Puppy 4.1.2 Bug Fixes

Please post any bugs you have found
Message
Author
User avatar
Botanic
Posts: 240
Joined: Wed 05 Sep 2007, 10:54
Location: Toulouse, France

#21 Post by Botanic »

i removed everything related to the second screen using mp and executed xwin.
Now it works...
But i have the feeling that i'm still in vesa mode

Code: Select all

	Driver      "vesa" #card0driver
:?:
Puppy 2.16 on HP Vectra VL400 - PIII 800 MHz - 320M RAM
Wary on Acer Aspire X3200 - Phenom X4 - 2.2GHz - 4G RAM

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

Restore detection of modems and module override function

#22 Post by rerwin »

Autodetection of Intel, mwave, and certain Agere modems disappeared from Puppy 4.1 after working (except agrserial) in 4.0. To compensate, I provided some workaround instructions for 4.1.2. Now I have restored the autodetection of all these PCI devices.
EDIT 2/15/09: For the ISA version of mwave modems, I corrected the forced-loading logic to bypass loading of the mwave module (as done, but differently, pre-4.1). The details are here:
http://www.murga-linux.com/puppy/viewto ... 523#270523
/EDIT

The fix uses new udev rules to trigger loading of the driver modules by augmenting the /sbin/pup_event_backend_modprobe script to accept as an optional argument a module name to be used if another module is not found using the normal method (so it could not override another module). In addition, /sbin/pup_event_backend_modprobe is corrected to restore the module override function that substitutes module dmfe for tulip for a particular NIC.
EDIT 2/15/09: Also, /etc/rc.d/rc.sysinit is corrected to skip modprobing mwave after extracting its firmware, so that the initialization script can perform the modprobe. /EDIT

The fix is attached as a dotpet, so users can apply it to 4.1.2. It contains only the following 3 files:
- /etc/udev/rules.d/60-udev-modem.rules
- /etc/rc.d/rc.sysinit
- /sbin/pup_event_backend_modprobe

I have tested with my Intel536-supported modem. However, the same implementation technique is used for all of the affected modems.
EDIT 2/15/09: Except for rc.sysinit, this dotpet is ready for 4.2; however, rc.sysinit has changed in 4.2, so that update should be applied manually, as described in the above-referenced posting about it. /EDIT
Richard

UPDATE 2/11: I have updated the above information and attached an updated version of the of this fix, to remove the no-longer-needed O2 Micro cardbus controller fix and restoring the "override" capability dropped from 4.1. The bugfix should resolve the dmfe-module issue described here:
http://www.murga-linux.com/puppy/viewto ... 076#264943

UPDATE 5/2: I have replaced the dotpet. It improves the udev rule and modem test, and makes the Intel536/7 drivers use device names created when they load (e.g., 535ep0). Would someone with an Intel 537 modem please tell me by PM whether it uses 5370 or some other name, since I do not have any such modem for testing? Do not use this dotpet on puppy 4.2, because the rc.sysinit file is wrong for it.

Update 6/21: I have upgraded the dotpet to support a second (and more generic) Intel537 modem. I have acquired a "537" modem for testing. This new version uses devices /dev/5360 and /dev/Intel5370, which appear when their driver is first loaded. If your connect wizard and pupdial still show either /dev/536 or /dev537, please use pupdial to "probe > ERASE" that setting and reboot to pick up the new device names.

UPDATE 8/22/09: Upgraded to "-5" for compatibility with the 3G-wireless (-12) support, to avoid regressing it if this package is installed with it.

Remember, this is only for 4.1.2. For 4.2.1, use the similar dotpet posted here: http://www.murga-linux.com/puppy/viewto ... 272#318272
Richard
Attachments
modem_and_override_fix-4.1.3-5.pet
Makes certain PCI modems usable and restores the module-override function.
Also restores support of ISA mwave modems.
(16.86 KiB) Downloaded 968 times
Last edited by rerwin on Sat 22 Aug 2009, 19:15, edited 14 times in total.

User avatar
edoc
Posts: 4729
Joined: Sun 07 Aug 2005, 20:16
Location: Southeast Georgia, USA
Contact:

#23 Post by edoc »

I am suffering serious problems with 4.12 on a laptop where prior versions have done quite well.

CUPS
gxine
serial
etc.

Is it possible that 4.12-Retro might resolve them or are these not the kind of things Retro usually fixes?
[b]Thanks! David[/b]
[i]Home page: [/i][url]http://nevils-station.com[/url]
[i]Don't google[/i] [b]Search![/b] [url]http://duckduckgo.com[/url]
TahrPup64 & Lighthouse64-b602 & JL64-603

User avatar
vtpup
Posts: 1420
Joined: Thu 16 Oct 2008, 01:42
Location: Republic of Vermont
Contact:

#24 Post by vtpup »

edoc, you don't say which laptop, but I use the following and ALL require 4.1.2 retro:

Thinkpad 600e Pentium II 400 mhz
Thinkpad 600e Pentium III (hack) 810 mhz
Thinkpad T30 Pentium IV 1.8 ghz

The last laptop, seemed to work well for a few weeks with 4.1.2 new kernel, but eventually I discovered it was unable to handle DVD writing, and sometimes reading as well. A change to 4.1.2 retro cleared this problem up.

I still have some problems with even 4.1.2 retro when shutting down the T30 laptop. I get kernel panics (after personal save file is checked) but before final blank screen powerdown. I haven't lost anything, but it is still a pain in the butt, and requires holding down the power button for 30 seconds to kill it. It is also an intermittent occurence.

Disabling ACPI seems to help, but hasn't totally eliminated it from happening on occasion. You might try diabling ACPI with the new kernel first before trying the retro version.

User avatar
edoc
Posts: 4729
Joined: Sun 07 Aug 2005, 20:16
Location: Southeast Georgia, USA
Contact:

#25 Post by edoc »

On the Z500 I was having the gxine & vlc video display problem - turns out it was a bug in xorg where the setup misreads the drivers and assigns an incorrect value.

Merely changing a 16 to a 24 in two locations of xorg.comf solved that problem.

Section "Screen"
Identifier "Screen0"
Device "Card0"
Monitor "Monitor0"
DefaultDepth 24
Subsection "Display"
Depth 24
Modes "1024x768"
EndSubsection
EndSection

I am hoping that the solution to CUPS is a similarly simple problem in a setup script.

I am also hoping that these bug fixes are useful in preventing these same problems in 4.2

Were you having CUPS-specific problems in 4.12-Normal and they resolved in 4.12-Retro?

I did observe a DVD writing problem with my Z500 under 4.11 but not 4.12
Last edited by edoc on Mon 26 Jan 2009, 16:50, edited 1 time in total.
[b]Thanks! David[/b]
[i]Home page: [/i][url]http://nevils-station.com[/url]
[i]Don't google[/i] [b]Search![/b] [url]http://duckduckgo.com[/url]
TahrPup64 & Lighthouse64-b602 & JL64-603

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

PetGet file updated to include 5 more mirrors

#26 Post by WB7ODYFred »

This updated petget is only related to finding resources on alternative mirrors Before repairing it would only download from ibiblior.org and ftp.nluug.nl The button "gatech.edu" was broken and would not download a pet package file using the "wget" ftp command.

Now, wherever in the world you are running puppylinux you can download from a realatively nearby puppylinux mirror.

The file /usr/sbin/petget is updated for downloading from the "www.gtlib.gatech.edu" puppylinux mirror site and also added
Austria, Australia, South Africa, Portugal, Thailand puppylinux mirrors.

Bosnian, Chec, and German puppylinux mirrors were also available that could be edited into the "petget" file. Someone living in that part of the world could customize to select their favorite puppylinux mirror.

This petget file has Pizzasgood's changes incorporated, to add quote marks around the variable "$ALIENFILE" mentioned above.

Report back any problems via private message to WB7ODYFred

I would like to add an Indian puppy linux mirror to file petget. You may edit in a new mirror website to match button value 18.
Edit lines 1622 - 1705 to add another mirror webisite in the return value 18 position.

Yes, the buttons are crowded, but PuppyLinux is headed for World Domination and needs the added mirror websites to succeed, Yeah!! :wink:

save original petget file by executing this command first before installing petget-4.1.2

cp /usr/sbin/petget /usr/sbin/petget_original

to undo changes

mv /usr/sbin/petget /usr/sbin/petget-4.1.2
cp /usr/sbin/petget_original /usr/sbin/petget

See the differences between the two files using "diff" command
diff -b /usr/sbin/petget /usr/sbin/petget_original

26-January-2009 Fred edited to make clear. Bugs fixed, mirrors added.
Attachments
petget-4.1.2.pet
file /usr/sbin/petget updated for gatech.edu mirror site and added
Austria, Australia, South Africa, Portugal, Thailand puppylinux mirrors.
Would like to add an Indian puppy linux mirror at button value 18.
Edit lines 1622 - 1854 to add another mirror
(20.59 KiB) Downloaded 1011 times
Last edited by WB7ODYFred on Tue 27 Jan 2009, 05:37, edited 2 times in total.

User avatar
edoc
Posts: 4729
Joined: Sun 07 Aug 2005, 20:16
Location: Southeast Georgia, USA
Contact:

#27 Post by edoc »

Fred - I am confused s this thread jumps all over the place and it is often difficult to be certain which posts apply to which sub-discussions within the thread. :roll:

Does your petget modification apply to the xorg or CUPS problems which may both occur during installation or is it only related to finding resources on alternative mirrors?
[b]Thanks! David[/b]
[i]Home page: [/i][url]http://nevils-station.com[/url]
[i]Don't google[/i] [b]Search![/b] [url]http://duckduckgo.com[/url]
TahrPup64 & Lighthouse64-b602 & JL64-603

User avatar
rjbrewer
Posts: 4405
Joined: Tue 22 Jan 2008, 21:41
Location: merriam, kansas

#28 Post by rjbrewer »

Edoc

This thread is about pet packaging bugs.

Your printer problem would probably be better addressed
in the hardware section of forum.

rjb

User avatar
edoc
Posts: 4729
Joined: Sun 07 Aug 2005, 20:16
Location: Southeast Georgia, USA
Contact:

#29 Post by edoc »

It was just that the video problem (gxine & vlc) was software problem created by an error in xorg during setup so I wondered if the CUPS problem - which did not exist in prior releases - may be the same.

The same printer and the same laptop suddenly will not communicate - the printer works fine elsewhere - the laptop will not address any printer.

As a result of all of that it seems to be that it is 99% probable that this is a bug in 4.12-Normal that is related to a software script error during setup rather than a hardware issue.

WDYT?
[b]Thanks! David[/b]
[i]Home page: [/i][url]http://nevils-station.com[/url]
[i]Don't google[/i] [b]Search![/b] [url]http://duckduckgo.com[/url]
TahrPup64 & Lighthouse64-b602 & JL64-603

User avatar
Patriot
Posts: 733
Joined: Thu 15 Jan 2009, 19:04

Xorg on QEMU

#30 Post by Patriot »

Hmmm .....

Recently I have the need to test internally made packages on fresh puppys and QEMU enables me to make those tests on different fresh puppy 4.x versions rapidly ...

I've read about how xorg does work on QEMU from here :
http://www.murga-linux.com/puppy/viewtopic.php?t=35469

After like the umph'teen times doing the same on fresh puppy boots, I figured it's good enough time to attempt a fix on this.

I've made a quick patch to xorgwizard to make a simple detection of QEMU. If puppy is running in QEMU then xorgwizard will copy a preconfigured-for-Qemu xorg.conf.new to /root ... this allows it to proceed on the next text screen for resolution selection. Testing of the selected resolution also will now work as intended.

A much more elegant solution would be to write a util to check for all VMs.

Code: Select all

xorgwizard : lines ~317
.
.
#autoprobe... 
# Patriot: skip any probing on QEMU; use preset instead
QEMUFlag=`dmesg | grep 'QEMU '`
if [ "$QEMUFlag" != "" ]; then
 cp /etc/X11/app-defaults/Xqemuconf /root/xorg.conf.new
else
 Xorg -configure > /tmp/xorgprobe.txt 2>&1
fi
#... config in /root/xorg.conf.new
.
.
I've only made preliminary testing using QEMU on windows. Attached is the patch I've made. Could the development team somehow evaluate if this is suitable enough to be included into future releases ?

Rgds
Attachments
xorgwizard-qemu.tar.gz
Contains patched xorgwizard & Xqemuconf
(18.9 KiB) Downloaded 1010 times

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

Mwave modem module loading bug affects old Thinkpads

#31 Post by rerwin »

EDITED 2/15/09 to replace previous posting:
Member vtpup has advocated for several months to repair the mwave modem firmware tgz file so that it contains a non-stripped version of the mwavem executable. However, in working with him I found another problem that is more basic. I discovered an omission in the forced loading of modules that prevents using the mwave driver on the ISA versions of that modem. Automatic detection can be expected only for PCI and USB devices; so drivers must be forced for ISA devices to make them work.

In the normal hardware detection and setup there is logic that skips actually loading the mwave driver module but causes the firmware/init-script to be expanded. That script then modprobes mwave. The fix is an easy edit to /etc/rc.d/rc.sysinit. Change line 298 from:

Code: Select all

  modprobe $MODULE $MPARAMS
to:

Code: Select all

  [ "$MODULE" != "mwave" ] && modprobe $MODULE $MPARAMS
Just copy the "[ "$MODULE" != "mwave" ] && " and paste it before the existing "modprobe $MODULE $MPARAMS". That will run the mwave setup in the correct sequence when forcong loading of the mwave driver (in BootManager). I have added this fix to my earlier posting for restoring modem functionality, here:
http://www.murga-linux.com/puppy/viewto ... 797#266797

Because my previous attachment of another "solution" is unproven and unsuitable for inclusion in Puppy, I have withdrawn it; I suspect it hints at an alternative solution, to be worked out after this rc.sysinit fix.

Thanks to vtpup for resuming interest in making mwave modems work in 4.1.. His recent posting about this is:
http://www.murga-linux.com/puppy/viewto ... 940#268940

Richard
Last edited by rerwin on Sun 15 Feb 2009, 23:00, edited 4 times in total.

User avatar
vtpup
Posts: 1420
Joined: Thu 16 Oct 2008, 01:42
Location: Republic of Vermont
Contact:

#32 Post by vtpup »

Hello rerwin,

I've just booted into 412 retro via pfix=ram on a TP600E to make sure everything is "fresh" for the test.

Steps:
1.) booted pfix=ram
2.) ran PupDial Modem ialup
3.) Did Probe (tried auto, tts0 and tts1)
4.) Got "Warning, no Modem Detected"
5.) Installed the new .pet from this post.
Tried again steps 1 through 4 with same result as in 5.

However, a few things should be checked before jumping to any conclusions:

a.) I didn't reboot because I was in pfix-ram. Maybe that would be needed
b.) This laptop just had its BIOS upgraded to the latest version this week and so it may not have its interrupts properly assigned to the internal modem and serial ports. That's half the battle on a Thinkpad,, as I remember it. Everything was reset to default in the BIOS refresh..

So let me work on this a little more, and we'll see if I can give you a more positive result.

Thanks for all your work on this, as well.

User avatar
vtpup
Posts: 1420
Joined: Thu 16 Oct 2008, 01:42
Location: Republic of Vermont
Contact:

#33 Post by vtpup »

deleted by request
Last edited by vtpup on Mon 02 Feb 2009, 01:13, edited 1 time in total.

User avatar
vtpup
Posts: 1420
Joined: Thu 16 Oct 2008, 01:42
Location: Republic of Vermont
Contact:

#34 Post by vtpup »

deleted by request
Last edited by vtpup on Mon 02 Feb 2009, 01:15, edited 1 time in total.

User avatar
vtpup
Posts: 1420
Joined: Thu 16 Oct 2008, 01:42
Location: Republic of Vermont
Contact:

#35 Post by vtpup »

deleted by request
Last edited by vtpup on Mon 02 Feb 2009, 01:17, edited 1 time in total.

User avatar
vtpup
Posts: 1420
Joined: Thu 16 Oct 2008, 01:42
Location: Republic of Vermont
Contact:

#36 Post by vtpup »

deleted by request
Last edited by vtpup on Mon 02 Feb 2009, 01:19, edited 1 time in total.

User avatar
vtpup
Posts: 1420
Joined: Thu 16 Oct 2008, 01:42
Location: Republic of Vermont
Contact:

#37 Post by vtpup »

deleted by request. Prior posts moved to Thinkpad TP600e Mwavemodem discussion at:

http://www.murga-linux.com/puppy/viewto ... 879#270879
Last edited by vtpup on Mon 02 Feb 2009, 01:24, edited 1 time in total.

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

#38 Post by rerwin »

vtpup,
Thanks for trying the mwave dotpet. I did not understand that it takes so much customizing to get it to work. But I can explain why you don't see anything happening.

With Puppy 4.1 the module loading technique changed so that simply using modprobe does not activate the firmware that does some of the magic. But there is a way to overcome that. Use the BootManager to force loading of mwaved with all its parameters. That will trigger the modem init script, which should handle the rest -- I think. (If it doesn't, we could make it so.)
Richard

EDIT: I read this last post first, then noticed your previous ones. So I may be addressing the wrong problem.

But first, I think we should go back to your other thread so that we don't hijack this one, which is for offering fixes. Could you move this dialog back there and reduce the size of your above posts? I think everyone would appreciate that.

Anyway, once you reboot after installing the dotpet, mwaved should get loaded automatically. We need to get that far first, maybe without your mods to the BIOS, considering that you have upgraded it. But I yield to your knowledge on that.

A-hah! The driver is not loading because of another change in module loading for 4.1. But I have made a dotpet to correct that, earlier in this thread, here:
http://www.murga-linux.com/puppy/viewto ... 797#266797
EDIT: A catch: This dotpet works only jfor the standard kernel. For the retro kernel, the fix is to insert a line into /etc/rc.d/MODULESCONFIG btween lines 10 and 11, so that lines 10-12 are:

Code: Select all

PCI_OVERRIDES='dmfe 0x00001282 0x00009102 #r_hughes reports tulip module not work.
mwave 0x00001014 0x0000007d #not in modinfo.
'
This is needed only in 4.1.2 retro, since 4.1 and 4.1.1 already contain the line (and others).

Richard
Last edited by rerwin on Mon 02 Feb 2009, 16:42, edited 1 time in total.

User avatar
vtpup
Posts: 1420
Joined: Thu 16 Oct 2008, 01:42
Location: Republic of Vermont
Contact:

#39 Post by vtpup »

But first, I think we should go back to your other thread so that we don't hijack this one, which is for offering fixes. Could you move this dialog back there and reduce the size of your above posts? I think everyone would appreciate that.
Apologies, didn't realize it was only for posting solutions since you requested test results here in this thread..

If the thread is reserved for solutions only, then perhaps it would be wise to only post tested software here and test it prior to inclusion. You could request testing results first elsewhere in a thread devoted to testing that package.

Apologies again...and thanks!

PaulBx1
Posts: 2312
Joined: Sat 17 Jun 2006, 03:11
Location: Wyoming, USA

#40 Post by PaulBx1 »

The net-setup.sh script contains this line:

Code: Select all

mv -f /etc/resolv.conf /etc/resolv.conf.$$
Unfortunately, over time that causes /etc to fill up with piles of resolv.conf files. A better way to do it would be this:

Code: Select all

if [ -f /etc/resolv.conf.-2 ]; then mv -f /etc/resolv.conf.-2 /etc/resolv.conf.-3; fi
if [ -f /etc/resolv.conf.-1 ]; then mv -f /etc/resolv.conf.-1 /etc/resolv.conf.-2; fi
if [ -f /etc/resolv.conf    ]; then mv -f /etc/resolv.conf    /etc/resolv.conf.-1; fi
Three backups should be enough.
I will post this in the network wizard thread too.

<later>
I just noticed rc.network has this same bit of code. In other words, when booting, rc.network may create another backup resolv.conf file. This seems to be an error. rc.network should execute the files created by the network wizard, not generate more of them. The comment there said the routine setupStaticIP was a hacked version of the same routine in net-setup.sh, however it appears it was not hacked enough!

Strangely, resolv.conf does not appear to be used anywhere. Not in /etc/rc.d anyway. I wonder why we go to all the trouble of making it and backing it up, if it never gets used. Maybe it's something in the boot process that uses it. Bind?

Post Reply