sitongia@hao.ucar.edu (Leonard Sitongia) (05/10/91)
I've been using an NCD X-19 terminal with xdm from MIT X11R4 displaying cmdtool from OpenWindows 2.0 served from a Sun-4/280 running SunOS 4.0.3 for some time. It is simply exec'd from the .openwin-menu. After I upgraded the Sun to SunOS 4.1, I have been unable to run the cmdtool on the Sun to display directly on the X terminal. If I exec a cmdtool which rlogins to the same Sun (in other words the same cmdtool which fails but in this case does a rlogin to the same system that the cmdtool is running on), it suceeds (using two ttyp?'s, though). The failing cmdtool appears normal, in that it opens ok, and echos input, but will not execute it. In any case, when the cmdtool is started up, the xdm-errors reports: ttysw-TIOCSPGRP: Interrupted system call The only discernible difference is from tracing the two cmdtools. In the case of the cmdtool which is rlogin'd, here is how it reads (I typed in the letter "a"): select (256, 0xf7fff5f0, 0xf7fff5d0, 0xf7fff5b0, 0xf7746894) = 0 gettimeofday (0x197d4, 0) = 0 write (3, "".., 168) = 168 ioctl (3, 0x4004667f, 0xf7fff2dc) = 0 gettimeofday (0xf7fff4b0, 0) = 0 setitimer (0, 0xf7fff420, 0) = 0 select (256, 0xf7fff5f0, 0xf7fff5d0, 0xf7fff5b0, 0) = 1 ioctl (3, 0x4004667f, 0xf7ffef4c) = 0 read (3, "".., 32) = 32 ioctl (3, 0x4004667f, 0xf7ffefe4) = 0 write (3, "".., 92) = 92 ioctl (3, 0x4004667f, 0xf7fff2dc) = 0 select (256, 0xf7fff5f0, 0xf7fff5d0, 0xf7fff5b0, 0) = 1 write (4, "a", 1) = 1 ioctl (3, 0x4004667f, 0xf7fff2dc) = 0 select (256, 0xf7fff5f0, 0xf7fff5d0, 0xf7fff5b0, 0) = 2 readv (4, 0xf774bd70, 2) = 2 gettimeofday (0xa9a0, 0) = 0 ioctl (3, 0x4004667f, 0xf7ffef4c) = 0 read (3, "".., 32) = 32 ioctl (3, 0x4004667f, 0xf7ffefe4) = 0 write (3, "".., 76) = 76 ioctl (3, 0x4004667f, 0xf7fff2dc) = 0 gettimeofday (0xf7fff4b0, 0) = 0 setitimer (0, 0xf7fff420, 0) = 0 select (256, 0xf7fff5f0, 0xf7fff5d0, 0xf7fff5b0, 0xf7746894) = But the cmdtool which is simply exec'd is doing this (again, I typed in "a"): select (256, 0xf7fff5f8, 0xf7fff5d8, 0xf7fff5b8, 0) = 1 ioctl (3, 0x4004667f, 0xf7ffef54) = 0 read (3, "".., 32) = 32 write (3, "".., 96) = 96 read (3, 0xf7ffdba4, 32) = -1 EWOULDBLOCK (Operation would block) select (4, 0xf7ffda70, 0, 0, 0) = 1 read (3, "".., 32) = 32 read (3, "".., 28) = 28 read (3, "".., 168) = 168 read (3, "".., 2688) = 1820 read (3, "".., 868) = 868 ioctl (3, 0x4004667f, 0xf7ffefec) = 0 write (3, "".., 188) = 188 ioctl (3, 0x4004667f, 0xf7fff2e4) = 0 read (3, "".., 32) = 32 ioctl (3, 0x4004667f, 0xf7fff1fc) = 0 ioctl (3, 0x4004667f, 0xf7fff2e4) = 0 select (256, 0xf7fff5f8, 0xf7fff5d8, 0xf7fff5b8, 0) = Can someone give me some insight into what is going wrong? Thanks. -- -Leonard E. Sitongia sitongia@ncar.ucar.edu System Manager (303) 497-1509 High Altitude Observatory P.O. Box 3000 Boulder CO 80307
fpb@ittc.wec.com (Frank P. Bresz) (05/13/91)
In article <11353@ncar.ucar.edu> sitongia@hao.ucar.edu (Leonard Sitongia) writes: >I've been using an NCD X-19 terminal with xdm from MIT X11R4 displaying >cmdtool from OpenWindows 2.0 served from a Sun-4/280 running SunOS 4.0.3 for >some time. It is simply exec'd from the .openwin-menu. >After I upgraded the Sun to SunOS 4.1, I have been unable to run the >cmdtool on the Sun to display directly on the X terminal. If I exec a >cmdtool which rlogins to the same Sun (in other words the same cmdtool >which fails but in this case does a rlogin to the same system that the >cmdtool is running on), it suceeds (using two ttyp?'s, though). >The failing cmdtool appears normal, in that it opens ok, and echos input, >but will not execute it. >In any case, when the cmdtool is started up, the xdm-errors reports: >ttysw-TIOCSPGRP: Interrupted system call >The only discernible difference is from tracing the two cmdtools. [ trace deleted] >Can someone give me some insight into what is going wrong? I can tell you what seems to be the problem, and offer a solution. We saw something very similar. It seems that Sun in it's Xview stuff must have the process group set, or some such nonsense. All of our GUIDE stuff which generates Xview code has the same problem. If I run 'xterm' I don't have this problem. Who knows if Sun will ever make X things work in nice conjunction with MIT X, I say this because I notice this when I use XDM. Best solution is to just use xterm it is a better utility anyway. I mean at least you can do file completion in a scrollable window. Come on Sun this bug is so old the cost of formaldehyde needed to keep it there must be phenomenal. -- | () () () | Frank P. Bresz | Westinghouse Electric Corporation | \ /\ / | fpb@ittc.wec.com | ITTC Simulators Department | \/ \/ | uunet!ittc!fpb | Those who can, do. Those who can't, simulate. | ---------- | +1 412 733 6749 | My opinions are mine, WEC don't want 'em.