isa/sudo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
4,602 Commits 1 Branch 0 Tags
2bc6801ca7a086ca43c8803117b9875d9c26732f
Commit Graph

69 Commits

Author SHA1 Message Date
Todd C. Miller
180785567c Add compress_transcript option. 2009-12-14 01:49:50 +00:00
Todd C. Miller
453d055b4c Can't rely on the shell sending us SIGCONT when transitioning from 
backgroup to foreground process.
 2009-11-22 15:58:37 +00:00
Todd C. Miller
84d4bc2df5 Start command out in foreground mode if stdout is a tty. Works 
around issues with some curses-based programs that don't handle
tcsetattr getting interrupted by a signal.  Still allows us to
avoid hogging the tty if the command is part of a pipeline.
 2009-11-15 21:53:38 +00:00
Todd C. Miller
dcf6602daa Use a socketpair to pass signals from parent to child. Child will 
now pass command status change info back via the socketpair.  This
allows the parent to distinguish between signals it has been sent
directly and signals the command has received.  It also means the
parent can once again print the signal notifications to the tty so
all writes to the pty master occur in the parent.  The command is
now always started in background mode with tty signals handled
by the parent.
 2009-11-15 21:42:17 +00:00
Todd C. Miller
febc3cbb1f Need to ignore SIGTT{IN,OU} in child when running the command in the 
background.  Also some minor cleanup.
 2009-11-01 15:14:58 +00:00
Todd C. Miller
78cd63e074 Instead of calling sigsuspend when waiting for SIGUSR[12] from parent, 
install the signal handlers w/o SA_RESTART and let them interrupt waitpid().
 2009-10-31 17:40:19 +00:00
Todd C. Miller
94370b319f Pass along SIGHUP and SIGTERM from parent to child. 2009-10-31 17:14:52 +00:00
Todd C. Miller
e6567f46b6 Close unused bits of script_fds in processes that don't need them. 
Restore default SIGCONT handler in child.
 2009-10-31 16:48:34 +00:00
Todd C. Miller
aee3382ed1 Update foreground/background status in SIGCONT handler in parent process. 2009-10-31 16:45:59 +00:00
Todd C. Miller
e62dd7645f Defer setting terminal into raw mode until just before we fork() 
and only do it if sudo is the foreground process.
If we get SIGTT{IN,OU} and sudo is already in the foreground be sure
to set raw mode before continuing the child.
 2009-10-25 14:55:14 +00:00
Todd C. Miller
c77ef16fcd Fix handling of SIGTTOU/SIGTTIN in program being run. We now only 
give the command the controlling tty if the main sudo process is the
foreground process.
 2009-10-24 18:43:16 +00:00
Todd C. Miller
61b57817a5 Don't bother with sudo_waitpid() here for now. 2009-10-24 18:32:44 +00:00
Todd C. Miller
50f5d55163 fix non-zlib case 2009-10-24 13:16:40 +00:00
Todd C. Miller
470d3a1925 Remove non-wroking code that crept into rev 1.55 2009-10-23 11:40:22 +00:00
Todd C. Miller
602a62c01e First pass at zlib support for transcript data files 2009-10-22 23:50:16 +00:00
Todd C. Miller
db502bb584 Add missing variable declaration for when TIOCSCTTY is not defined. 
Need to include sys/termio.h for TIOCSCTTY on some systems.
 2009-10-22 14:53:13 +00:00
Todd C. Miller
03e0e4c747 when resuming command, send SIGCONT to its pgrp not just pid 2009-10-22 13:52:59 +00:00
Todd C. Miller
8fa5a519a0 include selinux.h for is_selinux_enabled() proto 2009-10-22 13:47:32 +00:00
Todd C. Miller
b231f074a2 Don't use log_error() in the child process. 2009-10-22 13:40:41 +00:00
Todd C. Miller
425026fa11 Do I/O in parent instead of child since the parent can have both 
/dev/tty as well as the pty fds open.  The child just sets things
up and waits for its grandchild and writes the signal description
to the pty master if the command was killed by a signal.
 2009-10-22 00:58:41 +00:00
Todd C. Miller
27b221e439 Make comment at the top of script_exec() match reality. 2009-10-18 17:32:44 +00:00
Todd C. Miller
1a5afcbb15 Make transcript mode work as long as we can figure out our tty, even if 
it is not stdin.  We'd like to use /dev/tty but that won't be valid after
the setsid().
 2009-10-18 14:46:01 +00:00
Todd C. Miller
9544845ef8 Break out pty code into pty.c 2009-10-16 13:29:38 +00:00
Todd C. Miller
babf9292df Push ptem and ldterm for STERAMS-based systems when allocating a pty. 2009-10-16 00:07:11 +00:00
Todd C. Miller
7d79b1da35 Sprinkle some more O_NOCTTY and call grantpt() before unlockpt() 2009-10-15 21:48:19 +00:00
Todd C. Miller
3ece4c837f Call tcgetpgrp() in the parent, not the child and have the child spin 
until it is granted.  Fixes a race on darwin.
 2009-10-15 13:46:11 +00:00
Todd C. Miller
52f3374ce3 Only use TIOCNOTTY in the non-setsid case. If no TIOCSCTTY, just 
reopen slave.
 2009-10-15 13:38:11 +00:00
Todd C. Miller
7411b071e4 In script mode, if the command is killed by a signal, print the 
signal description as well as a core dump notification like the
shell does.
 2009-10-14 20:04:16 +00:00
Todd C. Miller
7bfe0fe957 Add missing WUNTRACED and store the signal that stopped the grandchild 
in suspended, not signo.
 2009-10-14 13:37:23 +00:00
Todd C. Miller
cd8ba12194 g/c unused code 2009-10-14 13:20:24 +00:00
Todd C. Miller
0d04ffc40c Associate the grandchild's pgrp with the tty instead of the child's 
and just get suspend notifications via SIGCHLD instead of directly.
This fixes a hang with programs that try to set terminal attributes
and is more consistent with how the shell handles things.
 2009-10-14 13:01:04 +00:00
Todd C. Miller
d72afb29a3 Move setpgid() of child into the parent side of the fork() where it 
belongs.
 2009-10-12 14:30:46 +00:00
Todd C. Miller
bc1769df24 fix typo 2009-10-11 22:02:02 +00:00
Todd C. Miller
88b93973fe Run command in its own pgrp (like the shell does) for easier signalling. 
No need to relay SIGINT or SIGQUIT to parent, just send to grandchild.
Don't want grandchild stopped events in the child (only termination).
Flush output after suspending grandchild before signalling parent.
 2009-10-11 21:41:36 +00:00
Todd C. Miller
a3185d3e25 Back out revision 1.34; the problem lies elsewhere. 2009-10-11 14:16:22 +00:00
Todd C. Miller
f44c2261fd Don't set stdout to blocking mode when flushing remaining output. It 
can cause us to hang when trying to exit.  Need to investigate why.
 2009-10-11 14:13:35 +00:00
Todd C. Miller
03e3ca08db Handle SIGTTOU and remove some debugging. 2009-10-11 13:25:23 +00:00
Todd C. Miller
09a7bb78fe Apparently we need to send SIGSTOP to the command as well as ourself 
when we get SIGTSTP, the kernel doesn't automatically stop the process
for us.
 2009-10-11 12:42:03 +00:00
Todd C. Miller
2cf1b81abf Use an extra process to act as the glue bewteen the sessions 
associated with the user's controlling tty (what the shell uses)
and the tty that sudo is using to do its logging.  Basically, this
means that if we get, e.g. SIGTSTP from the process sudo is running,
we relay the signal to the parent so it's shell can do the job control.
 2009-10-11 12:27:11 +00:00
Todd C. Miller
50b62b694d Already define O_NOCTTY in compat.h, don't need it here 2009-10-07 00:17:38 +00:00
Todd C. Miller
fe9fc5ad1b Retain NL to NLCR conversion on the real tty and skip it on the pty 
we allocate.  That way, if stdout is not a pty there are no extra carriage
returns.
 2009-09-30 02:12:35 +00:00
Todd C. Miller
e6331c0773 Fix log_output(); just pass in a string and a length. 2009-09-30 02:00:16 +00:00
Todd C. Miller
8e4379b264 do not use errno when complaining out lack of a tty 2009-09-28 11:44:26 +00:00
Todd C. Miller
577fb346fd Convert a do {} while into a while 2009-09-27 22:11:11 +00:00
Todd C. Miller
4283bbcb2f Add O_NOCTTY when opening a tty device. 
Explicitly disconnect from old tty before associatng with new one.
 2009-09-27 18:34:18 +00:00
Todd C. Miller
7d19478501 First cut at refactoring some of the selinux code so it can be used 
in conjunction with sudo's transcript support.
 2009-09-27 13:03:56 +00:00
Todd C. Miller
d111b0d9bb Use _PATH_SUDO_TRANSCRIPT instead of _PATH_SUDO_SESSDIR 2009-09-26 15:41:14 +00:00
Todd C. Miller
7bb38284b0 Go back to dropping out of the select() loop when the process dies; Linux 
ptys apparently don't behave the same as BSD in regards to select().
No need to flush remaining output to the transcript, only to stdout.
Add back code to check the master pty for additional data when we exit
the main select loop.
 2009-09-20 13:51:51 +00:00
Todd C. Miller
f68a24fb3b Fix final flush. 2009-09-19 11:43:24 +00:00
Todd C. Miller
3223231b1f Keep reading from the pty master -> log file until read returns <= 0. 
Do our best to write everything to stdout when flushing any remaining bits.
 2009-09-19 11:35:20 +00:00