[clang-format] do not insert spaces around inline asm symbolic names

[krasimir]

Summary

Fixes https://bugs.llvm.org/show_bug.cgi?id=45108.

The [ in such cases was mis-annotated as an TT_ArrayInitializerLSquare.

[hans]

Very nice, thanks!

clang/lib/Format/TokenAnnotator.cpp

502

Is the Style.isCpp() check necessary here?

If it's necessary, maybe that's because the setting of InsideInlineASM should be guarded by Style.isCpp() instead?

[krasimir]

Remove redundant check.

[krasimir]

clang/lib/Format/TokenAnnotator.cpp

502

Good point! I think it's unnecessary.

[hans]

lgtm, thanks for the quick fix!

[krasimir]

Merging commits into 1 patch

clang/lib/Format/FormatToken.h

#define LIST_TOKEN_TYPES                                                       \
   TYPE(FunctionDeclarationName)                                                \
   TYPE(FunctionLBrace)                                                         \
   TYPE(FunctionTypeLParen)                                                     \
   TYPE(ImplicitStringLiteral)                                                  \
   TYPE(InheritanceColon)                                                       \
   TYPE(InheritanceComma)                                                       \
   TYPE(InlineASMBrace)                                                         \
   TYPE(InlineASMColon)                                                         \
+  TYPE(InlineASMSymbolicNameLSquare)                                           \
   TYPE(JavaAnnotation)                                                         \
   TYPE(JsComputedPropertyName)                                                 \
   TYPE(JsExponentiation)                                                       \
   TYPE(JsExponentiationEqual)                                                  \
   TYPE(JsFatArrow)                                                             \
   TYPE(JsNonNullAssertion)                                                     \
   TYPE(JsNullishCoalescingOperator)                                            \
   TYPE(JsNullPropagatingOperator)                                              \

clang/lib/Format/TokenAnnotator.cpp

bool parseSquare() {
     bool ColonFound = false;
 
     unsigned BindingIncrease = 1;
     if (IsCppStructuredBinding) {
       Left->Type = TT_StructuredBindingLSquare;
     } else if (Left->is(TT_Unknown)) {
       if (StartsObjCMethodExpr) {
         Left->Type = TT_ObjCMethodExpr;
+      } else if (Style.isCpp() && InsideInlineASM) {
+        Left->Type = TT_InlineASMSymbolicNameLSquare;
       } else if (IsCpp11AttributeSpecifier) {
         Left->Type = TT_AttributeSquare;
       } else if (Style.Language == FormatStyle::LK_JavaScript && Parent &&
                  Contexts.back().ContextKind == tok::l_brace &&
                  Parent->isOneOf(tok::l_brace, tok::comma)) {
         Left->Type = TT_JsComputedPropertyName;
       } else if (Style.isCpp() && Contexts.back().ContextKind == tok::l_brace &&
                  Parent && Parent->isOneOf(tok::l_brace, tok::comma)) {

clang/unittests/Format/FormatTest.cpp

EXPECT_EQ(FormatStyle::LK_ObjC,
             guessLanguage("foo.h", "int(^foo[10])(char, float);"));
   EXPECT_EQ(FormatStyle::LK_ObjC,
             guessLanguage("foo.h", "int(^foo[kNumEntries])(char, float);"));
   EXPECT_EQ(
       FormatStyle::LK_ObjC,
       guessLanguage("foo.h", "int(^foo[(kNumEntries + 10)])(char, float);"));
 }
 
+TEST_F(FormatTest, FormatsInlineAsmSymbolicNames) {
+  // ASM symbolic names are identifiers that must be surrounded by [] without
+  // space in between:
+  // https://gcc.gnu.org/onlinedocs/gcc/Extended-Asm.html#InputOperands
+
+  // Example from https://bugs.llvm.org/show_bug.cgi?id=45108.
+  verifyFormat(R"(//
+asm volatile("mrs %x[result], FPCR" : [result] "=r"(result));
+)");
+
+  // A list of several ASM symbolic names.
+  verifyFormat(R"(asm("mov %[e], %[d]" : [d] "=rm"(d), [e] "rm"(*e));)");
+
+  // ASM symbolic names in inline ASM with inputs and outputs.
+  verifyFormat(R"(//
+asm("cmoveq %1, %2, %[result]"
+    : [result] "=r"(result)
+    : "r"(test), "r"(new), "[result]"(old));
+)");
+
+  // ASM symbolic names in inline ASM with no outputs.
+  verifyFormat(R"(asm("mov %[e], %[d]" : : [d] "=rm"(d), [e] "rm"(*e));)");
+}
+
 TEST_F(FormatTest, GuessedLanguageWithInlineAsmClobbers) {
   EXPECT_EQ(FormatStyle::LK_Cpp,
             guessLanguage("foo.h", "void f() {\n"
                                    "  asm (\"mov %[e], %[d]\"\n"
                                    "     : [d] \"=rm\" (d)\n"
                                    "       [e] \"rm\" (*e));\n"
                                    "}"));
   EXPECT_EQ(FormatStyle::LK_Cpp,