Re: [RFC PATCH v2 3/3] arch/x86/acrn: add hypercall for acrn_guest

[Zhao, Yakui]

On 2019å04æ06æ 03:19, Thomas Gleixner wrote:
> On Tue, 26 Mar 2019, Zhao Yakui wrote:
>
> > When acrn_hypervisor is detected, the hypercall is needed so that the
> > acrn guest can query/config some settings. For example: it can be used
> > to query the resources in hypervisor and manage the CPU/memory/device/
> > interrupt for Guest system.
> >
> > So the hypercall is added so that the kernel can communicate with the
> > low-level acrn-hypervisor. On x86 it is implemented by using vmacll when
>
> is implemented with the VMCALL instruction
>
> > the acrn hypervisor is enabled.
> >
> > +++ b/arch/x86/include/asm/acrn_hypercall.h
> > @@ -0,0 +1,84 @@
> > +/* SPDX-License-Identifier: GPL-2.0 */
> > +/*
> > + * acrn_hypercall.h : acrn hypervisor call API
>
> No file names in headers please. They are pointless and get out of sync
> when files are renamed.

Sure. It will be removed in next vesion.

> > + */
> > +
> > +#ifndef __ACRN_HYPERCALL_H__
> > +#define __ACRN_HYPERCALL_H__
>
> asm headers start with
>
>      __ASM_X86_....
This will be fixed.

> > +
> > +#include <linux/errno.h>
> > +
> > +#ifdef CONFIG_ACRN_GUEST
> > +
> > +/*
> > + * Hypercalls for ACRN Guest
> > + *
> > + * Hypercall number is passed in r8 register.
> > + * Return value will be placed in rax.
> > + * Up to 2 arguments are passed in rdi, rsi.
> > + */
> > +
> > +static inline long acrn_hypercall0(unsigned long hcall_id)
> > +{
> > +
>
> Remove the empty new line please.
>
> > +	register long result asm("rax");
> > +	register unsigned long r8 asm("r8") = hcall_id;
>
> Please order them the other way around, like a reverse christmas tree:
>
> 	register unsigned long r8 asm("r8") = hcall_id;
> 	register long result asm("rax");
>
> That's way simpler to read.

This will be fixed.

> > +	asm volatile(".byte 0x0F,0x01,0xC1\n"
> > +			: "=r"(result)
> > +			:  "r"(r8));
>
> Please mention in the changelog why this is implemented with bytes and not
> with the proper mnemonic. I assume it depends on binutils, so please
> document which version of binutils supports the mnemonic.

I check the binutils version mentioned in Document/changes.
(binutils, 2.20)
It seems that the "vmcall" is already supported.
So the "vmcall" mnemonic will be used to make it readable.

> And in the first function, i.e. hypercall0, add a comment above the asm
> volatile() to that effect as well.

Sure.


> Thanks,
>
> 	tglx