Re: [Qemu-devel] [PATCH] s390x: initialize virtio dev region

["Andreas Färber" <afaerber@suse.de>]

Am 11.11.2011 18:40, schrieb Alexander Graf:
> On 11/11/2011 05:44 PM, Peter Maydell wrote:
>> On 11 November 2011 16:24, Alexander Graf<agraf@suse.de>  wrote:
>>> On 11/11/2011 05:11 PM, Peter Maydell wrote:
>>>> Yes, so that's option (2) and you need to be using a
>>>> target_phys_addr_t.
>>> But ram_size is ram_addr_t and is the ram size that I have available
>>> to use,
>>> so it's exactly the address that I want. I don't see your point.
>>> Should go
>>> jump through random useless hoops of doing
>>>
>>>   target_phys_addr_t ram_end = ram_size;
>>>
>>> just because there are some subtile semantic differences between the two
>>> variables? They're integers at the end of the day. Both of them.
>> Because handing a ram_addr_t to something that wants a target_phys_addr_t
>> is a red flag that something might be wrong. (As is handing a variable
>> called _size to something that wants an address.)
>>
>> Force in Newtons and Force in pound-feet are also both just integers
>> at the end of the day, but that doesn't make them interchangeable:
>> http://en.wikipedia.org/wiki/Mars_Climate_Orbiter#Communications_loss
> 
> Or actually rather this one:
> 
> 
> Alex
> 
> diff --git a/hw/s390-virtio.c b/hw/s390-virtio.c
> index d936809..61b67e8 100644
> --- a/hw/s390-virtio.c
> +++ b/hw/s390-virtio.c
> @@ -167,6 +167,7 @@ static void s390_init(ram_addr_t my_ram_size,
>      uint8_t *storage_keys;
>      void *virtio_region;
>      target_phys_addr_t virtio_region_len;
> +    target_phys_addr_t virtio_region_start;
>      int i;
> 
>      /* s390x ram size detection needs a 16bit multiplier + an
> increment. So
> @@ -188,7 +189,9 @@ static void s390_init(ram_addr_t my_ram_size,
> 
>      /* clear virtio region */
>      virtio_region_len = my_ram_size - ram_size;
> -    virtio_region = cpu_physical_memory_map(ram_size,
> &virtio_region_len, true);
> +    virtio_region_start = ram_size;
> +    virtio_region = cpu_physical_memory_map(virtio_region_start,
> + &virtio_region_len, true);

Yes, ..._start reads better than ..._size there. Thanks.

Andreas

-- 
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg