Re: usb modules not loaded (w/ fix)
by Pete Zaitcev
On Sat, 10 Apr 2004 12:39:24 -0700
Jurgen Botz <jurgen(a)botz.org> wrote:
> Some time in the last week or so my system stopped loading
> the usb bus driver modules, uhci_hcd and ehci_hcd, at startup.
>
> In rc.sysinit, there is the following test:
>
> if ! LC_ALL=C fgrep -q "usb" /proc/devices 2>/dev/null ; then
> aliases=`/sbin/modprobe -c | awk '/^alias usb-controller/ { print
> $3 }'`
> ...
>
> Presumably the intent is that if usb already appears in
> /proc/devices, then the driver must already be loaded / or
> compiled into the kernel? Well, for whatever reason my system
> now shows a line:
>
> 180 usb
Bill, if you wish to probe, I suggest using /proc/bus/usb/devices and
check if it's present and non-empty. This should work on both 2.4 and
2.6. It needs usbcore to be loaded before it can be mounted.
-- Pete
20 years
Prelink hates me
by Scott Sloan
Prelink ran not 5 mins ago and I notice a huge increase in memory usage,
jumping from 15% to 94%, nothing changing but prelink running. I was
wondering if anyone else has had a similar occurrence?
--
Scott Sloan
------------
"I'm not a genius. I'm just passionately curious" -- Einstein
20 years
.desktop policy
by Gérard Milmeister
It is required that categories for .desktop files are selected from the
list in http://freedesktop.org/Standards/menu-spec/0.8/apa.html. For
example for DosEmu, one would choose Emulator. However most categories
will not be included in the gnome menu, according to
/etc/X11/desktop-menus/applications.menu. So one is forced to select one
of those categories mentioned in this file for the entry to appear at
all and the list from fdo doesn't make much sense and is rather
misleading for packagers.
On the other hand, applications.menu lists a directory named "Other"
which probably should catch the rest, but this menu doesn't appear at
all in my menu.
--
Gérard Milmeister
Tannenrauchstrasse 35
8038 Zürich
gemi(a)bluewin.ch
20 years
usb modules not loaded (w/ fix)
by Jurgen Botz
Some time in the last week or so my system stopped loading
the usb bus driver modules, uhci_hcd and ehci_hcd, at startup.
In rc.sysinit, there is the following test:
if ! LC_ALL=C fgrep -q "usb" /proc/devices 2>/dev/null ; then
aliases=`/sbin/modprobe -c | awk '/^alias usb-controller/ { print
$3 }'`
...
Presumably the intent is that if usb already appears in
/proc/devices, then the driver must already be loaded / or
compiled into the kernel? Well, for whatever reason my system
now shows a line:
180 usb
in /proc/devices even without any of the usb modules being loaded,
so the above test finds that doesn't execute its body.
My solution for now was to comment out the condition, executing
the body unconditionally.
:j
20 years
Re: [RFC] User Accesable Filesystem Hierarchy Standard
by Jamethiel Knorth
>Date: Wed, 7 Apr 2004 08:24:43 -0400
>From: Alan Cox <alan(a)redhat.com>
>
>On Wed, Apr 07, 2004 at 08:00:45AM -0400, Jamethiel Knorth wrote:
> > >"Other people fire shotguns at random without warning, lets all do
>that"
> >
> > More like, "People have a tendency to fire shotguns at random without
> > warning. Mayhaps we should expect them to."
>
>Which means stopping them from doing it without a lot of thought.
Certainly. And, requiring a root password for every install will cause
people to think little of giving a root password. If most installs do not
require a root password, those installs will require less thought, but they
will also be incapable of harming the root system and side-effects will be
less catastrophic.
Of course, even without this, any malware can install without a root
password. It merely installs to the home directory and it has every bit as
much influence as a program under this proposal would have. However, if a
root password is required to affect the core system, but not to do a common
install, any malware which affects the root system will have a much harder
time tricking people, as they will not be adjusted to seeing a request for
the root password.
Also, in a home desktop situation, the owner of the computer can easily
allow other people to install programs without risking them leaving the
system FUBAR.
> > The 10,000 private installations can be solved by a decent package
> > management system which will notify the administrator of multiple
> > installations. This system will also make it more likely an
>administrator
>
>You've never run a large student system have you 8)
Okay, let me see. Step 1, quota home-dir space. Now, if people want to
install stuff, they may run out of space. They can choose between storage
space in program space at their leisure. No step 2 required for this plan to
work.
Now, a really nice package manager which is tracking all the various user
databases could pop-up a notification to the sys-admin saying, "8000 users
have program X installed!" The sys-admin, getting this useful message can
then say, "Hmmm, maybe people want to use that program, I'll see about
installing it properly and giving them all messages that there is now a
central install," or "Who cares what they do with their private space! Never
show this message again!"
And, if users are not allowed to install programs, their home-directories
can be prevented from containing executable files.
> > which is currently rather lacking. The last time I ran into a group
> > project, the sharing of stuff was so much trouble, people decided to
>just
> > share out massive swaths of their home directories and hope no-one else
> > messed with them.
>
>ACLS fix most of this. With basic unix permissions you basically need an
>admin to set it up otherwise.
ACLS do fix this, I have heard many times. I have no doubt of that, and that
is great. The problem still remains that there needs to be a place to put
the shared information. Having a standardized way to handle group
directories would help with this. Right now, basically any distribution on
any architecture can handle home directories from any other setup because
they're basically standard. Sure, the existence of /home/ isn't actually
required, but everyone uses it that way.
Having a standard for the way these things are done would allow third-party
programs to more easily target mixed environments and multiple platforms.
Further, I haven't seen an incredibly easy way to manage ACLS. The nice
thing about a directory-base system is that File-Managers are extremely well
developed and powerful tools. If the problem can be solved with simple
tools, complex tools should be avoided. Obviously, if the simple tools
genuinely will not work, they should not be used. However, I see no reason
that these simpler tools will not work.
Also, this standard does other things.
It creates a standard place for users to put their own programs, when
currently it is something decided by users at random. Some people seem to
think that doing a private installation of a program should be a privilege
of the geek elite. I strongly disagree with this. There is no reason to
split users into groups like this, into the unskilled and the skilled. I am
strongly in favor of empowering users.
The other thing this does is organizes where configuration files are put.
Once the next draft is put up, it will even properly support having multiple
sets of configuration files, such that a user can actually have working
configurations for multiple distributions in one home directory. As this
currently doesn't work properly (I am told this, but do not know
personally), this is a major step forward.
My apologies for the tardy reply.
_________________________________________________________________
FREE pop-up blocking with the new MSN Toolbar get it now!
http://toolbar.msn.com/go/onm00200415ave/direct/01/
20 years
xinerama and gdm
by Eric Hattemer
Is there a better way to turn on xinerama than to add +xinerama into the
gdm.conf file? It gets overwritten every time gdm is upgraded. Is
there a particularly good reason xinerama is off by default?
-Eric Hattemer
20 years
rawhide report: 20040412 changes
by Build System
Updated Packages:
coreutils-5.2.1-5
-----------------
* Fri Apr 09 2004 Dan Walsh <dwalsh(a)redhat.com> 5.2.1-5
- Add ls -LZ fix
- Fix chcon to handle "."
* Wed Mar 17 2004 Tim Waugh <twaugh(a)redhat.com>
- Apply upstream fix for non-zero seconds for --date="10:00 +0100".
kernel-2.6.5-1.315
------------------
* Fri Apr 09 2004 Arjan van de Ven <arjanv(a)redhat.com>
- 2.6.5-mc3
- finish up the -mc2 merge
- latest 4g/4g patch from Ingo
- latest execshield patch from Ingo
- fix a few framebuffer bugs
policy-1.10.2-4
---------------
* Fri Apr 09 2004 Dan Walsh <dwalsh(a)redhat.com> 1.10.1-4
- Fix rpmscript to allow get_security
* Fri Apr 09 2004 Dan Walsh <dwalsh(a)redhat.com> 1.10.1-3
- Add games support
- Fix some samba issues
* Thu Apr 08 2004 Dan Walsh <dwalsh(a)redhat.com> 1.10.1-2
- Add apmd support for killof
rpmdb-fedora-1.91-0.20040412
----------------------------
zsh-4.2.0-1
-----------
* Sat Apr 10 2004 Jens Petersen <petersen(a)redhat.com> - 4.2.0-1
- update to 4.2.0 stable release
- zsh-4.0.7-bckgrnd-bld-102042.patch no longer needed
- add compinit and various commented config improvements to .zshrc
(Eric Hattemer,#114887)
- include zshprompt.pl in doc dir (Eric Hattemer)
- drop setenv function from zshrc
20 years
PCMCIA problems
by Shahms King
Currently, the PCMCIA startup script checks /proc/devices for pcmcia and
only loads the $PCIC module if it isn't defined. Unfortunately, at
bootup this is defined even if the $PCIC module hasn't been loaded.
This means that cardmgr thinks it doesn't have anything to manage and
none of my PCMCIA cards work. (Most important being the internal PCMCIA
wireless device). I'm not sure if this is actually the fault of the
PCMCIA startup script or if the appropriate module (yenta_socket) should
be loaded automagically and just isn't...
--
Shahms King <shahms(a)shahms.com>
20 years
fixfiles relabel errors
by Brian Millett
[root@mktg6nt selinux]# rpm -qa "*policy*"
policy-1.10.2-1
policycoreutils-1.10-1
checkpolicy-1.10-1
After upgrading 4/08/2004, I get the following trying to "fixfiles"
[root@mktg6nt root]# fixfiles relabel
Cleaning out /tmp
/usr/sbin/setfiles: read 1370 specifications
/usr/sbin/setfiles: invalid context system_u:object_r:default_t on line
number 39
/usr/sbin/setfiles: invalid context system_u:object_r:root_t on line
number 44
/usr/sbin/setfiles: invalid context system_u:object_r:home_root_t on line
number 53
/usr/sbin/setfiles: invalid context system_u:object_r:home_root_t on line
number 54
/usr/sbin/setfiles: invalid context system_u:object_r:user_home_dir_t on
line number 55
/usr/sbin/setfiles: invalid context system_u:object_r:user_home_dir_t on
line number 56
/usr/sbin/setfiles: invalid context system_u:object_r:user_home_t on line
number 57
/usr/sbin/setfiles: invalid context system_u:object_r:user_home_t on line
number 58
/usr/sbin/setfiles: invalid context system_u:object_r:mnt_t on line
number 62
/usr/sbin/setfiles: invalid context system_u:object_r:var_t on line
number 67
Exiting after 10 errors.
Now, I have SELINUX=disabled in the /etc/sysconfig/selinux. Is this
"normal" behavior?
--
Brian Millett
Enterprise Consulting Group "Shifts in paradigms
(314) 205-9030 often cause nose bleeds."
bpmATec-groupDOTcom Greg Glenn
20 years
rawhide report: 20040409 changes
by Build System
New package planner
A graphical project management tool.
Updated Packages:
checkpolicy-1.10-1
------------------
* Thu Apr 08 2004 Dan Walsh <dwalsh(a)redhat.com> 1.10-1
- Upgrade to the latest from NSA
ddd-3.3.8-4
-----------
* Fri Apr 09 2004 Than Ngo <than(a)redhat.com> 3.3.8-4
- fix gcc 3.4 build problem
e2fsprogs-1.35-7.1
------------------
* Thu Apr 08 2004 Thomas Woerner <twoerner(a)redhat.com> 1.35-7.1
- fixed 'check after next mount' for filesystems with maximum mount count -1
(#117109)
ethereal-0.10.3-1
-----------------
* Thu Apr 08 2004 Phil Knirsch <pknirsch(a)redhat.com> 0.10.3-1
- Update to 0.10.3 because of security issues.
gedit-2.6.0-2
-------------
* Thu Apr 08 2004 Dan Williams <dcbw(a)redhat.com> 1:2.6.0-2
- Fix dumb bug in ~/.recently-used patch where lockf() could
never succeed
gnome-panel-2.6.0-5
-------------------
* Thu Apr 08 2004 Mark McLoughlin <markmc(a)redhat.com> 2.6.0-5
- Fix problem with apm detection in %post on machines whose
APM bios doesn't have battery lifetime support
kdeartwork-3.2.1-3
------------------
* Thu Apr 08 2004 Than Ngo <than(a)redhat.com> 3.2.1-3
- fix file conflict with kdeedu
* Thu Apr 08 2004 Than Ngo <than(a)redhat.com> 3.2.1-2
- fix icontheme name, bug #119015
kdegames-3.2.1-4
----------------
* Thu Apr 08 2004 Than Ngo <than(a)redhat.com> 6:3.2.1-4
- fix dependency bug again, bug #120399
kernel-2.6.5-1.309
------------------
lvm-1.0.3-19.0
--------------
* Wed Apr 07 2004 Alasdair Kergon <agk(a)redhat.com> 1.0.3-19.0
- Add .lvm1 suffix to all binaries; LVM2 is now the default.
This package is only useful to Fedora users reverting to a 2.4 kernel.
* Tue Mar 02 2004 Elliot Lee <sopwith(a)redhat.com>
- rebuilt
* Fri Feb 13 2004 Elliot Lee <sopwith(a)redhat.com>
- rebuilt
lvm2-2.00.11-1.3
----------------
* Thu Apr 08 2004 Alasdair Kergon <agk(a)redhat.com> 2.00.11-1
- Fallback to using LVM1 tools when using a 2.4 kernel without device-mapper.
* Wed Apr 07 2004 Alasdair Kergon <agk(a)redhat.com> 2.00.10-2
- Install the full toolset, not just 'lvm'.
mod_auth_kerb-5.0-0.rc4.5
-------------------------
* Thu Apr 08 2004 Joe Orton <jorton(a)redhat.com> 5.0-0.rc4.5
- remove static globals
- add SSLRequireSSL
net-snmp-5.1.1-2
----------------
* Thu Apr 08 2004 Phil Knirsch <pknirsch(a)redhat.com> 5.1.1-2
- Added Kaj J. Niemi that fixes ipAdEntIfIndex problem (#119106)
- Added Kaj J. Niemi to shut up memshared message for 2.6 kernel (#119203)
pango-1.4.0-2
-------------
* Wed Mar 17 2004 Owen Taylor <otaylor(a)redhat.com> 1.4.0-2
- Fix location for modules file on ppc/ppc64 (#114399)
- Make the spec file check to avoid further mismatches
policy-1.10.2-1
---------------
* Thu Apr 08 2004 Dan Walsh <dwalsh(a)redhat.com> 1.10.1-6
- Lots of fixes for snmpd
* Wed Apr 07 2004 Dan Walsh <dwalsh(a)redhat.com> 1.10.1-5
- Force rebuild always
- Add cifs support
- Macroize sudo
policycoreutils-1.10-1
----------------------
* Thu Apr 08 2004 Dan Walsh <dwalsh(a)redhat.com> 1.10-1
- Upgrade to latest from NSA
rpmdb-fedora-1.91-0.20040409
----------------------------
selinux-doc-1.10-1
------------------
* Thu Apr 08 2004 Dan Walsh <dwalsh(a)redhat.com> 1.8-10
- Upgrade to match NSA
system-config-display-1.0.12-2
------------------------------
* Thu Apr 08 2004 Brent Fox <bfox(a)redhat.com> 1.0.12-2
- fix icon path (bug #120174)
system-config-keyboard-1.2.1-2
------------------------------
* Thu Apr 08 2004 Brent Fox <bfox(a)redhat.com> 1.2.1-2
- fix icon path (bug #120175)
* Wed Nov 12 2003 Brent Fox <bfox(a)redhat.com> 1.2.1-1
- renamed from redhat-config-keyboard
- add Obsoletes for redhat-config-keyboard
- make changes for Python2.3
* Mon Oct 13 2003 Brent Fox <bfox(a)redhat.com> 1.1.5-2
- pull in Croatian translation (bug #106617)
system-config-kickstart-2.5.10-2
--------------------------------
* Thu Apr 08 2004 Brent Fox <bfox(a)redhat.com> 2.5.10-2
- fix icon path (bug #120176)
system-config-language-1.1.5-2
------------------------------
* Thu Apr 08 2004 Brent Fox <bfox(a)redhat.com> 1.1.5-2
- fix icon path (bug #120177)
* Mon Jan 12 2004 Brent Fox <bfox(a)redhat.com> 1.1.5-1
- update locale-list (bug #107450)
* Fri Jan 09 2004 Brent Fox <bfox(a)redhat.com> 1.1.4-1
- enable TUI mode
system-config-nfs-1.2.3-2
-------------------------
system-config-rootpassword-1.1.3-2
----------------------------------
* Thu Jan 29 2004 Jeremy Katz <katzj(a)redhat.com> - 1.1.3-1
- fix firstboot module symlink
* Fri Dec 05 2003 Brent Fox <bfox(a)redhat.com> 1.1.2-1
- apply patch from bug #111423
* Thu Nov 20 2003 Brent Fox <bfox(a)redhat.com> 1.1.1-1
- fix path problem in startup script
xemacs-21.4.15-5
----------------
* Fri Apr 09 2004 Jens Petersen <petersen(a)redhat.com> - 21.4.15-4.9
- add xemacs-21.4.15-pui-120437.patch to fix pui problems in 21.4.15
(Ville Skyttä)
- buildrequire gpm-devel and build xemacs-nox with gpm (Ville Skyttä,120437)
- separate patches into lisp and non-lisp patches
- move gnuclient and gnuserv from xemacs-common to xemacs since they
require X libs (Ville Skyttä,110649)
20 years