louie@umd5.UUCP (05/01/85)
I've got GNU EMACS, version 15.22 running on a VAX 11/750 under ULTRIX 1.0 (which is 4.2BSD from DEC). It seems to work just fine except for one strange problem. When exiting from EMACS via a ^X-^C key sequence, it seems to 'leave' characters in the tty input queue. Typically, they are ^? (DEL), ~ (tilde), or @ (at-sign). Usually only one or two queued. I tried doing a TIOCFLUSH in RstTrm(), but that didn't seem to fix it. I am running with INTERRUPT_INPUT defined. I think it may be related to switching the tty back into cooked mode and disabling async I/O, but have no concrete reasons to support this claim. It doesn't occur with any other programs. Help! -- Louis A. Mamakos WA3YMH University of Maryland, Computer Science Center Internet: louie@umd5.arpa UUCP: {seismo!umcp-cs, ihnp4!rlgvax}!cvl!umd5!louie
smh@mit-eddie.UUCP (Steven M. Haflich) (05/03/85)
I too have experienced the problem with GNU EMACS emitting garbage characters to the tty upon exit. When using a VT100 emulator on a PC that reports ---BREAK-RECEIVED--- on framing errors, it bacame apparent that the problem is not extraneous well-formed characters, but some screwup resetting tty mode during exit, perhaps temporarily messing up speed or parity. I may get around to checking the code one of these days.
jeff@fluke.UUCP (Jeff Stearns) (05/06/85)
In article <4160@mit-eddie.UUCP> smh@mit-eddie.UUCP (Steven M. Haflich) writes: >I too have experienced the problem with GNU EMACS emitting garbage >characters to the tty upon exit. When using a VT100 emulator on a PC >that reports ---BREAK-RECEIVED--- on framing errors, it bacame >apparent that the problem is not extraneous well-formed characters, >but some screwup resetting tty mode during exit, perhaps temporarily >messing up speed or parity. I may get around to checking the code >one of these days. Sounds like a problem I discovered when working with the line printer spoolers. Some device filters do a series of ioctl's to change the tty modes while driving the printer. When finished, they switch 'em back - and a garbage character can be introduced into the tty output queue. It's an honest-to-god character *fragment*. Hang a good serial line analyzer on the line and you'll see the first few bits of the character go by, then WHAM - it gets cut off in mid air. I believe that it happens when the kernel is changing tty modes. I suspect that it may be sending a (premature) reset to the tty interface hardware while the last character is in the midst of dribbling out the UART. I found this problem on both 4.2BSD on a VAX, and on release 1.3 from Sun. I now see hints that Sun may have fixed the problem in their latest release. -- Jeff Stearns (206) 356-5064 John Fluke Mfg. Co. P.O. Box C9090 Everett WA 98043 {uw-beaver,decvax!microsof,ucbvax!lbl-csam,allegra,ssc-vax}!fluke!jeff