WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-devel

Re: [Xen-devel] [PATCH] xen: implement vector callback for evtchn delive

To: Stefano Stabellini <Stefano.Stabellini@xxxxxxxxxxxxx>, Keir Fraser <keir.fraser@xxxxxxxxxxxxx>
Subject: Re: [Xen-devel] [PATCH] xen: implement vector callback for evtchn delivery
From: Boris Derzhavets <bderzhavets@xxxxxxxxx>
Date: Wed, 26 May 2010 02:58:50 -0700 (PDT)
Cc: "xen-devel@xxxxxxxxxxxxxxxxxxx" <xen-devel@xxxxxxxxxxxxxxxxxxx>
Delivery-date: Wed, 26 May 2010 03:02:18 -0700
Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=yahoo.com; s=s1024; t=1274867930; bh=La1bGmzPh+CLucHD+c3yfT939X8kUm09QWJ2bg6s0Xs=; h=Message-ID:X-YMail-OSG:Received:X-Mailer:Date:From:Subject:To:Cc:In-Reply-To:MIME-Version:Content-Type; b=NE00zYiDH2T96/EiPdsWZf71KhlpOz4nV6ly4c19b+g1bPOBuatzCAVu0DMfSXXm28/90VfxfIMnUgCYueDin5c6/wgguk7xyDBvOzJRvrAxRnPDrfRaAChVIXSh3zi/xMZFAwzdiCSXEInzG1keJ2tdXqhphSNt+m9LIVsGOzY=
Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:X-YMail-OSG:Received:X-Mailer:Date:From:Subject:To:Cc:In-Reply-To:MIME-Version:Content-Type; b=5VrZalY3/qxdtNqlyJ4cv7A8e1l/tw385TmlILEiphGjBrx06YOZHbQU27PgWILZfvExNsNJ/1wOrOnBny+LsGM2tGL4m+ET2PAB+NQTG6f7XBwBg8UiPbkhCJb6iLpt0I/h1DGEXhzp6vDAfwWxQ5MlKshVzo1POJICbxUc8Vg=;
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
In-reply-to: <C8228CDF.15CAA%keir.fraser@xxxxxxxxxxxxx>
List-help: <mailto:xen-devel-request@lists.xensource.com?subject=help>
List-id: Xen developer discussion <xen-devel.lists.xensource.com>
List-post: <mailto:xen-devel@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=unsubscribe>
Sender: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx
When i run rpmbuild on F13 , i need pv-on-hvm.patch in SOURCES for xen-4.0.0.1.7.f12.src.rpm.
That was a reason of my request yesterday.

Boris.


--- On Wed, 5/26/10, Keir Fraser <keir.fraser@xxxxxxxxxxxxx> wrote:

From: Keir Fraser <keir.fraser@xxxxxxxxxxxxx>
Subject: Re: [Xen-devel] [PATCH] xen: implement vector callback for evtchn delivery
To: "Boris Derzhavets" <bderzhavets@xxxxxxxxx>, "Stefano Stabellini" <Stefano.Stabellini@xxxxxxxxxxxxx>
Cc: "xen-devel@xxxxxxxxxxxxxxxxxxx" <xen-devel@xxxxxxxxxxxxxxxxxxx>
Date: Wednesday, May 26, 2010, 3:31 AM

The patch isn't for 4.0. You can apply it by hand anyway. The missing
hvm_safe_pvclock doesn't matter.

K.

On 26/05/2010 07:51, "Boris Derzhavets" <bderzhavets@xxxxxxxxx> wrote:

> Patch bellow contains :-
>
> --- a/xen/common/kernel.c    Tue May 25 09:08:34 2010 +0100
> +++ b/xen/common/kernel.c    Tue May 25 10:44:07 2010 +0100
> @@ -260,7 +260,8 @@
>                               (1U << XENFEAT_highmem_assist) |
>                               (1U << XENFEAT_gnttab_map_avail_bits);
>              else
> -                fi.submap |= (1U << XENFEAT_hvm_safe_pvclock);
> +                fi.submap |= (1U << XENFEAT_hvm_safe_pvclock) |
> +                             (1U << XENFEAT_hvm_callback_vector);
> #endif
>
> However,  file xen/common/kernel.c under xen-4.0.0 doesn't contain entry
>
>       fi.submap |= (1U << XENFEAT_hvm_safe_pvclock);
>
> Boris.
>
> --- On Tue, 5/25/10, Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>
> wrote:
>>
>> From: Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>
>> Subject: Re: [Xen-devel] [PATCH] xen: implement vector callback for evtchn
>> delivery
>> To: "Keir Fraser" <Keir.Fraser@xxxxxxxxxxxxx>
>> Cc: "xen-devel@xxxxxxxxxxxxxxxxxxx" <xen-devel@xxxxxxxxxxxxxxxxxxx>, "Stefano
>> Stabellini" <Stefano.Stabellini@xxxxxxxxxxxxx>
>> Date: Tuesday, May 25, 2010, 5:55 AM
>>
>> On Mon, 24 May 2010, Keir Fraser wrote:
>>> Please add documentation to include/public/hvm/param.h about how to specify
>>> the new callback method, and detect when it is available. Move the new
>>> is_hvm_pv_blah macros out into an appropriate include/asm-x86 header file --
>>> They are not arch independent.
>>
>> Done.
>>
>> Signed-off-by: Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx
>> </mc/compose?to=stefano.stabellini@xxxxxxxxxxxxx> >
>>
>> ---
>>
>>
>> diff -r 12c79a476007 xen/arch/x86/hvm/irq.c
>> --- a/xen/arch/x86/hvm/irq.c    Tue May 25 09:08:34 2010 +0100
>> +++ b/xen/arch/x86/hvm/irq.c    Tue May 25 10:44:07 2010 +0100
>> @@ -185,16 +185,16 @@
>> 
>>  void hvm_assert_evtchn_irq(struct vcpu *v)
>>  {
>> -    if ( v->vcpu_id != 0 )
>> -        return;
>> -
>>      if ( unlikely(in_irq() || !local_irq_is_enabled()) )
>>      {
>>          tasklet_schedule(&v->arch.hvm_vcpu.assert_evtchn_irq_tasklet);
>>          return;
>>      }
>> 
>> -    hvm_set_callback_irq_level(v);
>> +    if ( is_hvm_pv_evtchn_vcpu(v) )
>> +        vcpu_kick(v);
>> +    else if ( v->vcpu_id == 0 )
>> +        hvm_set_callback_irq_level(v);
>>  }
>> 
>>  void hvm_set_pci_link_route(struct domain *d, u8 link, u8 isa_irq)
>> @@ -251,7 +251,7 @@
>> 
>>      via_type = (uint8_t)(via >> 56) + 1;
>>      if ( ((via_type == HVMIRQ_callback_gsi) && (via == 0)) ||
>> -         (via_type > HVMIRQ_callback_pci_intx) )
>> +         (via_type > HVMIRQ_callback_vector) )
>>          via_type = HVMIRQ_callback_none;
>> 
>>      spin_lock(&d->arch.hvm_domain.irq_lock);
>> @@ -297,6 +297,9 @@
>>          if ( hvm_irq->callback_via_asserted )
>>               __hvm_pci_intx_assert(d, pdev, pintx);
>>          break;
>> +    case HVMIRQ_callback_vector:
>> +        hvm_irq->callback_via.vector = (uint8_t)via;
>> +        break;
>>      default:
>>          break;
>>      }
>> @@ -312,6 +315,10 @@
>>      case HVMIRQ_callback_pci_intx:
>>          printk("PCI INTx Dev 0x%02x Int%c\n", pdev, 'A' + pintx);
>>          break;
>> +    case HVMIRQ_callback_vector:
>> +        printk("Set HVMIRQ_callback_vector to %u\n",
>> +               hvm_irq->callback_via.vector);
>> +        break;
>>      default:
>>          printk("None\n");
>>          break;
>> @@ -323,6 +330,10 @@
>>      struct hvm_domain *plat = &v->domain->arch.hvm_domain;
>>      int vector;
>> 
>> +    if (plat->irq.callback_via_type == HVMIRQ_callback_vector &&
>> +            vcpu_info(v, evtchn_upcall_pending))
>> +        return hvm_intack_vector(plat->irq.callback_via.vector);
>> +
>>      if ( unlikely(v->nmi_pending) )
>>          return hvm_intack_nmi;
>> 
>> @@ -363,6 +374,8 @@
>>      case hvm_intsrc_lapic:
>>          if ( !vlapic_ack_pending_irq(v, intack.vector) )
>>              intack = hvm_intack_none;
>> +        break;
>> +    case hvm_intsrc_vector:
>>          break;
>>      default:
>>          intack = hvm_intack_none;
>> diff -r 12c79a476007 xen/arch/x86/hvm/vmx/intr.c
>> --- a/xen/arch/x86/hvm/vmx/intr.c    Tue May 25 09:08:34 2010 +0100
>> +++ b/xen/arch/x86/hvm/vmx/intr.c    Tue May 25 10:44:07 2010 +0100
>> @@ -164,7 +164,8 @@
>>      {
>>          HVMTRACE_2D(INJ_VIRQ, intack.vector, /*fake=*/ 0);
>>          vmx_inject_extint(intack.vector);
>> -        pt_intr_post(v, intack);
>> +        if (intack.source != hvm_intsrc_vector)
>> +             pt_intr_post(v, intack);
>>      }
>> 
>>      /* Is there another IRQ to queue up behind this one? */
>> diff -r 12c79a476007 xen/common/kernel.c
>> --- a/xen/common/kernel.c    Tue May 25 09:08:34 2010 +0100
>> +++ b/xen/common/kernel.c    Tue May 25 10:44:07 2010 +0100
>> @@ -260,7 +260,8 @@
>>                               (1U << XENFEAT_highmem_assist) |
>>                               (1U << XENFEAT_gnttab_map_avail_bits);
>>              else
>> -                fi.submap |= (1U << XENFEAT_hvm_safe_pvclock);
>> +                fi.submap |= (1U << XENFEAT_hvm_safe_pvclock) |
>> +                             (1U << XENFEAT_hvm_callback_vector);
>>  #endif
>>              break;
>>          default:
>> diff -r 12c79a476007 xen/include/asm-x86/domain.h
>> --- a/xen/include/asm-x86/domain.h    Tue May 25 09:08:34 2010 +0100
>> +++ b/xen/include/asm-x86/domain.h    Tue May 25 10:44:07 2010 +0100
>> @@ -18,6 +18,10 @@
>>  #define is_pv_32on64_domain(d) (0)
>>  #endif
>>  #define is_pv_32on64_vcpu(v)   (is_pv_32on64_domain((v)->domain))
>> +
>> +#define is_hvm_pv_evtchn_domain(d) (is_hvm_domain(d) && \
>> +        d->arch.hvm_domain.irq.callback_via_type == HVMIRQ_callback_vector)
>> +#define is_hvm_pv_evtchn_vcpu(v) (is_hvm_pv_evtchn_domain(v->domain))
>> 
>>  #define VCPU_TRAP_NMI          1
>>  #define VCPU_TRAP_MCE          2
>> diff -r 12c79a476007 xen/include/asm-x86/hvm/hvm.h
>> --- a/xen/include/asm-x86/hvm/hvm.h    Tue May 25 09:08:34 2010 +0100
>> +++ b/xen/include/asm-x86/hvm/hvm.h    Tue May 25 10:44:07 2010 +0100
>> @@ -33,7 +33,8 @@
>>      hvm_intsrc_pic,
>>      hvm_intsrc_lapic,
>>      hvm_intsrc_nmi,
>> -    hvm_intsrc_mce
>> +    hvm_intsrc_mce,
>> +    hvm_intsrc_vector
>>  };
>>  struct hvm_intack {
>>      uint8_t source; /* enum hvm_intsrc */
>> @@ -44,6 +45,7 @@
>>  #define hvm_intack_lapic(vec) ( (struct hvm_intack) { hvm_intsrc_lapic, vec
>> } )
>>  #define hvm_intack_nmi        ( (struct hvm_intack) { hvm_intsrc_nmi,   2 }
>> )
>>  #define hvm_intack_mce        ( (struct hvm_intack) { hvm_intsrc_mce,   18 }
>> )
>> +#define hvm_intack_vector(vec)( (struct hvm_intack) { hvm_intsrc_vector, vec
>> } )
>>  enum hvm_intblk {
>>      hvm_intblk_none,      /* not blocked (deliverable) */
>>      hvm_intblk_shadow,    /* MOV-SS or STI shadow */
>> diff -r 12c79a476007 xen/include/asm-x86/hvm/irq.h
>> --- a/xen/include/asm-x86/hvm/irq.h    Tue May 25 09:08:34 2010 +0100
>> +++ b/xen/include/asm-x86/hvm/irq.h    Tue May 25 10:44:07 2010 +0100
>> @@ -54,12 +54,14 @@
>>          enum {
>>              HVMIRQ_callback_none,
>>              HVMIRQ_callback_gsi,
>> -            HVMIRQ_callback_pci_intx
>> +            HVMIRQ_callback_pci_intx,
>> +            HVMIRQ_callback_vector
>>          } callback_via_type;
>>      };
>>      union {
>>          uint32_t gsi;
>>          struct { uint8_t dev, intx; } pci;
>> +        uint32_t vector;
>>      } callback_via;
>> 
>>      /* Number of INTx wires asserting each PCI-ISA link. */
>> diff -r 12c79a476007 xen/include/public/features.h
>> --- a/xen/include/public/features.h    Tue May 25 09:08:34 2010 +0100
>> +++ b/xen/include/public/features.h    Tue May 25 10:44:07 2010 +0100
>> @@ -68,6 +68,9 @@
>>   */
>>  #define XENFEAT_gnttab_map_avail_bits      7
>> 
>> +/* x86: Does this Xen host support the HVM callback vector type? */
>> +#define XENFEAT_hvm_callback_vector        8
>> +
>>  /* x86: pvclock algorithm is safe to use on HVM */
>>  #define XENFEAT_hvm_safe_pvclock           9
>> 
>> diff -r 12c79a476007 xen/include/public/hvm/params.h
>> --- a/xen/include/public/hvm/params.h    Tue May 25 09:08:34 2010 +0100
>> +++ b/xen/include/public/hvm/params.h    Tue May 25 10:44:07 2010 +0100
>> @@ -33,6 +33,9 @@
>>   * val[63:56] == 1: val[55:0] is a delivery PCI INTx line, as follows:
>>   *                  Domain = val[47:32], Bus  = val[31:16],
>>   *                  DevFn  = val[15: 8], IntX = val[ 1: 0]
>> + * val[63:56] == 2: val[7:0] is a vector number, check for
>> + *                  XENFEAT_hvm_callback_vector to know if this delivery
>> + *                  method is available.
>>   * If val == 0 then CPU0 event-channel notifications are not delivered.
>>   */
>>  #define HVM_PARAM_CALLBACK_IRQ 0
>>
>> _______________________________________________
>> Xen-devel mailing list
>> Xen-devel@xxxxxxxxxxxxxxxxxxx </mc/compose?to=Xen-devel@xxxxxxxxxxxxxxxxxxx>
>> http://lists.xensource.com/xen-devel
>




_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
<Prev in Thread] Current Thread [Next in Thread>