Re: [PATCH 1/3] selftests: kvm: s390: Add uc_map_unmap VM test case

["Christoph Schlameuss" <schlameuss@linux.ibm.com>]

On Fri Aug 23, 2024 at 10:02 AM CEST, Janosch Frank wrote:
> On 8/19/24 6:03 PM, Christoph Schlameuss wrote:
> > On Fri Aug 16, 2024 at 4:29 PM CEST, Janosch Frank wrote:
> >> On 8/15/24 5:45 PM, Christoph Schlameuss wrote:
> >>> Add a test case verifying basic running and interaction of ucontrol VMs.
> >>> Fill the segment and page tables for allocated memory and map memory on
> >>> first access.
> >>>
> >>> * uc_map_unmap
> >>>     Store and load data to mapped and unmapped memory and use pic segment
> >>>     translation handling to map memory on access.
> >>>
> >>> Signed-off-by: Christoph Schlameuss <schlameuss@linux.ibm.com>
> >>> ---
> >>>    .../selftests/kvm/s390x/ucontrol_test.c       | 120 +++++++++++++++++-
> >>>    1 file changed, 119 insertions(+), 1 deletion(-)
> >>>
> >>
> >>> +static void uc_handle_exit_ucontrol(FIXTURE_DATA(uc_kvm) * self)
> >>> +{
> >>> +	struct kvm_run *run = self->run;
> >>> +
> >>> +	TEST_ASSERT_EQ(KVM_EXIT_S390_UCONTROL, run->exit_reason);
> >>> +	switch (run->s390_ucontrol.pgm_code) {
> >>> +	case PGM_SEGMENT_TRANSLATION:
> >>> +		pr_info("ucontrol pic segment translation 0x%llx\n",
> >>> +			run->s390_ucontrol.trans_exc_code);
> >>> +		/* map / make additional memory available */
> >>> +		struct kvm_s390_ucas_mapping map2 = {
> >>> +			.user_addr = (u64)gpa2hva(self, run->s390_ucontrol.trans_exc_code),
> >>> +			.vcpu_addr = run->s390_ucontrol.trans_exc_code,
> >>> +			.length = VM_MEM_EXT_SIZE,
> >>> +		};
> >>> +		pr_info("ucas map %p %p 0x%llx\n",
> >>> +			(void *)map2.user_addr, (void *)map2.vcpu_addr, map2.length);
> >>> +		TEST_ASSERT_EQ(0, ioctl(self->vcpu_fd, KVM_S390_UCAS_MAP, &map2));
> >>> +		break;
> >>
> >> Why is this necessary if you fix up the mapping in the test?
> >>
> > 
> > This is also used within the uc_skey test to make sure the remap does
> > work after the unmap.
>
> Maybe I'm blind because I'm still recovering but where exactly?

It is literally used in the last line of the test case. Calling uc_handle_exit()
again re-maps previously unmapped memory.

I can try to make that a little bit more obvious.

+	/* unmap and run loop again */
+	TH_LOG("ucas unmap %p %p 0x%llx",
+	       (void *)map2.user_addr, (void *)map2.vcpu_addr, map2.length);
+	rc = ioctl(self->vcpu_fd, KVM_S390_UCAS_UNMAP, &map2);
+	ASSERT_EQ(0, rc)
+		TH_LOG("ucas map result %d not expected, %s", rc, strerror(errno));
+	ASSERT_EQ(0, uc_run_once(self));
+	ASSERT_EQ(3, sync_regs->gprs[0]);
+	ASSERT_EQ(KVM_EXIT_S390_UCONTROL, run->exit_reason);
+	ASSERT_EQ(true, uc_handle_exit(self));  // <--- HERE
+}