[PATCH v2] xen: Improve x86's alloc_vcpu_guest_context()

[PATCH v2] xen: Improve x86's alloc_vcpu_guest_context()

[Andrew Cooper]

This essentially reverts c/s 2037f2adb "x86: introduce
alloc_vcpu_guest_context()", including the newer arm bits, but achieves
the same end goal by using the newer vmalloc() infrastructure.

For both x86 and ARM, {alloc,free}_vcpu_guest_context() become arch-local
static inlines (which avoids a call into a separate translation),
and removes an x86 scalability limit when compiling with a large NR_CPUS.

Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
---
CC: Jan Beulich <JBeulich@suse.com>
CC: Tim Deegan <tim@xen.org>
CC: Ian Campbell <ian.campbell@citrix.com>
CC: Stefano Stabellini <stefano.stabellini@citrix.com>
CC: Julien Grall <julien.grall@citrix.com>
CC: Roger Pau Monne <roger.pau@citrix.com>

v2:
 * Retain the use of x{malloc,free}() on ARM

Build tested on all architectures, functionally tested on x86.
---
 xen/arch/arm/domain.c        |   11 ----------
 xen/arch/x86/domain.c        |   46 ------------------------------------------
 xen/include/asm-arm/domain.h |   10 +++++++++
 xen/include/asm-x86/domain.h |   10 +++++++++
 xen/include/asm-x86/fixmap.h |    3 ---
 xen/include/xen/domain.h     |    6 ------
 6 files changed, 20 insertions(+), 66 deletions(-)

diff --git a/xen/arch/arm/domain.c b/xen/arch/arm/domain.c
index 5bdc2e9..575745c 100644
--- a/xen/arch/arm/domain.c
+++ b/xen/arch/arm/domain.c
@@ -466,17 +466,6 @@ void free_vcpu_struct(struct vcpu *v)
     free_xenheap_page(v);
 }
 
-struct vcpu_guest_context *alloc_vcpu_guest_context(void)
-{
-    return xmalloc(struct vcpu_guest_context);
-
-}
-
-void free_vcpu_guest_context(struct vcpu_guest_context *vgc)
-{
-    xfree(vgc);
-}
-
 int vcpu_initialise(struct vcpu *v)
 {
     int rc = 0;
diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c
index 5ff09e7..0b97912 100644
--- a/xen/arch/x86/domain.c
+++ b/xen/arch/x86/domain.c
@@ -48,7 +48,6 @@
 #include <asm/cpuidle.h>
 #include <asm/mpspec.h>
 #include <asm/ldt.h>
-#include <asm/fixmap.h>
 #include <asm/hvm/hvm.h>
 #include <asm/hvm/support.h>
 #include <asm/hvm/viridian.h>
@@ -272,51 +271,6 @@ void free_vcpu_struct(struct vcpu *v)
     free_xenheap_page(v);
 }
 
-static DEFINE_PER_CPU(struct page_info *[
-    PFN_UP(sizeof(struct vcpu_guest_context))], vgc_pages);
-
-struct vcpu_guest_context *alloc_vcpu_guest_context(void)
-{
-    unsigned int i, cpu = smp_processor_id();
-    enum fixed_addresses idx = FIX_VGC_BEGIN -
-        cpu * PFN_UP(sizeof(struct vcpu_guest_context));
-
-    BUG_ON(per_cpu(vgc_pages[0], cpu) != NULL);
-
-    for ( i = 0; i < PFN_UP(sizeof(struct vcpu_guest_context)); ++i )
-    {
-        struct page_info *pg = alloc_domheap_page(current->domain,
-                                                  MEMF_no_owner);
-
-        if ( unlikely(pg == NULL) )
-        {
-            free_vcpu_guest_context(NULL);
-            return NULL;
-        }
-        __set_fixmap(idx - i, page_to_mfn(pg), __PAGE_HYPERVISOR_RW);
-        per_cpu(vgc_pages[i], cpu) = pg;
-    }
-    return (void *)fix_to_virt(idx);
-}
-
-void free_vcpu_guest_context(struct vcpu_guest_context *vgc)
-{
-    unsigned int i, cpu = smp_processor_id();
-    enum fixed_addresses idx = FIX_VGC_BEGIN -
-        cpu * PFN_UP(sizeof(struct vcpu_guest_context));
-
-    BUG_ON(vgc && vgc != (void *)fix_to_virt(idx));
-
-    for ( i = 0; i < PFN_UP(sizeof(struct vcpu_guest_context)); ++i )
-    {
-        if ( !per_cpu(vgc_pages[i], cpu) )
-            continue;
-        clear_fixmap(idx - i);
-        free_domheap_page(per_cpu(vgc_pages[i], cpu));
-        per_cpu(vgc_pages[i], cpu) = NULL;
-    }
-}
-
 static int setup_compat_l4(struct vcpu *v)
 {
     struct page_info *pg;
diff --git a/xen/include/asm-arm/domain.h b/xen/include/asm-arm/domain.h
index 7ddaeaa..c3f5a95 100644
--- a/xen/include/asm-arm/domain.h
+++ b/xen/include/asm-arm/domain.h
@@ -300,6 +300,16 @@ static inline register_t vcpuid_to_vaffinity(unsigned int vcpuid)
     return vaff;
 }
 
+static inline struct vcpu_guest_context *alloc_vcpu_guest_context(void)
+{
+    return xmalloc(struct vcpu_guest_context);
+}
+
+static inline void free_vcpu_guest_context(struct vcpu_guest_context *vgc)
+{
+    xfree(vgc);
+}
+
 #endif /* __ASM_DOMAIN_H__ */
 
 /*
diff --git a/xen/include/asm-x86/domain.h b/xen/include/asm-x86/domain.h
index 59cf826..f0aeade 100644
--- a/xen/include/asm-x86/domain.h
+++ b/xen/include/asm-x86/domain.h
@@ -576,6 +576,16 @@ void domain_cpuid(struct domain *d,
 
 #define domain_max_vcpus(d) (is_hvm_domain(d) ? HVM_MAX_VCPUS : MAX_VIRT_CPUS)
 
+static inline struct vcpu_guest_context *alloc_vcpu_guest_context(void)
+{
+    return vmalloc(sizeof(struct vcpu_guest_context));
+}
+
+static inline void free_vcpu_guest_context(struct vcpu_guest_context *vgc)
+{
+    vfree(vgc);
+}
+
 #endif /* __ASM_DOMAIN_H__ */
 
 /*
diff --git a/xen/include/asm-x86/fixmap.h b/xen/include/asm-x86/fixmap.h
index 70eadff..1e24b11 100644
--- a/xen/include/asm-x86/fixmap.h
+++ b/xen/include/asm-x86/fixmap.h
@@ -47,9 +47,6 @@ enum fixed_addresses {
     FIX_COM_END,
     FIX_EHCI_DBGP,
     /* Everything else should go further down. */
-    FIX_VGC_END,
-    FIX_VGC_BEGIN = FIX_VGC_END
-      + PFN_UP(sizeof(struct vcpu_guest_context)) * NR_CPUS - 1,
     FIX_APIC_BASE,
     FIX_IO_APIC_BASE_0,
     FIX_IO_APIC_BASE_END = FIX_IO_APIC_BASE_0 + MAX_IO_APICS-1,
diff --git a/xen/include/xen/domain.h b/xen/include/xen/domain.h
index a469fe0..3dca3f6 100644
--- a/xen/include/xen/domain.h
+++ b/xen/include/xen/domain.h
@@ -31,12 +31,6 @@ struct vcpu *alloc_vcpu(
 struct vcpu *alloc_vcpu_struct(void);
 void free_vcpu_struct(struct vcpu *v);
 
-/* Allocate/free a vcpu_guest_context structure. */
-#ifndef alloc_vcpu_guest_context
-struct vcpu_guest_context *alloc_vcpu_guest_context(void);
-void free_vcpu_guest_context(struct vcpu_guest_context *);
-#endif
-
 /* Allocate/free a PIRQ structure. */
 #ifndef alloc_pirq_struct
 struct pirq *alloc_pirq_struct(struct domain *);
-- 
1.7.10.4

Re: [PATCH v2] xen: Improve x86's alloc_vcpu_guest_context()

[Julien Grall]

Hi Andrew,

On 17/09/15 17:52, Andrew Cooper wrote:
> This essentially reverts c/s 2037f2adb "x86: introduce
> alloc_vcpu_guest_context()", including the newer arm bits, but achieves
> the same end goal by using the newer vmalloc() infrastructure.
> 
> For both x86 and ARM, {alloc,free}_vcpu_guest_context() become arch-local
> static inlines (which avoids a call into a separate translation),
> and removes an x86 scalability limit when compiling with a large NR_CPUS.
> 
> Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
> ---
> CC: Jan Beulich <JBeulich@suse.com>
> CC: Tim Deegan <tim@xen.org>
> CC: Ian Campbell <ian.campbell@citrix.com>
> CC: Stefano Stabellini <stefano.stabellini@citrix.com>
> CC: Julien Grall <julien.grall@citrix.com>
> CC: Roger Pau Monne <roger.pau@citrix.com>

For the ARM bits:

Reviewed-by: Julien Grall <julien.grall@citrix.com>

Regards,

-- 
Julien Grall

Re: [PATCH v2] xen: Improve x86's alloc_vcpu_guest_context()

[Jan Beulich]

>>> On 17.09.15 at 18:52, <andrew.cooper3@citrix.com> wrote:
> This essentially reverts c/s 2037f2adb "x86: introduce
> alloc_vcpu_guest_context()", including the newer arm bits, but achieves
> the same end goal by using the newer vmalloc() infrastructure.
> 
> For both x86 and ARM, {alloc,free}_vcpu_guest_context() become arch-local
> static inlines (which avoids a call into a separate translation),
> and removes an x86 scalability limit when compiling with a large NR_CPUS.
> 
> Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>

Reviewed-by: Jan Beulich <jbeulich@suse.com>

(still awaiting an ARM maintainer ack for this to go in)

Re: [PATCH v2] xen: Improve x86's alloc_vcpu_guest_context()

[Ian Campbell]

On Mon, 2015-09-21 at 08:12 -0600, Jan Beulich wrote:
> > > > On 17.09.15 at 18:52, <andrew.cooper3@citrix.com> wrote:
> > This essentially reverts c/s 2037f2adb "x86: introduce
> > alloc_vcpu_guest_context()", including the newer arm bits, but achieves
> > the same end goal by using the newer vmalloc() infrastructure.
> > 
> > For both x86 and ARM, {alloc,free}_vcpu_guest_context() become arch
> > -local
> > static inlines (which avoids a call into a separate translation),
> > and removes an x86 scalability limit when compiling with a large
> > NR_CPUS.
> > 
> > Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
> 
> Reviewed-by: Jan Beulich <jbeulich@suse.com>
> 
> (still awaiting an ARM maintainer ack for this to go in)

Sorry, thought I had already:

Acked-by: Ian Campbell <ian.campbell@citrix.com>