Re: [Qemu-devel] [PATCH] cpu_defs: Simplify CPUTLB padding logic

[Paolo Bonzini]

On 06/07/2015 13:42, Richard Henderson wrote:
> On 07/06/2015 09:43 AM, Paolo Bonzini wrote:
>>
>>
>> On 05/07/2015 23:08, Peter Crosthwaite wrote:
>>> There was a complicated subtractive arithmetic for determining the
>>> padding on the CPUTLBEntry structure. Simplify this with a union.
>>>
>>> Signed-off-by: Peter Crosthwaite <address@hidden>
>>> ---
>>>   include/exec/cpu-defs.h | 23 ++++++++++++-----------
>>>   1 file changed, 12 insertions(+), 11 deletions(-)
>>>
>>> diff --git a/include/exec/cpu-defs.h b/include/exec/cpu-defs.h
>>> index 98b9cff..5093be2 100644
>>> --- a/include/exec/cpu-defs.h
>>> +++ b/include/exec/cpu-defs.h
>>> @@ -105,17 +105,18 @@ typedef struct CPUTLBEntry {
>>>          bit 3                      : indicates that the entry is
>>> invalid
>>>          bit 2..0                   : zero
>>>       */
>>> -    target_ulong addr_read;
>>> -    target_ulong addr_write;
>>> -    target_ulong addr_code;
>>> -    /* Addend to virtual address to get host address.  IO accesses
>>> -       use the corresponding iotlb value.  */
>>> -    uintptr_t addend;
>>> -    /* padding to get a power of two size */
>>> -    uint8_t dummy[(1 << CPU_TLB_ENTRY_BITS) -
>>> -                  (sizeof(target_ulong) * 3 +
>>> -                   ((-sizeof(target_ulong) * 3) & (sizeof(uintptr_t)
>>> - 1)) +
>>> -                   sizeof(uintptr_t))];
>>> +    union {
>>
>> The struct CPUTLBEntry can be changed to union CPUTLBEntry directly,
>> with no need for the anonymous struct.
> 
> Um, no it can't.  That would put all of the members at the same address.

Of course. :-(  With no need for the anonymous _union_.  *blush*.

>> Which compiler version started implementing anonymous structs?
> 
> A long long time ago -- gcc 2 era.

Great.  I now remember that the recent feature is anonymous tagged
structs, coming from the Plan 9 compiler.

Paolo

>> Or can we just add
>>
>>      __attribute__((__aligned__(1 << CPU_TLB_ENTRY_BITS)))
> 
> The structure isn't currently aligned, and it needn't be.  We only need
> the size to be a power of two for the addressing.
> 
> 
> 
> r~