bug#8821: 23.3; spins and eats up memory

bug#8821: 23.3; spins and eats up memory

[-1]

While using emacs in text mode (-nw) on my remote machine (over SSH), I 
lost connectivity. When I logged in later, I found that the emacs, which 
I had left in the background, was spinning and had consumed all of the 
memory. The following gives a basic idea,

$ ps -opid,%cpu,%mem,rss,vsz,args -p `pgrep emacs`
   PID %CPU %MEM   RSS    VSZ COMMAND
31367 95.6 87.6 1807420 3138696 emacs

I first encountered this problem in 23.1 and I was able to confirm it in 
23.3 as well.

Following is the looping strace output,

$ strace -o /tmp/1 -p `pgrep emacs`
$ grep -v gettimeofday /tmp/1 | sort | uniq
brk(0xb7585000)                         = 0xb7564000
brk(0xb7589000)                         = 0xb7564000
ioctl(3, FIONREAD, [-1074857400])       = -1 EIO (Input/output error)
ioctl(3, SNDCTL_TMR_STOP or TCSETSW, {B0 -opost -isig -icanon -echo 
...}) = -1 EIO (Input/output error)
ioctl(3, SNDCTL_TMR_TIMEBASE or TCGETS, 0xbfef13a8) = -1 EIO 
(Input/output error)
kill(31367, SIGHUP)                     = 0
mmap2(NULL, 1048576, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_NORESERVE, 
-1, 0) = -1 ENOMEM (Cannot allocate memory)
mmap2(NULL, 1048576, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, 
-1, 0) = -1 ENOMEM (Cannot allocate memory)
mmap2(NULL, 2097152, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_NORESERVE, 
-1, 0) = -1 ENOMEM (Cannot allocate memory)
rt_sigprocmask(SIG_BLOCK, [IO], [HUP TERM IO], 8) = 0
write(3, "\7", 1)                       = -1 EIO (Input/output error)

Following is the output from pstack,

$ pstack `pgrep emacs`
#1  0x00768e88 in mmap () from /lib/libc.so.6
#2  0x00701644 in new_heap () from /lib/libc.so.6
#3  0x00701e51 in arena_get2 () from /lib/libc.so.6
#4  0x007042c5 in malloc () from /lib/libc.so.6
#5  0x0816e843 in ?? ()
#6  0x0816f6c7 in ?? ()
#7  0x08185acb in ?? ()
#8  0x0813e7fe in ?? ()
#9  0x0813ec9d in ?? ()
#10 0x08141b13 in ?? ()
#11 0x08141b59 in ?? ()
#12 0x0807b0c0 in bind ()
#13 0x081a0616 in ?? ()
#14 0x081a2d3a in ?? ()
#15 0x08120ada in ?? ()
#16 0x08120c29 in ?? ()
#17 0x0818475c in ?? ()
#18 0x081208f5 in ?? ()
#19 0x08184821 in ?? ()
#20 0x08120971 in ?? ()
#21 0x08120d3b in ?? ()
#22 0x08120e62 in ?? ()
#23 0x081171fd in ?? ()
#24 0x006a8bb6 in __libc_start_main () from /lib/libc.so.6
#25 0x08054621 in bind ()

The following are the steps to reproduce the problem,

1. ssh localhost
2. emacs -nw
3. press CTRL-Z
4. press enter
5. disconnect by typing ~.
6. ps aux | grep emacs

Additional details,

$ uname -a
Linux lap.local 2.6.31.5-127.fc12.i686.PAE #1 SMP Sat Nov 7 21:25:57 EST 
2009 i686 i686 i386 GNU/Linux

$ cat /etc/redhat-release
Fedora release 12 (Constantine)

Lemme know if you need further details.

-- 
-1

bug#8821: 23.3; spins and eats up memory

[Chong Yidong]

> While using emacs in text mode (-nw) on my remote machine (over SSH),
> I lost connectivity. When I logged in later, I found that the emacs,
> which I had left in the background, was spinning and had consumed all
> of the memory.
>
> I first encountered this problem in 23.1 and I was able to confirm it
> in 23.3 as well.

Strange: this bug report sounds like Bug#4970/Bug#7951, but that is
supposed to be fixed in 23.3.

bug#8821: 23.3; spins and eats up memory

[-1]

>
> Strange: this bug report sounds like Bug#4970/Bug#7951, but that is
> supposed to be fixed in 23.3.

I just installed 23.3b and this bug is indeed fixed.

$ ssh localhost
Last login: Thu Dec  1 15:01:33 2011 from lap.local

$ emacs --version
GNU Emacs 23.3.1
Copyright (C) 2011 Free Software Foundation, Inc.
GNU Emacs comes with ABSOLUTELY NO WARRANTY.
You may redistribute copies of Emacs
under the terms of the GNU General Public License.
For more information about these matters, see the file named COPYING.

$ emacs -nw

[1]+  Stopped                 emacs -nw
$
$ Connection to localhost closed.
$ pgrep emacs
$

I must have made a mistake last time around when I installed 23.3.

-- 
-1

bug#8821: 23.3; spins and eats up memory

[Chong Yidong]

-1 <-1@0throot.com> writes:

> I just installed 23.3b and this bug is indeed fixed.

Thanks for checking.  Closing the bug.