MIPS64: [regexp] do not assume short external strings have a minimum size.

Port 3518e492

Original commit message:
    Short external strings do not cache the resource data, and may be used
    for compressible strings. The assumptions about their lengths is
    invalid and may lead to oob reads.

R=bmeurer@chromium.org
BUG=v8:4923,chromium:604897
LOG=N

Review URL: https://codereview.chromium.org/1902393004

Cr-Commit-Position: refs/heads/master@{#35683}

src/mips64/code-stubs-mips64.cc

@@ -1685,48 +1685,53 @@ void RegExpExecStub::Generate(MacroAssembler* masm) {
   __ mov(a3, subject);  // Make a copy of the original subject string.
   __ ld(a0, FieldMemOperand(subject, HeapObject::kMapOffset));
   __ lbu(a0, FieldMemOperand(a0, Map::kInstanceTypeOffset));
+
   // subject: subject string
-  // a3: subject string
   // a0: subject string instance type
+  // a3: subject string
   // regexp_data: RegExp data (FixedArray)
   // Handle subject string according to its encoding and representation:
-  // (1) Sequential string?  If yes, go to (5).
-  // (2) Anything but sequential or cons?  If yes, go to (6).
-  // (3) Cons string.  If the string is flat, replace subject with first string.
-  //     Otherwise bailout.
-  // (4) Is subject external?  If yes, go to (7).
-  // (5) Sequential string.  Load regexp code according to encoding.
+  // (1) Sequential string?  If yes, go to (4).
+  // (2) Sequential or cons?  If not, go to (5).
+  // (3) Cons string.  If the string is flat, replace subject with first string
+  //     and go to (1). Otherwise bail out to runtime.
+  // (4) Sequential string.  Load regexp code according to encoding.
   // (E) Carry on.
   /// [...]
 
   // Deferred code at the end of the stub:
-  // (6) Not a long external string?  If yes, go to (8).
-  // (7) External string.  Make it, offset-wise, look like a sequential string.
-  //     Go to (5).
-  // (8) Short external string or not a string?  If yes, bail out to runtime.
-  // (9) Sliced string.  Replace subject with parent.  Go to (4).
-
-  Label check_underlying;   // (4)
-  Label seq_string;         // (5)
-  Label not_seq_nor_cons;   // (6)
-  Label external_string;    // (7)
-  Label not_long_external;  // (8)
-
-  // (1) Sequential string?  If yes, go to (5).
+  // (5) Long external string?  If not, go to (7).
+  // (6) External string.  Make it, offset-wise, look like a sequential string.
+  //     Go to (4).
+  // (7) Short external string or not a string?  If yes, bail out to runtime.
+  // (8) Sliced string.  Replace subject with parent.  Go to (1).
+
+  Label check_underlying;   // (1)
+  Label seq_string;         // (4)
+  Label not_seq_nor_cons;   // (5)
+  Label external_string;    // (6)
+  Label not_long_external;  // (7)
+
+  __ bind(&check_underlying);
+  __ ld(a2, FieldMemOperand(subject, HeapObject::kMapOffset));
+  __ Daddu(a0, a2, Map::kInstanceTypeOffset);
+  __ lbu(a0, MemOperand(a0));
+
+  // (1) Sequential string?  If yes, go to (4).
   __ And(a1,
          a0,
          Operand(kIsNotStringMask |
                  kStringRepresentationMask |
                  kShortExternalStringMask));
   STATIC_ASSERT((kStringTag | kSeqStringTag) == 0);
-  __ Branch(&seq_string, eq, a1, Operand(zero_reg));  // Go to (5).
+  __ Branch(&seq_string, eq, a1, Operand(zero_reg));  // Go to (4).
 
-  // (2) Anything but sequential or cons?  If yes, go to (6).
+  // (2) Sequential or cons?  If not, go to (5).
   STATIC_ASSERT(kConsStringTag < kExternalStringTag);
   STATIC_ASSERT(kSlicedStringTag > kExternalStringTag);
   STATIC_ASSERT(kIsNotStringMask > kExternalStringTag);
   STATIC_ASSERT(kShortExternalStringTag > kExternalStringTag);
-  // Go to (6).
+  // Go to (5).
   __ Branch(&not_seq_nor_cons, ge, a1, Operand(kExternalStringTag));
 
   // (3) Cons string.  Check that it's flat.
@@ -1735,16 +1740,9 @@ void RegExpExecStub::Generate(MacroAssembler* masm) {
   __ LoadRoot(a1, Heap::kempty_stringRootIndex);
   __ Branch(&runtime, ne, a0, Operand(a1));
   __ ld(subject, FieldMemOperand(subject, ConsString::kFirstOffset));
+  __ jmp(&check_underlying);
 
-  // (4) Is subject external?  If yes, go to (7).
-  __ bind(&check_underlying);
-  __ ld(a0, FieldMemOperand(subject, HeapObject::kMapOffset));
-  __ lbu(a0, FieldMemOperand(a0, Map::kInstanceTypeOffset));
-  STATIC_ASSERT(kSeqStringTag == 0);
-  __ And(at, a0, Operand(kStringRepresentationMask));
-  __ Branch(&external_string, ne, at, Operand(zero_reg));  // Go to (7).
-
-  // (5) Sequential string.  Load regexp code according to encoding.
+  // (4) Sequential string.  Load regexp code according to encoding.
   __ bind(&seq_string);
   // subject: sequential subject string (or look-alike, external string)
   // a3: original subject string
@@ -1987,12 +1985,12 @@ void RegExpExecStub::Generate(MacroAssembler* masm) {
   __ TailCallRuntime(Runtime::kRegExpExec);
 
   // Deferred code for string handling.
-  // (6) Not a long external string?  If yes, go to (8).
+  // (5) Long external string?  If not, go to (7).
   __ bind(&not_seq_nor_cons);
-  // Go to (8).
+  // Go to (7).
   __ Branch(&not_long_external, gt, a1, Operand(kExternalStringTag));
 
-  // (7) External string.  Make it, offset-wise, look like a sequential string.
+  // (6) External string.  Make it, offset-wise, look like a sequential string.
   __ bind(&external_string);
   __ ld(a0, FieldMemOperand(subject, HeapObject::kMapOffset));
   __ lbu(a0, FieldMemOperand(a0, Map::kInstanceTypeOffset));
@@ -2012,20 +2010,20 @@ void RegExpExecStub::Generate(MacroAssembler* masm) {
   __ Dsubu(subject,
           subject,
           SeqTwoByteString::kHeaderSize - kHeapObjectTag);
-  __ jmp(&seq_string);    // Go to (5).
+  __ jmp(&seq_string);  // Go to (4).
 
-  // (8) Short external string or not a string?  If yes, bail out to runtime.
+  // (7) Short external string or not a string?  If yes, bail out to runtime.
   __ bind(&not_long_external);
   STATIC_ASSERT(kNotStringTag != 0 && kShortExternalStringTag !=0);
   __ And(at, a1, Operand(kIsNotStringMask | kShortExternalStringMask));
   __ Branch(&runtime, ne, at, Operand(zero_reg));
 
-  // (9) Sliced string.  Replace subject with parent.  Go to (4).
+  // (8) Sliced string.  Replace subject with parent.  Go to (4).
   // Load offset into t0 and replace subject string with parent.
   __ ld(t0, FieldMemOperand(subject, SlicedString::kOffsetOffset));
   __ SmiUntag(t0);
   __ ld(subject, FieldMemOperand(subject, SlicedString::kParentOffset));
-  __ jmp(&check_underlying);  // Go to (4).
+  __ jmp(&check_underlying);  // Go to (1).
 #endif  // V8_INTERPRETED_REGEXP
 }
 

test/cctest/cctest.status

@@ -86,9 +86,6 @@
   'test-func-name-inference/UpperCaseClass': [FAIL],
   'test-func-name-inference/LowerCaseClass': [FAIL],
 
-  # BUG(4923). MIPS64 port is missing.
-  'test-regexp/UncachedExternalString': [PASS, ['arch==mips64 or arch==mips64el', FAIL]],
-
   ##############################################################################
   # TurboFan compiler failures.