From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Huth Date: Fri, 18 Mar 2016 08:41:08 +0000 Subject: Re: [kvm-unit-tests PATCH 3/5] powerpc: check 64bit mode Message-Id: <56EBBF24.1050001@redhat.com> List-Id: References: <1458141183-27207-1-git-send-email-lvivier@redhat.com> <1458141183-27207-4-git-send-email-lvivier@redhat.com> <56EBBC31.6010301@redhat.com> <20160318083838.GC1393@fergus.ozlabs.ibm.com> In-Reply-To: <20160318083838.GC1393@fergus.ozlabs.ibm.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Paul Mackerras Cc: Laurent Vivier , kvm@vger.kernel.org, kvm-ppc@vger.kernel.org, drjones@redhat.com, dgibson@redhat.com, pbonzini@redhat.com On 18.03.2016 09:38, Paul Mackerras wrote: > On Fri, Mar 18, 2016 at 09:28:33AM +0100, Thomas Huth wrote: >> On 16.03.2016 16:13, Laurent Vivier wrote: >>> Check in MSR if the SF bit is set (64bit mode is enabled) >>> >>> Signed-off-by: Laurent Vivier >>> --- >>> powerpc/emulator.c | 14 ++++++++++++++f958ee745f70 >>> 1 file changed, 14 insertions(+) >>> >>> diff --git a/powerpc/emulator.c b/powerpc/emulator.c >>> index 1215c4f..b66c1d7 100644 >>> --- a/powerpc/emulator.c >>> +++ b/powerpc/emulator.c >>> @@ -32,12 +32,26 @@ static void test_illegal(void) >>> report_prefix_pop(); >>> } >>> >>> +static void test_64bit(void) >>> +{ >>> + uint64_t msr; >>> + >>> + report_prefix_push("64bit"); >>> + >>> + asm("mfmsr %[msr]": [msr] "=r" (msr)); >>> + >>> + report("detected", msr & 0x8000000000000000UL); >>> + >>> + report_prefix_pop(); >>> +} >>> + >>> int main(void) >>> { >>> handle_exception(0x700, program_check_handler, (void *)&is_invalid); >>> >>> report_prefix_push("emulator"); >>> >>> + test_64bit(); >>> test_illegal(); >>> >>> report_prefix_pop(); >> >> Reviewed-by: Thomas Huth >> >> But I still wonder how kvm-unit-tests worked at all before the 64-bit >> mode has been fixed in QEMU ... I mean kvm-unit-tests is compiled with >> 64-bit code, but it's still working when run in 32-bit? Something really >> strange was going on here... > > On Power processors, there's not as much difference between 32-bit and > 64-bit mode as you might expect. In 32-bit mode on a 64-bit > processor, you can use all 64 bits of the registers and execute 64-bit > instructions. The only things that change in 32-bit mode are that (a) > effective addresses are truncated to 32 bits and (b) dot-form > instructions (e.g. "add.") set CR0 based on the 32-bit result rather > than the 64-bit result. Ah, ok, ... and since kvm-unit-tests normally does not use the RAM > 4 GB, we didn't notice the truncation. Makes sense. Thanks for the clarification! Thomas