Wary 0.9.8 (098) feedback

Please post any bugs you have found
Message
Author
User avatar
BarryK
Puppy Master
Posts: 9392
Joined: Mon 09 May 2005, 09:23
Location: Perth, Western Australia
Contact:

Re: Loarding main sfs with pfix=ram from CD

#161 Post by BarryK »

shinobar wrote:I made a remastered custom CD of Wary-098.
Tested the custom CD of course with pfix=ram.
But it loads the main sfs from HDD original, instead of the CD modified.
PUPSTATE:

Code: Select all

PUPMODE=5
PDEV1='sr0'
DEV1FS='iso9660'
PUPSFS='sda3,ext3,/warry/wary_098.sfs'
The whole main sfs is loaded on RAM and the HDD partition /dev/sda3 is unmouted after, its nice, but...
It is a problem in case the PC with lower RAM.
The CD boot with pfix=ram means somewhat the HDD is unreliable, i think, so all the files should loaded from the CD.
Did you use the remaster program in Wary?

Yes, that problem with finding the wrong .sfs file would not occur if I had built Wary with the simplified filenames. Then, the file would be named 'puppy.sfs' and the init boot script would identify the correct one uniquely from its appended id-string.

Anyway, it is possible to do a remaster with new version number, and that would also fix the problem.
[url]https://bkhome.org/news/[/url]

User avatar
BarryK
Puppy Master
Posts: 9392
Joined: Mon 09 May 2005, 09:23
Location: Perth, Western Australia
Contact:

Re: Alsa again

#162 Post by BarryK »

zekebaby wrote:
BarryK wrote:Er, you didn't mention in what order you installed them.

If you want to test rerwin's latest upgrade, do not install my zzz PET.

If you want to test my zzz PET, it must be installed last. It replaces a couple of rerwin's scripts.
I installed zzz last, so the dumps I posted apply to zzz.

Now I'm going to do the same test without installing your zzz packages. I won't mix-and-match again.

Thanks for your support.

.Z.
Yeah, I won't post updates to my zzz PET to this thread, it causes confusion. I will post about it on my blog only.
[url]https://bkhome.org/news/[/url]

Sage
Posts: 5536
Joined: Tue 04 Oct 2005, 08:34
Location: GB

#163 Post by Sage »

Wary runs on old and new for me
Metoo, Billtoo!
Fair enough - Gateway used to be in Ireland. Emachines have been sold in Tesco and another of our cheapest mail order stockists - along with mainline Acer; shocking reputations. Nobody ever buys a second machine.
As for oldies, still got a couple of Amstrads in the loft. Shafted by those b*st*rds to your South because they were at least a couple of decades ahead with the technology. Alan Sugar came up with the bright idea of using the switched mode PSU from the monitor to power the mainboard, which was incorporated into the keyboard.
Shame Mr Churchill gave them all our IPR in exchange for food...

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

Diagnostic version of alsa initialization

#164 Post by rerwin »

To try to reconcile Barry's complaint about wary 098's failure to bark as expected, I set up a test PC and tried to recreate his situation. In the process, I created versions of the involved scripts that log trace information to show the order that events occur. My test specifically simulates the effect of the Agere and Intel537 driver variant switching. I ran with 098 plus the update-5 package, then added the "wired" versions of the "-5" scripts.

I attach a dotpet containing those wired scripts and an installation script that resets the variants so they are forced to switch on first use. The dotpet can be installed repeatedly, to reset those cases as needed.

My test results show that I cannot recreate any situation that disrupts the welcome barks! I encourage anyone, particularly Barry, who feels that the barks don't work when the Agere modem is installed, to try this package on top of the "-5" package.

For useful results, be sure to have pupdial CHOOSE > ERASE the modem selection, even if none is shown. This forces the full driver initialization process to occur at the next bootup, which is where the problem apparently exists. On subsequent tests, do both the CHOOSE-ERASE and reinstall the debug package, to reset the variant configuration. Please post or PM a pmodemdiag file if you find sound initialization not right.

Now, to try it on an old dual-CPU PC, to see how it keeps the two from stepping on each other.
Richard
Attachments
debug_alsa_init-1.pet
For documenting issues with ALSA sound initialization.
(21.26 KiB) Downloaded 354 times

User avatar
shinobar
Posts: 2672
Joined: Thu 28 May 2009, 09:26
Location: Japan
Contact:

Re: Loarding main sfs with pfix=ram from CD

#165 Post by shinobar »

shinobar wrote:I made a remastered custom CD of Wary-098.
Tested the custom CD of course with pfix=ram.
But it loads the main sfs from HDD original, instead of the CD modified.
PUPSTATE:

Code: Select all

PUPMODE=5
PDEV1='sr0'
DEV1FS='iso9660'
PUPSFS='sda3,ext3,/warry/wary_098.sfs'
The whole main sfs is loaded on RAM and the HDD partition /dev/sda3 is unmouted after, its nice, but...
It is a problem in case the PC with lower RAM.
The CD boot with pfix=ram means somewhat the HDD is unreliable, i think, so all the files should loaded from the CD.
BarryK wrote:Did you use the remaster program in Wary?
Yes. But remastering put aside. i will be back on remastering later.

What i'd like to say:
The live CD with pfix=ram option is often used for maintaining the HDD.
So the live CD with pfix=ram should not load anything from the HDD when we suspect the HDD, i think.
The Wary 098 now loads the main sfs from the HDD and keep mounting if the PC has small RAM?
Downloads for Puppy Linux [url]http://shino.pos.to/linux/downloads.html[/url]

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

modprobe_protect patch for Intel537 module timeout

#166 Post by rerwin »

Barry,
I have been concerned about the timeout of the Intel537 module showing up in the protect.log. Somehow, the modules.dep did not get/stay updated with the variant choice, something I need to investigate further. To prevent the timeout in such a case, I would like to insert a line into the script. Considering that the RC is approaching, I am describing the fix here, instead of making a package for it.

In /sbin/pup_event_backend_modprobe_protect, near the beginning of the "--modcheck=" case, please insert a line after line 110:

Code: Select all

 --modcheck=*)
  ENTITY="" #make visible here
  check_entity_claimed_func M || exit 1
  check_entity_claimed_func R && exit 1
  [ "`lsmod | grep "^$ENTITY "`" != "" ] && exit 0
  grep -q "^install $ENTITY /bin/true" /etc/modprobe.d/* && exit 1 #101008
  grep -q "/$ENTITY\.ko" /lib/modules/`uname -r`/modules.dep || exit 1 #101121
  sleep 1
  MAXRETRIES=90; RETRIES=1
It is the line annotated by "#101121" (line 7 above). That responds with "module not loaded".

The good news is that, while the welcome window mouseover popup had not appeared at bootup, with the fix, it returns! That might be what you were impacted by when you discovered the "woof woof" issue.

With this fix, the welcome should appear, but the connect wizard may show "no modem found", or the last device detected previously. I suspect that the workaround is CHOOSE-ERASE and then reboot. In my case, none of the variants had been selected, probably as a result my experimentation. Maybe there is some way the selection gets skipped; I will investigate. But this fix will lessen the impact. Thank you.
Richard

User avatar
bignono1
Posts: 360
Joined: Sun 17 May 2009, 07:30
Location: Q8

#167 Post by bignono1 »

Please check the download link,it is not working.

User avatar
BarryK
Puppy Master
Posts: 9392
Joined: Mon 09 May 2005, 09:23
Location: Perth, Western Australia
Contact:

Re: modprobe_protect patch for Intel537 module timeout

#168 Post by BarryK »

rerwin wrote:Barry,
I have been concerned about the timeout of the Intel537 module showing up in the protect.log. Somehow, the modules.dep did not get/stay updated with the variant choice, something I need to investigate further. To prevent the timeout in such a case, I would like to insert a line into the script. Considering that the RC is approaching, I am describing the fix here, instead of making a package for it.

In /sbin/pup_event_backend_modprobe_protect, near the beginning of the "--modcheck=" case, please insert a line after line 110:

Code: Select all

 --modcheck=*)
  ENTITY="" #make visible here
  check_entity_claimed_func M || exit 1
  check_entity_claimed_func R && exit 1
  [ "`lsmod | grep "^$ENTITY "`" != "" ] && exit 0
  grep -q "^install $ENTITY /bin/true" /etc/modprobe.d/* && exit 1 #101008
  grep -q "/$ENTITY\.ko" /lib/modules/`uname -r`/modules.dep || exit 1 #101121
  sleep 1
  MAXRETRIES=90; RETRIES=1
It is the line annotated by "#101121" (line 7 above). That responds with "module not loaded".

The good news is that, while the welcome window mouseover popup had not appeared at bootup, with the fix, it returns! That might be what you were impacted by when you discovered the "woof woof" issue.

With this fix, the welcome should appear, but the connect wizard may show "no modem found", or the last device detected previously. I suspect that the workaround is CHOOSE-ERASE and then reboot. In my case, none of the variants had been selected, probably as a result my experimentation. Maybe there is some way the selection gets skipped; I will investigate. But this fix will lessen the impact. Thank you.
Richard
Richard,
My to-do list has grown long again, and the next Wary will not be RC, and it is falling back a bit. The next release will be another beta, and may not come out for up to a week or so.
[url]https://bkhome.org/news/[/url]

Sage
Posts: 5536
Joined: Tue 04 Oct 2005, 08:34
Location: GB

#169 Post by Sage »

rerwin:
These issues are not restricted to Barry, nor to Winmodems/whatever. The failure of 'woof,woof', the non-appearance of the 'welcome' overlay, muted sound, etc., etc., are being reported by a wide range of our correspondents. Testing on a single machine will, in my experience, not suffice. The adverse effects are very new. My investigations with over a dozen boards, new and old, suggest that the effects are highly irreproducible, albeit heavily skewed towards the non-functional side. Notwithstanding, older systems seem less prone to some of the problems, although this might be subjective or small-sample slewed? Some combinations of the features not generally reported affect FULL installations in different, but irreproducible ways, eg settings not retained. I NEVER use Win/3G/USB modems - their interference with Linux systems is legendary; attempting, however successfully, to make these devices function should be regarded as an intellectual exercise alone. Rigour suggests that these are a red herring and should be disconnected in order to address the more fundamental issue that has now emerged.
All efforts to fix the present undesirable situation are welcome because Wary is proving to be a fantastic distro for old kit, for which it is/was intended. If it also works on new stuff, that should be regarded as a bonus - saving the planet must always take precedent!

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

#170 Post by rerwin »

Sage,
Thank you for laying it all out, there, about the sound issues. I was afraid I might be walking into a buzzsaw when I dared tinker with the ALSA initialization script, to pace it based on events instead of waiting regardless. I am responding to Barry's implication that the "protect" function, that attempts to ensure related operations occur in a predictable sequence where necessary, was at fault.

I over-promised when I suggested that I offer to resolve all sound problems. My attention to the init script is unrelated to modems, but to protect the reputation of the "protect" service as totally dependable.

The other issue I addressed was the intermittant detection of the sound card reported by zekebaby. My fix apparently resolved that. My goal there is simply that the script wait no longer than necessary to ensure it has what it needs to function.

BTW, I, too, have an issue with the cs4236 driver that my Aptiva seems to use. Not only does it not produce any sound, but on shutdown/reboot it spews trace lines implying an error, but that part scrolls away before I can see it. So, other that tightening up the start of the initialization script, I leave everything related to the ALSA wizard to Barry, as I normally do not need sound, nor understand much about linux support of it.
Richard

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

Small contribution to solving sound issues

#171 Post by rerwin »

I think I have sorted out the factors behind Barry's missing-welcome-woofwoof concern with his Agere modem. While it seemed to be triggered by the implementation of what I call "event gated pacing" of the ALSA initialization script, there was an underlying problem that appears to have been responsible. When I recreated CD booting, I also found that loading of a module timed out.

In my tests, I used my Intel537 modem, which is handled similarly to the Ageres. It was the Intel537 driver that failed to load. I found that the modules.dep entries for the agrmodem and intelmodem variants were not as designed. Consequently, the wrong variant was loaded for my 537EP card, which killed its loading. Modules.dep contained entries for all of the variants, instead of a selected one. Here's why:
intelmodem/537SP/Intel537.koHIDE:
intelmodem/537EP/Intel537.koHIDE:
intelmodem/537EA/Intel537.koHIDE:
intelmodem/537/Intel537.ko:
intelmodem/536EP/Intel536.ko:
. . .
agrmodem/11c11040/agrserial.ko:
agrmodem/11c11040/agrmodem.ko: initrd/kernel/drivers/usb/core/usbcore.ko kernel/sound/pci/hda/snd-hda-codec.ko kernel/sound/core/snd-pcm.ko kernel/sound/core/snd-page-alloc.ko kernel/sound/core/snd-timer.ko kernel/sound/core/snd.ko kernel/sound/soundcore.ko
agrmodem/06pci/agrserial.koHIDE:
agrmodem/06pci/agrmodem.koHIDE: kernel/sound/core/snd.ko kernel/sound/soundcore.ko
agrmodem/048pci/agrserial.koHIDE: agrmodem/11c11040/agrmodem.ko initrd/kernel/drivers/usb/core/usbcore.ko kernel/sound/pci/hda/snd-hda-codec.ko kernel/sound/core/snd-pcm.ko kernel/sound/core/snd-page-alloc.ko kernel/sound/core/snd-timer.ko kernel/sound/core/snd.ko kernel/sound/soundcore.ko
agrmodem/048pci/agrmodem.koHIDE:
The busybox depmod selects anything with the suffix beginning with .ko! This is the state of modules.dep on first boot from a CD.

I created a fix for this that uses, instead, the suffix ".HIDE", which does not confuse busybox depmod, and modified the agrsm and intel537ep firmware pinstall scripts accordingly. It is "6-delta" on my page 4 posting of modem updates.
http://www.murga-linux.com/puppy/viewto ... 639#467639

Package "-6-delta" is an addition to "-5" (since that is now in woof). It also contains a log message for the case where a module is checked (waited for) but is not in the modules.dep file, meaning it never actually started to load. The "protect" function so indicates to the caller, instead of waiting until timeout.

This is my best shot at eliminating some of the intermittent nature of sound card detection, by waiting for a sound driver to be loaded, before starting ALSA. I hope zekebaby can verify that it helps.
Richard

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

Intel537 and SmartLink PCI modems not working

#172 Post by rerwin »

Now that I have concluded my obsession with "module variants", I need to report issues with my above-named modems. The 537 starts to connect to the ISP, but claims "No dialtone". It works fine in 4.3.1+. The SmartLink does the repeating "NO CARRIER" thing, spewing connect-no-carrier sequences until manually disconnected.

The message log for the Intel contains this, FWIW:
Nov 23 10:29:05 puppypc user.info kernel: Intel 537EP card found
Nov 23 10:29:05 puppypc user.err kernel: 537ep:softcore_init_struct: driver serial already allocated device.
Nov 23 10:29:05 puppypc user.warn kernel: 537ep:softcore_init_struct: Unregistering serial driver.
Nov 23 10:29:05 puppypc user.info kernel: serial 0000:00:0a.0: PCI INT A disabled
I don't have solutions, but want to have this documented.
Richard

EDIT: On the other hand, my Intel 536 modem connects just fine.
Last edited by rerwin on Wed 24 Nov 2010, 03:52, edited 1 time in total.

tempestuous
Posts: 5464
Joined: Fri 10 Jun 2005, 05:12
Location: Australia

#173 Post by tempestuous »

rerwin wrote:The 537 starts to connect to the ISP, but claims "No dialtone".
I'm out of touch with Puppy's current dialup regime. Is the dialup connection done with pppd, or with wvdial?
I had that same problem several years ago (using a hardware modem, not soft modem).
The solution was to configure wvdial to not wait for dialtone.
The relevant configuration line in wvdial.conf is "Dial Command = ATDP"
Change it to "Dial Command = ATX3DT"

glenco
Posts: 37
Joined: Fri 14 May 2010, 13:19
Location: Milwaukee

#174 Post by glenco »

tempestuous
Pupdial already defaults to the ATX3DT command in WARY.

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

#175 Post by rerwin »

Tempestuous, glenco,
Thanks for the suggestion. I tried "unchecking" "Check dialtone" (which generates the "X3") just now, but instead of stopping at "NO DIALTONE", wvdial waits a minute after dialing, times out, then displays repeated NO CARRIER sequences until I disconnect.
Richard

User avatar
Sit Heel Speak
Posts: 2595
Joined: Fri 31 Mar 2006, 03:22
Location: downwind

posted: makedepend for the wary-098 devx

#176 Post by Sit Heel Speak »

Hi Barry,

makedepend is missing from the wary-098 devx, it is necessary for the compile of mesa-6.5.2.

Here is a .pet, compiled under wary-098.

Source:
http://xorg.freedesktop.org/releases/in ... .3.tar.bz2
29-Oct-2010 22:47 121K

Procedure:
CFLAGS= '-o2 -march=i486 -mtune=i486'
./configure --prefix=/usr
make
make check (1 test passed OK)
new2dir make install
(manually deleted the man file; 'strip --strip-unneeded *' has no effect on the binary)
dir2pet makedepend-1.0.3-i486

Philip
Attachments
makedepend-1.0.3-i486.pet.gz
makedepend-1.0.3-i486.pet for the devx of wary-098
(12.53 KiB) Downloaded 538 times

User avatar
shinobar
Posts: 2672
Joined: Thu 28 May 2009, 09:26
Location: Japan
Contact:

Unetbootin fails to boot up Wary

#177 Post by shinobar »

@Barry
Reported regarding Luci-238, but i confirmed Wary-098 also has the same problem.
It failes because of the 'pmedia=cd' option. Replacing 'pmedia=cd' to 'pmedia=usbflash' got success.

I remember the 'pmedia=cd' option searched all the media including usb devices. See the post:
http://www.murga-linux.com/puppy/viewto ... 545#470545

BTW, why you get back to underscore instead of minus for the name of main sfs files? 'wary_098.sfs' instead of 'wary-098.sfs'.
Downloads for Puppy Linux [url]http://shino.pos.to/linux/downloads.html[/url]

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

#178 Post by rerwin »

I hope I have made my final revision to the modem_update...-6 package: http://www.murga-linux.com/puppy/viewto ... 639#467639
1. It addresses Barry's issues with ttyUSB1 being detected instead of ttyUSB3, always picking the highest number as the data port.
2. It removes the E160E mode-switch and module-loading rules described as interfering rather than helping. Any mode switching is probably handled elsewhere.
3. It "fleshes out" the new capability to load (wireless) modem drivers for modems retaining their modem-state across reboots.
Richard

User avatar
Sit Heel Speak
Posts: 2595
Joined: Fri 31 Mar 2006, 03:22
Location: downwind

wary 098: compiled forcedeth doesn't work, from bkhome does

#179 Post by Sit Heel Speak »

An oddity, I'm curious what the explanation is.

I just compiled the wary-098 kernel, with the only change to .config being an attempt to enable configfs and cgroups via changes to 3 sections in make menuconfig as detailed in the three screencaps here.

But then I changed my mind and decided to boot using the "stock" kernel. However, all the drivers the compile produced under
/lib/modules are still there.

The forcedeth.ko (reverse-engineered nVidia integrated NIC driver) which the compile produced, will not load.

And yet the forcedeth.ko available inside

http://bkhome.org/sources/kernel-2.6.31 ... .14.tar.gz

does load.

Has anyone an explanation?

I am compiling in a frugal wary-098 install to HDD with a 10 G savefile (9.7 G free),
and
kernel_src-2.6.31.14-patched.sfs
and
wary_devx_098.sfs
are mounted by Boot Manager.

User avatar
Sit Heel Speak
Posts: 2595
Joined: Fri 31 Mar 2006, 03:22
Location: downwind

wary-098 --can't compile mesa, the X headers are borked?

#180 Post by Sit Heel Speak »

Hmm...probably something I miss in simple ignorance, but...

I can add makedepend to the devx.
I can fix /usr/include/limits.h.
I can link other headers where mesa-6.5.2 thinks they ought to be.
I can add the missing matypes.h from mesa-6.5.1.

...but when I'm faced with this many broken X headers in
xorg73_base_t2_DEV-7.3-w5.pet...

http://i54.tinypic.com/14m3r6o.jpg

...I am strongly encouraged to "attack in another direction," as the Marines say.

Maybe Lighthouse, maybe Joe's Slackware Puppy, maybe (*shudder*) Lucid.

Or maybe it's time to (*tosses off a two-finger shot of Laphroaig*) do my own CLFS Puppy?

Post Reply