Re: [Xen-devel] [PATCH] xen: arm: correct guest PSCI handling on 64-bit hypervisor.

[Julien Grall <julien.grall@xxxxxxxxxx>]

On 01/14/2014 05:32 PM, Ian Campbell wrote:
> Using ->rN truncates the 64-bit registers to 32-bits, which on X-gene chops
> off the top bit of the entry address for PSCI_UP.
> 
> Follow the pattern established in do_trap_hypercall.
> 
> Signed-off-by: Ian Campbell <ian.campbell@xxxxxxxxxx>

Acked-by: Julien Grall <julien.grall@xxxxxxxxxx>

> ---
> 
> Release argument: Only supporting single vcpu guests on arm64 would be
> unfortunate. There is no risk to arm32 since the ifdef ensures the code
> remains the same.
> ---
>  xen/arch/arm/traps.c |   17 ++++++++++++++---
>  1 file changed, 14 insertions(+), 3 deletions(-)
> 
> diff --git a/xen/arch/arm/traps.c b/xen/arch/arm/traps.c
> index fdf9440..62c9df2 100644
> --- a/xen/arch/arm/traps.c
> +++ b/xen/arch/arm/traps.c
> @@ -1065,23 +1065,34 @@ static void do_debug_trap(struct cpu_user_regs *regs, 
> unsigned int code)
>  }
>  #endif
>  
> +#ifdef CONFIG_ARM_64
> +#define PSCI_OP_REG(r) (r)->x0
> +#define PSCI_RESULT_REG(r) (r)->x0
> +#define PSCI_ARGS(r) (r)->x1, (r)->x2
> +#else
> +#define PSCI_OP_REG(r) (r)->r0
> +#define PSCI_RESULT_REG(r) (r)->r0
> +#define PSCI_ARGS(r) (r)->r1, (r)->r2
> +#endif
> +
>  static void do_trap_psci(struct cpu_user_regs *regs)
>  {
>      arm_psci_fn_t psci_call = NULL;
>  
> -    if ( regs->r0 >= ARRAY_SIZE(arm_psci_table) )
> +    if ( PSCI_OP_REG(regs) >= ARRAY_SIZE(arm_psci_table) )
>      {
>          domain_crash_synchronous();
>          return;
>      }
>  
> -    psci_call = arm_psci_table[regs->r0].fn;
> +    psci_call = arm_psci_table[PSCI_OP_REG(regs)].fn;
>      if ( psci_call == NULL )
>      {
>          domain_crash_synchronous();
>          return;
>      }
> -    regs->r0 = psci_call(regs->r1, regs->r2);
> +
> +    PSCI_RESULT_REG(regs) = psci_call(PSCI_ARGS(regs));
>  }
>  
>  #ifdef CONFIG_ARM_64
> 


-- 
Julien Grall

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel