printf format specifiers for different platforms

[rubenvb]

Bugzilla Link	13565
Resolution	FIXED
Resolved on	Jul 07, 2017 12:09
Version	trunk
OS	Windows NT
Blocks	llvm/llvm-bugzilla-archive#13707
CC	@compnerd,@erichkeane,@zmodem,@tritao,@belkadan,@tkremenek

Extended Description

Running this code through MinGW-w64 based Clang:
#define __STDC_FORMAT_MACROS
#include <inttypes.h>
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include
int main(int argc,char **argv)
{
uint64_t val=1234567890;
printf("%" PRId64"\n",val);
exit(0);
}
gives following warning:
M:\Development\x64\test>clang++ -std=c++11 test.cpp -c -Wall
test.cpp:11:15: warning: invalid conversion specifier 'I'
[-Wformat-invalid-specifier]
printf("%" PRId64"\n",val);
~~~^
M:/Development/mingw64/bin/../lib/clang/3.1/../../../x86_64-w64-mingw32/include\inttypes.h:42:17: note:
expanded from macro 'PRId64'
#define PRId64 "I64d"
^
1 warning generated.

The PRId64 is defined to a valid msvcrt (Windows C runtime) format specifier, all of which are documented here: http://msdn.microsoft.com/en-us/library/56e442dc.aspx

I dug through the code, and ended up in these files:
http://llvm.org/svn/llvm-project/cfe/trunk/lib/Analysis/PrintfFormatString.cpp
http://llvm.org/svn/llvm-project/cfe/trunk/lib/Analysis/ScanfFormatString.cpp

Which I believe need a new "case" in the "ParsePrintfSpecifier" function, which would look somewhat like this:

case 'I':
    if(triplet.third == "win32" || triplet.third == "mingw32")
        k = ConversionSpecifier::???;
    break;

And perhaps similar modifications for the glibc (which would check for glibc being used) and objc (which should check for objc being used) conversion specifier. I do not know what ??? should be, nor if this well then be properly handled, so I'm leaving this to the experts.

I'll gladly test a patch for this, but I'm too stupid and C-agnostic to even understand what the function in question is doing.

Related mailing list discussion:
http://lists.cs.uiuc.edu/pipermail/cfe-dev/2012-August/023600.html

PS: for testing on Windows, download the GCC dw2 4.6.3-1 (http://sourceforge.net/projects/mingw-w64/files/Toolchains%20targetting%20Win32/Personal%20Builds/rubenvb/release/i686-w64-mingw32-gcc-dw2-4.6.3-1-release-win32_rubenvb.7z/download) and assorted Clang 3.1 (http://sourceforge.net/projects/mingw-w64/files/Toolchains%20targetting%20Win32/Personal%20Builds/rubenvb/release/i686-w64-mingw32-clang-3.1-release-win32_rubenvb.7z/download) and extract to the same directory. But as I said, I can take care of the testing myself.

[belkadan]

The interesting bits (for when we actually do this):

Types:
%c/%s means wint_t/wchar_t* for wprintf
%C/%S means wint_t/wchar_t* for printf and int/char* for wprintf
%Z means UNICODE_STRING * (custom struct type)

http://msdn.microsoft.com/en-us/library/hf4y5e3w.aspx

Sizes:
%I32d/%I64d means __int32 or __int64
%Id means ptrdiff_t (or intptr_t)
%Iu means size_t (or uintptr_t)

%hc/%hC means char, and same for s and S
%lc/%lC means wchar_t, and same for s and S
%wc means wchar_t, and same for s

http://msdn.microsoft.com/en-us/library/tcxf1dw6.aspx

[llvmbot]

*** Bug llvm/llvm-bugzilla-archive#15848 has been marked as a duplicate of this bug. ***

[erichkeane]

The "I" specifiers were added in -r188937.

W,c,C,z, etc done here: llvm/llvm-bugzilla-archive#20808

[tritao]

mentioned in issue llvm/llvm-bugzilla-archive#13707

[llvmbot]

mentioned in issue llvm/llvm-bugzilla-archive#15848