KVM: arm64: Use pointer from memcpy() call for assignment in init_hyp_mode()

[PATCH] KVM: arm64: Use pointer from memcpy() call for assignment in init_hyp_mode()

Posted by Markus Elfring 3 months, 1 week ago

From: Markus Elfring <elfring@users.sourceforge.net>
Date: Thu, 30 Oct 2025 18:01:41 +0100

A pointer was assigned to a variable. The same pointer was used for
the destination parameter of a memcpy() call.
This function is documented in the way that the same value is returned.
Thus convert two separate statements into a direct variable assignment for
the return value from a memory copy action.

The source code was transformed by using the Coccinelle software.

Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
---
 arch/arm64/kvm/arm.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
index 870953b4a8a7..feab88c31703 100644
--- a/arch/arm64/kvm/arm.c
+++ b/arch/arm64/kvm/arm.c
@@ -2600,8 +2600,8 @@ static int __init init_hyp_mode(void)
 			goto out_err;
 		}

-		page_addr = page_address(page);
-		memcpy(page_addr, CHOOSE_NVHE_SYM(__per_cpu_start), nvhe_percpu_size());
+		page_addr = memcpy(page_address(page), CHOOSE_NVHE_SYM(__per_cpu_start),
+				   nvhe_percpu_size());
 		kvm_nvhe_sym(kvm_arm_hyp_percpu_base)[cpu] = (unsigned long)page_addr;
 	}

-- 
2.51.1

Re: [PATCH] KVM: arm64: Use pointer from memcpy() call for assignment in init_hyp_mode()

Posted by Mark Rutland 3 months, 1 week ago

On Thu, Oct 30, 2025 at 06:11:03PM +0100, Markus Elfring wrote:
> From: Markus Elfring <elfring@users.sourceforge.net>
> Date: Thu, 30 Oct 2025 18:01:41 +0100
> 
> A pointer was assigned to a variable. The same pointer was used for
> the destination parameter of a memcpy() call.
> This function is documented in the way that the same value is returned.
> Thus convert two separate statements into a direct variable assignment for
> the return value from a memory copy action.
> 
> The source code was transformed by using the Coccinelle software.
> 
> Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
> ---
>  arch/arm64/kvm/arm.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
> index 870953b4a8a7..feab88c31703 100644
> --- a/arch/arm64/kvm/arm.c
> +++ b/arch/arm64/kvm/arm.c
> @@ -2600,8 +2600,8 @@ static int __init init_hyp_mode(void)
>  			goto out_err;
>  		}
>  
> -		page_addr = page_address(page);
> -		memcpy(page_addr, CHOOSE_NVHE_SYM(__per_cpu_start), nvhe_percpu_size());
> +		page_addr = memcpy(page_address(page), CHOOSE_NVHE_SYM(__per_cpu_start),
> +				   nvhe_percpu_size());

This change makes the code harder to read, and harder to modify. It
saves no space.

As Dan said [1]:

| No one will thank you for making these changes...  :(  Please don't do
| it.

[1] https://lore.kernel.org/lkml/aQNsecHJSO2U68Fc@stanley.mountain/

Mark.

Re: [PATCH] KVM: arm64: Use pointer from memcpy() call for assignment in init_hyp_mode()

Posted by David Laight 3 months, 1 week ago

On Thu, 30 Oct 2025 17:49:51 +0000
Mark Rutland <mark.rutland@arm.com> wrote:

> On Thu, Oct 30, 2025 at 06:11:03PM +0100, Markus Elfring wrote:
> > From: Markus Elfring <elfring@users.sourceforge.net>
> > Date: Thu, 30 Oct 2025 18:01:41 +0100
> > 
> > A pointer was assigned to a variable. The same pointer was used for
> > the destination parameter of a memcpy() call.
> > This function is documented in the way that the same value is returned.
> > Thus convert two separate statements into a direct variable assignment for
> > the return value from a memory copy action.
> > 
> > The source code was transformed by using the Coccinelle software.
> > 
> > Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
> > ---
> >  arch/arm64/kvm/arm.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/arch/arm64/kvm/arm.c b/arch/arm64/kvm/arm.c
> > index 870953b4a8a7..feab88c31703 100644
> > --- a/arch/arm64/kvm/arm.c
> > +++ b/arch/arm64/kvm/arm.c
> > @@ -2600,8 +2600,8 @@ static int __init init_hyp_mode(void)
> >  			goto out_err;
> >  		}
> >  
> > -		page_addr = page_address(page);
> > -		memcpy(page_addr, CHOOSE_NVHE_SYM(__per_cpu_start), nvhe_percpu_size());
> > +		page_addr = memcpy(page_address(page), CHOOSE_NVHE_SYM(__per_cpu_start),
> > +				   nvhe_percpu_size());  
> 
> This change makes the code harder to read, and harder to modify. It
> saves no space.

It might save a register spill - but really isn't worth the effort.
memcpy() is really best treated as being 'void'.
Indeed most implementations would be better if it were 'void'.
Although you could define:
#define memcpy(d, s, l) ({ auto _d = d; void_memcpy(_d, s, l); _d})
so that the compiler would optimise away the save that memcpy() typically
has to do.

I even suspect that memcpy() is an old enough function that the return
value is 'what the implementation happened to leave in r0'.

	David


> 
> As Dan said [1]:
> 
> | No one will thank you for making these changes...  :(  Please don't do
> | it.
> 
> [1] https://lore.kernel.org/lkml/aQNsecHJSO2U68Fc@stanley.mountain/
> 
> Mark.
>