Date: Fri, 27 Jul 2012 20:29:02 +0800 From: David Xu <listlog2011@gmail.com> To: Bruce Evans <brde@optusnet.com.au> Cc: Garrett Cooper <yanegomi@gmail.com>, freebsd-bugs@FreeBSD.org, freebsd-gnats-submit@FreeBSD.org, davidxu@FreeBSD.org Subject: Re: kern/170203: [kern] piped dd's don't behave sanely when dealing with a fifo Message-ID: <5012898E.8060602@gmail.com> In-Reply-To: <20120727210330.L6869@besplex.bde.org> References: <201207262256.q6QMurVf077480@red.freebsd.org> <20120727103622.B933@besplex.bde.org> <501256C6.5000307@gmail.com> <20120727210330.L6869@besplex.bde.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 2012/7/27 19:08, Bruce Evans wrote: > On Fri, 27 Jul 2012, David Xu wrote: > >> On 2012/7/27 10:07, Bruce Evans wrote: >>> >>> I think it's working almost as expected. Large blocks give non-atomic >>> I/O, so the reader sees small blocks, then EOF when it gets ahead of >>> the writer. This always happens without SMP. >>> >>> Not is a bug (debugged below). There is no SIGPIPE at the start of >>> write() because there is a reader then, and no SIGPIPE for the next >>> write() because there is no next write() -- the current one doesn't >>> notice when the reader goes away. >>> >> After fixed dd to not open fifo output file in O_RDWR mode, I still >> found the >> writer is blocked there even the reader is already exited. > > I'm not sure that dd's open is a bug. It must be intentional to use > O_RDWR for some cases. > Don't know if original author even thought about FIFO. > POSIX (old 2001 draft) doesn't say anything about dd's open mode. > >> I think this is definitely a bug. if reader is exited, the writer >> should be aborted too, >> but I found it still be blocked in state "pipedwt", obviously, the >> code in >> /sys/fs/fifo_vnops.c wants to wake up the writer when the reader is >> closing the fifo, >> but it failed, because the bit flag PIPE_WANTW is forgotten to be set >> by writer, >> so it skips executing wakeup(), and then the writer has no chance to >> find EOF bit flag >> is set. > > Does this affect nameless pipes too? The old implementation presumably > doesn't have this bug. > It is easy to repeat the bug for named pipes, don't know if nameless pipes have same bug, I can not reproduce it yet. > Bruce > . >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5012898E.8060602>