Re: Question about the new kernel with PAE (Wheezy) - Report
On Tue, 21 Jun 2011 20:13:06 +0200, Sven Joachim wrote:
> On 2011-06-21 19:52 +0200, Camaleón wrote:
>
>> Well, if you agree with the update, the pae kernel installs despite it
>> warns about it will not work (and when you boot with it, it fails as
>> expected). You can still boot with the old kernel (good job!).
>>
>> So in the end you need to manually remove the pae kernel and install
>> the 486, as Gilbert suggested.
>>
>> I still think this should have been automagically done by the upgrade.
>
> I guess it's a bit difficult, since the package management cannot know
> that the kernel will not work with your CPU.
Well, after I chose "yes, update" it downloaded the packages and it
presented the warning prompt, so it did know something about my processor
specs before proceeding with the install.
But why go on? Why not "download, prepare the install, detect CPU
capabilities, stop and ask the user to a) install 486 kernel, b) don't do
anything or c) proceed anyway?
Yes, yes, I know. I am asking for too much ;-)
But just out of curiosity, what's the raw logic behind the routine that
decided to install a PAE kernel instead another one? Why the installer
took such option? :-?
> The alternative of downgrading to the -486 kernel for all former -686
> users is not very attractive either, since that means losing SMP
> support. This will certainly be mentioned in the Wheezy release notes
> eventually.
I can enable PAE/NX for the VM but never liked the PAE kernels. And only
have one processor available so SMP on/off wouldn't be noticed, right?
>> Why proceed with installing something that will not work and even set
>> it as the default boot option >:-P
>
> The installation has actually failed, leaving the -686-pae kernel
> unconfigured, has it not?
Yep. But I spent time, bandwith and hard disk space (that is a very
scarce resource in my small 8 GiB VM) in the process :-PP
> I think there is a problem with the
> upgrade-grub script that generates menu entries for all kernels, whether
> they work or not.
I understand the last installed kernel goes first, that seems pretty
logical, regardless it works or not, this cannot be foreseen by GRUB,
that will try to boot whatever it finds.
Greetings,
--
Camaleón
Reply to: