mysqli flaky test: ghsa-h35g-vwh6-m678-stmt-row-string

[morrisonlevi]

Description

This newish test is flaky:

--
     [*] Sending - Malicious Stmt Response for items [Extract heap through buffer over-read]: 01000001013000000203646566087068705f74657374056974656d73056974656d73046974656d046974656d0ce000c8000000fd011000000005000003fe00002200070000040000fa7465737405000005fe00002200
     
     Warning: mysqli_result::fetch_assoc(): Malformed server packet. Field length pointing after the end of packet in %s on line %d
013- [*] Received: 0500000019010000000100000001
013+ [*] Received: 050000001901000000
     [*] Server finished
     done!
========DONE========
FAIL GHSA-h35g-vwh6-m678 (mysqlnd leaks partial content of the heap - stmt row string buffer over-read) [ext/mysqli/tests/ghsa-h35g-vwh6-m678-stmt-row-string.phpt] 

The Received line comes from this code:

    public function read($bytes_len = 1024)
    {
        // wait 20ms to fill the buffer
        usleep(20000);
        $data = fread($this->conn, $bytes_len);
        if ($data) {
            fprintf(STDERR, "[*] Received: %s\n", bin2hex($data));
        }
    }

It seems as if there's a partial read of the data.

I don't have time right now to dig into the root and try and fix it. If no one else does either, we can at least add:

--FLAKY--
Sometimes get a partial read like:
013- [*] Received: 0500000019010000000100000001
013+ [*] Received: 050000001901000000

PHP Version

PHP 8.4

Operating System

No response

[bukka]

So this happens when the stream gets closed before it process the confirmation about peer side closing. I fixed the timing already before but seems like it's still a bit flaky. That part is not really important for the test so I decided to make it optional which should fix it hopefully.