Re: [PATCH v10 1/7] x86/tsc: remove tsc_disabled flag

2018-06-19 Thread Peter Zijlstra
On Tue, Jun 19, 2018 at 10:57:35PM +0200, Thomas Gleixner wrote: > I'm not arguing against removing it. I just refuse to accept changelogs > which suggest that there is no functional change. Fair enough..

Re: [PATCH v10 1/7] x86/tsc: remove tsc_disabled flag

2018-06-19 Thread Thomas Gleixner
On Tue, 19 Jun 2018, Peter Zijlstra wrote: > On Tue, Jun 19, 2018 at 09:12:45PM +0200, Thomas Gleixner wrote: > > On Tue, 19 Jun 2018, Peter Zijlstra wrote: > > > It does not; there is TSC usage even if you boot with notsc on. See how > > > it does not clear X86_FEATURE_TSC for instance. > > > >

Re: [PATCH v10 1/7] x86/tsc: remove tsc_disabled flag

2018-06-19 Thread Peter Zijlstra
On Tue, Jun 19, 2018 at 09:12:45PM +0200, Thomas Gleixner wrote: > On Tue, 19 Jun 2018, Peter Zijlstra wrote: > > It does not; there is TSC usage even if you boot with notsc on. See how > > it does not clear X86_FEATURE_TSC for instance. > > Well, kinda. There is some stuff in the apic calibration

Re: [PATCH v10 1/7] x86/tsc: remove tsc_disabled flag

2018-06-19 Thread Thomas Gleixner
On Tue, 19 Jun 2018, Peter Zijlstra wrote: > On Tue, Jun 19, 2018 at 07:32:49PM +0200, Thomas Gleixner wrote: > > On Fri, 15 Jun 2018, Pavel Tatashin wrote: > > > > > tsc_disabled is set when notsc is passed as kernel parameter. The reason > > > we > > > have notsc is to avoid timing problems on

Re: [PATCH v10 1/7] x86/tsc: remove tsc_disabled flag

2018-06-19 Thread Peter Zijlstra
On Tue, Jun 19, 2018 at 07:32:49PM +0200, Thomas Gleixner wrote: > On Fri, 15 Jun 2018, Pavel Tatashin wrote: > > > tsc_disabled is set when notsc is passed as kernel parameter. The reason we > > have notsc is to avoid timing problems on multi-socket systems. We already > > have a mechanism, howe

Re: [PATCH v10 1/7] x86/tsc: remove tsc_disabled flag

2018-06-19 Thread Pavel Tatashin
> > tsc_disabled is set when notsc is passed as kernel parameter. The reason we > > have notsc is to avoid timing problems on multi-socket systems. We already > > have a mechanism, however, to detect and resolve these issues by invoking > > tsc unstable path. Thus, make notsc to behave the same as

Re: [PATCH v10 1/7] x86/tsc: remove tsc_disabled flag

2018-06-19 Thread Thomas Gleixner
On Fri, 15 Jun 2018, Pavel Tatashin wrote: > tsc_disabled is set when notsc is passed as kernel parameter. The reason we > have notsc is to avoid timing problems on multi-socket systems. We already > have a mechanism, however, to detect and resolve these issues by invoking > tsc unstable path. Th

[PATCH v10 1/7] x86/tsc: remove tsc_disabled flag

2018-06-15 Thread Pavel Tatashin
tsc_disabled is set when notsc is passed as kernel parameter. The reason we have notsc is to avoid timing problems on multi-socket systems. We already have a mechanism, however, to detect and resolve these issues by invoking tsc unstable path. Thus, make notsc to behave the same as tsc=unstable.

[PATCH v10 1/7] x86/tsc: remove tsc_disabled flag

2018-02-09 Thread Pavel Tatashin
tsc_disabled is set when notsc is passed as kernel parameter. The reason we have notsc is to avoid timing problems on multi-socket systems. We already have a mechanism, however, to detect and resolve these issues by invoking tsc unstable path. Thus, make notsc to behave the same as tsc=unstable.