Re: [PATCH v2 0/7] KVM TSS cleanups and speedups

[Andy Lutomirski]

On Wed, Feb 22, 2017 at 7:13 AM, Paolo Bonzini <pbonzini@xxxxxxxxxx> wrote:
> On 21/02/2017 20:14, Andy Lutomirski wrote:
>> The first four patches here are intended to be straightforward
>> cleanups and to make a better base for Thomas' GDT series.  They may
>> be a slight speedup, too, because they remove an STR instruction
>> from the VMX entry path.
>>
>> The last two patches are a reasonably large speedup but need careful
>> review.
>>
>> FWIW, I can see lots of additional easy-ish speedups here.  For example:
>>
>>  - The GDT reload on VM exit isn't really needed at all.  Instead let's
>>    just change the kernel limit to 0xFFFF.  Doing that naively would
>>    waste memory, but doing it carefully on top of Thomas' series would
>>    be straightforward and almost free.
>>
>>  - RDMSR from MSR_GS_BASE is totally pointless.
>>
>>  - Once I or someone finishes the FSGSBASE series, we get a big speedup
>>    there.
>>
>>  - The LDT reload code should be split up and optimized better, I think.
>>
>> Changes from v1:
>>  - Fix some changelog typos.
>>  - Fix the bug that Paolo found.
>>  - Rename the helpers to make their usage more obvious.
>>  - Move clearing __tss_limit_invalid into force_reload_TR() as a tiny
>>    optimization.
>>  - Add a test case.  It doesn't test all the machinations, but at least
>>    it checks basic functionality.
>>
>> Andy Lutomirski (7):
>>   x86/asm: Define the kernel TSS limit in a macro
>>   x86/kvm/vmx: Don't fetch the TSS base from the GDT
>>   x86/kvm/vmx: Get rid of segment_base() on 64-bit kernels
>>   x86/kvm/vmx: Simplify segment_base()
>>   x86/asm/64: Drop __cacheline_aligned from struct x86_hw_tss
>>   x86/kvm/vmx: Defer TR reload after VM exit
>>   selftests/x86: Add a basic selftest for ioperm
>>
>>  arch/x86/include/asm/desc.h          |  62 +++++++++++--
>>  arch/x86/include/asm/processor.h     |  12 ++-
>>  arch/x86/kernel/ioport.c             |  11 +++
>>  arch/x86/kernel/process.c            |  10 ++
>>  arch/x86/kvm/vmx.c                   |  63 ++++++-------
>>  tools/testing/selftests/x86/Makefile |   2 +-
>>  tools/testing/selftests/x86/ioperm.c | 171 +++++++++++++++++++++++++++++++++++
>>  7 files changed, 284 insertions(+), 47 deletions(-)
>>  create mode 100644 tools/testing/selftests/x86/ioperm.c
>>
>
> I pushed and tagged before seeing this v2. :(  The differences seem to
> be x86-only, so I suppose Ingo can handle them if you resubmit.
>

I renamed the helpers to make it less likely that someone would repeat
my little buglet.  I can submit a patch that just has the differences,
but I think it should go in through your tree.  Would that work?

--Andy


-- 
Andy Lutomirski
AMA Capital Management, LLC