[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Not usefull kernel images take all first Debian sarge CD :(

"AKL. Mantas Kriauciunas" <mantas@akl.lt> wrote:
>I suggest always include only one latest 'stable' kernel version and (this 
>is very important) kernel-pcmcia-modules and kernel-headers of this kernel
>version on first Debian CD (386,586,686,k6,k7 variants, but no smp variants,
>because there are not so many smp systems and because of this they can 
>install kernel images from internet or next CD's).
>Kernel headers are very usefull for users, which should compile some modules
>for internet access, for example there are lots of "winmodems" and internal
>DSL modems, which reguires special kernel modules, not included in standard
>kernel, but included in driver CD of this hardware.

IMNSHO, the first CD should include *one* kernel image -- one which will work 
reasonably on *all* hardware of the appropriate architecture.

If there isn't such an image, then the first CD will need to include a few 
kernel images, so that there will be one of them which works reasonably on 
all hardware of the appropriate architecture.

In the case of the 386 architecture in particular:
* the 386 kernel is no good for SMP machines (due to emulation of the 486 
locking instructions) and is quite a bit slower than needed on 486+ machines
* the 486+ kernels don't work on 386 machines
So we need at least two kernels in that case.  From what I can tell the rest 
of the optimized kernel versions don't make such a significant difference -- 
although if debian-installer cleverly installs the "right" one, I guess we 
need them all on the first CD.

I'd actually advise shipping a 2.6 kernel as well as (rather than?) a 2.4 
kernel on the first CD, because 2.6 supports so much hardware that 2.4 

No kernel versions other than the most current 2.4 and 2.6 versions belong on 
the first CD, of course, unless they're needed for certain architectures (in 
which case maybe those architectures should be dropped, actually).

Reply to: