Re: [kvm-unit-tests PATCH 3/9] s390x: uv-host: Fence a destroy cpu test on z15

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 22/09/2021 09.18, Janosch Frank wrote:
Firmware will not give us the expected return code on z15 so let's
fence it for the z15 machine generation.

Signed-off-by: Janosch Frank <frankja@xxxxxxxxxxxxx>
---
  lib/s390x/asm/arch_def.h | 14 ++++++++++++++
  s390x/uv-host.c          | 11 +++++++----
  2 files changed, 21 insertions(+), 4 deletions(-)

diff --git a/lib/s390x/asm/arch_def.h b/lib/s390x/asm/arch_def.h
index aa80d840..c8d2722a 100644
--- a/lib/s390x/asm/arch_def.h
+++ b/lib/s390x/asm/arch_def.h
@@ -219,6 +219,20 @@ static inline unsigned short stap(void)
  	return cpu_address;
  }
+#define MACHINE_Z15A 0x8561
+#define MACHINE_Z15B	0x8562
+
+static inline uint16_t get_machine_id(void)
+{
+	uint64_t cpuid;
+
+	asm volatile("stidp %0" : "=Q" (cpuid));
+	cpuid = cpuid >> 16;
+	cpuid &= 0xffff;
+
+	return cpuid;
+}
+
  static inline int tprot(unsigned long addr)
  {
  	int cc;
diff --git a/s390x/uv-host.c b/s390x/uv-host.c
index 66a11160..5e351120 100644
--- a/s390x/uv-host.c
+++ b/s390x/uv-host.c
@@ -111,6 +111,7 @@ static void test_config_destroy(void)
  static void test_cpu_destroy(void)
  {
  	int rc;
+	uint16_t machineid = get_machine_id();
  	struct uv_cb_nodata uvcb = {
  		.header.len = sizeof(uvcb),
  		.header.cmd = UVC_CMD_DESTROY_SEC_CPU,
@@ -125,10 +126,12 @@ static void test_cpu_destroy(void)
  	       "hdr invalid length");
  	uvcb.header.len += 8;
- uvcb.handle += 1;
-	rc = uv_call(0, (uint64_t)&uvcb);
-	report(rc == 1 && uvcb.header.rc == UVC_RC_INV_CHANDLE, "invalid handle");
-	uvcb.handle -= 1;
+	if (machineid != MACHINE_Z15A && machineid != MACHINE_Z15B) {
+		uvcb.handle += 1;
+		rc = uv_call(0, (uint64_t)&uvcb);
+		report(rc == 1 && uvcb.header.rc == UVC_RC_INV_CHANDLE, "invalid handle");
+		uvcb.handle -= 1;
+	}

So this is a bug in the firmware? Any chance that it will still get fixed for the z15? If so, would it make sense to turn this into a report_xfail() instead?

 Thomas




[Index of Archives]     [KVM ARM]     [KVM ia64]     [KVM ppc]     [Virtualization Tools]     [Spice Development]     [Libvirt]     [Libvirt Users]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite Questions]     [Linux Kernel]     [Linux SCSI]     [XFree86]

  Powered by Linux