Re: [patch V2 21/28] x86/speculation: Prepare for conditional IBPB in switch_mm()

2018-11-26 Thread Ingo Molnar
* Thomas Gleixner wrote: > On Sun, 25 Nov 2018, Andy Lutomirski wrote: > > > On Nov 25, 2018, at 2:20 PM, Thomas Gleixner wrote: > > > On Sun, 25 Nov 2018, Andi Kleen wrote: > > > > > >>> The current check whether two tasks belong to the same context is using > > >>> the > > >>> tasks

Re: [patch V2 21/28] x86/speculation: Prepare for conditional IBPB in switch_mm()

2018-11-25 Thread Thomas Gleixner
On Sun, 25 Nov 2018, Andy Lutomirski wrote: > > On Nov 25, 2018, at 2:20 PM, Thomas Gleixner wrote: > > On Sun, 25 Nov 2018, Andi Kleen wrote: > > > >>> The current check whether two tasks belong to the same context is using > >>> the > >>> tasks context id. While correct, it's simpler to use

Re: [patch V2 21/28] x86/speculation: Prepare for conditional IBPB in switch_mm()

2018-11-25 Thread Thomas Gleixner
On Sun, 25 Nov 2018, Andi Kleen wrote: > On Sun, Nov 25, 2018 at 11:20:50PM +0100, Thomas Gleixner wrote: > > On Sun, 25 Nov 2018, Andi Kleen wrote: > > > > > > The current check whether two tasks belong to the same context is using > > > > the > > > > tasks context id. While correct, it's

Re: [patch V2 21/28] x86/speculation: Prepare for conditional IBPB in switch_mm()

2018-11-25 Thread Andi Kleen
On Sun, Nov 25, 2018 at 11:20:50PM +0100, Thomas Gleixner wrote: > On Sun, 25 Nov 2018, Andi Kleen wrote: > > > > The current check whether two tasks belong to the same context is using > > > the > > > tasks context id. While correct, it's simpler to use the mm pointer > > > because > > > it

Re: [patch V2 21/28] x86/speculation: Prepare for conditional IBPB in switch_mm()

2018-11-25 Thread Andy Lutomirski
> On Nov 25, 2018, at 2:20 PM, Thomas Gleixner wrote: > > On Sun, 25 Nov 2018, Andi Kleen wrote: > >>> The current check whether two tasks belong to the same context is using the >>> tasks context id. While correct, it's simpler to use the mm pointer because >>> it allows to mangle the

Re: [patch V2 21/28] x86/speculation: Prepare for conditional IBPB in switch_mm()

2018-11-25 Thread Thomas Gleixner
On Sun, 25 Nov 2018, Andi Kleen wrote: > > The current check whether two tasks belong to the same context is using the > > tasks context id. While correct, it's simpler to use the mm pointer because > > it allows to mangle the TIF_SPEC_IB bit into it. The context id based > > mechanism requires

Re: [patch V2 21/28] x86/speculation: Prepare for conditional IBPB in switch_mm()

2018-11-25 Thread Andi Kleen
> The current check whether two tasks belong to the same context is using the > tasks context id. While correct, it's simpler to use the mm pointer because > it allows to mangle the TIF_SPEC_IB bit into it. The context id based > mechanism requires extra storage, which creates worse code. [We

Re: [patch V2 21/28] x86/speculation: Prepare for conditional IBPB in switch_mm()

2018-11-25 Thread Thomas Gleixner
On Sun, 25 Nov 2018, Thomas Gleixner wrote: > /* > + * Use bit 0 to mangle the TIF_SPEC_IB state into the mm pointer which is > + * stored in cpu_tlb_state.last_user_mm_ibpb. > + */ > +#define LAST_USER_MM_IBPB0x1UL > + > +/* > + unsigned long next_tif = task_thread_info(next)->flags; > +

[patch V2 21/28] x86/speculation: Prepare for conditional IBPB in switch_mm()

2018-11-25 Thread Thomas Gleixner
The IBPB speculation barrier is issued from switch_mm() when the kernel switches to a user space task with a different mm than the user space task which ran last on the same CPU. An additional optimization is to avoid IBPB when the incoming task can be ptraced by the outgoing task. This