Re: Fwd: Solaris build of 7.4 problem with - Mailing list pgsql-bugs

I think we need some code in templates/solaris to identify the compiler
and use -pthread for gcc and -mt for Sun's compiler.

Would you please try the attached patch and let us know how it works.
If it works, we can include it in 7.4.1.

---------------------------------------------------------------------------

Seum-Lim Gan wrote:
> Hi Peter,
>
> You are right. I looked at the man page and it does not
> have the -pthread option but the -mt option.
>
> Perhaps there is a way to automate and detect which
> one it has ?
>
> I find in the man page that there is another option
> -###, which is like -# but does not actually execute:
>
>       The following options are interpreted by cc:
>
>       -#   Show each component as it is invoked (verbose mode).
>
>       -### Show each component as it is invoked, but, unlike the
>            -# option, do not actually execute.
>
> Maybe a better way is to do cc -flags and it will return
> all the flags and then grep for pthread and see if it is a valid
> option. Here is the output of cc -flags:
>
> -#              Verbose mode
> -###            Show compiler commands built by driver, no compilation
> -A<name[(tokens)]>      Preprocessor predicate assertion
> -B<[static|dynamic]>    Specify dynamic or static binding
> -C              Prevent preprocessor from removing comments
> -c              Compile only - produce .o files, suppress linking
> -D<name[=token]>        Associate name with token as if by #define
> -d[y|n]         dynamic [-dy] or static [-dn] option to linker
> -dalign         Expands to -xmemalign=8s
> -E              Compile source through preprocessor only, output to stdout
> -erroff=<t>     Suppress warnings specified by tags t(%none, %all, <tag list>)
> -errtags=<a>    Display messages with tags a(no, yes)
> -errwarn=<t>    Treats warnings specified by tags t(%none, %all, <tag
> list>) as errors
> -fast           Optimize using a selection of options
> -fd             Report old-style function definitions and declarations
> -flags          Show this summary of compiler options
> -fnonstd        Initialize floating-point hardware to non-standard preferences
> -fns[=<yes|no>] Select non-standard floating point mode
> -fround=<r>     Select the IEEE rounding mode in effect at startup
> -fsimple[=<n>]  Select floating-point optimization preferences <n>
> -fsingle        Use single-precision arithmetic (-Xt and -Xs modes only)
> -ftrap=<t>      Select floating-point trapping mode in effect at startup
> -G              Build a dynamic shared library
> -g              Compile for debugging
> -H              Print path name of each file included during compilation
> -h <name>       Assign <name> to generated dynamic shared library
> -I<dir>         Add <dir> to preprocessor #include file search path
> -i              Passed to linker to ignore any LD_LIBRARY_PATH setting
> -keeptmp        Keep temporary files created during compilation
> -KPIC           Compile position independent code with 32-bit addresses
> -Kpic           Compile position independent code
> -L<dir>         Pass to linker to add <dir> to the library search path
> -l<name>        Link with library lib<name>.a or lib<name>.so
> -mc             Remove duplicate strings from .comment section of output files
> -misalign       Expands to -xmemalign=1i
> -misalign2      Expands to -xmemalign=2i
> -mr             Remove all strings from .comment section of output files
> -mr,"string"    Remove all strings and append "string" to .comment section
> -mt             Specify options needed when compiling multi-threaded code
> -native         Optimize for the host system (-xtarget=native)
> -noqueue        Disable queuing of compiler license requests
> -O              Use default optimization level (-xO2)
> -o <outputfile> Set name of output file to <outputfile>
> -P              Compile source through preprocessor only, output to .i  file
> -p              Compile for profiling with prof
> -Q[y|n]         Emit/don't emit identification info to output file
> -qp             Compile for profiling with prof
> -R<dir[:dir]>   Build runtime search path list into executable
> -S              Compile and only generate assembly code (.s)
> -s              Strip symbol table from the executable file
> -U<name>        Delete initial definition of preprocessor symbol <name>
> -V              Report version number of each compilation phase
> -v              Do stricter semantic checking
> -W<c>,<arg>     Pass <arg> to specified component <c> (a,c,l,m,p,0,2,o)
> -w              Suppress compiler warning messages
> -Xa             Compile assuming ANSI C conformance, allow K & R
> extensions (default mode)
> -Xc             Compile assuming strict ANSI C conformance
> -Xs             Compile assuming (pre-ANSI) K & R C style code
> -Xt             Compile assuming K & R conformance, allow ANSI C
> -xa             Collect data for tcov basic block profiling (old format)
> -xalias_level=<a>       Enable optimizations based on the specified alias_level
> -xarch=<a>      Specify target architecture instruction set
> -xautopar       Enable automatic loop parallelization (requires
> WorkShop license)
> -xbuiltin[=<b>] When profitable inline, or substitute intrinisic
> functions for system functions, b={%all,%none}
> -xc99[=<a>]     Enable ISO C99 features, <a>={%all,%none}
> -xcache=<c>     Define cache properties for use by optimizer
> -xCC            Accept C++ style comments
> -xcg89          Expands to -xarch=v7 -xchip=old -xcache=64/32/1
> -xcg92          Expands to -xarch=v8 -xchip=super -xcache=16/32/4:1024/32/1
> -xchar_byte_order=<o>   Specify multi-char byte order <o> (default, high, low)
> -xchip=<c>      Specify the target processor for use by the optimizer
> -xcode=<c>      Generate different code for forming addresses
> -xcrossfile[=<n>]       Enable optimization and inlining across
> source files, n={0|1}
> -xcsi           Allow C source code using non-ISO C compliant locales
> -xdepend        Analyze loops for data dependencies
> -xe             Perform only syntax/semantic checking, no code generation
> -xexplicitpar   Parallelize loops explicitly marked with directives
> (requires WorkShop license)
> -xF             Compile for later mapfile reordering
> -xhelp=<f>      Display on-line help information f(flags, readme)
> -xildoff        Cancel -xildon
> -xildon         Enable use of the incremental linker, ild
> -xinline=[<a>,...,<a>]  Attempt inlining of specified user routines,
> <a>={%auto,func,no%func}
> -xipo[=<n>]     Enable optimization and inlining across source files, n={0|1|2}
> -xlibmieee      Force IEEE 754 return values for math routines in
> exceptional cases
> -xlibmil        Inline selected libm math routines for optimization
> -xlic_lib=sunperf       Link in the Sun supplied performance libraries
> -xlicinfo       Show license server information
> -xloopinfo      Show loops that parallelized (requires WorkShop license)
> -xM             Generate makefile dependencies
> -xM1            Generate makefile dependencies, but exclude /usr/include
> -xmaxopt=[off,1,2,3,4,5]        maximum optimization level allowed on
> #pragma opt
> -xmemalign[=<a><b>]     Controls memory alignment, <a>={1|2|4|8|16}, b={i|s}.
> -xMerge         Merge data segment into text segment
> -xnolib         Do not link with default system libraries
> -xnolibmil      Cancel -xlibmil on command line
> -xO<n>          Generate optimized code (n={1|2|3|4|5})
> -xopenmp        Enable OpenMP language extension
> -xP             Print prototypes for function definitions
> -xparallel      Perform parallel compilation
> -xpg            Compile for profiling with gprof
> -xprefetch[=<p>]        Specify instruction prefetch p
> (auto,no%auto,explicit,no%explicit,latx:<n>.<n>)
> -xprofile=<p>   Collect data for a profile or use a profile to
> optimize <p>={{collect,use}[:<path>],tcov}
> -xreduction     Recognize reduction operations in parallelized loops
> -xregs=<r>      Specify the usage of optional registers
> -xrestrict[=<f>]        Treat pointer valued function parameters as
> restricted f(%none,%all,<function-name list>)
> -xs             Allow debugging without object (.o) files
> -xsafe=mem      Assume that no memory based traps will occur
> -xsb            Compile for use with the WorkShop source browser
> -xsbfast        Generate only WorkShop source browser information, no
> compilation
> -xsfpconst      Represent unsuffixed floating point constants as
> single precision
> -xspace         Do not do optimizations that increase code size
> -xstrconst      Place string literals into read-only data segment
> -xtarget=<t>    Specify target system for optimization
> -xtemp=<dir>    Set directory for temporary files to <dir>
> -xtime          Report the execution time for each compilation phase
> -xtransition    Emit warnings for differences between K&R C and ANSI C
> -xtrigraphs[=<yes|no>]  Enable|disable trigraph translation
> -xunroll=n      Enable unrolling loops n times where possible
> -xvector<=yes|no>       Automatic Generation of calls to vector
> library functions
> -xvpara         Verbose parallelization warnings (requires WorkShop license)
> -Y<c>,<dir>     Specify <dir> for location of component <c> (a,c,l,m,p,0,2,o)
> -YA,<dir>       Change default directory searched for components
> -YI,<dir>       Change default directory searched for include files
> -YP,<dir>       Change default directory for finding libraries files
> -YS,<dir>       Change default directory for startup object files
> -Zll            Create lock_lint database files (.ll)
> -Zlp            Compile for looptool profiling
>
> Thanks.
>
> Gan
>
> At 7:41 am +0100 2003/11/26, Peter Eisentraut wrote:
> >Seum-Lim Gan writes:
> >
> >>  I got the output from the cc verbose (-#).
> >>  Looks like there is a "-hread" towards the end.
> >
> >Apparently, your compiler does not understand the -pthread option.  I've
> >read on the net that the Sun compiler options for thread-safety have
> >changed a few times, but I haven't found a system yet.  If you just want
> >to get going, edit src/template/solaris and replace -pthread by the right
> >option (-mt in your case?).
> >
> >--
> >Peter Eisentraut   peter_e@gmx.net
> >
> >
> >---------------------------(end of broadcast)---------------------------
> >TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org
>
>
> --
> +--------------------------------------------------------+
> | Seum-Lim GAN                 email : slgan@lucent.com  |
> | Lucent Technologies                                    |
> | 2000 N. Naperville Road, 6B-403F  tel : (630)-713-6665 |
> | Naperville, IL 60566, USA.        fax : (630)-713-7272 |
> |       web : http://inuweb.ih.lucent.com/~slgan         |
> +--------------------------------------------------------+
>
> ---------------------------(end of broadcast)---------------------------
> TIP 6: Have you searched our list archives?
>
>                http://archives.postgresql.org
>

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073