Lucid Puppy 5.2.8 - Updated ISO Version 005 - APR 05 2012

[bigpup]

Sound lost issue.

From a question I asked of Barry about lost sound.
His reply is in this topic on his blog.
May give some ideas to look at.
http://bkhome.org/blog/?viewDetailed=02616

Just an idea.
Based on this info, I wonder if the /etc/init.d/alsa should have been changed to 10alsa in Lucid 528. This seems to be the newer version and may not have all the supporting files that are in the newer version of Woof, because Lucid 528 is based on an older version of Woof..

[Aitch]

Tman made a multiple sound card wizard pet which worked for me in slacko 5.3....but the link to smokey's site is down/edit:fixed
http://smokey01.com/Tman/apps/Multiple_ ... Wizard.pet
info here
http://www.murga-linux.com/puppy/viewto ... 233#497233


Aitch

[rerwin]

I have discovered that the 2 latest fixes to snapmergepuppy (in wary/woof) are incomplete and are probably behind the problems being seen with USB flash drive installations! I will upload the corrected snapmergepuppy later today, but need some further testing of another mod to rc.shutdown.

EDIT: The patch is posted on the next page (51) of this thread.
http://www.murga-linux.com/puppy/viewto ... 693#587693
Richard

[playdayz]

OOPS: Reading the diff file is the problem. I haven't worked that out yet.

LOL. Yeah smokey01, that's the problem I had too

-----------------------------------------------------------------------------------------------

We are getting a lot of action on the usb/snd issue. So let's see what happens. One thing to remember is that 526 is OK but 527 or 528 is not. Yes bigpup, /etc/.init.d/alsa did change to 10alsa. That will be worth trying. Let's see what rerwin has--he probably knows the alsa and hardware config in Lucid as well as anyone. I am still suspicious of the change to initrd.gz from 526 to 527, so I may make an iso to test that, but I think I will wait for rerwin. It looks like several people think deleting /dev/snd on shutdown should cure the problem, but I am not sure if anything changed in that respect from 526 to 528. But of course any little change could have nasty repercussions.

[playdayz]

Tman made a multiple sound card wizard pet which worked for me in slacko 5.3....but the link to smokey's site is down/edit:fixed

Aitch, the Multiple sound Card Wizard is in Lucid 5.2.8. In Instant Update 002, Alsa Sound Wizard was also working correctly to change the default sound card.

--------------------------------------------------------------------------------------------------

shinobar and mavrothal, I will get to the config.guess and ffmpeg issues after this sound business. I do better with one thing at a time. shinobar, I have already figured out that when I compiled ffmpeg I did not enable external libmp3lame, libfaac, and libx264. Thanks.

--------------------------------------------------------------------------------------------------

Well, we might as well test this if we can. here is a lupu-528 *without* the change to initrd.gz. The change was made after 526 to allow booting on a computer with no drives attached, so this test will not do that, but we can see if it was this change to initrd.gz that caused the problem. Of course, we need a tester who can recreate the problem The only reason to test this is you have the "disappearing sound" issue.

http://diddywahdiddy.net/Puppy500/test01-lupu-528x.iso

---------------------------------------------------------------------------------------------

Notes on changes after 526

initrd.gz
alsa to 10alsa
Barry adds rm -f /dev/snd/* etc to rc.sysinit
pup_event_modprobe_backend_protect???

[Sylvander]

ASAP....

I'll try this new ISO...
Using a lupusave file on a Flash Drive.

[smokey01]

OOPS: Reading the diff file is the problem. I haven't worked that out yet.

LOL. Yeah smokey01, that's the problem I had too

I don't know what this means but you you might try this from a terminal window:

xdelta3 printdelta deltafile > delta.txt

We have even tried creating an empty ISO file and then applying the diff/delta file to it. That didn't work either.

It's a tough one.

[playdayz]

mavrothal, There is a new lupu_devx_528-4.sfs with /usr/share/libtool/config.guess. Thanks.

ftp://distro.ibiblio.org/puppylinux/pup ... _528-4.sfs
bf14b9eea9d67ad2f58dd11dc15832f7 lupu_devx_528-4.sfs

[shinobar]

I upload my proposal of /etc/init.d/10alsa and /usr/sbin/snapmergepuppy for your testing. I hope rechard will review them.
EDIT(2011-12-10): merged rerwin's patch-4 but skipped some:

10alsa:

Code: Select all

#111207 shinobar: wait loading modules, conditional saving asound.state (merged form the Quickset Wary, wary-511-01q)
#111210 shinobar: refresh asound.state at restart

You can simulate initial state by removing the file /etc/asound.state and reboot.

snapmergepuppy:

Code: Select all

#111205 rerwin: Remove saved whiteout file for directories that exist in snap layer; correct removal of saved file whiteouts; make any needed parent directory for opaque directory whiteouts.
#111204 shinobar: remove all /dev but symlinks, also remove whiteouts under /dev
#111210 shinobar: simplified removing whiteout

[Sylvander]

A. I downloaded test01-lupu-528x.iso and burned it to a CD-RW as per usual.

B. Booted it using the command:
puppy pfix=ram
So as not to use any of the existing lupusave files.

C. Once into the desktop of this Lupu:
1. Made various necessary config changes [language, region, keyboard].
2. Installed Pupsaveconfig-1.9.9 using the PPM, and used it to make a lupusave file [in a suitably named folder in the root of the ext3 only partition] on my 4GB Flash Drive, and rebooted.
The loudspeaker icon was still there after reboot.
3. Installed Xfe.
4. Installed Firefox-3.6.13-Lucid, and updated to 3.6.24.
5. Installed Wine-1.3.13-Lucid52.
6. Installed Frisbee-beta-2-Lucid52.
7. Installed Parcellite-0.9.3-lucid.

D. On the 4th reboot...
Prior to shutdown, I clicked the "Save" icon, to save during the session.
At shutdown, when the Pupsaveconfig->Save?=y/n was displayed I CHOSE TO SAVE...
Once back into the Lupu desktop, the loudspeaker icon was GONE.
I'm typing this whilst in the session after 4th reboot.

[shinobar]

Once back into the Lupu desktop, the loudspeaker icon was GONE.
I'm typing this whilst in the session after 4th reboot.

Click save icon to save the session, remove whole the /initrd/pup_ro1/dev and reboot without saving the session to see what happens.

[rerwin]

Here is a significant fix for USB flash-drive installations. It corrects the management of whiteout files, to avoid upsetting the layered file system (aufs/unionfs). It should be installed after Instant Update 002 (or 003).

The patch package is cumulative, containing patch-3 plus an upgrade to shutdown to support user-controlled saving of personal data. That feature is tested for the case where only the personal data is stored in a flash drive pup_save file. I am unable to try a flash-drive installation or the partition save on a flash drive, so welcome feedback from anyone needing that option for such installations.

Also, I include the more recent version of usb_modeswitch, 1.2.0, since it is already in wary. Keeping up.

After installing patch-4, clean out any residue from two /dev directories from the save layer. Run these commands:

• rm -r /initrd/pup_ro1/dev/snd
  rm -r /initrd/pup_ro1/dev/.udev

Then reboot. There will be no visible difference until the reboot is done. When you shut down after the reboot, you will be asked whether to save automatically or not. If "not", you will be asked whether to save the data now. Subsequently, no periodic automatic saves will be done. Click the "save" icon to cause saving, at any time. On new installations, to avoid any automatic saving, reboot again within a half hour, before the first automatic save, to see the "save" option request.

Next, I will check shinobar's latest offering, to see how it relates to what I have done.
Richard

[Sylvander]

@rerwin

A. OUCH!
I've only just deleted my latest lupusave for Lupu-528.003-ataflash...
But I have FIVE backup copies made at earlier points in time.
Number 5 [the latest] includes a state where the loudspeaker icon was still present.

B. Would this be suitable as a copy to try your fix upon?
No need to install instant update 002 or 003? [Because this ISO includes those, right?]

C. Should I be trying your fix before attempting that suggested by shinobar?

D. At this moment [typing this], I'm back using Lupu-526-ataflash, which is [of course] WORKING JUST FINE!

[playdayz]

Thanks for testing that test01 iso Sylvander. It seems to me that you could try shinobar's test easily (about deleting the /dev/snd). You could try it on any installation you have that is broken.

You could also try rerwin's patch-4 on an installation of lupu-528.003.

I will wait for any information from you plus for rerwin to examine shinobar's fix and see if he wants to incoporate it, then I will make a new iso from it and we can see about that.

Test01 does have Update 002 in it, but as I say, I think you could try shinobar's test on any one that is broken.

For general info, I think we are making some progress. At least we have determined several things that are not causing the problem. If the latest from rerwin and shinobar gets it then we will be in good shape. If not, ugh, we will try to revert various items to 526.

[bigpup]

Sylvander,

We all thank you, in willingness to help, as the lost sound test mouse. Your system seems to be the best test bed.

Just hang in there and be resolved to the final found fix.

To all that are working on the sound lost problem.
WE ALL THANK YOU!!
The easy answer is not always the best.
Do not be overcome by impatience, like our last test mouse.

[ETP]

Hunt script
The attached script when placed in the startup folder provides an early warning of possible corruption in the aufs filesystem and may prove useful for testing purposes. It could easily be modified to run a series of different tests. Please go easy on me as its my first bash at bash!

Code: Select all

#!/bin/bash
if grep -iq "whiteout" /tmp/bootkernel.log
then
		sleep 5
        		yaf-splash -close never -fontsize large -bg yellow -timeout 15 -text "WARNING! Aufs corruption. Reboot with pfix=fsck" &
else
		sleep 5
		yaf-splash -close never -fontsize large -bg blue -fg white -timeout 5 -text "All systems go!" &
fi

Line 127 of delayedrun

Currently by default the spash window exits on mouseover. Upon rebooting or restarting the X server the mouse pointer is over the window. Rather than seeing the message for 3 seconds as intended all you see (if anything) is a yellow flash. Including “-close never

[Karl Godt]

Reboot with pmedia=fsck

should be
Reboot with pfix=fsck

I don't use Puppy on USB much and seldom as frugal installation to HD .

If you want to work on aufs , you probably would need to read the
/usr/bin/auchk
/sbin/auplink
/usr/bin/aubrsync
shellscripts .

e2fsck/fsck does not check the AnoyherUnionFileSystem(aufs) but only ext2-4 .

The .wh. files are left as markers for deleted files from the read-only sfs to not show up again in the highest all-sfs merged layer '/' .

If there are nodes deleted in /dev/snd (included in the Lupu-5xy.sfs) they are still inside the sfs but won't show up on '/'/dev/snd .

To make them show up again you would need to put something like

Code: Select all

find /initrd/pup_rw -iname "*.wh.*" -exec rm {} \;

somewhere at the beginning of /etc/rc.d/rc.sysinit .

[ETP]

Karl
Thanks for pointing out my error which I have corrected in the post and attached file.
I take onboard your points and will examine the areas that you suggest.

Given my post on page 50 of this thread however, this script was only intended as a quick & dirty litmus test for possible corruption. Do you consider that as such it is a valid test? (i.e. is there ever any legitimate reason for the appearance of "whiteout" in the log?)

[ETP]

Karl
Thanks for pointing out my error which I have corrected in the post and attached file.
I take on board your points and will examine the areas that you suggest.

Given my post on page 50 of this thread however, this script was only intended as a quick & dirty litmus test for possible corruption. Do you consider that as such it is a valid test? (i.e. is there ever any legitimate reason for the appearance of "whiteout" in the log?)

[Karl Godt]

I can not remember many hassles for whiteout files , but one I will always remember :
I made an adjustment to rc.shutdown to save a copy of /etc/rc.d/PUPSTATE .

This is one of the most important configuration files , sourced by nearly every puppy.sh script .

This file gets created every time new @boot .

I made the copy using the mv command like
mv /etc/rc.d/PUPSTATE /etc/rc.d/PUPSTATE.prev
at the end of rc.shutdown .
mv deleted the /etc/rc.d/PUPSTATE file and created /etc/rc.d/PUPSTATE.prev but left a whiteout file .

This got me a corrupted /etc/rc.d/PUPSTATE file at boot and no fsck could help it . It took me a while to figure out , that i had to remove a/some whiteout files to make fsck work correctly .

Without knowing its PUPMODE Puppy will hang/crash/not work .

Until i recognized the whiteout file
i started to adjust the main boot/X/shutdown scripts with lots of unecessary code like
[ -f /etc/rc.d/PUPSTATE ] && . /etc/rc.d/PUPSTATE
if [ -z "$PUPMODE" ] ; then
dolotsofcode(){
#rdev,df,mount to guess the PUPMODE
}
fi