[PATCH] kvmtool: ARM: madvise mergeable and hugepage separately

[PATCH] kvmtool: ARM: madvise mergeable and hugepage separately

[Stefan Agner]

The madvise behavior is not a bit field and hence can not be or'ed.
Also madvise_behavior_valid checks the flag using a case statement
hence only one behavior is supposed to be supplied. Call madvise
twice, once for MERGEABLE and once for HUGEPAGE.

Signed-off-by: Stefan Agner <stefan@agner.ch>
---
 arm/kvm.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/arm/kvm.c b/arm/kvm.c
index ce40897..3cfa90a 100644
--- a/arm/kvm.c
+++ b/arm/kvm.c
@@ -80,7 +80,10 @@ void kvm__arch_init(struct kvm *kvm, const char *hugetlbfs_path, u64 ram_size)
 					SZ_2M);
 
 	madvise(kvm->arch.ram_alloc_start, kvm->arch.ram_alloc_size,
-		MADV_MERGEABLE | MADV_HUGEPAGE);
+		MADV_MERGEABLE);
+
+	madvise(kvm->arch.ram_alloc_start, kvm->arch.ram_alloc_size,
+		MADV_HUGEPAGE);
 
 	/* Create the virtual GIC. */
 	if (gic__create(kvm, kvm->cfg.arch.irqchip))
-- 
2.9.0

Re: [PATCH] kvmtool: ARM: madvise mergeable and hugepage separately

[Andre Przywara]

Hi,

On 08/08/16 08:01, Stefan Agner wrote:
> The madvise behavior is not a bit field and hence can not be or'ed.
> Also madvise_behavior_valid checks the flag using a case statement
> hence only one behavior is supposed to be supplied. Call madvise
> twice, once for MERGEABLE and once for HUGEPAGE.

thanks for that catch! Indeed we are only setting HUGEPAGE right now,
dropping MERGEABLE due to the OR'ing.

Acked-by: Andre Przywara <andre.przywara@arm.com>

Cheers,
Andre.

> 
> Signed-off-by: Stefan Agner <stefan@agner.ch>
> ---
>  arm/kvm.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/arm/kvm.c b/arm/kvm.c
> index ce40897..3cfa90a 100644
> --- a/arm/kvm.c
> +++ b/arm/kvm.c
> @@ -80,7 +80,10 @@ void kvm__arch_init(struct kvm *kvm, const char *hugetlbfs_path, u64 ram_size)
>  					SZ_2M);
>  
>  	madvise(kvm->arch.ram_alloc_start, kvm->arch.ram_alloc_size,
> -		MADV_MERGEABLE | MADV_HUGEPAGE);
> +		MADV_MERGEABLE);
> +
> +	madvise(kvm->arch.ram_alloc_start, kvm->arch.ram_alloc_size,
> +		MADV_HUGEPAGE);
>  
>  	/* Create the virtual GIC. */
>  	if (gic__create(kvm, kvm->cfg.arch.irqchip))
> 

Re: [PATCH] kvmtool: ARM: madvise mergeable and hugepage separately

[Suzuki K Poulose]

On 08/08/16 08:01, Stefan Agner wrote:
> The madvise behavior is not a bit field and hence can not be or'ed.
> Also madvise_behavior_valid checks the flag using a case statement
> hence only one behavior is supposed to be supplied. Call madvise
> twice, once for MERGEABLE and once for HUGEPAGE.
>
> Signed-off-by: Stefan Agner <stefan@agner.ch>
> ---
>  arm/kvm.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/arm/kvm.c b/arm/kvm.c
> index ce40897..3cfa90a 100644
> --- a/arm/kvm.c
> +++ b/arm/kvm.c
> @@ -80,7 +80,10 @@ void kvm__arch_init(struct kvm *kvm, const char *hugetlbfs_path, u64 ram_size)
>  					SZ_2M);
>
>  	madvise(kvm->arch.ram_alloc_start, kvm->arch.ram_alloc_size,
> -		MADV_MERGEABLE | MADV_HUGEPAGE);
> +		MADV_MERGEABLE);
> +
> +	madvise(kvm->arch.ram_alloc_start, kvm->arch.ram_alloc_size,
> +		MADV_HUGEPAGE);

Should we also do a MADV_DONTDUMP for guest RAM pages ? This is a security option
to strip off any guest data in case of crash in kvmtool. This could also help
to reduce the size of the kvmtool core dumps. I think Qemu does this already.

Suzuki