Precise Puppy RC2, October 20, 2012

[BarryK]

Hmmm... BK must have some 'thingy' about Leaf-Icons since even in beta-3 There Isn't Any Leaf (menu) and I have seen this mentioned since way back when.... even before Precise.

On the serious side... Where might one install .sfs files to be tested in Precise since we don't seem to have a mnt/home file? Do we create it ourselves? Currently running all tests on a live CD.

>>>---Indian------>

That's a new one. There should always be a /mnt/home symlink. I am running a frugal install of Precise right now, and I have /mnt/home.

[BarryK]

When I first run(pfix=ram) seamokey2.8 in console , the output as below:

Code: Select all

# mozstart 

(process:26029): Gtk-WARNING **: Locale not supported by C library.
	Using the fallback 'C' locale.
LoadPlugin: failed to initialize shared library /usr/lib/mozilla/plugins/gecko-mediaplayer-dvx.so [libcurl-gnutls.so.4: cannot open shared object file: No such file or directory]
LoadPlugin: failed to initialize shared library /usr/lib/mozilla/plugins/gecko-mediaplayer-qt.so [libcurl-gnutls.so.4: cannot open shared object file: No such file or directory]
LoadPlugin: failed to initialize shared library /usr/lib/mozilla/plugins/gecko-mediaplayer-rm.so [libcurl-gnutls.so.4: cannot open shared object file: No such file or directory]
LoadPlugin: failed to initialize shared library /usr/lib/mozilla/plugins/gecko-mediaplayer-wmp.so [libcurl-gnutls.so.4: cannot open shared object file: No such file or directory]
LoadPlugin: failed to initialize shared library /usr/lib/mozilla/plugins/gecko-mediaplayer.so [libcurl-gnutls.so.4: cannot open shared object file: No such file or directory]

(process:28412): Gtk-WARNING **: Locale not supported by C library.
	Using the fallback 'C' locale.
NOTE: child process received `Goodbye', closing down
# 

Seem Precise Puppy version 5.2.55 missing the libcurl-gnutls.so.
Thanks,
C.H.Lee

Precise Puppy alpha3 (5.2.55) does have libcurl-gnutls.so.4

[BarryK]

Hello Barry

This Puppy does not like ATI Radeon graphics.

If I set Radeon in the xorgwizard then X does not start.

If I copy ati_drv, radeon_drv, fbdev_drv and fbdevhw_drv to the modules drivers folder and delete vesa_drv, and delete Xorg.conf so that X auto configures, then X does start and load the correct drivers but there are errors in the Xorg.log and acceleration is disabled.

The errors are:

[ 12.893] (II) Loading /usr/lib/xorg/modules/drivers/radeon_drv.so
[ 12.893] (II) [KMS] No DRICreatePCIBusID symbol, no kernel modesetting.
[ 12.893] (WW) Falling back to old probe method for fbdev
[ 12.893] (II) Loading sub module "fbdevhw"
[ 12.893] (II) LoadModule: "fbdevhw"
[ 12.893] (II) Loading /usr/lib/xorg/modules/drivers/libfbdevhw.so
[ 12.900] (II) Module fbdevhw: vendor="X.Org Foundation"
[ 12.900] compiled for 1.11.4, module version = 0.0.2
[ 12.900] ABI class: X.Org Video Driver, version 11.0
[ 12.901] (EE) open /dev/fb0: No such device
[ 12.901] (WW) VGA arbiter: cannot open kernel arbiter, no multi-card support

and

[ 13.177] (WW) RADEON(0): Direct rendering disabled
[ 13.177] (EE) RADEON(0): Acceleration initialization failed
[ 13.177] (II) RADEON(0): Acceleration disabled

For comparison fatdog 600a2, which also has Xorg 1.11.4 and does work, gives this in the log:

[ 9.656] (II) Loading /usr/X11R7/lib64/xorg/modules/drivers/radeon_drv.so
[ 9.656] (II) [KMS] Kernel modesetting enabled.
[ 9.656] (WW) Falling back to old probe method for vesa
[ 9.656] (WW) Falling back to old probe method for fbdev
[ 9.656] (II) Loading sub module "fbdevhw"
[ 9.656] (II) LoadModule: "fbdevhw"
[ 9.661] (II) Loading /usr/X11R7/lib64/xorg/modules/libfbdevhw.so
[ 9.666] (II) Module fbdevhw: vendor="X.Org Foundation"
[ 9.666] compiled for 1.11.4, module version = 0.0.2
[ 9.666] ABI class: X.Org Video Driver, version 11.0
[ 9.666] (II) RADEON(0): Creating default Display subsection in Screen section

and

[ 9.850] (II) RADEON(0): Direct rendering enabled
[ 9.854] (II) RADEON(0): Setting EXA maxPitchBytes
[ 9.854] (II) EXA(0): Driver allocated offscreen pixmaps
[ 9.854] (II) EXA(0): Driver registered support for the following operations:
[ 9.854] (II) Solid
[ 9.854] (II) Copy
[ 9.854] (II) Composite (RENDER acceleration)
[ 9.854] (II) UploadToScreen
[ 9.854] (II) DownloadFromScreen
[ 9.854] (II) RADEON(0): Acceleration enabled

It looks like the kernel is lacking KMS

I did previously try setting Radeon in xorgwizard on Pemasu's version based on a2, and X did not start either. That had a different kernel. So I may be barking up the wrong tree.

If you want KMS, edit /etc/modprobe.d/radeon.conf, change to:

options radeon modeset=1

Then reboot.

[BarryK]

peebee,
Would you mind doing that again soon? I need to recompile the kernel in a running Precise Puppy, to avoid the gcc mismatch problem (current kernel compiled with Wary 5.3).

...I guess that I had better do that soon, as I know many people will want their nVidia, wl, etc., drivers.

I just remembered something. I don't have the 'wl' firmware in Woof. is it available in pemasu's Upup thread? I think that I recall, it is, plus mods to b43 firmware tarball?
...but, I also seem to recall, there are still some issues with that?

I just checked, the latest is 3.2.18, I might as well use that.

Hi Barry

No problem - once you issue the new precise then I'll recompile the wl driver.

There is actually no "firmware" as such with the wl driver - all the proprietary stuff comes as a binary image with a bit of source code wrapper that has to be recompiled for each kernel version seemingly.

That being said, rerwin has produced some broadcom "firmware" but actually as I understand it what he is doing is use the firmware mechanisms to determine which is the best Broadcom solution for each chip type.

So there are 3 types of driver :

b43 - this does have real firmware and there are later versions available than have been shipping in recent puppies - pemasu has provided later firmware versions in his pups

brcm - this comes as part of recent kernels and is for instance present in slacko - it does have some real firmware I think

wl - this is Broadcom proprietary and has no real firmware separate from the driver - it is all built-in - I have been providing versions for recent woof-based pups

Then there is rerwin's "firmware" which tries to manage it all and is still somewhat experimental.

This is only my understanding - I may well be wrong - but Richard will no doubt correct if I am....

Cheers
peebee

The b43 module in the 3.2.17 kernel used in Precise alph3, needs this firmware:

firmware: b43/ucode9.fw
firmware: b43/ucode5.fw
firmware: b43/ucode16_mimo.fw
firmware: b43/ucode15.fw
firmware: b43/ucode14.fw
firmware: b43/ucode13.fw
firmware: b43/ucode11.fw

All of which is present in the b43 firmware tarball.

[01micko]

First run of precise a3 seems pretty good.

First boot was troublesome though. Seems 'vesa' was loading for me (nvidia GS8400) and the resolution caused an "out of range" error for the monitor. Had to CTRL-ALT-BKSPC and run xorgwizard and then 'nouveau' was probed and chosen.

This was a problem for me in Racy, not Slacko though as nouveau loaded correctly there.

I do agree and can confirm Sage's issue with SiS graphics in Slacko and Racy when 'vesa' is loaded (recently discovered) causes "out of range" errors with the monitor, buit if the 'sis' driver loads it's fine. Yet to test Precise.

Once I got a desktop, connected straight up with my new Vodafone 'droid, browsing fine, no firstrun wizard , straight to the connect (bacon) screen. No option to change timezone, locale etc, of course because the default config is DISTRO_XORG_AUTO='yes'.

Surprised Abiword reasonably opened some of my docx homework.. which brings me to a point. Jemimah made a cool patch for glib worth consideration, it's in Slacko, forum thread. It allows the "help" to work from abiword, gnumeric, geany and many other gnomish gtk apps, homebank too. The patch also addresses mounted drives in the file browser, which need not be the case if /mnt was linked to /media, which seems to be the standard for all distros these days.. maybe an update for woof needed?

GLX not working:

[ 3423.088] (EE) AIGLX error: dlopen of /usr/lib/i386-linux-gnu/dri/nouveau_dri.so failed (libLLVM-3.0.so.1: cannot open shared object file: No such file or directory)
[ 3423.088] (EE) AIGLX: reverting to software rendering
[ 3423.088] (II) AIGLX: Screen 0 is not DRI capable
[ 3423.089] (EE) AIGLX error: dlopen of /usr/lib/i386-linux-gnu/dri/swrast_dri.so failed (libLLVM-3.0.so.1: cannot open shared object file: No such file or directory)
[ 3423.089] (EE) GLX: could not load software renderer

More tomorrow if I find the time.

Thanks

[BarryK]

I just got report that Precise Puppy might have same problem that I faced with Dpups.
Gtkdialog applications starts slower than they should. Radky has been the main bug finder and he has counted many times the starting time of gtkdialog apps.

It was detective work with strace command to trace the reason. At last I noticed that gtkdialog apps called gconv-modules.cache file. Which was missing. Woof packages-templates > glibc/usr/lib/gconv does not have template for it. Libc6 deb package provides that file.

So...here is the pet to fix that bug. Also woof might need fixing to include that file template for Upups and Dpups.

That is a very interesting discovery!

The problem is, Puppy has a cut-down set of files in /usr/lib/gconv, so the cache file is wrong.

I have generated a cache file for the cutdown set of files. Gunzip these and put them into /usr/lib/gconv, see it that works.

EDIT: I have improved these files, see my blog post:

http://bkhome.org/blog/?viewDetailed=02849

[BarryK]

Once I got a desktop, connected straight up with my new Vodafone 'droid, browsing fine, no firstrun wizard , straight to the connect (bacon) screen. No option to change timezone, locale etc, of course because the default config is DISTRO_XORG_AUTO='yes'.

Hi, I don't understand the above statement. First time that you get a desktop, QuickSetup runs, which gives you all of those settings.

[sszindian]

BK:

You may well have a mnt/home on your frugal install but I'm not testing that... I'm strictly testing a 'multi-session CD configuration'... and... 'It's NOT There'

>>>---Indian------>

[radky]

I just got report that Precise Puppy might have same problem that I faced with Dpups.
Gtkdialog applications starts slower than they should. Radky has been the main bug finder and he has counted many times the starting time of gtkdialog apps.

It was detective work with strace command to trace the reason. At last I noticed that gtkdialog apps called gconv-modules.cache file. Which was missing. Woof packages-templates > glibc/usr/lib/gconv does not have template for it. Libc6 deb package provides that file.

So...here is the pet to fix that bug. Also woof might need fixing to include that file template for Upups and Dpups.

That is a very interesting discovery!

The problem is, Puppy has a cut-down set of files in /usr/lib/gconv, so the cache file is wrong.

I have generated a cache file for the cutdown set of files. Gunzip these and put them into /usr/lib/gconv, see it that works.

Yes, this significantly improves load times for gtkdialog programs. For example, on my system with Precise alpha3, PupScan loads in 9 seconds with the updated gconv-modules and gconv-modules.cache, compared to 14 seconds without the updated files.

EDIT: After adding gconv-modules.cache to Racy 5.3 and Slacko 5.3.3., I see similar improvement in loading times for gtkdialog programs.

Thanks

[LateAdopter]

Hello Barry

Thanks for the reply

If you want KMS, edit /etc/modprobe.d/radeon.conf, change to:

options radeon modeset=1

Then reboot.

Since I don't know what I'm doing, I have just been looking at what the other Puppies say in the log.

They all load the Radeon driver and use it, and they say that KMS, DRI and GLX are enabled.

Precise Puppy says that KMS, DRI and GLX are disabled. It loads the Radeon driver but uses the VESA driver. It does not load the DRI and GLX extensions, although the files are present.

I deleted the save file and started again, to make sure that I hadn't broken anything with my random fiddling.

The initial setup screen says that it is using the Radeon driver but actually it is the VESA driver that is working. Then I shutdown and created the new savefile.

Then I edited the radeon.conf file as you said and rebooted.

The result was exactly the same as before.

I have been following this thread and pemasu's thread since they started and I have seen reports showing the VESA and NOUVEAU drivers in use but none for Radeon or Intel. There was a post saying that the Intel driver did not work.

So the configuration for these seems to be broken.

[peebee]

The b43 module in the 3.2.17 kernel used in Precise alph3, needs this firmware:

firmware: b43/ucode9.fw
firmware: b43/ucode5.fw
firmware: b43/ucode16_mimo.fw
firmware: b43/ucode15.fw
firmware: b43/ucode14.fw
firmware: b43/ucode13.fw
firmware: b43/ucode11.fw

All of which is present in the b43 firmware tarball.

Hi Barry

The b43 firmware in precise 5.2.55 is 2007 vintage:

[ 0.000000] Linux version 3.2.17 (root@puppypc) (gcc version 4.3.4 (GCC) ) #1 SMP Mon May 14 22:58:30 GMT-8 2012
[ 178.090041] b43-phy0: Loading firmware version 410.2160 (2007-05-26 15:32:10)

In pemasu's precise 5.3.3.3 he had a 2011 version of the firmware:

[ 0.000000] Linux version 3.3.2-upup (root@puppypc25823) (gcc version 4.6.3 (Ubuntu/Linaro 4.6.3-1ubuntu3) ) #1 SMP Wed Apr 18 20:00:55 EEST 2012
[ 27.313382] b43-phy0: Loading firmware version 666.2 (2011-02-23 01:15:07)

Presumably the 2011 version is "better" than the 2007.....
Cheers
peebee

[Keef]

My first try with Precise Puppy...
Manual frugal install to a Compaq Armada M700 - P3 850.
First attempts to boot had a NMI PCI system error 'for reason a9 on cpu0' - this occurs during 'Setting up services (networking. printing etc..)', then it hangs, after telling me it's 'Dazed and Confused'.
Eventually tracked it down to a CF card in a PCMCIA adapter. If it is removed the problem goes away. Also have a Linksys Wifi PCMCIA card, but this does not cause the problem.
The CF card does not need to be there, so not a great concern for me.
This particular error occurred once before for me. It was on a version of pupNgo with the 2.6.33 kernel. Never found out the cause at the time, but it may well have been the card/adapter.
Other than that, all seems ok so far

[01micko]

Once I got a desktop, connected straight up with my new Vodafone 'droid, browsing fine, no firstrun wizard , straight to the connect (bacon) screen. No option to change timezone, locale etc, of course because the default config is DISTRO_XORG_AUTO='yes'.

Hi, I don't understand the above statement. First time that you get a desktop, QuickSetup runs, which gives you all of those settings.

Hi Barry,

Yes it was a little late at night .. I'll explain.

At first boot I didn't get a desk top, just an "out of range" message on the monitor. I'm sure if the monitor could handle it I would have got the desktop and "quicksetup". I had to CTRL-ALT-BKSPCE to get to a prompt and run xorgwizard, so since quicksetup had likely already run it wasn't going to again. I hope that clarifies.

The real problem is firstly, nouveau didn't get selected for me, a problem with Racy too. I don't know why it works with Slacko, I use the stock woof method.

Secondly, vesa is causing these "out of range" messages. On one of my monitors I get that message but the desktop does show up, probably not good for the hardware. I know Sage has reported it and there have been a few instances reported in Slacko. I have it as a "known issue" on the Slacko homepage, http://01micko.com/slacko.html, with a note to do the "CTRL-ALT-BKSPCE routine to recover.

Cheers

[smokey01]

Barry a3 looking pretty good here in VirtualBox except for ayttm.

# ayttm
ayttm: error while loading shared libraries: libaudiofile.so.0: cannot open shared object file: No such file or directory
#

[linux28]

bug

[BarryK]

Hmmm... BK must have some 'thingy' about Leaf-Icons since even in beta-3 There Isn't Any Leaf (menu) and I have seen this mentioned since way back when.... even before Precise.

On the serious side... Where might one install .sfs files to be tested in Precise since we don't seem to have a mnt/home file? Do we create it ourselves? Currently running all tests on a live CD.

>>>---Indian------>

For the multi-session CD/DVD, there is a way of doing, but gee, that is a long time ago, I don't recall.

If you study the 'init' script in the 'initrd.gz' file, you should be able to figure out where Puppy searches for the .sfs files in the case of multisession.

The good thing about initrd.gz these days is all you have to do is click on it to open it up.

I seem to remember, if there are any sfs's on the CD, they can be mounted as layers, or maybe they get copied into ram and then can be mounted. ...it is too long ago, I forget all the details.

I also seem to remember, there was a place you could download them to, and they would get saved to CD, then at next boot could be loaded as a layer. Or maybe I am just dreaming, intended to implement that only.

[BarryK]

Hello Barry

This Puppy does not like ATI Radeon graphics.

If I set Radeon in the xorgwizard then X does not start.

If I copy ati_drv, radeon_drv, fbdev_drv and fbdevhw_drv to the modules drivers folder and delete vesa_drv, and delete Xorg.conf so that X auto configures, then X does start and load the correct drivers but there are errors in the Xorg.log and acceleration is disabled.

The errors are:

[ 12.893] (II) Loading /usr/lib/xorg/modules/drivers/radeon_drv.so
[ 12.893] (II) [KMS] No DRICreatePCIBusID symbol, no kernel modesetting.
[ 12.893] (WW) Falling back to old probe method for fbdev
[ 12.893] (II) Loading sub module "fbdevhw"
[ 12.893] (II) LoadModule: "fbdevhw"
[ 12.893] (II) Loading /usr/lib/xorg/modules/drivers/libfbdevhw.so
[ 12.900] (II) Module fbdevhw: vendor="X.Org Foundation"
[ 12.900] compiled for 1.11.4, module version = 0.0.2
[ 12.900] ABI class: X.Org Video Driver, version 11.0
[ 12.901] (EE) open /dev/fb0: No such device
[ 12.901] (WW) VGA arbiter: cannot open kernel arbiter, no multi-card support

and

[ 13.177] (WW) RADEON(0): Direct rendering disabled
[ 13.177] (EE) RADEON(0): Acceleration initialization failed
[ 13.177] (II) RADEON(0): Acceleration disabled

For comparison fatdog 600a2, which also has Xorg 1.11.4 and does work, gives this in the log:

[ 9.656] (II) Loading /usr/X11R7/lib64/xorg/modules/drivers/radeon_drv.so
[ 9.656] (II) [KMS] Kernel modesetting enabled.
[ 9.656] (WW) Falling back to old probe method for vesa
[ 9.656] (WW) Falling back to old probe method for fbdev
[ 9.656] (II) Loading sub module "fbdevhw"
[ 9.656] (II) LoadModule: "fbdevhw"
[ 9.661] (II) Loading /usr/X11R7/lib64/xorg/modules/libfbdevhw.so
[ 9.666] (II) Module fbdevhw: vendor="X.Org Foundation"
[ 9.666] compiled for 1.11.4, module version = 0.0.2
[ 9.666] ABI class: X.Org Video Driver, version 11.0
[ 9.666] (II) RADEON(0): Creating default Display subsection in Screen section

and

[ 9.850] (II) RADEON(0): Direct rendering enabled
[ 9.854] (II) RADEON(0): Setting EXA maxPitchBytes
[ 9.854] (II) EXA(0): Driver allocated offscreen pixmaps
[ 9.854] (II) EXA(0): Driver registered support for the following operations:
[ 9.854] (II) Solid
[ 9.854] (II) Copy
[ 9.854] (II) Composite (RENDER acceleration)
[ 9.854] (II) UploadToScreen
[ 9.854] (II) DownloadFromScreen
[ 9.854] (II) RADEON(0): Acceleration enabled

It looks like the kernel is lacking KMS

I did previously try setting Radeon in xorgwizard on Pemasu's version based on a2, and X did not start either. That had a different kernel. So I may be barking up the wrong tree.

If you want KMS, edit /etc/modprobe.d/radeon.conf, change to:

options radeon modeset=1

Then reboot.

I am wondering about that. It was on 01micko's advice that I set the default for radeon to modestting off for Woof builds.

Maybe it has matured enough so that it should now default to on.

Allright, I will do that for the next build, beta1, and we will see what happens.

[BarryK]

GLX not working:

[ 3423.088] (EE) AIGLX error: dlopen of /usr/lib/i386-linux-gnu/dri/nouveau_dri.so failed (libLLVM-3.0.so.1: cannot open shared object file: No such file or directory)
[ 3423.088] (EE) AIGLX: reverting to software rendering
[ 3423.088] (II) AIGLX: Screen 0 is not DRI capable
[ 3423.089] (EE) AIGLX error: dlopen of /usr/lib/i386-linux-gnu/dri/swrast_dri.so failed (libLLVM-3.0.so.1: cannot open shared object file: No such file or directory)
[ 3423.089] (EE) GLX: could not load software renderer

Oh... crap. File 'nouveau_dri.so' is in package:

http://packages.ubuntu.com/precise/libgl1-mesa-dri

And it has this dependency:

http://packages.ubuntu.com/precise/libllvm3.0

...look at the size of that package!!!!

I wonder if we could compile it ourselves to not have that dependency? Anyway, for now I will add it.

[soundNICK]

Bug fixes...or better just noticed missing libs...
_MAJOR SNIP_

Pemasu - just found bug fix version at your site...
yes... there are lots of inconsistencies to work out...
BUT - precise 5.3.3.3 fixed three things for me automatically...

1) It auto adjusts perfectly for my off-proportion flatscreen display

2) it gives me printing, with no special setup, in ALL printing apps

3) it uses all my dpupEXPRIMO compiled pets flawlessly

since 1) and 2) had been my only problems with
dpupExprimo, and were problems I couldnt work around,
I say HOORAY for the 5.3.3.3 bug fix version, all other possible
inconsistencies aside ! h-o-o-r-a-y : -)

I have had on and off trouble with internet connection...
this kindof hiccups sometimes, and needs connection
wizard.

blender pet didnt compile for me from repos... however...
found a blender SFS, which worked right away

I'll say it again... couldnt have asked for a more improved
puppy, but I got one anyway. Thanks to all involved in
developing precise.

soundNICK

[01micko]

Trouble with compiling without the libllmv dep is what else breaks down the track when people start adding stuff from precise repos.

-

Did some reading on archwiki about KMS and radeon. The problem I had with KMS switched on for radeon was no console, X was working fine. As you say it may be fixed by now but the archwiki seems to suggest only the X.org ATI driver supports KMS. I'll do some more tests with my reasonably modern radeon card and see what happens.