[Mesa-dev] [PATCH 2/7] gallium/auxiliary: Add u_bitcast.h header.

Roland Scheidegger sroland at vmware.com
Fri Jul 29 14:33:56 UTC 2016


Ah you're right it is guaranteed by c99 and newer (who cares about c++ :-)).

I was probably remembering some discussion for older c standards, where
it might not have been guaranteed by the standard but all compilers
would have honored the union aliasing anyway.

But in any case I'm still fine with using a memcpy - the linked article
even mentions gcc generates the same code.

Roland


Am 29.07.2016 um 15:42 schrieb Gustaw Smolarczyk:
> Is that really the case? I always thought that using union is one of
> the best ways to play it safe while not disabling strict aliasing
> rules.
> 
> Here is an answer on stack overflow that uses this reasoning:
> https://urldefense.proofpoint.com/v2/url?u=http-3A__stackoverflow.com_a_99010&d=CwIFaQ&c=Sqcl0Ez6M0X8aeM67LKIiDJAXVeAw-YihVMNtXt-uEs&r=Vjtt0vs_iqoI31UfJxBl7yv9I2FeiaeAYgMTLKRBc_I&m=yFSuLWBBbLWZDrPaxDVOCjO0cX88sn16LdG-wdamPNQ&s=Ay0ZWq7VAYWw11SblOvtJSj9jhhARSxr13nrT-y8iik&e= 
> 
> Here is another site with an extract from C11 standard:
> https://urldefense.proofpoint.com/v2/url?u=http-3A__dbp-2Dconsulting.com_tutorials_StrictAliasing.html&d=CwIFaQ&c=Sqcl0Ez6M0X8aeM67LKIiDJAXVeAw-YihVMNtXt-uEs&r=Vjtt0vs_iqoI31UfJxBl7yv9I2FeiaeAYgMTLKRBc_I&m=yFSuLWBBbLWZDrPaxDVOCjO0cX88sn16LdG-wdamPNQ&s=vIzvE6OQ57S1mnD6755iC9AOD57xXqJKUGtsk_DKBhs&e= 
> 
> Look especially at footnote 85 of 6.5.2.3.
> 
> Regards,
> Gustaw Smolarczyk
> 
> 2016-07-29 15:22 GMT+02:00 Roland Scheidegger <sroland at vmware.com>:
>> Am 29.07.2016 um 06:55 schrieb Connor Abbott:
>>> On Thu, Jul 28, 2016 at 7:58 PM, Roland Scheidegger <sroland at vmware.com> wrote:
>>>> Am 29.07.2016 um 00:35 schrieb Matt Turner:
>>>>> ---
>>>>>  src/gallium/auxiliary/Makefile.sources |  1 +
>>>>>  src/gallium/auxiliary/util/u_bitcast.h | 57 ++++++++++++++++++++++++++++++++++
>>>>>  2 files changed, 58 insertions(+)
>>>>>  create mode 100644 src/gallium/auxiliary/util/u_bitcast.h
>>>>>
>>>>> diff --git a/src/gallium/auxiliary/Makefile.sources b/src/gallium/auxiliary/Makefile.sources
>>>>> index e0311bf..26f7eee 100644
>>>>> --- a/src/gallium/auxiliary/Makefile.sources
>>>>> +++ b/src/gallium/auxiliary/Makefile.sources
>>>>> @@ -175,6 +175,7 @@ C_SOURCES := \
>>>>>       translate/translate_generic.c \
>>>>>       translate/translate_sse.c \
>>>>>       util/dbghelp.h \
>>>>> +     util/u_bitcast.h \
>>>>>       util/u_bitmask.c \
>>>>>       util/u_bitmask.h \
>>>>>       util/u_blend.h \
>>>>> diff --git a/src/gallium/auxiliary/util/u_bitcast.h b/src/gallium/auxiliary/util/u_bitcast.h
>>>>> new file mode 100644
>>>>> index 0000000..b1f9938
>>>>> --- /dev/null
>>>>> +++ b/src/gallium/auxiliary/util/u_bitcast.h
>>>>> @@ -0,0 +1,57 @@
>>>>> +/**************************************************************************
>>>>> + *
>>>>> + * Copyright © 2016 Intel Corporation
>>>>> + * All Rights Reserved.
>>>>> + *
>>>>> + * Permission is hereby granted, free of charge, to any person obtaining a
>>>>> + * copy of this software and associated documentation files (the
>>>>> + * "Software"), to deal in the Software without restriction, including
>>>>> + * without limitation the rights to use, copy, modify, merge, publish,
>>>>> + * distribute, sub license, and/or sell copies of the Software, and to
>>>>> + * permit persons to whom the Software is furnished to do so, subject to
>>>>> + * the following conditions:
>>>>> + *
>>>>> + * The above copyright notice and this permission notice (including the
>>>>> + * next paragraph) shall be included in all copies or substantial portions
>>>>> + * of the Software.
>>>>> + *
>>>>> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
>>>>> + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
>>>>> + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
>>>>> + * IN NO EVENT SHALL THE AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR
>>>>> + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
>>>>> + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
>>>>> + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
>>>>> + *
>>>>> + **************************************************************************/
>>>>> +
>>>>> +#ifndef U_BITCAST_H_
>>>>> +#define U_BITCAST_H_
>>>>> +
>>>>> +#include <string.h>
>>>>> +
>>>>> +#ifdef __cplusplus
>>>>> +extern "C" {
>>>>> +#endif
>>>>> +
>>>>> +static inline unsigned
>>>>> +u_bitcast_f2u(float f)
>>>>> +{
>>>>> +   unsigned u;
>>>>> +   memcpy(&u, &f, sizeof(u));
>>>>> +   return u;
>>>>> +}
>>>>> +
>>>>> +static inline float
>>>>> +u_bitcast_u2f(unsigned u)
>>>>> +{
>>>>> +   float f;
>>>>> +   memcpy(&f, &u, sizeof(f));
>>>>> +   return f;
>>>>> +}
>>>>> +
>>>>> +#ifdef __cplusplus
>>>>> +}
>>>>> +#endif
>>>>> +
>>>>> +#endif /* U_BITCAST_H_ */
>>>>>
>>>>
>>>> I think traditionally we've used the union fi from u_math.h to do this.
>>>> That said, I don't have anything against using an inline memcpy helper
>>>> function instead - both should hopefully get optimized away...
>>>
>>> Unfortunately, that won't work with the strict aliasing rules either
>>> since it's undefined behavior to reinterpret a union by dereferencing
>>> more than member. That is, once you assign to one field of a union,
>>> you can read from/write to that field only. You can, however,
>>> reinterpret anything as a char * and dereference, hence why memcopy
>>> etc. aren't undefined.
>>>
>>
>> That might be technically true, but I'm pretty sure it's not in
>> practice. Compilers don't warn about it, and are well prepared to not
>> misompile it, since it's very commonly used for avoiding aliasing problems.
>>
>> Roland
>>
>>
>> _______________________________________________
>> mesa-dev mailing list
>> mesa-dev at lists.freedesktop.org
>> https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.freedesktop.org_mailman_listinfo_mesa-2Ddev&d=CwIFaQ&c=Sqcl0Ez6M0X8aeM67LKIiDJAXVeAw-YihVMNtXt-uEs&r=Vjtt0vs_iqoI31UfJxBl7yv9I2FeiaeAYgMTLKRBc_I&m=yFSuLWBBbLWZDrPaxDVOCjO0cX88sn16LdG-wdamPNQ&s=u4_9w_rDhHuBVjADmDeFiOjw0J-dhP810cv6osVA7iw&e= 



More information about the mesa-dev mailing list