[kvm-devel] [patch 1/5] KVM: add basic paravirt support (v2)

2008-02-20 Thread Marcelo Tosatti
Add basic KVM paravirt support. Avoid vm-exits on IO delays.

Add KVM_GET_PARA_FEATURES ioctl so paravirt features can be reported in a
single bitmask. This allows the host to disable features on runtime if 
appropriate, which would require one ioctl per feature otherwise.

The limit of 32 features can be extended to 64 if needed, beyond that a new 
MSR is required.

v1->v2:
- replace KVM_CAP_CLOCKSOURCE with KVM_CAP_PARA_FEATURES
- cover FEATURE_CLOCKSOURCE

Signed-off-by: Marcelo Tosatti <[EMAIL PROTECTED]>
Cc: Anthony Liguori <[EMAIL PROTECTED]>

Index: kvm.paravirt2/arch/x86/Kconfig
===
--- kvm.paravirt2.orig/arch/x86/Kconfig
+++ kvm.paravirt2/arch/x86/Kconfig
@@ -382,6 +382,14 @@ config KVM_CLOCK
  provides the guest with timing infrastructure such as time of day, and
  system time
 
+config KVM_GUEST
+   bool "KVM Guest support"
+   select PARAVIRT
+   depends on !(X86_VISWS || X86_VOYAGER)
+   help
+This option enables various optimizations for running under the KVM
+hypervisor.
+
 source "arch/x86/lguest/Kconfig"
 
 config PARAVIRT
Index: kvm.paravirt2/arch/x86/kernel/Makefile
===
--- kvm.paravirt2.orig/arch/x86/kernel/Makefile
+++ kvm.paravirt2/arch/x86/kernel/Makefile
@@ -69,6 +69,7 @@ obj-$(CONFIG_DEBUG_RODATA_TEST)   += test_
 obj-$(CONFIG_DEBUG_NX_TEST)+= test_nx.o
 
 obj-$(CONFIG_VMI)  += vmi_32.o vmiclock_32.o
+obj-$(CONFIG_KVM_GUEST)+= kvm.o
 obj-$(CONFIG_KVM_CLOCK)+= kvmclock.o
 obj-$(CONFIG_PARAVIRT) += paravirt.o paravirt_patch_$(BITS).o
 
Index: kvm.paravirt2/arch/x86/kernel/kvm.c
===
--- /dev/null
+++ kvm.paravirt2/arch/x86/kernel/kvm.c
@@ -0,0 +1,52 @@
+/*
+ * KVM paravirt_ops implementation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * 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, write to the Free Software
+ * Foundation, 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
+ *
+ * Copyright (C) 2007, Red Hat, Inc., Ingo Molnar <[EMAIL PROTECTED]>
+ * Copyright IBM Corporation, 2007
+ *   Authors: Anthony Liguori <[EMAIL PROTECTED]>
+ */
+
+#include 
+#include 
+#include 
+#include 
+#include 
+
+/*
+ * No need for any "IO delay" on KVM
+ */
+static void kvm_io_delay(void)
+{
+}
+
+static void paravirt_ops_setup(void)
+{
+   pv_info.name = "KVM";
+   pv_info.paravirt_enabled = 1;
+
+   if (kvm_para_has_feature(KVM_FEATURE_NOP_IO_DELAY))
+   pv_cpu_ops.io_delay = kvm_io_delay;
+
+}
+
+void __init kvm_guest_init(void)
+{
+   if (!kvm_para_available())
+   return;
+
+   paravirt_ops_setup();
+}
Index: kvm.paravirt2/arch/x86/kernel/setup_32.c
===
--- kvm.paravirt2.orig/arch/x86/kernel/setup_32.c
+++ kvm.paravirt2/arch/x86/kernel/setup_32.c
@@ -784,6 +784,7 @@ void __init setup_arch(char **cmdline_p)
 */
vmi_init();
 #endif
+   kvm_guest_init();
 
/*
 * NOTE: before this point _nobody_ is allowed to allocate
Index: kvm.paravirt2/arch/x86/kernel/setup_64.c
===
--- kvm.paravirt2.orig/arch/x86/kernel/setup_64.c
+++ kvm.paravirt2/arch/x86/kernel/setup_64.c
@@ -452,6 +452,8 @@ void __init setup_arch(char **cmdline_p)
init_apic_mappings();
ioapic_init_mappings();
 
+   kvm_guest_init();
+
/*
 * We trust e820 completely. No explicit ROM probing in memory.
 */
Index: kvm.paravirt2/arch/x86/kvm/x86.c
===
--- kvm.paravirt2.orig/arch/x86/kvm/x86.c
+++ kvm.paravirt2/arch/x86/kvm/x86.c
@@ -788,7 +788,7 @@ int kvm_dev_ioctl_check_extension(long e
case KVM_CAP_USER_MEMORY:
case KVM_CAP_SET_TSS_ADDR:
case KVM_CAP_EXT_CPUID:
-   case KVM_CAP_CLOCKSOURCE:
+   case KVM_CAP_PARA_FEATURES:
r = 1;
break;
case KVM_CAP_VAPIC:
@@ -854,6 +854,15 @@ long kvm_arch_dev_ioctl(struct file *fil
r = 0;
break;
}
+   case KVM_GET_PARA_FEATURES: {
+   __u32 para_features = KVM_PARA_FEATURES;
+
+   r = -EFAULT;
+   if (copy_to_user(argp, ΒΆ

Re: [kvm-devel] [patch 1/5] KVM: add basic paravirt support (v2)

2008-02-21 Thread Avi Kivity
Marcelo Tosatti wrote:
> Add basic KVM paravirt support. Avoid vm-exits on IO delays.
>
> Add KVM_GET_PARA_FEATURES ioctl so paravirt features can be reported in a
> single bitmask. This allows the host to disable features on runtime if 
> appropriate, which would require one ioctl per feature otherwise.
>
> The limit of 32 features can be extended to 64 if needed, beyond that a new 
> MSR is required.
>
> v1->v2:
> - replace KVM_CAP_CLOCKSOURCE with KVM_CAP_PARA_FEATURES
> - cover FEATURE_CLOCKSOURCE
>
>
>   

I don't understand the motivation for this.  A handful of ioctl()s at 
init time are hardly time consuming.

There is the advantage that paravirt kernel advances are reflected 
automatically without changes in userspace, but sometimes this is a 
disadvantage (it means there is no way to disable it, for instance).


-- 
error compiling committee.c: too many arguments to function


-
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/
___
kvm-devel mailing list
kvm-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/kvm-devel


Re: [kvm-devel] [patch 1/5] KVM: add basic paravirt support (v2)

2008-02-21 Thread Marcelo Tosatti
On Thu, Feb 21, 2008 at 05:38:30PM +0200, Avi Kivity wrote:
> Marcelo Tosatti wrote:
> >Add basic KVM paravirt support. Avoid vm-exits on IO delays.
> >
> >Add KVM_GET_PARA_FEATURES ioctl so paravirt features can be reported in a
> >single bitmask. This allows the host to disable features on runtime if 
> >appropriate, which would require one ioctl per feature otherwise.
> >
> >The limit of 32 features can be extended to 64 if needed, beyond that a 
> >new MSR is required.
> >
> >v1->v2:
> >- replace KVM_CAP_CLOCKSOURCE with KVM_CAP_PARA_FEATURES
> >- cover FEATURE_CLOCKSOURCE
> >
> >
> >  
> 
> I don't understand the motivation for this.  A handful of ioctl()s at 
> init time are hardly time consuming.
> 
> There is the advantage that paravirt kernel advances are reflected 
> automatically without changes in userspace, but sometimes this is a 
> disadvantage (it means there is no way to disable it, for instance).

One ioctl per feature sounded like unnecessary code duplication. I have
no problems with it though. So you prefer that way?

-
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/
___
kvm-devel mailing list
kvm-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/kvm-devel


Re: [kvm-devel] [patch 1/5] KVM: add basic paravirt support (v2)

2008-02-21 Thread Avi Kivity
Marcelo Tosatti wrote:
> One ioctl per feature sounded like unnecessary code duplication. I have
> no problems with it though. So you prefer that way?
>   

Yes.  I think we can avoid code duplication.

-- 
error compiling committee.c: too many arguments to function


-
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse012070mrt/direct/01/
___
kvm-devel mailing list
kvm-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/kvm-devel