[clang] [clang-format] annotate inline assembly parens (PR #201703)
Gedare Bloom via cfe-commits
cfe-commits at lists.llvm.org
Tue Jun 9 12:04:07 PDT 2026
https://github.com/gedare updated https://github.com/llvm/llvm-project/pull/201703
>From a58b160f6dabdf02515b46d5faca98173ee45fbb Mon Sep 17 00:00:00 2001
From: Gedare Bloom <gedare at rtems.org>
Date: Thu, 4 Jun 2026 15:39:06 -0600
Subject: [PATCH 1/5] [clang-format] annotate inline assembly parens
Annotate the opening and closing parens of inline assembly.
This will make other improvements related to inline assembly easier.
---
clang/lib/Format/FormatToken.h | 1 +
clang/lib/Format/TokenAnnotator.cpp | 2 +
clang/lib/Format/UnwrappedLineParser.cpp | 38 +++++++++++++++----
clang/unittests/Format/TokenAnnotatorTest.cpp | 12 ++++++
4 files changed, 46 insertions(+), 7 deletions(-)
diff --git a/clang/lib/Format/FormatToken.h b/clang/lib/Format/FormatToken.h
index 556bb0f3dd0af..ed9ce435ea765 100644
--- a/clang/lib/Format/FormatToken.h
+++ b/clang/lib/Format/FormatToken.h
@@ -99,6 +99,7 @@ namespace format {
TYPE(InheritanceComma) \
TYPE(InlineASMBrace) \
TYPE(InlineASMColon) \
+ TYPE(InlineASMParen) \
TYPE(InlineASMSymbolicNameLSquare) \
TYPE(JavaAnnotation) \
TYPE(JsAndAndEqual) \
diff --git a/clang/lib/Format/TokenAnnotator.cpp b/clang/lib/Format/TokenAnnotator.cpp
index c47a512454476..f211325a3b249 100644
--- a/clang/lib/Format/TokenAnnotator.cpp
+++ b/clang/lib/Format/TokenAnnotator.cpp
@@ -2536,6 +2536,8 @@ class AnnotatingParser {
} else if (Current.is(tok::r_paren)) {
if (rParenEndsCast(Current))
Current.setType(TT_CastRParen);
+ if (Current.MatchingParen && Current.MatchingParen->is(TT_InlineASMParen))
+ Current.setType(TT_InlineASMParen);
if (Current.MatchingParen && Current.Next &&
!Current.Next->isBinaryOperator() &&
Current.Next->isNoneOf(
diff --git a/clang/lib/Format/UnwrappedLineParser.cpp b/clang/lib/Format/UnwrappedLineParser.cpp
index c83e82674dee1..92231e1598a12 100644
--- a/clang/lib/Format/UnwrappedLineParser.cpp
+++ b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -1508,23 +1508,47 @@ void UnwrappedLineParser::parseStructuralElement(
return;
}
switch (FormatTok->Tok.getKind()) {
- case tok::kw_asm:
+ case tok::kw_asm: {
+ bool DoNotFormat = false;
+ tok::TokenKind OpenType;
+ tok::TokenKind CloseType;
nextToken();
- if (FormatTok->is(tok::l_brace)) {
+ while (FormatTok &&
+ FormatTok->isOneOf(tok::kw_volatile, tok::kw_inline, tok::kw_goto)) {
+ nextToken();
+ }
+ if (FormatTok && FormatTok->is(tok::l_brace)) {
FormatTok->setFinalizedType(TT_InlineASMBrace);
+ OpenType = tok::l_brace;
+ CloseType = tok::r_brace;
+ DoNotFormat = true;
+ } else if (FormatTok && FormatTok->is(tok::l_paren)) {
+ OpenType = tok::l_paren;
+ CloseType = tok::r_paren;
+ FormatTok->setFinalizedType(TT_InlineASMParen);
+ }
+ if (DoNotFormat) {
+ FormatToken *OpenTok = FormatTok;
+ int NestLevel = 0;
nextToken();
while (FormatTok && !eof()) {
- if (FormatTok->is(tok::r_brace)) {
- FormatTok->setFinalizedType(TT_InlineASMBrace);
- nextToken();
- addUnwrappedLine();
- break;
+ if (FormatTok->is(OpenType)) {
+ ++NestLevel;
+ } else if (FormatTok->is(CloseType)) {
+ --NestLevel;
+ if (NestLevel < 1) {
+ FormatTok->setFinalizedType(OpenTok->getType());
+ nextToken();
+ addUnwrappedLine();
+ break;
+ }
}
FormatTok->Finalized = true;
nextToken();
}
}
break;
+ }
case tok::kw_namespace:
parseNamespace();
return;
diff --git a/clang/unittests/Format/TokenAnnotatorTest.cpp b/clang/unittests/Format/TokenAnnotatorTest.cpp
index 34106489d7f38..398c308d245ae 100644
--- a/clang/unittests/Format/TokenAnnotatorTest.cpp
+++ b/clang/unittests/Format/TokenAnnotatorTest.cpp
@@ -1930,9 +1930,11 @@ TEST_F(TokenAnnotatorTest, UnderstandsAsm) {
":);");
ASSERT_EQ(Tokens.size(), 10u) << Tokens;
EXPECT_TOKEN(Tokens[0], tok::kw_asm, TT_Unknown);
+ EXPECT_TOKEN(Tokens[1], tok::l_paren, TT_InlineASMParen);
EXPECT_TOKEN(Tokens[3], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[4], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[6], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[7], tok::r_paren, TT_InlineASMParen);
Tokens = annotate("asm volatile (\n"
"\"a_label:\"\n"
@@ -1941,9 +1943,11 @@ TEST_F(TokenAnnotatorTest, UnderstandsAsm) {
":);");
ASSERT_EQ(Tokens.size(), 11u) << Tokens;
EXPECT_TOKEN(Tokens[0], tok::kw_asm, TT_Unknown);
+ EXPECT_TOKEN(Tokens[2], tok::l_paren, TT_InlineASMParen);
EXPECT_TOKEN(Tokens[4], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[5], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[7], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[8], tok::r_paren, TT_InlineASMParen);
Tokens = annotate("__asm__(\n"
"\"a_label:\"\n"
@@ -1952,9 +1956,11 @@ TEST_F(TokenAnnotatorTest, UnderstandsAsm) {
": y);");
ASSERT_EQ(Tokens.size(), 11u) << Tokens;
EXPECT_TOKEN(Tokens[0], tok::kw_asm, TT_Unknown);
+ EXPECT_TOKEN(Tokens[1], tok::l_paren, TT_InlineASMParen);
EXPECT_TOKEN(Tokens[3], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[5], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[6], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[8], tok::r_paren, TT_InlineASMParen);
Tokens = annotate("__asm volatile (\n"
"\"a_label:\"\n"
@@ -1964,9 +1970,11 @@ TEST_F(TokenAnnotatorTest, UnderstandsAsm) {
":);");
ASSERT_EQ(Tokens.size(), 12u) << Tokens;
EXPECT_TOKEN(Tokens[0], tok::kw_asm, TT_Unknown);
+ EXPECT_TOKEN(Tokens[2], tok::l_paren, TT_InlineASMParen);
EXPECT_TOKEN(Tokens[5], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[6], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[8], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[9], tok::r_paren, TT_InlineASMParen);
Tokens = annotate("asm(\n"
"\"insn\"\n"
@@ -1975,9 +1983,11 @@ TEST_F(TokenAnnotatorTest, UnderstandsAsm) {
": \"memory\");");
ASSERT_EQ(Tokens.size(), 19u) << Tokens;
EXPECT_TOKEN(Tokens[0], tok::kw_asm, TT_Unknown);
+ EXPECT_TOKEN(Tokens[1], tok::l_paren, TT_InlineASMParen);
EXPECT_TOKEN(Tokens[3], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[13], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[14], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[16], tok::r_paren, TT_InlineASMParen);
Tokens = annotate("__asm__ volatile (\n"
"\"ldr r1, [r0, %%[sym]]\"\n"
@@ -1986,9 +1996,11 @@ TEST_F(TokenAnnotatorTest, UnderstandsAsm) {
");");
ASSERT_EQ(Tokens.size(), 21u) << Tokens;
EXPECT_TOKEN(Tokens[0], tok::kw_asm, TT_Unknown);
+ EXPECT_TOKEN(Tokens[2], tok::l_paren, TT_InlineASMParen);
EXPECT_TOKEN(Tokens[4], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[5], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[6], tok::l_square, TT_InlineASMSymbolicNameLSquare);
+ EXPECT_TOKEN(Tokens[18], tok::r_paren, TT_InlineASMParen);
}
TEST_F(TokenAnnotatorTest, UnderstandsObjCBlock) {
>From 42fd2fb6ef6483b7a434d1655493c8e5131d2a18 Mon Sep 17 00:00:00 2001
From: Gedare Bloom <gedare at rtems.org>
Date: Fri, 5 Jun 2026 11:22:24 -0600
Subject: [PATCH 2/5] Add annotation tests for asm qualifiers
---
clang/unittests/Format/TokenAnnotatorTest.cpp | 45 +++++++++++++++++++
1 file changed, 45 insertions(+)
diff --git a/clang/unittests/Format/TokenAnnotatorTest.cpp b/clang/unittests/Format/TokenAnnotatorTest.cpp
index 398c308d245ae..a178222d80eb2 100644
--- a/clang/unittests/Format/TokenAnnotatorTest.cpp
+++ b/clang/unittests/Format/TokenAnnotatorTest.cpp
@@ -2001,6 +2001,51 @@ TEST_F(TokenAnnotatorTest, UnderstandsAsm) {
EXPECT_TOKEN(Tokens[5], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[6], tok::l_square, TT_InlineASMSymbolicNameLSquare);
EXPECT_TOKEN(Tokens[18], tok::r_paren, TT_InlineASMParen);
+
+ Tokens = annotate("__asm__ inline (\n"
+ "\"ldr r1, [r0, %%[sym]]\"\n"
+ ":\n"
+ ": [sym] \"J\" (aaaaa(aaaa, aaaa))\n"
+ ");");
+ ASSERT_EQ(Tokens.size(), 21u) << Tokens;
+ EXPECT_TOKEN(Tokens[0], tok::kw_asm, TT_Unknown);
+ EXPECT_TOKEN(Tokens[2], tok::l_paren, TT_InlineASMParen);
+ EXPECT_TOKEN(Tokens[4], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[5], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[6], tok::l_square, TT_InlineASMSymbolicNameLSquare);
+ EXPECT_TOKEN(Tokens[18], tok::r_paren, TT_InlineASMParen);
+
+ Tokens = annotate("asm goto (\n"
+ "\"btl %1, %0\\n\\t\" \"jc %l2\"\n"
+ ":\n"
+ ": \"r\"(a), \"r\"(b)\n"
+ ": \"cc\"\n"
+ ": carry\n"
+ ");");
+ ASSERT_EQ(Tokens.size(), 23u) << Tokens;
+ EXPECT_TOKEN(Tokens[0], tok::kw_asm, TT_Unknown);
+ EXPECT_TOKEN(Tokens[2], tok::l_paren, TT_InlineASMParen);
+ EXPECT_TOKEN(Tokens[5], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[6], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[16], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[18], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[20], tok::r_paren, TT_InlineASMParen);
+
+ Tokens = annotate("__asm__ volatile inline goto (\n"
+ "\"btl %1, %0\\n\\t\" \"jc %l2\"\n"
+ ":\n"
+ ": \"r\"(a), \"r\"(b)\n"
+ ": \"cc\"\n"
+ ": carry\n"
+ ");");
+ ASSERT_EQ(Tokens.size(), 25u) << Tokens;
+ EXPECT_TOKEN(Tokens[0], tok::kw_asm, TT_Unknown);
+ EXPECT_TOKEN(Tokens[4], tok::l_paren, TT_InlineASMParen);
+ EXPECT_TOKEN(Tokens[7], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[8], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[18], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[20], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[22], tok::r_paren, TT_InlineASMParen);
}
TEST_F(TokenAnnotatorTest, UnderstandsObjCBlock) {
>From d70c52e84cabdf1db17ac182857d095a97f3a011 Mon Sep 17 00:00:00 2001
From: Gedare Bloom <gedare at rtems.org>
Date: Fri, 5 Jun 2026 11:25:01 -0600
Subject: [PATCH 3/5] Simplify logic for NULL condition
---
clang/lib/Format/UnwrappedLineParser.cpp | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/clang/lib/Format/UnwrappedLineParser.cpp b/clang/lib/Format/UnwrappedLineParser.cpp
index 92231e1598a12..354e518926653 100644
--- a/clang/lib/Format/UnwrappedLineParser.cpp
+++ b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -1517,12 +1517,14 @@ void UnwrappedLineParser::parseStructuralElement(
FormatTok->isOneOf(tok::kw_volatile, tok::kw_inline, tok::kw_goto)) {
nextToken();
}
- if (FormatTok && FormatTok->is(tok::l_brace)) {
+ if (!FormatTok)
+ break;
+ if (FormatTok->is(tok::l_brace)) {
FormatTok->setFinalizedType(TT_InlineASMBrace);
OpenType = tok::l_brace;
CloseType = tok::r_brace;
DoNotFormat = true;
- } else if (FormatTok && FormatTok->is(tok::l_paren)) {
+ } else if (FormatTok->is(tok::l_paren)) {
OpenType = tok::l_paren;
CloseType = tok::r_paren;
FormatTok->setFinalizedType(TT_InlineASMParen);
>From a978ad2957b0bff1fcc8c7c26a56bcf275f8a345 Mon Sep 17 00:00:00 2001
From: Gedare Bloom <gedare at rtems.org>
Date: Fri, 5 Jun 2026 16:22:20 -0600
Subject: [PATCH 4/5] Updates from review comments
---
clang/lib/Format/UnwrappedLineParser.cpp | 5 +++
clang/unittests/Format/TokenAnnotatorTest.cpp | 45 +++----------------
2 files changed, 10 insertions(+), 40 deletions(-)
diff --git a/clang/lib/Format/UnwrappedLineParser.cpp b/clang/lib/Format/UnwrappedLineParser.cpp
index 354e518926653..79ab2d460d295 100644
--- a/clang/lib/Format/UnwrappedLineParser.cpp
+++ b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -1509,6 +1509,9 @@ void UnwrappedLineParser::parseStructuralElement(
}
switch (FormatTok->Tok.getKind()) {
case tok::kw_asm: {
+ // Track whether to skip formatting the inline asm block with a variable.
+ // Formatting is skipped for any asm blocks inside of braces by default.
+ // A style option could be added to also skip formatting inside parens.
bool DoNotFormat = false;
tok::TokenKind OpenType;
tok::TokenKind CloseType;
@@ -1528,6 +1531,8 @@ void UnwrappedLineParser::parseStructuralElement(
OpenType = tok::l_paren;
CloseType = tok::r_paren;
FormatTok->setFinalizedType(TT_InlineASMParen);
+ } else {
+ break;
}
if (DoNotFormat) {
FormatToken *OpenTok = FormatTok;
diff --git a/clang/unittests/Format/TokenAnnotatorTest.cpp b/clang/unittests/Format/TokenAnnotatorTest.cpp
index a178222d80eb2..3ec74c7e705b4 100644
--- a/clang/unittests/Format/TokenAnnotatorTest.cpp
+++ b/clang/unittests/Format/TokenAnnotatorTest.cpp
@@ -2002,50 +2002,15 @@ TEST_F(TokenAnnotatorTest, UnderstandsAsm) {
EXPECT_TOKEN(Tokens[6], tok::l_square, TT_InlineASMSymbolicNameLSquare);
EXPECT_TOKEN(Tokens[18], tok::r_paren, TT_InlineASMParen);
- Tokens = annotate("__asm__ inline (\n"
- "\"ldr r1, [r0, %%[sym]]\"\n"
- ":\n"
- ": [sym] \"J\" (aaaaa(aaaa, aaaa))\n"
- ");");
- ASSERT_EQ(Tokens.size(), 21u) << Tokens;
- EXPECT_TOKEN(Tokens[0], tok::kw_asm, TT_Unknown);
- EXPECT_TOKEN(Tokens[2], tok::l_paren, TT_InlineASMParen);
- EXPECT_TOKEN(Tokens[4], tok::colon, TT_InlineASMColon);
- EXPECT_TOKEN(Tokens[5], tok::colon, TT_InlineASMColon);
- EXPECT_TOKEN(Tokens[6], tok::l_square, TT_InlineASMSymbolicNameLSquare);
- EXPECT_TOKEN(Tokens[18], tok::r_paren, TT_InlineASMParen);
-
- Tokens = annotate("asm goto (\n"
- "\"btl %1, %0\\n\\t\" \"jc %l2\"\n"
- ":\n"
- ": \"r\"(a), \"r\"(b)\n"
- ": \"cc\"\n"
- ": carry\n"
- ");");
- ASSERT_EQ(Tokens.size(), 23u) << Tokens;
- EXPECT_TOKEN(Tokens[0], tok::kw_asm, TT_Unknown);
- EXPECT_TOKEN(Tokens[2], tok::l_paren, TT_InlineASMParen);
- EXPECT_TOKEN(Tokens[5], tok::colon, TT_InlineASMColon);
- EXPECT_TOKEN(Tokens[6], tok::colon, TT_InlineASMColon);
- EXPECT_TOKEN(Tokens[16], tok::colon, TT_InlineASMColon);
- EXPECT_TOKEN(Tokens[18], tok::colon, TT_InlineASMColon);
- EXPECT_TOKEN(Tokens[20], tok::r_paren, TT_InlineASMParen);
-
- Tokens = annotate("__asm__ volatile inline goto (\n"
- "\"btl %1, %0\\n\\t\" \"jc %l2\"\n"
- ":\n"
- ": \"r\"(a), \"r\"(b)\n"
- ": \"cc\"\n"
- ": carry\n"
- ");");
- ASSERT_EQ(Tokens.size(), 25u) << Tokens;
+ Tokens = annotate("__asm__ volatile inline goto (\"nop\" : : : : l );");
+ ASSERT_EQ(Tokens.size(), 14u) << Tokens;
EXPECT_TOKEN(Tokens[0], tok::kw_asm, TT_Unknown);
EXPECT_TOKEN(Tokens[4], tok::l_paren, TT_InlineASMParen);
+ EXPECT_TOKEN(Tokens[6], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[7], tok::colon, TT_InlineASMColon);
EXPECT_TOKEN(Tokens[8], tok::colon, TT_InlineASMColon);
- EXPECT_TOKEN(Tokens[18], tok::colon, TT_InlineASMColon);
- EXPECT_TOKEN(Tokens[20], tok::colon, TT_InlineASMColon);
- EXPECT_TOKEN(Tokens[22], tok::r_paren, TT_InlineASMParen);
+ EXPECT_TOKEN(Tokens[9], tok::colon, TT_InlineASMColon);
+ EXPECT_TOKEN(Tokens[11], tok::r_paren, TT_InlineASMParen);
}
TEST_F(TokenAnnotatorTest, UnderstandsObjCBlock) {
>From bf324fe218d21caacc0e4247fb7ea111ebe8dadc Mon Sep 17 00:00:00 2001
From: Gedare Bloom <gedare at rtems.org>
Date: Mon, 8 Jun 2026 10:56:18 -0600
Subject: [PATCH 5/5] UnwrappedLineParser: comment on finalizing type
---
clang/lib/Format/UnwrappedLineParser.cpp | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/clang/lib/Format/UnwrappedLineParser.cpp b/clang/lib/Format/UnwrappedLineParser.cpp
index 79ab2d460d295..bf8320f6f51a4 100644
--- a/clang/lib/Format/UnwrappedLineParser.cpp
+++ b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -1509,8 +1509,8 @@ void UnwrappedLineParser::parseStructuralElement(
}
switch (FormatTok->Tok.getKind()) {
case tok::kw_asm: {
- // Track whether to skip formatting the inline asm block with a variable.
- // Formatting is skipped for any asm blocks inside of braces by default.
+ // Track whether to skip formatting inline asm by finalizing the tokens
+ // in the block. Formatting is skipped inside of braces by default.
// A style option could be added to also skip formatting inside parens.
bool DoNotFormat = false;
tok::TokenKind OpenType;
More information about the cfe-commits
mailing list