On 18/12/2019 13:01, Paolo Bonzini wrote:
> So far, specifying an accelerator that was not compiled in did not result
> in an error; fix that.
>
> While at it, clarify the mysterious "Back to TCG" message.
Every time I run a non native VM I have this message:
$ uname -m
x86_64
$ ./ppc64-softmmu/qemu-system-ppc64
qemu-system-ppc64: invalid accelerator kvm
qemu-system-ppc64: falling back to tcg
Is that expected?
At least it's annoying... it was silently falling back to tcg before
that patch.
Should the default be set to tcg to avoid the fallback?
Thanks,
Laurent
> Reviewed-by: Marc-André Lureau <marcandre.lur...@redhat.com>
> Signed-off-by: Paolo Bonzini <pbonz...@redhat.com>
> ---
> vl.c | 7 ++++++-
> 1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/vl.c b/vl.c
> index 0f620be..c350eef 100644
> --- a/vl.c
> +++ b/vl.c
> @@ -2723,6 +2723,8 @@ static int do_configure_accelerator(void *opaque,
> QemuOpts *opts, Error **errp)
> int ret;
>
> if (!ac) {
> + *p_init_failed = true;
> + error_report("invalid accelerator %s", acc);
> return 0;
> }
> ret = accel_init_machine(ac, current_machine);
> @@ -2777,6 +2779,9 @@ static void configure_accelerators(const char *progname)
> */
> if (accel_find(*tmp)) {
> qemu_opts_parse_noisily(qemu_find_opts("accel"), *tmp, true);
> + } else {
> + init_failed = true;
> + error_report("invalid accelerator %s", *tmp);
> }
> }
> } else {
> @@ -2796,7 +2801,7 @@ static void configure_accelerators(const char *progname)
>
> if (init_failed) {
> AccelClass *ac = ACCEL_GET_CLASS(current_machine->accelerator);
> - error_report("Back to %s accelerator", ac->name);
> + error_report("falling back to %s", ac->name);
> }
>
> if (use_icount && !(tcg_enabled() || qtest_enabled())) {
>
