2.6.31.14 "ide kernel"

Although there is some doubt whether we really need this (see previous blog thread), I have compiled the 2.6.31.14 kernel (same version as used in Wary 5.0) with the old IDE drivers and without SMP and without tickless configuration.

PET uploaded here:
http://distro.ibiblio.org/pub/linux/distributions/quirky/pet_packages-wary5/linux_kernel-2.6.31.14-uniproc-ide-w5.pet

The .config file can be found here:
http://bkhome.org/sources/kernel-2.6.31.14/uniproc-ide-kernel/

This cannot be used in the latest Woof, as some scripts do not properly recognise the /dev/hd* drive naming. I am working on that now.


Posted on 26 Jan 2011, 8:11


Comments:

Posted on 26 Jan 2011, 8:43 by 01micko
re ide kernel
Does this mean 2 wary versions?
Reason I ask is that nvidia drivers will likely need re-compilation against the "IDE" kernel.
Also, I recently compiled nividia-96 for spup (missing in wary) though I don't have that hardware to test. Forum member James C has an mx-400, did some testing for me and said the driver is successful. What that means is that I can compile all the current nvidia drivers from the nvidia site (71.xx, 96.xx, 173.xx and 260.xx). That would mean a slight rework of the video-upgrade wizard. Let me know if you want them, but the latter 2 would likely be unneeded for the IDE kernel as they are for later nvidias.

Also, pnethood-0.66.2 isn't at ibiblio yet.

Cheers


Posted on 26 Jan 2011, 10:33 by BarryK
Woof and /dev/hd*
Ok, I have restored support for /dev/hd* in the boot scripts in Woof.

While I was at it, I fixed a couple of other things in Woof. One fix was memory cards that show up as /dev/mmcblk0p1 (for example) -- Puppy should now be able to boot from these and/or use them for a save-file.

Another fix was in /etc/rc.d/PUPSTATE, I got ATADRIVES='usbcd' even though I have booted off an internal IDE/ATA optical drive.

01micko,
I don't know if the nVidia kernel modules would need to be recompiled. Perhaps, I will do a build and find out.

Regarding pnethood, clarf has released 0.6.7, which I will upgrade to very soon.



Posted on 26 Jan 2011, 22:05 by rodin.s
How to test it
How to test this kernel on Wary-5.0? Installing the pet is not enough. Is it for development only?


Posted on 26 Jan 2011, 23:42 by BarryK
Re test kernel
You have to wait until the next build of Wary (in Woof).



Posted on 5 Feb 2011, 4:25 by K Godt
First Impression : No problems !
On The 2.4 GHz Intel Single Core machine [ 448 MB SD-RAM 500 GB HD ] which is actually not faster than the 1045 MHz AMD-K7 [ 512 MB SD-RAM 2 HDs=80 GB ] machine due to the bottlenecks of RAM and DMA speed , it booted without any problems into my luci-218(pre-511)-full-install. All data is ok , the loadaverage is as low as usual , drivers loaded , Sound O.K.

When it is about ""old"" hardware I would rather change a CRT monitor for a newer [ LED ] one and replace the PSU,RAM,HD of a "desktop"-machine that has got 1000+ MHz . The former Xorgs ( 4-series )seems to work smoother with CRT monitors than the actual ones at the 5-Lupu-seies.

I really think that this IDE feature had been "psychologically" missing for wanting to support hardware that is more than lets say 5 years old. I really do not know if more than 10 or even 15 years would fit for the actual libraries and drivers .
I tried to compile 'dillo' for example , thought that these few K/MB would go fast and after 24 hours at luci-218 I have a 0.8.6 that will not download and 2.X would until now while disabling everything . "-D __cplusplus" and I think I can reinstall quite a few headers .....
So I think it is good to support older hardware but they also should at least be max RAM equipped and perhaps it is better to leave win3/95 on them than trying to crash them with demand for too much speed or just something new .

But nevertheless, this IDE-Kernel is I think a good improvement for reputation and fits without scandals into the today's puppy-kernels !

While compiling dillo wanted this and that and some libs really seem to need to be compiled for the architecture zlib because of the crypt-drivers , but many warning/error mesgs aren't true.

I must say that until today I do not know anything about how Xorg communicates with the kernel and it's [video-]modules. 'Nv' had been good enough until now .

Have Fun !