479699 - Sun Studio build bustage in LIR.h

Status: RESOLVED DUPLICATE of bug 474757

(Core :: JavaScript Engine, defect)

Description

[jag (Peter Annema)]

Building deps for /export/home/x86slave/xslave/mozilla-1.9.1-solaris-x86/build/js/src/jsapi.cpp using Sun Studio CC
/bin/CC -o jsapi.o -c -xldscope=hidden -DAVMPLUS_IA32 -DAVMPLUS_UNIX -DFEATURE_NANOJIT -DJS_TRACER -DOSTYPE=\"SunOS5\" -DOSARCH=SunOS -DEXPORT_JS_API  -DJS_USE_SAFE_ARENA  -I/export/home/x86slave/xslave/mozilla-1.9.1-solaris-x86/build/js/src -I.  -I./../../dist/include   -I./../../dist/include/js -I/export/home/x86slave/xslave/mozilla-1.9.1-solaris-x86/build/obj-firefox/dist/include/nspr -I/sdk/include -I/export/home/x86slave/xslave/mozilla-1.9.1-solaris-x86/build/js/src    -KPIC   -xlibmieee -xbuiltin=%all -features=tmplife -norunpath -mt  -DNDEBUG -DTRIMMED -xO4   -DMOZILLA_CLIENT -D_MOZILLA_CONFIG_H_ -DSOLARIS=1 -DNSCAP_DISABLE_DEBUG_PTR_TYPES=1 -DD_INO=d_ino -DSTDC_HEADERS=1 -DHAVE_ST_BLKSIZE=1 -DHAVE_SIGINFO_T=1 -DJS_HAVE_STDINT_H=1 -DJS_BYTES_PER_WORD=4 -DJS_BITS_PER_WORD_LOG2=5 -DJS_ALIGN_OF_POINTER=4 -DJS_BYTES_PER_DOUBLE=8 -DHAVE_INT16_T=1 -DHAVE_INT32_T=1 -DHAVE_INT64_T=1 -DHAVE_UINT=1 -DHAVE_UINT_T=1 -DHAVE_UINT16_T=1 -DHAVE_DIRENT_H=1 -DHAVE_SYS_BYTEORDER_H=1 -DHAVE_GETOPT_H=1 -DHAVE_MEMORY_H=1 -DHAVE_UNISTD_H=1 -DHAVE_NL_TYPES_H=1 -DHAVE_MALLOC_H=1 -DHAVE_X11_XKBLIB_H=1 -DHAVE_SYS_STATVFS_H=1 -DHAVE_SYS_STATFS_H=1 -DHAVE_SYS_VFS_H=1 -DHAVE_SYS_MOUNT_H=1 -DHAVE_MMINTRIN_H=1 -DNEW_H=\<new\> -DINCLUDE_MOZILLA_DTRACE=1 -DHAVE_LIBM=1 -DHAVE_LIBDL=1 -DHAVE_DLADDR=1 -DHAVE_LIBSOCKET=1 -D_REENTRANT=1 -DHAVE_RANDOM=1 -DHAVE_STRERROR=1 -DHAVE_LCHOWN=1 -DHAVE_FCHMOD=1 -DHAVE_SNPRINTF=1 -DHAVE_STATVFS=1 -DHAVE_MEMMOVE=1 -DHAVE_RINT=1 -DHAVE_STAT64=1 -DHAVE_LSTAT64=1 -DHAVE_TRUNCATE64=1 -DHAVE_STATVFS64=1 -DHAVE_FLOCKFILE=1 -DHAVE_GETPAGESIZE=1 -DHAVE_LOCALTIME_R=1 -DHAVE_STRTOK_R=1 -DHAVE_WCRTOMB=1 -DHAVE_MBRTOWC=1 -DHAVE_ICONV=1 -DHAVE_ICONV_WITH_CONST_INPUT=1 -DVA_COPY=va_copy -DHAVE_VA_COPY=1 -DHAVE_CPP_EXPLICIT=1 -DHAVE_CPP_TYPENAME=1 -DHAVE_CPP_MODERN_SPECIALIZE_TEMPLATE_SYNTAX=1 -DHAVE_CPP_PARTIAL_SPECIALIZATION=1 -DHAVE_CPP_ACCESS_CHANGING_USING=1 -DHAVE_CPP_AMBIGUITY_RESOLVING_USING=1 -DHAVE_CPP_NAMESPACE_STD=1 -DHAVE_CPP_UNAMBIGUOUS_STD_NOTEQUAL=1 -DHAVE_CPP_NEW_CASTS=1 -DHAVE_CPP_DYNAMIC_CAST_TO_VOID_PTR=1 -DNEED_CPP_UNUSED_IMPLEMENTATIONS=1 -DHAVE_I18N_LC_MESSAGES=1 -DSIZEOF_INT_P=4 -DMOZ_MEMORY_SIZEOF_PTR_2POW=2 -DMOZ_MEMORY=1 -DMOZ_MEMORY_SOLARIS=1 -DCPP_THROW_NEW=throw\(\) -DMOZ_DLL_SUFFIX=\".so\" -DXP_UNIX=1 -DUNIX_ASYNC_DNS=1 -DJS_THREADSAFE=1  /export/home/x86slave/xslave/mozilla-1.9.1-solaris-x86/build/js/src/jsapi.cpp
"/export/home/x86slave/xslave/mozilla-1.9.1-solaris-x86/build/js/src/nanojit/avmplus.h", line 100: Warning: The variable x has not yet been assigned a value.
NEXT ERROR "/export/home/x86slave/xslave/mozilla-1.9.1-solaris-x86/build/js/src/nanojit/LIR.h", line 190: Error: "," expected instead of "disp".
"/export/home/x86slave/xslave/mozilla-1.9.1-solaris-x86/build/js/src/nanojit/LIR.h", line 191: Error: Use ";" to terminate declarations.
"/export/home/x86slave/xslave/mozilla-1.9.1-solaris-x86/build/js/src/nanojit/LIR.h", line 208: Error: "," expected instead of "imm24".
"/export/home/x86slave/xslave/mozilla-1.9.1-solaris-x86/build/js/src/nanojit/LIR.h", line 216: Error: "," expected instead of "imm16".


In bug 468484 a change was made from using |signed int| to |signed int32_t| for __SUNPRO_C and __SUNPRO_CC. This won't compile, either due to int32_t being a typedef (and signed must be followed by a type name, not a typedef), or due to int32_t being something like |#define int32_t signed int|, resulting in |signed signed int|. I guess it's the former, though.

I'm thinking something like this as a fix:

  // Sun Studio requires explicitly declaring signed int bit-field
  #if defined(__SUNPRO_C) || defined(__SUNPRO_CC)
  typedef signed int _signed_int32_t
  #else
  typedef int32_t _signed_int32_t
  #endif

  struct sti_type
  {
    LOpcode          code:8;
    _signed_int32_t  disp:8;
    uint32_t         oprnd_1:8;
    uint32_t         oprnd_2:8;  
  };

Comment 1

[jag (Peter Annema)]

Attachment: Potential Sun Studio build bustage fix

What's up with all the tabs in that file, anyway?

Comment 2

[Andreas Gal :gal]

The indentation is really messy. We will fix it once all the merging with Adobe's code base is done. _signed_int32_t seems a really weird naming. How about bitfield_t. CCing Ed. Please get review from him. I am ok with whatever Ed likes.

Comment 3

[Jeff Walden [:Waldo]]

int32_t is by definition signed in C99; does the nanojit definition differ?  There shouldn't be a need to mark int32_t as signed at all.

Comment 4

[Leon Sha]

This is already fixed in mozilla-central. It just did not get the approval for 1.9.1.

Comment 5

[Ginn Chen]

Jeff,
on my box, int32_t is a typedef of "int", it is signed.
But we have to explicitly say "signed" for signed bitfield, otherwise it is implementation-defined. 
It is C90 6.5.2.1, C99 6.7.2.1.