Re: [PATCH] kvm-userspace: fix memslot assignment

[Avi Kivity <avi@qumranet.com>]

Christian Borntraeger wrote:
> Hello Avi,
>
> seems that I mixed up the slot initialization, instead of making the first 
> slot always 0 I made it always 1. Lets go back to Carstens variant, since I 
> dont like nested ifdefs. The compiler will remove the dead code anyway.
>
> Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
>
> ---
>  libkvm.c |    7 +++++--
>  1 file changed, 5 insertions(+), 2 deletions(-)
>
> Index: libkvm/libkvm.c
> ===================================================================
> --- libkvm.orig/libkvm.c
> +++ libkvm/libkvm.c
> @@ -78,7 +78,7 @@ int get_free_slot(kvm_context_t kvm)
>  	int i;
>  	int tss_ext;
>  
> -#if defined(KVM_CAP_SET_TSS_ADDR) && !defined(__s390__)
> +#if defined(KVM_CAP_SET_TSS_ADDR)
>  	tss_ext = ioctl(kvm->fd, KVM_CHECK_EXTENSION, KVM_CAP_SET_TSS_ADDR);
>  #else
>  	tss_ext = 0;
>   

This is really a no-op, since s390 will nack a KVM_CAP_SET_TSS_ADDR 
query.  Of course, the change is an improvement.

> @@ -92,8 +92,11 @@ int get_free_slot(kvm_context_t kvm)
>  	if (tss_ext > 0)
>  		i = 0;
>  	else
> +#if !defined(__s390__)
>  		i = 1;
> -
> +#else
> +		i = 0;
> +#endif
>   

ppc and ia64 also don't care about TSS_ADDR.  Care to make this depend 
on x86, rather than on s390?

-- 
error compiling committee.c: too many arguments to function