SEO

#1 (**permalink**) 01-12-2008, 05:32 AM

Why doesn't Sun fix this damn bug, so that it is POSIX compliant?

#mkdir data
#chmod 000 data
#ls data/.
#echo $?
0

This command should fail. HP fixed this bug 2 years ago.

Cmon Sun, this should be a fairly easy fix.

-Chuck

#2 (**permalink**) 01-12-2008, 05:32 AM

In article <3f671ca0$1_1@127.0.0.1>, Rhugga <rhugga@yahoo.com> wrote:

> Why doesn't Sun fix this damn bug, so that it is POSIX compliant?

> #mkdir data
> #chmod 000 data
> #ls data/.
> #echo $?
> 0

$ uname -rs
SunOS 5.8
$ mkdir data
$ chmod 000 data
$ ls data/.
data/.: Permission denied
$ echo $?
2
$

> This command should fail. HP fixed this bug 2 years ago.

Did you test as uid 0 (your # prompt indicates that you do)?
Does POSIX say that even uid 0 should get an error in your example?
If so, why should your example be special compared with all
other situations where privileges are ignored for uid 0?

--
Göran Larsson http://www.mitt-eget.com/

#3 (**permalink**) 01-12-2008, 05:32 AM

Michael Tosch wrote:

> In article <3f671ca0$1_1@127.0.0.1>, Rhugga <rhugga@yahoo.com> writes:
>
>>Why doesn't Sun fix this damn bug, so that it is POSIX compliant?
>>
>>#mkdir data
>>#chmod 000 data
>>#ls data/.
>>#echo $?
>>0
>>
>>This command should fail. HP fixed this bug 2 years ago.
>>
>>
>
>
> Your are probably using /usr/ucb/ls, /bin/ls works ok.
> Sun has very broken tools in /usr/ucb and NEVER fixes any of them!
>
> The typical fixes are:
>
> # rm /usr/ucb/ls /usr/ucb/sed; ln -s ../bin/ls; ln -s ../bin/sed
> Then edit /usr/ucb/file, replace $* by "$@"
>

Nah, I tried gnu ls, /usr/bin/ls and etc.. Besides, this bug is uch
deeper than the commands themselves. Some posts on another boardx
suggest a kernel bug. However, this problem does not exist on the -07
rev on the Solaris 9 kernel but it does on the -08 rev.

????

-Chuck

#4 (**permalink**) 01-12-2008, 05:32 AM

* Michael Tosch wrote:

> Your are probably using /usr/ucb/ls, /bin/ls works ok.
> Sun has very broken tools in /usr/ucb and NEVER fixes any of them!

/bin/ls does the same thing. Sun are *right* not to fix the `bugs'
in the /usr/ucb things, they're *meant* to behave the way the BSD ones
did, after all.

I'd be interested in knowing if the behaviour is OK by POSIX for UID 0
though. These cases are always hard to remember.

--tim

#5 (**permalink**) 01-12-2008, 05:33 AM

Rhugga <rhugga@yahoo.com> writes in comp.sys.sun.admin:
|Why doesn't Sun fix this damn bug, so that it is POSIX compliant?

File a bug report via your support contract if you need it fixed
and it will be fixed quickly. Otherwise, bug fixes tend not to
show up until the next full release, which is currently every 2-3 years.

--
__________________________________________________ ______________________
Alan Coopersmith alanc@alum.calberkeley.org
http://www.CSUA.Berkeley.EDU/~alanc/ aka: Alan.Coopersmith@Sun.COM
Working for, but definitely not speaking for, Sun Microsystems, Inc.

#6 (**permalink**) 01-12-2008, 05:34 AM

Rhugga <rhugga@yahoo.com> writes:

>Why doesn't Sun fix this damn bug, so that it is POSIX compliant?

>#mkdir data
>#chmod 000 data
>#ls data/.
>#echo $?
>0

>This command should fail. HP fixed this bug 2 years ago.

We didn't introduce this bug until Solaris 9 (less than
two years ago) and only for UFS filesystems.

Casper
--
Expressed in this posting are my opinions. They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.

#7 (**permalink**) 01-12-2008, 05:34 AM

Rhugga <rhugga@yahoo.com> writes:

>Why doesn't Sun fix this damn bug, so that it is POSIX compliant?

>#mkdir data
>#chmod 000 data
>#ls data/.
>#echo $?
>0

>This command should fail. HP fixed this bug 2 years ago.

BTW, the bug ID is 4789213. (filed 2002/12/05)

Casper
--
Expressed in this posting are my opinions. They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.

#8 (**permalink**) 01-12-2008, 05:34 AM

* Casper H S Dik wrote:

> We didn't introduce this bug until Solaris 9 (less than
> two years ago) and only for UFS filesystems.

It's there in solaris 8 2/02 with some fairly recent patch set (if the
bug is that even root should not be able to do this).

--tim

#9 (**permalink**) 01-12-2008, 05:34 AM

Tim Bradshaw <tfb@cley.com> writes:

>* Casper H S Dik wrote:

>> We didn't introduce this bug until Solaris 9 (less than
>> two years ago) and only for UFS filesystems.

>It's there in solaris 8 2/02 with some fairly recent patch set (if the
>bug is that even root should not be able to do this).

Root should be able to do it. (Root effectively ignores
file permissions except in one case: running executables by
root requires that at least one of the "x" bits is set)

Casper
--
Expressed in this posting are my opinions. They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.

#10 (**permalink**) 01-12-2008, 05:34 AM

* Casper H S Dik wrote:

> Root should be able to do it. (Root effectively ignores
> file permissions except in one case: running executables by
> root requires that at least one of the "x" bits is set)

Yes, that's what I would have assumed is the case (not being
intimately familiar with the POSIX specs...).

Is it the case that ordinary users can do this in Solaris 9 (which
would be really serious)? Or is the bug that they *can't* but ls
exits with 0 anyway? I did have a look for the bug report on sunsolve
but I couldn't find it either because the searching gadget is bust or
(more likely) because I can't drive it...

--tim

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

SEO

Why Doesn't Sun Fix This Frigging Bug???