bsafw@ncoast.UUCP (Brandon Allbery) (06/04/84)
Our installation (running Microsoft Xenix 1.03.02 (sorry Berkeley :-))) is capable of using multiplex files, but MS forgot to send us any documentation for it. Can anyone send us a copy of man/man2/mpx.2? and/or whatever other doc is available on the subject? If posting to the net isn't a good idea, mail to one of the addresses named in my signature: !bsafw or decvax!cwruecmp!ncoast!stuart. -- -------------------------------------------------------------------------------- Brandon Allbery decvax!cwruecmp!ncoast!bsafw "...he himself being one universe's prime MCI MAIL: 161-7070 example of utter, rambunctious free will!" USMail (core dump): 6504 Chestnut Road Independence, OH 44131
chris@umcp-cs.UUCP (06/05/84)
* From: bsafw@ncoast.UUCP Our installation (running Microsoft Xenix 1.03.02 ...) is capable of using multiplex files... But have they fixed the bugs that cause the system to crash and the disks to be corrupted if you use them? Perhaps that's why they didn't include documentation.... I'm not sure what the system call index number is, but the names in the C library are ``mpx'', ``chan'', ``extract'', ``attach'', ``detach'', and ``connect''. (They all go through a common syscall vector. There are some more names but they aren't generally very useful.) ``mpx'' creates multiplexed file nodes, ``chan'' creates a channel, ``attach'' attaches a channel to an mpx node, ``detach'' removes a channel from an mpx node, ``extract'' gets a file descriptor to a channel, and ``connect'' allows you to connect TTYs to the mpx file. There are a huge number of magic things that can happen on MPX files (M_WATCH, M_BLK, MXNBLK mode, ...), of which perhaps most work if you use them very carefully. Closing mpx nodes tends to hit the kernel bug and lose files, directories, etc. There is a race condition in which the inode gets freed twice. To sum up: if you can possibly do so, use PTYs rather than MPX files. (Does anyone know if sending mpx(2)/mpxio(5) manuals would violate copyright laws?) -- In-Real-Life: Chris Torek, Univ of MD Comp Sci (301) 454-7690 UUCP: {seismo,allegra,brl-bmd}!umcp-cs!chris CSNet: chris@umcp-cs ARPA: chris@maryland