uml: stop gdb from deleting breakpoints when running UML

Sometimes when UML is debugged gdb miss breakpoints.

When process traced by gdb do fork, debugger remove breakpoints from
child address space. There is possibility to trace more than one fork,
but this not work with UML, I guess (only guess) there is a deadlock -
gdb waits for UML and UML waits for gdb.

When clone() is called with SIGCHLD and CLONE_VM flags, gdb see this
as PTRACE_EVENT_FORK not as PTRACE_EVENT_CLONE and remove breakpoints
from child and at the same time from traced process, because either
have the same address space.

Maybe it is possible to do fix in gdb, but I'm not sure if there is
easy way to find out if traced and child processes share memory. So I
do fix for UML, it simply do not call clone() with both SIGCHLD and
CLONE_VM flags together.  Additionally __WALL flag is used for
waitpid() to assure not miss clone and normal process events.

[ jdike - checkpatch fixes ]
Signed-off-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

arch/um/drivers/net_user.c

@@ -201,7 +201,7 @@ static int change_tramp(char **argv, char *output, int output_len)
 	close(fds[1]);
 
 	if (pid > 0)
-		CATCH_EINTR(err = waitpid(pid, NULL, 0));
+		helper_wait(pid, 0, "change_tramp");
 	return pid;
 }
 

arch/um/drivers/slip_user.c

@@ -77,7 +77,7 @@ static int slip_tramp(char **argv, int fd)
 {
 	struct slip_pre_exec_data pe_data;
 	char *output;
-	int status, pid, fds[2], err, output_len;
+	int pid, fds[2], err, output_len;
 
 	err = os_pipe(fds, 1, 0);
 	if (err < 0) {
@@ -109,15 +109,7 @@ static int slip_tramp(char **argv, int fd)
 	read_output(fds[0], output, output_len);
 	printk("%s", output);
 
-	CATCH_EINTR(err = waitpid(pid, &status, 0));
-	if (err < 0)
-		err = errno;
-	else if (!WIFEXITED(status) || (WEXITSTATUS(status) != 0)) {
-		printk(UM_KERN_ERR "'%s' didn't exit with status 0\n", argv[0]);
-		err = -EINVAL;
-	}
-	else err = 0;
-
+	err = helper_wait(pid, 0, argv[0]);
 	close(fds[0]);
 
 out_free:

arch/um/drivers/slirp_user.c

@@ -79,7 +79,7 @@ static int slirp_open(void *data)
 static void slirp_close(int fd, void *data)
 {
 	struct slirp_data *pri = data;
-	int status,err;
+	int err;
 
 	close(fd);
 	close(pri->slave);
@@ -98,18 +98,9 @@ static void slirp_close(int fd, void *data)
 		       "(%d)\n", pri->pid, errno);
 	}
 #endif
-
-	CATCH_EINTR(err = waitpid(pri->pid, &status, WNOHANG));
-	if (err < 0) {
-		printk(UM_KERN_ERR "slirp_close: waitpid returned %d\n", errno);
-		return;
-	}
-
-	if (err == 0) {
-		printk(UM_KERN_ERR "slirp_close: process %d has not exited\n",
-		       pri->pid);
+	err = helper_wait(pri->pid, 1, "slirp_close");
+	if (err < 0)
 		return;
-	}
 
 	pri->pid = -1;
 }

arch/um/drivers/ubd_user.c

@@ -49,8 +49,7 @@ int start_io_thread(unsigned long sp, int *fd_out)
 		goto out_close;
 	}
 
-	pid = clone(io_thread, (void *) sp, CLONE_FILES | CLONE_VM | SIGCHLD,
-		    NULL);
+	pid = clone(io_thread, (void *) sp, CLONE_FILES | CLONE_VM, NULL);
 	if(pid < 0){
 		err = -errno;
 		printk("start_io_thread - clone failed : errno = %d\n", errno);

arch/um/include/os.h

@@ -214,7 +214,7 @@ extern int execvp_noalloc(char *buf, const char *file, char *const argv[]);
 extern int run_helper(void (*pre_exec)(void *), void *pre_data, char **argv);
 extern int run_helper_thread(int (*proc)(void *), void *arg,
 			     unsigned int flags, unsigned long *stack_out);
-extern int helper_wait(int pid);
+extern int helper_wait(int pid, int nohang, char *pname);
 
 
 /* tls.c */

arch/um/os-Linux/aio.c

@@ -218,7 +218,7 @@ static int init_aio_24(void)
 		goto out_close_pipe;
 
 	err = run_helper_thread(not_aio_thread, NULL,
-				CLONE_FILES | CLONE_VM | SIGCHLD, &aio_stack);
+				CLONE_FILES | CLONE_VM, &aio_stack);
 	if (err < 0)
 		goto out_close_pipe;
 
@@ -254,7 +254,7 @@ static int init_aio_26(void)
 	}
 
 	err = run_helper_thread(aio_thread, NULL,
-				CLONE_FILES | CLONE_VM | SIGCHLD, &aio_stack);
+				CLONE_FILES | CLONE_VM, &aio_stack);
 	if (err < 0)
 		return err;
 

arch/um/os-Linux/drivers/ethertap_user.c

@@ -94,7 +94,7 @@ static int etap_tramp(char *dev, char *gate, int control_me,
 		      int control_remote, int data_me, int data_remote)
 {
 	struct etap_pre_exec_data pe_data;
-	int pid, status, err, n;
+	int pid, err, n;
 	char version_buf[sizeof("nnnnn\0")];
 	char data_fd_buf[sizeof("nnnnnn\0")];
 	char gate_buf[sizeof("nnn.nnn.nnn.nnn\0")];
@@ -131,13 +131,7 @@ static int etap_tramp(char *dev, char *gate, int control_me,
 	}
 	if (c != 1) {
 		printk(UM_KERN_ERR "etap_tramp : uml_net failed\n");
-		err = -EINVAL;
-		CATCH_EINTR(n = waitpid(pid, &status, 0));
-		if (n < 0)
-			err = -errno;
-		else if (!WIFEXITED(status) || (WEXITSTATUS(status) != 1))
-			printk(UM_KERN_ERR "uml_net didn't exit with "
-			       "status 1\n");
+		err = helper_wait(pid, 0, "uml_net");
 	}
 	return err;
 }

arch/um/os-Linux/drivers/tuntap_user.c

@@ -107,7 +107,7 @@ static int tuntap_open_tramp(char *gate, int *fd_out, int me, int remote,
 		       "errno = %d\n", errno);
 		return err;
 	}
-	CATCH_EINTR(waitpid(pid, NULL, 0));
+	helper_wait(pid, 0, "tuntap_open_tramp");
 
 	cmsg = CMSG_FIRSTHDR(&msg);
 	if (cmsg == NULL) {

arch/um/os-Linux/helper.c

@@ -76,7 +76,7 @@ int run_helper(void (*pre_exec)(void *), void *pre_data, char **argv)
 	data.fd = fds[1];
 	data.buf = __cant_sleep() ? kmalloc(PATH_MAX, UM_GFP_ATOMIC) :
 					kmalloc(PATH_MAX, UM_GFP_KERNEL);
-	pid = clone(helper_child, (void *) sp, CLONE_VM | SIGCHLD, &data);
+	pid = clone(helper_child, (void *) sp, CLONE_VM, &data);
 	if (pid < 0) {
 		ret = -errno;
 		printk("run_helper : clone failed, errno = %d\n", errno);
@@ -101,7 +101,7 @@ int run_helper(void (*pre_exec)(void *), void *pre_data, char **argv)
 			ret = n;
 			kill(pid, SIGKILL);
 		}
-		CATCH_EINTR(waitpid(pid, NULL, 0));
+		CATCH_EINTR(waitpid(pid, NULL, __WCLONE));
 	}
 
 out_free2:
@@ -126,7 +126,7 @@ int run_helper_thread(int (*proc)(void *), void *arg, unsigned int flags,
 		return -ENOMEM;
 
 	sp = stack + UM_KERN_PAGE_SIZE - sizeof(void *);
-	pid = clone(proc, (void *) sp, flags | SIGCHLD, arg);
+	pid = clone(proc, (void *) sp, flags, arg);
 	if (pid < 0) {
 		err = -errno;
 		printk("run_helper_thread : clone failed, errno = %d\n",
@@ -134,7 +134,7 @@ int run_helper_thread(int (*proc)(void *), void *arg, unsigned int flags,
 		return err;
 	}
 	if (stack_out == NULL) {
-		CATCH_EINTR(pid = waitpid(pid, &status, 0));
+		CATCH_EINTR(pid = waitpid(pid, &status, __WCLONE));
 		if (pid < 0) {
 			err = -errno;
 			printk("run_helper_thread - wait failed, errno = %d\n",
@@ -150,14 +150,30 @@ int run_helper_thread(int (*proc)(void *), void *arg, unsigned int flags,
 	return pid;
 }
 
-int helper_wait(int pid)
+int helper_wait(int pid, int nohang, char *pname)
 {
-	int ret;
+	int ret, status;
+	int wflags = __WCLONE;
 
-	CATCH_EINTR(ret = waitpid(pid, NULL, WNOHANG));
+	if (nohang)
+		wflags |= WNOHANG;
+
+	if (!pname)
+		pname = "helper_wait";
+
+	CATCH_EINTR(ret = waitpid(pid, &status, wflags));
 	if (ret < 0) {
-		ret = -errno;
-		printk("helper_wait : waitpid failed, errno = %d\n", errno);
-	}
-	return ret;
+		printk(UM_KERN_ERR "%s : waitpid process %d failed, "
+		       "errno = %d\n", pname, pid, errno);
+		return -errno;
+	} else if (nohang && ret == 0) {
+		printk(UM_KERN_ERR "%s : process %d has not exited\n",
+		       pname, pid);
+		return -ECHILD;
+	} else if (!WIFEXITED(status) || WEXITSTATUS(status) != 0) {
+		printk(UM_KERN_ERR "%s : process %d didn't exit with "
+		       "status 0\n", pname, pid);
+		return -ECHILD;
+	} else
+		return 0;
 }

arch/um/os-Linux/process.c

@@ -101,7 +101,7 @@ void os_kill_process(int pid, int reap_child)
 {
 	kill(pid, SIGKILL);
 	if (reap_child)
-		CATCH_EINTR(waitpid(pid, NULL, 0));
+		CATCH_EINTR(waitpid(pid, NULL, __WALL));
 }
 
 /* This is here uniquely to have access to the userspace errno, i.e. the one
@@ -130,7 +130,7 @@ void os_kill_ptraced_process(int pid, int reap_child)
 	ptrace(PTRACE_KILL, pid);
 	ptrace(PTRACE_CONT, pid);
 	if (reap_child)
-		CATCH_EINTR(waitpid(pid, NULL, 0));
+		CATCH_EINTR(waitpid(pid, NULL, __WALL));
 }
 
 /* Don't use the glibc version, which caches the result in TLS. It misses some

arch/um/os-Linux/skas/process.c

@@ -64,7 +64,7 @@ void wait_stub_done(int pid)
 	int n, status, err;
 
 	while (1) {
-		CATCH_EINTR(n = waitpid(pid, &status, WUNTRACED));
+		CATCH_EINTR(n = waitpid(pid, &status, WUNTRACED | __WALL));
 		if ((n < 0) || !WIFSTOPPED(status))
 			goto bad_wait;
 
@@ -153,7 +153,7 @@ static void handle_trap(int pid, struct uml_pt_regs *regs,
 			panic("handle_trap - continuing to end of syscall "
 			      "failed, errno = %d\n", errno);
 
-		CATCH_EINTR(err = waitpid(pid, &status, WUNTRACED));
+		CATCH_EINTR(err = waitpid(pid, &status, WUNTRACED | __WALL));
 		if ((err < 0) || !WIFSTOPPED(status) ||
 		   (WSTOPSIG(status) != SIGTRAP + 0x80)) {
                         err = ptrace_dump_regs(pid);
@@ -255,16 +255,18 @@ int start_userspace(unsigned long stub_stack)
 		panic("start_userspace : mmap failed, errno = %d", errno);
 	sp = (unsigned long) stack + UM_KERN_PAGE_SIZE - sizeof(void *);
 
-	flags = CLONE_FILES | SIGCHLD;
+	flags = CLONE_FILES;
 	if (proc_mm)
 		flags |= CLONE_VM;
+	else
+		flags |= SIGCHLD;
 
 	pid = clone(userspace_tramp, (void *) sp, flags, (void *) stub_stack);
 	if (pid < 0)
 		panic("start_userspace : clone failed, errno = %d", errno);
 
 	do {
-		CATCH_EINTR(n = waitpid(pid, &status, WUNTRACED));
+		CATCH_EINTR(n = waitpid(pid, &status, WUNTRACED | __WALL));
 		if (n < 0)
 			panic("start_userspace : wait failed, errno = %d",
 			      errno);
@@ -314,7 +316,7 @@ void userspace(struct uml_pt_regs *regs)
 			      "pid=%d, ptrace operation = %d, errno = %d\n",
 			      pid, op, errno);
 
-		CATCH_EINTR(err = waitpid(pid, &status, WUNTRACED));
+		CATCH_EINTR(err = waitpid(pid, &status, WUNTRACED | __WALL));
 		if (err < 0)
 			panic("userspace - waitpid failed, errno = %d\n",
 			      errno);

arch/um/os-Linux/util.c

@@ -141,7 +141,7 @@ void os_dump_core(void)
 	 * nothing reasonable to do if that fails.
 	 */
 
-	while ((pid = waitpid(-1, NULL, WNOHANG)) > 0)
+	while ((pid = waitpid(-1, NULL, WNOHANG | __WALL)) > 0)
 		os_kill_ptraced_process(pid, 0);
 
 	abort();