Hide Forgot
Description of problem: For a fully updated "rawhide" system, launching "alsamixer" aborts with an error message: "alsamixer: function snd_ctl_open failed for default: No such device". Version-Release number of selected component (if applicable): alsa-utils-1.0.14-2.fc8 How reproducible: Always. Steps to Reproduce: 1. Launch "alsamixer" from the shell. Actual results: "alsamixer" aborts issuing "alsamixer: function snd_ctl_open failed for default: No such device". Expected results: Mixer panel is opened, and settings can be changed. Additional info: After logging in as root, the mixer panel is opened, and it is actually possible to modify the settings. Running "alsamixer" as an ordinary user was possible for both "Fedora Core 6" and "Fedora 7".
Aha. Can you please run system-config-soundcard, create /root/scsconfig.log and attach it here?
Created attachment 200581 [details] scsconfig.log for CS4236 on-board audio device of PR440FX (In reply to comment #1) > Aha. Can you please run system-config-soundcard, create /root/scsconfig.log and > attach it here?
I've had this problem on three test installs now. Trying to simplify troubleshooting, I did the following. Installed from the 7.92 liveCD. After booting up as normal user, the sound icon in the upper right was fine and I had sound. I then ran yum -y update. After rebooting, there was a small red X by the sound icon and sound no longer worked. However, running system-config- soundcard, which requires root permissions, worked. This may not be specific to 7.92 as, while trying to solve the issue, I ran across http://ditesh.gathani.org/blog/2007/09/07/much-ado-about-nothing/ (I receive the same error messages as mentioned in that link, but his solutions didn't work for me.) I can post any required logs, but it won't be till tomorrow evening. Thanks for your work on this.
Created attachment 219011 [details] scsconfig.log from my system-configure-soundcard I'm adding my own scsconfig.log in case it helps. Thank you again.
This seems to have been partially fixed by the latest kernel update (2.6.23- 0.224.rc9.git6.fc8). By partially, I mean that sound will now work for regular user if one changes the permissions for /dev/dsp and /dev/snd. Until today's update, doing a chmod on those two wouldn't solve the problem. (Before the problem arose, no chmod-ing was necessary.)
Ok, on a second machine, it seems as if by adding <sound>=/dev/dsp* /dev/snd/* to the top part of /etc/security/cosole.perms.d/50-default.perms and <console> 0666 <sound> 0600 root to the lower part, I can get sound working as normal user without the upgrade. I know that doing a chmod on those two on another machine didn't fix it before the upgrade. On this machine, now I don't remember if I even tried doing that. It's possible, that after trying that and having it do nothing on the other machine, I didn't bother.
Btw, the sound subsystem is not the only one which is affected by the current permission problems. I am also unable to access the floppy drive as an ordinary user using 'mtools'. I even have to be 'root' in order to be able to read from a diskette.
I understand that the audio should be available to the locally logged in user, but it isn't. I'm using icewm as window manager. $ id uid=500(proski) gid=500(proski) groups=10(wheel),14(uucp),500(proski),501(sbox) $ ls -al /dev/snd/ total 0 drwxr-xr-x 2 root root 220 Oct 19 11:57 . drwxr-xr-x 12 root root 4420 Oct 19 11:57 .. crw-rw---- 1 root root 116, 0 Oct 19 11:57 controlC0 crw-rw---- 1 root root 116, 24 Oct 19 11:57 pcmC0D0c crw-rw---- 1 root root 116, 16 Oct 19 11:57 pcmC0D0p crw-rw---- 1 root root 116, 25 Oct 19 11:57 pcmC0D1c crw-rw---- 1 root root 116, 26 Oct 19 11:57 pcmC0D2c crw-rw---- 1 root root 116, 27 Oct 19 11:57 pcmC0D3c crw-rw---- 1 root root 116, 20 Oct 19 11:57 pcmC0D4p crw-rw---- 1 root root 116, 1 Oct 19 11:57 seq crw-rw---- 1 root root 116, 33 Oct 19 11:57 timer I haven't changed anything in PAM or udev configuration. I think PAM should be responsible, not alsa-utils, as the later correctly refuses to work with insufficient permissions.
Same here on a F7 to F8 upgrade x86_64 machine. I see that no devices usually owned by the logged in user (e.g. floppy devices) are owned by the logged in user or have proper ACLs set up. In contrast, a machine with a fresh F8 installation has this: nils@gibraltar:~> ls -al /dev/snd/ total 0 drwxr-xr-x 2 root root 180 2007-11-02 11:46 . drwxr-xr-x 15 root root 4640 2007-11-08 15:48 .. crw-rw----+ 1 root root 116, 8 2007-11-02 11:46 controlC0 crw-rw----+ 1 root root 116, 7 2007-11-02 11:46 hwC0D0 crw-rw----+ 1 root root 116, 6 2007-11-02 11:46 pcmC0D0c crw-rw----+ 1 root root 116, 5 2007-11-02 11:46 pcmC0D0p crw-rw----+ 1 root root 116, 4 2007-11-02 11:46 pcmC0D1p crw-rw----+ 1 root root 116, 3 2007-11-02 11:46 seq crw-rw----+ 1 root root 116, 2 2007-11-02 11:46 timer E.g. the ACLS on the control file are: nils@gibraltar:~> getfacl /dev/snd/controlC0 getfacl: Removing leading '/' from absolute path names # file: dev/snd/controlC0 # owner: root # group: root user::rw- user:gdm:rw- user:nils:rw- group::rw- mask::rw- other::--- The non-working machine looks similar to comment #8, with no ACLs on the files. I've changed the product version to f8 FWIW and bumped severity (system without sound is no fun). I think that this is not a problem specific to the sound system as floppy devices are affected as well.
NB: This is the non-working machine (upgrade): root@wombat:~> find /dev -type c -o -type b | xargs ls -l | grep + root@wombat:~> ---> no ACLs added And this is the working machine (fresh install): [root@gibraltar ~]# find /dev -type c -o -type b | xargs ls -l | grep + crw-rw----+ 1 root root 14, 12 2007-11-02 11:46 /dev/adsp crw-rw----+ 1 root root 14, 4 2007-11-02 11:46 /dev/audio crw-rw-r--+ 1 root root 189, 809 2007-11-08 14:06 /dev/bus/usb/007/042 crw-rw----+ 1 root root 14, 3 2007-11-02 11:46 /dev/dsp crw-rw----+ 1 root root 14, 0 2007-11-02 11:46 /dev/mixer crw-rw----+ 1 root root 14, 1 2007-11-02 11:46 /dev/sequencer crw-rw----+ 1 root root 14, 8 2007-11-02 11:46 /dev/sequencer2 crw-rw----+ 1 root root 116, 8 2007-11-02 11:46 /dev/snd/controlC0 crw-rw----+ 1 root root 116, 7 2007-11-02 11:46 /dev/snd/hwC0D0 crw-rw----+ 1 root root 116, 6 2007-11-02 11:46 /dev/snd/pcmC0D0c crw-rw----+ 1 root root 116, 5 2007-11-02 11:46 /dev/snd/pcmC0D0p crw-rw----+ 1 root root 116, 4 2007-11-02 11:46 /dev/snd/pcmC0D1p crw-rw----+ 1 root root 116, 3 2007-11-02 11:46 /dev/snd/seq crw-rw----+ 1 root root 116, 2 2007-11-02 11:46 /dev/snd/timer brw-rw----+ 1 root disk 11, 0 2007-11-02 11:46 /dev/sr0 [root@gibraltar ~]# ---> ACLs for the logged in user added, not only to the sound devices but also to the fingerprint reader (/dev/bus/usb/007/042) and the optical drive (/dev/sr0)
Changing the component to hal as this seems more appropriate.
Versions here on both machines are: hal-0.5.10-1.fc8 hal-info-20071030-1.fc8
After some digging in /usr/share/hal/fdi/policy/10osvendor/20-acl-management.fdi and playing with /usr/libexec/hal-acl-tool I found that /var/lib/hal/acl-list contained binary garbage (why ever): root@wombat:~> cat /var/lib/hal/acl-list ��G�3"�G�3selinuxroot@wombat:~> After removing the file and restarting haldaemon, everything was back in working order again. Perhaps hald or hal-acl-tool should detect bogus contents in the file and in that case recreate it.
I've also found (through a post on Fedora forums) that in my particular case, I can resolve the non-root issue by enabling ConsoleKit (which is enabled by default, so the newcomer would probably not have run into this issue.) I have it disabled--however, if I enable it, I can remove the lines I mentioned from console.perms.d/50-default.perms and sound will work for all users. If I don't enable ConsoleKit, I have to have those lines for sound to work (for non- root users.)
Note that ConsoleKit is a core service that AFAIK is supposed to replace the /etc/security/console.perms* mechanism, i.e. things are not expected to work without it running.
Noted with thanks. However, this wasn't a problem till the updates (back when Fedora 8 was 7.92) Alas, poorly documented. I think that many people wind up finding the link below in trying to determine what services they can turn off. http://www.mjmwired.net/resources/mjm-services-f7.html There is no man page that I can find in Fedora for consolekit and the README in /usr/share/doc doesn't really indicate this. Again many thanks for the information. As the problem is frequently being mentioned on Fedora forums, I will pass this along.
Sorry to crash the party I filed this before I found this one! https://bugzilla.redhat.com/show_bug.cgi?id=370821 I have ConsoleKit running and I did: [root@freshy-desktop ~]# cat /var/lib/hal/acl-list /dev/sequencer /org/freedesktop/Hal/devices/computer_oss_sequencer u 42 /dev/sequencer2 /org/freedesktop/Hal/devices/computer_oss_sequencer_0 u 42 /dev/snd/seq /org/freedesktop/Hal/devices/computer_alsa_sequencer u 42 /dev/snd/timer /org/freedesktop/Hal/devices/computer_alsa_timer u 42 But still no go! I still cannot see any sound card device as "user" in my sound card Preferences gui I can how-ever see as "root" this message when I execute gnome-sound-properties "Unable to start the settings manager 'gnome-settings-daemon'. Without the GNOME settings manager running, some preferences may not take effect. This could indicate a problem with Bonobo, or a non-GNOME (e.g. KDE) settings manager may already be active and conflicting with the GNOME settings manager." and when I click OK I see the preferences gui with the correct sound card and all!
Here is some additional info: [engwnbie@freshy-desktop ~]$ aplay -l aplay: device_list:205: no soundcards found... [engwnbie@freshy-desktop ~]$ su Password: [root@freshy-desktop engwnbie]# aplay -l **** List of PLAYBACK Hardware Devices **** card 0: Live [Dell Sound Blaster Live!], device 0: emu10k1x [EMU10K1X Front] Subdevices: 1/1 Subdevice #0: subdevice #0 card 0: Live [Dell Sound Blaster Live!], device 1: emu10k1x [EMU10K1X Rear] Subdevices: 1/1 Subdevice #0: subdevice #0 card 0: Live [Dell Sound Blaster Live!], device 2: emu10k1x [EMU10K1X Center/LFE] Subdevices: 1/1 Subdevice #0: subdevice #0 [root@freshy-desktop engwnbie]#
(In reply to comment #14) > I've also found (through a post on Fedora forums) that in my particular case, I > can resolve the non-root issue by enabling ConsoleKit (which is enabled by > default, so the newcomer would probably not have run into this issue.) This is like the 10th bug like this.. In F9 and onwards ConsoleKit I'll make it impossible to disable ConsoleKit.
I was hit by this issue after a minimum F7 install [everything unchecked except for yum and dhclient] from DVD and updating and downloading the rest of the system over the network from the "rawhide" tree. At the end of this procedure, I found myself unable to control the audio devices or access the floppy disk drive as an ordinary user. I hadn't disabled any services deliberately. Why is this report that I had opened upon this experience closed as "NOTABUG" then?
Joachim, can you check the contents of /var/lib/hal/acl-list on your machine please?
After repeating the same procedure starting with an F8T3 minimum install, I didn't encounter this problem anymore. I will try the F7 one prior to a fresh F8 install one of these days.
I have installed current "rawhide" starting again from a minimum F7 install, albeit on a different system since the original one has been dismantled. The device permissions are correct now. I can access the floppy drive and use the ALSA mixer. The change of hardware should have no impact. This result seems to suggest that some intermediate package version from the "rawhide" tree messed things up.
I hadn't gotten around to enabling ConsoleKit at startup. After updating Fedora 8 last night (14 November, 2007) I find that editing console.perms.d/50- default-perms is no longer sufficient and that it is now definitely necessary to enable ConsoleKit. (The importance of ConsoleKit has been mentioned in this thread, I'm simply adding this information in case people find this thread in a search for solutions to their sound problems.) @davidz--rather than making it impossible to disable ConsoleKit, wouldn't it be far less effort to have a few lines of documentation somewhere, perhaps even a brief man page? It could basically direct them to the rather insufficient README, but at least add a note that this is one service which should not be disabled, whether or not the user is running Gnome. (The same probably goes for the avahi-daemon man page, which doesn't really indicate that it affects more than DNS.)
(In reply to comment #24) > @davidz--rather than making it impossible to disable ConsoleKit, wouldn't it be > far less effort No, I've spent 10+ hours this past month of dealing with users who think they know better, go and disable things they don't know about and then expect things to work. Sorry.
(In reply to comment #25) > No, I've spent 10+ hours this past month of dealing with users who think they > know better, go and disable things they don't know about and then expect things > to work. Sorry. That's always the case with new things that happen to be essential. You're aware that you can hide essential services from the uninitiated user in s-c-services with "# hide: true" in the chkconfig preamble of the init file?
(In reply to comment #26) > (In reply to comment #25) > > No, I've spent 10+ hours this past month of dealing with users who think they > > know better, go and disable things they don't know about and then expect things > > to work. Sorry. > > That's always the case with new things that happen to be essential. You're aware > that you can hide essential services from the uninitiated user in s-c-services > with "# hide: true" in the chkconfig preamble of the init file? I wasn't aware but the decision to move from initscripts to activation-on-demand also has to do with making the boot process faster and the system more robust by making the whole dependency problem that initscripts inflict on us go away. Besides, I bet the "optimizers" just use chkconfig(8) instead of s-c-services.
(In reply to comment #27) > I wasn't aware but the decision to move from initscripts to > activation-on-demand also has to do with making the boot process faster and > the system more robust by making the whole dependency problem that initscripts > inflict on us go away. Regardless of that, you'll still have to hide it from s-c-services eventually as admins should have the ability to enable and disable on-demand activated services just as SysVinit or xinetd services. > Besides, I bet the "optimizers" just use chkconfig(8) instead of s-c-services. There's that. But even if you move away from SysVinit, this bunch will find a way to somehow make it not run ;-).
(In reply to comment #28) > Regardless of that, you'll still have to hide it from s-c-services eventually as > admins should have the ability to enable and disable on-demand activated > services just as SysVinit or xinetd services. Disagree. Here's two reasons - It's just silly to turn this service off; it's not like it's using a lot of resources - By making it mandatory, code higher in the stack don't have to put in silly checks. Remember. Options leads to madness [1]. Like the current situation where the "optimizers" turn critical services off and expect their system to work. It's non-sense and increases the load on the support infrastructure for the distro. [1] : Cutting too many options like GNOME did in the early 2.x days is bad too. It's about finding a happy balance, a happy medium. Pretty much like everything else in life. > There's that. But even if you move away from SysVinit, this bunch will find a > way to somehow make it not run ;-). They can uninstall the package if they don't need it. It's fine, it's not needed for a minimal install.
@davidz and everyone else, I wasn't trying to be a troll here. David Z., thank you for taking the time to explain why you feel the best course would be to make it difficult to disable ConsoleKit. We users tend to forget how often you developers are blamed for things that are the fault of we users. Believe it or not, we do appreciate all the work you folks do for our benefit.
Created attachment 260581 [details] My /var/lib/hal/acl-list
Comment on attachment 260581 [details] My /var/lib/hal/acl-list I have the same problem. I did a clean install of Fedora 8 and my sound card doesn't work either, although when I run "Soundcard Detection" it does. I *haven't* disabled ConsoleKit, so that's not the source of the problem. Funny thing is that I also have a USB web-cam which has a microphone, and when I plug that in ALSA and the PulseAudio stuff *do* see it (as does "Soundcard Detection" BTW). In the /dev directory I see that the web-cam related nodes have an ACL setup so the logged in user can access the device. The sound card related nodes do not. My sound card BTW is a [Dell Sound Blaster Live!], same as Leo Canale in comments #17 and #18. I'm guessing hald is responsible for setting up the ACL's and maybe it's making a booboo somewhere? I'll try to see if I can get some more logging out of hald later (it's late).
(In reply to comment #32) It looks like I'm having exactly the same problem as Leo Canale in bug #370821. Sorry for the inconvenience.
(In reply to comment #29) > (In reply to comment #28) > > Regardless of that, you'll still have to hide it from s-c-services eventually as > > admins should have the ability to enable and disable on-demand activated > > services just as SysVinit or xinetd services. > > Disagree. Here's two reasons > > - It's just silly to turn this service off; it's not like it's using a lot > of resources > > - By making it mandatory, code higher in the stack don't have to put in > silly checks. I'm totally with you as far as ConsoleKit is concerned. Let me clarify what I meant: Even with on-demand-activated services, in many cases letting an admin expressly disable them is sensible. A tool for configuring this should be able to do the right thing with exceptions like ConsoleKit, i.e. hide them from the user. The notion of a service that is so important that it shouldn't be disabled isn't new if we talk about s-c-services, e.g. xfs was hidden so users couldn't mess with it.
(In reply to comment #33) > (In reply to comment #32) > > It looks like I'm having exactly the same problem as Leo Canale in gnome-volume-control. > Sorry for the inconvenience. Hans are you able to change your Display Settings? On my 2 machines that I have running F8 I have very little access to resources on both! If I log in as "root" on my Machine then everything is fine other than PulseAudio did not get configured with all the bit's that were needed. So my issues are that at start up the right amount of rights are set for the user. ie I cannot change the Display setting from what the system decided that they should be! As normal user sound devices, floppy drive and so on. If I change rights or change all the other bits that are out there I can get sound to work. But that is not the right solution. Both these 2 boxes have had Fedora Core 4 and on and they have never had any issues even when others were having display and sound issues, I never had them on these boxes. In short I want to help get these problems resolved properly so it will help everyone! I'm not interested in changing all kinds of settings and never get to the root of the problem "so to speak" My problems are simple I think. The normal user is not getting configured right from the first place. I wish I was able to help fix the problem, but I'm not that kind of engineer. I'm an electrical eng. Not software! David Z sorry not criticizing you! You are right more should read and understand what they read from all the "guides" that are out there! But we would not be here complaining if we did not have a problem. But you know how things can snow ball when things don't work. My first thing that I noticed was the little speaker thingy had a red x on it. So I too thought that my problem is due to sound. But thats not the root of the problem. There is some problem with access rights. Sorry to go on here! But how can we help get to the root of this problem? Leo.
For curiosity I downloaded the live CD and tested on my fist machine with Sound Blaster Card. I get the same problem as what I get from the installed DVD version no difference. Then I tried the live CD on another machine (ubuntu installed) with the Intel card, and everything is fine here is the output. [fedora@localhost ~]$ aplay -l **** List of PLAYBACK Hardware Devices **** card 0: I82801DBICH4 [Intel 82801DB-ICH4], device 0: Intel ICH [Intel 82801DB-ICH4] Subdevices: 1/1 Subdevice #0: subdevice #0 card 0: I82801DBICH4 [Intel 82801DB-ICH4], device 4: Intel ICH - IEC958 [Intel 82801DB-ICH4 - IEC958] Subdevices: 1/1 Subdevice #0: subdevice #0 [fedora@localhost ~]$ find /dev -type c -o -type b | xargs ls -l | grep + crw-rw----+ 1 root root 14, 12 2007-11-18 17:17 /dev/adsp crw-rw----+ 1 root root 14, 4 2007-11-18 17:17 /dev/audio crw-rw----+ 1 root root 14, 3 2007-11-18 17:17 /dev/dsp crw-rw----+ 1 root root 14, 0 2007-11-18 17:17 /dev/mixer crw-rw----+ 1 root root 14, 1 2007-11-18 17:17 /dev/sequencer crw-rw----+ 1 root root 14, 8 2007-11-18 17:17 /dev/sequencer2 crw-rw----+ 1 root root 116, 10 2007-11-18 17:17 /dev/snd/controlC0 crw-rw----+ 1 root root 116, 9 2007-11-18 17:17 /dev/snd/pcmC0D0c crw-rw----+ 1 root root 116, 8 2007-11-18 17:17 /dev/snd/pcmC0D0p crw-rw----+ 1 root root 116, 7 2007-11-18 17:17 /dev/snd/pcmC0D1c crw-rw----+ 1 root root 116, 6 2007-11-18 17:17 /dev/snd/pcmC0D2c crw-rw----+ 1 root root 116, 5 2007-11-18 17:17 /dev/snd/pcmC0D3c crw-rw----+ 1 root root 116, 4 2007-11-18 17:17 /dev/snd/pcmC0D4p crw-rw----+ 1 root root 116, 3 2007-11-18 17:17 /dev/snd/seq crw-rw----+ 1 root root 116, 2 2007-11-18 17:17 /dev/snd/timer brw-rw----+ 1 root disk 11, 0 2007-11-18 12:16 /dev/sr0 [fedora@localhost ~]$ Even the Live CD has the same problem, when boots and find what ever hardware it does not like it sets access rights wrong!
Now I'm finding that I have to enable ConsoleKit and, in addition, edit the 50- defaults-perm file for sound to work as non-root user. I will grant this might just be my machine, but right now, I don't have the time or diskspace to do a fresh install and see what happens.
(In reply to comment #37) > Now I'm finding that I have to enable ConsoleKit and, in addition, edit the 50- > defaults-perm file for sound to work as non-root user. I will grant this might > just be my machine, but right now, I don't have the time or diskspace to do a > fresh install and see what happens. > Read above it's not just you, and it has not only to do with ConsoleKit. I have never disabled anything on any of my installations. I always do a fresh install and never have had problems until F8! Just do a search on this bugzilla on F8 only and there are lot's of people having this problem, they just don't know it yet that it is the same thing! If I log on as root everything works!. I know I'm not supposed to do that. Just for testing. I love fedora because I used to administer a Sun network 20 + years ago and Fedora is the closest thing to that to me. I tried to install ubuntu on my server that has a fake raid 1 set up and I have to do all kind of messing around to getting it to work. So I'm in the middle of re-installing Fedora on it. It has no problem to recognize the raid, and I'm will to wait until they fix this issue. This is the only time I have had issues with Fedora since FC2!
FWIW, the fellow who maintains the mjmwired pages I mentioned in one of my early posts on this, has a page for Fedora 8 services now. He's updated it to mention some of the problems encountered when disabling ConsoleKit.
(In reply to comment #19) > > This is like the 10th bug like this.. In F9 and onwards ConsoleKit I'll make it > impossible to disable ConsoleKit. > What about the numerous reports where ConsoleKit is running but the audio permissions are still wrong? I just upgraded from F7 to F8, ConsoleKit is running but no audio. Apart from the latest reports here, Google shows there are many many similar issues. E.g. http://lindesk.com/2007/11/sound-issue-in-fedora-8/ (check comments) # chkconfig --list ConsoleKit ConsoleKit 0:off 1:off 2:on 3:on 4:on 5:on 6:off # /etc/init.d/ConsoleKit status console-kit-daemon (pid 1868) is running... # ls -l /dev/snd total 0 crw-rw----+ 1 root root 116, 10 Feb 14 23:01 controlC0 crw-rw----+ 1 root root 116, 13 Feb 14 23:01 controlC1 crw-rw----+ 1 root root 116, 9 Feb 14 23:01 pcmC0D0c crw-rw----+ 1 root root 116, 8 Feb 14 23:01 pcmC0D0p crw-rw----+ 1 root root 116, 7 Feb 14 23:01 pcmC0D1c crw-rw----+ 1 root root 116, 6 Feb 14 23:01 pcmC0D2c crw-rw----+ 1 root root 116, 5 Feb 14 23:01 pcmC0D3c crw-rw----+ 1 root root 116, 4 Feb 14 23:01 pcmC0D4p crw-rw----+ 1 root root 116, 12 Feb 14 23:01 pcmC1D0c crw-rw----+ 1 root root 116, 11 Feb 14 23:01 pcmC1D0p crw-rw----+ 1 root root 116, 3 Feb 14 23:01 seq crw-rw----+ 1 root root 116, 2 Feb 14 23:01 timer Modifying /etc/security/console.perms.d/50-default-perms fixes the issue for me. I think this bug should be reopened.
I agree, I have the same issue as parent: ConsoleKit running in levels 2345, permissions wrong in /dev/snd/* and /dev/dsp*. This IS still a bug.
Hi, I confirm that I have the same pb and that the authorizations are not updated by ConsoleKit. (tech infos below) Finally I put the following (ugly) hack in my /etc/rc.d/rc.local (called at startup) and it works (that is, all users have sound access): /etc/rc.d/rc.local: # Force access to the sound device # see https://bugzilla.redhat.com/show_bug.cgi?id=292201 chmod -R a+rw /dev/dsp chmod -R a+rw /dev/snd/* Hope it helps until we could have a cleaner solution... Mathieu ## TECHNICAL INFOS [root@thedoors ~]# aplay -l **** List of PLAYBACK Hardware Devices **** card 0: I82801DBICH4 [Intel 82801DB-ICH4], device 0: Intel ICH [Intel 82801DB-ICH4] Subdevices: 1/1 Subdevice #0: subdevice #0 card 0: I82801DBICH4 [Intel 82801DB-ICH4], device 4: Intel ICH - IEC958 [Intel 82801DB-ICH4 - IEC958] Subdevices: 1/1 Subdevice #0: subdevice #0 [root@thedoors ~]# uname -a Linux thedoors.vienna.argeo 2.6.23.1-42.fc8 #1 SMP Tue Oct 30 13:55:12 EDT 2007 i686 i686 i386 GNU/Linux