Patchwork qemu_vmalloc: align properly for transparent hugepages and KVM

login
register
mail settings
Submitter Avi Kivity
Date Sept. 5, 2011, 8:07 a.m.
Message ID <1315210025-17727-1-git-send-email-avi@redhat.com>
Download mbox | patch
Permalink /patch/113320/
State New
Headers show

Comments

Avi Kivity - Sept. 5, 2011, 8:07 a.m.
To make good use of transparent hugepages, KVM requires that guest-physical
and host-virtual addresses share the low 21 bits (as opposed to just the low
12 bits normally required).

Adjust qemu_vmalloc() to honor that requirement.  Ignore it for small regions
to avoid fragmentation.

Signed-off-by: Avi Kivity <avi@redhat.com>
---
 oslib-posix.c |   14 +++++++++++++-
 1 files changed, 13 insertions(+), 1 deletions(-)
Jan Kiszka - Sept. 5, 2011, 10:10 a.m.
On 2011-09-05 10:07, Avi Kivity wrote:
> To make good use of transparent hugepages, KVM requires that guest-physical
> and host-virtual addresses share the low 21 bits (as opposed to just the low
> 12 bits normally required).
> 
> Adjust qemu_vmalloc() to honor that requirement.  Ignore it for small regions
> to avoid fragmentation.
> 
> Signed-off-by: Avi Kivity <avi@redhat.com>
> ---
>  oslib-posix.c |   14 +++++++++++++-
>  1 files changed, 13 insertions(+), 1 deletions(-)
> 
> diff --git a/oslib-posix.c b/oslib-posix.c
> index 196099c..a304fb0 100644
> --- a/oslib-posix.c
> +++ b/oslib-posix.c
> @@ -35,6 +35,13 @@
>  extern int daemon(int, int);
>  #endif
>  
> +#if defined(__linux__) && defined(__x86_64__)
> +   /* Use 2MB alignment so transparent hugepages can be used by KVM */

Aren't transparent hugepages also available in TCG mode? Then just
remove "by KVM" from subject and comment.

Jan

> +#  define QEMU_VMALLOC_ALIGN (512 * 4096)
> +#else
> +#  define QEMU_VMALLOC_ALIGN getpagesize()
> +#endif
> +
>  #include "config-host.h"
>  #include "sysemu.h"
>  #include "trace.h"
> @@ -80,7 +87,12 @@ int qemu_daemon(int nochdir, int noclose)
>  void *qemu_vmalloc(size_t size)
>  {
>      void *ptr;
> -    ptr = qemu_memalign(getpagesize(), size);
> +    size_t align = QEMU_VMALLOC_ALIGN;
> +
> +    if (size < align) {
> +        align = getpagesize();
> +    }
> +    ptr = qemu_memalign(align, size);
>      trace_qemu_vmalloc(size, ptr);
>      return ptr;
>  }
Avi Kivity - Sept. 5, 2011, 10:21 a.m.
On 09/05/2011 01:10 PM, Jan Kiszka wrote:
> On 2011-09-05 10:07, Avi Kivity wrote:
> >  To make good use of transparent hugepages, KVM requires that guest-physical
> >  and host-virtual addresses share the low 21 bits (as opposed to just the low
> >  12 bits normally required).
> >
> >  Adjust qemu_vmalloc() to honor that requirement.  Ignore it for small regions
> >  to avoid fragmentation.
> >
> >  Signed-off-by: Avi Kivity<avi@redhat.com>
> >  ---
> >   oslib-posix.c |   14 +++++++++++++-
> >   1 files changed, 13 insertions(+), 1 deletions(-)
> >
> >  diff --git a/oslib-posix.c b/oslib-posix.c
> >  index 196099c..a304fb0 100644
> >  --- a/oslib-posix.c
> >  +++ b/oslib-posix.c
> >  @@ -35,6 +35,13 @@
> >   extern int daemon(int, int);
> >   #endif
> >
> >  +#if defined(__linux__)&&  defined(__x86_64__)
> >  +   /* Use 2MB alignment so transparent hugepages can be used by KVM */
>
> Aren't transparent hugepages also available in TCG mode? Then just
> remove "by KVM" from subject and comment.

They are, but they don't require the special alignment.
Anthony Liguori - Sept. 9, 2011, 6:34 p.m.
On 09/05/2011 03:07 AM, Avi Kivity wrote:
> To make good use of transparent hugepages, KVM requires that guest-physical
> and host-virtual addresses share the low 21 bits (as opposed to just the low
> 12 bits normally required).
>
> Adjust qemu_vmalloc() to honor that requirement.  Ignore it for small regions
> to avoid fragmentation.
>
> Signed-off-by: Avi Kivity<avi@redhat.com>

Applied.  Thanks.

Regards,

Anthony Liguori

> ---
>   oslib-posix.c |   14 +++++++++++++-
>   1 files changed, 13 insertions(+), 1 deletions(-)
>
> diff --git a/oslib-posix.c b/oslib-posix.c
> index 196099c..a304fb0 100644
> --- a/oslib-posix.c
> +++ b/oslib-posix.c
> @@ -35,6 +35,13 @@
>   extern int daemon(int, int);
>   #endif
>
> +#if defined(__linux__)&&  defined(__x86_64__)
> +   /* Use 2MB alignment so transparent hugepages can be used by KVM */
> +#  define QEMU_VMALLOC_ALIGN (512 * 4096)
> +#else
> +#  define QEMU_VMALLOC_ALIGN getpagesize()
> +#endif
> +
>   #include "config-host.h"
>   #include "sysemu.h"
>   #include "trace.h"
> @@ -80,7 +87,12 @@ int qemu_daemon(int nochdir, int noclose)
>   void *qemu_vmalloc(size_t size)
>   {
>       void *ptr;
> -    ptr = qemu_memalign(getpagesize(), size);
> +    size_t align = QEMU_VMALLOC_ALIGN;
> +
> +    if (size<  align) {
> +        align = getpagesize();
> +    }
> +    ptr = qemu_memalign(align, size);
>       trace_qemu_vmalloc(size, ptr);
>       return ptr;
>   }

Patch

diff --git a/oslib-posix.c b/oslib-posix.c
index 196099c..a304fb0 100644
--- a/oslib-posix.c
+++ b/oslib-posix.c
@@ -35,6 +35,13 @@ 
 extern int daemon(int, int);
 #endif
 
+#if defined(__linux__) && defined(__x86_64__)
+   /* Use 2MB alignment so transparent hugepages can be used by KVM */
+#  define QEMU_VMALLOC_ALIGN (512 * 4096)
+#else
+#  define QEMU_VMALLOC_ALIGN getpagesize()
+#endif
+
 #include "config-host.h"
 #include "sysemu.h"
 #include "trace.h"
@@ -80,7 +87,12 @@  int qemu_daemon(int nochdir, int noclose)
 void *qemu_vmalloc(size_t size)
 {
     void *ptr;
-    ptr = qemu_memalign(getpagesize(), size);
+    size_t align = QEMU_VMALLOC_ALIGN;
+
+    if (size < align) {
+        align = getpagesize();
+    }
+    ptr = qemu_memalign(align, size);
     trace_qemu_vmalloc(size, ptr);
     return ptr;
 }