Re: [PATCH v2 2/3] arch/arm64: fpsimd: Introduce fpsimd_save_and_flush_cpu_state() and use it

[Julien Grall]

Hi Dave,

On 4/16/19 12:53 PM, Dave Martin wrote:
> On Fri, Apr 12, 2019 at 06:14:19PM +0100, Julien Grall wrote:
> > The only external user of fpsimd_save() and fpsimd_flush_cpu_state() is
> > the KVM FPSIMD code.
> >
> > A follow-up patch will mandate hold of the FPSIMD context while
>
> This is a bit hard to read without understanding the following patch.
>
> Maybe elaborate a bit with something like
>
> "The following patch will introduce a mechanism to acquire ownership of the FPSIMD/SVE context for performing context management operations. Rather than having to export this [...]"

I will update the commit message with your suggestion.

> > modifying it. Rather than having to export the new helpers to get/put
> > the context, we can just introduce a new function to combine
> > fpsimd_save() and fpsimd_flush_cpu_state().
> >
> > This has also the advantage to remove any external call of fpsimd_save()
> > and fpsimd_flush_cpu_state(), so they can be turned static.
> >
> > Lastly, the new function can also be used in the PM notifier.
> >
> > Signed-off-by: Julien Grall <julien.grall@xxxxxxx>
> >
> > ---
> >      Changes in v2:
> >          - Patch added
> > ---
> >   arch/arm64/include/asm/fpsimd.h |  5 ++---
> >   arch/arm64/kernel/fpsimd.c      | 17 +++++++++++++----
> >   arch/arm64/kvm/fpsimd.c         |  3 +--
> >   3 files changed, 16 insertions(+), 9 deletions(-)
> >
> > diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h
> > index 32abbfd852e8..88c223f6ded6 100644
> > --- a/arch/arm64/include/asm/fpsimd.h
> > +++ b/arch/arm64/include/asm/fpsimd.h
> > @@ -48,11 +48,11 @@ struct task_struct;
> >   extern void fpsimd_save_state(struct user_fpsimd_state *state);
> >   extern void fpsimd_load_state(struct user_fpsimd_state *state);
> >   
> > -extern void fpsimd_save(void);
> > -
> >   extern void fpsimd_thread_switch(struct task_struct *next);
> >   extern void fpsimd_flush_thread(void);
> >   
> > +extern void fpsimd_save_and_flush_cpu_state(void);
> > +
>
> Nit, bit I think this makes more sense alongside the other flush
> functions.
>
> Can you just drop this new declaration in place of the old declaration
> for sve_flush_cpu_state()?

Sure, I will do it in the next version.

> >   extern void fpsimd_signal_preserve_current_state(void);
> >   extern void fpsimd_preserve_current_state(void);
> >   extern void fpsimd_restore_current_state(void);
> > @@ -63,7 +63,6 @@ extern void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *state,
> >   				     void *sve_state, unsigned int sve_vl);
> >   
> >   extern void fpsimd_flush_task_state(struct task_struct *target);
> > -extern void fpsimd_flush_cpu_state(void);
> >   
> >   /* Maximum VL that SVE VL-agnostic software can transparently support */
> >   #define SVE_VL_ARCH_MAX 0x100
> > diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c
> > index 577296bba730..9e4e4b6acd93 100644
> > --- a/arch/arm64/kernel/fpsimd.c
> > +++ b/arch/arm64/kernel/fpsimd.c
> > @@ -241,7 +241,7 @@ static void task_fpsimd_load(void)
> >    *
> >    * Softirqs (and preemption) must be disabled.
> >    */
> > -void fpsimd_save(void)
> > +static void fpsimd_save(void)
> >   {
> >   	struct fpsimd_last_state_struct const *last =
> >   		this_cpu_ptr(&fpsimd_last_state);
> > @@ -1117,12 +1117,22 @@ void fpsimd_flush_task_state(struct task_struct *t)
> >    * Invalidate any task's FPSIMD state that is present on this cpu.
> >    * This function must be called with softirqs disabled.
> >    */
> > -void fpsimd_flush_cpu_state(void)
> > +static void fpsimd_flush_cpu_state(void)
> >   {
> >   	__this_cpu_write(fpsimd_last_state.st, NULL);
> >   	set_thread_flag(TIF_FOREIGN_FPSTATE);
> >   }
> >   
> > +/*
> > + * Save the FPSIMD state to memory and invalidate cpu view.
> > + * This function must be called with softirqs (and preemption) disabled.
> > + */
> > +void fpsimd_save_and_flush_cpu_state(void)
> > +{
> > +	fpsimd_save();
> > +	fpsimd_flush_cpu_state();
> > +}
> > +
> >   #ifdef CONFIG_KERNEL_MODE_NEON
> >   
> >   DEFINE_PER_CPU(bool, kernel_neon_busy);
> > @@ -1279,8 +1289,7 @@ static int fpsimd_cpu_pm_notifier(struct notifier_block *self,
> >   {
> >   	switch (cmd) {
> >   	case CPU_PM_ENTER:
> > -		fpsimd_save();
> > -		fpsimd_flush_cpu_state();
> > +		fpsimd_save_and_flush_cpu_state();
>
> Any reason not to do this in kernel_neon_begin() too?

In the next patch, fpsimd_save_and_flush_cpu_state() will also call 
{get, put}_cpu_fpsimd_context(). So we avoid to expose the two new 
functions outside of the file.

I could introduce a double-underscore version of this function if you 
think it is worth it.

> >   		break;
> >   	case CPU_PM_EXIT:
> >   		break;
> > diff --git a/arch/arm64/kvm/fpsimd.c b/arch/arm64/kvm/fpsimd.c
> > index 6e3c9c8b2df9..01bd2e862506 100644
> > --- a/arch/arm64/kvm/fpsimd.c
> > +++ b/arch/arm64/kvm/fpsimd.c
> > @@ -113,8 +113,7 @@ void kvm_arch_vcpu_put_fp(struct kvm_vcpu *vcpu)
> >   		u64 *guest_zcr = &vcpu->arch.ctxt.sys_regs[ZCR_EL1];
> >   
> >   		/* Clean guest FP state to memory and invalidate cpu view */
>
> You can lose the comment here, since the call is now self-explanatory.

Sure.

> > -		fpsimd_save();
> > -		fpsimd_flush_cpu_state();
> > +		fpsimd_save_and_flush_cpu_state();
>
> [...]
>
> Otherwise, looks sensible.

Thank you for the review!

Cheers,

--
Julien Grall