All of lore.kernel.org
 help / color / mirror / Atom feed
From: Jes Sorensen <jes@sgi.com>
To: "Zhang, Xiantao" <xiantao.zhang@intel.com>
Cc: Avi Kivity <avi@qumranet.com>, "Luck, Tony" <tony.luck@intel.com>,
	"Xu, Anthony" <anthony.xu@intel.com>,
	Akio Takebe <takebe_akio@jp.fujitsu.com>,
	kvm-devel@lists.sourceforge.net,
	kvm-ia64-devel@lists.sourceforge.net, linux-ia64@vger.kernel.org,
	virtualization@lists.linux-foundation.org,
	Carsten Otte <cotte@de.ibm.com>
Subject: Re: [05/17][PATCH] kvm/ia64 : Add head files for kvm/ia64
Date: Mon, 31 Mar 2008 12:33:47 +0000	[thread overview]
Message-ID: <47F0DA2B.9070403@sgi.com> (raw)
In-Reply-To: <42DFA526FC41B1429CE7279EF83C6BDC01048246@pdsmsx415.ccr.corp.intel.com>

Hi Xiantao,

More comments.

Zhang, Xiantao wrote:
>>From 696b9eea9f5001a7b7a07c0e58514aa10306b91a Mon Sep 17 00:00:00 2001
> From: Xiantao Zhang <xiantao.zhang@intel.com>
> Date: Fri, 28 Mar 2008 09:51:36 +0800
> Subject: [PATCH] KVM:IA64 : Add head files for kvm/ia64
> 
> ia64_regs: some defintions for special registers
> which aren't defined in asm-ia64/ia64regs.

Please put missing definitions of registers into asm-ia64/ia64regs.h if
they are official definitions from the spec.

> kvm_minstate.h : Marcos about Min save routines.
> lapic.h: apic structure definition.
> vcpu.h : routions related to vcpu virtualization.
> vti.h  : Some macros or routines for VT support on Itanium.
> Signed-off-by: Xiantao Zhang <xiantao.zhang@intel.com>

> +/*
> + * Flushrs instruction stream.
> + */
> +#define ia64_flushrs() asm volatile ("flushrs;;":::"memory")
> +
> +#define ia64_loadrs() asm volatile ("loadrs;;":::"memory")

Please put these into include/asm-ia64/gcc_intrin.h

> +#define ia64_get_rsc()
> \
> +({
> \
> +	unsigned long val;
> \
> +	asm volatile ("mov %0=ar.rsc;;" : "=r"(val) :: "memory");
> \
> +	val;
> \
> +})
> +
> +#define ia64_set_rsc(val)					\
> +	asm volatile ("mov ar.rsc=%0;;" :: "r"(val) : "memory")

Please update the ia64_get/set_reg macros to handle the RSC register
and use those macros.

> +#define ia64_get_bspstore()
> \
> +({
> \
> +	unsigned long val;
> \
> +	asm volatile ("mov %0=ar.bspstore;;" : "=r"(val) :: "memory");
> \
> +	val;
> \
> +})

Ditto for for AR.BSPSTORE

> +#define ia64_get_rnat()
> \
> +({
> \
> +	unsigned long val;
> \
> +	asm volatile ("mov %0=ar.rnat;" : "=r"(val) :: "memory");
> \
> +	val;
> \
> +})

Ditto for AR.RNAT

> +static inline unsigned long ia64_get_itc(void)
> +{
> +	unsigned long result;
> +	result = ia64_getreg(_IA64_REG_AR_ITC);
> +	return result;
> +}

This exists in include/asm-ia64/delay.h

> +static inline void ia64_set_dcr(unsigned long dcr)
> +{
> +	ia64_setreg(_IA64_REG_CR_DCR, dcr);
> +}

Please just call ia64_setreg() in your code rather than defining a
wrapper for it.

> +#define ia64_ttag(addr)
> \
> +({
> \
> +	__u64 ia64_intri_res;
> \
> +	asm volatile ("ttag %0=%1" : "=r"(ia64_intri_res) : "r" (addr));
> \
> +	ia64_intri_res;
> \
> +})

Please add to include/asm-ia64/gcc_intrin.h instead.

> diff --git a/arch/ia64/kvm/lapic.h b/arch/ia64/kvm/lapic.h
> new file mode 100644
> index 0000000..152cbdc
> --- /dev/null
> +++ b/arch/ia64/kvm/lapic.h
> @@ -0,0 +1,27 @@
> +#ifndef __KVM_IA64_LAPIC_H
> +#define __KVM_IA64_LAPIC_H
> +
> +#include "iodev.h"

I don't understand why iodev.h is included here?

> --- /dev/null
> +++ b/arch/ia64/kvm/vcpu.h

The formatting of this file is dodgy, please try and make it comply with
the Linux standards in Documentation/CodingStyle

> +#define _vmm_raw_spin_lock(x)
> \
[snip]
> +
> +#define _vmm_raw_spin_unlock(x)				\

Could you explain the reasoning behind these two macros? Whenever I see
open coded spin lock modifications like these, I have to admit I get a
bit worried.

> +typedef struct kvm_vcpu VCPU;
> +typedef struct kvm_pt_regs REGS;
> +typedef enum { DATA_REF, NA_REF, INST_REF, RSE_REF } vhpt_ref_t;
> +typedef enum { INSTRUCTION, DATA, REGISTER } miss_type;

ARGH! Please see previous mail about typedefs! I suspect this is code
inherited from Xen ? Xen has a lot of really nasty and pointless
typedefs like these :-(

> +static inline void vcpu_set_dbr(VCPU *vcpu, u64 reg, u64 val)
> +{
> +	/* TODO: need to virtualize */
> +	__ia64_set_dbr(reg, val);
> +}
> +
> +static inline void vcpu_set_ibr(VCPU *vcpu, u64 reg, u64 val)
> +{
> +	/* TODO: need to virtualize */
> +	ia64_set_ibr(reg, val);
> +}
> +
> +static inline u64 vcpu_get_dbr(VCPU *vcpu, u64 reg)
> +{
> +	/* TODO: need to virtualize */
> +	return ((u64)__ia64_get_dbr(reg));
> +}
> +
> +static inline u64 vcpu_get_ibr(VCPU *vcpu, u64 reg)
> +{
> +	/* TODO: need to virtualize */
> +	return ((u64)ia64_get_ibr(reg));
> +}

More wrapper macros that really should just use ia64_get/set_reg()
directly in the code.


> diff --git a/arch/ia64/kvm/vti.h b/arch/ia64/kvm/vti.h
> new file mode 100644
> index 0000000..591ab22
[ship]
> +/* -*-  Mode:C; c-basic-offset:4; tab-width:4; indent-tabs-mode:nil -*-
> */

Evil formatting again!

Cheers,
Jes


WARNING: multiple messages have this Message-ID (diff)
From: Jes Sorensen <jes@sgi.com>
To: "Zhang, Xiantao" <xiantao.zhang@intel.com>
Cc: Avi Kivity <avi@qumranet.com>, "Luck, Tony" <tony.luck@intel.com>,
	"Xu, Anthony" <anthony.xu@intel.com>,
	Akio Takebe <takebe_akio@jp.fujitsu.com>,
	kvm-devel@lists.sourceforge.net,
	kvm-ia64-devel@lists.sourceforge.net, linux-ia64@vger.kernel.org,
	virtualization@lists.linux-foundation.org,
	Carsten Otte <cotte@de.ibm.com>
Subject: Re: [05/17][PATCH] kvm/ia64 : Add head files for kvm/ia64
Date: Mon, 31 Mar 2008 14:33:47 +0200	[thread overview]
Message-ID: <47F0DA2B.9070403@sgi.com> (raw)
In-Reply-To: <42DFA526FC41B1429CE7279EF83C6BDC01048246@pdsmsx415.ccr.corp.intel.com>

Hi Xiantao,

More comments.

Zhang, Xiantao wrote:
>>From 696b9eea9f5001a7b7a07c0e58514aa10306b91a Mon Sep 17 00:00:00 2001
> From: Xiantao Zhang <xiantao.zhang@intel.com>
> Date: Fri, 28 Mar 2008 09:51:36 +0800
> Subject: [PATCH] KVM:IA64 : Add head files for kvm/ia64
> 
> ia64_regs: some defintions for special registers
> which aren't defined in asm-ia64/ia64regs.

Please put missing definitions of registers into asm-ia64/ia64regs.h if
they are official definitions from the spec.

> kvm_minstate.h : Marcos about Min save routines.
> lapic.h: apic structure definition.
> vcpu.h : routions related to vcpu virtualization.
> vti.h  : Some macros or routines for VT support on Itanium.
> Signed-off-by: Xiantao Zhang <xiantao.zhang@intel.com>

> +/*
> + * Flushrs instruction stream.
> + */
> +#define ia64_flushrs() asm volatile ("flushrs;;":::"memory")
> +
> +#define ia64_loadrs() asm volatile ("loadrs;;":::"memory")

Please put these into include/asm-ia64/gcc_intrin.h

> +#define ia64_get_rsc()
> \
> +({
> \
> +	unsigned long val;
> \
> +	asm volatile ("mov %0=ar.rsc;;" : "=r"(val) :: "memory");
> \
> +	val;
> \
> +})
> +
> +#define ia64_set_rsc(val)					\
> +	asm volatile ("mov ar.rsc=%0;;" :: "r"(val) : "memory")

Please update the ia64_get/set_reg macros to handle the RSC register
and use those macros.

> +#define ia64_get_bspstore()
> \
> +({
> \
> +	unsigned long val;
> \
> +	asm volatile ("mov %0=ar.bspstore;;" : "=r"(val) :: "memory");
> \
> +	val;
> \
> +})

Ditto for for AR.BSPSTORE

> +#define ia64_get_rnat()
> \
> +({
> \
> +	unsigned long val;
> \
> +	asm volatile ("mov %0=ar.rnat;" : "=r"(val) :: "memory");
> \
> +	val;
> \
> +})

Ditto for AR.RNAT

> +static inline unsigned long ia64_get_itc(void)
> +{
> +	unsigned long result;
> +	result = ia64_getreg(_IA64_REG_AR_ITC);
> +	return result;
> +}

This exists in include/asm-ia64/delay.h

> +static inline void ia64_set_dcr(unsigned long dcr)
> +{
> +	ia64_setreg(_IA64_REG_CR_DCR, dcr);
> +}

Please just call ia64_setreg() in your code rather than defining a
wrapper for it.

> +#define ia64_ttag(addr)
> \
> +({
> \
> +	__u64 ia64_intri_res;
> \
> +	asm volatile ("ttag %0=%1" : "=r"(ia64_intri_res) : "r" (addr));
> \
> +	ia64_intri_res;
> \
> +})

Please add to include/asm-ia64/gcc_intrin.h instead.

> diff --git a/arch/ia64/kvm/lapic.h b/arch/ia64/kvm/lapic.h
> new file mode 100644
> index 0000000..152cbdc
> --- /dev/null
> +++ b/arch/ia64/kvm/lapic.h
> @@ -0,0 +1,27 @@
> +#ifndef __KVM_IA64_LAPIC_H
> +#define __KVM_IA64_LAPIC_H
> +
> +#include "iodev.h"

I don't understand why iodev.h is included here?

> --- /dev/null
> +++ b/arch/ia64/kvm/vcpu.h

The formatting of this file is dodgy, please try and make it comply with
the Linux standards in Documentation/CodingStyle

> +#define _vmm_raw_spin_lock(x)
> \
[snip]
> +
> +#define _vmm_raw_spin_unlock(x)				\

Could you explain the reasoning behind these two macros? Whenever I see
open coded spin lock modifications like these, I have to admit I get a
bit worried.

> +typedef struct kvm_vcpu VCPU;
> +typedef struct kvm_pt_regs REGS;
> +typedef enum { DATA_REF, NA_REF, INST_REF, RSE_REF } vhpt_ref_t;
> +typedef enum { INSTRUCTION, DATA, REGISTER } miss_type;

ARGH! Please see previous mail about typedefs! I suspect this is code
inherited from Xen ? Xen has a lot of really nasty and pointless
typedefs like these :-(

> +static inline void vcpu_set_dbr(VCPU *vcpu, u64 reg, u64 val)
> +{
> +	/* TODO: need to virtualize */
> +	__ia64_set_dbr(reg, val);
> +}
> +
> +static inline void vcpu_set_ibr(VCPU *vcpu, u64 reg, u64 val)
> +{
> +	/* TODO: need to virtualize */
> +	ia64_set_ibr(reg, val);
> +}
> +
> +static inline u64 vcpu_get_dbr(VCPU *vcpu, u64 reg)
> +{
> +	/* TODO: need to virtualize */
> +	return ((u64)__ia64_get_dbr(reg));
> +}
> +
> +static inline u64 vcpu_get_ibr(VCPU *vcpu, u64 reg)
> +{
> +	/* TODO: need to virtualize */
> +	return ((u64)ia64_get_ibr(reg));
> +}

More wrapper macros that really should just use ia64_get/set_reg()
directly in the code.


> diff --git a/arch/ia64/kvm/vti.h b/arch/ia64/kvm/vti.h
> new file mode 100644
> index 0000000..591ab22
[ship]
> +/* -*-  Mode:C; c-basic-offset:4; tab-width:4; indent-tabs-mode:nil -*-
> */

Evil formatting again!

Cheers,
Jes


  reply	other threads:[~2008-03-31 12:33 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-03-31  8:26 [05/17][PATCH] kvm/ia64 : Add head files for kvm/ia64 Zhang, Xiantao
2008-03-31 12:33 ` Jes Sorensen [this message]
2008-03-31 12:33   ` Jes Sorensen
2008-04-01  1:15   ` Zhang, Xiantao
2008-04-01  1:15     ` Zhang, Xiantao
2008-03-31 12:33 ` Jes Sorensen
2008-03-31 15:34 ` Carsten Otte
2008-03-31 15:34 ` Carsten Otte
2008-03-31 15:34   ` Carsten Otte
  -- strict thread matches above, loose matches on Subject: below --
2008-03-28  9:59 Zhang, Xiantao

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=47F0DA2B.9070403@sgi.com \
    --to=jes@sgi.com \
    --cc=anthony.xu@intel.com \
    --cc=avi@qumranet.com \
    --cc=cotte@de.ibm.com \
    --cc=kvm-devel@lists.sourceforge.net \
    --cc=kvm-ia64-devel@lists.sourceforge.net \
    --cc=linux-ia64@vger.kernel.org \
    --cc=takebe_akio@jp.fujitsu.com \
    --cc=tony.luck@intel.com \
    --cc=virtualization@lists.linux-foundation.org \
    --cc=xiantao.zhang@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.