(old)Puppy Linux Discussion Forum

peebee wrote:Both wl and b43 showed as loaded in lsmod

but netwiz and dmesg showed that wl was being used

Thanks for your thorough test. As for jim3630, this shows that b43 can be present if loaded after wl. All of the advice I have seen about both b43 and bcma/brcmsmac is to blacklist them. We demonstrate, now, that they can be loaded after wl binds to the device.

The significance of this is that multiple Broadcom wifi devices might be supported, even if they use different drivers. That is why a test of a PC with both wifi and ethernet Broadcom devices would be an interesting test. Or even with multiple Broadcom wifi devices.

If anyone has a Broadcom-based USB wifi device, could you please test with it? My new logic apparently works for boot-up time, but might not if a USB device is unplugged and re-plugged in. Or even if plugged in after boot-up. I plan to restore part of the logic to handle that (by unloading "dependers"), expecting that the dependers get loaded automatically without special logic (that is currently in uppup). I also expect to add the preference, "bcma:wl", so that only one of them (wl) will be loaded for dually supported devices.

In all cases, blacklisting of "wl" should cause the open-source drivers to be used.
Richard

rerwin wrote: To test the new driver, please blacklist wl in the BootManager, then reboot. I expect things should appear the same, but the diag file should confirm that the new driver is in control. Thanks.
Richard

Richard, et al.

as outlined done. continue not to tell any difference.. diag attached for your review.

noticed some output errors running the diag see below. don't know if significant or not.

# pmodemdiag
cat: /etc/puppyversion: No such file or directory
/usr/sbin/pmodemdiag: line 20: [: -ge: unary operator expected
Running PupScan to get hardware IDs and drivers - please close its window manually.
Collecting various modem-related files into an archive tarball:
pmodemdiag-20120502.tar.gz.
grep: /tmp/pup_event_backend/pup_event_alsa_cnt*: No such file or directory
grep: /tmp/rc_sysinit/*: No such file or directory
Diagnostic file collection complete.
Tarball pmodemdiag-20120502.tar.gz can be attached to a posting or personal
message (PM) in the Puppy forum, for analysis of dialup modem issues.
The content can be found in /tmp/pmodemdiag-20120502/.

hope helps and thanks again everyone for the hard work.

jim3630. In your second test...the bcma - brcmsmac driver combo was in use. Interesting thing is...that there isnt any indication of firmware loading. I wonder if this new bcma - brcmsmac combo work without previous firmware need.

Hmm...I should check the situation when I have time.

I could drop out those unneeded firmwares if the firmware loading is really out.

Jim3630. You could test it also, by removing /lib/modules/all-firmware/brcm/lib/firmware/brcm/bcm43xx-0.fw and /lib/modules/all-firmware/brcm/lib/firmware/brcm/bcm43xx_hdr-0.fw firmwares and then testing if the broadcom wireless works after rebooting....with wl blacklisted.

It would be interesting test...

pemasu wrote:jim3630. In your second test...the bcma - brcmsmac driver combo was in use. Interesting thing is...that there isnt any indication of firmware loading. I wonder if this new bcma - brcmsmac combo work without previous firmware need.

Hmm...I should check the situation when I have time.

I could drop out those unneeded firmwares if the firmware loading is really out.

Jim3630. You could test it also, by removing /lib/modules/all-firmware/brcm/lib/firmware/brcm/bcm43xx-0.fw and /lib/modules/all-firmware/brcm/lib/firmware/brcm/bcm43xx_hdr-0.fw firmwares and then testing if the broadcom wireless works after rebooting....with wl blacklisted.

It would be interesting test...

pemasu

in my early morning haste after reading Richards post booted into 5333 which the previous night did the test.

did not install the test pet again a second time before blacklisting wl and reboot.

do have two 5333 savefiles. should have deleted the first before starting this test. bothhave different names perhaps choose the wrong one this am.

will remove /lib/modules/all-firmware/brcm/lib/firmware/brcm/bcm43xx-0.fw and /lib/modules/all-firmware/brcm/lib/firmware/brcm/bcm43xx_hdr-0.fw firmwares and then testing if the broadcom wireless works after rebooting....with wl blacklisted.

after will delete extra savefile and redo this am test

pemasu wrote: It would be interesting test...

pemasu

done as outlined and it worked!

will go back and do this am test over on pfix=ram fresh with wl blacklisted.

Richard

here is a pfix=ram fresh 5333 with your test pet installed, rebooted with a savefile created then wl blacklisted and rebooted.

wifi detected on wan0! diag attached for your review. deleted the diag file in previous post so this file has same name but is not same file.

thanks

# modinfo brcmsmac
filename: /lib/modules/3.3.2-upup/kernel/drivers/net/wireless/brcm80211/brcmsmac.ko
license: Dual BSD/GPL
description: Broadcom 802.11n wireless LAN driver.
author: Broadcom Corporation
alias: bcma:m04BFid0812rev18cl*
alias: bcma:m04BFid0812rev17cl*
depends: mac80211,brcmutil,cfg80211,cordic,bcma,crc8
intree: Y
vermagic: 3.3.2-upup SMP mod_unload modversions 686

It does not show firmware dependency and jim3630 definitively used brcmsmac after removal of those previously needed firmwares.

Broadcom brcmsmac driver
Support for both 32 and 64 bit Linux kernels
Firmware installation
Copy brcm/bcm43xx-0.fw and brcm/bcm43xx_hdr-0.fw to /lib/firmware/brcm (or wherever firmware is normally installed on your system).

http://linuxwireless.org/en/users/Drive ... mac_driver

So...this is so new setup that linux wireless page is not up to date.
I will remove those firmwares from woof now....

For all USB chips listed: cp brcm/brcmfmac43236b.bin /lib/firmware/brcm/

That firmware has been preinstalled for brcmfmac driver. I think it was Volhout who previously tried to get it working with his usb broadcom wireless....or then I remember the poster wrongly....

EDIT: yep...I remembered right.
http://murga-linux.com/puppy/viewtopic. ... 470#621470

rerwin wrote:pemasu, jim3630 & Broadcom wifi users,
I have updated the Broadcom-wifi support to include the new brcm open-source driver and logic to select the correct driver for every supported Broadcom device.

Clean frugal install of Precise 5.3.3.3 on my Dell D430 laptop

Installed your experimental pet, the rebooted

1st boot w/savefile

- b43 and wl modules loaded according to lsmod
- Network Wizard sees wlan0 as b43
- dmesg shows first loading wl followed by b43

All subsequent boots show the exact same thing (both modules loaded). No blacklisting was performed.

Wifi works (woohoo!)

Thanks!

# seamonkey
GLib-GIO-Message: Using the 'memory' GSettings backend. Your settings will not be saved or shared with other applications.

zekebaby wrote:1st boot w/savefile

- b43 and wl modules loaded according to lsmod
- Network Wizard sees wlan0 as b43
- dmesg shows first loading wl followed by b43

All subsequent boots show the exact same thing (both modules loaded). No blacklisting was performed.

Thanks for your report. Very interesting. If both modules loaded, wl should have been the active driver. This tells me that perhaps the Network Wizard (and possibly SNS & Frisbee) may be impacted by the presence of multiple drivers supporting the same device.

I now have another experimental version that should avoid the duplication for a particular device, but allow in some cases the presence of multiple Broadcom wifi cards. It should handle, say, a card supported by the wl driver plus one using only b43.

My caveat is because of the way the b43 driver is accessed. It can be invoked the normal way via a modalias for the vendor & product IDS. But some that use b43 get to it via an modalias to ssb, which then, apparently, decides which b43(legacy) driver to use. With the ssb route, the preference for wl is ignored. The new package corrects for that in the .conf file for b43. That should prevent b43 from loading with wl for the same device. If two b43 devices are present but only one uses the ssb route, the other will not be supported. That is necessary because there is no way to determine which one does not use the ssb route (but loads ssb as a dependency).

The package adds some preferences, so that three are required: ssb:wl b43:wl bcma:wl. As before, the proprietary driver is preferred, but can be overridden by blacklisting wl in the BootManager. However, for the official release of an ISO, blacklisting it in the ISO might be better, so that the open source drivers are used by default. If a user finds them unacceptable, he/she can un-blacklist wl to activate the proprietary driver.

Please try everything you can think of with this package and provide a pmodemdiag file if you see anything not quite right. (The pmodemdiag script is updated to avoid the reported error messages.) Thanks.
Richard

Package removed for more debugging. Please discontinue testing it. I must be away for a few hours, though. R

Hello. Me again.

1) I prepared an iso of PPr-5.333 with the devx.sfs and LO-x.sfs on it.
Upon shutdown of first boot I requested "save to file" on a half-empty 8Gb usb stick.

a) It took forever.
b) It created a 512 Mb savefile without asking me for the size. This was only a test, 128 Mb would have been enough.
c) It copied the devx sfs and the LibreOffice sfs to the stick, but not its own operational *precise*.sfs. It claimed there was not enough space left, whereas there were over 3Gb left!

2) This is about mtpaint, but may affect other programs.
The default mtpaint in precise is v. 3.31. The localisation provided in esmourguit's French package is for v. 3.4*. The top menu bar was bilingual (one French title, one English title, one French Title, one English title, and so on) and the French entries were dummy entries until I installed mtpaint 3.4.3 (the latest pet, anyway). Then the menu entries appeared ok, in French only, and were fully operational.

I don't know if other localisations, or localized programs, do this. The bigger packages (abiword, gnumeric, LO, for ex.) have their own localisations more or less "embedded", and they work ok, but the smaller programs? We'd have to go through all of them one by one. (hehe)

rerwin wrote:Please try everything you can think of with this package and provide a pmodemdiag file if you see anything not quite right. (The pmodemdiag script is updated to avoid the reported error messages.) Thanks.
Richard

Hi Richard

Experiment 2 kills wl for me........

On 1st reboot after installing the pet b43 is in use

Blacklist b43 - reboot - no wifi - no sign of wl

I've attached a pmodemdiag - however this is from the "old" version - the new one just hangs and produces no output in the terminal or anywhere else.

Cheers
peebee

# beep-media-player-2
** Message: Domain: dbus-glib-error-quark, code: 25, message: Process /usr/lib/beep-media-player-2-bin exited with status 127






dbus-glib-error???

Calligra Office, formerly known as K-Office, is in the Ubuntu Precise repo. It should be possible to configure PPM to use the Ubuntu repos and then choose Calligra and Examine Dependencies--and get it working that way. I made quite a few of the original Lucid packages in this way. PPM will download the deps as deb files and then they can be installed one at a time or else undeb'ed and installed all at once.

All 3 precise pangolin repos have been added already. Calligra with all dependencies....234 Mb

Calligra with all dependencies....234 Mb

LOL. So much for saving space over Libre

pemasu wrote:All 3 precise pangolin repos have been added already. Calligra with all dependencies....234 Mb

234MB would not owe dependent on the?I want the new 3.334 what?How my hard drive space, a large capacity does not matter, only hope that the library the Qi install software easily, and when they release?

Fresh savefile installed on HDD -- Running main program from CD.

So far all appears to be OK, no major issues to report yet!

On a minor issue.... eth0 doesn't always stay connected on a reboot? I didn't have this with 5-3.3.2? and haven't been able to configure wlan0 for my rt2800usb yet? Had this problem back in original Precise on BK's thread and pemasu made that 'patch' and all worked then. I didn't have that problem with 5-3.3.2 and it configured and worked nice however, 5-3.3.2 was an 'update' to BK's original build and the 'patch' carried over to 5-3.3.2 where 5-3.3.3 is a fresh install without the patch. I will do some more testing on this and if I just can't get it going, I'll try installing that 'patch' again and see what happens then.

Other than that... looking and working good!