Re: svn commit: r205165 - head/lib/libc/gen
On Wed, 17 Mar 2010, [utf-8] Dag-Erling Sm??rgrav wrote: Bruce Evans b...@optusnet.com.au writes: Even if the child causes the flush, the content of stdout and stderr is not flushed normally since it is redirected to /dev/null. Unflushed input in stdin is handled more brokenly: although stdin is redirected to /dev/null, input on it can still be read via stdin's buffer. Inheriting unflushed input on other streams is a feature (unless these streams are open on fd's 0-2; then these streams will have the same corruption as std* streams open on their normal fd's 0-2). how about Index: gen/daemon.c === --- gen/daemon.c(revision 204870) +++ gen/daemon.c(working copy) @@ -37,6 +37,7 @@ #include errno.h #include fcntl.h #include paths.h +#include stdio.h #include stdlib.h #include signal.h #include unistd.h @@ -81,6 +82,9 @@ (void)chdir(/); if (!noclose (fd = _open(_PATH_DEVNULL, O_RDWR, 0)) != -1) { + fpurge(stdin); + fflush(stdout); + fflush(stderr); (void)_dup2(fd, STDIN_FILENO); (void)_dup2(fd, STDOUT_FILENO); (void)_dup2(fd, STDERR_FILENO); ? That seems like a reasonable start. I wanted to flush/purge all streams but that is more dangerous of course. I misread the man page about fflush(NULL) purging anything. It is only fpurge() that purges, but I didn't notice that fflush()'s man page is also about fpurge(). fpurge() hasn't caught up with the 1980's developments that made fflush(NULL) operate on all streams, so there is no way to purge all streams without using stdio internals (_fwalk()). The above is missing purging of stdout and stderr. Someone may have freopen()ed these in rw mode. Callers of daemon() shouldn't do this, but if fpurge(NULL) worked right then it would be easy to purge stdout and stderr when purging stdin and all other streams (if this is what we want to do). Bruce___ svn-src-all@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-all To unsubscribe, send any mail to svn-src-all-unsubscr...@freebsd.org
Re: svn commit: r205165 - head/lib/libc/gen
Bruce Evans b...@optusnet.com.au writes: Even if the child causes the flush, the content of stdout and stderr is not flushed normally since it is redirected to /dev/null. Unflushed input in stdin is handled more brokenly: although stdin is redirected to /dev/null, input on it can still be read via stdin's buffer. Inheriting unflushed input on other streams is a feature (unless these streams are open on fd's 0-2; then these streams will have the same corruption as std* streams open on their normal fd's 0-2). how about Index: gen/daemon.c === --- gen/daemon.c(revision 204870) +++ gen/daemon.c(working copy) @@ -37,6 +37,7 @@ #include errno.h #include fcntl.h #include paths.h +#include stdio.h #include stdlib.h #include signal.h #include unistd.h @@ -81,6 +82,9 @@ (void)chdir(/); if (!noclose (fd = _open(_PATH_DEVNULL, O_RDWR, 0)) != -1) { + fpurge(stdin); + fflush(stdout); + fflush(stderr); (void)_dup2(fd, STDIN_FILENO); (void)_dup2(fd, STDOUT_FILENO); (void)_dup2(fd, STDERR_FILENO); ? DES -- Dag-Erling Smørgrav - d...@des.no ___ svn-src-all@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-all To unsubscribe, send any mail to svn-src-all-unsubscr...@freebsd.org
Re: svn commit: r205165 - head/lib/libc/gen
On Mon, 15 Mar 2010, Poul-Henning Kamp wrote: In message 20100316024446.a24...@delplex.bde.org, Bruce Evans writes: On Tue, 16 Mar 2010, Bruce Evans wrote: Due to the way that daemon() works, it is really an error to have any open streams when it is called. This is also undocumented, except implicitly. The errors are: - unflushed output on stdout and stderr won't get flushed normally by the child. stdout and stderr are redirected so it will go there if the child erroneously (?) uses these streams or simply calls exit() which will give the default flush. The in-core FILE buffers are copied in the child process, so they should most certainly not be flushed, but rather, as they correctly are, discarded, so that when the child causes the flush, the content is only output once. Nope. Even if the child causes the flush, the content of stdout and stderr is not flushed normally since it is redirected to /dev/null. Unflushed input in stdin is handled more brokenly: although stdin is redirected to /dev/null, input on it can still be read via stdin's buffer. Inheriting unflushed input on other streams is a feature (unless these streams are open on fd's 0-2; then these streams will have the same corruption as std* streams open on their normal fd's 0-2). As described in the unquoted part of my reply above, all streams should be flushed by the parent before the fork(), so that they are flushed normally. daemon(3) has a lot to say about problems with open fd's 0-2, but it says nothing about the extra problems with open streams on these descriptors. It documents as a feature that fd's above 2 are inherited normally. It also says too little about the implementation always using fork() (it cross-references fork() but doesn't say that the context is always a child context on successful completion). Bruce ___ svn-src-all@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-all To unsubscribe, send any mail to svn-src-all-unsubscr...@freebsd.org
svn commit: r205165 - head/lib/libc/gen
Author: phk Date: Mon Mar 15 08:58:35 2010 New Revision: 205165 URL: http://svn.freebsd.org/changeset/base/205165 Log: Comment a fine point, so it does not get lost when people borrow code from FreeBSD for other purposes. Modified: head/lib/libc/gen/daemon.c Modified: head/lib/libc/gen/daemon.c == --- head/lib/libc/gen/daemon.c Mon Mar 15 00:29:15 2010(r205164) +++ head/lib/libc/gen/daemon.c Mon Mar 15 08:58:35 2010(r205165) @@ -64,6 +64,10 @@ daemon(nochdir, noclose) case 0: break; default: + /* +* A fine point: _exit(0), not exit(0), to avoid triggering +* atexit(3) processing +*/ _exit(0); } ___ svn-src-all@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-all To unsubscribe, send any mail to svn-src-all-unsubscr...@freebsd.org
Re: svn commit: r205165 - head/lib/libc/gen
On Mon, 15 Mar 2010, Poul-Henning Kamp wrote: Log: Comment a fine point, so it does not get lost when people borrow code from FreeBSD for other purposes. Modified: head/lib/libc/gen/daemon.c Modified: head/lib/libc/gen/daemon.c == --- head/lib/libc/gen/daemon.c Mon Mar 15 00:29:15 2010(r205164) +++ head/lib/libc/gen/daemon.c Mon Mar 15 08:58:35 2010(r205165) @@ -64,6 +64,10 @@ daemon(nochdir, noclose) case 0: break; default: + /* +* A fine point: _exit(0), not exit(0), to avoid triggering +* atexit(3) processing +*/ _exit(0); } This point is obvious -- the programmer used _exit() for a reason, and reading _exit(3) gives the reason. _exit(3) only mentions atexit(3) cryptically as etc., but it makes it clear that there are more reasons than to avoid triggering atexit(3) -- exit(3) also does things like flushing open streams which are logically and possibly implementationally independent of atexit(3). They _should_ be implementationally independent to avoid bloating all programs with dynamic allocation of the atexit() table. The unobvious points are: - why avoid triggering atexit(), etc. processing in daemon()? - why isn't this documented? Some callers of daemon() need to know that it won't flush open streams, etc., so that they can flush and their open streams, etc., before calling daemon(). Callers should do this anyway before calling any function that can exit (especially exit() itself), so that they can actually check that the output went out, but most are sloppy. Callers should know if they have any open streams but they would have a hard time telling what else they are missing from not calling exit(). There may be profiling cleanup (profiling won't work right for the child?) and other magic destructors. Bruce ___ svn-src-all@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-all To unsubscribe, send any mail to svn-src-all-unsubscr...@freebsd.org