Re: [PHP4BETA] cvs: /php4/ext/standard info.c

Colin Viebrock wrote: > > cmv Wed Apr 5 13:29:50 2000 EDT > > Modified files: > /php4/ext/standard info.c > Log: > fixed the segfault problem (probably not the prettiest way, but) [ ... ] > + int tmp_zend_debug; > > #ifdef PHP_WIN32 > char php_windows_uname[256]; > @@ -168,9 +169,10 @@ > php_info_print_table_row(2, "Configure Command", CONFIGURE_COMMAND ); > #endif > php_info_print_table_row(2, "php.ini Path", CONFIGURATION_FILE_PATH > ); > -/* why does this cause seg faults? > - php_info_print_table_row(2, "ZEND_DEBUG", ZEND_DEBUG ); > -*/ > + > + sprintf(tmp_zend_debug, "%d", ZEND_DEBUG); > + php_info_print_table_row(2, "ZEND_DEBUG", tmp_zend_debug ); > + Funny, I thought sprintf takes a (char *) as its first arg. I don't see why this wouldn't seg fault quite regularly; you're writing into whatever address happens to be on the stack. Moreover, the value of tmp_zend_debug in the call to php_info_print_table_row is undefined and gets dereferenced as a char* in php_info_print_table_row. Curious that this works for you . . . . Anyway, the reason it segfaulted before is that php_info_print_table_row expects all char*s as args. ZEND_DEBUG if I follow the .h files correctly is either set to 1 or unset. I think the way it should be written is: #ifdef ZEND_DEBUG php_info_print_table_row(2, "ZEND_DEBUG", "true" ) #else php_info_print_table_row(2, "ZEND_DEBUG", "false" ) #endif Ciao, Andreas