daid kahl пишет:
>
>     > Also, you should avoid using oldconfig except for really minor
>     kernel
>     > upgrades. I know this is mentioned in documentation elsewhere, but
>     > just a useful reminder.
>     >
>     > ~daid
>
>     This has been discussed on this list before. Running make oldconfig
>     works fine. I, and a lot of others, have said this many times. I
>     configed one kernel about 5 years ago and have used oldconfig ever
>     since. It is faster and less prone to problems than starting from
>     scratch.
>
>     If you are going from 2.4 to 2.6, then you should start fresh. I
>     recently went from 2.6.23 to a 2.6.28 with no problem, other than
>     trying
>     to figure out that new group stuff.
>
>     Dale
>
>
> Huh. I was going from 2.6.23 to 2.6.26 about 2 weeks ago and using 
> oldconfig was giving me some major trouble as I recall. Not sure 
> specifically now, but I know CONFIG_HPET_TIMER had to be turned off 
> later during manual configuration, or my machine was hanging after 
> some ACPI scripts. This behavior is replicated elsewhere for similar 
> hardware, though the author claimed it was fixed after 2.6.25 which I 
> found to clearly not be the case. I had also not manually configured 
> my own kernel either, so I thought it's worth the experience.
>
> I could save my current configuration for 2.6.26 and try oldconfig 
> again from 2.6.23.9 to make another 2.6.26 kernel and see what the 
> problem was.
>
> ~daid
Hm, may be it's off-topic but I've seen such behavior on one of my 
servers - some ACPI events were hanging server with last kernels. The 
only thing I had to be done is to turn off ACPI in BIOS. Now service 
local isn't working and CPU frequency is not scaling also. Is this 
normal or can this parameter (CONFIG_HPET_TIMER) help me?

Best regards,
Evgeniy B.