On Tue, 2014-02-25 at 13:37 +0000, Ian Jackson wrote:
> Cyril Brulebois writes ("Re: Bug#602506: HP DL165 boot crash with lenny i386 686 but OK with -bigmem or amd64"):
> > Either way, kernel selection was adjusted over the last release cycles,
> > especially after kernel flavours were reduced to a bare minimum. I doubt
> > this bug is still current, so closing for now.
>
> Fair enough. I don't have a reasonable way to try to repro this right
> now. If I do get a chance to try this with wheezy I will reopen this
> bug if it is still present.
>
> Thanks for your consideration.
This is a 12-core system. In lenny, the 686 flavour supported only 8
CPUs whereas the 686-bigmem flavour supported 32, and I suspect that
this led to the boot failure (though I would have expected the extra
CPUs to simply be unused - and obviously that did work for the 486
(single CPU) kernel in the installer). Also, I expect that this system
has at least 4GB of RAM, some of which would have been inaccessible with
the 686 flavour even if it could boot.
I believe this was fixed in squeeze as:
- Starting with linux 2.6.30-1, both 686 and 686-bigmem flavours
supported up to 32 CPUs
- I changed d-i's kernel flavour selection to use 686-bigmem if PAE is
supported and there is RAM with a physical address above 4GB, so the 686
flavour would not be selected for this system
In wheezy the 686-pae flavour (renamed 686-bigmem) is installed on all
systems supporting PAE. Not that I would recommend installing i386 on a
64-bit capable system, anyway.
Ben.
--
Ben Hutchings
Everything should be made as simple as possible, but not simpler.
- Albert Einstein
Attachment:
signature.asc
Description: This is a digitally signed message part