CR-Client: Enabling 150Cay to build with GCC 4.x

Eric Hyche wrote:
> Greg may want to eyeball these since he reviewed
> the original CR, but they look good to me.

I think they are fine.

--greg.

> 
> Eric
> 
> ==============================================
> Eric Hyche (ehychereal.com)
> Technical Lead
> Embedded Player and Technologies
> RealNetworks, Inc. 
> 
> 
>> -----Original Message-----
>> From: datatype-dev-bounceshelixcommunity.org 
>> [mailto:datatype-dev-bounceshelixcommunity.org] On
Behalf Of 
>> Daniel Yek
>> Sent: Wednesday, July 26, 2006 11:07 PM
>> To: datatype-devhelixcommunity.org; 
>> common-devhelixcommunity.org; audio-devhelixcommunity.org
>> Subject: [datatype-dev] CR-Client: Enabling 150Cay
to build 
>> with GCC 4.x
>>
>>
>> This is the follow-up of my previous CR (that is
postponed 
>> for way too 
>> long...):
>> http://lists.helixcommunity.org/pipermail/common-d
ev/2006-June
>> /003574.html
>>
>> I have done what Greg suggested - compiling the
distribution 
>> components and 
>> tested the functionality.
>>
>>
>> Modified by: dyekreal.com
>> Date: 07/25/2006
>> Project: Helix Player
>>
>> Synopsis: Integrate fixes for g2mp4combo piacpu,
sem_t, and 
>> math64.h into 
>> 150 Cay to fix GCC 4.x issues.
>>
>> Overview:
>> Integrate Greg's fix g2mp4combo piacpu.c: fixed
'ebx' 
>> register clobbering 
>> while using the -fPIC compiler flag.
>>
>> sem_t is now defined as an union, so HXsem_t can no
longer 
>> subclass it.
>> Greg fixed this in HEAD already:
>> http://lists.helixcommunity.org/pipermail/common-d
ev/2006-June
>> /003528.html
>>
>> For more information about math64.h issue, see the
following URL:
>> http://lists.helixcommunity.org/pipermail/audio-de
v/2006-June/
>> 000717.html
>>
>>
>> Files Modified:
>>
datatype-restricted/rm/video/codec/g2mp4combo/env/piacpu.cpp

>> - Fix for 
>> 'ebx' register getting clobbered while using the
-fPIC compiler flag.
>>
>> common/system/pub/platform/unix/pthreadthreads.h -
Define 
>> HXsem_t to be 
>> sem_t, rather than subclassing.
>> audio/fixptutil/pub/math64.h - Integrate Greg's
fixes for GCC 4.x.x
>>
>>
>> Image Size and Heap Use impact (Client -Only):
>> None
>>
>> Platforms and Profiles Affected:
>> g2mp4combo piacpu fix affects Linux, Unix, VXWORKS
platforms.
>> sem_t fix affects all Unix platforms, math64.h fix
affects 
>> all x86 platforms.
>>
>> Distribution Libraries Affected:
>> g2mp4combo?
>>
>> Distribution library impact and planned action:
>> <Is an update required and if so when will it
occur ?>
>>
>> Platforms and Profiles Build Verified:
>> Linux, hxclient_1_5_0_cayenne_restricted BIF, 
>> helix-client-all-defines profile.
>> Linux, hxplay_gtk_current BIF,
helix-client-all-defines-free profile.
>>
>> Platforms and Profiles Functionality verified:
>> Linux, hxclient_1_5_0_cayenne_restricted BIF, 
>> helix-client-all-defines, splay.
>>
>> Branch: hxclient_1_5_0_cayenne
>>
>> Copyright assignment: I am a RealNetworks employee.
>>
>>
>>
>> Index: piacpu.cpp
>>
============================================================
=======
>> RCS file: 
>>
/cvsroot/rarvcode-mpeg4combo/codec/g2mp4combo/env/piacpu.cpp
,v
>> retrieving revision 1.2.2.1
>> diff -u -w -r1.2.2.1 piacpu.cpp
>> --- piacpu.cpp  27 Apr 2005 19:31:57 -0000     
1.2.2.1
>> +++ piacpu.cpp  26 Jul 2006 03:14:51 -0000
>>  -242,6 +242,7 
>>          I32 has_mmx = 0, has_kni = 0, has_sse2 =
0;
>>
>>       __asm__ __volatile ("          \
>> +      push %%ebx                \n \
>>         pushfl                    \n \
>>         popl %%eax                \n \
>>         movl %%eax, %%ebx         \n \
>>  -282,10 +283,11 
>>         and $0x4000000, %%eax     \n \
>>         mov %%eax, %5             \n \
>>   jumphere:                       \n \
>> +      pop %%ebx                 \n \
>>   "
>>   :"=m" (family),"=m"
(model), "=m" (stepping), "=m"
(has_mmx), "=m" 
>> (has_kni), "=m" (has_sse2)
>>   : /*no input*/
>> -:
"eax","ebx","ecx","
;edx"
>> +:
"eax","ecx","edx"
>>   );
>>
>>       cpu_info.ia.family     =
PIA_IA_CPU_Family(family);
>>
>>
>>
>>
>>
>> Index: pthreadthreads.h
>>
============================================================
=======
>> RCS file:
/cvsroot/common/system/pub/platform/unix/pthreadthreads.h,v
>> retrieving revision 1.8
>> diff -u -w -r1.8 pthreadthreads.h
>> --- pthreadthreads.h    27 Oct 2004 23:41:10 -0000 
    1.8
>> +++ pthreadthreads.h    27 Jul 2006 02:23:37 -0000
>>  -102,10 +102,7 
>>
>>   #ifndef _MAC_UNIX
>>
>> -struct HXsem_t : public sem_t
>> -{
>> -    char padding[64]; /* Flawfinder: ignore */ //
different 
>> linux versions 
>> have different binary reps blechhhh!
>> -};
>> +typedef sem_t HXsem_t;
>>
>>   
>>
//==========================================================
==
>> ===========
>>   //
>>
>>
>>
>>
>> Index: math64.h
>>
============================================================
=======
>> RCS file: /cvsroot/audio/fixptutil/pub/math64.h,v
>> retrieving revision 1.26.2.6
>> diff -u -w -r1.26.2.6 math64.h
>> --- math64.h    30 Apr 2006 18:50:29 -0000     
1.26.2.6
>> +++ math64.h    27 Jul 2006 02:48:56 -0000
>>  -1,5 +1,5 
>>   /* ***** BEGIN LICENSE BLOCK *****
>> - * Source last modified: $Id: math64.h,v 1.26.2.6
2006/04/30 
>> 18:50:29 
>> abockover Exp $
>> + * Source last modified: $Id: math64.h,v 1.31
2006/06/06 
>> 22:14:02 gwright 
>> Exp $
>>    *
>>    * Portions Copyright (c) 1995-2004 RealNetworks,
Inc. All 
>> Rights Reserved.
>>    *
>>  -123,63 +123,37 
>>   // GCC / i386
>>   
>>
////////////////////////////////////////////////////////////
//
>> /////////////////////////
>>
>> -#elif !defined(_MAC_UNIX) &&
defined(__GNUC__) && 
>> (defined(__i386__) || 
>> defined(__amd64__)) && !defined(_NO_GNU_AS)
>> +#elif !defined(_MAC_UNIX) &&
defined(__GNUC__) && 
>> (defined(__i386__) || 
>> defined(__amd64__))
>>
>>   #define HAVE_PLATFORM_MACROS
>>
>>   /* Compute a * b / c, using 64-bit intermediate
result */
>>   static __inline__ int MulDiv64(register int x,
register int 
>> y, register 
>> int z)
>>   {
>> -    /* we specify four alternatives here, one for
each 
>> permutation of 
>> memory or
>> -       register operand in the multiplier and the
divisor. All are 
>> commutative in
>> -       the multiplication arguments, one of which
needs to 
>> be in eax when we
>> -       start. */
>> -
>> -    __asm__ volatile ("imull %2\n\t"
>> -                      "idivl %3\n"
>> -                      : "+a,a,a,a" (x)
>> -                      : "%0,%0,%0,%0"
(x), "m,r,m,r" (y), 
>> "m,m,r,r" (z)
>> -                      : "edx") ;
>> -    return x ;
>> +    return (int)(((INT64)x*(INT64)y)/(INT64)z);
>>   }
>>
>>   /* Compute (a * b) >> 32, using 64-bit
intermediate result */
>>   static __inline__ int MulShift32(int x, int y)
>>   {
>> -    int z ;
>> -    /* we specify two alternatives here. The first
one can read the 
>> multiplier from
>> -       memory, the second from from a register.
Both return 
>> the result in 
>> eax,edx
>> -       and are commutative in the arguments, one
of which 
>> needs to be in 
>> eax when we
>> -       start. */
>> -    __asm__ volatile ("imull %3" :
"=d,d" (z), "+a,a" (x): 
>> "%1,1" (x), 
>> "m,r" (y)) ;
>> -    return z ;
>> +    return (int)(((INT64)x*(INT64)y)>>32);
>>   }
>>
>>   /* Compute (a * b) >> 31, using 64-bit
intermediate result */
>>   static __inline__ int MulShift31(int x, int y)
>>   {
>> -    int zhi ;
>> -    __asm__ volatile ("imull %3\n\t"
>> -                      "shrdl
$31,%1,%0": "+a,a" (x), "=d,d"
(zhi) : 
>> "%0,%0" (x), "m,r" (y)) ;
>> -    return x ;
>> +    return (int)(((INT64)x*(INT64)y)>>31);
>>   }
>>
>> -/* Compute (a * b) >> 30, using 64-bit
intermediate result */
>>   static __inline__ int MulShift30(int x, int y)
>>   {
>> -    int zhi ;
>> -    __asm__ volatile ("imull %3\n\t"
>> -                      "shrdl
$30,%1,%0" : "+a,a" (x),
"=d,d" (zhi) : 
>> "%0,%0" (x), "m,r" (y)) ;
>> -    return x ;
>> +    return (int)(((INT64)x*(INT64)y)>>30);
>>   }
>>
>>   /* Compute (a * b) >> n, using 64-bit
intermediate result */
>>   static __inline__ int MulShiftN(register int x,
register 
>> int y, register 
>> int n)
>>   {
>> -    int zhi ;
>> -    __asm__ volatile ("imull %3\n\t"
>> -                      "shrdl
%%cl,%1,%0" : "+a,a" (x), 
>> "=d,d" (zhi) : 
>> "%0,%0" (x), "m,r" (y),
"c,c" (n)) ;
>> -    return x ;
>> +    return (int)(((INT64)x*(INT64)y)>>n);
>>   }
>>
>>   #ifdef TIMING
>>  -196,6 +170,9 
>>   #endif
>>
>>   #ifdef DEBUG
>> +#  ifdef ASSERT
>> +#    undef ASSERT
>> +#  endif
>>   #define ASSERT(x) if (!(x)) __asm__ __volatile
("int $3" :: )
>>   #endif
>>
>>
>>
>>
>> -- 
>> Daniel Yek
>>
>>
>> _______________________________________________
>> Datatype-dev mailing list
>> Datatype-devhelixcommunity.org
>> http://lists.helixcommunity.org/mailman/listinfo/da
tatype-dev
>>
> 
> 
> _______________________________________________
> Datatype-dev mailing list
> Datatype-devhelixcommunity.org
> http://lists.helixcommunity.org/mailman/listinfo/da
tatype-dev
> 

_______________________________________________
Common-dev mailing list
Common-devhelixcommunity.org
http://lists.helixcommunity.org/mailman/listinfo/comm
on-dev

CR-Client: Enabling 150Cay to build with GCC 4.x

Thread: CR-Client: Enabling 150Cay to build with GCC 4.x