| 
     
      
      
      From: Barry J. <zen...@ze...> - 2021-05-13 16:24:01
       
   | 
Relevant lines from build log:
-------------------------------
libtool: compile:  g++ -std=gnu++14 -DHAVE_CONFIG_H -I. -I../.. 
-Wreturn-type -ffast-math -O2 -g -pipe -Wformat -Werror=format-security 
-Wp,-D_FORTIFY_SOURCE=2 -fstack-protector --param=ssp-buffer-size=4 
-fasynchronous-unwind-tables -pthread -c RTMath.cpp  -fPIC -DPIC -o 
.libs/RTMath.o
RTMath.cpp:83:8: error: #error "Sorry, LinuxSampler lacks time stamp 
code for your system."
    83 |     #  error "Sorry, LinuxSampler lacks time stamp code for 
your system."
       |        ^~~~~
RTMath.cpp:84:8: error: #error "Please report this error and the CPU you 
are using to the LinuxSampler developers mailing list!"
    84 |     #  error "Please report this error and the CPU you are 
using to the LinuxSampler developers mailing list!"
       |        ^~~~~
In file included from RTMath.h:29,
                  from RTMath.cpp:24:
global_private.h:99:40: warning: dynamic exception specifications are 
deprecated in C++11 [-Wdeprecated]
    99 | inline int ToInt(const std::string& s) 
throw(LinuxSampler::Exception) {
       |                                        ^~~~~
global_private.h:106:44: warning: dynamic exception specifications are 
deprecated in C++11 [-Wdeprecated]
   106 | inline float ToFloat(const std::string& s) 
throw(LinuxSampler::Exception) {
       |                                            ^~~~~
RTMath.cpp: In static member function 'static RTMathBase::time_stamp_t 
RTMathBase::CreateTimeStamp()':
RTMath.cpp:86:1: warning: no return statement in function returning 
non-void [-Wreturn-type]
    86 | }
       | ^
make[3]: *** [Makefile:558: RTMath.lo] Error 1
-----------------------------------------------
We have two patches applied:
linuxsampler-2.1-mga-arm_RTMath.cpp.patch
linuxsampler-2.1-mga-arm_atomic.h.patch
Not certain which build machine was used but one is Amazon Graviton 2 
and the others are older Graviton.
Any help appreciated.
Thanks,
Barry Jackson
 | 
| 
     
      
      
      From: Barry J. <zen...@ze...> - 2021-05-13 16:37:00
       
   | 
On 13/05/2021 17:23, Barry Jackson wrote: > Not certain which build machine was used but one is Amazon Graviton 2 > and the others are older Graviton. It was the Graviton 2 on further investigation.  | 
| 
     
      
      
      From: Christian S. <sch...@li...> - 2021-05-13 17:23:52
       
   | 
On Donnerstag, 13. Mai 2021 18:23:49 CEST Barry Jackson wrote: > Relevant lines from build log: > ------------------------------- > libtool: compile: g++ -std=gnu++14 -DHAVE_CONFIG_H -I. -I../.. > -Wreturn-type -ffast-math -O2 -g -pipe -Wformat -Werror=format-security > -Wp,-D_FORTIFY_SOURCE=2 -fstack-protector --param=ssp-buffer-size=4 > -fasynchronous-unwind-tables -pthread -c RTMath.cpp -fPIC -DPIC -o > .libs/RTMath.o > RTMath.cpp:83:8: error: #error "Sorry, LinuxSampler lacks time stamp > code for your system." > 83 | # error "Sorry, LinuxSampler lacks time stamp code for > your system." > > | ^~~~~ Already filed: https://bugs.linuxsampler.org/cgi-bin/show_bug.cgi?id=193 I think some people are using a std lib call to generate a time stamp instead, which is not real-time safe though. A CPU/asm solution for generating the time stamp would be better. Best regards, Christian Schoenebeck  | 
| 
     
      
      
      From: Barry J. <zen...@ze...> - 2021-05-13 20:20:42
       
   | 
On 13/05/2021 18:23, Christian Schoenebeck wrote: > Already filed: > https://bugs.linuxsampler.org/cgi-bin/show_bug.cgi?id=193 > > I think some people are using a std lib call to generate a time stamp instead, > which is not real-time safe though. A CPU/asm solution for generating the time > stamp would be better. > That bug is over 6 years old! So for now we just drop gig/linuxsampler/gigedit from all arm builds I guess. Is a solution being worked on? Regards, Barry  | 
| 
     
      
      
      From: Christian S. <sch...@li...> - 2021-05-14 09:22:00
       
   | 
On Donnerstag, 13. Mai 2021 22:20:26 CEST Barry Jackson wrote: > On 13/05/2021 18:23, Christian Schoenebeck wrote: > > Already filed: > > https://bugs.linuxsampler.org/cgi-bin/show_bug.cgi?id=193 > > > > I think some people are using a std lib call to generate a time stamp > > instead, which is not real-time safe though. A CPU/asm solution for > > generating the time stamp would be better. > > That bug is over 6 years old! > > So for now we just drop gig/linuxsampler/gigedit from all arm builds I > guess. Makes sense. > Is a solution being worked on? Not that I know of. The way to go is to read out the cpu cycle register, which is the fastest solution and real-time safe. The other asm code mentioned should already been taken care of. Best regards, Christian Schoenebeck  |