PCRE_UTF8 modifier causes "Warning: Unknown error" on first run

[atdt]

To reproduce:

<?php preg_split('//u', 'x'); exit(error_get_last() ? 1 : 0);

The first time this is run, the output is:

Warning: Unknown error in /home/ori/bug66386.php on line 1
    #0 at [/home/ori/bug66386.php:1]

There is no error on subsequent invocations. If I make a modification to the file (including adding whitespace), the error recurs. So I imagine it is an artifact of compilation.

[ptarjan]

I can't repo with my version of preg.

I'm assuming it comes from this line:

hhvm/hphp/runtime/base/preg.cpp

Line 1360 in 1ecb444

raise_warning("Unknown error");

(you can use gdb and set a breakpoint on raise_error to make sure).

And yes, it looks like the regex cache is kicking in? I don't have much... is this blocking something?

[atdt]

Yep, it's a blocker, and yes, that's the line. The value returned from pcre_exec is -21, which is:

PCRE_ERROR_RECURSIONLIMIT (-21)

       The internal recursion limit, as specified by the match_limit_recursion
       field in a pcre_extra structure (or defaulted)  was  reached.  See  the
       description above.