https://github.com/fyrsta7 created 
https://github.com/llvm/llvm-project/pull/215837

Clang-format currently changes compact TableGen bit ranges such as
`Inst{24-20}` to `Inst{24 -20}`. TableGen lexes the lower bound as a negative
integer token, so the generic rule that separates adjacent word-like tokens
inserts the unwanted space.

Keep adjacent numeric tokens together in TableGen when the second token is a
negative integer. This preserves the established compact spelling of
hyphenated bit ranges. Add coverage for both a single range and a
comma-separated range list, plus a release note for the user-visible
formatting fix.

Tests:

- `build-pr/tools/clang/unittests/Format/FormatTests` (1276 passed)
- `ninja -C build-pr check-clang-format` (33 passed)
- `python3 clang/tools/clang-format/git-clang-format --binary 
build-pr/bin/clang-format --diff origin/main`
- `git diff --check origin/main...HEAD`

Fixes #177051

Assisted-by: OpenAI Codex


>From d10cd6bb26ca4ee652779adaecadecb9974bf97f Mon Sep 17 00:00:00 2001
From: Yuwei Zhao <[email protected]>
Date: Wed, 12 Aug 2026 23:41:26 +0800
Subject: [PATCH] [clang-format] Preserve compact TableGen bit ranges

TableGen lexes the lower bound of a hyphenated bit range as a negative
integer token. Avoid separating adjacent numeric tokens in TableGen so ranges
such as {24-20} remain compact.

Fixes #177051

Assisted-by: OpenAI Codex
---
 clang/docs/ReleaseNotes.md                    | 3 +++
 clang/lib/Format/TokenAnnotator.cpp           | 6 ++++++
 clang/unittests/Format/FormatTestTableGen.cpp | 7 +++++++
 3 files changed, 16 insertions(+)

diff --git a/clang/docs/ReleaseNotes.md b/clang/docs/ReleaseNotes.md
index 9648ad429d040..daaf7ca983465 100644
--- a/clang/docs/ReleaseNotes.md
+++ b/clang/docs/ReleaseNotes.md
@@ -568,6 +568,9 @@ features cannot lower the translation-unit ABI level;
 - Add `SpacesInBlockComments` option to control spacing after `/*` and
   before `*/` in ordinary block comments.
 
+- Fixed formatting of TableGen bit ranges so that hyphen separators remain
+  compact, as in `{24-20}`. (#GH177051)
+
 ### libclang
 
 - visit identifier initializers in lambda capture as VarDecl instead of 
VariableRef. Warning: this changes behaviour.
diff --git a/clang/lib/Format/TokenAnnotator.cpp 
b/clang/lib/Format/TokenAnnotator.cpp
index b6c33279b0aca..349ce2b28ae4a 100644
--- a/clang/lib/Format/TokenAnnotator.cpp
+++ b/clang/lib/Format/TokenAnnotator.cpp
@@ -5208,6 +5208,12 @@ bool TokenAnnotator::spaceRequiredBefore(const 
AnnotatedLine &Line,
   const bool IsVerilog = Style.isVerilog();
   assert(!IsVerilog || !IsCpp);
 
+  // TableGen lexes the lower bound of a bit range as a negative integer.
+  if (Style.isTableGen() && Left.is(tok::numeric_constant) &&
+      Right.is(tok::numeric_constant) && Right.TokenText.starts_with("-")) {
+    return false;
+  }
+
   // Never ever merge two words.
   if (Keywords.isWordLike(Right, IsVerilog) &&
       Keywords.isWordLike(Left, IsVerilog)) {
diff --git a/clang/unittests/Format/FormatTestTableGen.cpp 
b/clang/unittests/Format/FormatTestTableGen.cpp
index df20cc26e1094..da0f1b155f2f1 100644
--- a/clang/unittests/Format/FormatTestTableGen.cpp
+++ b/clang/unittests/Format/FormatTestTableGen.cpp
@@ -216,6 +216,13 @@ TEST_F(FormatTestTableGen, ValueSuffix) {
                "}");
 }
 
+TEST_F(FormatTestTableGen, BitRanges) {
+  verifyFormat("def Ranges {\n"
+               "  let Inst{24-20} = rs2;\n"
+               "  let Inst{19-15, 11-7} = rs1;\n"
+               "}");
+}
+
 TEST_F(FormatTestTableGen, PasteOperator) {
   verifyFormat("def Paste#\"Operator\" { string Paste = \"Paste\"#operator; 
}");
 

_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to