Upup Precise 5.3.3.3 with kernel 3.3.2

For talk and support relating specifically to Puppy derivatives
Message
Author
User avatar
BarryK
Puppy Master
Posts: 9392
Joined: Mon 09 May 2005, 09:23
Location: Perth, Western Australia
Contact:

#346 Post by BarryK »

I have addressed many of the issues posted about in this forum thread, applied them to Woof. See blog thread:

http://bkhome.org/blog/?viewDetailed=02839

And, Precise Puppy alpha3 is now out:

http://bkhome.org/blog/?viewDetailed=02842
[url]https://bkhome.org/news/[/url]

User avatar
Terryphi
Posts: 761
Joined: Wed 02 Jul 2008, 09:32
Location: West Wales, Britain.

#347 Post by Terryphi »

Thanks, Barry. Everything tested so far is running well for me. I took your advice and kept away for PPM for now.
[b]Classic Opera 12.16 browser SFS package[/b] for Precise, Slacko, Racy, Wary, Lucid, etc available[url=http://terryphillips.org.uk/operasfs.htm]here[/url] :)

mrokosz
Posts: 25
Joined: Sat 28 Jan 2012, 11:58

Compression method

#348 Post by mrokosz »

Barry,
Thanks for the distro, works great, but I am using on really slow computer P3 500MHz 256MB RAM frugal installation, so the software is not as responsive as with with the little bigger but faster for old computers distro version, I am talking about version around 169 MB size of the iso. If there is no problem I prefer to use sfs with smaller compression ratio, it just works faster

mrokosz
P.S. Everything works great, Pemasu package problems not there any more. I guess it will replace Pemasu distro

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

Re: Compression method

#349 Post by BarryK »

mrokosz wrote:P.S. Everything works great, Pemasu package problems not there any more. I guess it will replace Pemasu distro
Pemasu is a bit busy right now, see earlier post this Forum thread. But, he is a Puppy addict like the rest of us, and will be back to do another build from latest Woof -- so, whatever development we do with alpha3 will be going into Woof, and pemasu will pick up from there, the very latest. I mean, nothing will be lost here, it is all cumulative.
[url]https://bkhome.org/news/[/url]

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

Broadcom wifi driver management experiments

#350 Post by rerwin »

jim3630, peebee, pemasu, Broadcom 43xx users,
I have attached my refined experimental package to select and load the appropriate Broadcom wifi driver for each variant of the 43xx devices -- b43, b43legacy, brcmsmac, wl, wl (older version). It is also intended to support b44 ethernet devices even with the wl loaded; but I cannot, myself, verify that that would be successful.

The logic ensures that only the wl driver is loaded even if a device is also supported by b43 or brcmsmac. As before, to avoid using the (default) wl driver, simply blacklist it in the BootManager. Also as before, if the wl driver cannot handle a device (device not detected), it will be automatically blacklisted so that another driver gets loaded at the next boot-up. (You can un-blacklist it in the BootManager.)

If you experience a problem with the main experiment-4 package, you can instead use either the 4a or 4b packages to ensure that only wl or brcmsmac (respectively) are available for use. That is to eliminate any influence by the main-package logic or the other drivers.

Note that brcmsmac cannot be used on an adhoc network, so please avoid doing that. Please send me a pmodemdiag file if you encounter a problem with these packages.

Note also that jim3630 has more success using frisbee as the network interface for his brcmsmac-using device. But I hope to also have SNS and the network wizard work with the newer (wl, brcmsmac) drivers, but need the expertise of those more familiar with the workings of wifi connection.

Thanks for any feedback you can provide.
Richard

UPDATE 6/15/2012: The "5delta" package is added to address the cases where the proprietary driver fails for some (BCM4312 - hardware ID 14e4:4312) models that can also use the open-source driver. In experiment-4, a reboot was necessary to get the open-source driver to take over. The new package makes the substitution during the first boot-up after its installation. Although the logic appears to function as intended, it must be tried with the target device installed, to verify whether the substitution is truly effective and allows immediate connection. Please send me a pmodemdiag file if you can run the test. You should see driver b43 used, instead of no detection during first boot-up. TIA.

UPDATE 7/13/2012: The "5-6delta" package is added for minor mods and to move the b43 firmware files from the b43 "tarball"/directory into /lib/firmware/b43, to avoid a shutdown hang related to the absence of the firmware during the first boot-up. With that change, the module is successfully changed from wl to b43 if wl fails, all during the first boot-up. This was verified with dpup-exprimo.

UPDATE 7/14/2012: Re-uploaded "5-6delta" after download number 6, to remove the moving of b43 firmware during the installation and to add a test so that the b43 driver is not loaded if the firmware is not in /lib/firmware (ready for use). This prevents the b43-firmware-related shutdown hang if the firmware is not ready, making the implementation safe no matter where the b43 firmware is.
Richard
Attachments
precise_broadcom_experiment-5-6delta.pet
Additional refinement to experiment-4 and 5delta, to switch driver wl to b43 during first boot-up.
Applies to BCM4312 wifi device (which may be named BCM4311).
No firmware relocation, but switches drivers immediately only if b43 firmware is ready.
(2.16 KiB) Downloaded 553 times
precise_broadcom_experiment-5delta.pet
Addition to experiment-4, to switch driver wl to b43 during first boot-up.
Applies to BCM4312 wifi device (which may be named BCM4311).
(2.9 KiB) Downloaded 558 times
precise_broadcom_experiment-4a-wl_only.pet
(3.39 KiB) Downloaded 607 times
Last edited by rerwin on Sat 14 Jul 2012, 16:16, edited 5 times in total.

User avatar
peebee
Posts: 4370
Joined: Sun 21 Sep 2008, 12:31
Location: Worcestershire, UK
Contact:

Re: Broadcom wifi driver management experiments

#351 Post by peebee »

rerwin wrote:Thanks for any feedback you can provide.
Richard
Hi Richard

Experiment 4 works as expected on my HP550 laptop.

Pristine, frugal install - loaded Exp4 pet - rebooted.

wl driver loaded and in use - wpa connection established with Frisbee.

Survived reboot

Blacklisted wl - rebooted - b43 loaded and in use and Frisbee established a connection automatically

Blacklisted b43 - rebooted - no wifi loaded

Unblacklisted both b43 and wl - rebooted - wl back in use

Cheers
Peter

p.s. notice that BK has now compiled kernel 3.2.18 under Precise Alpha 3 with I have enabled the brcm network drivers
ImageLxPup = Puppy + LXDE
Main version used daily: LxPupSc; Assembler of UPups, ScPup & ScPup64, LxPup, LxPupSc & LxPupSc64

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

#352 Post by BarryK »

Precise Puppy 5.2.58 (beta1) is out:

http://bkhome.org/blog/?viewDetailed=02856

Note, it has the brcm and wl modules, but doesn't have any of the resolution logic that you guys have been working out with rerwin.
[url]https://bkhome.org/news/[/url]

User avatar
peebee
Posts: 4370
Joined: Sun 21 Sep 2008, 12:31
Location: Worcestershire, UK
Contact:

#353 Post by peebee »

BarryK wrote:Precise Puppy 5.2.58 (beta1) is out:

http://bkhome.org/blog/?viewDetailed=02856

Note, it has the brcm and wl modules, but doesn't have any of the resolution logic that you guys have been working out with rerwin.
Hi Barry

Don't think wl.ko is in the iso.....

I've compiled and uploaded a multi-kernel version:
which includes Precise 5.2.58 k3.2.18

This includes the latest rerwin logic.

Cheers
peebee
Last edited by peebee on Sun 03 Jun 2012, 22:59, edited 2 times in total.
ImageLxPup = Puppy + LXDE
Main version used daily: LxPupSc; Assembler of UPups, ScPup & ScPup64, LxPup, LxPupSc & LxPupSc64

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

#354 Post by BarryK »

peebee wrote:
BarryK wrote:Precise Puppy 5.2.58 (beta1) is out:

http://bkhome.org/blog/?viewDetailed=02856

Note, it has the brcm and wl modules, but doesn't have any of the resolution logic that you guys have been working out with rerwin.
Hi Barry

Don't think wl.ko is in the iso.....

I've compiled and uploaded a compatible version:
Precise 5.2.58 k3.2.18 Upload 0306120920 [wlan0]

This includes the latest rerwin logic.
Oh, yes of course. Thanks for the PET.

Why is the module in /opt? Why can't it be inside /lib/modules/3.2.18?
[url]https://bkhome.org/news/[/url]

User avatar
tronkel
Posts: 1116
Joined: Fri 30 Sep 2005, 11:27
Location: Vienna Austria
Contact:

ATI fglrx hack

#355 Post by tronkel »

Just been playing with the latest Precise Beta 5.2.38 released earlier today. Another li'l beaut. Like its Upup predecessors nice and snappy!

Just as with Ubuntu Precise Pangolin itself the ATI video card causes grief - nothing new here. With ATI graphics under Linux you have two choices. Either install the open source Radeon driver or the proprietary FGLRX driver. The current Radeon open source driver has been broken in recent kernel versions - so that just leaves the proprietary option.

You need to have both the devx and the kernel sources activated on your system. The bad news is that it does build - but with errors that result in an extremely slow frame rate performance.

The better news is that there is a tiny hack available that seems to work around the problem - see here:

http://www.murga-linux.com/puppy/viewtopic.php?t=77735

It has rescued me on more than one occasion. So thanks for the tip - well worth a try if you're experiencing this sort of problem with ATI.

This is ilewis's first forum post actually.
Life is too short to spend it in front of a computer

User avatar
peebee
Posts: 4370
Joined: Sun 21 Sep 2008, 12:31
Location: Worcestershire, UK
Contact:

#356 Post by peebee »

BarryK wrote:Why is the module in /opt? Why can't it be inside /lib/modules/3.2.18?
It enables me to build multi kernel versions for puppies like slacko which come with 2 kernel flavours - and I like to keep the same template for all puppies I am supporting. The pinstall hard links the wl.ko to the correct version for the kernel.

shinobar uses a similar but I think less elegant technique for his nvidia sfs's which also support multi kernels but you end up with lots of surplus /lib/modules/ kernel specific directories.

If there is a better way to support multi-kernels with one pet when the drivers are kernel dependent then I'm ready and eager to learn.

[Later edit] - I have now extended the technique to make a single mutli-kernel version of the driver - this uses a base-driver and xdelta3 delta files to create the kernel specific drivers at install time. See here[/edit]

Cheers
peebee
Last edited by peebee on Sun 03 Jun 2012, 23:06, edited 2 times in total.
ImageLxPup = Puppy + LXDE
Main version used daily: LxPupSc; Assembler of UPups, ScPup & ScPup64, LxPup, LxPupSc & LxPupSc64

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

#357 Post by rjbrewer »

A non-pae version with i810 xorg drivers would be
nice to try.

Inspiron 700m, Pent.M 1.6Ghz, 1Gb ram.
Msi Wind U100, N270 1.6>2.0Ghz, 1.5Gb ram.
Eeepc 8g 701, 900Mhz, 1Gb ram.
Full installs

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

Re: Broadcom wifi driver management experiments

#358 Post by rerwin »

I just now discovered a minor bug in the experiment-4 package that modifies the firmware.dep file incorrectly, affecting 2 entries for Broadcom wifi and ethernet. I have re-uploaded experiment-4 with the correction to the relevant edit command.

To the first 6 downloaders: If you need the corrected package, please re-download the replaced copy of the experment-4 pet package, from here.
http://murga-linux.com/puppy/viewtopic. ... 889#630889

The bug should not affect you unless your Broadcom device uses either of the b43legacy or b44 drivers. To correct the problem directly, please edit /etc/modules/firmware.dep.<kernel version>. The incorrect entries are

Code: Select all

b43legacy
b44:b44.ko:b43legacy.ko
But should be

Code: Select all

b43legacy:b43legacy.ko
b44:b44.ko
Sorry for letting that slip through.
Richard

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

Broadcom driver management code updated to experiment 6

#359 Post by rerwin »

I have added minor mods to the set of packages to automate the selection of the appropriate Broadcom wifi driver, "experiment-5-6delta" above:
http://murga-linux.com/puppy/viewtopic. ... 889#630889

Just be sure that you have installed experiment-4, then 5delta, then 5-6delta. I am using multiple deltas because pemasu has been including each level in various exprimo versions. You need install only those of these packages not already installed.

We discovered that the b43 firmware needs to be in place in /lib/firmware before the first boot-up of a puppy. Otherwise, the driver failure due to the absence of the firmware seems to result in a serious hang during shutdown. If you have a Broadcom 4312 (aka, 4311) wifi device that produces a failure message (in dmesg), try making a remaster of a fresh puppy plus packages -4, -5delta and -5-6delta, to see the benefit of its connecting on the first boot-up.
Richard

User avatar
lassie
Posts: 2
Joined: Wed 11 Jul 2012, 23:16

Re: Broadcom driver management code updated to experiment 6

#360 Post by lassie »

rerwin wrote:I have added minor mods to the set of packages to automate the selection of the appropriate Broadcom wifi driver, "experiment-5-6delta" above:
http://murga-linux.com/puppy/viewtopic. ... 889#630889
I will try this with my BCM4311, great that you try to automate the setup :-)

I have tried to get this chip working wireless for months now. Sometimes it is okay even for a couple of reboots, but most of the times the card isn't found at all when checking with 'lspci'.

What I have found out is what you wrote:
1). The card needs to be detected, probably with 'broadcom_wl-K3.1.10-slacko_4gA' and 'ssb' in the ADDLIST. But that only works sometimes (and I don't understand when the success happends)
2). When the card is detected, 'wl' has to be removed and instead 'b43' has to be activated (with 'modprobe')
3). Then I see the wireless networks and are able to connect.

Now I will try your package!

User avatar
lassie
Posts: 2
Joined: Wed 11 Jul 2012, 23:16

Re: Broadcom driver management code updated to experiment 6

#361 Post by lassie »

lassie wrote:
rerwin wrote:I have added minor mods to the set of packages to automate the selection of the appropriate Broadcom wifi driver, "experiment-5-6delta" above:
http://murga-linux.com/puppy/viewtopic. ... 889#630889
I will try this with my BCM4311, great that you try to automate the setup :-)

Now I will try your package!
I tried with the pet but no success...

Then I restarted from RAM (no save file) and suddenly the BCM4311 card was visible with lspci... Now I do a backup of the whole file system, maybe helpful next time when troubleshooting this hell of a chip... After next reboot I swear it won't be visible again!

What's the easiest way to take a snapshot of the computer as is now? (I mean when the wireless is up and the other functions also seems to work.)

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

Re: Broadcom driver management code updated to experiment 6

#362 Post by rerwin »

lassie wrote: I tried with the pet but no success...

Then I restarted from RAM (no save file) and suddenly the BCM4311 card was visible with lspci... Now I do a backup of the whole file system, maybe helpful next time when troubleshooting this hell of a chip... After next reboot I swear it won't be visible again!

What's the easiest way to take a snapshot of the computer as is now? (I mean when the wireless is up and the other functions also seems to work.)
The way to obtain the info needed to troubleshoot your wifi device and my experiments is by running the pmodemdiag script. It was originally for working with analog and wireless modems, but has been expanded to get some networking info in a addition.

If your wifi device is still not working, please boot with pfix=ram, install the experiment package, reboot and make a pupsave file, then (if wifi not detected or not working) run pmodemdiag and send me the tar.gz file it produces.

BTW, be sure to install experiment-4, then 5delta, then 5-6delta, to get the full experiment 6 configuration.

I am working on a simplified version of the experiment. So, any info you provide could be useful to me. Or might justify moving on to the new version.
Richard

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

Broadcom 43xx candidate for woof - please test

#363 Post by rerwin »

Attached is what I hope to be my final implementation for the various Broadcom 43xx wifi cards. It is much simpler than the previous experiments and replaces all of the experiment packages on the previous page (24). The main difference is that, instead of having all open-source drivers wait for the proprietary driver to run, now the proprietary driver waits for the open-source drivers to do whatever they do and then reloads them as necessary.

This "experiment-7" omits the logic to switch between versions of the proprietary wl driver (5.100 vs. 5.60), because peebee's multi-kernel driver packages support only one of them (whichever the user chooses to install). It simplifies the selection between proprietary and open-source drivers for the case where wl produces an error message, by utilizing an earlier message that identifies the problematic devices, thus avoiding the need to wait for wl to run to see the error message.

I have separated the two parts of the previous experiments to create a "universal" part intended for all puppies (including lucid pup) and a package containing the brcm drivers for precise puppy 5.3.3.3. The "universal" package can be installed after installing peebee's broadcom_sta_v5.100.82.112_patched_x86_32-multi.pet package (or the 5.60 equivalent) for testing. Once "experiment-7" is verified by those of you with the Broadcom wifi devices, I hope that peebee will adopt this version for his packages. My goal is to repackage this experiment as a submission for incorporation into woof, for all new puppies (unless it makes more sense to keep it with the proprietary-driver packages).

To have this package install correctly, be sure to add it to an existing pupsave environment. This is because it includes file removals, which are not preserved in the initial creation of a pupsave file. (Lupuplus ISOs contain a fix for this, but it has not yet been accepted for newer woof-based puppies.) For precise pup 5.2.60, a "small fixes" package containing the fix is here:
http://www.murga-linux.com/puppy/viewto ... 525#633525
Richard
Attachments
precise_broadcom_experiment-7.pet
Intended for any puppy.
Contains only new modprobe config files.
Ensures that residue from previous experiments is removed.
(5.26 KiB) Downloaded 524 times

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

#364 Post by rerwin »

I have now corrected the bug I found and have re-uploaded the experiment package. Testing with the previous upload should still be valid except that b43 would be loaded along with wl. Peebee told me some time ago that the presence of b43 does not impact his test results. But the advice is to avoid having them both run, to avoid possible interference, so that's my design. Likewise for wl and brcmsmac.

Sorry for inconveniencing the 3 of you who downloaded the original.
Richard

User avatar
peebee
Posts: 4370
Joined: Sun 21 Sep 2008, 12:31
Location: Worcestershire, UK
Contact:

Re: Broadcom 43xx candidate for woof - please test

#365 Post by peebee »

rerwin wrote:Attached is what I hope to be my final implementation for the various Broadcom 43xx wifi cards.
Hi Richard

Testing reports on my HP550 laptop....

Booted a pristine version of precise 5.3.3.3 - rebooted with savefile creation - loaded Experiment 7 V2 pet - rebooted - used Frisbee to connect WPA wifi - checked wl was in use - blacklisted wl - rebooted - wifi reconnected automatically - checked b43 was in use.

All as expected with my b43 version.

Then booted a pristine version of slacko 5.3.3 - installed my wl driver pet - rebooted with savefile creation - loaded Experiment 7 V2 pet - rebooted - used Frisbee to connect WPA wifi - checked wl was in use - blacklisted wl - rebooted - wifi reconnected automatically - checked b43 was in use.

Again - all as expected.

Cheers
peebee
ImageLxPup = Puppy + LXDE
Main version used daily: LxPupSc; Assembler of UPups, ScPup & ScPup64, LxPup, LxPupSc & LxPupSc64

Post Reply