meanings are the same as those for the ieee_flags subroutine.
When r is tozero, negative, or positive, this flag sets the rounding direction mode to round-to-zero, round-to-negative-infinity, or round-to-positive-infinity respectively when a program begins execution. When r is nearest or the -fround flag is not used, the rounding direction mode is not altered from its initial value (round-to-nearest by default).
This option is effective only if used when compiling the main program.
Allows the optimizer to make simplifying assumptions concerning floating-point arithmetic.
If n is present, it must be 0, 1, or 2. The defaults are:
Permits no simplifying assumptions. Preserve strict IEEE 754 conformance.
Allows conservative simplifications. The resulting code does not strictly conform to IEEE 754, but numeric results of most programs are unchanged.
With -fsimple=1, the optimizer can assume the following:
With -fsimple=1, the optimizer is not allowed to optimize completely without regard to roundoff or exceptions. In particular, a floating-point computation cannot be replaced by one that produces different results with rounding modes held constant at runtime. The -fast macroflag includes -fsimple=1.
Permits aggressive floating point optimizations that may cause many programs to produce different numeric results due to changes in rounding. For example, -fsimple=2 permits the optimizer to replace all computations of x/y in a given loop with x*z, where x/y is guaranteed to be evaluated at least once in the loop, z=1/y, and the values of y and z are known to have constant values during execution of the loop.
Even with -fsimple=2, the optimizer is not permitted to introduce a floating point exception in a program that otherwise produces none.
(-Xt and -Xs modes only) Causes the compiler to evaluate float expressions as single precision rather than double precision. This option has no effect if the compiler is used in either -Xa or -Xc modes, as float expressions are already evaluated as single precision.
(Intel) Causes the compiler to convert the value of a floating-point expression or function to the type on the left-hand side of an assignment, when that expression or function is assigned to a variable, or when the expression is cast to a shorter floating-point type, rather than leaving the value in a register. Due to rounding and truncation, the results may be different from those that are generated from the register value. This is the default mode.
To turn off this option, use the -nofstore option.
Sets the IEEE 754 trapping mode in effect at startup.
t is a comma-separated list that consists of one or more of the following: %all, %none, common, [no%]invalid, [no%]overflow, [no%]underflow, [no%]division, [no%]inexact.
This option sets the IEEE 754 trapping modes that are established at program initialization. Processing is left-to-right. The common exceptions, by definition, are invalid, division by zero, and overflow.
Example: -ftrap=%all,no%inexact means set all traps, except inexact.
The meanings are the same as for the ieee_flags subroutine, except that:
If you compile one routine with -ftrap=t, compile all routines of the program with the same -ftrap=t option; otherwise, you can get unexpected results.
Passes the option to the link editor to produce a shared object rather than a dynamically linked executable. This option is passed to ld(1), and cannot be used with the -dn option.
Produces additional symbol table information for debugging with dbx(1) and the Performance Analyzer analyzer(1).
In link-only invocations of the compiler, the -g option makes the incremental linker (ild) the default instead of the linker (ld). With -g, the compiler's default behavior is to automatically invoke ild in place of ld unless you specify -G or any source file on the command line. Use -xildoff to disable the use of ild. For more information, see the ild(1) man page. See Section A.3.86, -xildoff and Section A.3.87, -xildon.
If you specify -g, and the optimization level is -xO3 or lower, the compiler provides best-effort symbolic information with almost full optimization. Tail-call optimization and back-end inlining are disabled.
If you specify -g and the optimization level is -xO4, the compiler provides best-effort symbolic information with full optimization.
Compile with the -g option to use the full capabilities of the Performance Analyzer. While some performance analysis features do not require -g, you must compile with -g to view annotated source, some function level information, and compiler commentary messages. See the analyzer(1) man page and "Compiling Your Program for Data Collection and Analysis" in Program Performance Analysis Tools for more information.
The commentary messages that are generated with -g describe the optimizations and transformations that the compiler made while compiling your program. Use the er_src(1) command to display the messages, which are interleaved with the source code.
For more information on debugging, see the Debugging a Program With dbx manual.
Prints to standard error, one per line, the path name of each file included during the current compilation. The display is indented so as to show which files are included by other files.
Here, the program sample.c includes the files, stdio.h and math.h; math.h includes the file, floatingpoint.h, which itself includes functions that use sys/ieeefp.h:
% cc -H sample.c /usr/include/stdio.h /usr/include/math.h /usr/include/floatingpoint.h /usr/include/sys/ieeefp.h |
Assigns a name to a shared dynamic library as a way to have different versions of a library. In general, the name after -h should be the same as the file name given after the -o option. The space between -h and name is optional.
The linker assigns the specified name to the library and records the name in the library file as the intrinsic name of the library. If there is no -hname option, then no intrinsic name is recorded in the library file.
When the runtime linker loads the library into an executable file, it copies the intrinsic name from the library file into the executable, into a list of needed shared library files. Every executable has such a list. If there is no intrinsic name of a shared library, then the linker copies the path of the shared library file instead.
-I dir adds dir to the list of directories that are searched for #include files with relative file names, that is, those not beginning with a / (slash). -I values accumulate. See Section 2.13, How to Specify Include Files for a discussion of the search order used to find the include files.
-I- gives you more control over the algorithm that the compiler uses when searching for include files. -I- values do not accumulate. This section first describes the default search algorithms, then it describes the effect of -I- on these algorithms.
For more information on the search pattern of the compiler, see Section 2.13, How to Specify Include Files.
Passes the option to the linker to ignore any LD_LIBRARY_PATH or LD_LIBRARY_PATH_64 setting.
(SPARC) The -KPIC command is equivalent to -xcode=pic32. See also Section A.3.75, -xcode[=v].
(Intel) -KPIC is identical to -Kpic.
(SPARC) The -Kpic command is equivalent to -xcode=pic13. See Section A.3.75, -xcode[=v].
(Intel) Generate position-independent code for use in shared libraries (small model). Permits references to, at most, 2**11 unique external symbols.
Retains temporary files created during compilation instead of deleting them automatically.
Adds dir to the list of directories searched for libraries by ld(1). This option and its arguments are passed to ld(1).
Links with object library libname.so, or libname.a. The order of libraries in the command-line is important, as symbols are resolved from left to right.
This option must follow the sourcefile arguments.
Removes duplicate strings from the .comment section of the object file. When you use the -mc flag, mcs -c is invoked.
(SPARC) -misalign is equivalent to -xmemalign=1i. See Section A.3.102, -xmemalign=ab.
(SPARC) -misalign2 is equivalent to -xmemalign=2i. See Section A.3.102, -xmemalign=ab.
-mr removes all strings from the .comment section. When you use this flag, mcs -d -a is invoked.
-mr,string removes all strings from the .comment section and inserts string in that section of the object file. If string contains embedded blanks, it must be enclosed in quotation marks. A null string results in an empty .comment section. This option is passed as -d -astring to mcs.
Macro option that expands to -D_REENTRANT -lthread. If you are doing your own multithread coding, you must use this option in the compile and link steps. To obtain faster execution, this option requires a multiprocessor system. On a single-processor system, the resulting executable usually runs more slowly with this option.
This option is a synonym for -xtarget=native.
(Intel) Does not convert the value of a floating-point expression or function to the type on the left-hand side of an assignment, when that expression or function is assigned to a variable or is cast to a shorter floating-point type; rather, it leaves the value in a register. See also Section A.3.25, -fstore.
Names the output file filename (as opposed to the default, a.out). filename cannot be the same as sourcefile, since cc does not overwrite the source file. This option and its arguments are passed to ld(1).
Runs the source file through the C preprocessor only. It then puts the output in a file with a .i suffix. Unlike -E, this option does not include preprocessor-type line number information in the output. See also the -E option.
Prepares the object code to collect data for profiling with prof(1). This option invokes a runtime recording mechanism that produces a mon.out file at normal termination.
Emits or does not emit identification information to the output file. -Qy is the default.
If -Qy is used, identification information about each invoked compilation tool is added to the .comment section of output files, which is accessible with mcs. This option can be useful for software administration.
-Qn suppresses this information.
Passes a colon-separated list of directories used to specify library search directories to the runtime linker. If present and not null, it is recorded in the output object file and passed to the runtime linker.
If both LD_RUN_PATH and the -R option are specified, the -R option takes precedence.
Directs cc to produce an assembly source file but not to assemble the program.
Removes all symbolic debugging information from the output object file. This option cannot be specified with -g.
Undefines the preprocessor symbol name. This option removes any initial definition of the preprocessor symbol name created by -D on the same command line including those placed there by the command-line driver.
-U has no effect on any preprocessor directives in source files. You can give multiple -U options on the command line.
If the same name is specified for both -D and -U on the command line, name is undefined, regardless of the order the options appear. In the following example, -U undefines __sun:
Preprocessor statements of the following form in test.c will not take effect because __sun is undefined.
See Section A.3.7, -Dname[=tokens] for a list of predefined symbols.
Directs cc to print the name and version ID of each component as the compiler executes.
Directs the compiler to perform stricter semantic checks and to enable other lint-like checks. For example, the code:
compiles and executes without problem. With -v, it still compiles; however, the compiler displays this warning:
-v does not give all the warnings that lint(1) does. Try running the above example through lint.
Passes the argument arg to a specified component c. See TABLE 1-1 for a list of components.
Each argument must be separated from the preceding only by a comma. All -W arguments are passed after the regular command-line arguments. A comma can be part of an argument by using the escape character \ (backslash) immediately before the comma. All -W arguments are passed after the regular command-line arguments.
For example, -Wa,-o,objfile passes -o and objfile to the assembler, in that order. Also, -Wl,-I,name causes the linking phase to override the default name of the dynamic linker, /usr/lib/ld.so.1.
The order in which the argument(s) are passed to a tool with respect to the other specified command line options may change.
c can be one of the following:
|
cc driver[1] |
|
Suppresses compiler warning messages.
This option overrides the error_messages pragma.
The -X (note uppercase X) options specify varying degrees of compliance to the ISO C standard. The value of -xc99 affects which version of the ISO C standard the -X option applies. The -xc99 option defaults to -xc99=%all which supports a subset of the 1999 ISO/IEC C standard. -xc99=%none supports the 1990 ISO/IEC C standard. See Appendix D for a discussion of supported 1999 ISO/IEC features. See Appendix F for a discussion of differences between ISO/IEC C and K&R C.
(c = conformance) Issues errors and warnings for programs that use non-ISO C constructs. This option is strictly conformant ISO C, without K&R C compatibility extensions. The predefined macro _ _STDC_ _ has a value of 1 with the -Xc option.
This is the default compiler mode. ISO C plus K&R C compatibility extensions, with semantic changes required by ISO C. Where K&R C and ISO C specify different semantics for the same construct, the compiler uses the ISO C interpretation. If the -Xa option is used in conjunction with the -xtransition option, the compiler issues warnings about the different semantics. The predefined macro _ _STDC_ _ has a value of 0 with the -Xa option.
(t = transition) This option uses ISO C plus K&R C compatibility extensions without semantic changes required by ISO C. Where K&R C and ISO C specify different semantics for the same construct, the compiler uses the K&R C interpretation. If you use the -Xt option in conjunction with the -xtransition option, the compiler issues warnings about the different semantics. The predefined macro _ _STDC_ _ has a value of 0 with the -Xt option.
(s = K&R C) Attempts to warn about all language constructs that have differing behavior between ISO C and K&R C. The compiler language includes all features compatible with K&R C. This option invokes cpp for preprocessing. _ _STDC_ _ is not defined in this mode.
(Intel) Optimizes for the 80386 processor.
(Intel) Optimizes for the 80486 processor.
This option is now considered obsolete. Use -xprofile=tcov instead.
(SPARC) The compiler uses the -xalias_level option to determine what assumptions it can make in order to perform optimizations using type-based alias-analysis. This option places the indicated alias level into effect for the translation units being compiled.
If you do not specify the -xalias_level command, the compiler assumes -xalias_level=any. If you specify -xalias_level without a value, the default is -xalias_level=layout.
The -xalias_level option requires optimization level -xO3 or above. If optimization is set lower, a warning is issued and the -xalias_level option is ignored.
Remember that if you issue the -xalias_level option but you fail to adhere to all of the assumptions and restrictions about aliasing described for any of the alias levels, the behavior of your program is undefined.
Replace l with one of the terms in the following table.
Specify instruction set architecture (ISA).
This option limits the code generated by the compiler to the instructions of the specified instruction set architecture. This option does not guarantee use of any target-specific instructions. However, use of this option may affect the portability of a binary program.
If you compile and link in separate steps, make sure you specify the same value for -xarch in both steps.
Architectures that are accepted by -xarch keyword isa are shown in TABLE A-9:
|
generic, native, v7, v8a, v8, v8plus, v8plusa, v8plusb, v9, v9a, v9b |
|
Note that although -xarch can be used alone, it is part of the expansion of the -xtarget option and may be used to override the -xarch value that is set by a specific -xtarget option. For example:
overrides the -xarch=v8 set by -xtarget=ultra2
If you use this option with optimization, the appropriate choice can provide good performance of the executable on the specified architecture. An inappropriate choice results in a binary program that is not executable on the intended target platform.
The following table details the performance of an executable that is compiled with a given -xarch option and then executed by various SPARC processors. The purpose of this table is to help you identify the best -xarch option for your executable given a particular target machine. Start by identifying the range of machines that are of interest to you and then consider the cost of maintaining multiple binaries versus the benefit of extracting the last iota of performance from newer machines.
If you are compiling your executable with the v8plus or v8plusa instruction set, consider compiling with v9 or v9a instead. The v8plus and v8plusa options are provided so that programs can take advantage of some SPARC V9 and UltraSPARC features prior to the availability of Solaris 7 software with its support for 64-bit programs. Programs compiled with the v8plus or v8plusa option are not portable to SPARC V8 or older machines. You can recompile such programs with v9 or v9a, respectively, to take full advantage of all the features of SPARC V9 and UltraSPARC. The V8+ Technical Specification white paper, part number 802-7447-10, is available through your Sun representative and explains the limitations of v8plus and v8plusa.
For any particular choice, the generated executable may run much more slowly on earlier architectures. Also, although quad-precision (REAL*16 and long double) floating-point instructions are available in many of these instruction set architectures, the compiler does not use these instructions in the code it generates.
The following table gives details for each of the -xarch keywords on SPARC platforms.
(SPARC) Turns on automatic parallelization for multiple processors. Does dependence analysis (analyze loops for inter-iteration data dependence) and loop restructuring. If optimization is not at -xO3 or higher, optimization is raised to -xO3 and a warning is emitted.
Avoid -xautopar if you do your own thread management.
To achieve faster execution, this option requires a multiple processor system. On a single-processor system, the resulting binary usually runs slower.
To determine how many processors you have, use the psrinfo command:
% psrinfo 0 on-line since 01/12/95 10:41:54 1 on-line since 01/12/95 10:41:54 3 on-line since 01/12/95 10:41:54 4 on-line since 01/12/95 10:41:54 |
To request a number of processors, set the PARALLEL environment variable. The default is 1.
If you use -xautopar and compile and link in one step, then linking automatically includes the microtasking library and the threads-safe C runtime library. If you use -xautopar and compile and link in separate steps, then you must also link with -xautopar.
Use the -xbuiltin[=(%all|%none)] command when you want to improve the optimization of code that calls standard library functions. Many standard library functions, such as the ones defined in math.h and stdio.h, are commonly used by various programs. This command lets the compiler substitute intrinsic functions or inline system functions where profitable for performance.
If you do not specify -xbuiltin, the default is -xbuiltin=%none, which means no functions from the standard libraries are substituted or inlined. If you specify -xbuiltin, but do not provide any argument, the default is -xbuiltin%all, which means the compiler substitutes intrinsics or inlines standard library functions as it determines the optimization benefit.
If you compile with -fast, then -xbuiltin is set to %all.
|
Note - -xbuiltin only inlines global functions defined in system header files, never static functions defined by the user. |
When you specify -xc99=%none and -xCC, the compiler accepts the C++-style comments. In particular, // can be used to indicate the start of a comment.
The -xc99 flag controls compiler recognition of the implemented features from the C99 standard (ISO/IEC 9899:1999, Programming Language - C).
o can be a comma separated list comprised of the following:
|
[Do not] Enable the 1999 C standard library semantics of routines that appeared in both the 1990 and 1999 C standard. |
|
If you do not specify -xc99, the compiler defaults to -xc99=%all,no%lib. If you specify -xc99 without any values, the option is set to -xc99=%all.
The compiler accepts %all, %none, and no%lib as synonyms for all, none, and no_lib respectively.
|
Note - Though the compiler support-level defaults to the features of C99 listed in Appendix D, the standard headers provided by Solaris software in /usr/include do not yet conform with the 1999 ISO/IEC C standard. If you encounter error messages, try using -xc99=%none to obtain the 1990 ISO/IEC C standard behavior for these headers. |
Defines the cache properties for use by the optimizer. c must be one of the following:
The si/li/ai are defined as follows:
Although this option can be used alone, it is part of the expansion of the -xtarget option; its primary use is to override a value supplied by the -xtarget option.
This option specifies the cache properties that the optimizer can use. It does not guarantee that any particular cache property is used. The following table lists the -xcache values.
Example: -xcache=16/32/4:1024/32/1 specifies the following:
-xcg89 is a macro for: -xarch=v7 -xchip=old -xcache=64/32/1.
-xcg92 is a macro for: -xarch=v8 -xchip=super
The option is provided solely for the purpose of easing the migration of code from systems where the char type is defined as unsigned. Unless you are migrating from such a system, do not use this option. Only code that relies on the sign of a char type needs to be rewritten to explicitly specify signed or unsigned.
You can substitute one of the following for o:
If you do not specify -xchar, the compiler assumes -xchar=s.
If you specify -xchar, but do not specify a value, the compiler assumes -xchar=s.
The -xchar option changes the range of values for the type char only for code compiled with -xchar. This option does not change the range of values for type char in any system routine or header file. In particular, the value of CHAR_MAX and CHAR_MIN, as defined by limits.h, do not change when this option is specified. Therefore, CHAR_MAX and CHAR_MIN no longer represent the range of values encodable in a plain char.
If you use -xchar, be particularly careful when you compare a char against a predefined system macro because the value in the macro may be signed. This is most common for any routine that returns an error code which is accessed through a macro. Error codes are typically negative values so when you compare a char against the value from such a macro, the result is always false. A negative number can never be equal to any value of an unsigned type.
It is strongly recommended that you never use -xchar to compile routines for any interface exported through a library. The Solaris ABI specifies type char as signed, and system libraries behave accordingly. The effect of making char unsigned has not been extensively tested with system libraries. Instead of using this option, modify your code so that it does not depend on whether type char is signed or unsigned. The sign of type char varies among compilers and operating systems.
Produce an integer constant by placing the characters of a multi-character character-constant in the specified byte order. You can substitute one of the following values for o:
Compiling with -xcheck=stkovf adds a runtime check for stack overflow of the main thread in a singly-threaded program as well as slave-thread stacks in a multithreaded program. If a stack overflow is detected, a SIGSEGV is generated. If your application needs to handle a SIGSEGV caused by a stack overflow differently than it handles other address-space violations, see sigaltstack(2).
You can substitute one of the following values for o:
If you do not specify -xcheck, the compiler defaults to -xcheck=%none. If you specify -xcheck without any arguments, the compiler defaults to -xcheck=%all which turns on the runtime check for stack overflow.
The -xcheck option does not accumulate on the command line. The compiler sets the flag in accordance with the last occurrence of the command.
Specifies the target processor for use by the optimizer.
c must be one of the following: generic, old, super, super2, micro, micro2, hyper, hyper2, powerup, ultra, ultra2, ultra2e, ultra2i, ultra3, ultra3cu, 386, 486, pentium, pentium_pro.
Although this option can be used alone, it is part of the expansion of the -xtarget option; its primary use is to override a value supplied by the -xtarget option.
This option specifies timing properties by specifying the target processor.
(SPARC) Specify code address space.
The default is -xcode=abs32 for SPARC V7 and V8, and -xcode=abs64 for SPARC and UltraSPARC V9 (with -xarch=v9|v9a).
When building shared dynamic libraries with -xarch=v9 or v9a or v9b on 64-bit Solaris operating environments, you can specify -xcode=pic13 or -xcode=pic32 but are not required to do so.
There are two nominal performance costs with -xcode=pic13 and -xcode=pic32 on SPARC:
When considering the above costs, remember that the use of -xcode=pic13 and -xcode=pic32 can significantly reduce system memory requirements, due to the effect of library code sharing. Every page of code in a shared library compiled -xcode=pic13 or -xcode=pic32 can be shared by every process that uses the library. If a page of code in a shared library contains even a single non-pic (that is, absolute) memory reference, the page becomes nonsharable, and a copy of the page must be created each time a program using the library is executed.
The easiest way to tell whether or not a .o file has been compiled with -xcode=pic13 or -xcode=pic32 is with the nm command:
A .o file containing position-independent code contains an unresolved external reference to _GLOBAL_OFFSET_TABLE_, as indicated by the letter U.
To determine whether to use -xcode=pic13 or -xcode=pic32, use nm to identify the number of distinct global and static variables used or defined in the library. If the size of _GLOBAL_OFFSET_TABLE_ is under 8,192 bytes, you can use -Kpic. Otherwise, you must use -xcode=pic32.
(SPARC) Enables optimization and inlining across source files. If specified, n can be 0 or 1.
Normally the scope of the compiler's analysis is limited to each separate file on the command line. For example, -xO4's automatic inlining is limited to subprograms defined and referenced within the same source file.
With -xcrossfile, the compiler analyzes all the files named on the command line as if they had been concatenated into a single source file. -xcrossfile is only effective when used with -xO4 or -xO5.
The files produced from this compilation are interdependent due to possible inlining, and must be used as a unit when they are linked into a program. If any one routine is changed and the files recompiled, they must all be recompiled. As a result, using this option affects the construction of make files.
The default is -xcrossfile=0, and no crossfile optimizations are performed. -xcrossfile is equivalent to -xcrossfile=1.
Allows the C compiler to accept source code written in locales that do not conform to the ISO C source character code requirements. These locales include: ja_JP.PCK.
The compiler translation phases required to handle such locales may result in significantly longer compilation times. You should only use this option when you compile source files that contain source characters from one of these locales.
The compiler does not recognize source code written in locales that do not conform to the ISO C source character code requirements unless you specify -xcsi.
The C compiler is migrating the format of debugger information from the stabs format to the dwarf format as specified in 'DWARF Debugging Information Format'. If you maintain software which reads debugging information, you now have the option to transition your tools from the stabs format to the dwarf format. The default setting for this release is -xdebugformat=stabs.
Use this option as a way of accessing the new format for the purpose of porting tools. There is no need to use this option unless you maintain software which reads debugger information, or unless a specific tool tells you that it requires debugger information in one of these formats.
-xdebugformat=stabs generates debugging information using the stabs standard format.
-xdebugformat=dwarf generates debugging information using the dwarf standard format.
If you do not specify -xdebugformat, the compiler assumes -xdebugformat=stabs. It is illegal to specify the option without an argument.
This option affects the format of the data that is recorded with the -g option. Some small amount of debugging information is recorded even without -g, and the format of that information is also controlled with this option. So -xdebugformat has an effect even when -g is not used.
The dbx and Performance Analyzer software understand both stabs and dwarf format so using this option does not have any effect on the functionality of either tool.
(SPARC) Analyzes loops for inter-iteration data dependencies and does loop restructuring.
Loop restructuring includes loop interchange, loop fusion, scalar replacement, and elimination of "dead" array assignments. If optimization is not at -xO3 or higher, the compiler raises optimization to -xO3 and issues a warning.
If you do not specify -xdepend, the default is -xdepend=no which means the compiler does not analyze loops for data dependencies. If you specify -xdepend, but do not specify an argument, the compiler sets the option to -xdepend=yes which means the compiler analyzes loops for data dependencies.
Dependency analysis is also included with -xautopar or -xparallel. The dependency analysis is done at compile time.
Dependency analysis may help on single-processor systems. However, if you try -xdepend on single-processor systems, you should not use either -xautopar or -xexplicitpar. If either of them is on, then the -xdepend optimization is done for multiple-processor systems.
This option is a macro for -###.
Performs only syntax and semantic checking on the source file, but does not produce any object or executable code.
(SPARC) Generates parallelized code based on specification of #pragma MP directives. You do the dependency analysis: analyze and specify loops for inter-iteration data dependencies. The software parallelizes the specified loops. If optimization is not at -xO3 or higher, optimization is raised to -xO3 and a warning is issued. Avoid -xexplicitpar if you do your own thread management.
To get faster code, this option requires a multiprocessor system. On a single-processor system, the generated code usually runs slower.
If you identify a loop for parallelization, and the loop has dependencies, you can get incorrect results, possibly different ones with each run, and with no warnings. Do not apply an explicit parallel pragma to a reduction loop. The explicit parallelization is done, but the reduction aspect of the loop is not done, and the results can be incorrect.
In summary, to parallelize explicitly:
The following is an example of inserting a parallel pragma immediately before the loop:
If you use -xexplicitpar and compile and link in one step, then linking automatically includes the microtasking library and the threads-safe C runtime library. If you use -xexplicitpar and compile and link in separate steps, then you must also link with -xexplicitpar.
Do not specify -xexplicitpar and -xopenmp together.
Enables optimal reordering of functions and variables by the linker.
This option instructs the compiler to place functions and/or data variables into separate section fragments, which enables the linker, using directions in a mapfile specified by the linker's -M option, to reorder these sections to optimize program performance. Generally, this optimization is only effective when page fault time constitutes a significant fraction of program run time.
Reording of variables can help solve the following problems which negatively impact run-time performance:
Reordering variables and functions for optimal performance requires the following operations:
1. Compiling and linking with -xF.
2. Following the instructions in the "Program Performance Analysis" Tools manual regarding how to generate a mapfile for functions or following the instructions in the "Linker and Libraries Guide" regarding how to generate a mapfile for data.
3. Relinking with the new mapfile by using the linker's -M option.
4. Re-executing under the Analyzer to verify improvement.
v can be one or more of the following:
|
[Do not] fragment global data (variables with external linkage) into separate sections. |
|
If you do not specify -xF, the default is -xF=%none. If you specify -xF without any arguments, the default is -xF=%none,func.
See also analyzer(1), debugger(1), ld(1) man pages
Displays on-line help information.
f must be either flags, or readme.
-xhelp=flags displays a summary of the compiler options.
-xhelp=readme displays the README file.
(SPARC) Enables compiler support for hardware counter-based profiling.
When -xhwcprof=[enable|disable] is enabled, the compiler generates information that helps tools match hardware-counter data reference and miss events with associated instructions. Corresponding data-types and structure-members may also be identified in conjunction with symbolic information (produced with -g). This information can be useful in performance analysis and it is not easily identified from profiles based on code addresses, source statements, or routines.
You can compile a specified set of object files with -xhwcprof. However, -xhwcprof is most useful when applied to all object files in the application. This will provide coverage to identify and correlate all memory references distributed in the application's object files.
If you are compiling and linking in separate steps, use -xhwcprof at link time as well. Future extensions to -xhwcprof may require its use at link time.
An instance of -xhwcprof=enable or -xhwcprof=disable overrides all previous instances of -xhwcprof in the same command line.
-xhwcprof is disabled by default. Specifying -xhwcprof without any arguments is the equivalent to -xhwcprof=enable.
-xhwcprof requires that optimization be turned on and that the debug data format be set to DWARF (-xdebugformat=dwarf).
The combination of -xhwcprof and -g increases compiler temporary file storage requirements by more than the sum of the increases due to -xhwcprof and -g specified alone.
The following command compiles example.c and specifies support for hardware counter profiling and symbolic analysis of data types and structure members using DWARF symbols:
For more information on hardware counter-based profiling, see the Program Performance Analysis Tools manual.
Turns off the incremental linker and forces the use of ld. This option is the default if you do not use the -g option, or you do use the -G option, or any source files are present on the command line. Override this default by using the -xildon option.
Turns on the incremental linker and forces the use of ild in incremental mode. This option is the default if you use the -g option, and you do not use the -G option, and there are no source files present on the command line. Override this default by using the -xildoff option.
The format of the list for -xinline is as follows: [{%auto,func_name,no%func_name}[,{%auto,func_name,no%func_name}]...]
-xinline tries to inline only those functions specified in the optional list. The list is either empty, or comprised of a comma-separated list of func_name, no%func_name, or %auto, where func_name is a function name. -xinline only has an effect at -xO3 or higher.
The list of values accumulates from left to right. So for a specification of -xinline=%auto,no%foo the compiler attempts to inline all functions except foo. For a specification of -xinline=%bar,%myfunc,no%bar the compiler only tries to inline myfunc.
When you compile with optimization set at -xO4 or above, the compiler normally tries to inline all references to functions defined in the source file. You can restrict the set of functions the compiler attempts to inline by specifying the -xinline option. If you specify only -xinline=, that is you do not name any functions or %auto, this indicates that none of the routines in the source files are to be inlined. If you specify a list of func_name and no%func_name without specifying %auto, the compiler only attempts to inline those functions specified in the list. If %auto is specified in the list of values with the -xinline option at optimization level set at -xO4 or above, the compiler attempts to inline all functions that are not explicitly excluded by no%func_name.
A function is not inlined if any of the following conditions apply. No warning is issued.
If you specify multiple -xinline options on the command line, they do not accumulate. The last -xinline on the command line specifies what functions the compiler attempts to inline.
(SPARC) Replace a with 0, 1, or 2. -xipo without any arguments is equivalent -xipo=1. -xipo=0 is the default setting and turns off -xipo.
This compiler performs whole-program optimizations by invoking an interprocedural analysis component. Unlike -xcrossfile, -xipo performs optimizations across all object files in the link step, and is not limited to just the source files of the compile command. With -xipo=1, the compiler performs inlining across all source files. With -xipo=2, the compiler performs interprocedural aliasing analysis as well as optimizations of memory allocation and layout to improve cache performance.
The -xipo option generates significantly larger object files due to the additional information needed to perform optimizations across files. However, this additional information does not become part of the final executable binary file. Any increase in the size of the executable program is due to the additional optimizations performed.The object files created in the compilation steps have additional analysis information compiled within them to permit crossfile optimizations to take place at the link step.
-xipo is particularly useful when compiling and linking large multi-file applications. Object files compiled with this flag have analysis information compiled within them that enables interprocedural analysis across source and pre-compiled program files.
However, analysis and optimization is limited to the object files compiled with -xipo, and does not extend to object files in the libraries.
-xipo is multiphased, so you need to specify -xipo for each step if you compile and link in separate steps.
In this example, compilation and linking occur in a single step:
The optimizer performs crossfile inlining across all three source files. This is done in the final link step, so the compilation of the source files need not all take place in a single compilation and could take place over a number of separate compilations, each specifying -xipo.
In this example, compilation and linking occur in separate steps:
cc -xipo -xO4 -c part1.c part2.c cc -xipo -xO4 -c part3.c cc -xipo -xO4 -o prog part1.o part2.o part3.o |
A restriction is that libraries, even if compiled with -xipo, do not participate in crossfile interprocedural analysis, as this example shows:
cc -xipo -xO4 one.c two.c three.c ar -r mylib.a one.o two.o three.o ... cc -xipo -xO4 -o myprog main.c four.c mylib.a |
Here interprocedural optimizations are performed between one.c, two.c and three.c, and between main.c and four.c, but not between main.c or four.c and the routines on mylib.a. (The first compilation may generate warnings about undefined symbols, but the interprocedural optimizations are performed because it is a compile and link step.)
Other important information about -xipo:
Specify the -xjobs option to set how many processes the compiler creates to complete its work. This option can reduce the build time on a multi-cpu machine. Currently, -xjobs works only with the -xipo option. When you specify -xjobs=n, the interprocedural optimizer uses n as the maximum number of code generator instances it can invoke to compile different files.
Generally, a safe value for n is 1.5 multiplied by the number of available processors. Using a value that is many times the number of available processors can degrade performance because of context switching overheads among spawned jobs. Also, using a very high number can exhaust the limits of system resources such as swap space.
You must always specify -xjobs with a value. Otherwise an error diagnostic is issued and compilation aborts.
Multiple instances of -xjobs on the command line override each other until the right-most instance is reached.
The following example compiles more quickly on a system with two processors than the same command without the -xjobs option.
Specify the -xldscope option to change the default linker scoping for the definition of extern symbols. Changing the default can result in faster and safer shared libraries because the implementation is better hidden.
v must be one of the following:
If you do not specify -xldscope, the compiler assumes -xldscope=global. The compiler issues an error if you specify -xldscope without an argument. Multiple instances of this option on the command line override each other until the rightmost instance is reached.
If you intend to allow a client to override a function in a library, you must be sure that the function is not generated inline during the library build. The compiler inlines a function if you specify the function name with -xinline, if you compile at -xO4 or higher in which case inlining can happen automatically, if you use the inline specifier, if you use the inline pragma, or if you are using cross-file optimization.
For example, suppose library ABC has a default allocator function that can be used by library clients, and is also used internally in the library:
processor.
void* ABC_allocator(size_t size) { return malloc(size); }
If you build the library at -xO4 or higher, the compiler inlines calls to ABC_allocator that occur in library components. If a library client wants to replace ABC_allocator with a customized version, the replacement will not occur in library components that called ABC_allocator. The final program will include different versions of the function.
Library functions declared with the __hidden or __symbolic specifiers can be generated inline when building the library. They are not supposed to be overridden by clients. See Linker Scoping Specifiers.
Library functions declared with the __global specifier, should not be declared inline, and should be protected from inlining by use of the -xinline compiler option.
See also -xinline, -xO, -xcrossfile, #pragma inline
Forces IEEE 754 style return values for math routines in exceptional cases. In such cases, no exception message is printed, and you should not rely on errno.
Inlines some library routines for faster execution. This option selects the appropriate assembly language inline templates for the floating-point option and platform for your system.
-xlibmil inlines a function regardless of any specification of the function as part of the -xinline flag.
(SPARC) Links in the Sun-supplied performance libraries.
Returns information about the license file used, the license tokens accepted, the serial number, the RTUs, trial license and the number of days to expiration. This option does not request compilation or check out a license.
Instructs the compiler to perform link-time optimizations on relocatable object files. These optimizations are performed at link time by analyzing the object binary code. The object files are not rewritten but the resulting executable code may differ from the original object codes.
You must use -xlinkopt on at least some of the compilation commands for -xlinkopt to be useful at link time. The optimizer can still perform some limited optimizations on object binaries that are not compiled with -xlinkopt.
-xlinkopt optimizes code coming from static libraries that appear on the compiler command line, but it skips and does not optimize code coming from shared (dynamic) libraries that appear on the command line. You can also use -xlinkopt when you build shared libraries (compiling with -G ).
level sets the level of optimizations performed, and must be 0, 1, or 2. The optimization levels are:
If you compile in separate steps, -xlinkopt must appear on both compile and link steps:
example% cc -c -xlinkopt a.c b.c
example% cc -o myprog -xlinkopt=2 a.o
Note that the level parameter is only used when the compiler is linking. In the example above, the post- optimization level used is 2 even though the object binaries were compiled with an implied level of 1.
Specifying -xlinkopt without a level parameter implies -xlinkopt=1.
This option is most effective when you use it to compile the whole program, and with profile feedback. Profiling reveals the most and least used parts of the code and building directs the optimizer to focus its effort accordingly. This is particularly important with large applications where optimal placement of code performed at link time can reduce instruction cache misses. Typically, this compiles as follows:
example% cc -o progt -xO5 -xprofile=collect:prog file.c example% progt example% cc -o prog -xO5 -xprofile=use:prog -xlinkopt file.c |
For details on using profile feedback, see -xprofile=p.
You cannot use the link-time post-optimizer with the incremental linker, ild. -xlinkopt sets the default linker to be ld. If you enable the incremental linker explicitly with -xildon and also specify -xlinkopt, -xlinkopt is disabled.
Do not use the -zcompreloc linker option when you compile with -xlinkopt.
Note that compiling with this option increases link time slightly. Object file sizes also increase, but the size of the executable remains the same. Compiling with -xlinkopt and -g increases the size of the executable by including debugging information.
(SPARC) Shows which loops are parallelized and which are not. Gives a short reason for not parallelizing a loop. The -xloopinfo option is valid only if -xautopar, or -xparallel, or -xexplicitpar is specified; otherwise, the compiler issues a warning.
To achieve faster execution, this option requires a multiprocessor system. On a single-processor system, the generated code usually runs slower.
Runs only the preprocessor on the named C programs, requesting that it generate makefile dependencies and send the result to the standard output (see make(1) for details about make files and dependencies).
Collects dependencies like -xM, but excludes /usr/include files. For example:
more hello.c #include<stdio.h> main() { (void)printf("hello\n"); } cc -xM hello.c hello.o: hello.c hello.o: /usr/include/stdio.h |
Compiling with -xM1 does not report header file dependencies:
-xM1 is not available under -Xs mode.
Merges data segments into text segments. Data initialized in the object file produced by this compilation is read-only and (unless linked with ld -N) is shared between processes.
where v is one of off, 1, 2, 3, 4, 5. This command limits the level of pragma opt to the level specified. The default value is -xmaxopt=off which causes pragma opt to be ignored. If you specify -xmaxopt without supplying an argument, that is the equivalent of specifying -xmaxopt=5.
Specify maximum assumed memory alignment and behavior of misaligned data accesses. There must be a value for both a (alignment) and b (behavior). a specifies the maximum assumed memory alignment and b specifies the behavior for misaligned memory accesses. The following table lists the alignment and behavior values for -xmemalign
|
Raise signal SIGBUS for alignments less or equal to 4,otherwise interpret access and continue execution. |
|||
For memory accesses where the alignment is determinable at compile time, the compiler generates the appropriate load/store instruction sequence for that alignment of data.
For memory accesses where the alignment cannot be determined at compile time, the compiler must assume an alignment to generate the needed load/store sequence.
The -xmemalign flag allows the user to specify the maximum memory alignment of data to be assumed by the compiler in these indeterminable situations. It also specifies the error behavior to be followed at run time when a misaligned memory access does take place.
Here are the default values for -xmemalign. The following default values only apply when no -xmemalign flag is present:
Here is the default when -xmemalign flag is present but no value is given:
The following table shows how you can use -xmemalign to handle different alignment situations.
Use the -xnativeconnect option when you want to include interface information inside object files and subsequent shared libraries so that the shared library can interface with code written in the Java[tm] programming language (Java code). You must also include -xnativeconnect when you build the shared library with the cc -G command.
When you compile with -xnativeconnect, you are providing the maximum, external, visibility of the native code interfaces. The Native Connector Tool (NCT) enables the automatic generation of Java code and the Java[tm] Native Interface (JNI) code. Using -xnativeconnect with NCT can make functions in C shared libraries callable from Java code. For more information on how to use the NCT, see the online help.
a can be one of the following:
If you do not specify -xnativeconnect, the compiler sets the option to -xnativeconnect=%none. If you specify only -xnativeconnect, the compiler sets the option to -xnativeconnect=interfaces.
-xnativeconnect=interfaces forces the generation of Binary Interface Descriptors (BIDS).
Does not link any libraries by default; that is, no -l options are passed to ld(1). Normally, the cc driver passes -lc to ld.
When you use -xnolib, you have to pass all the -l options yourself. For example:
links libm statically and the other libraries dynamically.
Does not inline math library routines. Use it after the -fast option. For example:
Optimizes the object code; note the uppercase letter O followed by the digit 1, 2, 3, 4, or 5. Generally, the higher the level of optimization, the better the runtime performance. However, higher optimization levels can result in increased compilation time and larger executable files.
In a few cases, -xO2 might perform better than the others, and -xO3 might outperform -xO4. Try compiling with each level to see if you have one of these rare cases.
If the optimizer runs out of memory, it tries to recover by retrying the current procedure at a lower level of optimization and resumes subsequent procedures at the original level specified in the command-line option.
There are five levels that you can use with -xO. The following sections describe how they operate on the SPARC platform and the x86 platform.
|
Does basic local and global optimization. This is induction variable elimination, local and global common subexpression elimination, algebraic simplification, copy propagation, constant propagation, loop-invariant optimization, register allocation, basic block merging, tail recursion elimination, dead code elimination, tail call elimination, and complex expression expansion. The -xO2 level does not assign global, external, or indirect references or definitions to registers. It treats these references and definitions as if they were declared volatile. In general, the -xO2 level results in minimum code size. |
|
|
Performs like -xO2, but also optimizes references or definitions for external variables. Loop unrolling and software pipelining are also performed. This level does not trace the effects of pointer assignments. When compiling either device drivers, or programs that modify external variables from within signal handlers, you may need to use the volatile type qualifier to protect the object from optimization. In general, the -xO3 level results in increased code size. |
|
|
Performs like -xO3, but also automatically inlines functions contained in the same file; this usually improves execution speed. If you want to control which functions are inlined, see -xinline=list. This level traces the effects of pointer assignments, and usually results in increased code size. |
|
|
Attempts to generate the highest level of optimization. Uses optimization algorithms that take more compilation time or that do not have as high a certainty of improving execution time. Optimization at this level is more likely to improve performance if it is done with profile feedback. See -xprofile=p. |
The default is no optimization. However, this is only possible if you do not specify an optimization level. If you specify an optimization level, there is no option for turning optimization off.
If you are trying to avoid setting an optimization level, be sure not to specify any option that implies an optimization level. For example, -fast is a macro option that sets optimization at -xO5. All other options that imply an optimization level give a warning message that optimization has been set. The only way to compile without any optimization is to delete all options from the command line or make file that specify an optimization level.
When you use -xO with the -g option, a limited amount of debugging is available. For more information, see "Debugging Optimized Code" in Chapter 1 of Debugging a Program With dbx.
If you optimize at -xO3 or -xO4 with very large procedures (thousands of lines of code in the same procedure), the optimizer may require a large amount of virtual memory. In such cases, machine performance may degrade.
For more information on debugging, see the Debugging a Program With dbx manual. For more information on optimization, see the Program Performance Analysis Tools manual.
(SPARC) Use the -xopenmp option to enable explicit parallelization with OpenMP directives.The following table lists the values for i:
If you specify -xopenmp, but do not include a value, the compiler assumes -xopenmp=parallel. If you do not specify -xopenmp, the compiler assumes -xopenmp=none.
If you are debugging an OpenMP program with dbx, compile with -g and -xopenmp=noopt so you can breakpoint within parallel regions and display the contents of variables.
Do not specify -xopenmp, with either -xexplicitpar, or -xparallel.
The default for -xopenmp might change in future releases. You can avoid warning messages by explicitly specifying an appropriate optimization.
If you compile and link in separate steps, also specify -xopenmp on the link step. This is especially important when you compile libraries that contain OpenMP directives.
For more information on how to compile a program that is OpenMP compliant, see Section 3.2, Parallelizing for OpenMP.
For information that is specific to the C implementation of OpenMP, see Appendix G.
For a complete summary of the OpenMP Fortran 95, C, and C++ application program interface (API) for building multiprocessing applications, see the OpenMP API User's Guide.
Prints prototypes for all K&R C functions defined in this module.
(SPARC) Sets the preferred page size for the stack and the heap.
The n value must be one of the following: 8K, 64K, 512K, 4M, 32M, 256M, 2G, 16G, or default.
You must specify a valid page size for the Solaris operating environment on the target platform, as returned by getpagesize(3C). If you do not specify a valid pagesize, the request is silently ignored at run-time. The Solaris operating environment offers no guarantee that the page size request will be honored.
You can use pmap(1) or meminfo(2) to determine page size of the target platform.
The -xpagesize option has no effect unless you use it at compile time and at link time.
|
Note - This feature is not available on the Solaris 7 and Solaris 8 operating environments. A program compiled with this option will not link on the Solaris 7 and Solaris 8 operating environments. |
If you specify -xpagesize=default, the Solaris operating environment sets the page size. -xpagesize without an argument is the equivalent to -xpagesize=default.
Compiling with this option has the same effect as setting the LD_PRELOAD environment variable to mpss.so.1 with the equivalent options, or running the Solaris 9 command ppgsz(1) with the equivalent options before running the program. See the Solaris 9 man pages for details.
This option is a macro for -xpagesize_heap and -xpagesize_stack. These two options accept the same arguments as -xpagesize: 8K, 64K, 512K, 4M, 32M, 256M, 2G, 16G, or default. You can set them both with the same value by specifying -xpagesize or you can specify them individually with different values.
(SPARC) Set the page size in memory for the heap. n can be 8K, 64K, 512K, 4M, 32M, 256M, 2G, 16G, or default. You must specify a valid page size for the Solaris operating environment on the target platform, as returned by getpagesize(3C). If you do not specify a valid page size, the request is silently ignored at run-time.
You can use pmap(1) or meminfo(2) to determine page size at the target platform.
If you specify -xpagesize_heap=default, the Solaris operating environment sets the page size. -xpagesize_heap without an argument is the equivalent to -xpagesize_heap=default.
Compiling with this option has the same effect as setting the LD_PRELOAD environment variable to mpss.so.1 with the equivalent options, or running the Solaris 9 command ppgsz(1) with the equivalent options before running the program. See the Solaris 9 man pages for details.
|
Note - This feature is not available on the Solaris 7 and Solaris 8 operating environments. A program compiled with this option will not link on the Solaris 7 and Solaris 8 operating environments. |
(SPARC) Set the page size in memory for the stack. n can be 8K, 64K, 512K, 4M, 32M, 256M, 2G, 16G, or default. You must specify a valid page size for the Solaris operating environment on the target platform, as returned by getpagesize(3C). If you do not specify a valid page size, the request is silently ignored at run-time. You can use pmap(1) or meminfo(2) to determine page size at the target platform.
If you specify -xpagesize_stack=default, the Solaris operating environment sets the page size. -xpagesize_stack without an argument is the equivalent to -xpagesize_stack=default.
Compiling with this option has the same effect as setting the LD_PRELOAD environment variable to mpss.so.1 with the equivalent options, or running the Solaris 9 command ppgsz(1) with the equivalent options before running the program. See the Solaris 9 man pages for details.
|
Note - This feature is not available on the Solaris 7 and Solaris 8 operating environments. A program compiled with this option will not link on the Solaris 7 and Solaris 8 operating environments. |
(SPARC) Parallelizes loops both automatically by the compiler and explicitly specified by the programmer. The -xparallel option is a macro, and is equivalent to specifying all three of -xautopar, -xdepend, and -xexplicitpar. With explicit parallelization of loops, there is a risk of producing incorrect results. If optimization is not at -xO3 or higher, optimization is raised to -xO3 and a warning is issued.
Avoid -xparallel if you do your own thread management. Do not use -xparallel if you are issuing -xopenmp. -xparallel sets -xexplicitpar which should not be used if you specify -xopenmp.
To get faster code, this option requires a multiprocessor system. On a single-processor system, the generated code usually runs slower.
If you compile and link in one step, -xparallel links with the microtasking library and the threads-safe C runtime library. If you compile and link in separate steps, and you compile with -xparallel, then link with -xparallel
This compiler option activates the precompiled-header feature. The precompiled-header feature reduces compile time for applications whose source files share a common set of include files containing a large amount of source code. The compiler collects information about a sequence of header files from one source file, and then uses that information when recompiling that source file, and when compiling other source files that have the same sequence of headers. The information that the compiler collects is stored in a precompiled-header file. You can take advantage of this feature through the -xpch and -xpchstop options in combination with the #pragma hdrstop directive.
When you specify -xpch=v, v can be collect:pch_filename or use:pch_filename. The first time you use -xpch, you must specify the collect mode. The compilation command that specifies -xpch=collect must only specify one source file. In the following example, the -xpch option creates a precompiled-header file called myheader.cpch based on the source file a.c:
cc -xpch=collect:myheader a.c
A valid precompiled-header filename always has the suffix .cpch. When you specify pch_filename, you can add the suffix or let the compiler add it for you. For example, if you specify cc -xpch=collect:foo a.c, the precompiled-header file is called foo.cpch.
When you create a precompiled-header file, pick a source file that contains the common sequence of include files across all the source files with which the precompiled-header file is to be used. The common sequence of include files must be identical across these source files. Remember, only one source filename value is legal in collect mode. For example, cc -xpch=collect:foo bar.c is valid, whereas cc -xpch=collect:foo bar.c foobar.c is invalid because it specifies two source files.
Specify -xpch=use:pch_filename to use a precompiled-header file. You can specify any number of source files with the same sequence of include files as the source file that was used to create the precompiled-header file. For example, your command in use mode could look like this: cc -xpch=use:foo.cpch foo.c bar.c foobar.c.
You should only use an existing precompiled-header file if the following are true. If any of the following is not true, you should recreate the precompiled-header file:
In order to share a precompiled-header file across multiple source files, those source files must share a common set of include files as their initial sequence of tokens. This initial sequence of tokens is known as the viable prefix. The viable prefix must be interpreted consistently across all the source files that use the same precompiled-header file.
The viable prefix begins with the first token of each source file and ends with the either a #pragma hdrstop or the last token of the #include directive for the header file named in the -xpchstop option.
The viable prefix of a source file can only be comprised of comments and any of the following pre-processor directives:
#include
#if/ifdef/ifndef/else/elif/endif
#define/undef
#ident (if identical, passed through as is)
#pragma (if identical)
Any of these may reference macros. The #else, #elif, and #endif directives must match within the viable prefix.
Within the viable prefix of each file that shares a precompiled-header file, each corresponding #define and #undef directive must reference the same symbol (in the case of #define, each one must reference the same value). Their order of appearance within each viable prefix must be the same as well. Each corresponding pragma must also be the same and appear in the same order across all the files sharing a precompiled header.
A header file that is incorporated into a precompiled-header file must not violate the following. The results of compiling a program that violates any of these constraints is undefined.
Make the following additions to your make files in order to incorporate -xpch into your builds. Here's an easy way to use the KEEP_STATE facility of both make and dmake.
CFLAGS=-xpch=use:shared
shared.cpch : bar.c
cc -xpch=collect:shared bar.c -xe
bar.o : bar.c shared.cpch
cc $(CFLAGS) bar.c -c
pong.o : pong.c shared.cpch
cc $(CFLAGS) pong.c -c
Use the -xpchstop=file option to specify the last include file of the viable prefix for the precompiled-header file created by the -xpch option. Using -xpchstop on the command line is equivalent to placing a hdrstop pragma after the first include-directive that references file in each of the source files that you specify with the cc command.
In the following example, the -xpchstop option specifies that the viable prefix for the precompiled-header file ends with the include of projectheader.h. Therefore, privateheader.h is not a part of the viable prefix.
example% cat a.c #include <stdio.h> #include <strings.h> #include "projectheader.h" #include "privateheader.h" . . . example% cc -xpch=collect:foo.cpch a.c -xpchstop=projectheader.h -c |
(Intel) Optimizes for the Pentium processor.
Prepares the object code to collect data for profiling with gprof(1). It invokes a runtime recording mechanism that produces a gmon.out file at normal termination.
(SPARC) Enable prefetch instructions on those architectures that support prefetch, such as UltraSPARC II. (-xarch=v8plus, v9plusa, v9, or v9a)
Explicit prefetching should only be used under special circumstances that are supported by measurements.
val must be one of the following:
|
Adjust the compiler's assumed prefetch-to-load and prefetch-to-store latencies by the specified factor. See Section A.3.117.1, Prefetch Latency Ratio |
|
|
[Disable] Enable automatic generation of prefetch instructions |
|
If you do not specify -xprefetch, the default is -xprefetch=no%auto,explicit. If you specify -xprefetch without a value, that is equivalent to -xprefetch=auto,explicit.
The sun_prefetch.h header file provides the macros that you can use to specify explicit prefetch instructions. The prefetches are approximately at the place in the executable that corresponds to where the macros appear.
The prefetch latency is the hardware delay between the execution of a prefetch instruction and the time the data being prefetched is available in the cache.
The factor must be a positive number of the form n.n.
The compiler assumes a prefetch latency value when determining how far apart to place a prefetch instruction and the load or store instruction that uses the prefetched data. The assumed latency between a prefetch and a load may not be the same as the assumed latency between a prefetch and a store.
The compiler tunes the prefetch mechanism for optimal performance across a wide range of machines and applications. This tuning may not always be optimal. For memory-intensive applications, especially applications intended to run on large multiprocessors, you may be able to obtain better performance by increasing the prefetch latency values. To increase the values, use a factor that is greater than 1 (one). A value between .5 and 2.0 will most likely pro vide the maximum performance.
For applications with datasets that reside entirely within the external cache, you may be able to obtain better performance by decreasing the prefetch latency values. To decrease the values, use a factor that is less than one.
To use the latx:factor suboption, start with a factor value near 1.0 and run performance tests against the application. Then increase or decrease the factor, as appropriate, and run the performance tests again. Continue adjusting the factor and running the performance tests until you achieve optimum performance. When you increase or decrease the factor in small steps, you will see no performance difference for a few steps, then a sudden difference, then it will level off again.
Use the -xprefetch_level option to control the aggressiveness of automatic insertion of prefetch instructions as determined with -xprefetch=auto. l must be 1, 2, or 3. The compiler becomes more aggressive, or in other words, introduces more prefetches with each, higher, level of -xprefetch_level.
The appropriate value for the -xprefetch_level depends on the number of cache misses the application may have. Higher -xprefetch_level values have the potential to improve the performance of applications.
This option is effective only when it is compiled with -xprefetch=auto, with optimization level 3 or greater, and generate code for a platform that supports prefetch (v8plus, v8plusa, v9, v9a, v9b, generic64, native64).
-xprefetch_level=1 enables automatic generation of prefetch instructions. -xprefetch_level=2 enables additional generation beyond level 1 and -xprefetch_level=3 enables additional generation beyond level 2.
The default is -xprefetch_level=1 when you specify -xprefetch=auto.
Use this option to collect and save execution-frequency data so you can then use the data in subsequent runs to improve performance. This option is only valid when you specify optimization at level -xO2 or above.
Compiling with high optimization levels (for example -xO5) is enhanced by providing the compiler with runtime-performance feedback. In order to produce runtime-performance feedback, you must compile with -xprofile=collect, run the executable against a typical data set, and then recompile at the highest optimization level and with -xprofile=use.
Profile collection is safe for multithreaded applications. That is, profiling a program that does its own multitasking ( -mt ) produces accurate results.
p must be collect[:name], use[:name], or tcov.
Collects and saves execution-frequency data for later use by the optimizer with -xprofile=use. The compiler generates code to measure statement execution-frequency.
The name is the name of the program that is being analyzed. This name is optional. If name is not specified, a.out is assumed to be the name of the executable.
You can set the environment variables SUN_PROFDATA and SUN_PROFDATA_DIR to control where a program compiled with -xprofile=collect stores the profile data. If set, the -xprofile=collect data is written to SUN_PROFDATA_DIR/SUN_PROFDATA.
These environment variables similarly control the path and names of the profile data files written by tcov, as described in the tcov(1) man page.
If these environment variables are not set, the profile data is written to name.profile/feedback in the current directory, where name is the name of the executable or the name specified in the -xprofile=collect:name flag. -xprofile does not append .profile to name if name already ends in .profile. If you run the program several times, the executions-frequency data accumulates in the feedback file; that is, output from prior executions is not lost.
If you are compiling and linking in separate steps, make sure that any object files compiled with -xprofile=collect are also linked with -xprofile=collect.
The program is optimized by using the execution-frequency data generated and saved in the feedback files from a previous execution of the program that was compiled with -xprofile=collect.
The name is the name of the program that is being analyzed. This name is optional. If name is not specified, a.out is assumed to be the name of the executable.
Except for the -xprofile option which changes from -xprofile=collect to -xprofile=use, the source files and other compiler options must be exactly the same as those used for the compilation that created the compiled program which in turn generated the feedback file. The same version of the compiler must be used for both the collect build and the use build as well. If compiled with -xprofile=collect:name, the same program name name must appear in the optimizing compilation: -xprofile=use:name.
The association between an object file and its profile data is based on the UNIX pathname of the object file when it is compiled with -xprofile=collect. In some circumstances, the compiler will not associate an object file with its profile data: the object file has no profile data because it was not previously compiled with -xprofile=collect, the object file is not linked in a program with -xprofile=collect, the program has never been executed.
The compiler can also become confused if an object file was previously compiled in a different directory with -xprofile=collect and this object file shares a common basename with other object files compiled with -xprofile=collect but they cannot be uniquely identified by the names of their containing directories. In this case, even if the object file has profile data, the compiler will not be able to find it in the feedback directory when the object file is recompiled with -xprofile=use.
All of these situations cause the compiler to lose the association between an object file and its profile data. Therefore, if an object file has profile data but the compiler is unable to associate it with the object file's pathname when you specify -xprofile=use, use the -xprofile_pathmap option to identify correct directory. See -xprofile_pathmap.
Basic block coverage analysis using "new" style tcov.
The -xprofile=tcov option is the new style of basic block profiling for tcov. It has similar functionality to the -xa option, but correctly collects data for programs that have source code in header files. See Section A.3.62, -xa for information on the old style of profiling, the tcov(1) man page, and Program Performance Analysis Tools for more details.
Code instrumentation is performed similarly to that of the -xa option, but .d files are no longer generated. Instead, a single file is generated, the name of which is based on the final executable. For example, if the program is run out of /foo/bar/myprog.profile, the data file is stored in /foo/bar/myprog.profile/myprog.tcovd.
The -xprofile=tcov and the -xa options are compatible in a single executable. That is, you can link a program that contains some files that have been compiled with -xprofile=tcov, and others with -xa. You cannot compile a single file with both options.
When running tcov, you must pass it the -x option to make it use the new style of data. If not, tcov uses the old .d files, if any, by default for data, and produces unexpected output.
Unlike the -xa option, the TCOVDIR environment variable has no effect at compile-time. However, its value is used at program runtime. See tcov(1) and Program Performance Analysis Tools for more details.
|
Note - tcov's code coverage report can be unreliable if there is inlining of routines due to -xO4 or -xinline. |
When you use -xprofile=collect to compile a program for profile collection and -xprofile=use to compile a program for profile feedback, the source files and compiler options other than -xprofile=collect and -xprofile=use must be identical in both compilations.
The profile feedback directory names specified by the -xprofile=use:name option are accumulated from all instances of the option in a single invocation of the compiler. For example, assume that profile directories a.profile, b.profile and c.profile are created as a result of executing profiled binaries named a, b, and c respectively.
All three profile directories are used. Any valid profile feedback data pertaining to a particular object file is accumulated from the specified feedback directories when the object file is compiled.
If both -xprofile=collect and -xprofile=use are specified in the same command line, the rightmost -xprofile option in the command line is applied as follows:
-xhwcprof, -xprofile_ircache, -xprofile_pathmap
Use -xprofile_ircache[=path] with -xprofile=collect|use to improve compilation time during the use phase by reusing compilation data saved from the collect phase.
With large programs, compilation time in the use phase can improve significantly because the intermediate data is saved. Note that the saved data could increase disk space requirements considerably.
When you use -xprofile_ircache[=path], path overrides the location where the cached files are saved. By default, these files are saved in the same directory as the object file. Specifying a path is useful when the collect and use phases happen in two different directories. Here's a typical sequence of commands:
example% cc -xO5 -xprofile=collect -xprofile_ircache t1.c t2.c example% a.out // run collects feedback data example% cc -xO5 -xprofile=use -xprofile_ircache t1.c t2.c |
Use the -xprofile_pathmap=collect_prefix:use_prefix option when you are also specifying the -xprofile=use command. Use -xprofile_pathmap when both of the following are true and the compiler is unable to find profile data for an object file that is compiled with -xprofile=use.
The collect-prefix is the prefix of the UNIX pathname of a directory tree in which object files were compiled using -xprofile=collect.
The use-prefix is the prefix of the UNIX pathname of a directory tree in which object files are to be compiled using -xprofile=use.
If you specify multiple instances of -xprofile_pathmap, the compiler processes them in the order of their occurrence. Each use-prefix specified by an instance of -xprofile_pathmap is compared with the object file pathname until either a matching use-prefix is identified or the last specified use-prefix is found not to match the object file pathname.
(SPARC) Turns on reduction recognition during automatic parallelization. -xreduction must be specified with -xautopar, or -xparallel.
When reduction recognition is enabled, the compiler parallelizes reductions such as dot products, maximum and minimum finding. These reductions yield different roundoffs than obtained by unparallelized code.
(SPARC) Specifies the usage of registers for the generated code.
r is a comma-separated list that consists of one or more of the following: [no%]appl, [no%]float.
|
[Does not] Allow the compiler to generate code using the application registers as scratch registers. The application registers are: g2, g3, g4 (v8a, v8, v8plus, v8plusa, v8plusb) It is strongly recommended that all system software and libraries be compiled using -xreg=no%appl. System software (including shared libraries) must preserve these registers' values for the application. Their use is intended to be controlled by the compilation system and must be consistent throughout the application. For more information on SPARC instruction sets, see -xarch=isa. In the SPARC ABI, these registers are described as application registers. Using these registers can increase performance because fewer load and store instructions are needed. However, such use can conflict with some old library programs written in assembly code. |
|
|
[Does not] Allow the compiler to generate code by using the floating-point registers as scratch registers for integer values. Use of floating-point values may use these registers regardless of this option. If -xregs=no%float, a source program cannot contain any floating-point code. |
The default is -xregs=appl,float.
It is strongly recommended that you compile code intended for shared libraries that will link with applications, with -xregs=no%appl,float. At the very least, the shared library should explicitly document how it uses the application registers so that applications linking with those libraries know how to cope with the issue.
For example, an application using the registers in some global sense (such as using a register to point to some critical data structure) would need to know exactly how a library with code compiled without -xregs=no%appl is using the application registers in order to safely link with that library.
(SPARC) Treats pointer-valued function parameters as restricted pointers . f is %all, %none, or a comma-separated list of one or more function names: {%all|%none|fn[,fn...]}.
If a function list is specified with this option, pointer parameters in the specified functions are treated as restricted; if -xrestrict=%all is specified, all pointer parameters in the entire C file are treated as restricted. Refer to Section 3.8.2, Restricted Pointers, for more information.
This command-line option can be used on its own, but it is best used with optimization. For example, the command:
treats all pointer parameters in the file prog.c as restricted pointers. The command:
treats all pointer parameters in the function agc in the file prog.c as restricted pointers.
The default is %none; specifying -xrestrict is equivalent to specifying -xrestrict=%all.
Allows debugging by dbx without object files.
This option causes all the debug information to be copied into the executable. This has little impact on dbx performance or the run-time performance of the program, but it does take more disk space.
(SPARC) Allows the compiler to assume no memory-based traps occur.
This option grants permission to use the speculative load instruction on V9 machines. It is only effective when you specify -xO5 optimization and -xarch=v8plus|v8plusa|v9|v9a.
Use this option to generates extra symbol table information for the Source Browser. This option is not valid with the -Xs mode of the compiler.
If you are compiling and linking in separate steps, be sure to specify -xsb in both the compile step and the link step otherwise you will see error messages from the linker.
If you do not use -xsb to link objects that were compiled with -xsb, you limit the source browser data to those references used by the executable that was created with the link step. Also, if you do not specify -xsb in separate compile and link steps, some symbol references in the source browser database may be lost.
By including -xsb in both the compile step and the separate link step, you ensure that all symbol references in both objects are visible to the source browser when the objects are compiled in different ways in the same directory and linked with different executables.
Creates the database for the Source Browser. Does not compile source into an object file. This option is not valid with the -Xs mode of the compiler.
Represents unsuffixed floating-point constants as single precision, instead of the default mode of double precision. Not valid with -Xc.
Does no optimizations or parallelization of loops that increase code size.
Example: The compiler will not unroll loops or parallelize loops if it increases code size.
Inserts string literals into the read-only data section of the text segment instead of the default data segment. Duplicate strings will be eliminated and the remaining copy shared amongst references in the code.
Specifies the target system for instruction set and optimization.
The value of t must be one of the following: native, generic, system-name (SPARC, x86).
The -xtarget option is a macro that permits a quick and easy specification of the -xarch, -xchip, and -xcache combinations that occur on real systems. The only meaning of -xtarget is in its expansion.
|
Gets the best performance on the host system. The compiler generates code for the best performance on the host system. It determines the available architecture, chip, and cache properties of the machine on which the compiler is running. |
|
|
Gets the best performance for generic architecture, chip, and cache. The compiler expands -xtarget=generic to: |
|
|
Gets the best performance for the specified system. You select a system name from TABLE A-32 that lists the mnemonic encodings of the actual system name and numbers. |
The performance of some programs may benefit by providing the compiler with an accurate description of the target computer hardware. When program performance is critical, the proper specification of the target hardware could be very important. This is especially true when running on the newer SPARC processors. However, for most programs and older SPARC processors, the performance gain is negligible and a generic specification is sufficient.
Each specific value for -xtarget expands into a specific set of values for the -xarch, -xchip, and -xcache options. Use the fpversion(1) command to determine the expansion of -xtarget=native on a running system. See TABLE A-32 for the values.
For example, -xtarget=sun4/15 is equivalent to: -xarch=v8a -xchip=micro -xcache=2/16/1.
|
Note - The expansion of -xtarget for a specific host platform might not result in the same -xarch, -xchip, or -xcache settings as -xtarget=native when compiling on that platform. |
The following table lists the -xtarget values for the Intel Architecture:
Sets the directory for temporary files used by cc to dir. No space is allowed within this option string. Without this option, temporary files go into /tmp. -xtemp has precedence over the TMPDIR environment variable.
(SPARC) Specify -xthreadvar to control the implementation of thread local variables. Use this option in conjunction with the __thread declaration specifier to take advantage of the compiler's thread-local storage facility. After you declare the thread variables with the __thread specifier, specify -xthreadvar to enable the use of thread-local storage with position dependent code (non-PIC code) in dynamic (shared) libraries. For more information on how to use __thread, see Thread Local Storage Specifier.
If you do not specify -xthreadvar, the default used by the compiler depends upon whether or not position-independent code is enabled. If position-independent code is enabled, the option is set to -xthreadvar=dynamic. If position-independent code is disabled, the option is set to -xthreadvar=no%dynamic.
If you specify -xthreadvar, but do not specify any values, the option is set to -xthreadvar=dynamic.
If there is non-position-independent code within a dynamic library, you must specify -xthreadvar.
The linker cannot support the thread-variable equivalent of non-PIC code in dynamic libraries. Non-PIC thread variables are significantly faster, and hence should be the default for executables.
Using thread variables on different versions of Solaris software requires different options on the command line.
See Also: -xcode, -KPIC, -Kpic
Reports the time and resources used by each compilation component.
Issues warnings for the differences between K&R C and Sun ISO C.
The -xtransition option issues warnings in conjunction with the -Xa and -Xt options. You can eliminate all warning messages about differing behavior through appropriate coding. The following warnings no longer appear unless you issue the -xtransition option:
The -xtrigraphs option determines whether the compiler recognizes trigraph sequences as defined by the ISO C standard.
By default, the compiler assumes -xtrigraphs=yes and recognizes all trigraph sequences throughout the compilation unit.
If your source code has a literal string containing question marks (?) that the compiler is interpreting as a trigraph sequence, you can use the -xtrigraph=no suboption to turn off the recognition of trigraph sequences. The -xtrigraphs=no option turns off recognition of all trigraphs throughout the entire compilation unit.
Consider the following example source file named trigraphs_demo.c.
Here is the output if you compile this code with -xtrigraphs=yes.
Here is the output if you compile this code with -xtrigraphs=no.
Suggests to the optimizer to unroll loops n times. n is a positive integer. When n is 1, it is a command, and the compiler unrolls no loops. When n is greater than 1, the -xunroll=n merely suggests to the compiler that it unroll loops n times.
Use this option if your code contains a string literal that you want the compiler to convert to UTF-16 strings in the object file. This option enables recognition of the U"ASCII_string" string literals as an array of type unsigned short.
The default is -xustr=no. -xustr=ascii_utf16_ushort enables compiler recognition of U"ASCII_string" string literals.
Enable automatic generation of calls to the vector library functions.
-xvector=yes permits the compiler to transform math library calls within loops into single calls to the equivalent vector math routines when such transformations are possible. Such transformations could result in a performance improvement for loops with large loop counts.
If you do not specify -xvector, the default is -xvector=no. -xvector=no undoes a previously specified -xvector=yes. If you specify -xvector but do not supply a value, the default is -xvector=yes.
If you use -xvector on the command line without previously specifying -xdepend, -xvector triggers -xdepend. The -xvector option also raises the optimization level to -x03 if optimization is not specified or optimization is set lower than -x03.
The compiler includes the libmvec libraries in the load step. If you compile and link with separate commands, be sure to use -xvector in the linking cc command.
(SPARC) Use the -xvis=[yes|no] command when you are using the assembly-language templates defined in the VIS[tm] instruction-set Software Developers Kit (VSDK). The default is -xvis=no. Specifying -xvis is equivalent to specifying -xvis=yes.
The VIS instruction set is an extension to the SPARC v9 instruction set. Even though the UltraSPARC processors are 64-bit, there are many cases, especially in multimedia applications, when the data are limited to eight or 16 bits in size. The VIS instructions can process four 16-bit data with one instruction so they greatly improve the performance of applications that handle new media such as imaging, linear algebra, signal processing, audio, video and networking.
For more information on the VSDK, see http://www.sun.com/processors/vis/.
(SPARC) Warns about loops that have #pragma MP directives specified when the loop may not be properly specified for parallelization. For example, when the optimizer detects data dependencies between loop iterations, it issues a warning.
Use -xvpara with the -xexplicitpar option or the -xparallel option and the #pragma MP. See Section 3.8.3, Explicit Parallelization and Pragmas for more information.
Specifies a new directory dir for the location of component c. c can consist of any of the characters representing components that are listed under the -W option.
If the location of a component is specified, then the new path name for the tool is dir/tool. If more than one -Y option is applied to any one item, then the last occurrence holds.
Changes the default directory searched for components.
Changes the default directory searched for include files.
Changes the default directory for finding library files.
Changes the default directory for startup object files.
(SPARC) Creates the program database for lock_lint, but does not generate executable code. Refer to the lock_lint(1) man page for more details.
cc recognizes -a, -e, -r, -t, -u, and -z and passes these options and their arguments to ld. cc passes any unrecognized options to ld with a warning.
| C User's Guide | 817-5064-10 |
|
Copyright © 2004, Sun Microsystems, Inc. All rights reserved.