Hello Imre, thanks for the report. I can't reproduce this issue myself,
nor could I find anything in the source code that would suggest dash
would try to quit. In fact I'm not even sure that dash knows that it's
happened.
Here's the tail end of strace output (run on the same terminal as dash,
to force intermingling of syscall output and terminal output), showing
dash handling a write(1) without trouble, and checking for completed
jobs after I hit enter several times:
geteuid() = 1000
getegid() = 1000
rt_sigaction(SIGINT, NULL, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, 8) = 0
rt_sigaction(SIGINT, {sa_handler=0x55a82089e200, sa_mask=~[RTMIN RT_1],
sa_flags=SA_RESTORER, sa_restorer=0x7fd3be19df20}, NULL, 8) = 0
rt_sigaction(SIGQUIT, NULL, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, 8) = 0
rt_sigaction(SIGQUIT, {sa_handler=SIG_IGN, sa_mask=~[RTMIN RT_1],
sa_flags=SA_RESTORER, sa_restorer=0x7fd3be19df20}, NULL, 8) = 0
rt_sigaction(SIGTERM, NULL, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, 8) = 0
rt_sigaction(SIGTERM, {sa_handler=SIG_IGN, sa_mask=~[RTMIN RT_1],
sa_flags=SA_RESTORER, sa_restorer=0x7fd3be19df20}, NULL, 8) = 0
openat(AT_FDCWD, "/dev/tty", O_RDWR) = 3
fcntl(3, F_DUPFD, 10) = 10
close(3) = 0
fcntl(10, F_SETFD, FD_CLOEXEC) = 0
ioctl(10, TIOCGPGRP, [23306]) = 0
getpgrp() = 23306
rt_sigaction(SIGTSTP, NULL, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, 8) = 0
rt_sigaction(SIGTSTP, {sa_handler=SIG_IGN, sa_mask=~[RTMIN RT_1],
sa_flags=SA_RESTORER, sa_restorer=0x7fd3be19df20}, NULL, 8) = 0
rt_sigaction(SIGTTOU, NULL, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, 8) = 0
rt_sigaction(SIGTTOU, {sa_handler=SIG_IGN, sa_mask=~[RTMIN RT_1],
sa_flags=SA_RESTORER, sa_restorer=0x7fd3be19df20}, NULL, 8) = 0
rt_sigaction(SIGTTIN, NULL, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, 8) = 0
rt_sigaction(SIGTTIN, {sa_handler=SIG_DFL, sa_mask=~[RTMIN RT_1],
sa_flags=SA_RESTORER, sa_restorer=0x7fd3be19df20}, NULL, 8) = 0
setpgid(0, 23308) = 0
ioctl(10, TIOCSPGRP, [23308]) = 0
wait4(-1, 0x7ffcdf89717c, WNOHANG|WSTOPPED, NULL) = -1 ECHILD (No child
processes)
write(2, "$ ", 2$ ) = 2
read(0,
Message from sarnold@hunt on pts/13 at 17:41 ...
hello
EOF
"\n", 8192) = 1
wait4(-1, 0x7ffcdf89717c, WNOHANG|WSTOPPED, NULL) = -1 ECHILD (No child
processes)
write(2, "$ ", 2$ ) = 2
read(0,
"\n", 8192) = 1
wait4(-1, 0x7ffcdf89717c, WNOHANG|WSTOPPED, NULL) = -1 ECHILD (No child
processes)
write(2, "$ ", 2$ ) = 2
read(0, ^C0x55a820aa9a60, 8192) = ? ERESTARTSYS (To be restarted if
SA_RESTART is set)
--- SIGINT {si_signo=SIGINT, si_code=SI_KERNEL} ---
rt_sigprocmask(SIG_SETMASK, [], ~[KILL STOP RTMIN RT_1], 8) = 0
write(2, "\n", 1
) = 1
wait4(-1, 0x7ffcdf89717c, WNOHANG|WSTOPPED, NULL) = -1 ECHILD (No child
processes)
write(2, "$ ", 2$ ) = 2
read(0, "", 8192) = 0
write(2, "\n", 1
) = 1
ioctl(10, TIOCSPGRP, [23306]) = 0
setpgid(0, 23306) = 0
close(10) = 0
exit_group(0) = ?
+++ exited with 0 +++
I tried both dash in a terminal (urxvt) and ssh -t sarnold@localhost /bin/dash
-- both behaved identically.
Could you try to reproduce this while you're attached to the shell with
gdb or perhaps perf trace?
Thanks
** Changed in: dash (Ubuntu)
Status: New => Incomplete
--
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to dash in Ubuntu.
https://bugs.launchpad.net/bugs/1814905
Title:
dash exits when receiving wall message
Status in dash package in Ubuntu:
Incomplete
Bug description:
When using "/bin/sh" (dash) as a shell for ssh, or just starting
/bin/sh in a terminal, when the shell receives a wall message (from
another user or root), it just quits.
How to reproduce:
- start a terminal (xterm, gnome-terminal, etc), usually starts with bash
- from bash run 'sh' command to open a dash shell
- start another terminal, and run: echo "foo" | wall
- the broadcast message appears on both terminals
- in the first terminal you are thrown back to bash, because dash exits
I've tried this on Ubuntu 14.04, 16.04, 18.04 (64bits) latest updates
and it's happening on all of them. I've also tried with latest dash
from git, compiled on Ubuntu 16.04, same problem, that's why I don't
think it's dash's problem.
Also tried on CentOS 7, there is no problem there.
The problem appears this morning (6 February) after updating to latest
version of 16.04, 18.04. I also have a box with 16.04 but not the
latest updates, and on that box the problem does not happen.
If one uses dash as a login shell, and enters on a remote box with SSH
for example, the connection can be terminated by another user simply
wall'ing a broadcast message (or a system daemon like NUT). This could
be taken as a potential security issue.
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/dash/+bug/1814905/+subscriptions
--
Mailing list: https://launchpad.net/~touch-packages
Post to : [email protected]
Unsubscribe : https://launchpad.net/~touch-packages
More help : https://help.launchpad.net/ListHelp
