|
|
php.net | support | documentation | report a bug | advanced search | search howto | statistics | random bug | login |
[2000-01-16 23:51 UTC] simon at airdmhor dot gen dot nz
Long time php4/cvs user, using the same build script as I always do. Very recently without sybase, but with it this time :
./configure --prefix=/home/simonr/www --with-apache=../apache_1.3.9 --enable-track-vars --enable-debugger --with-xml --with-pgsql --with-sybase-ct=/usr/sybase --with-gd --with-config-file-path=/home/simonr/www/conf/php.ini
And I'm getting inconsistant results on the same script. It either :
a) works fine,
b) reports an error on a line that's in the middle of a comment (ie: Fatal error: Cannot redeclare class xxdebug in class.debug.php on line 3), or
c) fails to respond to the HTTP request at all
I'm also getting :
[Mon Jan 17 17:43:45 2000] Script: '/home/simonr/www/htdocs/crash.php'
---------------------------------------
php_sybase_ct.c(338) : Block 0x081DFF58 status:
Beginning: Cached (allocated on php_sybase_ct.c:319, 8 bytes)
End: OK
---------------------------------------
[Mon Jan 17 17:43:45 2000] [notice] child pid 629 exit signal Segmentation fault (11)
happening with trivial scripts. I'm still working on narrowing down exactly what code triggers it.
Mostly the segfault doesn't generate a core file, and when it does, the backtrace only has 1 line -
#0 0x8060bbd in ap_get_server_built ()
More soon.
PatchesPull RequestsHistoryAllCommentsChangesGit/SVN commits
|
|||||||||||||||||||||||||||
Copyright © 2001-2026 The PHP GroupAll rights reserved. |
Last updated: Sun Oct 11 15:00:01 2026 UTC |
Hmmm, got the latest snapshot (php4-200001162045) and hitting reload a few times on a trivial source easily generates : [Mon Jan 17 18:07:40 2000] [notice] child pid 12602 exit signal Segmentation fault (11), possible coredump in /home/simonr/www The file is : <?php session_start(); ?> I guess the session stuff is still broken? The bt looks a bit more useful : (gdb) bt #0 0x8088dc6 in zend_hash_add_or_update () #1 0x807e6e1 in do_bind_function_or_class () #2 0x807e8d9 in do_early_binding () #3 0x80db7a6 in zendparse () #4 0x808d5a5 in require_file () #5 0x808d366 in require_filename () #6 0x808020c in do_require () #7 0x80dbaf2 in zendparse () #8 0x808cf4b in v_compile_files () #9 0x808ce74 in compile_files () #10 0x80772b9 in php_execute_script () #11 0x8090c13 in apache_php_module_main () #12 0x8075694 in send_php () #13 0x80756de in send_parsed_php () #14 0x80eaf09 in ap_invoke_handler () #15 0x80fef6f in ap_some_auth_required () #16 0x80fefd6 in ap_process_request () #17 0x80f62f6 in ap_child_terminate () #18 0x80f6578 in ap_child_terminate () #19 0x80f6908 in ap_child_terminate () #20 0x80f6e7c in ap_child_terminate () #21 0x80f74ac in main () #22 0x401dfcb3 in __libc_start_main (main=0x80f7124 <main>, argc=1, argv=0xbffffa44, init=0x805ea0c <_init>, fini=0x812bd8c <_fini>, rtld_fini=0x4000a350 <_dl_fini>, stack_end=0xbffffa3c) at ../sysdeps/generic/libc-start.c:78 (gdb) But there's another unrelated problem! See next message.Another crash happened. Hmm, maybe it's a broken build system? I think the sysadmin upgraded some stuff a couple of weeks ago. How would I tell? [Mon Jan 17 18:31:50 2000] Script: '/home/simonr/www/htdocs/class.crash.php' --------------------------------------- zend_execute_API.c(216) : Block 0x081DAEB8 status: zend_variables.c(63) : Actual location (location was relayed) Beginning: Overrun (magic=0x2A8FCC84, expected=0x7312F8DC) End: Unknown --------------------------------------- [Mon Jan 17 18:31:51 2000] [notice] child pid 12754 exit signal Segmentation fault (11), possible coredump in /home/simonr/www (gdb) bt #0 0x8082615 in destroy_zend_class () #1 0x808a2e3 in zend_hash_clean () #2 0x808a451 in zend_hash_apply () #3 0x807cbbb in shutdown_compiler () #4 0x8086af3 in zend_deactivate () #5 0x8076762 in php_request_shutdown () #6 0x80e7a79 in ap_run_cleanup () #7 0x80e6123 in ap_clear_pool () #8 0x80e61a3 in ap_destroy_pool () #9 0x80f6353 in ap_child_terminate () #10 0x80f6578 in ap_child_terminate () #11 0x80f6908 in ap_child_terminate () #12 0x80f6e7c in ap_child_terminate () #13 0x80f74ac in main () #14 0x401dfcb3 in __libc_start_main (main=0x80f7124 <main>, argc=1, argv=0xbffffa44, init=0x805ea0c <_init>, fini=0x812bd8c <_fini>, rtld_fini=0x4000a350 <_dl_fini>, stack_end=0xbffffa3c) at ../sysdeps/generic/libc-start.c:78 (gdb)Oh, BTW, a script as trivial as <?php class Database { }; ?> has been causing apache to segfault as described in the first note above. Another example below. It looks like a rogue pointer is trashing whatever... Maybe the one that's in php_sybase_ct.c:338 as described in apache's error_log, or something trashing it. That function reads : PHP_RSHUTDOWN_FUNCTION(sybase) { efree(sybase_globals.appname); // line 338 if (sybase_globals.server_message) { efree(sybase_globals.server_message); } return SUCCESS; } Does it need to check for a NULL pointer too? I haven't got time to run purify on it tonight - maybe tomorrow. (gdb) bt #0 0x8088dc6 in zend_hash_add_or_update () #1 0x807f4fc in do_begin_class_declaration () #2 0x80dbc51 in zendparse () #3 0x808cf4b in v_compile_files () #4 0x808ce74 in compile_files () #5 0x80772b9 in php_execute_script () #6 0x8090c13 in apache_php_module_main () #7 0x8075694 in send_php () #8 0x80756de in send_parsed_php () #9 0x80eaf09 in ap_invoke_handler () #10 0x80fef6f in ap_some_auth_required () #11 0x80fefd6 in ap_process_request () #12 0x80f62f6 in ap_child_terminate () #13 0x80f6578 in ap_child_terminate () #14 0x80f6908 in ap_child_terminate () #15 0x80f6e7c in ap_child_terminate () #16 0x80f74ac in main () #17 0x401dfcb3 in __libc_start_main (main=0x80f7124 <main>, argc=1, argv=0xbffffa44, init=0x805ea0c <_init>, fini=0x812bd8c <_fini>, rtld_fini=0x4000a350 <_dl_fini>, stack_end=0xbffffa3c) at ../sysdeps/generic/libc-start.c:78 (gdb)