[clang] [llvm] [RISCV] Allow extra underscores in parseNormalizedArchString and parseArchString. (PR #91532)

[Craig Topper via cfe-commits]

https://github.com/topperc updated https://github.com/llvm/llvm-project/pull/91532

>From 0f5fcf5035d66fd4c1d1e97525e9266d8460b297 Mon Sep 17 00:00:00 2001
From: Craig Topper <craig.topper at sifive.com>
Date: Wed, 8 May 2024 13:40:31 -0700
Subject: [PATCH] [RISCV] Allow extra underscores in parseNormalizedArchString
 and parseArchString.

Allow double underscores and trailing underscores. gcc and binutils
allow extra underscores without error.
---
 clang/test/Driver/riscv-arch.c                   |  5 -----
 llvm/lib/TargetParser/RISCVISAInfo.cpp           | 16 +++-------------
 llvm/unittests/TargetParser/RISCVISAInfoTest.cpp | 15 +--------------
 3 files changed, 4 insertions(+), 32 deletions(-)

diff --git a/clang/test/Driver/riscv-arch.c b/clang/test/Driver/riscv-arch.c
index ddf617bbb6237..418d8e91595de 100644
--- a/clang/test/Driver/riscv-arch.c
+++ b/clang/test/Driver/riscv-arch.c
@@ -308,11 +308,6 @@
 // RV32-SMINOR0: error: invalid arch name 'rv32ist2p0',
 // RV32-SMINOR0: unsupported standard supervisor-level extension 'st'
 
-// RUN: not %clang --target=riscv32-unknown-elf -march=rv32ixabc_ -### %s \
-// RUN: -fsyntax-only 2>&1 | FileCheck -check-prefix=RV32-XSEP %s
-// RV32-XSEP: error: invalid arch name 'rv32ixabc_',
-// RV32-XSEP: extension name missing after separator '_'
-
 // RUN: not %clang --target=riscv32-unknown-elf -march=rv32ixabc_a -### %s \
 // RUN: -fsyntax-only 2>&1 | FileCheck -check-prefix=RV32-PREFIX %s
 // RV32-PREFIX: error: invalid arch name 'rv32ixabc_a',
diff --git a/llvm/lib/TargetParser/RISCVISAInfo.cpp b/llvm/lib/TargetParser/RISCVISAInfo.cpp
index c553e330a878b..14b757d04ce37 100644
--- a/llvm/lib/TargetParser/RISCVISAInfo.cpp
+++ b/llvm/lib/TargetParser/RISCVISAInfo.cpp
@@ -452,7 +452,7 @@ RISCVISAInfo::parseNormalizedArchString(StringRef Arch) {
   // and separated by _. Split by _ and then extract the name and version
   // information for each extension.
   SmallVector<StringRef, 8> Split;
-  Arch.split(Split, '_');
+  Arch.split(Split, '_', /*MaxSplit=*/-1, /*KeepEmpty=*/false);
   for (StringRef Ext : Split) {
     StringRef Prefix, MinorVersionStr;
     std::tie(Prefix, MinorVersionStr) = Ext.rsplit('p');
@@ -651,10 +651,6 @@ RISCVISAInfo::parseArchString(StringRef Arch, bool EnableExperimentalExtension,
     break;
   }
 
-  if (Arch.back() == '_')
-    return createStringError(errc::invalid_argument,
-                             "extension name missing after separator '_'");
-
   // Skip baseline.
   StringRef Exts = Arch.drop_front(1);
 
@@ -697,16 +693,10 @@ RISCVISAInfo::parseArchString(StringRef Arch, bool EnableExperimentalExtension,
   Exts.consume_front("_");
 
   SmallVector<StringRef, 8> SplitExts;
-  // Only split if the string is not empty. Otherwise the split will push an
-  // empty string into the vector.
-  if (!Exts.empty())
-    Exts.split(SplitExts, '_');
+  Exts.split(SplitExts, '_', /*MaxSplit=*/-1, /*KeepEmpty=*/false);
 
   for (auto Ext : SplitExts) {
-    if (Ext.empty())
-      return createStringError(errc::invalid_argument,
-                               "extension name missing after separator '_'");
-
+    assert(!Ext.empty());
     do {
       if (RISCVISAUtils::AllStdExts.contains(Ext.front())) {
         if (auto E = processSingleLetterExtension(
diff --git a/llvm/unittests/TargetParser/RISCVISAInfoTest.cpp b/llvm/unittests/TargetParser/RISCVISAInfoTest.cpp
index f9e386a85fea8..dc2a8b44cf15d 100644
--- a/llvm/unittests/TargetParser/RISCVISAInfoTest.cpp
+++ b/llvm/unittests/TargetParser/RISCVISAInfoTest.cpp
@@ -38,8 +38,7 @@ TEST(ParseNormalizedArchString, RejectsInvalidBaseISA) {
 }
 
 TEST(ParseNormalizedArchString, RejectsMalformedInputs) {
-  for (StringRef Input :
-       {"rv64i2p0_", "rv32i2p0__a2p0", "rv64e2p", "rv32i", "rv64ip1"}) {
+  for (StringRef Input : {"rv64e2p", "rv32i", "rv64ip1"}) {
     EXPECT_EQ(
         toString(RISCVISAInfo::parseNormalizedArchString(Input).takeError()),
         "extension lacks version in expected format");
@@ -518,18 +517,6 @@ TEST(ParseArchString,
       "unsupported standard user-level extension 'zba1p0m'");
 }
 
-TEST(ParseArchString, RejectsDoubleOrTrailingUnderscore) {
-  EXPECT_EQ(
-      toString(RISCVISAInfo::parseArchString("rv64i__m", true).takeError()),
-      "extension name missing after separator '_'");
-
-  for (StringRef Input :
-       {"rv32ezicsr__zifencei", "rv32i_", "rv32izicsr_", "rv64im_"}) {
-    EXPECT_EQ(toString(RISCVISAInfo::parseArchString(Input, true).takeError()),
-              "extension name missing after separator '_'");
-  }
-}
-
 TEST(ParseArchString, RejectsDuplicateExtensionNames) {
   EXPECT_EQ(toString(RISCVISAInfo::parseArchString("rv64ii", true).takeError()),
             "invalid standard user-level extension 'i'");