Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:48246
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: error (pb1.pair.com: domain juniper.net from 64.18.2.219 cause and error)
To: "internals@lists.php.net" <internals@lists.php.net>
Date: Thu, 6 May 2010 22:49:22 -0700
Thread-Topic: PHP  upgrade from 5.2.3 to 5.3.2 
Thread-Index: AcrqtH3PWY0xl7GuRIO0AC1W1xLVFwAAd4twAEg6cbAAAnpoYAA9R0gAADSDDSA=
Message-ID: <DBDEC20E6E29304D8AE5EAA493609B355743F9611E@EMBX02-HQ.jnpr.net>
Accept-Language: en-US
Content-Language: en-US
acceptlanguage: en-US
Content-Type: multipart/alternative;
	boundary="_000_DBDEC20E6E29304D8AE5EAA493609B355743F9611EEMBX02HQjnprn_"
MIME-Version: 1.0
Subject: PHP  upgrade from 5.2.3 to 5.3.2 
From: bnagwani@juniper.net (Bharat Nagwani)

--_000_DBDEC20E6E29304D8AE5EAA493609B355743F9611EEMBX02HQjnprn_
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

Hello php experts,

We are doing php upgrade from 5.2.3 to 5.3.2 (unix on Octeon/MIPS) and gett=
ing a compilation error

*       Issue with Octeon platform
o       Round() API throws "Maximum execution time exceeded" error.  Analyz=
ed the issue and Found the problem to be that "mult" function (src/dist/php=
/Zend/zend_strtod.c) is being called indefinitely from "pow5mult" function =
(src/dist/php/Zend/zend_strtod.c).  It might be some problem with byte alig=
nment on this platform.
*       We have filed a bug
o       http://bugs.php.net/bug.php?id=3D51701   -  "Maximum execution time=
 exceeded" error with round() API on octeon platform

Any idea about this


Here is a response from one of our platform developers after doing some ana=
lysis

In pow5mult():
------------------------
        for(;;) {
                if (k & 1) {
                        b1 =3D mult(b, p5);
                        Bfree(b);
                        b =3D b1;
                }
                if (!(k >>=3D 1)) {
                        break;
                }
                if (!(p51 =3D p5->next)) {
                        if (!(p51 =3D p5->next)) {
                                p51 =3D p5->next =3D mult(p5,p5);
                                p51->next =3D 0;
                        }
                }
                p5 =3D p51;
        }
-------------------

As it is suspected that mult is being called continuously from pow5mult: do=
 we know:
1.       How b and p5 are changing over each iteration?
2.      If the computation done by mult() in each of these iteration is cor=
rect or not?
3.      What is the impact of this computation on k?
4.      The loop will break only with change in k but I don't see k changin=
g here in mult5pow().

Can anyone throw some light on this?

Thanks


--_000_DBDEC20E6E29304D8AE5EAA493609B355743F9611EEMBX02HQjnprn_--