[Bug 257093] [NEW] tmpreaper should automatically avoid /tmp/aquota.{user, group}

2008-08-11 Thread Steven Black
Public bug reported:

The old quota file format used quota.* names. These names have explicit
protection in tmpreaper's /etc/cron.daily file.

However, the new quota files have a different name. They're named
aquota.*. These are just as important as the older quota-prefixed files.

** Affects: tmpreaper (Ubuntu)
 Importance: Undecided
 Status: New

-- 
tmpreaper should automatically avoid /tmp/aquota.{user,group}
https://bugs.launchpad.net/bugs/257093
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs


[Bug 257093] Re: tmpreaper should automatically avoid /tmp/aquota.{user, group}

2008-08-11 Thread Steven Black

** Attachment added: "Patch for (installation path) /etc/cron.daily/tmpreaper"
   http://launchpadlibrarian.net/16713158/tmpreaper-aquota.patch

-- 
tmpreaper should automatically avoid /tmp/aquota.{user,group}
https://bugs.launchpad.net/bugs/257093
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs


[Bug 258369] [NEW] mailagent package has wrong version number (3.73 should be 3.0.73)

2008-08-15 Thread Steven Black
Public bug reported:

Binary package hint: mailagent

Mailagent has an incorrect package version.

In aptitude it reports itself as:

3.73-26, however mailagent -V reports:
mailagent 3.0 PL73

PL appears to stand for "patch level" which seems to imply it should
actually be 3.0.73.

This is corroborated by the copyright file stating:
It may also be found at most of the
 Comprehensive Perl Archive Network sites, in the location
 <$CPAN/authors/id/RAM/[EMAIL PROTECTED]>.

If you go to http://search.cpan.org/~ram/ (as the CPAN referenced author
ID 'RAM') you see no trace of a mailagent-3.73. Instead you see:
mailagent-3.0.73

It would seem there is no mailagent-3.73 released. The version number is
way off, particularly if Raphael Manfredi decides to release a mailagent
3.1 at some point.

An "epoch version" can, and should, be used to correct the version
number issue.

This would make the next version, according to the debian/changelog
file: 1:3.0.73-1

Cheers,
Steven Black

** Affects: mailagent (Ubuntu)
 Importance: Undecided
 Status: New

-- 
mailagent package has wrong version number (3.73 should be 3.0.73)
https://bugs.launchpad.net/bugs/258369
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs


[Bug 234409] [NEW] friendly-recovery violates the Linux Filesystem Hierarchy Standard

2008-05-23 Thread Steven Black
Public bug reported:

Binary package hint: friendly-recovery

friendly-recovery uses files in /usr, but runs in single-user mode.

This is a violation of the Linux Filesystem Hierarchy Standard, which
states that "The contents of the root filesystem must be adequate to
boot, restore, recover, and/or repair the system." (See
http://www.pathname.com/fhs/pub/fhs-2.3.html#PURPOSE2 )

friendly-recovery, due to the current install locations:

(1) prevents the system from booting in recovery mode if the /usr partition is 
corrupted.
(2) prevents /usr from being resized, as it can never be unmounted

I'm a fan of LVM, and found #2 a huge problem for me. A huge benefit of
LVM is that it allows you to dynamically increase the size of
partitions. This benefit is negated if you can't unmount some of your
partitions due to software using it.

** Affects: friendly-recovery (Ubuntu)
 Importance: Undecided
 Status: New

-- 
friendly-recovery violates the Linux Filesystem Hierarchy Standard
https://bugs.launchpad.net/bugs/234409
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs


[Bug 234409] Re: friendly-recovery violates the Linux Filesystem Hierarchy Standard

2008-05-23 Thread Steven Black
My preference:

Instead of /usr/share, the recovery menu logic moves to /lib. Either
whiptail needs to also move to /bin, or you need to fall-back to a
shell-based solution if whiptail is unavailable (such as the partition
not available).

-- 
friendly-recovery violates the Linux Filesystem Hierarchy Standard
https://bugs.launchpad.net/bugs/234409
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs


[Bug 234421] [NEW] single-user mode should work when libraries are broken

2008-05-23 Thread Steven Black
Public bug reported:

Binary package hint: friendly-recovery

It would be great if the recovery-menu project could add a static-
compiled stub that tries to run the recovery-menu with bash, and if that
fails it would try to fall back to a known static-compiled shell (such
as sash, busybox-static, or bash-static) then falling back to
/sbin/sulogin if nothing else is available.

In my opinion, attempting to fall-back to a static-compiled shell before
falling back to sulogin does not significantly present any greater
security concern. The Ubuntu default is to have the root password locked
out, and administrators which like security will know that the best way
to protect single-user mode is to add a GRUB-based password, as that
retains the security of having the root account locked out. With the
default installation of recovery-menu, Ubuntu has moved in a direction
requiring boot-loader security to prevent the system from being
manipulated in single-user mode. (With just the three options now
available, it may be possible for a person to "fix" another person's X
configuration to something unexpected.)

Additionally, with the recovery-menu logic automatically falling back to
a staticly compiled shell, we can hopefully avoid static shell packages
shadowing the root account to create a "root account with a static
shell", like sash does with the "sashroot" account in hardy.

** Affects: friendly-recovery (Ubuntu)
 Importance: Undecided
 Status: New

-- 
single-user mode should work when libraries are broken
https://bugs.launchpad.net/bugs/234421
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs


[Bug 234434] [NEW] sash creates 'sashroot' account

2008-05-23 Thread Steven Black
*** This bug is a security vulnerability ***

Public security bug reported:

Binary package hint: friendly-recovery

When you install sash it clones your root account to create a 'sashroot'
account.

This is useless with Ubuntu, as Ubuntu has root's account locked out.
This means sash is cloning a locked out acount, which does no one any
good.

Additionally, on systems with root account passwords, this is a security
concern. Root account passwords are serious business, and they regularly
should be changed. By cloning the password, you are effectively by-
passing the normal processes in place in an institution to regulate the
root password.

As an example:
Company A has a system administrator, Evil-Bill. Now, Evil-Bill knows the 
institution will change the root passwords on all the systems when he leaves. 
He also knows that packages are not strictly watched. (How many places actually 
strictly monitor packages and the user accounts they each create?) Before he 
leaves, he installs 'sash'. They change all the root accounts, but they miss 
his backdoor account 'sashroot'. A few weeks after he has left, he logs in and 
performs his evil.

Note that this security concern occurs on Ubuntu systems in cases where
the administrator thought that creating a password for the root user
increased security of single-user mode, or in cases where administrative
policy at an institution requires setting/changing the root password on
a regular basis.


Another solution to the problem addressed by creating the 'sashroot' account 
would be to use standard package logic to ask the user if their 
root/single-user sessions should use a potentially more reliable staticly 
compiled shell. Then all packages providing static shells should offer 
alternatives. A name such as /bin/static-sh could be used as the common 
alternative name. Then with the root account set to /bin/static-sh, things 
should just work. (You could go as far as making /bin/bash a super-low 
recommendation for /bin/static-sh, so that if things didn't get cleared up 
properly when all the static shells were removed, the root account would still 
be accessable.)

** Affects: sash (Ubuntu)
 Importance: Undecided
 Status: New

** Visibility changed to: Public

-- 
sash creates 'sashroot' account
https://bugs.launchpad.net/bugs/234434
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs


[Bug 234434] Re: sash creates 'sashroot' account

2008-05-23 Thread Steven Black
Correcting the package this affects.

** Changed in: sash (Ubuntu)
Sourcepackagename: friendly-recovery => sash

-- 
sash creates 'sashroot' account
https://bugs.launchpad.net/bugs/234434
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs


[Bug 234434] Re: sash creates 'sashroot' account

2008-05-23 Thread Steven Black
I wanted to add that when sash is purged, it does not remove the
sashroot account. -- However the shell gets changed to bash.

This means that if packages are being monitored by regular daily reports
pulled from a dpkg -l list, you can get around this showing up in logs
by installing sash and immediately purging sash.

-- 
sash creates 'sashroot' account
https://bugs.launchpad.net/bugs/234434
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs


[Bug 368127] Re: ToME Crashes Randomly

2009-08-04 Thread Steven Black
I ran across this bug report when I was experiencing a very similar
problem -- one which I think Adam may be experiencing.

Adam, is it really crashing randomly, or does it reliably crash when you
try to save the game? When it crashed via saving the game, it will
likely leave a save-file around which will also crash if you try to load
it. If that's the problem you're encountering, then it is the same
problem I've been having.

That problem has been reported upstream, and upstream has a patch:
http://wiki.t-o-m-e.net/BugReport930

I can confirm that after applying the patch on that page, I can save new
games. The old partially-saved games (from earlier attempts to save the
game) are a total waste and need to be removed. The way to check (and
remove the crap that won't load) is simple: If the game saved properly,
it should show up in the list of saved games (on the "New
Character"/"Load Character" screen).

The saved files are placed in "~/.tome/2.3/save/". The bogus files show
up in pairs, like: PLAYER.new  PLAYER.pnc -- PLAYER (the default save
file name) isn't usable until these two files are removed. Proper save
files will have no extension (just PLAYER) with an additional
user.$UID.svg descriptor file (appears to be one-line per save file, or
per character -- I'm not sure. $UID is my system user-id).

I'd love to see this patch rolled in, as the game is extremely
frustrating without it. A new player starts a new game, has a great
time. tries to save the game, and *poof* -- the game is lost.

As the URL says, this bug is more visible on Ubuntu systems due to the
use of Fortify. This particular bug wouldn't cause memory corruption
until the user saves and that usually happens at application termination
-- it is quite possible that folks not using Fortify would have never
noticed the bug. The upstream folks are also working on 3.0. We need
this patch.

-- 
ToME Crashes Randomly
https://bugs.launchpad.net/bugs/368127
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs


[Bug 368127] Re: ToME Crashes Randomly

2010-07-23 Thread Steven Black
It looks like wiki.t-o-m-e.net is currently off-line.

Worst case this patch should be easy to recreate.

There's a buffer overflow in the save-game logic. Since it only happened
saving the game, and most folks save the game before they quit, folks
never noticed the buffer overflow... until Valgrind became the default
with Ubuntu.

If you build the code from source (with debugging symbols) and run gdb
on it it should point you to the source of the error. To reproduce the
problem all you need to do is create a new game and try to immediately
save it.

I've had some issues with some of my hardware since I commented on this
ticket and I'm fairly certain I no longer have the original patch on any
of my systems.

This should be a straight-forward process, so I'm going to try it myself
and whip up a patch.

-- 
ToME Crashes Randomly
https://bugs.launchpad.net/bugs/368127
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs


[Bug 368127] Re: ToME Crashes Randomly

2010-07-23 Thread Steven Black
Okay. I'm not sure how similar this is to the original patch, but the
following three line patch takes care of the issue.

There may be some other consequence of this. I didn't increase the
buffer so it is getting truncated, but I don't see it.

I'm just using strncpy instead of strcpy to make sure we don't write too
much. I also explicitly set the last character of the array to be a NUL
byte so it remains terminated even if it hits the limit.

It worked with my test on Ubuntu 10.04, but it was a fairly quick test.

Again, this fixes total breakage of this package in Ubuntu so this
patch, the original patch, or a similar patch should be rolled in. The
game is frustratingly unusable without a fix for this bug.


** Patch added: "tome-savefix.patch"
   http://launchpadlibrarian.net/52403339/tome-savefix.patch

-- 
ToME Crashes Randomly
https://bugs.launchpad.net/bugs/368127
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs