Recent changes to bugs

gzread inside a class breakes debuger

Sinisa Culic — Mon, 07 Mar 2011 14:24:52 -0000

calling a gzread to read a segment inside a class breaks the debuger AND THE FILE ITSELF!!!! meaning you lose the archive you were working with and all data in it!!!!!!!

i am using

with ubuntu OPS

Missing includes with GCC 4.3 (fix included)

Warbothong — Fri, 15 Oct 2010 13:56:47 -0000

GCC 4.3 has apparently changed some of the #includes of its standard libraries, which requires adding a few more explicit includes to DBG's sources. I've attached an archive of those files which are affected, updated appropriately. I've also put in some conditional includes to make PATH_MAX work on Linux. If you're having compilation problems for dbg-cli then try replacing some of the 2.15.5 source files with these counterparts. Worked for me on Fedora 11, x86.

Vanilla Ubuntu x64 10.04 Installation

Chris Landress — Fri, 30 Apr 2010 22:24:27 -0000

This is my first bug submission so I hope I am doing this correctly:

Steps to recreate:
1) Downloaded Ubuntu Server x64 10.04
2) Did a vanilla install without adding any services
3) Installed all the required packages : apache2 php php5-dev php-config apache2 mysql5 etc etc etc
4) ./configure --enable-dbg=shared --with-dbg-profiler --with-php-config=/usr/bin/php-config --prefix=/usr/lib/php5/20090626
5) make

See attached for compile log. I've followed these same steps on several versions of Ubuntu in the past with no issues.

DBG does not find HTTP_SERVER_VARS in &EG(symbol_table)

Martin Grimm — Sun, 04 Feb 2007 23:00:54 -0000

Hi, dmdmitry, don't know, if you're still using the tracker as the last bug report is from 2004, but I'll try here first before using the NuSphere forums.

My webserver is Apache/2.2.0 with a stock OpenSuSE PHP 5.1.2 running on an OpenSuSE Linux 10.1 kernel 2.6.16.27-0.6-smp

I'm trying to remote debug with a win 2000 client.

I tested this with a self compiled DBG 2.15.1 as well as with a premade 2.13.1 extension. I even compiled 2.15.1 --with-dbg-trace and added some trace messages of my own to find out a little more about this problem (as I'm not really good with C, that's all I was able to do about it).

Problem is: DBG always assumes "localhost" as address of the listening client.

When using an URL in the form of http://www/dbgtest/debug.php?DBGSESSID=1@clienthost:10001 I always get an error "failed to establish connection to client host on localhost:10001", although I'm not connecting from localhost but another host on the network. DBG is working as intended, whenever I replace "clienthost" with the IP address or name of the debugging client.

DBG recognises the clienthost option in the request in dbg_cmd create_debugger_socket(...) and enters get_redirected_address(...) as it should.

When DBG tries to get the HTTP_SERVER_VARS with

rv = zend_hash_find(&EG(symbol_table), "HTTP_SERVER_VARS", sizeof("HTTP_SERVER_VARS"), (void **) &httpvars);
if (rv != FAILURE && ((*httpvars)->type==IS_ARRAY)) {
...

it fails and get_redirected_address returns NULL. create_debugger_socket sets DBG(client_address) to "localhost" as a consequence.

Now the strange thing is, when I check the the Apache Environment with phpinfo() there's no problem: REMOTE_ADDR is set to the address of my debugging client.

I don't really know if this problem is DBG or Zend related or if another extension might be the problem (although I tested it without loading other extensions at some time). It might be related to http://support.nusphere.com/viewtopic.php?t=2336

I'll attach a phpinfo() I created from my debugging host connecting to the server. Please contact me if you need more information or if you have any hints regarding a solution to this problem.

2.11.26 - FreeBSD patches

Edwin Groothuis — Sat, 11 Sep 2004 09:19:06 -0000

Please review the following patches:

The first includes are to make sure that inet_isaddr
and gethostbyname are prototyped properly.
The wcnt is an unused variable (-Wall)
The last one is a long vs int issue.

--- dbg_cmd.c.orig Sat Sep 11 18:46:24 2004
+++ dbg_cmd.c Sat Sep 11 18:48:11 2004
@@ -20,6 +20,12 @@
#include "dbg_bp.h"
#include "SAPI.h"

+#include <sys/types.h>
+#include <sys/socket.h>
+#include <netinet/in.h>
+#include <arpa/inet.h>
+#include <netdb.h>
+
#include "php_network.h"

@@ -789,7 +795,7 @@

/* waiting and looking for debugger host responce */
inline void dbg_ack_loop(TSRMLS_D2(DBG, E)) {
- int ret_val = 0, wcnt = 0;
+ int ret_val = 0;
dbg_header_struct hdr;
dbg_packet pack;

@@ -1102,7 +1108,7 @@
DBG(req_client_port) =
atol(numbuf);
if (DBG(req_client_port) <= 0
|| DBG(req_client_port) > 32767) DBG(req_client_port) =
DEFAULT_PORT;
}
- snprintf(reqbuf, sizeof(reqbuf),
DBGSESSVAR "=%s@%s:%d", DBG(session_id),
DBG(req_client_ip_address), DBG(req_client_port));
+ snprintf(reqbuf, sizeof(reqbuf),
DBGSESSVAR "=%s@%s:%ld", DBG(session_id),
DBG(req_client_ip_address), DBG(req_client_port));
DBG(req_sess_var) = estrdup(reqbuf);
DBG_TRACE(("parsed req IP=%s, PORT=%ld
(%s), SID=%s\n", SON(DBG(req_client_ip_address)),
DBG(req_client_port), numbuf, SON(DBG(session_id))));
} else {

dbg-cli compile error on FreeBSD 5.2.1 RELEASE

Joseph Mingrone — Mon, 02 Aug 2004 20:00:47 -0000

Compile error:

In file included from main.cpp:38:
../getopt/getopt.h:108: error: declaration of C function `int getopt()'
conflicts with
/usr/include/unistd.h:374: error: previous declaration `int getopt(int,
char*
const*, const char*)' here
*** Error code 1

Stop in /usr/home/joey/dbg/dbg-cli-2.11.0/dbg.
*** Error code 1

Stop in /usr/home/joey/dbg/dbg-cli-2.11.0.
*** Error code 1

Stop in /usr/home/joey/dbg/dbg-cli-2.11.0

dbg-cli compile error on Solaris 8

Justin Clift — Thu, 15 Jan 2004 08:08:55 -0000

Hi,

Attempting to compile the dbg-cli on Solaris 8, gives
me this compilation error:

if g++ -DHAVE_CONFIG_H -I. -I. -I..
-DDBG_USE_STDALLOCA -I../pcre -g -O2 -MT
dbgbase_globals.o -MD -MP -MF ".deps/dbgbase_globals.Tpo" \ -c -o dbgbase_globals.o `test -f
'dbgbase_globals.cpp' || echo './'`dbgbase_globals.cpp; \ then mv -f ".deps/dbgbase_globals.Tpo"
".deps/dbgbase_globals.Po"; \ else rm -f ".deps/dbgbase_globals.Tpo"; exit 1; \ fi
dbgbase_globals.cpp: In function `int
dbh_create_listen_socket(const char*,
short unsigned int)':
dbgbase_globals.cpp:581: error: `INADDR_NONE'
undeclared (first use this
function)
dbgbase_globals.cpp:581: error: (Each undeclared
identifier is reported only
once for each function it appears in.)
dbgbase_globals.cpp: In function `int
dbh_check_new_client_connect(int,
sockaddr*, int)':
dbgbase_globals.cpp:623: error: invalid conversion from
`int*' to `socklen_t*'
make[2]: *** [dbgbase_globals.o] Error 1
make[2]: Leaving directory
`/export/home/jclift/install/dbg-cli-2.11.0/dbgbase'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory
`/export/home/jclift/install/dbg-cli-2.11.0'
make: *** [all] Error 2

Using these tools:

libtool 1.5
automake 1.7.9
autoconf 2.58
gcc 3.3

Any ideas?

Regards and best wishes,

Justin Clift

line nrs wrong with source files starting with a shebang

Sven Schnitzke — Tue, 25 Mar 2003 13:31:48 -0000

When a source file starts with a shebang (e.g. #!/usr/local/bin/php -q) the line numbers
reported by DBG are one too big. This is not the case when the string is moved to col 2
of the line.

This almost seems as if it is by design. If it is please define the exact circumstances.

Experienced using dbg dll ver 2.11.5 for 4.3rc2, PHP4.4-dev on WIN98SE.

create_function returns incomplete string

Sven Schnitzke — Tue, 25 Mar 2003 12:19:02 -0000

When evaluating the expression "create_function('$var', 'return $var*2;') the result should be the
name of the newly created function (according to PHP docs). DBG returns a string fragment
instead >s:9:"< .
I experience that using PHP4-4dev, DBG 2.11.4 (and .5) using php_dbg.dll 4.3rc2 on WIN98SE.

PHP give Invalid Page Fault with DBG

Larry O'Leary — Fri, 27 Sep 2002 08:33:14 -0000

I am using PHP 4.2.1 on Win98SE
After installing the 2.11.2 win32 binaries PHP would
crash with:

PHP caused an invalid page fault in
module PHP4TS.DLL at 0167:100a44f1.
Registers:
EAX=012b1120 CS=0167 EIP=100a44f1
EFLGS=00010202
EBX=100a44e0 SS=016f ESP=0063fc64
EBP=007913bc
ECX=00000017 DS=016f ESI=01172c38 FS=32df
EDX=00791330 ES=016f EDI=00760650 GS=0000
Bytes at CS:EIP:
8b 3e 85 c0 74 09 8d 4e 08 51 ff d0 83 c4 04 8a
Stack dump:
00760650 00760650 00000000 012b1654
007913bc 00760650 1010b620 00791630
78001d42 012b270b 00760650 00791630
1008d003 00000001 0000000c 00760650

If I disabled the loading of the dbg extension,
everything would be fine. I finally went back to the
2.11 version and everything is fixed.

The crash would happen almost immidiately. At first
PHP would crash even without any commands:
C:\&gt;PHP
<CTRL+Z>
An Invalid Page Fault would occur...
Then after a few times of doing this I finally got PHP
not to crash until I had it output plain HTML text...
C:>PHP
Some Test Text
<CTRL+Z>
Then it would crash with the exact same IPF
Then, it would only crash if I used PHP code.
C:> PHP
<?php
phpinfo();
?>
This would output all the phpinfo() and then PHP
would crash with the IPF.

After going back to the prvious release, everything
appears okay.

Thanks for you attention to this matter.