Re: [Xen-devel] [PATCH 5 of 5] Ignore memory events with an obviously invalid gfn

[Tim Deegan <tim@xxxxxxx>]

At 10:12 -0500 on 05 Dec (1323079924), Andres Lagar-Cavilla wrote:
>  xen/arch/x86/mm/mem_sharing.c  |  2 ++
>  xen/arch/x86/mm/p2m.c          |  4 ++++
>  xen/include/public/mem_event.h |  7 +++++++
>  3 files changed, 13 insertions(+), 0 deletions(-)
> 
> 
> Ring semantics require that for every request, a response be put. This
> allows consumer to place a dummy response if need be.
> 
> diff -r c71f4c5b42f0 -r 790cd814bee8 xen/include/public/mem_event.h
> --- a/xen/include/public/mem_event.h
> +++ b/xen/include/public/mem_event.h
> @@ -76,6 +76,13 @@ typedef struct mem_event_st {
>  
>  DEFINE_RING_TYPES(mem_event, mem_event_request_t, mem_event_response_t);
>  
> +#define INVALID_RSP_GFN -1UL
> +

On a 32-bit system this doesn't do what you want.

Actually, can you use a flag for this, please?  Wedging it into the gfn
field seems a bit of a hack, and we have plenty of flag bits left.

Tim.

> +#define DUMMY_MEM_EVENT_RSP(_r)         ((_r)->gfn == INVALID_RSP_GFN)
> +#define MAKE_MEM_EVENT_RSP_DUMMY(_r)    ((_r)->gfn = INVALID_RSP_GFN)
> +#define DECLARE_DUMMY_MEM_EVENT_RSP(_name) \
> +    mem_event_response_t _name = { .gfn = INVALID_RSP_GFN }
> +
>  #endif
>  
>  /*
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxxxxxxxx
> http://lists.xensource.com/xen-devel

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel