Re: [PATCH v3 4/5] x86/mwait-idle: disable IBRS during long idle

2022-10-14 Thread Jan Beulich
On 14.10.2022 10:03, Roger Pau Monné wrote: > On Thu, Oct 13, 2022 at 04:15:04PM +0200, Jan Beulich wrote: >> On 13.10.2022 15:10, Roger Pau Monné wrote: >>> On Thu, Oct 13, 2022 at 02:17:54PM +0200, Jan Beulich wrote: On 13.10.2022 14:03, Roger Pau Monné wrote: > On Thu, Aug 18, 2022 at 0

Re: [PATCH v3 4/5] x86/mwait-idle: disable IBRS during long idle

2022-10-14 Thread Roger Pau Monné
On Thu, Oct 13, 2022 at 04:15:04PM +0200, Jan Beulich wrote: > On 13.10.2022 15:10, Roger Pau Monné wrote: > > On Thu, Oct 13, 2022 at 02:17:54PM +0200, Jan Beulich wrote: > >> On 13.10.2022 14:03, Roger Pau Monné wrote: > >>> On Thu, Aug 18, 2022 at 03:04:51PM +0200, Jan Beulich wrote: > From

Re: [PATCH v3 4/5] x86/mwait-idle: disable IBRS during long idle

2022-10-13 Thread Jan Beulich
On 13.10.2022 15:10, Roger Pau Monné wrote: > On Thu, Oct 13, 2022 at 02:17:54PM +0200, Jan Beulich wrote: >> On 13.10.2022 14:03, Roger Pau Monné wrote: >>> On Thu, Aug 18, 2022 at 03:04:51PM +0200, Jan Beulich wrote: From: Peter Zijlstra Having IBRS enabled while the SMT sibling i

Re: [PATCH v3 4/5] x86/mwait-idle: disable IBRS during long idle

2022-10-13 Thread Roger Pau Monné
On Thu, Oct 13, 2022 at 02:17:54PM +0200, Jan Beulich wrote: > On 13.10.2022 14:03, Roger Pau Monné wrote: > > On Thu, Aug 18, 2022 at 03:04:51PM +0200, Jan Beulich wrote: > >> From: Peter Zijlstra > >> > >> Having IBRS enabled while the SMT sibling is idle unnecessarily slows > >> down the runnin

Re: [PATCH v3 4/5] x86/mwait-idle: disable IBRS during long idle

2022-10-13 Thread Jan Beulich
On 13.10.2022 14:03, Roger Pau Monné wrote: > On Thu, Aug 18, 2022 at 03:04:51PM +0200, Jan Beulich wrote: >> From: Peter Zijlstra >> >> Having IBRS enabled while the SMT sibling is idle unnecessarily slows >> down the running sibling. OTOH, disabling IBRS around idle takes two >> MSR writes, whic

Re: [PATCH v3 4/5] x86/mwait-idle: disable IBRS during long idle

2022-10-13 Thread Roger Pau Monné
On Thu, Aug 18, 2022 at 03:04:51PM +0200, Jan Beulich wrote: > From: Peter Zijlstra > > Having IBRS enabled while the SMT sibling is idle unnecessarily slows > down the running sibling. OTOH, disabling IBRS around idle takes two > MSR writes, which will increase the idle latency. > > Therefore,

[PATCH v3 4/5] x86/mwait-idle: disable IBRS during long idle

2022-08-18 Thread Jan Beulich
From: Peter Zijlstra Having IBRS enabled while the SMT sibling is idle unnecessarily slows down the running sibling. OTOH, disabling IBRS around idle takes two MSR writes, which will increase the idle latency. Therefore, only disable IBRS around deeper idle states. Shallow idle states are bounde