[kvmarm] [PATCH 08/29] arm64: KVM: user space interface

[mst@redhat.com (Michael S. Tsirkin)]

On Fri, Mar 08, 2013 at 04:46:15AM +0100, Marc Zyngier wrote:
> On Thu, 7 Mar 2013 10:09:03 +0200, "Michael S. Tsirkin" <mst@redhat.com>
> wrote:
> > On Tue, Mar 05, 2013 at 03:47:24AM +0000, Marc Zyngier wrote:
> >> Provide the kvm.h file that defines the user space visible
> >> interface.
> >> 
> >> Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
> >> ---
> >>  arch/arm64/include/uapi/asm/kvm.h | 112
> >>  ++++++++++++++++++++++++++++++++++++++
> >>  1 file changed, 112 insertions(+)
> >>  create mode 100644 arch/arm64/include/uapi/asm/kvm.h
> >> 
> >> diff --git a/arch/arm64/include/uapi/asm/kvm.h
> >> b/arch/arm64/include/uapi/asm/kvm.h
> >> new file mode 100644
> >> index 0000000..f5525f1
> >> --- /dev/null
> >> +++ b/arch/arm64/include/uapi/asm/kvm.h
> >> @@ -0,0 +1,112 @@
> >> +/*
> >> + * Copyright (C) 2012 - ARM Ltd
> >> + * Author: Marc Zyngier <marc.zyngier@arm.com>
> >> + *
> >> + * Derived from arch/arm/include/uapi/asm/kvm.h:
> >> + * Copyright (C) 2012 - Virtual Open Systems and Columbia University
> >> + * Author: Christoffer Dall <c.dall@virtualopensystems.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/>.
> >> + */
> >> +
> >> +#ifndef __ARM_KVM_H__
> >> +#define __ARM_KVM_H__
> >> +
> >> +#define KVM_SPSR_EL1	0
> >> +#define KVM_NR_SPSR	1
> >> +
> >> +#ifndef __ASSEMBLY__
> >> +#include <asm/types.h>
> >> +#include <asm/ptrace.h>
> >> +
> >> +#define __KVM_HAVE_GUEST_DEBUG
> >> +#define __KVM_HAVE_IRQ_LINE
> >> +
> >> +#define KVM_REG_SIZE(id)						\
> >> +	(1U << (((id) & KVM_REG_SIZE_MASK) >> KVM_REG_SIZE_SHIFT))
> >> +
> >> +struct kvm_regs {
> >> +	struct user_pt_regs regs;	/* sp = sp_el0 */
> >> +
> >> +	unsigned long	sp_el1;
> >> +	unsigned long	elr_el1;
> >> +
> >> +	unsigned long	spsr[KVM_NR_SPSR];
> >> +};
> >> +
> > 
> > Using long in uapi is generally a mistake: with gcc it has
> > different size depending on whether you build in 64 and 32 bit mode.
> > I think it is better to make it __u64.
> 
> Ah, well spotted. Definitely a mistake.
> 
> >> +/* Supported Processor Types */
> >> +#define KVM_ARM_TARGET_CORTEX_A57	0
> >> +#define KVM_ARM_NUM_TARGETS		1
> >> +
> >> +/* KVM_ARM_SET_DEVICE_ADDR ioctl id encoding */
> >> +#define KVM_ARM_DEVICE_TYPE_SHIFT	0
> >> +#define KVM_ARM_DEVICE_TYPE_MASK	(0xffff << KVM_ARM_DEVICE_TYPE_SHIFT)
> >> +#define KVM_ARM_DEVICE_ID_SHIFT		16
> >> +#define KVM_ARM_DEVICE_ID_MASK		(0xffff << KVM_ARM_DEVICE_ID_SHIFT)
> >> +
> >> +/* Supported device IDs */
> >> +#define KVM_ARM_DEVICE_VGIC_V2		0
> >> +
> >> +/* Supported VGIC address types  */
> >> +#define KVM_VGIC_V2_ADDR_TYPE_DIST	0
> >> +#define KVM_VGIC_V2_ADDR_TYPE_CPU	1
> >> +
> >> +#define KVM_VGIC_V2_DIST_SIZE		0x1000
> >> +#define KVM_VGIC_V2_CPU_SIZE		0x2000
> >> +
> >> +struct kvm_vcpu_init {
> >> +	__u32 target;
> >> +	__u32 features[7];
> >> +};
> >> +
> >> +struct kvm_sregs {
> >> +};
> >> +
> >> +struct kvm_fpu {
> >> +};
> >> +
> >> +struct kvm_guest_debug_arch {
> >> +};
> >> +
> >> +struct kvm_debug_exit_arch {
> >> +};
> >> +
> > 
> > This is a problem too: structure alignment is different
> > in -m32 versus -m64 modes, which will affect the offset of
> > the following fields. I think it's best to add a "padding"
> > field in there, and size it to multiple of 8 bytes.
> 
> Sorry, which field are you referring to?

Sorry, I was wrong here: while it seems to be compler dependent gcc
returns 0 as size of an empty structure.

> > I think the same is a good idea for other empty stuctures,
> > since otherwise the padding is implicit and
> > not initialized by gcc, and it is hard not too leak
> > info to userspace when you copy these structures out.
> > 
> > And, it'll be handy if you want to extend the structures
> > down the line.
> 
> Sure, what would make some sense.
> 
> Thanks,
> 
>          M.
> -- 
> Fast, cheap, reliable. Pick two.