[IA] Recognize hexadecimal escape sequences

[void]

Summary

Implement support for hexadecimal escape sequences to match how GNU 'as'
handles them. I.e., read all hexadecimal characters and truncate to the
lower 16 bits.

[jcai19]

llvm/test/MC/AsmParser/directive_ascii.s

46 ↗

(On Diff #223283)

0x64 is smaller than 0xff. Maybe we can have a test case for values bigger than that?

[void]

Adding testcase for hex over 0xFF.

[jcai19]

Thank you for the fix! LGTM. Please give some time to @nickdesaulniers for comments.

[thakis]

I reverted this in r373898 since MC/AsmParser/directive_ascii.s failed on bots. I didn't look into it, but maybe it's because there's no bounds checking on the i + 1 index.

[void]

In D68483#1697207, @thakis wrote:

I reverted this in r373898 since MC/AsmParser/directive_ascii.s failed on bots. I didn't look into it, but maybe it's because there's no bounds checking on the i + 1 index.

Doh! Asserts weren't turned on in my build tree. Fixed in D68598. Very sorry for the breakage.

llvm/trunk/lib/MC/MCParser/AsmParser.cpp

bool AsmParser::parseEscapedString(std::string &Data) {
   StringRef Str = getTok().getStringContents();
   for (unsigned i = 0, e = Str.size(); i != e; ++i) {
     if (Str[i] != '\\') {
       Data += Str[i];
       continue;
     }
 
     // Recognize escaped characters. Note that this escape semantics currently
-    // loosely follows Darwin 'as'. Notably, it doesn't support hex escapes.
+    // loosely follows Darwin 'as'.
     ++i;
     if (i == e)
       return TokError("unexpected backslash at end of string");
 
+    // Recognize hex sequences similarly to GNU 'as'.
+    if (Str[i] == 'x' || Str[i] == 'X') {
+      if (!isHexDigit(Str[i + 1]))
+        return TokError("invalid hexadecimal escape sequence");
+
+      // Consume hex characters. GNU 'as' reads all hexadecimal characters and
+      // then truncates to the lower 16 bits. Seems reasonable.
+      unsigned Value = 0;
+      while (isHexDigit(Str[i + 1]))
+        Value = Value * 16 + hexDigitValue(Str[++i]);
+
+      Data += (unsigned char)(Value & 0xFF);
+      continue;
+    }
+
     // Recognize octal sequences.
     if ((unsigned)(Str[i] - '0') <= 7) {
       // Consume up to three octal characters.
       unsigned Value = Str[i] - '0';
 
       if (i + 1 != e && ((unsigned)(Str[i + 1] - '0')) <= 7) {
         ++i;
         Value = Value * 8 + (Str[i] - '0');

llvm/trunk/test/MC/AsmParser/directive_ascii.s

         
 # CHECK: TEST6:
 # CHECK: .byte 66
 # CHECK: .byte 0
 # CHECK: .byte 67
 # CHECK: .byte 0
 TEST6:
         .string "B", "C"
+
+# CHECK: TEST7:
+# CHECK: .ascii "dk"
+TEST7:
+        .ascii "\x64\Xa6B"