Debian 'Stable' not so stable.

Debian 'Stable' not so stable.

Post by alex.gma » Fri, 18 Nov 2005 07:08:32


I have a USB DVD-RW. When I unplug it while /cdrom is mounted, and then
do `umount /cdrom`, a Segmentation Fault occurs (Don't tell me to
unmount first, then unplug. I know. But a segfault is always a
programmer's error) Needless to say, nothing of the kind ever happened
to me while using XP on the same hardware.
 
 
 

Debian 'Stable' not so stable.

Post by Peter T. B » Fri, 18 Nov 2005 07:16:14


No it doesn't. You will likely get a kenrel oops, look in your log. You
have just wriecked the programming assumptions - it's like taking a
hammer to your hard disk while the



No it isn't. This is a userspace manifestaton of a kaput kernel. You have
an ex-kernel. Do not cut the wires to your hard disk!


Of course it did. There's no avoiding death after disappearing file
system media! If you are going to do that kind of thing you will need to
use the supermunt driver to manage the media for you, and present a
facsimile of the device to the kernel when it has been yanked.

Peter

 
 
 

Debian 'Stable' not so stable.

Post by Keith Kell » Fri, 18 Nov 2005 07:21:15


Why don't you file a bug report with Debian and/or the maintainers of
mount? (Or automount, if you're using that for your cdrom; or hotplug,
if you're using that for your cdrom. But it doesn't sound like either
of those are involved.)

Oh, and umount first.


Needless to say, I don't care what XP does or doesn't do. A bug is
relative to what the software should be doing, not to what some other
piece of software does or doesn't do.

--keith

--
XXXX@XXXXX.COM
(try just my userid to email me)
AOLSFAQ= http://www.yqcomputer.com/
see X- headers for PGP signature information
 
 
 

Debian 'Stable' not so stable.

Post by alex.gma » Fri, 18 Nov 2005 08:50:20


You can't read? I'm telling you what DOES happen, not what will likely
happen. I don't need to look in dmesg.

$ umount /cdrom
Segmentation Fault
$
 
 
 

Debian 'Stable' not so stable.

Post by OTID » Fri, 18 Nov 2005 11:15:41


LOL, Peter, this is what you get for answering trolls. What a moron,
and so uppity, too.

It has to be noted, however, that the word "stable" may mean different
things. It may mean "robust", but also "unchanging". In the later case,
bugs fixes developed long ago are not accepted in order not to jeopardize
the predictability of a stable version. In Fedora Core 4, unplugging USB
CD-ROM while mounted is perfectly ok, for example. However, I see no
value in incorporating disruptive bugfixes into Debian (especially
"stable").

-- Pete
 
 
 

Debian 'Stable' not so stable.

Post by Peter T. B » Fri, 18 Nov 2005 15:58:20


Read your kernel log.

(if you don't get an ops, then you have survived, and a segfault is a
good result from such a bizarre thing to do).

Peter
 
 
 

Debian 'Stable' not so stable.

Post by Peter T. B » Fri, 18 Nov 2005 16:04:37


You think there is a kernel bugfix here? Surely he CANNOT remove a cd
while it is mounted because the openers count would be positive and
hence the driver would not allow the "open cd door" ioctl. Therefore he
has forced it by "cutting the cables" (yanking them out). That's physical
destruction.

When the medium is destroyed in-situ like that, it's hard to know what to
do. I would imagine a kernel oops would follow. But if it doesn't, then
at least the kernel will error out all requests, and the file system
code would maybe oops (where'd my inode go)? If not, then maybe the
application layer will segfault. The latter seems to me to be a good
result, but I daresay the author of the app will prefer to catch the
segfault and print out "owwwww".


Peter
 
 
 

Debian 'Stable' not so stable.

Post by alex.gma » Fri, 18 Nov 2005 17:17:42


Are you saying that I defied the laws of physics?


Wow, what an incredible command of logic!

You are saying:

unplugging cables == cutting cables
cutting cables == physical destruction

Therefore, unplugging cables is physical destruction

Analogously,

sitting in a chair == sitting on a toilet
sitting on a toilet == defecation

Therefore sitting in a chair implies defecation.

Do you get it now? Equivalence of one aspect of objects does not imply
equivalence in other aspects. Unplugging a cable is not physical
destruction. Under XP, I plug and unplug this external DVD-RW all the
time. If Linux can't handle this, tough shit, scored one up for XP.

P.S. And by the way, please see a shrink, you just don't sound very
sane to me. Really.
 
 
 

Debian 'Stable' not so stable.

Post by alex.gma » Fri, 18 Nov 2005 17:36:45


You said segmentation fault can not occur. But it does. Therefore, you
are wrong. What's in the kernel log will not change that. omprende?
 
 
 

Debian 'Stable' not so stable.

Post by Peter T. B » Fri, 18 Nov 2005 17:38:06


Physics? The laws of O/Ss, yes. You had to physically do something that
is forbidden via the provided interfaces, such as knocking the wall down
with a machanical ball and chain instead of walking in through the door.



Good.


Yep.


Yep.


Nope.


It isn't. Sitting in a chair is the same thing as loading a chair,
however.


It doesn't. Loading a toilet is sitting on it.


No it doesn't.


Of course not. I provided proper abstractions both times: unplugging is
an instance of cutting the connection, and cutting the connection is an
instance of destroying the connection, therefore you destroyed the
connection (wires).


Of course it is. In what way is it different from cutting them?

Would you mind yanking the cables from your TV a few times?



CDs are not intrinsically hot-pluggable and neither are hard disks.
That you can physically destroy the wires doesn't excuse you when you do
it. But I must add that you are mistaking what happens - it likely IS
the case that the device driver copes OK with the disappearing device,
but you probably need a FS emulation driver above the device driver if
you are going to be doing that sort of thing in order to keep the fs
code happy with mutating bytes on a readonly medium - use the correct
mount options. You get the same results as yanking a floppy the way you
are ging about it.


It handles it fine.


Keep your foul-mouthed insults to yourself.


Peter
 
 
 

Debian 'Stable' not so stable.

Post by Peter T. B » Fri, 18 Nov 2005 17:54:26


I cannot be "wrong" about something only you know - it is not in the
domain of discourse. Any time I say "you have chosen the number 7", you
can say "no - you're wrong". I am telling you what has PROBABLY happened,
and if you go and look where I advised, you will be able to tell us
whether that is the case or not, and thus bring something more to the
discussion.

What's in the kernel log will not change that. _Comprende?

Cease insulting behaviour, NOW. If you are interested in learning what is
going on, start again, and make it evident.

Peter
 
 
 

Debian 'Stable' not so stable.

Post by alex.gma » Fri, 18 Nov 2005 18:05:08


OMFG. Read the quote above.

I wrote "a Segmentation Fault occurs" and you replied "No, it doesn't".

Now, you are saying that you couldn't have been wrong, BECAUSE you were
talking about something you could NOT know?!

Did they put DSL in the ward?

I'm out of this thread. I want to keep *MY* sanity.
 
 
 

Debian 'Stable' not so stable.

Post by Peter T. B » Fri, 18 Nov 2005 18:28:15


Cease rudeness.


That's my opinion. Check. I told you how.

Are you interested in conversation, or is insults where you get off the
bus at?

Peter
 
 
 

Debian 'Stable' not so stable.

Post by Kasper Dup » Fri, 18 Nov 2005 19:32:30


If you just want to complain you have come to the wrong group.

If you actually want a problem solved, then for .... sake tell
us what the error messages in the log says. And also tell us
what kernel version you are using (or even better reproduce
the problem with the newest kernel and then tell us what it
says).

--
Kasper Dupont
Note to self: Don't try to allocate
256000 pages with GFP_KERNEL on x86.
 
 
 

Debian 'Stable' not so stable.

Post by Rich Walke » Sat, 19 Nov 2005 02:04:56

"Peter T. Breuer" < XXXX@XXXXX.COM > writes:




*BUT HE HAS AN EXTERNAL USB ONE*

Users expect USB devices to be hot-pluggable.

This is a corner case, and it causes umount to fail.

sheesh.

cheers, Rich.

--
rich walker | Shadow Robot Company | XXXX@XXXXX.COM
technical director 251 Liverpool Road |
need a Hand? London N1 1LX | +UK 20 7700 2487
www.shadow.org.uk/products/newhand.shtml