Re: [Qemu-ppc] [Qemu-devel] [PATCH] Fix circular dependency for HOST_LONG_BITS qemu-common.h <-> bswap.h

[Andreas Färber]

Am 13.02.2013 18:41, schrieb Anthony Liguori:
> David Gibson <address@hidden> writes:
> 
>> On Tue, Feb 05, 2013 at 11:42:30AM +0100, Andreas Färber wrote:
>>> Am 05.02.2013 01:07, schrieb Peter Maydell:
>>>> On 4 February 2013 23:52, Richard Henderson <address@hidden> wrote:
>>>>> On 2013-02-04 15:30, David Gibson wrote:
>>>>>> Anthony, Richard, anyone?
>>>>>>
>>>>>> Please apply - qemu has now been build-broken on all big endian
>>>>>> platforms for a month.
>>>>>
>>>>>
>>>>> I know.  See also my bswap.h patch which also fixes the width
>>>>> of long vs uintptr_t.  No one seems willing to pick these up...
>>>>
>>>> In both cases, the patch:
>>>>  * was sent out after the soft freeze
>>>>  * doesn't have a "for-1.4" tag
>>>>  * doesn't have a summary line that clearly says "fixes build
>>>>    failure" either
>>>>  * hasn't got a Reviewed-by: tag from anybody
>>>
>>> I ack'ed it, which in my terminology usually means that I reviewed and
>>> tested it.
>>>
>>>>
>>>> so I'm not terribly surprised they haven't got picked up.
>>>> You could start by reviewing each others' patches :-)
>>>
>>> Personally I see no reason to keep around misnamed HOST_LONG_BITS at all
>>> when we can easily calculate its value using sizeof(uintptr_t) * 8 or
>>> replace it by different conditions as suggested by rth. I thus prefer
>>
>> Ok, I missed rth's patch to do this differently.  Note that sizeof()
>> will not work in this case, because we need the correct value at cpp
>> time.
>>
>>> his patch and have been waiting for Blue to pick it up for 1.4.
>>
>> Sure, whatever.  Can we please just get whichever damn fix *in*.
> 
> For the sake of completeness, "rth's patch" means:
> 
> commit 91107fdf4443d2171e06840e87277bb7a047343b
> Author: Richard Henderson <address@hidden>
> Date:   Mon Feb 4 16:21:06 2013 -0800
> 
>     bswap: Fix width of swap in leul_to_cpu
> 
> Correct?

Yes.

> If so, this was committed before you sent this note.  Can someone
> confirm if we still have a problem on big endian hosts?

Around Central European lunch time today things seemed to compile fine,
and `make check` worked, too. (Thanks!)

Regards,
Andreas

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