Re: [RFC 0/5] parker: PARtitioned KERnel

[H. Peter Anvin]

On September 24, 2025 1:14:26 PM PDT, Fam Zheng <fam@xxxxxxxxxx> wrote:
>On Wed, Sep 24, 2025 at 9:10 PM Fam Zheng <fam@xxxxxxxxxx> wrote:
>>
>>
>>
>> On Wed, Sep 24, 2025 at 8:02 PM H. Peter Anvin <hpa@xxxxxxxxx> wrote:
>>>
>>> On September 24, 2025 8:22:54 AM PDT, Dave Hansen <dave.hansen@xxxxxxxxx> wrote:
>>> >On 9/23/25 08:31, Fam Zheng wrote:
>>> >> In terms of fault isolation or security, all kernel instances share
>>> >> the same domain, as there is no supervising mechanism. A kernel bug
>>> >> in any partition can cause problems for the whole physical machine.
>>> >> This is a tradeoff for low-overhead / low-complexity, but hope in
>>> >> the future we can take advantage of some hardware mechanism to
>>> >> introduce some isolation.
>>> >I just don't think this is approach is viable. The buck needs to stop
>>> >_somewhere_. You can't just have a bunch of different kernels, with
>>> >nothing in charge of the system as a whole.
>>> >
>>> >Just think of bus locks. They affect the whole system. What if one
>>> >kernel turns off split lock detection? Or has a different rate limit
>>> >than the others? What if one kernel is a big fan of WBINVD? How about
>>> >when they use resctrl to partition an L3 cache? How about microcode updates?
>>> >
>>> >I'd just guess that there are a few hundred problems like that. Maybe more.
>>> >
>>> >I'm not saying this won't be useful for a handful of folks in a tightly
>>> >controlled environment. But I just don't think it has a place in
>>> >mainline where it needs to work for everyone.
>>>
>>> Again, this comes down to why a partitioning top level hypervisor is The Right Thing[TM].
>>>
>>> IBM mainframes are, again, the archetype here, having done it standard since VM/370 in 1972. This was running on machines with a *maximum* of 4 MB memory.
>>>
>>> This approach works.
>>>
>>> Nearly every OS on these machines tend to run under a *second* level hypervisor, although that isn't required.
>>
>>
>I'm trying to think about the hypervisor approach you mentioned, but
>if it doesn't provide memory and I/O isolation, what is the advantage
>over this RFC? (if it doesn I think then we're talking about a
>specially configured KVM which does 1:1 vcpu pinning etc).
>
>
>Sorry, forgot to turn off email html mode in my previous message..
>
>
>Fam
>

The difference is that this is highly invasive to the OS, which affects developers and users not wanting this feature.