[Patch, libstdc++/63497] Avoid dereferencing invalid iterator in regex_executor

[Jonathan Wakely]

On 21/10/14 09:45 -0700, Tim Shen wrote:
>On Tue, Oct 21, 2014 at 3:25 AM, Jonathan Wakely <jwakely@redhat.com> wrote:
>> Did you manage to produce a testcase that crashed on trunk?
>
>Oh I forgot to mention that I've tried my best to make a testcase that
>crash the trunk, but failed :).
>
>I'm not sure if I should directly put an assert in the code and make a
>testcase to explode it. Now I think it's better to do it.

Only if it's likely to catch problems in future. If you'd be putting
it in only to make a testcase fail then it's not worth it.

>> Is it really necessary to modify _M_current here?
>> Couldn't you do:
>>
>>       auto __pre = _M_current;
>>       if (_M_is_word(*--__pre))
>>         __left_is_word = true;
>>
>> Then the function could remain const, couldn't it?
>
>That's exactly what I did in the early version of this patch. But
>later I changed because I assume that copying an iterator is
>potentially expensive, but mutating is cheaper.

In general iterators are always passed by value and should be cheap to
copy. Inside regex the iterator is usually a const char* so is very
cheap to copy.

>Making this function const may bring some optimization, doesn't it?
>But I have no idea how much it will bring and if it's worthy.

It's unlikely (see http://www.gotw.ca/gotw/081.htm).

I just don't see the point in making it a non-const function just to
perform a micro-optimisation.

If you were passing an integer to a function would you do
  f(i-1);
or
  --i;
  f(i);
  ++i; 
?

The first form seems obviously better to me.

You could even simplify it further using std::prev:

      if (_M_is_word(*std::prev(__pre)))
        __left_is_word = true;