Problem: strace -ff fails on 2.4.21-rc1

Problem: strace -ff fails on 2.4.21-rc1

Post by Chuck Ebber » Mon, 12 May 2003 17:50:09





> >   I just found out minicom is spawing /sbin/lockdev which is setgrp
> > 'lock'.  Would that cause ptrace failure??

> AFAIK that could have caused the failure. Please test 2.4.21-rc2 whcih
> has fixes for many ptrace problems.

  I can now strace minicom and its children with 2.4.21-rc2-ac1 but it
hangs on exit.  Both child processes exit successfully:

 # tail -5 minicom.trc.5775

10:16:49.299253 kill(5774, SIG_0)       = 0
10:16:49.299418 unlink("/var/lock/LOCKDEV") = 0
10:16:49.299649 close(3)                = 0
10:16:49.299849 umask(022)              = 02
10:16:49.300043 _exit(0)                = ?

 # tail -8 minicom.trc.5776

10:17:05.497676 kill(5774, SIG_0)       = 0
10:17:05.497845 unlink("/var/lock/LCK...5774") = 0
10:17:05.498068 unlink("/var/lock/LCK..ttyS1") = 0
10:17:05.498283 unlink("/var/lock/LCK.004.065") = 0
10:17:05.498616 unlink("/var/lock/LOCKDEV") = 0
10:17:05.498870 close(4)                = 0
10:17:05.499065 umask(022)              = 02
10:17:05.499337 _exit(0)                = ?

However strace and minicom are hung up somehow and the screen is
black with a blinking cursor at row 1 column 1.  The other ttys all
work OK and killing minicom cleans everything up.

 # tail -3 minicom.trc

10:17:05.499678 --- SIGCHLD (Child exited) ---
10:17:05.499760 rt_sigaction(SIGCHLD, {SIG_DFL}, {SIG_IGN}, 8) = 0
10:17:05.499971 close(3

strace        S C015FF82   960  5773    822  5774               (NOTLB)
Call Trace:    [<c015ff82>] [<c011dba8>] [<c0108b73>]
minicom       S 00000000  2400  5774   5773                     (NOTLB)
Call Trace:    [<c0122f24>] [<c0124095>] [<c021195b>] [<c0220d79>] [<c020d997>]
  [<c0124095>] [<c0124159>] [<c020e01a>] [<c013c52e>] [<c01183a1>] [<c013b1d5>]
  [<c013b23b>] [<c0108bdf>]

Proc;  strace

Quote:>>EIP; c015ff82 <ext3_file_write+22/b0>   <=====

Trace; c015ff82 <ext3_file_write+22/b0>
Trace; c011dba8 <sys_wait4+3c8/400>
Trace; c0108b73 <system_call+33/38>
Proc;  minicom
Quote:>>EIP; 00000000 Before first symbol

Trace; c0122f24 <schedule_timeout+14/a0>
Trace; c0124095 <wake_up_parent+25/40>
Trace; c021195b <tty_wait_until_sent+9b/e0>
Trace; c0220d79 <rs_close+129/1f0>
Trace; c020d997 <release_dev+247/500>
Trace; c0124095 <wake_up_parent+25/40>
Trace; c0124159 <do_notify_parent+a9/c0>
Trace; c020e01a <tty_release+2a/60>
Trace; c013c52e <fput+4e/100>
Trace; c01183a1 <schedule+351/3b0>
Trace; c013b1d5 <filp_close+95/a0>
Trace; c013b23b <sys_close+5b/70>
Trace; c0108bdf <tracesys+1f/23>

Configuration:
        2.4.21-rc2-ac1
        SMP kernel on 1-CPU SMP machine (PII Xeon, 440GX)

To reproduce:
        # strace -ff -q -tt -o minicom.trc minicom

Note:
        I also get lockups while minicom is running:

        Proc;  strace
        >>EIP; c015ff82 <ext3_file_write+22/b0>   <=====
        Trace; c015ff82 <ext3_file_write+22/b0>
        Trace; c011dba8 <sys_wait4+3c8/400>
        Trace; c0108b73 <system_call+33/38>
        Proc;  minicom
        >>EIP; 00000000 Before first symbol
        Trace; c0122f24 <schedule_timeout+14/a0>
        Trace; c021105a <read_chan+37a/720>
        Trace; c020cf5f <tty_read+cf/120>
        Trace; c013b706 <sys_read+96/110>
        Trace; c0108bdf <tracesys+1f/23>

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

 
 
 

Problem: strace -ff fails on 2.4.21-rc1

Post by Bernhard Kaind » Tue, 13 May 2003 01:50:09





> > >   I just found out minicom is spawing /sbin/lockdev which is setgrp
> > > 'lock'.  Would that cause ptrace failure??

> > AFAIK that could have caused the failure. Please test 2.4.21-rc2 whcih
> > has fixes for many ptrace problems.

>   I can now strace minicom and its children with 2.4.21-rc2-ac1 but it
> hangs on exit.  Both child processes exit successfully:

Very strange, does this work with 2.4.20? What's your version of strace?

Quote:> However strace and minicom are hung up somehow and the screen is
> black with a blinking cursor at row 1 column 1.  The other ttys all
> work OK and killing minicom cleans everything up.

Hm, this sounds like there could be some error/loop opening the device,
which could be the effect of another side effect of the original ptrace
fix, which my fixes which are included in 2.4.21-rc2 don't fix.

This side effect causes that if a system call needs a module loaded,
it is not loaded and only an error from request_module() is sent to
the kernel log. The attached patch on top of 2.4.21-rc2, fixes this
remaining problem.

I'm not writing much info about it now, except that I think that it does
not open any securiy hole, but I would like to give it a little more
testing on SMP machines. On single CPU it fixed the "modprobe rejected"
problem fine for me, without sacrifying ptrace securitey.

I'm uncertain if it would help in your case.


Quote:>   (BTW does minicom work for you on 2.5?  It fails with the "No child
> processes" message on 2.5.6x here but works on 2.4 when it's not being
> traced.  Just the very act of running it under strace on 2.4 makes it
> fail the same way it does on 2.5 here.  And stracing it on 2.5.66 made
> it start working again!  Something very strange is going on...)

Very strange, maybe a "tail minicom.trc.*" at the time when it's*
helps to get some idea.

Another note: suid is ignored when you are tracing the task which runs
exec() for a setuid program.

So if minicom relies on having the setgid gid of /sbin/lockdev honored
has in your case, the only ways I can find to get it working inder strace are:

- Change the locking config (temporary, for the debug) so that /sbin/lockdev
  does not need to be setgid.

- Don't have ptrace follow fork mode activated when forking the child
  which exec()'s /sbin/lockdev.

Bernd

  ptrace-kmod-2.4.21-rc2.diff
< 1K Download

 
 
 

Problem: strace -ff fails on 2.4.21-rc1

Post by Daniel Jacobowit » Tue, 13 May 2003 02:40:12






> > > >   I just found out minicom is spawing /sbin/lockdev which is setgrp
> > > > 'lock'.  Would that cause ptrace failure??

> > > AFAIK that could have caused the failure. Please test 2.4.21-rc2 whcih
> > > has fixes for many ptrace problems.

> >   I can now strace minicom and its children with 2.4.21-rc2-ac1 but it
> > hangs on exit.  Both child processes exit successfully:

> Very strange, does this work with 2.4.20? What's your version of strace?

> > However strace and minicom are hung up somehow and the screen is
> > black with a blinking cursor at row 1 column 1.  The other ttys all
> > work OK and killing minicom cleans everything up.

> Hm, this sounds like there could be some error/loop opening the device,
> which could be the effect of another side effect of the original ptrace
> fix, which my fixes which are included in 2.4.21-rc2 don't fix.

> This side effect causes that if a system call needs a module loaded,
> it is not loaded and only an error from request_module() is sent to
> the kernel log. The attached patch on top of 2.4.21-rc2, fixes this
> remaining problem.

> I'm not writing much info about it now, except that I think that it does
> not open any securiy hole, but I would like to give it a little more
> testing on SMP machines. On single CPU it fixed the "modprobe rejected"
> problem fine for me, without sacrifying ptrace securitey.

> I'm uncertain if it would help in your case.


> >   (BTW does minicom work for you on 2.5?  It fails with the "No child
> > processes" message on 2.5.6x here but works on 2.4 when it's not being
> > traced.  Just the very act of running it under strace on 2.4 makes it
> > fail the same way it does on 2.5 here.  And stracing it on 2.5.66 made
> > it start working again!  Something very strange is going on...)

> Very strange, maybe a "tail minicom.trc.*" at the time when it's*
> helps to get some idea.

> Another note: suid is ignored when you are tracing the task which runs
> exec() for a setuid program.

> So if minicom relies on having the setgid gid of /sbin/lockdev honored
> has in your case, the only ways I can find to get it working inder strace are:

> - Change the locking config (temporary, for the debug) so that /sbin/lockdev
>   does not need to be setgid.

> - Don't have ptrace follow fork mode activated when forking the child
>   which exec()'s /sbin/lockdev.

Run strace as root, and use strace -u?  Then it should not remove the
setuid-ness, I think.

--
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://www.veryComputer.com/
Please read the FAQ at  http://www.veryComputer.com/

 
 
 

Problem: strace -ff fails on 2.4.21-rc1

Post by Chuck Ebber » Tue, 13 May 2003 03:10:04


Quote:> Another note: suid is ignored when you are tracing the task which runs
> exec() for a setuid program.

  You're right -- an ordinary user cannot trace minicom's setgroupid
children with -ff (but root now can, where with rc1 that failed too.)

  And this is no longer an strace/ptrace problem at all.

  Minicom hangs on exit (ctrl-a, q, enter) even when strace is not
running (can't believe I just now thought to try that.)

minicom       S C01183A1  4472  1197    834                     (NOTLB)
Using defaults from ksymoops -t elf32-i386 -a i386
Call Trace:    [<c0122f24>] [<c0134a31>] [<c021195b>] [<c0220d79>] [<c020d997>]
  [<c011690d>] [<c020e01a>] [<c013c52e>] [<c013b1d5>] [<c013b23b>] [<c0108b73>]
Warning (Oops_read): Code line not seen, dumping what data is available

Proc;  minicom

Quote:>>EIP; c01183a1 <schedule+351/3b0>   <===== ignore this bogus address

Trace; c0122f24 <schedule_timeout+14/a0>
Trace; c0134a31 <__alloc_pages+41/170>
Trace; c021195b <tty_wait_until_sent+9b/e0>
Trace; c0220d79 <rs_close+129/1f0>
Trace; c020d997 <release_dev+247/500>
Trace; c011690d <do_page_fault+11d/43b>
Trace; c020e01a <tty_release+2a/60>
Trace; c013c52e <fput+4e/100>
Trace; c013b1d5 <filp_close+95/a0>
Trace; c013b23b <sys_close+5b/70>
Trace; c0108b73 <system_call+33/38>

It's hung up somewhere inside schedule().

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

 
 
 

Problem: strace -ff fails on 2.4.21-rc1

Post by Russell Kin » Tue, 13 May 2003 10:30:14



> Proc;  minicom
> >>EIP; c01183a1 <schedule+351/3b0>   <===== ignore this bogus address
> Trace; c0122f24 <schedule_timeout+14/a0>
> Trace; c0134a31 <__alloc_pages+41/170>
> Trace; c021195b <tty_wait_until_sent+9b/e0>
> Trace; c0220d79 <rs_close+129/1f0>
> Trace; c020d997 <release_dev+247/500>
> Trace; c011690d <do_page_fault+11d/43b>
> Trace; c020e01a <tty_release+2a/60>
> Trace; c013c52e <fput+4e/100>
> Trace; c013b1d5 <filp_close+95/a0>
> Trace; c013b23b <sys_close+5b/70>
> Trace; c0108b73 <system_call+33/38>

> It's hung up somewhere inside schedule().

Wait 30 seconds and see if it exits by itself.  I bet you have hardware
RTS/CTS handshaking enabled on the serial port, but without anything
connected.

When a port is closed, we wait up to 30 seconds (or a user specified time
period) for any characters in the transmit queue to be sent.  If CTS is
inactive and we're using RTS/CTS handshaking, we can't send any characters,
so we'll wait the full timeout.

--

             http://www.arm.linux.org.uk/personal/aboutme.html

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

 
 
 

Problem: strace -ff fails on 2.4.21-rc1

Post by Chuck Ebber » Tue, 13 May 2003 14:00:22




>> It's hung up somewhere inside schedule().

>Wait 30 seconds and see if it exits by itself.  I bet you have hardware
>RTS/CTS handshaking enabled on the serial port, but without anything
>connected.

  It hangs forever (over an hour anyway.)  And it's definitely taking the
MAX_TIMEOUT case (wait forever) in schedule_timeout().

  Yes, RTS/CTS is enabled but nothing is connected... and just attaching
a nullmodem adapter that lights up CTS, DTR and CD makes the problem go
away.

  Does that mean this is a minicom problem because it didn't specify
a timeout?

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

 
 
 

1. Problem: strace -ff fails on 2.4.21-rc1

  Is this a bug or a feature?  Machine only has 1 CPU, if that makes
any difference:

# strace -q -o minicom.trc -tt -ff minicom
upeek: ptrace(PTRACE_PEEKUSER, ... ): Operation not permitted
detach: ptrace(PTRACE_CONT, ...): Operation not permitted
Device /dev/ttyS1 lock failed: No child processes.
# uname -a
Linux d2 2.4.21-rc1 #1 SMP Wed May 7 06:05:31 EDT 2003 i686 unknown

  (Ignore that minicom error message about the lock failing -- that's
a separate problem...)

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in

More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

2. Security Advisory with Netscape Servers on HP-UX

3. v4l won't compile 2.4.21-rc1 & rc1-ac3

4. FAQs for system recovery

5. 2.4.21-rc1 compile fails on serveraid

6. help needed

7. 2.4.21-rc2 and rc1 compile problem in fs/exec.c + PATCH

8. CFS - NFS - WEBNFS - PCNFS - Tarantella - SAMBA - whats best

9. problems after upgrading from 2.4.20-pre7 to 2.4.21-rc1 (maybe nfs related)

10. Matrox FB problem in latest 2.4.21-rc1-BK

11. ethernet address FF:FF:FF:FF:FF:FF

12. "giant packet from ff:ff:ff:ff:ff:ff"

13. Oops in firewire (2.4.21-pre5 with 2.4.21-pre4 firewire driver)