Hi, On Thu, Oct 7, 2010 at 8:41 AM, NAKAMURA Takumi <[email protected]> wrote: > Konbanwa, Nico. > > I think it would be better below eg. ; > > // RUN: %clang_cc1 -triple i686-pc-linux -emit-llvm %s -o - | > FileCheck %s -check-prefix=W32 > // RUN: %clang_cc1 -triple i686-pc-win32 -emit-llvm %s -o - | > FileCheck %s -check-prefix=W16 > > // W32: private constant [12 x i8] c"A\00\00\00B\00\00\00\00\00\00\00" > // W16: private constant [6 x i8] c"A\00B\00\00\00" > > Please see also http://llvm.org/docs/TestingGuide.html#FileCheck
I will look into this, thanks for the pointer. > It would be best if clang has the option wchar_t size, not to depend on > triplet. > > ps. it seems buggy to encode utf16(\u) and utf32(\U) on "L" literal > for W16 environment. \u UCNs are always <= 0xffff, so their utf16 is the same as the character itself. What does MSVC do for UCNs > 0xffff in L literal strings? gcc converts to utf16 if -fshort-wchar is used, so I'm emulating this for now. If you think it's better to do something else on windows, let me know what. Thanks, Nico > > > ...Takumi > > 2010/10/7 Nico Weber <[email protected]>: >> On Thu, Oct 7, 2010 at 6:42 AM, NAKAMURA Takumi <[email protected]> wrote: >>> Francois, >>> >>> I guess it would be due to wchar_t is 16 bit on win32. >>> I saw the same result on mingw. >> >> That'll be my guess, too. D'oh! >> >>> It passes to add -triple i686-linux for cc1 driver. >> >> What's the usual strategy to test platform-dependent codegen stuff? Is >> adding a triple to the CHECK line ok? >> >> Nico >> >>> >>> ...Takumi >>> >>> 2010/10/7 Francois Pichet <[email protected]>: >>>> I don't know why but the test fails on Windows: >>>> any idea? >>>> >>>> 1>Command 1: "FileCheck" >>>> "D:\Dev\llvm\llvm_trunk\tools\clang\test\CodeGen\string-literal.c" >>>> 1>Command 1 Result: 1 >>>> 1>Command 1 Output: >>>> 1>Command 1 Stderr: >>>> 1>D:\Dev\llvm\llvm_trunk\tools\clang\test\CodeGen\string-literal.c:11:12: >>>> error: expected string not found in input >>>> 1> // CHECK: private constant [12 x i8] c"A\00\00\00B\00\00\00\00\00\00\00" >>>> 1> ^ >>>> 1><stdin>:7:1: note: scanning from here >>>> 1>@.str = private constant [6 x i8] c"A\00B\00\00\00" >>>> 1>^ >>>> 1><stdin>:8:10: note: possible intended match here >>>> 1>@.str1 = private constant [10 x i8] c"4\12\00\00\0B\F0\10\00\00\00" >>>> 1> ^ >>>> 1>-- >>>> >>>> On Wed, Oct 6, 2010 at 12:57 AM, Nico Weber <[email protected]> wrote: >>>>> Author: nico >>>>> Date: Tue Oct 5 23:57:26 2010 >>>>> New Revision: 115743 >>>>> >>>>> URL: http://llvm.org/viewvc/llvm-project?rev=115743&view=rev >>>>> Log: >>>>> Add support for 4-byte UCNs like \U12345678. Warn about UCNs in c90 mode. >>>>> >>>>> Added: >>>>> cfe/trunk/test/CodeGen/string-literal-short-wstring.c >>>>> cfe/trunk/test/Lexer/wchar.c >>>>> Modified: >>>>> cfe/trunk/include/clang/Basic/DiagnosticLexKinds.td >>>>> cfe/trunk/lib/Lex/LiteralSupport.cpp >>>>> cfe/trunk/test/CodeGen/string-literal.c >>>>> cfe/trunk/test/Lexer/c90.c >>>>> >>>>> Modified: cfe/trunk/include/clang/Basic/DiagnosticLexKinds.td >>>>> URL: >>>>> http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Basic/DiagnosticLexKinds.td?rev=115743&r1=115742&r2=115743&view=diff >>>>> ============================================================================== >>>>> --- cfe/trunk/include/clang/Basic/DiagnosticLexKinds.td (original) >>>>> +++ cfe/trunk/include/clang/Basic/DiagnosticLexKinds.td Tue Oct 5 >>>>> 23:57:26 2010 >>>>> @@ -98,6 +98,10 @@ >>>>> def ext_string_too_long : Extension<"string literal of length %0 exceeds >>>>> " >>>>> "maximum length %1 that %select{C90|ISO C99|C++}2 compilers are >>>>> required to " >>>>> "support">, InGroup<OverlengthStrings>; >>>>> +def warn_ucn_escape_too_large : ExtWarn< >>>>> + "character unicode escape sequence too long for its type">; >>>>> +def warn_ucn_not_valid_in_c89 : ExtWarn< >>>>> + "unicode escape sequences are only valid in C99 or C++">; >>>>> >>>>> //===----------------------------------------------------------------------===// >>>>> // PTH Diagnostics >>>>> >>>>> Modified: cfe/trunk/lib/Lex/LiteralSupport.cpp >>>>> URL: >>>>> http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Lex/LiteralSupport.cpp?rev=115743&r1=115742&r2=115743&view=diff >>>>> ============================================================================== >>>>> --- cfe/trunk/lib/Lex/LiteralSupport.cpp (original) >>>>> +++ cfe/trunk/lib/Lex/LiteralSupport.cpp Tue Oct 5 23:57:26 2010 >>>>> @@ -172,8 +172,8 @@ >>>>> SourceLocation Loc, Preprocessor &PP, >>>>> bool wide, >>>>> bool Complain) { >>>>> - // FIXME: Add a warning - UCN's are only valid in C++ & C99. >>>>> - // FIXME: Handle wide strings. >>>>> + if (!PP.getLangOptions().CPlusPlus && !PP.getLangOptions().C99) >>>>> + PP.Diag(Loc, diag::warn_ucn_not_valid_in_c89); >>>>> >>>>> // Save the beginning of the string (for error diagnostics). >>>>> const char *ThisTokBegin = ThisTokBuf; >>>>> @@ -218,13 +218,34 @@ >>>>> } >>>>> if (wide) { >>>>> (void)UcnLenSave; >>>>> - assert(UcnLenSave == 4 && >>>>> - "ProcessUCNEscape - only ucn length of 4 supported"); >>>>> - // little endian assumed. >>>>> - *ResultBuf++ = (UcnVal & 0x000000FF); >>>>> - *ResultBuf++ = (UcnVal & 0x0000FF00) >> 8; >>>>> - *ResultBuf++ = (UcnVal & 0x00FF0000) >> 16; >>>>> - *ResultBuf++ = (UcnVal & 0xFF000000) >> 24; >>>>> + assert((UcnLenSave == 4 || UcnLenSave == 8) && >>>>> + "ProcessUCNEscape - only ucn length of 4 or 8 supported"); >>>>> + >>>>> + if (!PP.getLangOptions().ShortWChar) { >>>>> + // Note: our internal rep of wide char tokens is always >>>>> little-endian. >>>>> + *ResultBuf++ = (UcnVal & 0x000000FF); >>>>> + *ResultBuf++ = (UcnVal & 0x0000FF00) >> 8; >>>>> + *ResultBuf++ = (UcnVal & 0x00FF0000) >> 16; >>>>> + *ResultBuf++ = (UcnVal & 0xFF000000) >> 24; >>>>> + return; >>>>> + } >>>>> + >>>>> + // Convert to UTF16. >>>>> + if (UcnVal < (UTF32)0xFFFF) { >>>>> + *ResultBuf++ = (UcnVal & 0x000000FF); >>>>> + *ResultBuf++ = (UcnVal & 0x0000FF00) >> 8; >>>>> + return; >>>>> + } >>>>> + PP.Diag(Loc, diag::warn_ucn_escape_too_large); >>>>> + >>>>> + typedef uint16_t UTF16; >>>>> + UcnVal -= 0x10000; >>>>> + UTF16 surrogate1 = 0xD800 + (UcnVal >> 10); >>>>> + UTF16 surrogate2 = 0xDC00 + (UcnVal & 0x3FF); >>>>> + *ResultBuf++ = (surrogate1 & 0x000000FF); >>>>> + *ResultBuf++ = (surrogate1 & 0x0000FF00) >> 8; >>>>> + *ResultBuf++ = (surrogate2 & 0x000000FF); >>>>> + *ResultBuf++ = (surrogate2 & 0x0000FF00) >> 8; >>>>> return; >>>>> } >>>>> // Now that we've parsed/checked the UCN, we convert from UTF32->UTF8. >>>>> >>>>> Added: cfe/trunk/test/CodeGen/string-literal-short-wstring.c >>>>> URL: >>>>> http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/string-literal-short-wstring.c?rev=115743&view=auto >>>>> ============================================================================== >>>>> --- cfe/trunk/test/CodeGen/string-literal-short-wstring.c (added) >>>>> +++ cfe/trunk/test/CodeGen/string-literal-short-wstring.c Tue Oct 5 >>>>> 23:57:26 2010 >>>>> @@ -0,0 +1,14 @@ >>>>> +// RUN: %clang_cc1 -emit-llvm -fshort-wchar %s -o - | FileCheck %s >>>>> + >>>>> +int main() { >>>>> + // This should convert to utf8. >>>>> + // CHECK: internal constant [10 x i8] >>>>> c"\E1\84\A0\C8\A0\F4\82\80\B0\00", align 1 >>>>> + char b[10] = "\u1120\u0220\U00102030"; >>>>> + >>>>> + // CHECK: private constant [6 x i8] c"A\00B\00\00\00" >>>>> + void *foo = L"AB"; >>>>> + >>>>> + // This should convert to utf16. >>>>> + // CHECK: private constant [10 x i8] c" \11 \02\C8\DB0\DC\00\00" >>>>> + void *bar = L"\u1120\u0220\U00102030"; >>>>> +} >>>>> >>>>> Modified: cfe/trunk/test/CodeGen/string-literal.c >>>>> URL: >>>>> http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/string-literal.c?rev=115743&r1=115742&r2=115743&view=diff >>>>> ============================================================================== >>>>> --- cfe/trunk/test/CodeGen/string-literal.c (original) >>>>> +++ cfe/trunk/test/CodeGen/string-literal.c Tue Oct 5 23:57:26 2010 >>>>> @@ -1,7 +1,16 @@ >>>>> -// RUN: %clang_cc1 -emit-llvm %s -o - >>>>> +// RUN: %clang_cc1 -emit-llvm %s -o - | FileCheck %s >>>>> >>>>> int main() { >>>>> + // CHECK: internal constant [10 x i8] c"abc\00\00\00\00\00\00\00", >>>>> align 1 >>>>> char a[10] = "abc"; >>>>> >>>>> + // This should convert to utf8. >>>>> + // CHECK: internal constant [10 x i8] >>>>> c"\E1\84\A0\C8\A0\F4\82\80\B0\00", align 1 >>>>> + char b[10] = "\u1120\u0220\U00102030"; >>>>> + >>>>> + // CHECK: private constant [12 x i8] >>>>> c"A\00\00\00B\00\00\00\00\00\00\00" >>>>> void *foo = L"AB"; >>>>> + >>>>> + // CHECK: private constant [12 x i8] >>>>> c"4\12\00\00\0B\F0\10\00\00\00\00\00" >>>>> + void *bar = L"\u1234\U0010F00B"; >>>>> } >>>>> >>>>> Modified: cfe/trunk/test/Lexer/c90.c >>>>> URL: >>>>> http://llvm.org/viewvc/llvm-project/cfe/trunk/test/Lexer/c90.c?rev=115743&r1=115742&r2=115743&view=diff >>>>> ============================================================================== >>>>> --- cfe/trunk/test/Lexer/c90.c (original) >>>>> +++ cfe/trunk/test/Lexer/c90.c Tue Oct 5 23:57:26 2010 >>>>> @@ -27,3 +27,7 @@ >>>>> "sdjflksdjf lksdjf skldfjsdkljflksdjf kldsjflkdsj fldks jflsdkjfds" >>>>> "sdjflksdjf lksdjf skldfjsdkljflksdjf kldsjflkdsj fldks jflsdkjfds"; >>>>> } >>>>> + >>>>> +void test3() { >>>>> + (void)L"\u1234"; // expected-error {{unicode escape sequences are >>>>> only valid in C99 or C++}} >>>>> +} >>>>> >>>>> Added: cfe/trunk/test/Lexer/wchar.c >>>>> URL: >>>>> http://llvm.org/viewvc/llvm-project/cfe/trunk/test/Lexer/wchar.c?rev=115743&view=auto >>>>> ============================================================================== >>>>> --- cfe/trunk/test/Lexer/wchar.c (added) >>>>> +++ cfe/trunk/test/Lexer/wchar.c Tue Oct 5 23:57:26 2010 >>>>> @@ -0,0 +1,6 @@ >>>>> +// RUN: %clang_cc1 -fsyntax-only -fshort-wchar -verify %s >>>>> + >>>>> +void f() { >>>>> + (void)L"\U00010000"; // expected-warning {{character unicode escape >>>>> sequence too long for its type}} >>>>> +} >>>>> + >>>>> >>>>> >>>>> _______________________________________________ >>>>> cfe-commits mailing list >>>>> [email protected] >>>>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits >>>>> >>>> >>>> _______________________________________________ >>>> cfe-commits mailing list >>>> [email protected] >>>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits >>>> >>> >>> _______________________________________________ >>> cfe-commits mailing list >>> [email protected] >>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits >>> >> > _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
