Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 18 Jul 2002 15:41:26 +0100 (BST)
From:      Gavin Atkinson <gavin@ury.york.ac.uk>
To:        <freebsd-ports@freebsd.org>
Cc:        <cy@freebsd.org>
Subject:   screen(1) issues with 4.6?
Message-ID:  <Pine.BSF.4.33.0207181023030.19954-100000@ury.york.ac.uk>

next in thread | raw e-mail | index | archive | help

Hi,

(This may be same as PR 39479, so cy@ cc'd as he is owner of that PR)

I'm seeing some strange problems with a couple of machine running
4.6-RELEASE and screen(1) installed as a package from ports. Installing
from source via ports works fine.

It seems that screen will use as much processor power as it can get hold
on. Worse still, this seems to be mostly system calls, thus making the
system pretty sluggish...

last pid: 61504;  load averages:  1.00,  1.00,  1.00    up 0+11:50:05  10:25:19
26 processes:  4 running, 22 sleeping
CPU states: 19.0% user,  0.0% nice, 81.0% system,  0.0% interrupt,  0.0% idle
Mem: 15M Active, 8352K Inact, 15M Wired, 5216K Cache, 14M Buf, 15M Free
Swap: 600M Total, 80K Used, 600M Free

  PID USERNAME PRI NICE  SIZE    RES STATE    TIME   WCPU    CPU COMMAND
  147 gavin     54   0  1720K  1084K RUN    635:51 98.83% 98.83% screen
  101 root       2   0  1800K  1100K select   0:04  0.00%  0.00% sendmail
61382 root      28   0  1716K  1204K RUN      0:03  0.00%  0.00% ppp

using ktrace reveals that the program is stuck in this tight loop:

 147 screen   CALL  seteuid(0x3e8)
 147 screen   RET   seteuid 0
 147 screen   CALL  setegid(0x3e8)
 147 screen   RET   setegid 0
 147 screen   CALL  open(0x8089ca0,0x4,0)
 147 screen   NAMI  "/tmp/screens/S-gavin/147.ttyp0.endor"
 147 screen   RET   open 4
 147 screen   CALL  seteuid(0)
 147 screen   RET   seteuid 0
 147 screen   CALL  setegid(0x3e8)
 147 screen   RET   setegid 0
 147 screen   CALL  select(0x400,0xbfbff440,0xbfbff3c0,0,0)
 147 screen   RET   select 1
 147 screen   CALL  fcntl(0x4,0x4,0)
 147 screen   RET   fcntl 0
 147 screen   CALL  read(0x4,0x8088480,0xc30)
 147 screen   GIO   fd 4 read 0 bytes
       ""
 147 screen   RET   read 0
 147 screen   CALL  close(0x4)
 147 screen   RET   close 0

So my suspition is that select() is returning when it shouldn't. As I say,
this only affects two machines, both are running 4.6-RELEASE. I am in the
process of upgrading one to -STABLE, so i'll reply to say if it fixed the
problem, though it does look as if it is the binary package at fault. I
can give shell access to somebody to the box I am about to upgrade to
-STABLE, if nobody else can recreate the problem. In the mean-time, is it
possible to delete the binary package from the FTP servers?

Gavin


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-ports" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.33.0207181023030.19954-100000>