#39221 [Com]: php5ts.dll faulting every few minutes on apache2.2
ID: 39221 Comment by: niels at keurentjes dot net Reported By: kris at k-software dot org Status: No Feedback Bug Type: Apache2 related Operating System: windows server 2003 x64 PHP Version: 5CVS-2006-10-20 (snap) New Comment: I'm still having the problems, albeit less frequently after some performance tuning, but still too often. In my case, it's mainly some Ajax pages, inherently causing a lot of pageviews, that trigger the problem. I'm not in a position to run snapshots on the involved production machines, but I'm highly curious whether the problem was really attacked and or fixed now. Previous Comments: [2007-03-27 16:23:53] php at blazemonger dot com The problem appears to be gone in today's latest snapshot of PHP 5.2. I hammered MediaWiki for 15-20 minutes and didn't see a single error. [2007-03-27 13:46:52] php at blazemonger dot com Following up my previous comment: this is easily reproduced if I set up 3-4 scripts to simultaneously pound MediaWiki with requests to the Special:Random page, e.g., # repeat 1000 times wget -O - http://servername/wiki/Special:Random I have reproduced this on both the 32-bit and 64-bit versions of Windows 2003 Server SP1, with the software versions given earlier. [2007-03-25 03:02:40] php at blazemonger dot com I can reproduce this PHP crash easily in our MediaWiki 1.9.3 installation. Visit the main page and click the "Random page" link (in the left navigation menu) repeatedly, as fast as you can, about 50 times. The error is thrown into the event log: Faulting application httpd.exe, version 2.2.4.0, faulting module php5ts.dll, version 5.2.1.1, fault address 0xae66. and Apache becomes completely unresponsive (will not respond to any page requests) and must be restarted. Windows 2003 Server SP1 64-bit (2 GB RAM) Dual core AMD Opteron (2.81 GHz, Processor 8220 SE) Apache 2.2.4 with SSL PHP 5.2.1 (apache2handler) MediaWiki 1.9.3 MySQL 5.0.27-community-nt This is a VMware virtual machine if that matters. ---- [2007-02-16 11:01:41] niels at keurentjes dot net I have the exact same issue: a stock Win2k3 setup with Apache 2.2.4 and PHP 5.2.1 on a Dual Xeon HT setup, and regular crashes, several times per day, sometimes 15 in 30 minutes when the site is being hit severely. I have performed several recent software upgrades, the problems also occurred in the exact same fasion in Apache 2.0.59 + PHP 5.1.4, Apache 2.0.59 + PHP 5.2.0 and Apache 2.2.4 + PHP 5.2.1. I am currently testing whether hard-forcing uniprocessor operation using ImageCfg (http://www.robpol86.com/Pages/imagecfg.php) works. I have set the httpd.exe process to the following: --- httpd.exe contains the following configuration information: Process Affinity Mask: 0001 httpd.exe contains a Subsystem Version of 4.0 Image can only run in uni-processor mode on multi-processor systems --- Notably the crashes seemed to become less frequent when I disabled all caching on the site. At first I suspected this pointed towards problems with too many files open simultaneously, but I am now after reading these problems pretty sure that it merely slowed down the site enough not to trigger the synchronization issues or race conditions. [2007-01-04 04:05:22] judexidoneus_g6 at hotmail dot com So, there is no solution for this bug? The remainder of the comments for this report are too long. To view the rest of the comments, please view the bug report online at http://bugs.php.net/39221 -- Edit this bug report at http://bugs.php.net/?id=39221&edit=1
#39221 [Com]: php5ts.dll faulting every few minutes on apache2.2
ID: 39221 Comment by: niels at keurentjes dot net Reported By: kris at k-software dot org Status: No Feedback Bug Type: Apache2 related Operating System: windows server 2003 x64 PHP Version: 5CVS-2006-10-20 (snap) New Comment: I have the exact same issue: a stock Win2k3 setup with Apache 2.2.4 and PHP 5.2.1 on a Dual Xeon HT setup, and regular crashes, several times per day, sometimes 15 in 30 minutes when the site is being hit severely. I have performed several recent software upgrades, the problems also occurred in the exact same fasion in Apache 2.0.59 + PHP 5.1.4, Apache 2.0.59 + PHP 5.2.0 and Apache 2.2.4 + PHP 5.2.1. I am currently testing whether hard-forcing uniprocessor operation using ImageCfg (http://www.robpol86.com/Pages/imagecfg.php) works. I have set the httpd.exe process to the following: --- httpd.exe contains the following configuration information: Process Affinity Mask: 0001 httpd.exe contains a Subsystem Version of 4.0 Image can only run in uni-processor mode on multi-processor systems --- Notably the crashes seemed to become less frequent when I disabled all caching on the site. At first I suspected this pointed towards problems with too many files open simultaneously, but I am now after reading these problems pretty sure that it merely slowed down the site enough not to trigger the synchronization issues or race conditions. Previous Comments: [2007-01-04 04:05:22] judexidoneus_g6 at hotmail dot com So, there is no solution for this bug? [2006-11-07 07:44:44] paradiso at gmail dot com This is really a bug. on my p4 2.4c + 865G + 1G ram, win2k3 server, apache 2.2.3.0 & php 5.2, this happens all the time. [2006-10-31 01:00:01] php-bugs at lists dot php dot net No feedback was provided for this bug for over a week, so it is being suspended automatically. If you are able to provide the information that was originally requested, please do so and change the status of the bug back to "Open". [2006-10-22 14:14:23] kris at k-software dot org Error in CPU IDs in previous comment, from microsoft support article ID 252867: Processor enumeration on computers that use hyperthreading first assign processor numbers to the primary logical processor for each processor and then assign numbers to the secondary. For example, for dual physical processor computers with hyperthreading, the first processor has logical processor 0 and 2, and the second processor has logical processor 1 and 3. [2006-10-22 13:59:13] kris at k-software dot org It appears that setting the CPU affinity of the apache http process to a single physical processor stops the crashing... Setting the server to NOT use the second hyperthreaded core of the xeons (i.e. CPUs 0+2 or 1+3) seems to solve this random crashing. Will test with a prefork MPM. The remainder of the comments for this report are too long. To view the rest of the comments, please view the bug report online at http://bugs.php.net/39221 -- Edit this bug report at http://bugs.php.net/?id=39221&edit=1
#32442 [NEW]: serialize/unserialize floats with de-locales
From: niels at monarch dot de Operating system: Linux PHP version: 4.3.10 PHP Bug Type: Variables related Bug description: serialize/unserialize floats with de-locales Description: When unserializing a previously serialized float that contains a exponent-expressed value (ie. 5Mrd), the result is wrong when locale is set to setlocale(LC_ALL,"de_DE"); (with compiled de-locales, of course) With the locales set to "C", everything works fine. In my opinion, this is a ',','.'-number-format-handling error. Reproduce code: --- setlocale (LC_ALL,"de_DE"); echo "test:".(float)3.14152; $fl1=(float)50; // 5Mrd $str=serialize ($fl1); $fl2=unserialize($str); var_dump ($fl1); var_dump ($fl2); phpinfo(); Expected result: test:3,14152 float(5.0E+9) float(5.0E+9) Actual result: -- test:3,14152 (OK) float(5.0E+9) float(5) (hrmm, not quite!) -- Edit bug report at http://bugs.php.net/?id=32442&edit=1 -- Try a CVS snapshot (php4): http://bugs.php.net/fix.php?id=32442&r=trysnapshot4 Try a CVS snapshot (php5.0): http://bugs.php.net/fix.php?id=32442&r=trysnapshot50 Try a CVS snapshot (php5.1): http://bugs.php.net/fix.php?id=32442&r=trysnapshot51 Fixed in CVS:http://bugs.php.net/fix.php?id=32442&r=fixedcvs Fixed in release:http://bugs.php.net/fix.php?id=32442&r=alreadyfixed Need backtrace: http://bugs.php.net/fix.php?id=32442&r=needtrace Need Reproduce Script: http://bugs.php.net/fix.php?id=32442&r=needscript Try newer version: http://bugs.php.net/fix.php?id=32442&r=oldversion Not developer issue: http://bugs.php.net/fix.php?id=32442&r=support Expected behavior: http://bugs.php.net/fix.php?id=32442&r=notwrong Not enough info: http://bugs.php.net/fix.php?id=32442&r=notenoughinfo Submitted twice: http://bugs.php.net/fix.php?id=32442&r=submittedtwice register_globals:http://bugs.php.net/fix.php?id=32442&r=globals PHP 3 support discontinued: http://bugs.php.net/fix.php?id=32442&r=php3 Daylight Savings:http://bugs.php.net/fix.php?id=32442&r=dst IIS Stability: http://bugs.php.net/fix.php?id=32442&r=isapi Install GNU Sed: http://bugs.php.net/fix.php?id=32442&r=gnused Floating point limitations: http://bugs.php.net/fix.php?id=32442&r=float No Zend Extensions: http://bugs.php.net/fix.php?id=32442&r=nozend MySQL Configuration Error: http://bugs.php.net/fix.php?id=32442&r=mysqlcfg
#28209 [Com]: strtotime("now")
ID: 28209
Comment by: niels dot hansen at pola dot dk
Reported By: gamin at centras dot lt
Status: Assigned
Bug Type: Date/time related
Operating System: WinXP HE SP1
PHP Version: 5.0.0RC2
Assigned To: derick
New Comment:
The functionality is maintained in RC3.
I needs the realtime for logging of process. My workaround is to do
this:
$str=date('d F Y h:i:s A');
$timestamp = strtotime("$str");
Previous Comments:
[2004-06-15 12:42:47] wb at pro-net dot co dot uk
So how would one, for example, use strtotime() to get the current time
+ 10 mins?
[2004-05-19 12:33:35] [EMAIL PROTECTED]
That's feasible for the rewrite, but for the current parser.
[2004-05-19 12:10:35] wb at pro-net dot co dot uk
Just thought that i would point out that in PHP 4 now is real now and
not GNU now so form a BC point of view i think now should be real now.
[2004-05-19 12:10:17] wb at pro-net dot co dot uk
Just thought that i would point out that in PHP 4 now is real now and
not GNU now so form a BC point of view i think now should be real now.
[2004-05-19 12:04:25] wb at pro-net dot co dot uk
Hi derick,
Yes that is strange, What about adding an optional parameter to the
function to set toggle between real now and gnu now?
Something like:
int strtotime ( string time [, int now] [, bool gnu = false])
The remainder of the comments for this report are too long. To view
the rest of the comments, please view the bug report online at
http://bugs.php.net/28209
--
Edit this bug report at http://bugs.php.net/?id=28209&edit=1
#17115 [Com]: lambda functions produce segfault with static vars
ID: 17115
Comment by: [EMAIL PROTECTED]
Reported By: [EMAIL PROTECTED]
Status: No Feedback
Bug Type: Reproducible crash
Operating System: Redhat 7.1
PHP Version: 4.2.0
New Comment:
I see the same bug in 4.2.3. I haven't tried "latest".
OS: Debian testing/unstable
Configure Command: '../configure' '--prefix=/usr'
'--prefix=/usr' '--with-regex=php'
'--enable-force-cgi-redirect'
'--with-config-file-path=/etc/php4/cgi' '--disable-rpath'
'--disable-pear' '--disable-debug' '--enable-memory-limit'
'--with-layout=GNU' '--enable-calendar' '--enable-sysvsem'
'--enable-sysvshm' '--enable-track-vars'
'--enable-trans-sid' '--enable-bcmath' '--with-bz2'
'--enable-ctype' '--with-db2' '--with-iconv'
'--enable-exif' '--enable-filepro' '--enable-ftp'
'--with-gettext' '--enable-mbstring'
'--with-pcre-regex=/usr' '--enable-shmop'
'--enable-sockets' '--enable-wddx' '--disable-xml'
'--with-expat-dir=/usr' '--enable-yp' '--with-zlib'
'--without-pgsql' '--with-openssl=/usr' '--disable-static'
'--without-mm' '--without-mysql' '--without-sybase-ct'
$ sudo apt-get install php4/unstable
Reading Package Lists... Done
Building Dependency Tree... Done
Selected version 4:4.2.3-3 (Debian:unstable) for php4
Sorry, php4 is already the newest version.
0 packages upgraded, 0 newly installed, 0 to remove and 0
not upgraded.
$ php4 -v
4.2.3
$ php4
^DSegmentation fault
$ _
I get similar results when running php under Apache.
Error.log says: "child pid 12848 exit signal Segmentation
fault (11)"
Previous Comments:
[2002-07-18 01:00:11] [EMAIL PROTECTED]
No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".
[2002-06-17 21:09:48] [EMAIL PROTECTED]
Neither of those examples crash with this snapshot:
http://snaps.php.net/php4-latest.tar.gz
[2002-05-09 07:41:59] [EMAIL PROTECTED]
Code is -
When run from the command-line this core-dumps under 4.2.0. Running
4.1.1 compiled with --apxs for apache produces incorrect results -
$s is one []
$s is two []
$s is three []
But if used in array_map() different results are produced (still using
4.1.1)
code -
", $f("two"), "", $f("three"), "";
echo "";
echo join(" | ", array_map($f, array("abc","def","ghi"));
?>
result -
$s is one []
$s is two []
$s is three []
$s is abc [] | $s is def [1] | $s is ghi [2]
But take you the first three calls to $f and the page never returns.
And there are many more bizarre results like 4.1.1 (but there is always
a coredump in 4.2.0).
Details
configure line (4.2.0) -
'./configure' '--enable-cli' '--disable-short-tags'
'--enable-overload' '--enable-sockets' '--with-readline'
configure line (4.1.1) -
'./configure' '--with-apache=../apache_1.3.20' '--with-gd'
'--with-mysql=/usr/local/mysql' '--enable-debug=no'
'--enable-track-vars=yes' '--enable-magic-quotes=yes' '--with-gdbm'
'--with-ndbm' '--with-db' '--with-png-dir=/usr/local'
'--with-zlib-dir=/usr/local' '--with-jpeg-dir=/usr/local'
gdb bt (running 4.2.0 from the commandline) -
#0 0x080e7cd5 in zend_hash_find (ht=0x8183e1c, arKey=0x8183bb4 "foo",
nKeyLength=4, pData=0xbfffdf74) at zend_hash.c:861
#1 0x080fab7a in zend_fetch_var_address (opline=0x8182b1c,
Ts=0xbfffdfa0, type=1) at ./zend_execute.c:560
#2 0x080fc6b8 in execute (op_array=0x8183ee8) at
./zend_execute.c:1239
#3 0x080dcdc3 in call_user_function_ex (function_table=0x8153530,
object_pp=0x0, function_name=0x818199c, retval_ptr_ptr=0xbfffe304,
param_count=1,
params=0x8184154, no_separation=0, symbol_table=0x0) at
zend_execute_API.c:517
#4 0x080f8118 in zif_array_map (ht=2, return_value=0x8184044,
this_ptr=0x0, return_value_used=1) at array.c:3228
#5 0x080fe283 in execute (op_array=0x8181ac4) at
./zend_execute.c:1598
#6 0x080e3efe in zend_execute_scripts (type=8, retval=0x0,
file_count=3) at zend.c:810
#7 0x08062b71 in php_execute_script (primary_file=0xb7f0) at
main.c:1381
#8 0x08060b7c in main (argc=3, argv=0xb89c) at cgi_main.c:785
#9 0x40136177 in __libc_start_main (main=0x8060374 , argc=3,
ubp_av=0xb89c, init=0x805efc8 <_init>, fini=0x8102db0 <_fini>,
rtld_fini=0x4000e184 <_dl_fini>, stack_end=0xb88c) at
../sysdeps/generic/libc-start.c:129
--
Edit this bug report at http://bugs.php.net/?id=17115&edit=1
Bug #17272 Updated: Failure to read Flash MX file size
ID: 17272
Updated by: [EMAIL PROTECTED]
Reported By: [EMAIL PROTECTED]
Status: Suspended
Bug Type: Feature/Change Request
Operating System: N/A
PHP Version: 4.1.2
New Comment:
I ran into the same problem when Flash 6 came out, so I tried to fix
the problem with PHP code. The SWC format is a regular SWF file, but
with a portion of the file compressed with zlib. Unfortunately the
width and height of the SWC file is located inside the compressed
portion.
SWF:
[HEADER - 3 bytes] 'FWS'
[VERSION - 1 byte] In which version was this file created
[SIZE - 4 bytes] size of file
[TAGS] All tags
SWC:
[HEADER - 3 bytes] 'CWS'
[VERSION - 1 byte] In which version was this file created
[SIZE - 4 bytes] size of uncompressed file
[ZLIBSTREAM] Zlib compressed stream of all tags
To decompress the file you need to copy the first 8 bytes of the SWC
file to a buffer and change to first byte of the buffer to 'C'. Then
you need to decompress the remaining bytes of the SWC file and append
it to the buffer.
The following PHP function uses this method to decompress the file:
function phpAds_SWFDecompress($buffer)
{
if (function_exists('gzuncompress') &&
substr($buffer, 0, 3) == swf_tag_compressed &&
ord(substr($buffer, 3, 1)) >= 6)
{
$output = 'F';
$output .= substr ($buffer, 1, 7);
$output .= gzuncompress (substr ($buffer, 8));
return ($output);
}
else
return ($buffer);
}
If you are only interested in retrieving the dimensions of the SWC file
it can be even easier. Just strip the first 8 bytes of the SWC file and
decompress the remaining bytes. The tags for the width and height
should be first two tags inside the decompressed stream.
Previous Comments:
[2002-05-17 13:27:46] [EMAIL PROTECTED]
I'm suspending this until somebody can show me some documentation on
the new fileformat. Apperently there are two Flash 6 formats, the SWF
one (which works fine and is the same as Flash 5) and the SWC one,
which is some compressed format.
Derick
[2002-05-16 13:22:22] [EMAIL PROTECTED]
Not really a bug, as it was not written for this :)
Anyway, if you could mail me empty flash movies with the following
size, I'll see what I can do:
255x127
640x480
400x256
(and perhaps some more)
Derick
[2002-05-16 12:22:48] [EMAIL PROTECTED]
The problem is with the getimagesize() function and its detection
routines which don't detect Flash 6/MX content.
[2002-05-16 12:00:44] [EMAIL PROTECTED]
Flash file size detection seems to not work for Flash files published
in the new Flash "MX" (version 6). When published to version 5 or
lower, it works.
--
Edit this bug report at http://bugs.php.net/?id=17272&edit=1
