Skip to contentSkip to conversation

unkillable process

DISCUSSION Albert Cahalan 4 participants 4 messages Oct 14, 2004

lore
Share & tools
lore
#0 lore raw

It's really bad when a task group leader exits.
The process becomes unkillable.

This is with the 2.6.8-rc1 kernel. I haven't seen
any mention of this getting fixed since then.
Here's the top of the /proc/*/status file:

Name:   a.out
State:  Z (zombie)
SleepAVG:       59%
Tgid:   9662
Pid:    9662
PPid:   1
TracerPid:      0
Uid:    1000    1000    1000    1000
Gid:    1000    1000    1000    1000
FDSize: 0
Groups: 500 1000 
Threads:        9

Here's the code:

/////////////////////////////////////////////////////////////// #include <sys/types.h>
#include <unistd.h>
#include <signal.h>
#include <stdio.h>
#include <sched.h>

#ifndef CLONE_THREAD
#define CLONE_THREAD 0x00010000
#endif
#ifndef CLONE_DETACHED
#define CLONE_DETACHED 0x00400000
#endif
#ifndef CLONE_STOPPED
#define CLONE_STOPPED 0x02000000
#endif

#define FLAGS (CLONE_FS|CLONE_FILES|CLONE_SIGHAND|CLONE_VM|CLONE_THREAD|CLONE_DETACHED)

static pid_t one;

static void die(int signo){
(void)signo;
_exit(0);
}

static void hang(void){
for(;;) pause();
}

static int clone_fn(void *vp){
(void)vp;
hang();
return 0; // keep gcc happy
}

static long clone_stack_data[2048];
#ifdef __hppa__
static long *clone_stack = &clone_stack_data[0]; #else
static long *clone_stack = &clone_stack_data[2048]; #endif

int main(int argc, char *argv[]){
pid_t minime;
int i = 8;
(void)argc;
(void)argv;

  one = getpid();
  signal(SIGHUP,die);
  if(fork()) hang();    // parent later killed as readyness signal

  while(i--){
    // better be stopped... they share a stack
    minime = clone(clone_fn, clone_stack, FLAGS | CLONE_STOPPED, NULL);
    if(minime==-1){
      perror("no clone");
      kill(one,SIGKILL);
      _exit(8);
    }
  }

kill(one,SIGHUP); // let the shell know we're ready

_exit(0); // make task group leader a zombie
return 0; // keep gcc happy
} /////////////////////////////////////////////////////////////////////

#1 lore raw
Quoted 5 lines · Albert Cahalan

It's really bad when a task group leader exits.
The process becomes unkillable.

This is with the 2.6.8-rc1 kernel.

That's a pretty old kernel.

Quoted 2 lines

...
Here's the code:

I can't get it to misbehave with current -linus. Can you upgrade and retest?

#2 lore raw
In reply to Albert Cahalan
Quoted 2 lines · Albert Cahalan

It's really bad when a task group leader exits.
The process becomes unkillable.

I have been having zombie problems since 2.6.9-rc1. I run a boinc climateprediction program (related to seti@home) which leaves defunct "cp" processes about. Killing the climatepredictor (called hadsm3um_4.03_i686-pc-linux-gnu) which spawns them causes these zombie cp things to get reaped.

Quoted 3 lines

This is with the 2.6.8-rc1 kernel. I haven't seen
any mention of this getting fixed since then.
Here's the top of the /proc/*/status file:

I tried it with 2.6.9-rc3 just now and it doesn't make zombies for me.

climateprediction still makes defunct cp.

(I fired up 2.6.9-rc4 but it somehow wouldn't load the driver for my ethernet 3c59x. That's another issue, but I have no idea if the problem has been fixed there since I am stopped by another problem.)

I skimmed over the changelogs but I have found anything looking like a change in this area. I am not sure what the right keyword(s) to search for on this topic would be. I didn't grovel through them yet, but perhaps someone on the list knows what is going on.

Signature · 1 line
Johan KULLSTAM
#3 lore raw
Quoted 10 lines · Johan Kullstam
Quoted 2 lines · Albert Cahalan

It's really bad when a task group leader exits.
The process becomes unkillable.

I have been having zombie problems since 2.6.9-rc1. I run a boinc climateprediction program (related to seti@home) which leaves defunct "cp" processes about. Killing the climatepredictor (called hadsm3um_4.03_i686-pc-linux-gnu) which spawns them causes these zombie cp things to get reaped.

I believe this is not related. Just a bug in the program missing SIGCHLD and not calling waitpid.