llvm-addr2line: assume addresses on the command line are hexadecimal rather than attempting to guess the base based on the form of the number.

Summary:
This matches the behavior of GNU addr2line. We previously treated
hexadecimal addresses as binary if they started with 0b, otherwise as
octal if they started with 0, otherwise as decimal.

This only affects llvm-addr2line; the behavior of llvm-symbolize is
unaffected.

Reviewers: ikudrin, rupprecht, jhenderson

Subscribers: llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D73306

docs/CommandGuide/llvm-addr2line.rst

@@ -17,6 +17,11 @@ GNU's :program:`addr2line`.
 
 Here are some of those differences:
 
+-  ``llvm-addr2line`` interprets all addresses as hexadecimal and ignores an
+   optional ``0x`` prefix, whereas ``llvm-symbolizer`` attempts to determine
+   the base from the literal's prefix and defaults to decimal if there is no
+   prefix.
+
 -  ``llvm-addr2line`` defaults not to print function names. Use `-f`_ to enable
    that.
 

test/lit.cfg.py

@@ -146,10 +146,10 @@ tools = [
 # FIXME: Why do we have both `lli` and `%lli` that do slightly different things?
 tools.extend([
     'dsymutil', 'lli', 'lli-child-target', 'llvm-ar', 'llvm-as',
-    'llvm-bcanalyzer', 'llvm-config', 'llvm-cov', 'llvm-cxxdump', 'llvm-cvtres',
+    'llvm-addr2line', 'llvm-bcanalyzer', 'llvm-config', 'llvm-cov',
-    'llvm-diff', 'llvm-dis', 'llvm-dwarfdump', 'llvm-exegesis', 'llvm-extract',
+    'llvm-cxxdump', 'llvm-cvtres', 'llvm-diff', 'llvm-dis', 'llvm-dwarfdump',
-    'llvm-isel-fuzzer', 'llvm-ifs', 'llvm-install-name-tool',
+    'llvm-exegesis', 'llvm-extract', 'llvm-isel-fuzzer', 'llvm-ifs',
-    'llvm-jitlink', 'llvm-opt-fuzzer', 'llvm-lib',
+    'llvm-install-name-tool', 'llvm-jitlink', 'llvm-opt-fuzzer', 'llvm-lib',
     'llvm-link', 'llvm-lto', 'llvm-lto2', 'llvm-mc', 'llvm-mca',
     'llvm-modextract', 'llvm-nm', 'llvm-objcopy', 'llvm-objdump',
     'llvm-pdbutil', 'llvm-profdata', 'llvm-ranlib', 'llvm-rc', 'llvm-readelf',

test/tools/llvm-symbolizer/input-base.test

@@ -0,0 +1,33 @@
+# llvm-symbolizer infers the number base from the form of the address.
+RUN: llvm-symbolizer -e /dev/null -a 0x1234 | FileCheck %s
+RUN: llvm-symbolizer -e /dev/null -a 0X1234 | FileCheck %s
+RUN: llvm-symbolizer -e /dev/null -a 4660 | FileCheck %s
+RUN: llvm-symbolizer -e /dev/null -a 011064 | FileCheck %s
+RUN: llvm-symbolizer -e /dev/null -a 0b1001000110100 | FileCheck %s
+RUN: llvm-symbolizer -e /dev/null -a 0B1001000110100 | FileCheck %s
+RUN: llvm-symbolizer -e /dev/null -a 0o11064 | FileCheck %s
+
+# llvm-symbolizer / StringRef::getAsInteger only accepts the 0o prefix in lowercase.
+RUN: llvm-symbolizer -e /dev/null -a 0O1234 | FileCheck %s --check-prefix=INVALID-NOT-OCTAL-UPPER
+
+# llvm-addr2line always requires hexadecimal, but accepts an optional 0x prefix.
+RUN: llvm-addr2line -e /dev/null -a 0x1234 | FileCheck %s
+RUN: llvm-addr2line -e /dev/null -a 0X1234 | FileCheck %s
+RUN: llvm-addr2line -e /dev/null -a 1234 | FileCheck %s
+RUN: llvm-addr2line -e /dev/null -a 01234 | FileCheck %s
+RUN: llvm-addr2line -e /dev/null -a 0b1010 | FileCheck %s --check-prefix=HEXADECIMAL-NOT-BINARY
+RUN: llvm-addr2line -e /dev/null -a 0B1010 | FileCheck %s --check-prefix=HEXADECIMAL-NOT-BINARY
+RUN: llvm-addr2line -e /dev/null -a 0o1234 | FileCheck %s --check-prefix=INVALID-NOT-OCTAL-LOWER
+RUN: llvm-addr2line -e /dev/null -a 0O1234 | FileCheck %s --check-prefix=INVALID-NOT-OCTAL-UPPER
+
+CHECK: 0x1234
+CHECK-NEXT: ??
+
+HEXADECIMAL-NOT-BINARY: 0xb1010
+HEXADECIMAL-NOT-BINARY: ??
+
+INVALID-NOT-OCTAL-LOWER: 0o1234
+INVALID-NOT-OCTAL-LOWER-NOT: ??
+
+INVALID-NOT-OCTAL-UPPER: 0O1234
+INVALID-NOT-OCTAL-UPPER-NOT: ??

tools/llvm-symbolizer/llvm-symbolizer.cpp

@@ -181,7 +181,7 @@ enum class Command {
   Frame,
 };
 
-static bool parseCommand(StringRef InputString, Command &Cmd,
+static bool parseCommand(bool IsAddr2Line, StringRef InputString, Command &Cmd,
                          std::string &ModuleName, uint64_t &ModuleOffset) {
   const char kDelimiters[] = " \n\r";
   ModuleName = "";
@@ -218,15 +218,21 @@ static bool parseCommand(StringRef InputString, Command &Cmd,
   // Skip delimiters and parse module offset.
   Pos += strspn(Pos, kDelimiters);
   int OffsetLength = strcspn(Pos, kDelimiters);
-  return !StringRef(Pos, OffsetLength).getAsInteger(0, ModuleOffset);
+  StringRef Offset(Pos, OffsetLength);
+  // GNU addr2line assumes the offset is hexadecimal and allows a redundant
+  // "0x" or "0X" prefix; do the same for compatibility.
+  if (IsAddr2Line)
+    Offset.consume_front("0x") || Offset.consume_front("0X");
+  return !Offset.getAsInteger(IsAddr2Line ? 16 : 0, ModuleOffset);
 }
 
-static void symbolizeInput(StringRef InputString, LLVMSymbolizer &Symbolizer,
+static void symbolizeInput(bool IsAddr2Line, StringRef InputString,
-                           DIPrinter &Printer) {
+                           LLVMSymbolizer &Symbolizer, DIPrinter &Printer) {
   Command Cmd;
   std::string ModuleName;
   uint64_t Offset = 0;
-  if (!parseCommand(StringRef(InputString), Cmd, ModuleName, Offset)) {
+  if (!parseCommand(IsAddr2Line, StringRef(InputString), Cmd, ModuleName,
+                    Offset)) {
     outs() << InputString << "\n";
     return;
   }
@@ -340,12 +346,12 @@ int main(int argc, char **argv) {
           std::remove_if(StrippedInputString.begin(), StrippedInputString.end(),
                          [](char c) { return c == '\r' || c == '\n'; }),
           StrippedInputString.end());
-      symbolizeInput(StrippedInputString, Symbolizer, Printer);
+      symbolizeInput(IsAddr2Line, StrippedInputString, Symbolizer, Printer);
       outs().flush();
     }
   } else {
     for (StringRef Address : ClInputAddresses)
-      symbolizeInput(Address, Symbolizer, Printer);
+      symbolizeInput(IsAddr2Line, Address, Symbolizer, Printer);
   }
 
   return 0;