TahrPup 5.8.4

[666philb]

hello @666PhilB

Testing TahrPUP running in a KVM capable PC, shows a good speed increase in TahrPUP. But, there is a nagging issue that I wonder if you would consider resolving.

Video driver
The OOTB video card of a KVM virtual guest PC has a Cirrus card emulated. All other PUPs tested have a Cirrus module/driver and when the guests boot,they all find the Cirrus card and use it.

But, in TahrPUP,this is not occurring and there is no way to get X desktop via console of TahrPUP.

Fortunately, QEMU offers an option (-vga std) to set the VM guest PC as VESA video. Below is screenshots of QEMU setup for VESA and using FirstRUN to change the video map to match my 4X3 monitor from the TahrPUP default for a 16X10 origin.

Hope this is helpful

I hope this patch, applied to the Ubuntu 14.0.4, against the kernel which enables some pentium M hidden PAE.
https://lkml.org/lkml/2014/3/7/180

i've uploaded a new 3.14.11 kernel which hopefully has the force PAE patch applied, and the cirrus qemu module inbuilt https://www.dropbox.com/s/3x121v1gt8ib9 ... .11.tar.gz

[666philb]

fixes for tahrpup will now be coming through quickpet.

click the update button in quickpet to get the latest quickpet.
there should now be a tahr bugfix button (in update) which will install the latest fixes.

updates ffconvert thanks shinobar
updates PUI thanks 01micko
fixes dir2pet thanks billtoo
increases contrast of highlighting in default theme

[01micko]

i can't get ndiswrapper to compile

Code: Select all

# make
make -C utils
make[1]: Entering directory `/root/ndiswrapper-1.59/utils'
gcc -g -Wall -I../driver  -o loadndisdriver loadndisdriver.c
make[1]: Leaving directory `/root/ndiswrapper-1.59/utils'
make -C driver
make[1]: Entering directory `/root/ndiswrapper-1.59/driver'
Makefile:33: *** Multiple copies of version.h found, please clean your build tree.  Stop.
make[1]: Leaving directory `/root/ndiswrapper-1.59/driver'
make: *** [driver] Error 2
# 

does anyone know what the problem is?

In /usr/src/linux there are 2 (or maybe more) version.h files. Move the other(s) out of the way, and you will get further. It's trial and error as to which one, but I did it a while back for 3.12/3.13

[666philb]

cheers 01micko,

got rid off all version.h's apart from /usr/src/linux/include/generated/uapi/linux/version.h
the official ndiswrapper from sourceforge still wouldn't compile but i found this patched one that did https://github.com/Schwartz/ndiswrapper-patched-3.14/

ndiswrapper pets here https://www.dropbox.com/sh/5i98qm00abz5 ... 2SaUZUGGXa
choose the one that matches your kernel

[shinobar]

i've uploaded a new 3.14.11 kernel which hopefully has the force PAE patch applied, and the cirrus qemu module inbuilt https://www.dropbox.com/s/3x121v1gt8ib9 ... .11.tar.gz

Great!
Now TahrPup is running on Pentium M 1.6GHz, my notebook compaq nc4010.

title tahr 5.8.3 (sda2/tahr-5.8.3)
uuid 3eff6dbd-90d8-40d5-a966-12255895d295
kernel /tahr-5.8.3/vmlinuz forcepae pmedia=atahd psubdir=tahr-5.8.3 pfix=fsck
initrd /tahr-5.8.3/initrd.gz

vmlinuz need to be add ID string, otherwise PDEV1 is blank. See /etc/rc.d/PUPSTATE.

Code: Select all

# echo -n 't140711144545ZZZZ5.8.3XXXXXXXXXX' >> vmlinuz

[gcmartin]

In the past, several members who have laptops with Pentium M's have shared that their PCs wont boot PAE kernels. 01Micko posted on his website

Please note that Intel Pentium M processors with a 400Mhz FSB (see glossary) do not run the PAE version. These were marketed around 2005 as "Centrino".

Recently, @Shinobar proposed and presented a resolution (patch) that exposes the PAE feature so that both PAE and non-PAE PUP distros will boot on capable Pentium Ms.

Is there any chance that there is an ISO built with the PATCH so that this could be tested. Further, an ISO like this would allow members who have Pentium Ms who, indeed, have PAE boot issues to report impact results of patch's use.

Each of my Pentium Ms boot without the need for the patch. If there is, somewhere, a PAE ISO with the patch, I volunteer to boot my Pentium Ms AND report what happens when booting without the kernel parm as well as what happens booting with the kernel parm on my Pentium Ms. This would present an impact report of the patch on systems which already have the hardware feature in the CPU.

Anyone with knowledge of such a PUP distro, please post.

[slavvo67]

GCMartin:

To simplify, are you basically saying the result is a PAE and Non-PAE in one?

If that is possible, it would be fantastic! I have 2 PC's that will run non-PAE and only one PAE.

Best,

Slavvo67

[gcmartin]

Yes @Slavvo67. There are a few Intel CPUs which were manufactured and sold to motherboard vendors with a "hidden" PAE feature. The most noted of these are some Pentium Ms.

@Shinobar posted for developers a kernel patch which when included in the distro, allows Linux to expose that hidden feature for system use. This means that those system can now take advantage of all features which was previously not available to the OS at boot time. This, for example, would allow all PCs to boot PhatSlacko (or the same PUP distro without regard for whether distro is PAE or nonPAE; as is the case for AMD CPUs and most every other Intel CPU).

Hope this help in understanding

[shinobar]

i've uploaded a new 3.14.11 kernel which hopefully has the force PAE patch applied, and the cirrus qemu module inbuilt https://www.dropbox.com/s/3x121v1gt8ib9 ... .11.tar.gz

I made .iso, which does not mean any TahrPup derivative, but just for making it easy to test 666philb's new kernel.
It does not contain any fix other than replacing the vmlinuz add ID string and zdrv. The boot menu is add 'forcepae' parameter. It runs both with non-PAE Pentium M CPU and with any other PAE capable CPU.

261f685dd699c2127f6302bf324c5768 tahr-5.8.3-forcepae.iso
http://shino.pos.to/party/bridge.cgi?pu ... rpup/test/

Run with 'puppy pfix=ram' boot option (no need 'forcepae') if you have TahrPup on your HDD. Otherwise, Puppy runs from the HDD.

The tahr-5.8.3-forcepae.iso is made using Mkdist:
http://www.murga-linux.com/puppy/viewtopic.php?t=94034

[imago]

hi
can you say what for ydrv need?
i use adrv for make save to sfs module

[gcmartin]

Thanks @Shinobar

Turns out the distro looks the same, per se, no matter if I boot with the boot parm or without. Preliminary pics attached. If there is any additional testing desired, request and I'll comply.

[666philb]

hi
can you say what for ydrv need?
i use adrv for make save to sfs module

hi imago . this explains http://www.murga-linux.com/puppy/viewtopic.php?t=83675

[666philb]

vmlinuz need to be add ID string, otherwise PDEV1 is blank. See /etc/rc.d/PUPSTATE.

Code: Select all

# echo -n 't140711144545ZZZZ5.8.3XXXXXXXXXX' >> vmlinuz

hi shinobar,

can you explain this a bit more... does that command edit the vmlinuz? and do i need to run it on the other kernels i've created?

thanks

[666philb]

gcmartin,

does the iso that shinobar created have the qemu cirrus driver you requested?

[666philb]

GCMartin:

To simplify, are you basically saying the result is a PAE and Non-PAE in one?

If that is possible, it would be fantastic! I have 2 PC's that will run non-PAE and only one PAE.

Best,

Slavvo67

if shinobars iso doesn't work for you there's a non pae kernel here https://www.dropbox.com/sh/czf7b44ob03v ... DYR1PaDpca unzip it and copy over the zdrv.sfs and vmlinuz replacing the original ones

[imago]

hi
can you say what for ydrv need?
i use adrv for make save to sfs module

hi imago . this explains http://www.murga-linux.com/puppy/viewtopic.php?t=83675

i want know how boot ydrv, before main puppy sfs or after? for example adrv boots before main puppy sfs

[mavrothal]

i want know how boot ydrv, before main puppy sfs or after? for example adrv boots before main puppy sfs

The init mounts the main sfs, adrv, ydrv and zdrv together

[shinobar]

Puppy loads in the order, pupsave, adrv, ydrv, main sfs, zdrv, and then extra sfs.

[pemasu]

I just tested Shinobar`s packaged tahr puppy iso with 3.14.11 kernel.
Several things are splendid in it.
- intel/nvidia 710M dual graphics works great. No need to remove nouveau.ko kernel module. Suspend and resume without freezing works with that kernel. 3.14 kernel and up has finally fized that nouveau.ko freezing problem. 3.15 kernel is ok also.
- Powertop works. There is sufficient config options enabled in kernel so that powertop gets information. Busybox included one is broken as always, but from ppm Powertop 2.5 version works. You get power usage information when in battery with laptop by using powertop. Great way to check how energy efficient is the kernel. Intel_pstate cpu freguency governor has been used in this kernel. Meaning no more ondemand option without disabling it. I test how responding this intel_pstate is. Previously I have disabled it with boot parameter: intel_pstate=disable
- Ath9k.ko wireless kernel module loads now with this kernel automatically. No need to load by using /root/Startup/folder with script to load it. Great !
- Xrandr shows me 1360x768 screen resolution option. And I obtain it by feeding it to the xorg.conf file. The default 1920x1080 resolution is too small for my eyes...due to aging and presbyopia. None of recent Slackos provide me 1360x768 resolution so Slacko`s are beyond me.
- The flsynclient and touchpad gets me delicate touchpad responsiveness....previously I couldnt finetune touchpad much with Dpup Wheezy or Upup Raring because the touchpad movements turned to awkward like drunken. Now I can finetune touchpad with Tahr Puppy. I suppose xorg version and kernel improvements are behind this progress.

- Overall...after finetuning things for my liking....I have great Puppy distro under my fingers here. Thanks 666philb of this Tahr Puppy.

http://i.imgur.com/3CyhJfT.jpg

[shinobar]

vmlinuz need to be add ID string, otherwise PDEV1 is blank. See /etc/rc.d/PUPSTATE.

Code: Select all

# echo -n 't140711144545ZZZZ5.8.3XXXXXXXXXX' >> vmlinuz

hi shinobar,

can you explain this a bit more... does that command edit the vmlinuz?

The vmlinuz and the main sfs has hidden ID string.
You can see it typing like
# tail -c 32 puppy_thar_5.3.8.sfs
# tail -c 32 vmlinuz

Puppy need to know from which vmlinuz it booted off. The ID string is a label to identify which. The ID string value is written in the file DISTRO_SPECS.

do i need to run it on the other kernels i've created?

Yes. The ID string is automatically add by the woof. You need manually add the ID string to the new vmlinuz if you do not make a distro by the woof. It means the kernel is not portable between Puppy distro or version, and replacing kernel is not simple even if you use the huge kernel/zdrv system.