Re: command-line debugging broken?

An update to this. It only happens if an extension is dl()'ed. If it is loaded via an "extension=" in php.ini it works fine. -Rasmus On Mon, 22 Oct 2001, Rasmus Lerdorf wrote: > The leak notices and other --enable-debug messages seem to be brokwn right > now. > > ie. > > stick a stray char *s = emalloc(20); somewhere and try running php > script.php from the command line. I get a segfault that looks like this: > > #0 0x406ab75c in _IO_vfprintf (s=0xbfffeca0, format=0x8186e40 "%s(%d) : > Freeing 0x%.8lX (%d bytes), script=%s\n", > ap=0xbfffedd0) at ../sysdeps/i386/i486/bits/string.h:530 > #1 0x406cc615 in _IO_vsnprintf (string=0xbffff010 "", maxlen=512, > format=0x8186e40 "%s(%d) : Freeing 0x%.8lX (%d bytes), script=%s\n", > args=0xbfffedcc) at vsnprintf.c:131 > #2 0x406b3afb in __snprintf (s=0xbffff010 "", maxlen=512, > format=0x8186e40 "%s(%d) : Freeing 0x%.8lX (%d bytes), script=%s\n") at > snprintf.c:37 > #3 0x0806f0b3 in php_message_handler_for_zend (message=4, data=0x827b560) at main.c:582 > #4 0x08145fab in zend_message_dispatcher (message=4, data=0x827b560) at zend.c:616 > #5 0x08135fd5 in shutdown_memory_manager (silent=0, clean_cache=0) at zend_alloc.c:502 > #6 0x0806f75e in php_request_shutdown (dummy=0x0) at main.c:743 > #7 0x0806e096 in main (argc=2, argv=0xbffff934) at cgi_main.c:775 > #8 0x406706b7 in __libc_start_main (main=0x806d62c <main>, argc=2, ubp_av=0xbffff934, > init=0x8069ee4 <_init>, > fini=0x8185db0 <_fini>, rtld_fini=0x4000db64 <_dl_fini>, stack_end=0xbffff92c) > at ../sysdeps/generic/libc-start.c:129 > > (gdb) up > #3 0x0806f0b3 in php_message_handler_for_zend (message=4, data=0x827b560) > at main.c:582 > 582 snprintf(memory_leak_buf, 512, "%s(%d) : > Freeing 0x%.8lX (%d bytes), script=%s\n", t->filename, t->lineno, (unsigned > long)ptr, t->size, SAFE_FILENAME(SG(request_info).path_translated)); > (gdb) p *t > $2 = {magic = 1930623196, filename = 0x40017cab <Address 0x40017cab out of bounds>, > lineno = 123, reported = 0, > orig_filename = 0x0, orig_lineno = 0, pNext = 0x827b4e0, pLast = 0x0, size = 20, persistent = > 0, cached = 0} > > That mem_header looks a bit messed up. I could of course be stepping on > memory elsewhere. Anybody else seeing this? > > -Rasmus > > >