Re: [Xen-devel] [PATCH v5 10/12] x86/hyperv: provide percpu hypercall input page

[Roger Pau Monné <roger.pau@xxxxxxxxxx>]

On Wed, Jan 29, 2020 at 08:20:32PM +0000, Wei Liu wrote:
> Hyper-V's input / output argument must be 8 bytes aligned an not cross
> page boundary. One way to satisfy those requirements is to use percpu
> page.
> 
> For the foreseeable future we only need to provide input for TLB
> and APIC hypercalls, so skip setting up an output page.
> 
> We will also need to provide an ap_setup hook for secondary cpus to
> setup its own input page.
> 
> Signed-off-by: Wei Liu <liuwe@xxxxxxxxxxxxx>

Reviewed-by: Roger Pau Monné <roger.pau@xxxxxxxxxx>

Just some nits below.

> ---
> v5:
> 1. Adjust to new ap_setup
> 2. Change variable name to hv_pcpu_input_page
> 
> v4:
> 1. Change wording in commit message
> 2. Prevent leak
> 3. Introduce a private header
> 
> v3:
> 1. Use xenheap page instead
> 2. Drop page tracking structure
> 3. Drop Paul's review tag
> ---
>  xen/arch/x86/guest/hyperv/hyperv.c  | 31 +++++++++++++++++++++++++++++
>  xen/arch/x86/guest/hyperv/private.h | 29 +++++++++++++++++++++++++++
>  2 files changed, 60 insertions(+)
>  create mode 100644 xen/arch/x86/guest/hyperv/private.h
> 
> diff --git a/xen/arch/x86/guest/hyperv/hyperv.c 
> b/xen/arch/x86/guest/hyperv/hyperv.c
> index 4387b6541e..f0facccbad 100644
> --- a/xen/arch/x86/guest/hyperv/hyperv.c
> +++ b/xen/arch/x86/guest/hyperv/hyperv.c
> @@ -27,7 +27,10 @@
>  #include <asm/guest/hyperv-tlfs.h>
>  #include <asm/processor.h>
>  
> +#include "private.h"
> +
>  struct ms_hyperv_info __read_mostly ms_hyperv;
> +DEFINE_PER_CPU_READ_MOSTLY(void *, hv_pcpu_input_page);

I would drop the 'pcpu_' from the name, as you already know it's
per-cpu because of the accessors you have to use.

>  
>  static uint64_t generate_guest_id(void)
>  {
> @@ -127,14 +130,42 @@ static void __init setup_hypercall_page(void)
>      }
>  }
>  
> +static int setup_hypercall_pcpu_arg(void)
> +{
> +    void *mapping;

There's no need for the local variable, you can just assign to
this_cpu(hv_pcpu_input_page) directly, as a failure will just set it
to NULL (as it already was).

> +
> +    if ( this_cpu(hv_pcpu_input_page) )
> +        return 0;
> +
> +    mapping = alloc_xenheap_page();
> +    if ( !mapping )
> +    {
> +        printk("Failed to allocate hypercall input page for CPU%u\n",
> +               smp_processor_id());

I find it clearer to have the CPU%u prefix at the begging of the line,
but it's up to you.

> +        return -ENOMEM;
> +    }
> +
> +    this_cpu(hv_pcpu_input_page) = mapping;
> +
> +    return 0;
> +}
> +
>  static void __init setup(void)
>  {
>      setup_hypercall_page();
> +    if ( setup_hypercall_pcpu_arg() )
> +        panic("Hypercall percpu arg setup failed\n");

Could you add "HyperV hypercall...", just hypercall page is too
generic IMO.

Thanks, Roger.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel