Hi Drew, On Mon, Jan 13, 2025 at 04:29:21PM +0100, Andrew Jones wrote: > On Fri, Jan 10, 2025 at 01:58:48PM +0000, Alexandru Elisei wrote: > > Newer architecture features are supported by qemu TCG on newer CPUs. When > > writing a test for such architecture features, it is necessary to pass the > > correct -cpu argument to qemu. Make it easier on users and test authors > > alike by making 'max' the default value for --processor. The 'max' CPU > > model contains all the features of the cortex-a57 CPU (the old default), so > > no regression should be possible. > > > > A side effect is that, by default, the compiler will not receive a -mcpu > > argument for compiling the code. The expectation is that this is fine, > > since support for -mcpu=$PROCESSOR has only been added for arm64 in the > > last commit. > > > > The default for arm (cortex-a15) has been kept unchanged, because passing > > --processor=max will cause compilation to break. If the user wants the qemu > > CPU model to be 'max', the user will also have to supply a suitable compile > > CPU target via --cflags=-mcpu=<cpu> configure option. > > > > Signed-off-by: Alexandru Elisei <alexandru.elisei@xxxxxxx> > > --- > > configure | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > diff --git a/configure b/configure > > index 46964d36a7d8..3ab0ec208e10 100755 > > --- a/configure > > +++ b/configure > > @@ -14,7 +14,7 @@ function get_default_processor() > > default_processor="cortex-a15" > > ;; > > "arm64" | "aarch64") > > - default_processor="cortex-a57" > > + default_processor="max" > > ;; > > *) > > default_processor=$arch > > -- > > 2.47.1 > > > > Another reason to introduce a new parameter (qemu_cpu) is that we can also > change arm32 to 'max', reducing divergence between arm32 and arm64. That sounds very sensible, I'll do that for the next iteration. Thanks, Alex
