> On Sun, 2007-08-12 at 07:58 +0100, Al Viro wrote:
>> On Sun, Aug 12, 2007 at 03:21:57AM +0200, Rene Herman wrote:
>>> @@ -2609,13 +2609,13 @@ static void do_cciss_request(request_queue_t *q)
>>> } else {
>>> c->Request.CDBLen = 16;
>>> c->Request.CDB[1]= 0;
>>> - c->Request.CDB[2]= (start_blk >> 56) & 0xff; //MSB
>>> - c->Request.CDB[3]= (start_blk >> 48) & 0xff;
>>> - c->Request.CDB[4]= (start_blk >> 40) & 0xff;
>>> - c->Request.CDB[5]= (start_blk >> 32) & 0xff;
>>> - c->Request.CDB[6]= (start_blk >> 24) & 0xff;
>>> - c->Request.CDB[7]= (start_blk >> 16) & 0xff;
>>> - c->Request.CDB[8]= (start_blk >> 8) & 0xff;
>>> + c->Request.CDB[2]= ((u64)start_blk >> 56) & 0xff; //MSB
>>> + c->Request.CDB[3]= ((u64)start_blk >> 48) & 0xff;
>>> + c->Request.CDB[4]= ((u64)start_blk >> 40) & 0xff;
>>> + c->Request.CDB[5]= ((u64)start_blk >> 32) & 0xff;
>>> + c->Request.CDB[6]= ((u64)start_blk >> 24) & 0xff;
>>> + c->Request.CDB[7]= ((u64)start_blk >> 16) & 0xff;
>>> + c->Request.CDB[8]= ((u64)start_blk >> 8) & 0xff;
>> put_unaligned(cpu_to_be64(start_blk), &c->Request.CDB[2]);
>>
>> which is what's happening here anyway.
>
> Well ... this was debated a while ago ad nauseam:
>
>
http://marc.info/?t=117699555300010
>
> The main objection to what you propose is that it forces the u64
> coercion even in the 32 bit start_blk case. The preferred solution as a
> result of that debate was simply to us a macro Andrew introduced:
>
> upper_32_bits()
>
> Which will silently replace zero in the non LBD case. I actually
> thought this had already been done.
