Subject: Re: port-i386/6862: Problems with va_start and gcc
To: None <>
From: Andreas Gustafsson <>
List: netbsd-bugs
Date: 01/22/1999 17:00:44
>    if the second parameter of va_start() is of type unsigned char
>    the macro does not calculate the address of the variable argument
>    list correctly, because the address of the unsigned char variable
>    is "of sequence".

I don't think using a char (signed or unsigned, doesn't matter) as the
second parameter of va_start is legal C.  At least that is my
interpretation of the ISO/IEC JTC1/SC22/WG14 N843 draft; it would be
nice if someone with access to the ANSI C standard could double-check
this.  See PR #6431 for more information.
Andreas Gustafsson,