Hi, due to my limited hardware equipment I implemented a mechanism that automatically detects X server deadlocks and recovers from them as long as the Kernel isn't locked either. It consists of a C programme which detects the lockup and a shell script which starts the xserver and uses kbd_mode and savetextmode/textmode to make the system usable again after a lockup. This means the script must run with superuser privileges. However, it will try to make X clients run with a normal UID taken from $LOGNAME or $SUDO_USER.
The lockup detection uses two processes communicating through a pipe.
One of them performs dummy X operations and sends a heart beat through
the pipe. It will freeze when the X server locks. The other process
receives the heart beats and detects timeouts. It indicates a lockup in
the return value which is used in the shell script.
The script and the C programme are attached. The script is currently
configured for my setup. Changing it is just a matter of changing 5
variables.
Regards,
Felix
--
__\|/__ ___ ___ ___
__Tschüß_______\_6 6_/___/__ \___/__ \___/___\___You can do anything,___
_____Felix_______\Ä/\ \_____\ \_____\ \______U___just not everything____
[EMAIL PROTECTED] >o<__/ \___/ \___/ at the same time!
/* compile with:
* gcc -o xpoll -lX11 xpoll.c
*/
#include <X11/Xlib.h>
#include <stdio.h>
#include <unistd.h>
#include <fcntl.h>
#include <errno.h>
#include <sys/wait.h>
/* messages sent through the pipe */
#define NO_SERVER 'n'
#define HEART_BEAT 'h'
/* return codes */
#define NORMAL_EXIT 0
#define LOCKUP 1
#define NO_CONNECT 2
#define OTHER_ERR 3
/* constants */
#define MAX_TRY_OPEN 10
#define BEAT_INTERVAL 2
#define CHECK_INTERVAL 1
#define TIMEOUT 4
/* file descriptors of the pipeline */
int stethoscope[2];
void poll_x () {
Display *display;
Window root_win;
int i;
char buffer[1];
/* try to open display for MAX_TRY_OPEN seconds */
for (i = 0; i <= MAX_TRY_OPEN; ++i) {
if ((display = XOpenDisplay (NULL)))
break;
sleep (CHECK_INTERVAL);
}
if (display == NULL) {
/* signal "no connect" to parent */
buffer[0] = NO_SERVER;
write (stethoscope[1], buffer, 1);
close (stethoscope[1]);
return;
}
root_win = XDefaultRootWindow (display);
while (1) {
XWindowAttributes attr;
/* use XGetWindowAttributes to create some activity without actually
* drawing something */
XGetWindowAttributes (display, root_win, &attr);
/* if it returns, X is not locked and we send a heart beat */
buffer[0] = HEART_BEAT;
write (stethoscope[1], buffer, 1);
sleep (BEAT_INTERVAL);
}
/* will never get here */
}
int monitor_child () {
/* counts the number of CHECK_INTERVALs in which there was no heart beat
* give some extra time in the beginning while the X server starts up */
int time = -MAX_TRY_OPEN;
while (1) {
int r;
char buffer[1];
/* try to read something from the non-blocking pipe */
r = read (stethoscope[0], buffer, 1);
if (r == -1 && errno != EAGAIN) {
perror ("read");
return OTHER_ERR;
} else if (r == 0 || (r == -1 && errno == EAGAIN)) {
if (time++ == TIMEOUT)
break; /* timeout expired */
sleep (CHECK_INTERVAL);
} else {
time = 0;
if (buffer[0] == NO_SERVER)
return NO_CONNECT;
else if (buffer[0] != HEART_BEAT) {
fprintf (stderr, "unexpected message\n");
return OTHER_ERR;
}
}
}
/* timeout expired, either child died (X server exit) or lockup */
return NORMAL_EXIT;
}
int main () {
pid_t pid;
int flags, r;
/* make a pipe */
if (pipe (stethoscope) == -1) {
perror ("pipe");
return OTHER_ERR;
}
/* make the reading end non-blocking */
if ((flags = fcntl (stethoscope[0], F_GETFL, 0)) == -1) {
perror ("fcntl");
return OTHER_ERR;
}
if (fcntl (stethoscope[0], F_SETFL, flags | O_NONBLOCK) == -1) {
perror ("fcntl");
return OTHER_ERR;
}
/* fork a child process */
pid = fork ();
if (pid == 0) {
close (stethoscope[0]);
/* child will poll the x-server */
poll_x ();
return 0;
} else if (pid > 0) {
close (stethoscope[1]);
/* parent will monitor the child */
r = monitor_child ();
} else {
perror ("fork");
return OTHER_ERR;
}
if (r == NORMAL_EXIT) {
/* if child completed then it's really a normal exit
* otherwise it's a lockup */
pid_t rpid = waitpid (pid, NULL, WNOHANG);
if (rpid == -1) {
perror ("waitpid");
return OTHER_ERR;
} else if (rpid == 0) {
fprintf (stderr, "X server locked up\n");
/* ok, lockup */
r = LOCKUP;
} else
fprintf (stderr, "X server exited normally\n");
} else {
/* wait for child to complete */
if (waitpid (pid, NULL, 0) == -1) {
perror ("waitpid");
return OTHER_ERR;
}
if (r == NO_CONNECT)
fprintf (stderr, "Could not connect to an X server\n");
}
return r;
}
glserver
Description: application/shellscript
