Lucid Puppy 5.28 X/ALSA/FLASH crash

Please post any bugs you have found
Message
Author
dio444
Posts: 33
Joined: Thu 20 Nov 2008, 11:57

#31 Post by dio444 »

Well, after something like 20 hours of use it just crashed. Same symptom as before, I was playing flash video and playing a flash game, and editing a libre-office file. It just instantly dropped to prompt as before, xwin failed to start it. After a little consideration, I issued a "halt" rather than rebooting so that the session was not saved, and then cold booted. During boot it did an fsck on both the main host FS and the savefile, but it came back up perfectly. No alsa failure. I still don't know what the actual problem is, but running this way (session save interval 0) is MUCH better than before when this was taking out my savefile every time it happened.

Tom

User avatar
rerwin
Posts: 2017
Joined: Wed 24 Aug 2005, 22:50
Location: Maine, USA

#32 Post by rerwin »

PaulBx1 wrote:Running alsa wizard kills the sound and the sound icon. Booting from a pupsave, no sound. OK booting pfix=ram.
. . .
BTW I am running 5.2.6 on another machine and have seen no alsa problem with it. So, something between 5.2.6 and 5.2.8...
PaulBx1,
Could you give us more details, so that I know where to look for a cause? Could you obtain a pmodemdiag file before you run the wizard and then afterward? Then a third one after you reboot and find no sound? Please tell us, too, what you did leading up to running the wizard, and why it was necessary.

To get the files, run pmodemdiag from a command line. Then rename the tar.gz file it produces in your root directory, to annotate it uniquely (-1, -2,& -3 are good enough). The tool always uses the same name for a particular day, and will not overwrite an existing such file. Oh, it is good for more than just modem diagnosis.

I would like to work on this, but need the evidence for clues. Thanks for reporting your problem.

UPDATE: After a couple of hours to think about this, I think you can ignore the above request for the diag files. I net out the issue to be simply that in 5.2.6 your ISA/legacy sound driver was always loaded automatically at bootup, after the initial setup with the wizard, but that that does not happen with 5.2.8. Right?

All of the scripts that deal with sound are the same in those two releases. The one thing that changed was the version of udev, which controls hardware detection and module selection. My guess is that udev no longer generates an event for the older sound cards -- maybe because of a difference in configuration parameters, or something else. Unless playdayz knows of a reason for not seeing these devices -- because he upgraded the udev version for lupu -- I will look for a way/workaround to restore the startup of such modules.

To confirm this approach, could you get a diag file after a good reboot in 5.2.6, and also after an unsuccessful reboot in 5.2.8 (with the correct driver already set up)? That way I can see if there is a difference in the event generation at bootup time. If we can confirm that the required event is no longer being created, I should be able to compensate for that, somehow. Since the alsawizard retains the driver name, the issue becomes that of when to decide that it needs to be loaded. Thanks for your help.
Richard

User avatar
WB7ODYFred
Posts: 169
Joined: Sun 14 Dec 2008, 02:15
Location: Oregon & Washington

Pmodemdiag-20110924-01 file posted here.

#33 Post by WB7ODYFred »

No, I have not got a flash crash. I am still working on getting sound operational on the I3 core Dell sutdio 1558 Model PP39L.
I will be reading other older forum posts to see what to do to
make sound operational. I ran the Setup --> Alsa wizard. But get some input/output error message and can not hear the 2 puppy barks.

Rerwin Yes, I think this might be a solution from my other reading.
"option snd-hda-intel model=dell-m6-dmic" in file /etc/modprobe.d/snd_hda_intel.conf

I don't have access to this computer to test, but will try at next opportunity. Maybe I need to use the Multicard alsa configuration script, too. Since you
point out that I have two sound card devices. Will try to disable the first one.

I attached my pmodemdiag file.
Also note on a fresh 5.2.8 USB install, the firmware file iwlwifi-6000-4.ucode for Intel Wireless N-6200 chip was missing in directory /lib/firmware
Went to Intel WIreless website http://intellinuxwireless.org/?n=downloads


Fred
Last edited by WB7ODYFred on Tue 27 Sep 2011, 07:52, edited 1 time in total.

User avatar
ilanrab
Posts: 100
Joined: Sun 01 Apr 2007, 14:01

Re: SFS!

#34 Post by ilanrab »

dio444 wrote:...
2) It's NOT that simply having a pupsave is the problem...
Indeed. I am going to provide some more info. Hopefully it will help in isolating the problem. This is not related to flash. It's related to running the ALSA Wizard, amixer, and retrovol (under certain conditions).

A few weeks ago I upgraded from 526 to TwoHeadedDog (528), via a frugal file copy, from the puppy ISO image to my Dell Netbook system. The system actually resides on a bootable 16GB flash drive. All went well. The sound icon appeared on the taskbar, I was able to use amixer and retrovol with no problems.
I've been running this system with no problems from the sound, the flash, or VLC.

Today, I replaced the 526 system, that runs on a Dell Desktop PC, with the solid TwoHeadedDog system that I've been using on my netbook (see the paragraph above.) Everything seems to be working fine, except for the sound.
The sound icon never appeared on the taskbar.
When I run amixer, I get: "amixer: Mixer attach default error: Input/output error"
When I run retrovol, I get: "retrovol: hcontrol.c:573: snd_hctl_load: Assertion `hctl->count == 0' failed. Aborted"

Since the Dell desktop has a different sound chip, I ran the ALSA wizard to remedy the situation. No change. The problem persists.
If I reboot the Desktop system, with a brand new (clean) pupsave file, the sound icon appears, and amixer/retrovol operate just fine.

So. I went back to the netbook (which has a working sound capability -- see the 1st paragraph, above) and I clicked on the following:
1. Desktop setup icon (PupControl comes up)
2. Hardware tab.
3. Sound Setup.
4. Single card.
5. Alsawizrd button in the Alsawizrd Warning dialog.
NOTE: at this point, the sound icon, on the taskbar, disappeared.
6. OK (searching sound cards progress dialog appeared).
7. OK .
8. NO.

The taskbar sound icon is gone. Running retrovol brings it back up, as long as the shell is up. Running the sound controls, on the netbook, does not give me the errors that I get on the Dell desktop system.
[b]ir[/b]

cthisbear
Posts: 4422
Joined: Sun 29 Jan 2006, 22:07
Location: Sydney Australia

#35 Post by cthisbear »

WB7ODYFred:

On a recent Toshiba...my tests.

""""""""""

Next sound >> as reported by windows it is

>> Conexant Smart Audio HD


At boot to the desktop >> Nothing happens...
Is Fatdog a barker?

So in Alsa Equaliser Mixer
I changed to >> default >> and it is now showing
>> Conexant Cx20590


Try my test mp3...as there is no bark.

= xine engine error
Input plugin failed to open mrl /mnt.

Alsa Equaliser control is off...>> I turn it on.

Alsa Mixer shows...But mp3 fails again.

Chipset says alsaequal

Hit F6 >> highlight says >> enter device name

I select >> Default

Card >> HDA Intel PCH

Chip now shows >> Conexant CX20590

Hit F5 >> give Master a few more hits upwards

Esc...Still errors. bugger bugger

But Xine plays off my usb...yeah.

Open up Pmusic to add the mp3 from SDA2.
It decides to play.

""""""""""

See if any of that works 4 you....Chris.

User avatar
rerwin
Posts: 2017
Joined: Wed 24 Aug 2005, 22:50
Location: Maine, USA

#36 Post by rerwin »

WB7ODYFred,
Although I have been silent since you made your report, I have been working on your issue, from the module-loading perspective. But I fear we may have run into a "brick wall." The design of the module loading function is based on the notion that a particular driver is associated with a single device. But at least some of the HDA sound drivers seem to involve the same driver, snd-hda-intel, being used by multiple, related devices. Your Dell does that.

In your case, there is the basic HDA device as well as an HDMI device. They both use the snd-hda-intel module but apparently require separate card numbers, which get assigned when a module is loaded. Once a module is loaded and has its sound card number, it is not loaded a second time, so does not assign another card number for the second device.

Considering that the module loading design is at the end of its life, having been replaced by the version now in woof and wary, it may not make sense to try to rework it outside of Barry's efforts. Although I could try a few simple-minded things that might luckily work, I think this calls for Barry's expertise and committment to wary/woof to resolve.

The first thing to establish is whether wary supports your HDA system, at all. If it does, I could dig into his code looking for a solution that would suffice in lucid pup, if not too complicated. I am not competent to design something for this from scratch, but might be a able to plagarize his logic. So, please try your system with the latest wary, to give us a starting point. Thanks.

Having said all of that, I have some little things to try separately with lucid pup 5.2.8, as experiments:
- Tell the driver the "model" of the device, in case it cannot already figure that out, by creating file /etc/modprobe.d/snd-hda-intel.conf with an "option" statement: echo "option snd-hda-intel model=dell-m6-dmic" > /etc/modprobe.d/snd-hda-intel.conf
- Try preventing the HDMI device from getting the driver, which would probably disable the HDMI output, but that might not matter to you. That would be done by editing a set of lines in /etc/rc.d/MODULESCONFIG, to read:

Code: Select all

PCI_OVERRIDES='
(none) 0x0000127a 0x00004321 #Rockwell riptide modem not supported by driver
(none) 0x00001002 0x0000aa38 #Ignore ATI Technologies Inc HDA HDMI Audio device RV710/730 
'
The second item would ensure that the main HDA card gets slot 0. If neither works alone, try them together.

In the long run, though, we will probably make more progress by testing in wary.

cthisbear, ilanrab,
I remain eager to resolve the issue of the alsawizard messing things up. Could each of you install the attached package that contains only an ALSA information collector, run it and attach its output to a posting? I think the simplest way to use it, right now, is from the command line:
  • alsa-info.sh --stdout > alsainfo.txt
Then gzip the alsainfo.txt file (or whatever you name it) for attachment. Thanks.
Richard
Attachments
alsa-info-1.pet
ALSA data collection tool
(7.88 KiB) Downloaded 367 times

User avatar
rerwin
Posts: 2017
Joined: Wed 24 Aug 2005, 22:50
Location: Maine, USA

#37 Post by rerwin »

WB7ODYFred, ilanrab, cthisbear, PaulBx1,
The alsawizard apparently does not actually restart the sound card after probing for the cards. Wary has a simple fix for that, which I have picked up and put into a dot-pet package, posted here:
http://www.murga-linux.com/puppy/viewto ... 146#568146

Please install that before running the wizard -- when you have sound already working. No need to reboot first. Thanks.
Richard

dio444
Posts: 33
Joined: Thu 20 Nov 2008, 11:57

#38 Post by dio444 »

Well, I had it crash again twice in the last hour... I've developed a plan for saving the work that is lost when not saving the session: I copy it into a real filesystem before issuing the halt. It's still really annoying to be having to workaround this issue. I really like LuPu in concept, but this issue is really a showstopper for any remotely serious use.

rerwin: I assume that your lack of mention of me in any of your recent posts means that you think my issue is different from the other people's issues? I attach the alsainfo.txt as you requested from some of the others just in case it might be any use. I did find that the forum refused to accept a txt attachment, so I fixed that. I also noticed that it seems that most of us having issues are running Dell hardware. Seems unlikely to be pure coincidence.

Anyway, based on rerwin's statements regarding the sound in lupu vs wary, I think I'm going to spend the night testing out the latest wary in hopes that it doesn't exhibit this show-stopping issue for me.

Tom
Attachments
alsainfo.txt.gz
(3.94 KiB) Downloaded 332 times

User avatar
rerwin
Posts: 2017
Joined: Wed 24 Aug 2005, 22:50
Location: Maine, USA

#39 Post by rerwin »

dio444 wrote:rerwin: I assume that your lack of mention of me in any of your recent posts means that you think my issue is different from the other people's issues?
Well, yes, because of the flash media factor. Thanks very much for the alsa-info file. Assuming it is telling the truth, I pick up on this:
  • Driver version: 1.0.21
    Library version: 1.0.24.1
    Utilities version: 1.0.24.2
I think it is time to try compiling the ALSA driver for ALSA 1.0.24, in case there is a mismatch somewhere.

But there is an experiment you could try to test the theory that there is a mismatch. Tempestuous has created a package of the 1.0.21 library and utilities that I posted here:
http://murga-linux.com/puppy/viewtopic. ... 562#566562

You might install it (and reboot) and see what happens. One user (peebee) tried it for another issue but saw no improvement. Note that whatever benefit 1.0.24 provides will be absent, so expect that.
Richard

User avatar
rerwin
Posts: 2017
Joined: Wed 24 Aug 2005, 22:50
Location: Maine, USA

#40 Post by rerwin »

WB7ODYFred, ilanrab, cthisbear,
I have posted a cumulative set of lupu 5.2.8 fixes that corrects an intermittent problem with verifying the loading of sound driver modules.
http://murga-linux.com/puppy/viewtopic. ... 604#568604

It could be a cause of the disappearance of sound function after a reboot. Please install it before running more sound tests. Thanks.
Richard

dio444
Posts: 33
Joined: Thu 20 Nov 2008, 11:57

#41 Post by dio444 »

Well I gave wary5141 a spin last night, and within 2 hours I dropped it again. It seems to have similar problems to lupu528, though when X crashed (3 times in that period) it didn't drop to console shell, it dropped to console login, and I failed to login. The account credentials must be right, root:woofwoof, because supplying something different responds as invalid and that credential just sits there and never makes it to shell.

Now, I'm back in Lupu528. This is a fresh save file. So far: installed the 528-001 instant update, iron12-lucid, flash (using the included downloader), the alsa-downgrade and alsa-info, reboot, create save, load devx. Writing now, about to reboot and go into normal use pattern. I'll post any results. (I do NOT plan to use the session save 0 workaround).

Tom

PS: It's possible the connection to flash is a red herring. I am just addicted to flash video/radio for YouTube&other site politics&religion discussions, so it's literally running constantly while I'm in puppy.

PPS: The instant update installed was the one on ibiblio I d/l'd the other day. I just got yours, and have it ready to go in my pet archive if you think I should run a test with it.
Last edited by dio444 on Thu 29 Sep 2011, 07:36, edited 1 time in total.

dio444
Posts: 33
Joined: Thu 20 Nov 2008, 11:57

#42 Post by dio444 »

Allrighty. The dreaded crash happened within about an hour. Downgraded alsa doesn't seem to have helped. Reboot was required to get X back, alsa broken exactly as before. Just in case you can see anything from the crashed version, I re-ran the alsa-info script, and attach the result.

I'm leaving this save file broken for future reference in case anyone can think of any info to glean from it. Back to the puppy that's working fine for me "MyWolfe028". If anyone wants info from it to compare I can post that too.

Tom
Attachments
alsa-info.txt.gz
(2.63 KiB) Downloaded 346 times

User avatar
rerwin
Posts: 2017
Joined: Wed 24 Aug 2005, 22:50
Location: Maine, USA

#43 Post by rerwin »

dio444,
I am relieved to hear that wary has the same problem. Crashing after a long period just sounds like something is running out of some kind of resource, usually memory. I think the most effective route, now, is to report the problem in the wary feedback (bugs) thread, so that Barry can get involved. I am sure he has better insight into the possibilities for resolution. He is the person to make the significant modifications to the long-term puppy.

About my latest patch posting: you could add it to your installed packages, but I doubt it would fix anything regarding the crashing.
Richard

dio444
Posts: 33
Joined: Thu 20 Nov 2008, 11:57

#44 Post by dio444 »

rerwin said:
I think the most effective route, now, is to report the problem in the wary feedback (bugs) thread, so that Barry can get involved.
Ok... I have a few hours to conduct some wary tests. I wasn't particularly interested in wary, I was much more interested in lupu. But if the issue is the same (and I'm not positive it is the same, just similar), It totally makes sense to try and get closer to the source of all things puppy with it... ;)

Tom

dio444
Posts: 33
Joined: Thu 20 Nov 2008, 11:57

#45 Post by dio444 »

Well, I've been using wary 5.1.4.1 for about 2 weeks now as my primary, and while it has crashed a few times in ways I can't explain, it is NOT the same. Wary has yet to eat its savefile like LuPu was reliably eating its savefile. I've had to reboot to get X back after a crash, but have yet to need to start my save file over to recover. This LuPu 5.28 crash that this post is about is definately unique to LuPu.

Tom

dio444
Posts: 33
Joined: Thu 20 Nov 2008, 11:57

Answer Found!

#46 Post by dio444 »

So, I finally figured out the real problem causing all of that stuff I experienced in this thread: the HDD on this machine was failing (according to the drive's SMART info, it had 7.6 years of spin time, and was failing about 50% of all operations on the first try, though it was not obvious there was an issue because it was almost always succeeding in the first few retries so did not return a fail state to the OS), and as it's got kinda minimal RAM (256mb), the swap space on the HDD was seeing heavy use. Heavy use of the failing HDD as virtual memory is, of course, likely to generate all sorts of funny problems as occasionally something in memory gets corrupted. Since replacing the HDD about a month ago, all such symptoms disappeared.

Anyway, sorry for resurrecting this zombie thread, but I wanted to post what the actual issue was and the solution.

Tom

cthisbear
Posts: 4422
Joined: Sun 29 Jan 2006, 22:07
Location: Sydney Australia

#47 Post by cthisbear »

dio444

Thanks for the problem solving mate.

Chris.

aarf

#48 Post by aarf »

Nah, i'd say not solved. Only read first and solution posts but i have posted this same x crashing in luci 004 thread . Mine is SSD and not failing. Problem only occurs in the luci and not in any of the other puppies i use.

Post Reply