Subject: [PATCH] build: use CPU tuning supplied by the build environment

The CFFI build appends -march=native and -mtune=native based on the build
host. These options fail with the x86_64-hosted AArch64 cross compiler.

Remove the host CPU selection, including the PowerPC variant, and let
OpenEmbedded provide target tuning through its compiler and flags.
Keep the remaining compiler options and audio functionality unchanged.

Upstream-Status: Inappropriate [OE-specific compiler tuning]
---
 build_ffi_module.py | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/build_ffi_module.py b/build_ffi_module.py
--- a/build_ffi_module.py
+++ b/build_ffi_module.py
@@ -793,13 +793,7 @@
     compiler_args = ["-g1", "-O3", "-ffast-math"]
     libraries = ["m", "pthread", "dl"]
 
-    # determine appropriate compiler tuning options
-    if platform.system() == "Darwin":
-        compiler_args += []         # apple compiler doesn't seem to like any tune/cpu options
-    elif "ppc64" in platform.system():
-        compiler_args += ["-mcpu=native"]
-    else:
-        compiler_args += ["-mtune=native", "-march=native"]
+    # CPU and architecture flags are provided by the build environment.
 
 
 ffibuilder.set_source("_miniaudio", """