Page 5 of 11
Posted: Sat 28 Apr 2012, 15:07
by vtpup
thanks 01micko for the explanation. I don't think it's worth pursuing further then.
The only program this seems to affect is gmplayer. VLC doesn't have the view re-size problem. So I can simply make that the default.
Otherwise, graphics intensive and typically difficult programs seem to run well. For instance, I can run Google Sketchup (3D CAD) in Wine.
So doesn't seem worth it jto try to solve ust for a gmplayer quirk, unless I find some other problem down the road.
I'm still curious for future reference, how one could recover from a messed up video driver .pet installation. Is there a way to remove the .pet and restore the former Xorg system?
I imagine it would be a sequence performed in the console before starting X. Can you remove a .pet from the command line?
Seemed like if there was something wrong with xorg, you used to be able to boot into xvesa and correct it. But I don't think I saw that as an option this last time -- or maybe it didn't work. Can't quite remember.
Posted: Sat 28 Apr 2012, 15:58
by 666philb
pemasu wrote:# find /root/my-documents/x_xorg76_mega_pkg_DEV-7.6-5-w5c -name "Xv*"
/root/my-documents/x_xorg76_mega_pkg_DEV-7.6-5-w5c/usr/X11R7/include/X11/extensions/Xvlib.h
/root/my-documents/x_xorg76_mega_pkg_DEV-7.6-5-w5c/usr/X11R7/include/X11/extensions/Xv.h
http://her.gr.distfiles.macports.org/mi ... -5-w5c.pet
DISTRO_PKGS_SPECS:
yes|x_xorg76_mega_pkg|x_xorg76_mega_pkg|exe,dev,doc,nls| #RACY Xorg 7.6 mega-pkg
hi pemasu.... is that for me?
i'm still getting the same error
Posted: Sat 28 Apr 2012, 16:50
by Karl Godt
There are two possibilities :
1) The configure script is testing something not correctly (buggy or lacks options)
2) the needed library does not include the needed function (function not enabled at compile time, function not implemented (anymore) in library or library not found in the search PATH)
Posted: Sat 28 Apr 2012, 21:08
by rjbrewer
One Gmplayer solution with Racy:
My Latitude D820 has 945gm graphics.
The default boot loads vesa, which gives the big yellow
border when trying to use Mplayer full screen.
Running Xorgwizard and choosing "intel Intel" solves the
problem.
Having a PAE kernel is useless for machines with 945 graphics.
It will not read past 3.3 of the 4Gb of ram.
edit:
Deleteing the i915 files from the new (3.xx) kernel versions
increases glxgears speed from 20fps to 330 fps on the
Latitude.
Posted: Mon 30 Apr 2012, 13:41
by vtpup
After trying out 5 other late kernel Puppies and testing for a week, Racy is the clear winner on this brand new dual proccesor machine. It is far superior to the others in performing graphics intensive CADs through Wine.
Racy 53 also worked perfectly out of the box with wireless, and sound, and the screen dimming keys also worked.
Racy will now be the day-to-day operating system for this computer. It replaces Puppy 4.3.2 v3 and my old Thinkpad T43, which gave years of stable use as a heavy duty work computer.
Thank you Barry for this solid high-performance OS!
(EDIT: blacklisting i915 module caused big problems with this particular computer, overloading the proc when clicking on desktop icons. Restoring i915 to the loaded modules solved the problem. I'm not messing with the graphics system any more. Performance is great using vanilla setup and simply choosing "Probe" in xorgwizard.)
Posted: Mon 30 Apr 2012, 15:13
by vtpup
It looks like the new Puppy wallpaper setter actually alters image files in /Usr/share/backgrounds/.
I put an image in there, and running the wallpaper setter actually cropped the image and saved the new file over the old one!
In Nathan wallpaper setter, the image wasn't overwritten in /.../backgrounds, and you could select it and show it uncropped as a tile or centered, etc.
This new wallpaper setter actually messes up the image, and once that happens you can never go back to showing the whole image as a tile or centered. It seems to serially re-write the image every time you make a change.
I'd like to get rid of this app and bring back Nathan Wallpaper setter on this computer, if possible.
Programs like these should never re-write an image!
EDIT: Clarification -- I was using an image that was larger than the screen pixel dimensions, and the monitor is a wide screen.
Basically the wallpaper setter cropped the image to wide screen aspect ratio, then saved it over the original image.
This destroyed a lot of the original image. When I tried to "center" or "tile" it to see the original whole image, it didn't work because the new cropped and saved image exactly fit the screen. I had expected that the original image would be retained and fit onto the screen with gray space at the edges when "centered".
I've subsequently found that if I use an image that is smaller than the screen, centering and tiling work. I do not believe under these circumstances that the image is re-written. But I'm not sure about the stretch option which may rewrite the image. Haven't tested that.
Posted: Mon 30 Apr 2012, 16:54
by rjbrewer
vtpup wrote:
(EDIT: blacklisting i915 module caused big problems with this particular computer, overloading the proc when clicking on desktop icons. Restoring i915 to the loaded modules solved the problem. I'm not messing with the graphics system any more. Performance is great using vanilla setup and simply choosing "Probe" in xorgwizard.)
Trashing and blacklisting are 2 totally different things.
Posted: Mon 30 Apr 2012, 20:01
by vtpup
RJBrewer, would you explain further in relation to i915 files:
1.) Why it would be desirable to trash them? What do they do and why would they interfere with graphics speed?
2.) What is the difference in system modules loaded and running between either trashing all i915 files, or blacklisting the i915 module?
Posted: Tue 01 May 2012, 13:19
by rjbrewer
vtpup wrote:RJBrewer, would you explain further in relation to i915 files:
1.) Why it would be desirable to trash them? What do they do and why would they interfere with graphics speed?
2.) What is the difference in system modules loaded and running between either trashing all i915 files, or blacklisting the i915 module?
If I go to setup-configure startup-blacklist modules and
blacklist i915; i915 still appears in the loaded list and the
blacklist list.
It generally has no effect or a negative effect on glxgears.
Trashing a few major i915 modules increases glxgears fps
on both my 855gm and 945gm laptops in some (but not
all) cases.
On the 945gm Latitude the fps jumped from 20fps to 300
fps.
Unfortunately, when I rebooted it wasn't possible to restart
x. May be related to running frugal as I don't have a hard drive
for that machine yet and I always use full installs if possible.
i915 modeset was a hot topic back when Lupu was just
getting started and was pushed as a cure for intel graphic
black screen and shutdown problems.
I found that using i810 module was a much better solution
on my 855gm laptop.
I have no way of knowing what is going to work without
setting up experimental installs.
Posted: Tue 01 May 2012, 14:32
by vtpup
rjbrewer wrote:
On the 945gm Latitude the fps jumped from 20fps to 300
fps.
Unfortunately, when I rebooted it wasn't possible to restart
x. May be related to running frugal as I don't have a hard drive
for that machine yet and I always use full installs if possible
Same here, and I use frugal installs.
Were you able to recover from the X problem?
I'd really like to know how such a system (say from a blacklisted i915) could be repaired without trashing the personal savefile and starting over.
Is it possible to un-blacklist a video module if you can't start X and you get a black screen with blinking cursor, and no key response.
I did try pfix=clean in the grub menu.lst stanza but that made no difference to the start up problem. Didn't try pfix=purge.
Seems like there might be a less extreme alternative means if you could boot to terminal and then load the blacklisted module, and then do an xwin to boot to X with module in place. Then use Bootmanager to un-blacklist the module. Then re-start.
But what pfix line boots to cursor, and what are the commands to load a module from console, say i915?
I'd like to just know this stuff in case of any similar future X problems.
Posted: Tue 01 May 2012, 16:05
by Karl Godt
vtpup wrote:rjbrewer wrote:
On the 945gm Latitude the fps jumped from 20fps to 300
fps.
Unfortunately, when I rebooted it wasn't possible to restart
x. May be related to running frugal as I don't have a hard drive
for that machine yet and I always use full installs if possible
Same here, and I use frugal installs.
Were you able to recover from the X problem?
I'd really like to know how such a system (say from a blacklisted i915) could be repaired without trashing the personal savefile and starting over.
Is it possible to un-blacklist a video module if you can't start X and you get a black screen with blinking cursor, and no key response.
I did try pfix=clean in the grub menu.lst stanza but that made no difference to the start up problem. Didn't try pfix=purge.
Seems like there might be a less extreme alternative means if you could boot to terminal and then load the blacklisted module, and then do an xwin to boot to X with module in place. Then use Bootmanager to un-blacklist the module. Then re-start.
But what pfix line boots to cursor, and what are the commands to load a module from console, say i915?
I'd like to just know this stuff in case of any similar future X problems.
Newer Intel graphic chips (900+) and goo' old Xorg-1.3.0 intel_drv.so from Puppy-4.3 don't play very well together .
Newer Xorg-1.7+ since Lupu should be fine with i915.ko and Xorg's driver intel(_drv.so) .
Blacklisting using the bootmanager adds to the modprobe configuration file used to autoload kernel drivers at boot time (/tmp/modprobe.conf) . Xorg itself seems to be capable to load drivers itself and won't care about /tmp/modprobe.conf . Former puppy modprobe-v3.6 seems to be patched to automatically obey /etc/i[modprobe.d]/[modprobe.conf]* blacklist DRIVERNAME entries , while it is not the default . To make modprobe command obey blacklist entries , modprobe has to be aliased in configuration files like "alias modprobe='modprobe -b'" in /etc/profile , same as already "ls" is aliased there .
To load i915 for example :
or
What modeset actually does is still not clear to me ...
There is also the insmod command, but it requires the full path/driver-file-name.extension :
Code: Select all
insmod /lib/modules/`uname -r`/kernel/drivers/gpu/drm/i915/i915.ko[.gz]
To remove a module :
or
Posted: Tue 01 May 2012, 16:16
by rjbrewer
vtpup wrote:
Were you able to recover from the X problem?
I'd really like to know how such a system (say from a blacklisted i915) could be repaired without trashing the personal savefile and starting over.
Is it possible to un-blacklist a video module if you can't start X and you get a black screen with blinking cursor, and no key response.
Guess that's why constant backups are necessary with
frugal.
With full installs; I just pop in the cd, rerun the install, choose
upgrade instead of wipe, and it fixes any file problems
without destroying my setups.
Posted: Tue 01 May 2012, 17:31
by npierce
vtpup wrote:Can you remove a .pet from the command line?
While I don't know about Racy 5.3, this worked from the text console in Puppy 4.3.1:
Notice the minus sign before the name, and the lack of ".pet" after it.
(The opposite was not true: petget could
not install from the text console in 4.3.1, it only downloaded the pet.)
vtpup wrote:But what pfix line boots to cursor . . .
Posted: Tue 01 May 2012, 18:24
by vtpup
Thanks very much Karl Godt and npierce, that information is very very valuable to understand, and to have. I'm writing it in a little notebook I keep by the computer. I guess it would be a good idea to add all of the pfix commands there, too.
that -packagename is very cool......
Posted: Sat 05 May 2012, 15:19
by vtpup
Possible bug -- just mentioning it here as a bug report -- I've raised a separate request for personal help on this in the regular User's forum.
Symptoms:
After a couple weeks of using Racy53, I'm suddenly getting many instances of a process called kworker. These use up processor resources to the point that the OS is inoperable a few seconds after X starts. I have to hard reset with the on-off button. Menu shutdown, Alt-Ctl-Del doesn't work. Can't open a terminal for poweroff, either. Well maybe if I could do it quick enough.....
The computer is listed in the sig below. This computer and Racy53 were working fine yesterday evening. I haven't installed any hardware or software since then, or made any changes at all. I did not have any hard shutdowns, or suspends before this occurrence. Now I have, of course.
The computer hardware seems to be fine, I am dual booted right now into Upup Precise to write this.
Posted: Sat 05 May 2012, 16:22
by Karl Godt
Haven't had such problems, googling revealed these :
https://porteus.org/forum/viewtopic.php?f=41&t=250
http://lwn.net/Articles/403891/
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=blob;f=Documentation/workqueue.txt;h=01c513fac40ece66b73a1709f6df496b5aade8d2;hb=HEAD
The links are from one page ( in german ) with several additional likks to try :
http://forum.ubuntuusers.de/topic/kworker-frisst-cpu/
The best possible solution seems to be
Code: Select all
echo N >/sys/module/drm_kms_helper/parameters/poll
which may be related to your radeon? craphics chip/card .
and permanent solution /etc/modprobe.d/local.conf
options drm_kms_helper poll=N
So we likely need to know the drivers which are loaded .
Posted: Sat 05 May 2012, 17:15
by vtpup
Thank you Karl. The problem is suddenly gone, but I would still like to try to pin it down if we can, anyway.
The only difference I can see between multiple attempts to start up this morning, and starting up now, is that I moved my physical location. I was at a coffeeshop this morning, and am home now, so I'm wondering if it is related to wireless connection processes.
My understanding now is that kworker processes are actually place holder processes for other processes. Possibly wireless was one of those.
In older kernels pprocess or top would list actual processes running, but now these are seen as kworker processes -- so it is a lot harder to tell what is actually causing the hang up.
I like the older system better where you could actually see what was going on and what the cause was.
Anyway, to answer your questions, Karl, I don't have Radeon graphics. It is a laptop with integrated graphics using Intel Second Generation Core Processor graphics.
To get a list of drivers loaded would I run modprobe -l ?
Thanks again.
Posted: Sat 05 May 2012, 18:53
by Karl Godt
No vetty , it was
lsmod command , one of the very few commands without ARGV[S]
If the list is very long , then maybe filter it by
|grep "PATTERN"
btw my acer has died and i preferably wont use an acer again the next time .
My preferences are asus,gigabyte,elitegroup,dell,aoc,eizo ... ... ?
My preferences are full installations, also ... ...
Posted: Sat 05 May 2012, 19:38
by vtpup
Code: Select all
Module Size Used by
parport_pc 19580 0
lp 5996 0
parport 21043 2 parport_pc,lp
i915 290754 2
drm_kms_helper 17982 1 i915
drm 121018 3 i915,drm_kms_helper
i2c_algo_bit 3556 1 i915
cpufreq_ondemand 4432 2
acpi_cpufreq 4481 1
mperf 787 1 acpi_cpufreq
fan 1714 0
snd_hda_codec_hdmi 16805 1
snd_hda_codec_realtek 203051 1
uvcvideo 47297 0
videodev 52792 1 uvcvideo
arc4 962 2
ath9k 62987 0
mac80211 134600 1 ath9k
ath9k_common 1242 1 ath9k
ath9k_hw 238866 2 ath9k,ath9k_common
ath 10746 2 ath9k,ath9k_hw
cfg80211 113680 3 ath9k,mac80211,ath
pcspkr 1195 0
snd_hda_intel 16395 0
snd_hda_codec 52722 3 snd_hda_codec_hdmi,snd_hda_codec_realtek,snd_hda_intel
snd_hwdep 3730 1 snd_hda_codec
snd_pcm_oss 28135 0
snd_mixer_oss 10062 1 snd_pcm_oss
snd_pcm 47219 4 snd_hda_codec_hdmi,snd_hda_intel,snd_hda_codec,snd_pcm_oss
snd_seq_dummy 915 0
snd_seq_oss 18929 0
snd_seq_midi 3280 0
atl1c 23866 0
i2c_i801 6154 0
i2c_core 12704 6 i915,drm_kms_helper,drm,i2c_algo_bit,videodev,i2c_i801
snd_rawmidi 11899 1 snd_seq_midi
snd_seq_midi_event 3596 2 snd_seq_oss,snd_seq_midi
snd_seq 32450 6 snd_seq_dummy,snd_seq_oss,snd_seq_midi,snd_seq_midi_event
snd_timer 11590 2 snd_pcm,snd_seq
snd_seq_device 3509 5 snd_seq_dummy,snd_seq_oss,snd_seq_midi,snd_rawmidi,snd_seq
snd 33474 13 snd_hda_codec_hdmi,snd_hda_codec_realtek,snd_hda_intel,snd_hda_codec, snd_hwdep,snd_pcm_oss,snd_mixer_oss,snd_pcm,snd_seq_oss,snd_rawmidi,snd_seq,snd_timer,snd_seq_device
soundcore 3255 1 snd
snd_page_alloc 4765 2 snd_hda_intel,snd_pcm
shpchp 18143 0
sparse_keymap 2056 0
rfkill 10068 1 cfg80211
intel_agp 8065 1 i915
intel_gtt 9446 3 i915,intel_agp
agpgart 18222 3 drm,intel_agp,intel_gtt
wmi 5859 0
battery 8251 0
thermal 6066 0
video 9504 1 i915
evdev 5773 1
button 3315 1 i915
processor 20797 1 acpi_cpufreq
thermal_sys 9477 4 fan,thermal,video,processor
hwmon 961 1 thermal_sys
ac 2303 0
fuse 47724 0
aufs 121827 47
fontconfig missing in GTK
Posted: Sun 06 May 2012, 14:40
by Mobeus
fontconfig is absent in the GTK includes. Compiling with
Code: Select all
`pkg-config --cflags --libs gtk+-2.0`
errors with
Code: Select all
/usr/lib/gcc/i486-t2-linux-gnu/4.3.4/../../../../i486-t2-linux-gnu/bin/ld: cannot find -lfontconfig