in the morning i did emerge rsync emerge -u world emerge -pu world told me it would update ... 4 packages, i think - not including screen. when i came back to the computer, it behaved strange on creating a new virtual screen in 'screen' - cpu usage went up to 100% - and it took about 3 seconds for the new virtual screen to be created. this even held on after restarting and so on, i did an strace screen and found the following: close(3) = 0 open("/lib/libc.so.6", O_RDONLY) = 3 read(3, "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0\200\202"..., 1024) = 10 24 fstat64(3, {st_mode=S_IFREG|0755, st_size=1359445, ...}) = 0 old_mmap(NULL, 1215136, PROT_READ|PROT_EXEC, MAP_PRIVATE, 3, 0) = 0x40098000 mprotect(0x401b7000, 39584, PROT_NONE) = 0 old_mmap(0x401b7000, 24576, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED, 3, 0x11 e000) = 0x401b7000 old_mmap(0x401bd000, 15008, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_ANON YMOUS, -1, 0) = 0x401bd000 close(3) = 0 munmap(0x40016000, 33792) = 0 getrlimit(0x7, 0xbfffe5b8) = 0 close(1023) = -1 EBADF (Bad file descriptor) close(1022) = -1 EBADF (Bad file descriptor) close(1021) = -1 EBADF (Bad file descriptor) close(1020) = -1 EBADF (Bad file descriptor) close(1019) = -1 EBADF (Bad file descriptor) close(1018) = -1 EBADF (Bad file descriptor) close(1017) = -1 EBADF (Bad file descriptor) and so on, down to close(8) = -1 EBADF (Bad file descriptor) close(7) = -1 EBADF (Bad file descriptor) close(6) = -1 EBADF (Bad file descriptor) close(5) = -1 EBADF (Bad file descriptor) close(4) = -1 EBADF (Bad file descriptor) close(3) = -1 EBADF (Bad file descriptor) those files were never opened. i guess it has something to do with a library update from the portage tree, don't remeber what the library was called but i think there was something including 'utils' or 'tool' in it's name. this is an athlon-tbird processor, i am running a 2.4.18-kernel.
just wanted to 'w3m http://www.gentoo.org' in screen, the following was in the first line: == ERROR ===: [/var/tmp/portage/gpm-1.20.0-r3/work/gpm-1.20.0/src/liblow.c(261)]: Gentoo L | Graphics | ScreenShots | Articles | Install: x86 / PowerPC/ strncmp/isdigit/consolename failedumentation | Changelog | Bugs | Projects | this also didn't happen before.
(for the screen problem:) it doesn't even have to do with those close()-calls... i'm not sure about what's the matter, 'strace -t -f screen' didn't help me, sorry. maybe you can figure out.
I'm getting the same problem since update.
someone posted to gentoo-users that removing /etc/termcap will stop the problem, as the time seems to be taken by parsing that file... I tried this and it worked for me. The question is, will there be any negative effects by removing /etc/termcap?
It seems removing termcap does help, but... I took a termcap file (size: 702559, aprox. equal to the one from gentoo) from a mandrake 8.0 system, (termcap-11.0.1-4mdk) where screen (screen-3.9.8-3mdk) starts instantly and put it on my gentoo system and screen is still slow. The question is: why in the strace there is no sign of screen opening termcap and why is screen trying to close 1021 file descriptors ? (taken from grep EBADF typescript |wc -l)
Not a clue and not really up my ally.
NAz, crack at this?
strangely enough, i'm no longer seeing this after installing with gentoo 1.4_rc1
I've had this bug assigned to me for a bit, and allthough I've been quiet in this thread I have been trying to test it under every condition I can think of. I haven't been able to produce any errors similar to this, or otherwise. Testing has been done on three 1.2 machines and one 1.4_rc1 machine. Removing the /etc/termcap file doesn't really seem to be a good longterm fix, as that file is used in many different ways. Also, I think that it being fixed on a fresh install shouldn't really be so suprising. I think it's related more to a shift in libraries on a currently installed system. That makes it that much harder to track down. I'm still "stress testing" screen to make it break.
Umm.. Ok. I'm still the big screen user I've allways been, and I'm still not able to reproduce anything remotely similar to what's been reported. Lately I've tried angles such as gpm and ncurses, but haven't been able to dig up issues there. Is anyone still having these issues??
It's coming up on a week since my last post. Besides the initial reporter there are three of you on the CC list. If there's no response that any of you are still having this issue with screen this bug will be closed on Oct. 31. BOO!
I'm declaring this bug dead.