Yep, the spec goes right. a corresponding va_end() dtor should be applied to ap once ap has been initialized by a va_start(). IMO no va_end() is needed without a preceding va_start(), and it doesn't matter if ap is used between va_start() and va_end().
BTW, could anyone commit this patch if there seems no problem? Moriyoshi [EMAIL PROTECTED] (Marcus Börger) wrote: > Some comments on ISO9899 standard > 7.15.1.3-2 Read between the lines: without va_end the behaviour is undefined. > What ever that means i guess you have to call va_end and that requires > va_start. > > 7.15.1.4-3 Says do not call va_start twice without va_end. > > marcus > > > ISO/IEC 9899:1999 (E) ©ISO/IEC > > 7.15.1.3 The va_end macro > Synopsis > 1 #include <stdarg.h> > void va_end(va_list ap); > Description > 2 The va_end macro facilitates a normal return from the function whose variable > argument list was referred to by the expansion of va_start, or the function > containing > the expansion of va_copy, that initialized the va_list ap. The va_end macro may > modify ap so that it is no longer usable (without an intervening invocation > of va_start > or va_copy). If there is no corresponding invocation of the va_start or va_copy > macro, or if the va_end macro is not invoked before the return, the behavior is > undefined. > Returns > 3 The va_end macro returns no value. > > 7.15.1.4 The va_start macro > Synopsis > 1 #include <stdarg.h> > void va_start(va_list ap, parmN); > Description > 2 The va_start macro shall be invoked before any access to the unnamed > arguments. > 3 The va_start macro initializes ap for subsequent use by va_arg and va_end. > va_start (or va_copy) shall not be invoked again for the same ap without an > intervening invocation of va_end for the same ap. > (...) > > > At 10:47 08.11.2002, Moriyoshi Koizumi wrote: > >See http://www.opengroup.org/onlinepubs/007908799/xsh/stdarg.h.html > >This appears to imply that va_start() can be used more than twice. > > > >And I don't think va_start() always has to be invoked. > > > >Moriyoshi > > > >[EMAIL PROTECTED] (Marcus Börger) wrote: > > > > > I am not sure if va_start can be called twice in a row (rekursive). > > > Manual does not say anything about that. > > > > > > How about: > > > > > > cvs -z3 -q diff zend_hash.c (in directory S:\php4-HEAD\Zend) > > > Index: zend_hash.c > > > =================================================================== > > > RCS file: /repository/ZendEngine2/zend_hash.c,v > > > retrieving revision 1.93 > > > diff -u -r1.93 zend_hash.c > > > --- zend_hash.c 5 Nov 2002 18:22:02 -0000 1.93 > > > +++ zend_hash.c 8 Nov 2002 09:32:48 -0000 > > > @@ -722,9 +722,13 @@ > > > > > > HASH_PROTECT_RECURSION(ht); > > > > > > - va_start(args, num_args); > > > p = ht->pListHead; > > > + if (p == NULL) { > > > + va_start(args, num_args); > > > + va_end(args); > > > + } > > > while (p != NULL) { > > > + va_start(args, num_args); > > > hash_key.arKey = p->arKey; > > > hash_key.nKeyLength = p->nKeyLength; > > > hash_key.h = p->h; > > > @@ -733,8 +737,8 @@ > > > } else { > > > p = p->pListNext; > > > } > > > + va_end(args); > > > } > > > - va_end(args); > > > > > > HASH_UNPROTECT_RECURSION(ht); > > > } > > > > > > > > > marcus > > > > > > At 09:52 08.11.2002, Moriyoshi Koizumi wrote: > > > >Hi, > > > > > > > >The attached patch is a probable fix for bug #19566. I guess the bug > > > >is that va_list is not properly initialized before each callback function > > > >call. I've tested it in PPC linux, and it works fine. > > > > > > > >Regards, > > > >Moriyoshi > > > > > > > > > > > >-- > > > >PHP Development Mailing List <http://www.php.net/> > > > >To unsubscribe, visit: http://www.php.net/unsub.php -- PHP Development Mailing List <http://www.php.net/> To unsubscribe, visit: http://www.php.net/unsub.php
