> You say:
>> This is the indentation in c-mode (it is the same in php-mode):
>>
>> <?   php
>> void foo() {
>> 	bar();
>>      }
>> ?>
>
> In that little code snippet, could you please do C-u C-c C-s on each
> line, to dump CC Mode's idea of each line's syntax, then post the
> results.

Here it is:

  <?php				/* ((topmost-intro 1)) */
  void foo () {			/* ((topmost-intro-cont 1)) */
    bar ();			/* ((defun-block-intro 1)) */
    }				/* ((defun-close 3)) */
  ?>				/* ((topmost-intro 125)) */

And the flawless one:

  <?php				/* ((topmost-intro 1)) */
  void foo ()			/* ((topmost-intro-cont 1)) */
  {				/* ((defun-open 1)) */
    bar ();			/* ((defun-block-intro 81)) */
  }				/* ((defun-close 81)) */
  ?>				/* ((topmost-intro 152)) */

>> I am not sure if it's a bug or the ancient php-mode just needs to
>> accommodate for the new emacs.
>
> "<" and ">" are nasty tokens in C-like languages.  They can appear as
> "i < 10", "#include <stdio.h>" and in C++ templates.  It could be that
> the PHP "<?" is getting confused with one of these other uses.

I think that sound reasonable.

> Looking forward to hearing from you again,

Glad if I can help.

	Thomas