Merged main:5e007afa9d4f into amd-gfx:4c17e1c71592

Local branch amd-gfx 4c17e1c Merged main:c00a708fc954 into amd-gfx:6f6131a974b2
Remote branch main 5e007af [AMDGPU] Handle hazard in v_scalef32_sr_fp4_* conversions (llvm#118589)

Author: SC llvm team

.ci/generate_test_report.py

@@ -5,6 +5,7 @@
 # python3 -m unittest discover -p generate_test_report.py
 
 import argparse
+import os
 import subprocess
 import unittest
 from io import StringIO
@@ -267,6 +268,46 @@ def test_report_dont_list_failures(self):
             ),
         )
 
+    def test_report_dont_list_failures_link_to_log(self):
+        self.assertEqual(
+            _generate_report(
+                "Foo",
+                [
+                    junit_from_xml(
+                        dedent(
+                            """\
+          <?xml version="1.0" encoding="UTF-8"?>
+          <testsuites time="0.02">
+          <testsuite name="Bar" tests="1" failures="1" skipped="0" time="0.02">
+          <testcase classname="Bar/test_1" name="test_1" time="0.02">
+            <failure><![CDATA[Output goes here]]></failure>
+          </testcase>
+          </testsuite>
+          </testsuites>"""
+                        )
+                    )
+                ],
+                list_failures=False,
+                buildkite_info={
+                    "BUILDKITE_ORGANIZATION_SLUG": "organization_slug",
+                    "BUILDKITE_PIPELINE_SLUG": "pipeline_slug",
+                    "BUILDKITE_BUILD_NUMBER": "build_number",
+                    "BUILDKITE_JOB_ID": "job_id",
+                },
+            ),
+            (
+                dedent(
+                    """\
+          # Foo
+
+          * 1 test failed
+
+          Failed tests and their output was too large to report. [Download](https://buildkite.com/organizations/organization_slug/pipelines/pipeline_slug/builds/build_number/jobs/job_id/download.txt) the build's log file to see the details."""
+                ),
+                "error",
+            ),
+        )
+
     def test_report_size_limit(self):
         self.assertEqual(
             _generate_report(
@@ -308,7 +349,13 @@ def test_report_size_limit(self):
 # listed. This minimal report will always fit into an annotation.
 # If include failures is False, total number of test will be reported but their names
 # and output will not be.
-def _generate_report(title, junit_objects, size_limit=1024 * 1024, list_failures=True):
+def _generate_report(
+    title,
+    junit_objects,
+    size_limit=1024 * 1024,
+    list_failures=True,
+    buildkite_info=None,
+):
     if not junit_objects:
         return ("", "success")
 
@@ -354,11 +401,21 @@ def plural(num_tests):
         report.append(f"* {tests_failed} {plural(tests_failed)} failed")
 
     if not list_failures:
+        if buildkite_info is not None:
+            log_url = (
+                "https://buildkite.com/organizations/{BUILDKITE_ORGANIZATION_SLUG}/"
+                "pipelines/{BUILDKITE_PIPELINE_SLUG}/builds/{BUILDKITE_BUILD_NUMBER}/"
+                "jobs/{BUILDKITE_JOB_ID}/download.txt".format(**buildkite_info)
+            )
+            download_text = f"[Download]({log_url})"
+        else:
+            download_text = "Download"
+
         report.extend(
             [
                 "",
                 "Failed tests and their output was too large to report. "
-                "Download the build's log file to see the details.",
+                f"{download_text} the build's log file to see the details.",
             ]
         )
     elif failures:
@@ -381,13 +438,23 @@ def plural(num_tests):
 
     report = "\n".join(report)
     if len(report.encode("utf-8")) > size_limit:
-        return _generate_report(title, junit_objects, size_limit, list_failures=False)
+        return _generate_report(
+            title,
+            junit_objects,
+            size_limit,
+            list_failures=False,
+            buildkite_info=buildkite_info,
+        )
 
     return report, style
 
 
-def generate_report(title, junit_files):
-    return _generate_report(title, [JUnitXml.fromfile(p) for p in junit_files])
+def generate_report(title, junit_files, buildkite_info):
+    return _generate_report(
+        title,
+        [JUnitXml.fromfile(p) for p in junit_files],
+        buildkite_info=buildkite_info,
+    )
 
 
 if __name__ == "__main__":
@@ -399,7 +466,18 @@ def generate_report(title, junit_files):
     parser.add_argument("junit_files", help="Paths to JUnit report files.", nargs="*")
     args = parser.parse_args()
 
-    report, style = generate_report(args.title, args.junit_files)
+    # All of these are required to build a link to download the log file.
+    env_var_names = [
+        "BUILDKITE_ORGANIZATION_SLUG",
+        "BUILDKITE_PIPELINE_SLUG",
+        "BUILDKITE_BUILD_NUMBER",
+        "BUILDKITE_JOB_ID",
+    ]
+    buildkite_info = {k: v for k, v in os.environ.items() if k in env_var_names}
+    if len(buildkite_info) != len(env_var_names):
+        buildkite_info = None
+
+    report, style = generate_report(args.title, args.junit_files, buildkite_info)
 
     if report:
         p = subprocess.Popen(

bolt/lib/Core/BinaryEmitter.cpp

@@ -46,13 +46,17 @@ BreakFunctionNames("break-funcs",
   cl::Hidden,
   cl::cat(BoltCategory));
 
-static cl::list<std::string>
-FunctionPadSpec("pad-funcs",
-  cl::CommaSeparated,
-  cl::desc("list of functions to pad with amount of bytes"),
-  cl::value_desc("func1:pad1,func2:pad2,func3:pad3,..."),
-  cl::Hidden,
-  cl::cat(BoltCategory));
+cl::list<std::string>
+    FunctionPadSpec("pad-funcs", cl::CommaSeparated,
+                    cl::desc("list of functions to pad with amount of bytes"),
+                    cl::value_desc("func1:pad1,func2:pad2,func3:pad3,..."),
+                    cl::Hidden, cl::cat(BoltCategory));
+
+cl::list<std::string> FunctionPadBeforeSpec(
+    "pad-funcs-before", cl::CommaSeparated,
+    cl::desc("list of functions to pad with amount of bytes"),
+    cl::value_desc("func1:pad1,func2:pad2,func3:pad3,..."), cl::Hidden,
+    cl::cat(BoltCategory));
 
 static cl::opt<bool> MarkFuncs(
     "mark-funcs",
@@ -70,11 +74,12 @@ X86AlignBranchBoundaryHotOnly("x86-align-branch-boundary-hot-only",
   cl::init(true),
   cl::cat(BoltOptCategory));
 
-size_t padFunction(const BinaryFunction &Function) {
+size_t padFunction(const cl::list<std::string> &Spec,
+                   const BinaryFunction &Function) {
   static std::map<std::string, size_t> FunctionPadding;
 
-  if (FunctionPadding.empty() && !FunctionPadSpec.empty()) {
-    for (std::string &Spec : FunctionPadSpec) {
+  if (FunctionPadding.empty() && !Spec.empty()) {
+    for (const std::string &Spec : Spec) {
       size_t N = Spec.find(':');
       if (N == std::string::npos)
         continue;
@@ -319,6 +324,32 @@ bool BinaryEmitter::emitFunction(BinaryFunction &Function,
     Streamer.emitCodeAlignment(Function.getAlign(), &*BC.STI);
   }
 
+  if (size_t Padding =
+          opts::padFunction(opts::FunctionPadBeforeSpec, Function)) {
+    // Handle padFuncsBefore after the above alignment logic but before
+    // symbol addresses are decided.
+    if (!BC.HasRelocations) {
+      BC.errs() << "BOLT-ERROR: -pad-before-funcs is not supported in "
+                << "non-relocation mode\n";
+      exit(1);
+    }
+
+    // Preserve Function.getMinAlign().
+    if (!isAligned(Function.getMinAlign(), Padding)) {
+      BC.errs() << "BOLT-ERROR: user-requested " << Padding
+                << " padding bytes before function " << Function
+                << " is not a multiple of the minimum function alignment ("
+                << Function.getMinAlign().value() << ").\n";
+      exit(1);
+    }
+
+    LLVM_DEBUG(dbgs() << "BOLT-DEBUG: padding before function " << Function
+                      << " with " << Padding << " bytes\n");
+
+    // Since the padding is not executed, it can be null bytes.
+    Streamer.emitFill(Padding, 0);
+  }
+
   MCContext &Context = Streamer.getContext();
   const MCAsmInfo *MAI = Context.getAsmInfo();
 
@@ -373,7 +404,7 @@ bool BinaryEmitter::emitFunction(BinaryFunction &Function,
   emitFunctionBody(Function, FF, /*EmitCodeOnly=*/false);
 
   // Emit padding if requested.
-  if (size_t Padding = opts::padFunction(Function)) {
+  if (size_t Padding = opts::padFunction(opts::FunctionPadSpec, Function)) {
     LLVM_DEBUG(dbgs() << "BOLT-DEBUG: padding function " << Function << " with "
                       << Padding << " bytes\n");
     Streamer.emitFill(Padding, MAI->getTextAlignFillValue());

bolt/lib/Passes/ReorderFunctions.cpp

@@ -28,7 +28,9 @@ extern cl::OptionCategory BoltOptCategory;
 extern cl::opt<unsigned> Verbosity;
 extern cl::opt<uint32_t> RandomSeed;
 
-extern size_t padFunction(const bolt::BinaryFunction &Function);
+extern size_t padFunction(const cl::list<std::string> &Spec,
+                          const bolt::BinaryFunction &Function);
+extern cl::list<std::string> FunctionPadSpec, FunctionPadBeforeSpec;
 
 extern cl::opt<bolt::ReorderFunctions::ReorderType> ReorderFunctions;
 cl::opt<bolt::ReorderFunctions::ReorderType> ReorderFunctions(
@@ -304,8 +306,12 @@ Error ReorderFunctions::runOnFunctions(BinaryContext &BC) {
                           return false;
                         if (B->isIgnored())
                           return true;
-                        const size_t PadA = opts::padFunction(*A);
-                        const size_t PadB = opts::padFunction(*B);
+                        const size_t PadA =
+                            opts::padFunction(opts::FunctionPadSpec, *A) +
+                            opts::padFunction(opts::FunctionPadBeforeSpec, *A);
+                        const size_t PadB =
+                            opts::padFunction(opts::FunctionPadSpec, *B) +
+                            opts::padFunction(opts::FunctionPadBeforeSpec, *B);
                         if (!PadA || !PadB) {
                           if (PadA)
                             return true;

bolt/test/AArch64/pad-before-funcs.s

@@ -0,0 +1,29 @@
+# Test checks that --pad-before-funcs is working as expected.
+# It should be able to introduce a configurable offset for the _start symbol.
+# It should reject requests which don't obey the code alignment requirement.
+
+# RUN: llvm-mc -filetype=obj -triple aarch64-unknown-unknown %s -o %t.o
+# RUN: %clang %cflags %t.o -o %t.exe -Wl,-q -Wl,--section-start=.text=0x4000
+# RUN: llvm-bolt %t.exe -o %t.bolt.0 --pad-funcs-before=_start:0
+# RUN: llvm-bolt %t.exe -o %t.bolt.4 --pad-funcs-before=_start:4
+# RUN: llvm-bolt %t.exe -o %t.bolt.8 --pad-funcs-before=_start:8
+
+# RUN: not llvm-bolt %t.exe -o %t.bolt.8 --pad-funcs-before=_start:1 2>&1 | FileCheck --check-prefix=CHECK-BAD-ALIGN %s
+
+# CHECK-BAD-ALIGN: user-requested 1 padding bytes before function _start(*2) is not a multiple of the minimum function alignment (4).
+
+# RUN: llvm-objdump --section=.text --disassemble %t.bolt.0 | FileCheck --check-prefix=CHECK-0 %s
+# RUN: llvm-objdump --section=.text --disassemble %t.bolt.4 | FileCheck --check-prefix=CHECK-4 %s
+# RUN: llvm-objdump --section=.text --disassemble %t.bolt.8 | FileCheck --check-prefix=CHECK-8 %s
+
+# Trigger relocation mode in bolt.
+.reloc 0, R_AARCH64_NONE
+
+.section .text
+.globl _start
+
+# CHECK-0: 0000000000400000 <_start>
+# CHECK-4: 0000000000400004 <_start>
+# CHECK-8: 0000000000400008 <_start>
+_start:
+    ret

clang-tools-extra/clang-tidy/bugprone/CMakeLists.txt

@@ -16,6 +16,7 @@ add_clang_library(clangTidyBugproneModule STATIC
   ChainedComparisonCheck.cpp
   ComparePointerToMemberVirtualFunctionCheck.cpp
   CopyConstructorInitCheck.cpp
+  CrtpConstructorAccessibilityCheck.cpp
   DanglingHandleCheck.cpp
   DynamicStaticInitializersCheck.cpp
   EasilySwappableParametersCheck.cpp
@@ -26,11 +27,8 @@ add_clang_library(clangTidyBugproneModule STATIC
   ForwardingReferenceOverloadCheck.cpp
   ImplicitWideningOfMultiplicationResultCheck.cpp
   InaccurateEraseCheck.cpp
-  IncorrectEnableIfCheck.cpp
-  ReturnConstRefFromParameterCheck.cpp
-  SuspiciousStringviewDataUsageCheck.cpp
-  SwitchMissingDefaultCaseCheck.cpp
   IncDecInConditionsCheck.cpp
+  IncorrectEnableIfCheck.cpp
   IncorrectRoundingsCheck.cpp
   InfiniteLoopCheck.cpp
   IntegerDivisionCheck.cpp
@@ -45,15 +43,16 @@ add_clang_library(clangTidyBugproneModule STATIC
   MultipleNewInOneExpressionCheck.cpp
   MultipleStatementMacroCheck.cpp
   NoEscapeCheck.cpp
-  NondeterministicPointerIterationOrderCheck.cpp
   NonZeroEnumToBoolConversionCheck.cpp
+  NondeterministicPointerIterationOrderCheck.cpp
   NotNullTerminatedResultCheck.cpp
   OptionalValueConversionCheck.cpp
   ParentVirtualCallCheck.cpp
   PointerArithmeticOnPolymorphicObjectCheck.cpp
   PosixReturnCheck.cpp
   RedundantBranchConditionCheck.cpp
   ReservedIdentifierCheck.cpp
+  ReturnConstRefFromParameterCheck.cpp
   SharedPtrArrayMismatchCheck.cpp
   SignalHandlerCheck.cpp
   SignedCharMisuseCheck.cpp
@@ -74,7 +73,9 @@ add_clang_library(clangTidyBugproneModule STATIC
   SuspiciousReallocUsageCheck.cpp
   SuspiciousSemicolonCheck.cpp
   SuspiciousStringCompareCheck.cpp
+  SuspiciousStringviewDataUsageCheck.cpp
   SwappedArgumentsCheck.cpp
+  SwitchMissingDefaultCaseCheck.cpp
   TaggedUnionMemberCountCheck.cpp
   TerminatingContinueCheck.cpp
   ThrowKeywordMissingCheck.cpp
@@ -85,7 +86,6 @@ add_clang_library(clangTidyBugproneModule STATIC
   UnhandledExceptionAtNewCheck.cpp
   UnhandledSelfAssignmentCheck.cpp
   UniquePtrArrayMismatchCheck.cpp
-  CrtpConstructorAccessibilityCheck.cpp
   UnsafeFunctionsCheck.cpp
   UnusedLocalNonTrivialVariableCheck.cpp
   UnusedRaiiCheck.cpp