[PATCH v2 09/21] arm64: KVM: Implement guest entry

Marc Zyngier marc.zyngier at arm.com
Tue Dec 1 10:41:11 PST 2015 

On 01/12/15 15:29, Christoffer Dall wrote:
> On Fri, Nov 27, 2015 at 06:50:03PM +0000, Marc Zyngier wrote:
>> Contrary to the previous patch, the guest entry is fairly different
>> from its assembly counterpart, mostly because it is only concerned
>> with saving/restoring the GP registers, and nothing else.
>>
>> Signed-off-by: Marc Zyngier <marc.zyngier at arm.com>
>> ---
>>  arch/arm64/kvm/hyp/Makefile |   1 +
>>  arch/arm64/kvm/hyp/entry.S  | 155 ++++++++++++++++++++++++++++++++++++++++++++
>>  arch/arm64/kvm/hyp/hyp.h    |   2 +
>>  3 files changed, 158 insertions(+)
>>  create mode 100644 arch/arm64/kvm/hyp/entry.S
>>
>> diff --git a/arch/arm64/kvm/hyp/Makefile b/arch/arm64/kvm/hyp/Makefile
>> index ec14cac..1e1ff06 100644
>> --- a/arch/arm64/kvm/hyp/Makefile
>> +++ b/arch/arm64/kvm/hyp/Makefile
>> @@ -7,3 +7,4 @@ obj-$(CONFIG_KVM_ARM_HOST) += vgic-v3-sr.o
>>  obj-$(CONFIG_KVM_ARM_HOST) += timer-sr.o
>>  obj-$(CONFIG_KVM_ARM_HOST) += sysreg-sr.o
>>  obj-$(CONFIG_KVM_ARM_HOST) += debug-sr.o
>> +obj-$(CONFIG_KVM_ARM_HOST) += entry.o
>> diff --git a/arch/arm64/kvm/hyp/entry.S b/arch/arm64/kvm/hyp/entry.S
>> new file mode 100644
>> index 0000000..2c4449a
>> --- /dev/null
>> +++ b/arch/arm64/kvm/hyp/entry.S
>> @@ -0,0 +1,155 @@
>> +/*
>> + * Copyright (C) 2015 - ARM Ltd
>> + * Author: Marc Zyngier <marc.zyngier at arm.com>
>> + *
>> + * This program is free software; you can redistribute it and/or modify
>> + * it under the terms of the GNU General Public License version 2 as
>> + * published by the Free Software Foundation.
>> + *
>> + * This program is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>> + * GNU General Public License for more details.
>> + *
>> + * You should have received a copy of the GNU General Public License
>> + * along with this program.  If not, see <http://www.gnu.org/licenses/>.
>> + */
>> +
>> +#include <linux/linkage.h>
>> +
>> +#include <asm/asm-offsets.h>
>> +#include <asm/assembler.h>
>> +#include <asm/fpsimdmacros.h>
>> +#include <asm/kvm.h>
>> +#include <asm/kvm_arm.h>
>> +#include <asm/kvm_asm.h>
>> +#include <asm/kvm_mmu.h>
>> +
>> +#define CPU_GP_REG_OFFSET(x)	(CPU_GP_REGS + x)
>> +#define CPU_XREG_OFFSET(x)	CPU_GP_REG_OFFSET(CPU_USER_PT_REGS + 8*x)
>> +
>> +	.text
>> +	.pushsection	.hyp.text, "ax"
>> +
>> +.macro save_common_regs ctxt
>> +	stp	x19, x20, [\ctxt, #CPU_XREG_OFFSET(19)]
>> +	stp	x21, x22, [\ctxt, #CPU_XREG_OFFSET(21)]
>> +	stp	x23, x24, [\ctxt, #CPU_XREG_OFFSET(23)]
>> +	stp	x25, x26, [\ctxt, #CPU_XREG_OFFSET(25)]
>> +	stp	x27, x28, [\ctxt, #CPU_XREG_OFFSET(27)]
>> +	stp	x29, lr,  [\ctxt, #CPU_XREG_OFFSET(29)]
>> +.endm
>> +
>> +.macro restore_common_regs ctxt
>> +	ldp	x19, x20, [\ctxt, #CPU_XREG_OFFSET(19)]
>> +	ldp	x21, x22, [\ctxt, #CPU_XREG_OFFSET(21)]
>> +	ldp	x23, x24, [\ctxt, #CPU_XREG_OFFSET(23)]
>> +	ldp	x25, x26, [\ctxt, #CPU_XREG_OFFSET(25)]
>> +	ldp	x27, x28, [\ctxt, #CPU_XREG_OFFSET(27)]
>> +	ldp	x29, lr,  [\ctxt, #CPU_XREG_OFFSET(29)]
>> +.endm
>> +
>> +.macro save_host_regs reg
>> +	save_common_regs \reg
>> +.endm
>> +
>> +.macro restore_host_regs reg
>> +	restore_common_regs \reg
>> +.endm
>> +
>> +.macro save_guest_regs
>> +	// x0 is the vcpu address
>> +	// x1 is the return code, do not corrupt!
>> +	// x2 is the cpu context
> 
> this is confusing because the caller says x2 is free, so are these the
> inputs or invariants preserved in the function, or?
> 
> note that you'll avoid this kind of confusion by inlining this stuff in
> __guest_exit.

Indeed. I might just do that.

>> +	// x3 is a tmp register
>> +	// Guest's x0-x3 are on the stack
>> +
>> +	add	x2, x0, #VCPU_CONTEXT
>> +
>> +	// Compute base to save registers
> 
> misleading comment?

Of course. Isn't that the very purpose of a comment? I'm confused... ;-)

>> +	stp	x4, x5,   [x2, #CPU_XREG_OFFSET(4)]
>> +	stp	x6, x7,   [x2, #CPU_XREG_OFFSET(6)]
>> +	stp	x8, x9,   [x2, #CPU_XREG_OFFSET(8)]
>> +	stp	x10, x11, [x2, #CPU_XREG_OFFSET(10)]
>> +	stp	x12, x13, [x2, #CPU_XREG_OFFSET(12)]
>> +	stp	x14, x15, [x2, #CPU_XREG_OFFSET(14)]
>> +	stp	x16, x17, [x2, #CPU_XREG_OFFSET(16)]
>> +	str	x18,      [x2, #CPU_XREG_OFFSET(18)]
>> +
>> +	pop	x6, x7			// x2, x3
>> +	pop	x4, x5			// x0, x1
> 
> hard to review when I haven't seen the code that calls this, but I'll
> assume we store things in register order on the stack.

Indeed. I've basically lifted that code from the previous version, so
some things may have stuck...

>> +
>> +	stp	x4, x5, [x2, #CPU_XREG_OFFSET(0)]
>> +	stp	x6, x7, [x2, #CPU_XREG_OFFSET(2)]
>> +
>> +	save_common_regs x2
>> +.endm
>> +
>> +.macro restore_guest_regs
>> +	// Assume vcpu in x0, clobbers everything else
> 
> nit: clobbers everything (x0 gets nuked too)
> 
>> +
>> +	add	x2, x0, #VCPU_CONTEXT
>> +
>> +	// Prepare x0-x3 for later restore
>> +	ldp	x4, x5, [x2, #CPU_XREG_OFFSET(0)]
>> +	ldp	x6, x7, [x2, #CPU_XREG_OFFSET(2)]
>> +	push	x4, x5		// Push x0-x3 on the stack
>> +	push	x6, x7
> 
> why do you need x2 and x3 later? can't you just make do with x0 and x1
> and move the cpu context pointer to x1 ?

Maybe I can optimize this a bit indeed.

>> +
>> +	// x4-x18
>> +	ldp	x4, x5,   [x2, #CPU_XREG_OFFSET(4)] 
>> +	ldp	x6, x7,   [x2, #CPU_XREG_OFFSET(6)] 
>> +	ldp	x8, x9,   [x2, #CPU_XREG_OFFSET(8)] 
>> +	ldp	x10, x11, [x2, #CPU_XREG_OFFSET(10)]
>> +	ldp	x12, x13, [x2, #CPU_XREG_OFFSET(12)]
>> +	ldp	x14, x15, [x2, #CPU_XREG_OFFSET(14)]
>> +	ldp	x16, x17, [x2, #CPU_XREG_OFFSET(16)]
>> +	ldr	x18,      [x2, #CPU_XREG_OFFSET(18)]
>> +
>> +	// x19-x29, lr
>> +	restore_common_regs x2
>> +
>> +	// Last bits of the 64bit state
>> +	pop	x2, x3
>> +	pop	x0, x1
>> +
>> +	// Do not touch any register after this!
>> +.endm
>> +
>> +/*
>> + * u64 __guest_enter(struct kvm_vcpu *vcpu,
>> + *		     struct kvm_cpu_context *host_ctxt);
>> + */
>> +ENTRY(__guest_enter)
>> +	// x0: vcpu
>> +	// x1: host_ctxt
>> +	// x2, x3: parameter registers
>> +	// x4-x18: clobbered by macros
>> +
>> +	save_host_regs x1
>> +
>> +	// Preserve vcpu & host_ctxt for use at exit time
>> +	stp	x0, x1, [sp, #-16]!
> 
> why is this not a simple push?

This *is* a simple push. Catalin has been threatening me to remove
push/pop from the convenience macros, and I've started toying with the
idea. Maybe I'll should bite the bullet and convert them all.

> 
>> +
>> +	restore_guest_regs
> 
> do we ever reuse any of the above macros?  If not, perhaps it's more
> clear to simply inline them here?

Yeah, that's clearly better.

>> +	eret
>> +ENDPROC(__guest_enter)
>> +
>> +ENTRY(__guest_exit)
>> +	// x0: vcpu
>> +	// x1: return code
>> +	// x2-x3: free
>> +	// x4-x29,lr: vcpu regs
>> +	// vcpu x0-x3 on the stack
>> +	save_guest_regs
>> +
>> +	// Restore vcpu & host_ctxt from the stack
>> +	// (preserving return code in x1)
>> +	ldp	x0, x2, [sp], #16
> 
> why is this not a regular pop?

See above.

>> +	restore_host_regs x2
>> +
>> +	mov	x0, x1
>> +	ret
>> +ENDPROC(__guest_exit)
>> +
>> +	/* Insert fault handling here */
>> diff --git a/arch/arm64/kvm/hyp/hyp.h b/arch/arm64/kvm/hyp/hyp.h
>> index 2581232..7ac8e11 100644
>> --- a/arch/arm64/kvm/hyp/hyp.h
>> +++ b/arch/arm64/kvm/hyp/hyp.h
>> @@ -50,5 +50,7 @@ void __debug_restore_state(struct kvm_vcpu *vcpu,
>>  void __debug_cond_save_host_state(struct kvm_vcpu *vcpu);
>>  void __debug_cond_restore_host_state(struct kvm_vcpu *vcpu);
>>  
>> +u64 __guest_enter(struct kvm_vcpu *vcpu, struct kvm_cpu_context *host_ctxt);
>> +
>>  #endif /* __ARM64_KVM_HYP_H__ */
>>  
>> -- 
>> 2.1.4
>>
> 
> Thanks,
> -Christoffer
> 

Thanks,

	M.
-- 
Jazz is not dead. It just smells funny...

More information about the linux-arm-kernel mailing list