On Fri 02 Oct 09, 1:21 PM, Ken Bloom <[email protected]> said: > When I ssh from little-cat-a to cat-in-the-hat, exiting the login shell > on cat-in-the-hat doesn't end my ssh connection. After running ps on > cat-in-the-hat, I noticed that the login shell in question was still > sitting around as a zombie process, meaning that sshd isn't reaping the > shell. > > Here's an strace of the problem (from sshd on cat-in-the-hat), attaching > to the sshd process (as root) after I am already logged in. You can see > that sshd is changing the signal mask an awful lot (blocking and > unblocking SIGCHLD), and it appears sshd isn't responding to SIGCHLD. > (It's definitley not calling waitpid to reap the shell.) > > Anybody seen this before? Anybody know how to fix it? Anybody know where > to report a bug? > > ii openssh-server 1:5.1p1-7 secure shell server, an rshd > replacement > > > [bl...@cat-in-the-hat ~]$ sudo strace -p 8124 > [sudo] password for bloom: > Process 8124 attached - interrupt to quit > select(13, [3 6 8 9 12], [], NULL, NULL) = 1 (in [3]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > read(3, "?m\3\270^\316,\314\0v\375\240\36\rG\263\203`f\210\210\264\340 > \366\3\341\366\230_X\5\276"..., 16384) = 48 > select(13, [3 6 8 9 12], [10], NULL, NULL) = 1 (out [10]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > write(10, "e", 1) = 1 > ioctl(10, SNDCTL_TMR_TIMEBASE or TCGETS, {B38400 opost isig -icanon > -echo ...}) = 0 > select(13, [3 6 8 9 12], [], NULL, NULL) = 1 (in [12]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > read(12, "e", 16384) = 1 > select(13, [3 6 8 9 12], [3], NULL, NULL) = 1 (out [3]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > write(3, "X\263\2A\350\235\233\355,\3149\223\364\221\265Ut\346\225h\275 > \215z\234Z\3039\31\356x[\27"..., 48) = 48 > select(13, [3 6 8 9 12], [], NULL, NULL) = 1 (in [3]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > read(3, "\233`\205l\271\25\217\245\311j\307\2614\267t\226D\212\275h&q > \314+\262\23\25\321\302\261D\235"..., 16384) = 48 > select(13, [3 6 8 9 12], [10], NULL, NULL) = 1 (out [10]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > write(10, "x", 1) = 1 > ioctl(10, SNDCTL_TMR_TIMEBASE or TCGETS, {B38400 opost isig -icanon > -echo ...}) = 0 > select(13, [3 6 8 9 12], [], NULL, NULL) = 1 (in [12]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > read(12, "x", 16384) = 1 > select(13, [3 6 8 9 12], [3], NULL, NULL) = 1 (out [3]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > write(3, "\206\331V\270\377\4\356\34rA\211c\336\200-e\264M\205\244\263 > \250Qg$\227\307\205\270\36\201t"..., 48) = 48 > select(13, [3 6 8 9 12], [], NULL, NULL) = 1 (in [3]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > read(3, "<\326\326|\204\343[G\22\230\211\231\2\240\324xs~\257\35\250TT > \212,>\1\376U\231\374\264"..., 16384) = 48 > select(13, [3 6 8 9 12], [10], NULL, NULL) = 1 (out [10]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > write(10, "i", 1) = 1 > ioctl(10, SNDCTL_TMR_TIMEBASE or TCGETS, {B38400 opost isig -icanon > -echo ...}) = 0 > select(13, [3 6 8 9 12], [], NULL, NULL) = 1 (in [12]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > read(12, "i", 16384) = 1 > select(13, [3 6 8 9 12], [3], NULL, NULL) = 1 (out [3]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > write(3, "\320\230 \245\237\230[\261\33\16K1\207\270\241\270\310\\tkU > \231\n\347\232H2*LENf"..., 48) = 48 > select(13, [3 6 8 9 12], [], NULL, NULL) = 1 (in [3]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > read(3, "\236\347\243\35\22\235\241v\247k\246<\r\305a\271\335V\305&\30 > \324-=\204\256...@\302`\333"..., 16384) = 48 > select(13, [3 6 8 9 12], [10], NULL, NULL) = 1 (out [10]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > write(10, "t", 1) = 1 > ioctl(10, SNDCTL_TMR_TIMEBASE or TCGETS, {B38400 opost isig -icanon > -echo ...}) = 0 > select(13, [3 6 8 9 12], [], NULL, NULL) = 1 (in [12]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > read(12, "t", 16384) = 1 > select(13, [3 6 8 9 12], [3], NULL, NULL) = 1 (out [3]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > write(3, ")(\353\317\2G(<\374e\346\253\277&R\214\37719\\\306\341\241d\5 > \247\237\3059\35B}"..., 48) = 48 > select(13, [3 6 8 9 12], [], NULL, NULL) = 1 (in [3]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > read(3, "i2xeL\20\215\266~s\264\264/fIZe\314v\304\216E\323GO\236\34\331 > \364\250[l"..., 16384) = 48 > select(13, [3 6 8 9 12], [10], NULL, NULL) = 1 (out [10]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > write(10, "\r", 1) = 1 > select(13, [3 6 8 9 12], [], NULL, NULL) = 1 (in [12]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > read(12, "\r\n", 16384) = 2 > select(13, [3 6 8 9 12], [3], NULL, NULL) = 2 (in [12], out [3]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > read(12, "logout\r\n", 16384) = 8 > write(3, "v\216`<\244\240\264\32D\331\240\212R\250-+\346\325\237b\255 > \351{{y\327D4(\250\371\313"..., 48) = 48 > select(13, [3 6 8 9 12], [3], NULL, NULL) = 1 (out [3]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > write(3, "\261\10\331\20fV\225\231\351=\2111\34\225G\366r\222+a\335\264 > \244\266\r\17\2424^\311\0027"..., 48) = 48 > select(13, [3 6 8 9 12], [], NULL, NULL) = 1 (in [12]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > read(12, 0x7fff72ac4d40, 16384) = -1 EIO (Input/output error) > close(12) = 0 > select(13, [3 6 8 9], [3], NULL, NULL) = 1 (out [3]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > write(3, "\217\254\305\327\4\235N\266r\214\6h\341\256hg\362\227kNa\322 > \376\33\306\363i\233\243\260x\311", 32) = 32 > select(13, [3 6 8 9], [], NULL, NULL) = 1 (in [3]) > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > read(3, "", 16384) = 0 > rt_sigprocmask(SIG_BLOCK, [CHLD], ~[KILL ALRM STOP RTMIN RT_1], 8) = 0 > rt_sigprocmask(SIG_SETMASK, ~[KILL ALRM STOP RTMIN RT_1], NULL, 8) = 0 > close(10) = 0 > shutdown(8, 2 /* send and receive */) = 0 > close(8) = 0 > close(8) = -1 EBADF (Bad file descriptor) > close(8) = -1 EBADF (Bad file descriptor) > shutdown(9, 2 /* send and receive */) = 0 > close(9) = 0 > close(9) = -1 EBADF (Bad file descriptor) > close(9) = -1 EBADF (Bad file descriptor) > brk(0x7fc905223000) = 0x7fc905223000 > write(5, "\0\0\0\17\34", 5) = 5 > write(5, "\0\0\0\n/dev/pts/1", 14) = 14 > close(13) = 0 > shutdown(3, 2 /* send and receive */) = 0 > close(3) = 0 > write(5, "\0\0\0\1=", 5) = 5 > exit_group(0) = ? > Process 8124 detached
Ken, two comments: 1. ltrace can sometimes be more useful than strace. Not that it necessarily would be here. Just saying because sometimes the library calls are more transparent than the system calls. 2. Perhaps you have something running in the background that's preventing the shell from exiting? Pete -- GPG Fingerprint: B9F1 6CF3 47C4 7CD8 D33E 70A9 A3B9 1945 67EA 951D Last night I dreamt of 09-f9-11-02-9d-74-e3-5b-d8-41-56-c5-63-56-88-c0 Facebook: http://facebook.com/coffeeAchiever "The fundamental cause of trouble in the world today [email protected] is that the stupid are cocksure while the intelligent http://www.dirac.org are full of doubt" --- Bertrand Russell _______________________________________________ vox-tech mailing list [email protected] http://lists.lugod.org/mailman/listinfo/vox-tech
