Re: [Csnd-dev] [EXTERNAL] [Csnd-dev] building on RPi...

You’re probably doing something wrong. I added to my Custom.cmake

set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -DPFFFT_SIMD_DISABLE")

and I get this kind of command line

/Library/Developer/CommandLineTools/usr/bin/cc -DCS_DEFAULT_PLUGINDIR=\"/Users/victor/Library/Frameworks/CsoundLib64.framework/Versions/7.0/Resources/Opcodes64\" -DCS_DEFAULT_USER_PLUGINDIR=\"Library/csound/7.0/plugins64\" -DCsoundLib64_EXPORTS -DHAVE_EXECINFO -DHAVE_SOCKETS -DHAVE_SPRINTF_L -DHAVE_STRLCAT -DHAVE_STRTOD_L -DHAVE_STRTOK_R -DMACOSX -DNO_FLTK_THREADS -DPIPES -DSNDFILE_MP3 -DUSE_LIBSNDFILE -DUSE_LRINT -D_CSOUND_RELEASE_ -I/usr/local/include -I/Users/victor/src/csound/./H -I/Users/victor/src/csound/./include -I/Users/victor/src/csound/./Engine -I/Users/victor/src/csound/./util -I/Users/victor/src/csound/. -I/Users/victor/src/csound/debug -I/Users/victor/src/csound/debug/include -I/System/Library/Frameworks -DHAVE_CLOCK_GETTIME  -ffast-math -fomit-frame-pointer -DPFFFT_SIMD_DISABLE


========================
Prof. Victor Lazzarini
Maynooth University
Ireland

> On 14 Nov 2023, at 13:10, Rory Walsh <[email protected]> wrote:
> 
> I don't see any mention of PFFFT here. I've tried set cmake method, and I've also tried -DPFFFT_SIMD_DISABLE=ON but no dice
> 
> /usr/bin/cc -DCS_DEFAULT_PLUGINDIR=\"/usr/local/lib/csound/plugins-6.0\" -DCS_DEFAULT_USER_PLUGINDIR=\".local/lib/csound/6.0/plugins\" -DHAVE_EXECINFO -DHAVE_SOCKETS -DHAVE_STRTOD_L -DHAVE_STRTOK_R -DLINUX -DNO_FLTK_THREADS -DPIPES -D_CSOUND_RELEASE_ -D_GNU_SOURCE -Dcsound_EXPORTS -I/home/rory/sourcecode/csound/./H -I/home/rory/sourcecode/csound/./include -I/home/rory/sourcecode/csound/./Engine -I/home/rory/sourcecode/csound/./util -I/home/rory/sourcecode/csound/. -I/home/rory/sourcecode/csound/build -I/home/rory/sourcecode/csound/build/include -ffast-math -fomit-frame-pointer -DNDEBUG -fvisibility=hidden -std=gnu99 -DSNDFILE_MP3 -O3  -fPIC -D__BUILDING_LIBCSOUND -DPARCS -DHAVE_DIRENT_H -DHAVE_FCNTL_H -DHAVE_UNISTD_H -DHAVE_STDINT_H -DHAVE_SYS_TIME_H -DHAVE_SYS_TYPES_H -DHAVE_TERMIOS_H -DHAVE_VALUES_H -MD -MT CMakeFiles/csound.dir/OOps/midiops.c.o -MF CMakeFiles/csound.dir/OOps/midiops.c.o.d -o CMakeFiles/csound.dir/OOps/midiops.c.o -c /home/rory/sourcecode/csound/OOps/midiops.c
> In file included from /home/rory/sourcecode/csound/OOps/pffft.c:158:
> /usr/lib/gcc/arm-linux-gnueabihf/12/include/arm_neon.h: In function 'passf2_ps':
> /usr/lib/gcc/arm-linux-gnueabihf/12/include/arm_neon.h:1957:1: error: inlining failed in call to 'always_inline' 'vsubq_f32': target specific option mismatch
>  1957 | vsubq_f32 (float32x4_t __a, float32x4_t __b)
> 
> On Tue, 14 Nov 2023 at 13:03, Victor Lazzarini <[email protected]> wrote:
> Show the full cc command that fails here
> 
> Prof. Victor Lazzarini 
> Maynooth University
> Ireland
> 
>> On 14 Nov 2023, at 12:33, Rory Walsh <[email protected]> wrote:
>> 
>>  It seems there is no NEON support for RPi Zero, so I added the following to the Custom.cmake file:
>> 
>> set(CMAKE_FLAGS "${CMAKE_FLAGS} -DPFFFT_SIMD_DISABLE")
>> 
>> but building still fails:
>> 
>> In file included from /home/rory/sourcecode/csound/OOps/pffft.c:158:
>> /usr/lib/gcc/arm-linux-gnueabihf/12/include/arm_neon.h: In function 'passf2_ps':
>> /usr/lib/gcc/arm-linux-gnueabihf/12/include/arm_neon.h:1957:1: error: inlining failed in call to 'always_inline' 'vsubq_f32': target specific option mismatch
>>  1957 | vsubq_f32 (float32x4_t __a, float32x4_t __b)
>>       | ^~~~~~~~~
>> /home/rory/sourcecode/csound/OOps/pffft.c:165:21: note: called from here
>>   165 | #  define VSUB(a,b) vsubq_f32(a,b)
>>       |                     ^~~~~~~~~~~~~~
>> /home/rory/sourcecode/csound/OOps/pffft.c:284:23: note: in expansion of macro 'VSUB'
>>   284 |       ch[l1ido + 1] = VSUB(cc[1], cc[ido+1]);
>>       |                       ^~~~
>> 
>> etc....
>> 
>> On Tue, 14 Nov 2023 at 12:28, Rory Walsh <[email protected]> wrote:
>> Turns out the Custom.cmake set the build back to debug. Thanks, I'm getting further now.
>> 
>> On Tue, 14 Nov 2023 at 12:22, Rory Walsh <[email protected]> wrote:
>> Csound 6. I call build with Release enabled, but the errors persist?   
>> 
>> cmake ---build . --config=Release 
>> 
>> On Tue, 14 Nov 2023 at 12:20, Victor Lazzarini <[email protected]> wrote:
>> Do you want to build Csound 6 or 7?
>> ========================
>> Prof. Victor Lazzarini
>> Maynooth University
>> Ireland
>> 
>> > On 14 Nov 2023, at 12:08, Rory Walsh <[email protected]> wrote:
>> > 
>> > *Warning*
>> > This email originated from outside of Maynooth University's Mail System. Do not reply, click links or open attachments unless you recognise the sender and know the content is safe.
>> > I'm having an issue building on a RPi Zero 2 w. Some warnings are being treated as errors(see below), but I can't seem to disable this behaviour in Cmake even though I have added the following to my Custom.cmake file: 
>> > 
>> > set(CMAKE_COMPILE_WARNING_AS_ERROR OFF)
>> > 
>> > Anyone have any ideas?
>> > 
>> > /home/rory/sourcecode/csound/Engine/fgens.c: In function 'isstrcod':
>> > /home/rory/sourcecode/csound/Engine/fgens.c:46:9: error: unused variable 'sel' [-Werror=unused-variable]
>> >    46 |     int sel = (byte_order()+1)&1;
>> >       |         ^~~
>> > cc1: all warnings being treated as errors
>> >