There seems to be an issue of ksh forwarding signals to recycled process
ids.
This was reported by a customer while running some script but looks like
the same can be reproduced with the following
steps.
1. ssh into a test machine.
2. Run some background job.
for eg. sleep 10 &
Note the process id.
3. In another terminal, create a process with the process id observed in
step 2.
The attached C file can be used for this.
Run as ./cpid process-id.
4. Forcibly disconnect the ssh connection in terminal 1 by
entering ~.
We'll observe that the process created in step 3 is killed. Looks like ksh
forwards the SIGHUP to the recycled pid.
I've tested this in the 2011-02-08, 2012-08-01 and the latest 2014 version
on Solaris env.
This is also observed in Ubuntu 14(ksh2012-08-01).
I'm currently going through the job control code to figure out why this is
happening.
But as I'm quite new to the code, it might take me a while to figure
something out.
If someone has observed this earlier/investigated this, please let me know.
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <signal.h>
#include <wait.h>
int
main(int argc, char *argv[])
{
pid_t pid, rpid, opid;
int i, status, npid;
if (argc != 2) {
(void) fprintf(stderr, "Usage\n");
exit(1);
}
rpid = atoi(argv[1]);
opid = getpid();
for (;;) {
if ((pid = fork()) == 0) {
(void) setpgrp();
pause();
_exit(0);
}
if (pid == rpid)
break;
(void) kill(pid, SIGKILL);
(void) waitpid(pid, NULL, 0);
if (opid < rpid && pid > rpid)
(void) printf("Unable to create pid %d\n", rpid);
opid = pid;
}
(void) printf("pid %d is ready\n", pid);
i = waitpid(pid, &status, 0);
(void) printf("waitpid return %d, status 0x%4.4x\n", i, status);
return (0);
}
_______________________________________________
ast-developers mailing list
ast-developers@lists.research.att.com
http://lists.research.att.com/mailman/listinfo/ast-developers
