Compiling assembly-code on ARMv7: CLang vs. GNU

Описание: Hello everyone! I hope this video has helped solve your questions and issues. This video is shared because a solution has been found for the question/problem. I create videos for questions that have solutions. If you have any other issues, feel free to reach out to me on Instagram:   / ky.emrah  

Below, you can find the text related to the question/problem. In the video, the question will be presented first, followed by the answers. If the video moves too fast, feel free to pause and review the answers. If you need more detailed information, you can find the necessary sources and links at the bottom of this description. I hope this video has been helpful, and even if it doesn't directly solve your problem, it will guide you to the source of the solution. I'd appreciate it if you like the video and subscribe to my channel!Compiling assembly-code on ARMv7: CLang vs. GNU

I'm trying to build x265 on an ARMv7 system, and am getting the following error with one of the assembly-sources:
/usr/bin/c++ -mcpu=native -mfloat-abi=hard -mfpu=vfp -marm -DPIC -c .../source/common/arm/blockcopy8.S -o blockcopy8.S.o
.../source/common/arm/blockcopy8.S:835:21: error: expected immediate or register in shift operand

/usr/bin/c++ -mcpu=native -mfloat-abi=hard -mfpu=vfp -marm -DPIC -c .../source/common/arm/blockcopy8.S -o blockcopy8.S.o
.../source/common/arm/blockcopy8.S:835:21: error: expected immediate or register in shift operand

The compiler -- /usr/bin/c++ is clang-16.0.6. Using g++ 13.2.0 the same code compiles without a problem, which must be why the developers have never seen the problem.
The problematic line reads:
/usr/bin/c++
problematic line
mov r0, r0, lsr 16

mov r0, r0, lsr 16

How do I correct it so it compiles with clang just as well as it does with g++?


Tags: assembly,clang,neon,armv7Source of the question:
https://stackoverflow.com/questions/7...

Question and source license information:
https://meta.stackexchange.com/help/l...
https://stackoverflow.com/