Apache OpenOffice (AOO) Bugzilla – Issue 6431
Program hangs after splash screen until lpc process killed
Last modified: 2004-03-15 14:11:19 UTC
Running on Linux (Red Hat 7.2), installed OpenOffice and used it successfully. Had system crashes (was running in 1600x1200 resolution, so probably crashes in XFree86). Could run in 1280x1024 mode quite reliably, but the program wouldn't start up - it would just hang at the splash screen. Installed some newer drivers for X. Tried again in 1600x1200 mode. Noticed a strange thing - an lpc status process that seemed idle. When that process was killed, the OpenOffice splash screen vanished and the program started up. I see only this occurrence of lpc in the binaries when I do a "strings" on everything: /opt/OpenOffice.org1.0/program/libpsp641li.so /usr/sbin/lpc status lpc status $ ps al F UID PID PPID PRI NI VSZ RSS WCHAN STAT TTY TIME COMMAND 000 126 20421 20409 13 0 2692 1440 wait4 S pts/4 0:00 -sh 000 126 16893 20421 9 0 82560 32916 pipe_w S pts/4 0:05 /opt/OpenOffice.org1.0/program/soffice.bin 040 126 16960 16893 9 0 82560 32916 do_pol S pts/4 0:00 /opt/OpenOffice.org1.0/program/soffice.bin 040 126 16961 16960 9 0 82560 32916 nanosl S pts/4 0:00 /opt/OpenOffice.org1.0/program/soffice.bin 040 126 16962 16960 9 0 82560 32916 wait_f S pts/4 0:00 /opt/OpenOffice.org1.0/program/soffice.bin 040 126 16963 16960 9 0 82560 32916 nanosl S pts/4 0:00 /opt/OpenOffice.org1.0/program/soffice.bin 040 126 16965 16960 9 0 82560 32916 do_pol S pts/4 0:00 /opt/OpenOffice.org1.0/program/soffice.bin 000 126 16966 16893 9 0 2740 1196 tcp_da S pts/4 0:00 /usr/sbin/lpc status 000 126 16987 20421 19 0 3284 1356 - R pts/4 0:00 ps al Seems like something is blocked on I/O. luke
SBA->TM: "Framework"
TM->HR/CP: Any ideas ?
HR->TM: PL might have any idea
Ooo tries "lpc status" on startup to discover which print queues are available. If the lpc command hangs, then Ooo hangs also, because it waits for lpc's output. The question is why does it hang ? Does this also happen via the normal command line ? Ooo tries a popen("lpc status", "r") which essentially starts '/bin/sh -c "lpc status"' and passes its output to Ooo via pipe.
Hmm, our sysadmins aren't in yet, but as far as I know we don't use that printing environment here. If I do lpc from the command line, it outputs octal 0377 and exits with a status of 0. Are you trying to read more bytes than that, and blocking on input? (Sounds unlikely.) The problem has stopped of its own (after being perfectly repeatable for a week or so). So, perhaps something has enabled lpc to return, where previously it was hanging. Anyway, assuming that lpc status is going to work even on systems that use other printing systems, may be assuming too much. I'll find out more after our Unix sysadmins get in. luke
Problem has been reproducible and will be fixed in a staroffice 6.1beta and a corresponding OO version.
.
Too late for beta2, re-targeted to 1.1 RC.
TM->PL: Possible solution might be to build in a timer within office start. As discussed via phone, please have a look, thanks !
Will build in a timeout for the "popen"; this will let OOo continue, but without automatic print queues. I guess the next issue in this regard will then be that startup takes too long :-)
*** Issue 14035 has been marked as a duplicate of this issue. ***
moved print command evaluation into a separate thread so it does not slow down office startup; this means that system print queues and print command (that is the default print command) will potentially never be available - which is a probably acceptable if the print system doesn't work anyway. fixed in CWS vcl11
Review done!
pl->us: you can test this by delaying lpc/lpstat artifially; e.g. place this shell script (on Solaris) into the program directory, name it lpstat and make it executable: <----- snip #!/bin/sh echo delaying $0 1>&2 sleep 30 echo wakeup $0 1>&2 /usr/bin/lpstat $* <----- snip This should not delay office startup anymore and after the wakeup additional printer should be available.
Okay, I've added that command early in my PATH, and if the problem recurs, I'll pass on the results. luke
Verified in cws vcl11. Setting to fixed.
Setting to verified.
OK in OOo 1.1 RC2 (srx645_m13s2.8662) Closing issue.
*** Issue 17322 has been marked as a duplicate of this issue. ***
*** Issue 25550 has been marked as a duplicate of this issue. ***
*** Issue 26191 has been marked as a duplicate of this issue. ***