isa/sudo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

The change in 818e82ecbbfc that caused to exit when the monitor

Browse Source 
dies created a race condition between the monitor exiting and the
status being read.  All we really want to do is make sure that
select() notifies us that there is a status change when the monitor
dies unexpectedly so shutdown the socketpair connected to the monitor
for writing when it dies.  That way we can still read the status
that is pending on the socket and select() on Linux will tell us
that the fd is ready.
This commit is contained in:
Todd C. Miller
2012-01-25 16:29:08 -05:00
parent d11e7febbc
commit 1877c455d1
2 changed files with 13 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
src/exec.c
View File

@@ -74,7 +74,7 @@ struct sigforward {
TQ_DECLARE(sigforward)
static struct sigforward_list sigfwd_list;
static int handle_signals(int fd, pid_t child, int log_io,
static int handle_signals(int sv[2], pid_t child, int log_io,
struct command_status *cstat);
static void forward_signals(int fd);
static void schedule_signal(int signo);
@@ -343,7 +343,7 @@ sudo_execute(struct command_details *details, struct command_status *cstat)
forward_signals(sv[0]);
}
if (FD_ISSET(signal_pipe[0], fdsr)) {
n = handle_signals(signal_pipe[0], child, log_io, cstat);
n = handle_signals(sv, child, log_io, cstat);
if (n == 0) {
/* Child has exited, cstat is set, we are done. */
break;
@@ -442,7 +442,7 @@ sudo_execute(struct command_details *details, struct command_status *cstat)
* Returns -1 on error, 0 on child exit, else 1.
*/
static int
handle_signals(int fd, pid_t child, int log_io, struct command_status *cstat)
handle_signals(int sv[2], pid_t child, int log_io, struct command_status *cstat)
{
unsigned char signo;
ssize_t nread;
@@ -483,17 +483,11 @@ handle_signals(int fd, pid_t child, int log_io, struct command_status *cstat)
/*
* On BSD we get ECONNRESET on sv[0] if monitor dies
* and select() will return with sv[0] readable.
* On Linux that doesn't appear to happen so we
* treat the monitor dying as a fatal error.
* Note that the wait status we return is that of
* the monitor and not the command; unfortunately
* that is the best that we can do here.
* On Linux that doesn't appear to happen so if the
* monitor dies, shut down the socketpair to force a
* select() notification.
*/
if (!WIFSTOPPED(status)) {
cstat->type = CMD_WSTATUS;
cstat->val = status;
debug_return_int(0);
}
(void) shutdown(sv[0], SHUT_WR);
} else {
if (WIFSTOPPED(status)) {
/*