From owner-freebsd-bugs Tue Jun 24 12:50:03 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA02527 for bugs-outgoing; Tue, 24 Jun 1997 12:50:03 -0700 (PDT) Received: (from gnats@localhost) by hub.freebsd.org (8.8.5/8.8.5) id MAA02518; Tue, 24 Jun 1997 12:50:01 -0700 (PDT) Date: Tue, 24 Jun 1997 12:50:01 -0700 (PDT) Message-Id: <199706241950.MAA02518@hub.freebsd.org> To: freebsd-bugs Cc: From: Bill Fenner Subject: Re: bin/3942: mmap fails on large file on CDROM Reply-To: Bill Fenner Sender: owner-bugs@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk The following reply was made to PR bin/3942; it has been noted by GNATS. From: Bill Fenner To: freebsd-gnats-submit@freebsd.org Cc: Subject: Re: bin/3942: mmap fails on large file on CDROM Date: Tue, 24 Jun 1997 12:43:41 PDT A little more info: When performing the "tail" experiment, the following messages often get logged: Jun 24 19:26:16 sundae /kernel: cd0(ncr0:6:0): MEDIUM ERROR info:12e73 asc:15,0 Random positioning error Jun 24 19:26:16 sundae /kernel: vm_fault: pager input (probably hardware) error, PID 16521 failure Jun 24 19:26:16 sundae /kernel: pid 16521 (tail), uid 0: exited on signal 11 However, dd'ing the same file succeeds at reading the block in question, so I'm not sure that I believe there's a medium error in the file. Also, I suspect that the zero-filled page comes from debugging the process; does the debugging fault handler insert a zero-filled page to let a process continue from accessing an unmapped page? And presumably, it makes its way into the buffer cache because of the merged VM/buffer cache (this might be a bug of its own, that debugging a process that uses mmap() can insert stuff into the buffer cache...) I guess these two errors are from trying to flush the modified buffer back to the CDROM: Jun 24 19:33:42 sundae /kernel: vnode_pager_putpages: I/O error 45 Jun 24 19:33:43 sundae /kernel: vnode_pager_putpages: residual I/O 4096 at 38239 Bill