ACK/Cmnt: [SRU][Xenial][PULL] Guests using IBRS incur a large performance penalty (LP: #1764956)
Stefan Bader
stefan.bader at canonical.com
Wed Jan 9 08:28:45 UTC 2019
On 08.01.19 17:32, Juerg Haefliger wrote:
> On Tue, 8 Jan 2019 14:43:34 +0100
> Stefan Bader <stefan.bader at canonical.com> wrote:
>
>> On 19.12.18 11:03, Juerg Haefliger wrote:
>>> git://git.launchpad.net/~juergh/+git/xenial-linux lp1764956-v2
>>
>>
>> Glancing over the changes I noticed that "UBUNTU: SAUCE: x86/speculation:
>> Cleanup IBPB runtime control handling" adds use of spec_ctrl_mutex
>
> spec_ctrl_mutex is already defined (introduced by the original IBRS/IBPB
> patches).
Hm, ok, sorry, then better not change it. Not that obvious looking at the patch.
>
>
>> but only
>> "UBUNTU: SAUCE: x86/speculation: Cleanup IBRS runtime control handling" which
>> follows next actually seems to declare it.
>
> That patch just moves the definition from kernel/smp.c to kernel/sysctl.c.
>
>
>> In the end that does not matter only
>> could make bisection a pain.
>
>
> IIRC, I've compile-tested every single patch.
>
>
>> Minor pedantic nitbit: is not "UBUNTU: SAUCE: x86/speculation: Move RSB_CTXSW
>> hunk" moving the hunk further down?
>
> Nope. The diff shows a different hunk being moved down as a result of moving
> RSB_CTXSW up. So the title is misleading and should probably read 'Move
> X86_FEATURE_IBPB hunk'.
Actually could have been fooled by the way the diff shows. There was another
patch which said move something before but then looked like moving something
after. But that was just because what got moved was a hunk before the hunk that
was claimed to be moved. And indeed, if you move something from before to
after, then the thing after is implicitly moved to the front. :/
.Stefan
>
> ...Juerg
>
>
>> Apart from that the other changes roughly seemed to make sense but I have to
>> admit that I went rather quickly over the non-SAUCE ones. The delta is just too
>> big and I rely a bit on the successful testing.
>>
>> So if fixing up the first 2 patches (which should be doable when applying)
>>
>> Acked-by: Stefan Bader <stefan.bader at canonical.com>
>>
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <https://lists.ubuntu.com/archives/kernel-team/attachments/20190109/a1cb6698/attachment.sig>
More information about the kernel-team
mailing list