Mailing List Archive

Break sequences acting up?
Hi all,

I'm having some fun with version 8.1.1 on Debian (Sarge) on an HP
Prolient, with a Rocketport 550 4x serial card.

I have these entries in the config file:


default full {
rw *;
}
default * {
logfile /var/log/conserver/&;
timestamp "";
include full;
}

break 8 {
delay 250;
string "\d\z\ds\d\z\ds\d\z\ds";
}

break 9 {
delay 250;
string "\d\z\ds\d\z\ds\d\z\ds";
}

access * {
allowed 127.0.0.1;
}
console willow {
master localhost;
type device;
device /dev/ttyr3;
baud 115200;
parity none;
}





My keystrokes follow the ">>>" below:

>>> ctrl-e c l 9
willow:~# [halt sent]
SysRq : HELP : loglevel0-8 reBoot tErm kIll saK showMem Off showPc unRaw
Sync showTasks Unmount
9SysRq : HELP : loglevel0-8 reBoot tErm kIll saK showMem Off showPc
unRaw Sync showTasks Unmount
9SysRq : Emergency Sync
Syncing device 68:01 ... OK
Syncing device 68:03 ... OK
Done.
>>> ctrl-e c l 8
[halt sent]
SysRq : HELP : loglevel0-8 reBoot tErm kIll saK showMem Off showPc unRaw
Sync showTasks Unmount
8SysRq : HELP : loglevel0-8 reBoot tErm kIll saK showMem Off showPc
unRaw Sync showTasks Unmount
8SysRq : Emergency Sync
Syncing device 68:01 ... OK
Syncing device 68:03 ... OK
Done.
>>> ctrl-e c l ?
[halt list]
0 - 250ms, `\z'
1 - 250ms, `\z'
2 - 250ms, `\r~^b'
3 - 250ms, `#.'
4 - 600ms, `\r\d~\d^b'
8 - 250ms, `\d\z\ds\d\z\ds\d\z\ds'
9 - 250ms, `\d\z\ds\d\z\ds\d\z\ds'

>>> carriage-return
-bash: 9988: command not found





This seem pretty bizarre to me - my interpretation of what the Linux
console is seeing is this:

I type ctrl-e c l 9

Instead of seeing:

<break> s <break> s <break> s

It sees:

<break> "something or other" 9 <break> "something or other" 9 <break> s


Any ideas?

Tim.
Re: Break sequences acting up? [ In reply to ]
On Tue, May 18, 2004 at 03:37:40PM +0100, Tim Small wrote:
> Instead of seeing:
>
> <break> s <break> s <break> s
>
> It sees:
>
> <break> "something or other" 9 <break> "something or other" 9 <break> s
>
>
> Any ideas?

yeah...it kinda makes sense. not in a "that's the way it's supposed to
be", but in "a bug could make it behave that way" sorta way.

would you be willing to do the same thing, but have conserver in debug
mode (use -DD, for the level of detail that might be useful)? there
might be a password in there you would want to strip out, but i'd love
to see the rest of the output. it would help me determine where the bug
actually is (gotta be a bug...just not sure how to reproduce it - can't
so far).

Bryan
Re: Break sequences acting up? [ In reply to ]
On Tue, May 18, 2004 at 12:05:27PM -0700, Bryan Stansell wrote:
> to see the rest of the output. it would help me determine where the
> bug actually is (gotta be a bug...just not sure how to reproduce it -
> can't so far).

no need for the debug output...i was able to reproduce it. it didn't
*look* like there was a problem when under solaris (where i first
tried it...though closer examination of debug output did show the
problem) but it was loud and clear under linux.

i've attached a patch for 8.1.5, but it works with 8.1.1 (and all
versions in between) as well (and possibly earlier, but i haven't
checked).

depending on the break sequences you use, you may or may not see the
problem. if you have \d or \z sequences with non-\d/\z characters, you
could very well hit the problem (but it depends on how things are
ordered). a simple series of \z and \d (one or more) won't trigger the
bug.

Bryan
Re: Break sequences acting up? [ In reply to ]
Bryan Stansell wrote:

>On Tue, May 18, 2004 at 12:05:27PM -0700, Bryan Stansell wrote:
>
>
>>to see the rest of the output. it would help me determine where the
>>bug actually is (gotta be a bug...just not sure how to reproduce it -
>>can't so far).
>>
>>
>
>no need for the debug output...i was able to reproduce it. it didn't
>*look* like there was a problem when under solaris (where i first
>tried it...though closer examination of debug output did show the
>problem) but it was loud and clear under linux.
>
>
Excellent - that's what I call service! Got the two messages together
this morning (GMT+1) - have applied the patch, and the behaviour is now
sane...

Cheers,

Tim.