Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit 35000870 authored Apr 15, 2012 by Anton Blanchard Committed by Benjamin Herrenschmidt Apr 30, 2012

powerpc: Optimise enable_kernel_altivec

Add two optimisations to enable_kernel_altivec:

- enable_kernel_altivec has already determined if we need to
save the previous task's state but we call giveup_altivec
in both cases, requiring an extra branch in giveup_altivec. Create
giveup_altivec_notask which only turns on the VMX bit in the
MSR.

- We write the VMX MSR bit each time we call enable_kernel_altivec
even it was already set. Check the bit and branch out if we have
already set it. The classic case for this is vectored IO
where we have to copy multiple buffers to or from userspace.

The following testcase was used to confirm this patch improves
performance:

http://ozlabs.org/~anton/junkcode/copy_to_user.c



Since the current breakpoint for using VMX in copy_tofrom_user is
4096 bytes, I'm using buffers of 4096 + 1 cacheline (4224) bytes.
A benchmark of 16 entry readvs (-s 16):

time copy_to_user -l 4224 -s 16 -i 1000000

completes 5.2% faster on a POWER7 PS700.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

parent 8cd3c23d

arch/powerpc/include/asm/switch_to.h

+1 −0

Original line number	Original line	Diff line number	Diff line
	@@ -40,6 +40,7 @@ static inline void discard_lazy_cpu_state(void)
	#ifdef CONFIG_ALTIVEC		#ifdef CONFIG_ALTIVEC
	extern void flush_altivec_to_thread(struct task_struct *);		extern void flush_altivec_to_thread(struct task_struct *);
	extern void giveup_altivec(struct task_struct *);		extern void giveup_altivec(struct task_struct *);
			extern void giveup_altivec_notask(void);
	#else		#else
	static inline void flush_altivec_to_thread(struct task_struct *t)		static inline void flush_altivec_to_thread(struct task_struct *t)
	{		{

arch/powerpc/kernel/process.c

+1 −1

Original line number	Original line	Diff line number	Diff line
	@@ -124,7 +124,7 @@ void enable_kernel_altivec(void)
	if (current->thread.regs && (current->thread.regs->msr & MSR_VEC))		if (current->thread.regs && (current->thread.regs->msr & MSR_VEC))
	giveup_altivec(current);		giveup_altivec(current);
	else		else
	giveup_altivec(NULL); /* just enable AltiVec for kernel - force */		giveup_altivec_notask();
	#else		#else
	giveup_altivec(last_task_used_altivec);		giveup_altivec(last_task_used_altivec);
	#endif /* CONFIG_SMP */		#endif /* CONFIG_SMP */

arch/powerpc/kernel/vector.S

+10 −0

Original line number	Original line	Diff line number	Diff line
	@@ -89,6 +89,16 @@ _GLOBAL(load_up_altivec)
	/* restore registers and return */		/* restore registers and return */
	blr		blr

			_GLOBAL(giveup_altivec_notask)
			mfmsr r3
			andis. r4,r3,MSR_VEC@h
			bnelr /* Already enabled? */
			oris r3,r3,MSR_VEC@h
			SYNC
			MTMSRD(r3) /* enable use of VMX now */
			isync
			blr

	/*		/*
	* giveup_altivec(tsk)		* giveup_altivec(tsk)
	* Disable VMX for the task given as the argument,		* Disable VMX for the task given as the argument,