[LLVM][MIPS] Fix different definition of off_t in LLDB and LLVM

[nitesh.jain]

Summary

Fix is to define _LARGEFILE_SOURCE and _FILE_OFFSET_BITS=64. This fix will cause llvm::sys::fs::file_size() to return correct object size.

[krytarowski]

Is this just GNU specific? BSD moved to 64-bit off_t on 32-bit platforms 20+ years ago.

It's perhaps no-op, but it might be noted in the commit message what platforms are supposed to be improved.

[beanz]

Can you please move this check into HandleLLVMOptions.cmake? By putting it into a module that is vended as part of LLVM's packaging then LLVM subprojects can have consistent settings when building out-of-tree.

This would enable @zturner's request to remove the duplicated code from LLDB.

[nitesh.jain]

Update diff as per suggestion.

[nitesh.jain]

In D32125#728166, @krytarowski wrote:

Is this just GNU specific? BSD moved to 64-bit off_t on 32-bit platforms 20+ years ago.

It's perhaps no-op, but it might be noted in the commit message what platforms are supposed to be improved.

Yes.

[beanz]

This looks good to me.

[emaste]

In D32125#728166, @krytarowski wrote:

Is this just GNU specific? BSD moved to 64-bit off_t on 32-bit platforms 20+ years ago.

It's perhaps no-op, but it might be noted in the commit message what platforms are supposed to be improved.

I'd say it's better to put a comment in the source where it's set.

[nitesh.jain]

In D32125#734500, @emaste wrote:

In D32125#728166, @krytarowski wrote:

Is this just GNU specific? BSD moved to 64-bit off_t on 32-bit platforms 20+ years ago.

It's perhaps no-op, but it might be noted in the commit message what platforms are supposed to be improved.

I'd say it's better to put a comment in the source where it's set.

Added comment in the source.

[sdardis]

Hi Nitesh,

this commit broke clang-cmake-mips. Can you investigate?

http://lab.llvm.org:8011/builders/clang-cmake-mips/builds/3189

Thanks,
Simon

[nitesh.jain]

In D32125#736543, @sdardis wrote:

Hi Nitesh,

this commit broke clang-cmake-mips. Can you investigate?

http://lab.llvm.org:8011/builders/clang-cmake-mips/builds/3189

Thanks,
Simon

Hi Simon,

The assertion has been fixed in revision rL301307.

Thanks
Nitesh

llvm/trunk/cmake/modules/HandleLLVMOptions.cmake

if( LLVM_BUILD_32_BITS )
     set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -m32")
     set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -m32")
     set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -m32")
     set(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} -m32")
     set(CMAKE_MODULE_LINKER_FLAGS "${CMAKE_MODULE_LINKER_FLAGS} -m32")
   endif( LLVM_BUILD_32_BITS )
 endif( CMAKE_SIZEOF_VOID_P EQUAL 8 AND NOT WIN32 )
 
+# If building on a GNU specific 32-bit system, make sure off_t is 64 bits
+# so that off_t can stored offset > 2GB
+if( CMAKE_SIZEOF_VOID_P EQUAL 4 )
+  add_definitions( -D_LARGEFILE_SOURCE )
+  add_definitions( -D_FILE_OFFSET_BITS=64 )
+endif()
+
 if( XCODE )
   # For Xcode enable several build settings that correspond to
   # many warnings that are on by default in Clang but are
   # not enabled for historical reasons.  For versions of Xcode
   # that do not support these options they will simply
   # be ignored.
   set(CMAKE_XCODE_ATTRIBUTE_GCC_WARN_ABOUT_RETURN_TYPE "YES")
   set(CMAKE_XCODE_ATTRIBUTE_GCC_WARN_ABOUT_MISSING_NEWLINE "YES")