Hi Drew,
On 5/13/21 6:43 PM, Andrew Jones wrote:
> On Thu, May 13, 2021 at 07:18:44PM +0200, Andrew Jones wrote:
>> [..]
>> Thanks Alex,
>>
>> I think a better fix is this untested one below, though. If you can test
>> it out and confirm it also resolves the issue, then I'll add this patch
>> to the series.
>>
>> Thanks,
>> drew
>>
>>
>> diff --git a/arm/micro-bench.c b/arm/micro-bench.c
>> index 95c418c10eb4..deafd5695c33 100644
>> --- a/arm/micro-bench.c
>> +++ b/arm/micro-bench.c
>> @@ -273,16 +273,22 @@ static void hvc_exec(void)
>> asm volatile("mov w0, #0x4b000000; hvc #0" ::: "w0");
>> }
>>
>> -static void mmio_read_user_exec(void)
>> +/*
>> + * FIXME: Read device-id in virtio mmio here in order to
>> + * force an exit to userspace. This address needs to be
>> + * updated in the future if any relevant changes in QEMU
>> + * test-dev are made.
>> + */
>> +static void *userspace_emulated_addr;
>> +
>> +static bool mmio_read_user_prep(void)
>> {
>> - /*
>> - * FIXME: Read device-id in virtio mmio here in order to
>> - * force an exit to userspace. This address needs to be
>> - * updated in the future if any relevant changes in QEMU
>> - * test-dev are made.
>> - */
>> - void *userspace_emulated_addr = (void*)0x0a000008;
>> + userspace_emulated_addr = (void*)ioremap(0x0a000008, 8);
>> + return true;
>> +}
>>
>> +static void mmio_read_user_exec(void)
>> +{
>> readl(userspace_emulated_addr);
>> }
>>
>> @@ -309,14 +315,14 @@ struct exit_test {
>> };
>>
>> static struct exit_test tests[] = {
>> - {"hvc", NULL, hvc_exec, NULL, 65536, true},
>> - {"mmio_read_user", NULL, mmio_read_user_exec, NULL, 65536, true},
>> - {"mmio_read_vgic", NULL, mmio_read_vgic_exec, NULL, 65536, true},
>> - {"eoi", NULL, eoi_exec, NULL, 65536, true},
>> - {"ipi", ipi_prep, ipi_exec, NULL, 65536, true},
>> - {"ipi_hw", ipi_hw_prep, ipi_exec, NULL, 65536, true},
>> - {"lpi", lpi_prep, lpi_exec, NULL, 65536, true},
>> - {"timer_10ms", timer_prep, timer_exec, timer_post, 256, true},
>> + {"hvc", NULL, hvc_exec, NULL, 65536, true},
>> + {"mmio_read_user", mmio_read_user_prep, mmio_read_user_exec, NULL, 65536, true},
>> + {"mmio_read_vgic", NULL, mmio_read_vgic_exec, NULL, 65536, true},
>> + {"eoi", NULL, eoi_exec, NULL, 65536, true},
>> + {"ipi", ipi_prep, ipi_exec, NULL, 65536, true},
>> + {"ipi_hw", ipi_hw_prep, ipi_exec, NULL, 65536, true},
>> + {"lpi", lpi_prep, lpi_exec, NULL, 65536, true},
>> + {"timer_10ms", timer_prep, timer_exec, timer_post, 256, true},
>> };
>>
>> struct ns_time {
>>
> I still haven't tested it (beyond compiling), but I've tweaked this a bit.
> You can see it here
>
> https://gitlab.com/rhdrjones/kvm-unit-tests/-/commit/71938030d160e021db3388037d0d407df17e8e5e
>
> The whole v4 of this series is here
>
> https://gitlab.com/rhdrjones/kvm-unit-tests/-/commits/efiprep
Had a look at the patch, looks good; in my suggestion I wrongly thought that readl
reads a long (64 bits), not an uint32_t value:
Reviewed-by: Alexandru Elisei <alexandru.elisei@xxxxxxx>
I also ran some tests on the v4 series from your repo.
Qemu TCG on x86 machine:
- arm compiled with arm-linux-gnu-gcc and arm-none-eabi-gcc
- arm64, 4k and 64k pages.
Odroid-c4:
- arm, both compilers, under kvmtool
- arm64, 4k, 16k and 64k pages under qemu KVM and kvmtool
Rockpro64:
- arm, both compilers, under kvmtool
- arm64, 4k and 64k pages, under qemu KVM and kvmtool.
The ITS migration tests I had to run manually on the rockpro64 (Odroid has a
gicv2) because it looks like the run script wasn't detecting the prompt to start
migration. I'm guessing something on my side, because I had issues with the
migration tests before. Nonetheless, those tests ran just fine manually under qemu
and kvmtool, so everything looks correct to me:
Tested-by: Alexandru Elisei <alexandru.elisei@xxxxxxx>
Thanks,
Alex
