meson.build: Don't misdetect posix_memalign() on Windows

Currently we incorrectly think that posix_memalign() exists on
Windows.  This is because of a combination of:

 * the msys2/mingw toolchain/libc claim to have a
   __builtin_posix_memalign when there isn't a builtin of that name
 * meson will assume that if you have a __builtin_foo that
   counts for has_function('foo')

Specifying a specific include file via prefix: causes meson to not
treat builtins as sufficient and actually look for the function
itself; see this meson pull request which added that as the official
way to get the right answer:
  mesonbuild/meson#1150

Currently this misdectection doesn't cause problems because we only
use CONFIG_POSIX_MEMALIGN in oslib-posix.c; however that will change
in a following commit.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20220226180723.1706285-6-peter.maydell@linaro.org
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>

meson.build

@@ -1619,7 +1619,9 @@ config_host_data.set('CONFIG_CLOCK_ADJTIME', cc.has_function('clock_adjtime'))
 config_host_data.set('CONFIG_DUP3', cc.has_function('dup3'))
 config_host_data.set('CONFIG_FALLOCATE', cc.has_function('fallocate'))
 config_host_data.set('CONFIG_POSIX_FALLOCATE', cc.has_function('posix_fallocate'))
-config_host_data.set('CONFIG_POSIX_MEMALIGN', cc.has_function('posix_memalign'))
+# Note that we need to specify prefix: here to avoid incorrectly
+# thinking that Windows has posix_memalign()
+config_host_data.set('CONFIG_POSIX_MEMALIGN', cc.has_function('posix_memalign', prefix: '#include <stdlib.h>'))
 config_host_data.set('CONFIG_PPOLL', cc.has_function('ppoll'))
 config_host_data.set('CONFIG_PREADV', cc.has_function('preadv', prefix: '#include <sys/uio.h>'))
 config_host_data.set('CONFIG_SEM_TIMEDWAIT', cc.has_function('sem_timedwait', dependencies: threads))