On 18.09.19 11:47, Alex Bennée wrote:
>
> David Hildenbrand <da...@redhat.com> writes:
>
>> Let's add the simple test based on the example from the PoP.
>>
>> Signed-off-by: David Hildenbrand <da...@redhat.com>
>> ---
>> tests/tcg/s390x/Makefile.target | 1 +
>> tests/tcg/s390x/mvo.c | 25 +++++++++++++++++++++++++
>> 2 files changed, 26 insertions(+)
>> create mode 100644 tests/tcg/s390x/mvo.c
>>
>> diff --git a/tests/tcg/s390x/Makefile.target
>> b/tests/tcg/s390x/Makefile.target
>> index 151dc075aa..6a3bfa8b29 100644
>> --- a/tests/tcg/s390x/Makefile.target
>> +++ b/tests/tcg/s390x/Makefile.target
>> @@ -6,3 +6,4 @@ TESTS+=ipm
>> TESTS+=exrl-trt
>> TESTS+=exrl-trtr
>> TESTS+=pack
>> +TESTS+=mvo
>> diff --git a/tests/tcg/s390x/mvo.c b/tests/tcg/s390x/mvo.c
>> new file mode 100644
>> index 0000000000..5546fe2a97
>> --- /dev/null
>> +++ b/tests/tcg/s390x/mvo.c
>> @@ -0,0 +1,25 @@
>> +#include <stdint.h>
>> +#include <stdio.h>
>> +
>> +int main(void)
>> +{
>> + uint8_t dest[6] = {0xff, 0x77, 0x88, 0x99, 0x0c, 0xff};
>> + uint8_t src[5] = {0xee, 0x12, 0x34, 0x56, 0xee};
>> + uint8_t expected[6] = {0xff, 0x01, 0x23, 0x45, 0x6c, 0xff};
>> + int i;
>> +
>> + asm volatile (
>> + " mvo 0(4,%[dest]),0(3,%[src])\n"
>> + :
>> + : [dest] "d" (dest + 1),
>> + [src] "d" (src + 1)
>> + : "memory");
>> +
>> + for (i = 0; i < sizeof(expected); i++) {
>> + if (dest[i] != expected[i]) {
>> + fprintf(stderr, "bad data\n");
>> + return 1;
>> + }
>> + }
>> + return 0;
>> +}
>
> Reviewed-by: Alex Bennée <alex.ben...@linaro.org>
>
> but...
>
> can this test be expanded to check the page cross cases that caused you
> so much trouble to track down?
I might add a MVC test that tries to reproduce this. But with
speculative page faults and things like that it might not be very easy
to reproduce. However, I can give it a try.
Thanks!
>
> --
> Alex Bennée
>
--
Thanks,
David / dhildenb
