[kvm-unit-tests 0/2] s390x: Fixup test fencing

[kvm-unit-tests 0/2] s390x: Fixup test fencing

[Janosch Frank]

Migration tests have been fenced for way too long in TCG due to a TCG
bug that has been fixed ages ago.

The skey access tests will always fail in TCG so it makes more sense
to fence them in the test itself rather than to filter the CI results
so that humans won't see the fail.

Janosch Frank (2):
  s390x: unittests: Reenable tcg for migration tests
  s390x: skey: Fence TCG protection checks

 s390x/skey.c        | 8 ++++++++
 s390x/unittests.cfg | 5 -----
 2 files changed, 8 insertions(+), 5 deletions(-)

-- 
2.53.0

[kvm-unit-tests 1/2] s390x: unittests: Reenable tcg for migration tests

[Janosch Frank]

The TCG fixes for this issue have been in QEMU for quite a while.
It's time to reenable these tests.

QEMU fix discussion:
https://lore.kernel.org/qemu-devel/20240312201458.79532-1-philmd@linaro.org/

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
---
 s390x/unittests.cfg | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/s390x/unittests.cfg b/s390x/unittests.cfg
index 8c39d4f1..64bc998d 100644
--- a/s390x/unittests.cfg
+++ b/s390x/unittests.cfg
@@ -15,19 +15,14 @@ test_args = 'test 123'
 [selftest-migration]
 file = selftest-migration.elf
 groups = selftest migration
-# TODO: Remove accel=kvm once the following TCG migration fix has been merged:
-# https://lore.kernel.org/qemu-devel/20240219061731.232570-1-npiggin@gmail.com/
-accel = kvm
 
 [selftest-migration-skip]
 file = selftest-migration.elf
 groups = selftest migration
 test_args = "skip"
 
-# This fails due to a QEMU TCG bug so KVM-only until QEMU is fixed upstream
 [migration-memory]
 file = memory-verify.elf
-accel = kvm
 groups = migration
 
 [intercept]
-- 
2.53.0

[kvm-unit-tests 2/2] s390x: skey: Fence TCG protection checks

[Janosch Frank]

TCG does not implement the skey memory protection, skip those tests
under TCG instead of filtering them in the CI logs.

This way the test will PASS when run by humans and there's no
confusion why it fails anymore.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
---
 s390x/skey.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/s390x/skey.c b/s390x/skey.c
index bb769730..1fc95857 100644
--- a/s390x/skey.c
+++ b/s390x/skey.c
@@ -14,6 +14,7 @@
 #include <vmalloc.h>
 #include <css.h>
 #include <mmu.h>
+#include <hardware.h>
 #include <asm/page.h>
 #include <asm/facility.h>
 #include <asm/mem.h>
@@ -736,6 +737,13 @@ int main(void)
 	test_set();
 	test_set_mb();
 	test_chg();
+
+	if  (detect_host() == HOST_IS_TCG) {
+		report_skip("No actual access protection in TCG for skeys.");
+		report_prefix_pop();
+		return report_summary();
+	}
+
 	test_test_protection();
 	test_store_cpu_address();
 	test_diag_308();
-- 
2.53.0

Re: [kvm-unit-tests 2/2] s390x: skey: Fence TCG protection checks

[Claudio Imbrenda]

On Tue,  4 Aug 2026 14:16:29 +0000
Janosch Frank <frankja@linux.ibm.com> wrote:

> TCG does not implement the skey memory protection, skip those tests
> under TCG instead of filtering them in the CI logs.
> 
> This way the test will PASS when run by humans and there's no
> confusion why it fails anymore.
> 
> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>

Reviewed-by: Claudio Imbrenda <imbrenda@linux.ibm.com>

> ---
>  s390x/skey.c | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff --git a/s390x/skey.c b/s390x/skey.c
> index bb769730..1fc95857 100644
> --- a/s390x/skey.c
> +++ b/s390x/skey.c
> @@ -14,6 +14,7 @@
>  #include <vmalloc.h>
>  #include <css.h>
>  #include <mmu.h>
> +#include <hardware.h>
>  #include <asm/page.h>
>  #include <asm/facility.h>
>  #include <asm/mem.h>
> @@ -736,6 +737,13 @@ int main(void)
>  	test_set();
>  	test_set_mb();
>  	test_chg();
> +
> +	if  (detect_host() == HOST_IS_TCG) {
> +		report_skip("No actual access protection in TCG for skeys.");
> +		report_prefix_pop();
> +		return report_summary();
> +	}
> +
>  	test_test_protection();
>  	test_store_cpu_address();
>  	test_diag_308();