Re: [Xen-devel] [PATCH v4 2/7] x86/hyperv: setup hypercall page

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 22/01/2020 20:23, Wei Liu wrote:
> Use the top-most addressable page for that purpose. Adjust e820 code
> accordingly.
>
> We also need to register Xen's guest OS ID to Hyper-V. Use 0x300 as the
> OS type.
>
> Signed-off-by: Wei Liu <liuwe@xxxxxxxxxxxxx>
> ---
> XXX the decision on Xen's vendor ID is pending.

Presumably this is pending a published update to the TLFS?  (And I
presume using 0x8088 is out of the question?  That is an X in the bottom
byte, not a reference to an 8 bit microprocessor.)

> diff --git a/xen/arch/x86/e820.c b/xen/arch/x86/e820.c
> index 082f9928a1..5a4ef27a0b 100644
> --- a/xen/arch/x86/e820.c
> +++ b/xen/arch/x86/e820.c
> @@ -36,6 +36,22 @@ boolean_param("e820-verbose", e820_verbose);
> @@ -357,6 +373,21 @@ static unsigned long __init find_max_pfn(void)
>              max_pfn = end;
>      }
>  
> +#ifdef CONFIG_HYPERV_GUEST
> +    {
> +     /*
> +      * We reserve the top-most page for hypercall page. Adjust
> +      * max_pfn if necessary.

It might be worth leaving a "TODO: Better algorithm/guess?" here.

> +      */
> +        unsigned int phys_bits = find_phys_addr_bits();
> +        unsigned long hcall_pfn =
> +          ((1ull << phys_bits) - 1) >> PAGE_SHIFT;

(1ull << (phys_bits - PAGE_SHIFT)) - 1 is equivalent, and doesn't
require a right shift.  I don't know if the compiler is smart enough to
make this optimisation automatically.

> +
> +        if ( max_pfn >= hcall_pfn )
> +          max_pfn = hcall_pfn - 1;

Indentation looks weird.

> @@ -446,13 +477,7 @@ static uint64_t __init mtrr_top_of_ram(void)
>           return 0;
>  
>      /* Find the physical address size for this CPU. */
> -    eax = cpuid_eax(0x80000000);
> -    if ( (eax >> 16) == 0x8000 && eax >= 0x80000008 )
> -    {
> -        phys_bits = (uint8_t)cpuid_eax(0x80000008);
> -        if ( phys_bits > PADDR_BITS )
> -            phys_bits = PADDR_BITS;
> -    }
> +    phys_bits = find_phys_addr_bits();
>      addr_mask = ((1ull << phys_bits) - 1) & ~((1ull << 12) - 1);

Note for whomever is next doing cleanup in this area.  This wants to be
& PAGE_MASK.

> diff --git a/xen/arch/x86/guest/hyperv/hyperv.c 
> b/xen/arch/x86/guest/hyperv/hyperv.c
> index 8d38313d7a..f986c1a805 100644
> --- a/xen/arch/x86/guest/hyperv/hyperv.c
> +++ b/xen/arch/x86/guest/hyperv/hyperv.c
> @@ -72,6 +82,43 @@ const struct hypervisor_ops *__init hyperv_probe(void)
>      return &ops;
>  }
>  
> +static void __init setup_hypercall_page(void)
> +{
> +    union hv_x64_msr_hypercall_contents hypercall_msr;
> +    union hv_guest_os_id guest_id;
> +    unsigned long mfn;
> +
> +    rdmsrl(HV_X64_MSR_GUEST_OS_ID, guest_id.raw);
> +    if ( !guest_id.raw )
> +    {
> +        guest_id.raw = generate_guest_id();
> +        wrmsrl(HV_X64_MSR_GUEST_OS_ID, guest_id.raw);
> +    }
> +
> +    rdmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);
> +    if ( !hypercall_msr.enable )
> +    {
> +        mfn = ((1ull << paddr_bits) - 1) >> HV_HYP_PAGE_SHIFT;
> +        hypercall_msr.enable = 1;
> +        hypercall_msr.guest_physical_address = mfn;
> +        wrmsrl(HV_X64_MSR_HYPERCALL, hypercall_msr.as_uint64);

Is it worth reading back, and BUG() if it is different?  It will be a
more obvious failure than hypercalls disappearing mysteriously.

> +    } else {
> +        mfn = hypercall_msr.guest_physical_address;
> +    }

Style.

Otherwise, LGTM.

~Andrew

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