Issues with kernel 2.6.0-test1 and other thoughts
Greeting list,
I must start out by saying that I am totally thrilled by the new kernel. I
finally figured out what was causing all those dumb unresolved symbols errors,
so I spent some time tinkering with it today and got it running on both of my
unstable machines. I still have a couple of problems, though.
First machine:
Toshiba Satellite 2805-S401 notebook
P-III 700 MHz mobile processor, Intel PIIX4 440BX chipset
S3 Savage IX-MV 8 MB graphics, 256 MB RAM, 20 BB HDD
Positive:
- ACPI actually works. My BIOS was blacklisted in the 2.4.x ACPI, but
now it works. I get temperature, battery stat, AC stat, etc.
- Preemptible kernel seems much more responsive
- Integration of ALSA drivers is very nice (rather than build modules)
Negative:
- Kernel's SpeedStep driver does not yet support my chipset
- USB mouse tracking is a touch fast (more annoying than anything else)
Second machine:
Custom built
AMD Athlon XP 2500+, 333 MHz FSB
Biostar motherboard with nForce2 chipset
Radeon 9000 Pro 128 MB graphics, 1 GB Dual-channel DDR RAM, 120 GB HDD
Positive:
- System seems more responsive (but this is difficult to judge since
I have only had the machine a short while)
- nVidia's AGPGART patch is now incorporated into the kernel
- nVidia's modified audio driver is now incorporated into the kernel
(in the newly added ALSA drivers section)
- The Radeon framebuffer driver now actually supports the R250 chip
in the Radeon 9000 Pro (haven't actually compiled it in yet, though)
Negative:
- The binary nvnet.o module will not compile because of changes to the
PCI and net device interfaces (I guess). I spent a couple of hours
hacking at this, but each time I solved one proble I created at least
two more. Has anyone with an nForce2 mobo got this working?
- The system is under a constant 50% load in the new kernel. When I pull
up top and let it sit, it show 50% system and 50% idle, but none of
the processes in the list show more than 1% utilization (and I am
sorting by decsending CPU utilization)
- There is a still problem with the vesafb.c that I had to manually fix
(apparently it tries to reserve the top 128 MB of the first 1 GB of
address space for the frame buffer, which causes problems with systems
with 1+ GB of RAM, like mine)
If anyone can help out with the nForce networking and the strange CPU load,
I would really appreciate it.
-Roberto
___________________________________________________
Yahoo! Messenger - Nueva versión GRATIS
Super Webcam, voz, caritas animadas, y más...
http://messenger.yahoo.es
Reply to: