[plug] [rant] 2.6 Kernels

[Michael Collard]

On Wed, 2004-09-01 at 16:40, Leon Blackwell wrote:
> On Wed, Sep 01, 2004 at 04:16:34PM +0800, Michael Collard wrote:
> > With the number of bugs, its resembling something that Microsoft make.
> 
> Hey... let's not get nasty  :)

Thats not nasty. Nasty would be installing windows.

> > the kernel has stability problems with nvidia drivers causing lockups
> > etc.
> 
> Are you running the most recent nvidia drivers?  There were some issues
> with earlier versions, but nvidia has fixed these.  Of course, if nvidia
> had released code for their drivers, I'm sure they would have been fixed
> long before then  :)

Been down that road. I'm running 6111 which are really great NVidia
drivers. I've already proven the kernel to be the cause of the lockups.
AFAIK, all versions pre 2.6.8.0-rc1 will cause lock ups after a week or
so. Most mine made it was 5 days on 2.6.7 or 2.6.3 kernel.

> > So far the least broken kernel I've found is 2.6.8.0-rc2 which only has
> > one of the problems mentioned above. More or less just wanted to give
> > all the pluggers a warning about recent 2.6 kernels, don't want other
> > people to go through what I've been through. 
> 
> I have to say I'm in the "concerned" camp with regard to the way the
> linux kernel is now being developed.  Without stable/unstable branching,
> I just don't see how we're going to ensure that end users get a usable
> product.  At this point, the idea seems to be to leave this to the
> distros, who will pick and choose stable patches, rather than simply
> moving through the kernel versions as they are released.
> people at this point, and "if it ain't broke, don't fix it".

This is pretty much the point I'm making. I though the 2.[1,3,5,7].
kernels were meant to be the unstable ones. So far from the 2.6 tree, I
see 3 versions (including RC) that I would consider running on my
hardware. I mean, I am not meant to be finding bugs in the kernel, I'm
nowhere near as knowledgeable as the actual kernel developers. 

I didn't even get a acknowledgement of my bug report to the LKML (Linux
Kernel Mailing List) about my reproducible loop oops bug. I mean, f@&K,
I don't even know if it will be fixed or if my posts were just ignored.
You can understand how I'm caught between a user and a developer. I
don't want to fix this stuff, just use it(tm).

Regards
Michael Collard