Date: Wed, 31 May 2006 22:39:32 +1000 (EST) From: Bruce Evans <bde@zeta.org.au> To: Maxim Konovalov <maxim@macomnet.ru> Cc: freebsd-bugs@FreeBSD.org Subject: Re: bin/98141: [patch] Update for regression/sockets/unix_cmsg Message-ID: <20060531221622.I30824@delplex.bde.org> In-Reply-To: <200605301350.k4UDoIAj049864@freefall.freebsd.org> References: <200605301350.k4UDoIAj049864@freefall.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 30 May 2006, Maxim Konovalov wrote: > > o Use PRIdMAX macro for outputting intmax_t values. > > What is it for? I see nothing is using this macro in the base system > and it is defined as "jd" on all our platforms. Unfortunately, the PRI* mistake is used in a few places in the base system. At least: libc_r/test/ fsck_ffs/ (of all places) find/ getconf/ memcontrol/ Its use in find/ is an especially good example of how not to use it. In fsck_ffs/, PRI_i64 is used to print some types that are certain to match, but in find/ PRI_i64 is used to print st_blocks and st_size which have opaque types. st_blocks happens to have type int64_t, but POSIX reqires it to have type blkcnt_t which can be any signed integer type. st_size has type off_t which happens to match. find/ also prints st_ino using a cast to u_long on the same line that it prints st_blocks without a cast. Bruce
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20060531221622.I30824>