Re: [Qemu-devel] [QEMU][RFC PATCH 3/6] memory: Add xen memory hook

[Avi Kivity]

On 03/23/2012 06:37 PM, Jan Kiszka wrote:
> On 2012-03-23 16:08, Julien Grall wrote:
> > On 03/22/2012 05:44 PM, Jan Kiszka wrote:
> >>>
> >>>   static void core_region_nop(MemoryListener *listener,
> >>> diff --git a/ioport.c b/ioport.c
> >>> index 78a3b89..073ed75 100644
> >>> --- a/ioport.c
> >>> +++ b/ioport.c
> >>> @@ -28,6 +28,7 @@
> >>>   #include "ioport.h"
> >>>   #include "trace.h"
> >>>   #include "memory.h"
> >>> +#include "hw/xen.h"
> >>>
> >>>   /***********************************************************/
> >>>   /* IO Port */
> >>> @@ -155,6 +156,11 @@ int register_ioport_read(pio_addr_t start, int
> >>> length, int size,
> >>>                        i);
> >>>           ioport_opaque[i] = opaque;
> >>>       }
> >>> +
> >>> +    if (xen_enabled()) {
> >>> +        xen_map_iorange(start, length, 0);
> >>> +    }
> >>> +
> >>>       return 0;
> >>>   }
> >>>
> >>> @@ -175,7 +181,13 @@ int register_ioport_write(pio_addr_t start, int
> >>> length, int size,
> >>>                        i);
> >>>           ioport_opaque[i] = opaque;
> >>>       }
> >>> +
> >>> +    if (xen_enabled()) {
> >>> +        xen_map_iorange(start, length, 0);
> >>> +    }
> >>> +
> >>>       return 0;
> >>> +
> >>>   }
> >>>
> >>>   static uint32_t ioport_readb_thunk(void *opaque, uint32_t addr)
> >>> @@ -260,6 +272,11 @@ void isa_unassign_ioport(pio_addr_t start, int
> >>> length)
> >>>           ioport_destructor_table[start](ioport_opaque[start]);
> >>>           ioport_destructor_table[start] = NULL;
> >>>       }
> >>> +
> >>> +    if (xen_enabled()) {
> >>> +        xen_unmap_iorange(start, length, 0);
> >>> +    }
> >>> +
> >>>       for(i = start; i<  start + length; i++) {
> >>>           ioport_read_table[0][i] = NULL;
> >>>           ioport_read_table[1][i] = NULL;
> >>>      
> >> memory_listener_register(xen_hooks, system_io)?
> >>    
> > QEMU doesn't seem to call region_add/region_del for ioport.
> > Moreover, some of ioport are directly register without
> > using memory hook (for example cirrus vga).
> > 
> > What is the best way to do it ?
>
> I haven't looked at details. Maybe it is just a combination of "use case
> not yet considered, but can easily be added" and "need to switch legacy
> code to new scheme". Then this still remains the better option than this
> hook. Avi?

Just the second - region_add/del will be called, but only for ioports
registered via the MemoryRegion APIs.

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