[kvm-unit-tests PATCH v3 0/9] s390x: smp: Improve smp code and reset checks

[kvm-unit-tests PATCH v3 0/9] s390x: smp: Improve smp code and reset checks

[Janosch Frank]

Let's extend sigp reset testing and clean up the smp library as well.

GIT: https://github.com/frankjaa/kvm-unit-tests/tree/smp_cleanup

v3:
	* Added patch to introduce cpu loop in cpu setup
	* Added patch that removes cpu loops in favor of the previously introduced one
	* Fixed inline assembly for fpc dirtying
	* Moved cpu stop hunk from first into the second patch
	* Reworked patch #4 commit message and added a comment when waiting for PU

v2:
	* Added cpu stop to test_store_status()
	* Added smp_cpu_destroy() to the end of smp.c main()
	* New patch that prints cpu id on interrupt errors
	* New patch that reworks cpu start in the smp library (needed for lpar)
	* nullp is now an array

Janosch Frank (9):
  s390x: smp: Cleanup smp.c
  s390x: smp: Only use smp_cpu_setup once
  s390x: Add cpu id to interrupt error prints
  s390x: smp: Rework cpu start and active tracking
  s390x: smp: Wait for cpu setup to finish
  s390x: smp: Loop if secondary cpu returns into cpu setup again
  s390x: smp: Remove unneeded cpu loops
  s390x: smp: Test all CRs on initial reset
  s390x: smp: Dirty fpc before initial reset test

 lib/s390x/interrupt.c | 20 +++++------
 lib/s390x/smp.c       | 52 ++++++++++++++++-----------
 s390x/cstart64.S      |  4 +++
 s390x/smp.c           | 82 +++++++++++++++++++++++++++----------------
 4 files changed, 97 insertions(+), 61 deletions(-)

-- 
2.20.1

[kvm-unit-tests PATCH v3 1/9] s390x: smp: Cleanup smp.c

[Janosch Frank]

Let's remove a lot of badly formatted code by introducing the
wait_for_flag() function.

Also let's remove some stray spaces.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
---
 s390x/smp.c | 42 +++++++++++++++++++++++-------------------
 1 file changed, 23 insertions(+), 19 deletions(-)

diff --git a/s390x/smp.c b/s390x/smp.c
index ab7e46c..8d8e3a5 100644
--- a/s390x/smp.c
+++ b/s390x/smp.c
@@ -22,6 +22,13 @@
 
 static int testflag = 0;
 
+static void wait_for_flag(void)
+{
+	while (!testflag) {
+		mb();
+	}
+}
+
 static void cpu_loop(void)
 {
 	for (;;) {}
@@ -30,20 +37,17 @@ static void cpu_loop(void)
 static void test_func(void)
 {
 	testflag = 1;
-	mb();
 	cpu_loop();
 }
 
 static void test_start(void)
 {
 	struct psw psw;
-	psw.mask =  extract_psw_mask();
+	psw.mask = extract_psw_mask();
 	psw.addr = (unsigned long)test_func;
 
 	smp_cpu_setup(1, psw);
-	while (!testflag) {
-		mb();
-	}
+	wait_for_flag();
 	report(1, "start");
 }
 
@@ -115,24 +119,24 @@ static void ecall(void)
 	testflag = 1;
 	while (lc->ext_int_code != 0x1202) { mb(); }
 	report(1, "ecall");
-	testflag= 1;
+	testflag = 1;
 }
 
 static void test_ecall(void)
 {
 	struct psw psw;
-	psw.mask =  extract_psw_mask();
+	psw.mask = extract_psw_mask();
 	psw.addr = (unsigned long)ecall;
 
 	report_prefix_push("ecall");
-	testflag= 0;
+	testflag = 0;
 	smp_cpu_destroy(1);
 
 	smp_cpu_setup(1, psw);
-	while (!testflag) { mb(); }
-	testflag= 0;
+	wait_for_flag();
+	testflag = 0;
 	sigp(1, SIGP_EXTERNAL_CALL, 0, NULL);
-	while(!testflag) {mb();}
+	wait_for_flag();
 	smp_cpu_stop(1);
 	report_prefix_pop();
 }
@@ -147,7 +151,7 @@ static void emcall(void)
 	mask = extract_psw_mask();
 	mask |= PSW_MASK_EXT;
 	load_psw_mask(mask);
-	testflag= 1;
+	testflag = 1;
 	while (lc->ext_int_code != 0x1201) { mb(); }
 	report(1, "ecall");
 	testflag = 1;
@@ -156,18 +160,18 @@ static void emcall(void)
 static void test_emcall(void)
 {
 	struct psw psw;
-	psw.mask =  extract_psw_mask();
+	psw.mask = extract_psw_mask();
 	psw.addr = (unsigned long)emcall;
 
 	report_prefix_push("emcall");
-	testflag= 0;
+	testflag = 0;
 	smp_cpu_destroy(1);
 
 	smp_cpu_setup(1, psw);
-	while (!testflag) { mb(); }
-	testflag= 0;
+	wait_for_flag();
+	testflag = 0;
 	sigp(1, SIGP_EMERGENCY_SIGNAL, 0, NULL);
-	while(!testflag) { mb(); }
+	wait_for_flag();
 	smp_cpu_stop(1);
 	report_prefix_pop();
 }
@@ -177,7 +181,7 @@ static void test_reset_initial(void)
 	struct cpu_status *status = alloc_pages(0);
 	struct psw psw;
 
-	psw.mask =  extract_psw_mask();
+	psw.mask = extract_psw_mask();
 	psw.addr = (unsigned long)test_func;
 
 	report_prefix_push("reset initial");
@@ -208,7 +212,7 @@ static void test_reset(void)
 {
 	struct psw psw;
 
-	psw.mask =  extract_psw_mask();
+	psw.mask = extract_psw_mask();
 	psw.addr = (unsigned long)test_func;
 
 	report_prefix_push("cpu reset");
-- 
2.20.1

Re: [kvm-unit-tests PATCH v3 1/9] s390x: smp: Cleanup smp.c

[David Hildenbrand]

On 17.01.20 11:46, Janosch Frank wrote:
> Let's remove a lot of badly formatted code by introducing the
> wait_for_flag() function.
> 
> Also let's remove some stray spaces.
> 
> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
> Reviewed-by: Thomas Huth <thuth@redhat.com>
> Reviewed-by: Cornelia Huck <cohuck@redhat.com>
> ---
>  s390x/smp.c | 42 +++++++++++++++++++++++-------------------
>  1 file changed, 23 insertions(+), 19 deletions(-)
> 
> diff --git a/s390x/smp.c b/s390x/smp.c
> index ab7e46c..8d8e3a5 100644
> --- a/s390x/smp.c
> +++ b/s390x/smp.c
> @@ -22,6 +22,13 @@
>  
>  static int testflag = 0;
>  
> +static void wait_for_flag(void)
> +{
> +	while (!testflag) {
> +		mb();
> +	}

No need for the {}


Reviewed-by: David Hildenbrand <david@redhat.com>

-- 
Thanks,

David / dhildenb

[kvm-unit-tests PATCH v3 2/9] s390x: smp: Only use smp_cpu_setup once

[Janosch Frank]

Let's stop and start instead of using setup to run a function on a
cpu.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>
---
 s390x/smp.c | 21 ++++++++++++++-------
 1 file changed, 14 insertions(+), 7 deletions(-)

diff --git a/s390x/smp.c b/s390x/smp.c
index 8d8e3a5..555ed72 100644
--- a/s390x/smp.c
+++ b/s390x/smp.c
@@ -46,7 +46,7 @@ static void test_start(void)
 	psw.mask = extract_psw_mask();
 	psw.addr = (unsigned long)test_func;
 
-	smp_cpu_setup(1, psw);
+	smp_cpu_start(1, psw);
 	wait_for_flag();
 	report(1, "start");
 }
@@ -102,6 +102,7 @@ static void test_store_status(void)
 	report(1, "status written");
 	free_pages(status, PAGE_SIZE * 2);
 	report_prefix_pop();
+	smp_cpu_stop(1);
 
 	report_prefix_pop();
 }
@@ -130,9 +131,8 @@ static void test_ecall(void)
 
 	report_prefix_push("ecall");
 	testflag = 0;
-	smp_cpu_destroy(1);
 
-	smp_cpu_setup(1, psw);
+	smp_cpu_start(1, psw);
 	wait_for_flag();
 	testflag = 0;
 	sigp(1, SIGP_EXTERNAL_CALL, 0, NULL);
@@ -165,9 +165,8 @@ static void test_emcall(void)
 
 	report_prefix_push("emcall");
 	testflag = 0;
-	smp_cpu_destroy(1);
 
-	smp_cpu_setup(1, psw);
+	smp_cpu_start(1, psw);
 	wait_for_flag();
 	testflag = 0;
 	sigp(1, SIGP_EMERGENCY_SIGNAL, 0, NULL);
@@ -185,7 +184,7 @@ static void test_reset_initial(void)
 	psw.addr = (unsigned long)test_func;
 
 	report_prefix_push("reset initial");
-	smp_cpu_setup(1, psw);
+	smp_cpu_start(1, psw);
 
 	sigp_retry(1, SIGP_INITIAL_CPU_RESET, 0, NULL);
 	sigp(1, SIGP_STORE_STATUS_AT_ADDRESS, (uintptr_t)status, NULL);
@@ -216,7 +215,7 @@ static void test_reset(void)
 	psw.addr = (unsigned long)test_func;
 
 	report_prefix_push("cpu reset");
-	smp_cpu_setup(1, psw);
+	smp_cpu_start(1, psw);
 
 	sigp_retry(1, SIGP_CPU_RESET, 0, NULL);
 	report(smp_cpu_stopped(1), "cpu stopped");
@@ -225,6 +224,7 @@ static void test_reset(void)
 
 int main(void)
 {
+	struct psw psw;
 	report_prefix_push("smp");
 
 	if (smp_query_num_cpus() == 1) {
@@ -232,6 +232,12 @@ int main(void)
 		goto done;
 	}
 
+	/* Setting up the cpu to give it a stack and lowcore */
+	psw.mask = extract_psw_mask();
+	psw.addr = (unsigned long)cpu_loop;
+	smp_cpu_setup(1, psw);
+	smp_cpu_stop(1);
+
 	test_start();
 	test_stop();
 	test_stop_store_status();
@@ -240,6 +246,7 @@ int main(void)
 	test_emcall();
 	test_reset();
 	test_reset_initial();
+	smp_cpu_destroy(1);
 
 done:
 	report_prefix_pop();
-- 
2.20.1

[kvm-unit-tests PATCH v3 3/9] s390x: Add cpu id to interrupt error prints

[Janosch Frank]

It's good to know which cpu broke the test.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
---
 lib/s390x/interrupt.c | 20 ++++++++++----------
 1 file changed, 10 insertions(+), 10 deletions(-)

diff --git a/lib/s390x/interrupt.c b/lib/s390x/interrupt.c
index 05f30be..773752a 100644
--- a/lib/s390x/interrupt.c
+++ b/lib/s390x/interrupt.c
@@ -107,8 +107,8 @@ static void fixup_pgm_int(void)
 void handle_pgm_int(void)
 {
 	if (!pgm_int_expected)
-		report_abort("Unexpected program interrupt: %d at %#lx, ilen %d\n",
-			     lc->pgm_int_code, lc->pgm_old_psw.addr,
+		report_abort("Unexpected program interrupt: %d on cpu %d at %#lx, ilen %d\n",
+			     lc->pgm_int_code, stap(), lc->pgm_old_psw.addr,
 			     lc->pgm_int_id);
 
 	pgm_int_expected = false;
@@ -119,8 +119,8 @@ void handle_ext_int(void)
 {
 	if (!ext_int_expected &&
 	    lc->ext_int_code != EXT_IRQ_SERVICE_SIG) {
-		report_abort("Unexpected external call interrupt (code %#x): at %#lx",
-			     lc->ext_int_code, lc->ext_old_psw.addr);
+		report_abort("Unexpected external call interrupt (code %#x): on cpu %d at %#lx",
+			     stap(), lc->ext_int_code, lc->ext_old_psw.addr);
 		return;
 	}
 
@@ -137,18 +137,18 @@ void handle_ext_int(void)
 
 void handle_mcck_int(void)
 {
-	report_abort("Unexpected machine check interrupt: at %#lx",
-		     lc->mcck_old_psw.addr);
+	report_abort("Unexpected machine check interrupt: on cpu %d at %#lx",
+		     stap(), lc->mcck_old_psw.addr);
 }
 
 void handle_io_int(void)
 {
-	report_abort("Unexpected io interrupt: at %#lx",
-		     lc->io_old_psw.addr);
+	report_abort("Unexpected io interrupt: on cpu %d at %#lx",
+		     stap(), lc->io_old_psw.addr);
 }
 
 void handle_svc_int(void)
 {
-	report_abort("Unexpected supervisor call interrupt: at %#lx",
-		     lc->svc_old_psw.addr);
+	report_abort("Unexpected supervisor call interrupt: on cpu %d at %#lx",
+		     stap(), lc->svc_old_psw.addr);
 }
-- 
2.20.1

[kvm-unit-tests PATCH v3 4/9] s390x: smp: Rework cpu start and active tracking

[Janosch Frank]

The architecture specifies that processing sigp orders may be
asynchronous, and this is indeed the case on some hypervisors, so we
need to wait until the cpu runs before we return from the setup/start
function.

As there was a lot of duplicate code, a common function for cpu
restarts has been introduced.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
---
 lib/s390x/smp.c | 50 ++++++++++++++++++++++++++++---------------------
 1 file changed, 29 insertions(+), 21 deletions(-)

diff --git a/lib/s390x/smp.c b/lib/s390x/smp.c
index f57f420..84e681d 100644
--- a/lib/s390x/smp.c
+++ b/lib/s390x/smp.c
@@ -104,35 +104,46 @@ int smp_cpu_stop_store_status(uint16_t addr)
 	return rc;
 }
 
+static int smp_cpu_restart_nolock(uint16_t addr, struct psw *psw)
+{
+	int rc;
+	struct cpu *cpu = smp_cpu_from_addr(addr);
+
+	if (!cpu)
+		return -1;
+	if (psw) {
+		cpu->lowcore->restart_new_psw.mask = psw->mask;
+		cpu->lowcore->restart_new_psw.addr = psw->addr;
+	}
+	rc = sigp(addr, SIGP_RESTART, 0, NULL);
+	if (rc)
+		return rc;
+	/*
+	 * The order has been accepted, but the actual restart may not
+	 * have been performed yet, so wait until the cpu is running.
+	 */
+	while (!smp_cpu_running(addr))
+		mb();
+	cpu->active = true;
+	return 0;
+}
+
 int smp_cpu_restart(uint16_t addr)
 {
-	int rc = -1;
-	struct cpu *cpu;
+	int rc;
 
 	spin_lock(&lock);
-	cpu = smp_cpu_from_addr(addr);
-	if (cpu) {
-		rc = sigp(addr, SIGP_RESTART, 0, NULL);
-		cpu->active = true;
-	}
+	rc = smp_cpu_restart_nolock(addr, NULL);
 	spin_unlock(&lock);
 	return rc;
 }
 
 int smp_cpu_start(uint16_t addr, struct psw psw)
 {
-	int rc = -1;
-	struct cpu *cpu;
-	struct lowcore *lc;
+	int rc;
 
 	spin_lock(&lock);
-	cpu = smp_cpu_from_addr(addr);
-	if (cpu) {
-		lc = cpu->lowcore;
-		lc->restart_new_psw.mask = psw.mask;
-		lc->restart_new_psw.addr = psw.addr;
-		rc = sigp(addr, SIGP_RESTART, 0, NULL);
-	}
+	rc = smp_cpu_restart_nolock(addr, &psw);
 	spin_unlock(&lock);
 	return rc;
 }
@@ -192,10 +203,7 @@ int smp_cpu_setup(uint16_t addr, struct psw psw)
 	lc->sw_int_crs[0] = 0x0000000000040000UL;
 
 	/* Start processing */
-	rc = sigp_retry(cpu->addr, SIGP_RESTART, 0, NULL);
-	if (!rc)
-		cpu->active = true;
-
+	smp_cpu_restart_nolock(addr, NULL);
 out:
 	spin_unlock(&lock);
 	return rc;
-- 
2.20.1

Re: [kvm-unit-tests PATCH v3 4/9] s390x: smp: Rework cpu start and active tracking

[David Hildenbrand]

On 17.01.20 11:46, Janosch Frank wrote:
> The architecture specifies that processing sigp orders may be
> asynchronous, and this is indeed the case on some hypervisors, so we
> need to wait until the cpu runs before we return from the setup/start
> function.
> 
> As there was a lot of duplicate code, a common function for cpu
> restarts has been introduced.
> 
> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
> Reviewed-by: Cornelia Huck <cohuck@redhat.com>
> ---
>  lib/s390x/smp.c | 50 ++++++++++++++++++++++++++++---------------------
>  1 file changed, 29 insertions(+), 21 deletions(-)
> 
> diff --git a/lib/s390x/smp.c b/lib/s390x/smp.c
> index f57f420..84e681d 100644
> --- a/lib/s390x/smp.c
> +++ b/lib/s390x/smp.c
> @@ -104,35 +104,46 @@ int smp_cpu_stop_store_status(uint16_t addr)
>  	return rc;
>  }
>  
> +static int smp_cpu_restart_nolock(uint16_t addr, struct psw *psw)
> +{
> +	int rc;
> +	struct cpu *cpu = smp_cpu_from_addr(addr);

I'd exchange these two (reverse christmas tree)

> +
> +	if (!cpu)
> +		return -1;

-EINVAL?

> +	if (psw) {
> +		cpu->lowcore->restart_new_psw.mask = psw->mask;
> +		cpu->lowcore->restart_new_psw.addr = psw->addr;
> +	}

Does this make sense to have optional? (the other CPU will execute
random crap if not set, won't it?)

> +	rc = sigp(addr, SIGP_RESTART, 0, NULL);
> +	if (rc)
> +		return rc;
> +	/*
> +	 * The order has been accepted, but the actual restart may not
> +	 * have been performed yet, so wait until the cpu is running.
> +	 */
> +	while (!smp_cpu_running(addr))
> +		mb();

Should you make sure to stop the CPU before issuing the restart?
Otherwise you will get false positives if it is still running (but
hasn't processed the RESTART yet)



-- 
Thanks,

David / dhildenb

Re: [kvm-unit-tests PATCH v3 4/9] s390x: smp: Rework cpu start and active tracking

[Thomas Huth]

On 20/01/2020 13.06, David Hildenbrand wrote:
> On 17.01.20 11:46, Janosch Frank wrote:
[...]
>> +
>> +	if (!cpu)
>> +		return -1;
> 
> -EINVAL?

-ENOERRNOHEADERINKVMUNITTESTS ;-)

 Thomas

Re: [kvm-unit-tests PATCH v3 4/9] s390x: smp: Rework cpu start and active tracking

[David Hildenbrand]

On 20.01.20 14:16, Thomas Huth wrote:
> On 20/01/2020 13.06, David Hildenbrand wrote:
>> On 17.01.20 11:46, Janosch Frank wrote:
> [...]
>>> +
>>> +	if (!cpu)
>>> +		return -1;
>>
>> -EINVAL?
> 
> -ENOERRNOHEADERINKVMUNITTESTS ;-)
> 

-EDAMMIT :)

-- 
Thanks,

David / dhildenb

Re: [kvm-unit-tests PATCH v3 4/9] s390x: smp: Rework cpu start and active tracking

[Janosch Frank]

[-- Attachment #1.1: Type: text/plain, Size: 2127 bytes --]

On 1/20/20 1:06 PM, David Hildenbrand wrote:
> On 17.01.20 11:46, Janosch Frank wrote:
>> The architecture specifies that processing sigp orders may be
>> asynchronous, and this is indeed the case on some hypervisors, so we
>> need to wait until the cpu runs before we return from the setup/start
>> function.
>>
>> As there was a lot of duplicate code, a common function for cpu
>> restarts has been introduced.
>>
>> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
>> Reviewed-by: Cornelia Huck <cohuck@redhat.com>
>> ---
>>  lib/s390x/smp.c | 50 ++++++++++++++++++++++++++++---------------------
>>  1 file changed, 29 insertions(+), 21 deletions(-)
>>
>> diff --git a/lib/s390x/smp.c b/lib/s390x/smp.c
>> index f57f420..84e681d 100644
>> --- a/lib/s390x/smp.c
>> +++ b/lib/s390x/smp.c
>> @@ -104,35 +104,46 @@ int smp_cpu_stop_store_status(uint16_t addr)
>>  	return rc;
>>  }
>>  
>> +static int smp_cpu_restart_nolock(uint16_t addr, struct psw *psw)
>> +{
>> +	int rc;
>> +	struct cpu *cpu = smp_cpu_from_addr(addr);
> 
> I'd exchange these two (reverse christmas tree)

Christmas is over

> 
>> +
>> +	if (!cpu)
>> +		return -1;
> 
> -EINVAL?
> 
>> +	if (psw) {
>> +		cpu->lowcore->restart_new_psw.mask = psw->mask;
>> +		cpu->lowcore->restart_new_psw.addr = psw->addr;
>> +	}
> 
> Does this make sense to have optional? (the other CPU will execute
> random crap if not set, won't it?)

Well, I have restarts in the smp test and I don't want to always pass a
psw if I know what the last restart psw was.
Simply restarting into test_func or wait_for_flag is certainly no problem.

> 
>> +	rc = sigp(addr, SIGP_RESTART, 0, NULL);
>> +	if (rc)
>> +		return rc;
>> +	/*
>> +	 * The order has been accepted, but the actual restart may not
>> +	 * have been performed yet, so wait until the cpu is running.
>> +	 */
>> +	while (!smp_cpu_running(addr))
>> +		mb();
> 
> Should you make sure to stop the CPU before issuing the restart?
> Otherwise you will get false positives if it is still running (but
> hasn't processed the RESTART yet)

Good point


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [kvm-unit-tests PATCH v3 4/9] s390x: smp: Rework cpu start and active tracking

[David Hildenbrand]

> Well, I have restarts in the smp test and I don't want to always pass a
> psw if I know what the last restart psw was.
> Simply restarting into test_func or wait_for_flag is certainly no problem.
> 
>>

Makes sense.

>>> +	rc = sigp(addr, SIGP_RESTART, 0, NULL);
>>> +	if (rc)
>>> +		return rc;
>>> +	/*
>>> +	 * The order has been accepted, but the actual restart may not
>>> +	 * have been performed yet, so wait until the cpu is running.
>>> +	 */
>>> +	while (!smp_cpu_running(addr))
>>> +		mb();
>>
>> Should you make sure to stop the CPU before issuing the restart?
>> Otherwise you will get false positives if it is still running (but
>> hasn't processed the RESTART yet)
> 
> Good point
> 


-- 
Thanks,

David / dhildenb

[kvm-unit-tests PATCH v3 5/9] s390x: smp: Wait for cpu setup to finish

[Janosch Frank]

We store the user provided psw address into restart new, so a psw
restart does not lead us through setup again.

Also we wait on smp_cpu_setup() until the cpu has finished setup
before returning. This is necessary for z/VM and LPAR where sigp is
asynchronous.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
---
 lib/s390x/smp.c  | 2 ++
 s390x/cstart64.S | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/lib/s390x/smp.c b/lib/s390x/smp.c
index 84e681d..9dad146 100644
--- a/lib/s390x/smp.c
+++ b/lib/s390x/smp.c
@@ -204,6 +204,8 @@ int smp_cpu_setup(uint16_t addr, struct psw psw)
 
 	/* Start processing */
 	smp_cpu_restart_nolock(addr, NULL);
+	/* Wait until the cpu has finished setup and started the provided psw */
+	while (lc->restart_new_psw.addr != psw.addr) { mb(); }
 out:
 	spin_unlock(&lock);
 	return rc;
diff --git a/s390x/cstart64.S b/s390x/cstart64.S
index 86dd4c4..9af6bb3 100644
--- a/s390x/cstart64.S
+++ b/s390x/cstart64.S
@@ -159,6 +159,8 @@ smp_cpu_setup_state:
 	xgr	%r1, %r1
 	lmg     %r0, %r15, GEN_LC_SW_INT_GRS
 	lctlg   %c0, %c0, GEN_LC_SW_INT_CRS
+	/* We should only go once through cpu setup and not for every restart */
+	stg	%r14, GEN_LC_RESTART_NEW_PSW + 8
 	br	%r14
 
 pgm_int:
-- 
2.20.1

Re: [kvm-unit-tests PATCH v3 5/9] s390x: smp: Wait for cpu setup to finish

[Cornelia Huck]

On Fri, 17 Jan 2020 05:46:36 -0500
Janosch Frank <frankja@linux.ibm.com> wrote:

> We store the user provided psw address into restart new, so a psw
> restart does not lead us through setup again.
> 
> Also we wait on smp_cpu_setup() until the cpu has finished setup
> before returning. This is necessary for z/VM and LPAR where sigp is
> asynchronous.
> 
> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
> Reviewed-by: David Hildenbrand <david@redhat.com>
> ---
>  lib/s390x/smp.c  | 2 ++
>  s390x/cstart64.S | 2 ++
>  2 files changed, 4 insertions(+)

Reviewed-by: Cornelia Huck <cohuck@redhat.com>

Re: [kvm-unit-tests PATCH v3 5/9] s390x: smp: Wait for cpu setup to finish

[David Hildenbrand]

On 17.01.20 11:46, Janosch Frank wrote:
> We store the user provided psw address into restart new, so a psw
> restart does not lead us through setup again.
> 
> Also we wait on smp_cpu_setup() until the cpu has finished setup
> before returning. This is necessary for z/VM and LPAR where sigp is
> asynchronous.
> 
> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
> Reviewed-by: David Hildenbrand <david@redhat.com>
> ---
>  lib/s390x/smp.c  | 2 ++
>  s390x/cstart64.S | 2 ++
>  2 files changed, 4 insertions(+)
> 
> diff --git a/lib/s390x/smp.c b/lib/s390x/smp.c
> index 84e681d..9dad146 100644
> --- a/lib/s390x/smp.c
> +++ b/lib/s390x/smp.c
> @@ -204,6 +204,8 @@ int smp_cpu_setup(uint16_t addr, struct psw psw)
>  
>  	/* Start processing */
>  	smp_cpu_restart_nolock(addr, NULL);
> +	/* Wait until the cpu has finished setup and started the provided psw */
> +	while (lc->restart_new_psw.addr != psw.addr) { mb(); }

while (lc->restart_new_psw.addr != psw.addr)
	mb();



-- 
Thanks,

David / dhildenb

[kvm-unit-tests PATCH v3 6/9] s390x: smp: Loop if secondary cpu returns into cpu setup again

[Janosch Frank]

Up to now a secondary cpu could have returned from the function it was
executing and ending up somewhere in cstart64.S. This was mostly
circumvented by an endless loop in the function that it executed.

Let's add a loop to the end of the cpu setup, so we don't have to rely
on added loops in the tests.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
---
 s390x/cstart64.S | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/s390x/cstart64.S b/s390x/cstart64.S
index 9af6bb3..5fd8d2f 100644
--- a/s390x/cstart64.S
+++ b/s390x/cstart64.S
@@ -162,6 +162,8 @@ smp_cpu_setup_state:
 	/* We should only go once through cpu setup and not for every restart */
 	stg	%r14, GEN_LC_RESTART_NEW_PSW + 8
 	br	%r14
+	/* If the function returns, just loop here */
+0:	j	0
 
 pgm_int:
 	SAVE_REGS
-- 
2.20.1

Re: [kvm-unit-tests PATCH v3 6/9] s390x: smp: Loop if secondary cpu returns into cpu setup again

[Cornelia Huck]

On Fri, 17 Jan 2020 05:46:37 -0500
Janosch Frank <frankja@linux.ibm.com> wrote:

> Up to now a secondary cpu could have returned from the function it was
> executing and ending up somewhere in cstart64.S. This was mostly
> circumvented by an endless loop in the function that it executed.
> 
> Let's add a loop to the end of the cpu setup, so we don't have to rely
> on added loops in the tests.
> 
> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
> ---
>  s390x/cstart64.S | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/s390x/cstart64.S b/s390x/cstart64.S
> index 9af6bb3..5fd8d2f 100644
> --- a/s390x/cstart64.S
> +++ b/s390x/cstart64.S
> @@ -162,6 +162,8 @@ smp_cpu_setup_state:
>  	/* We should only go once through cpu setup and not for every restart */
>  	stg	%r14, GEN_LC_RESTART_NEW_PSW + 8
>  	br	%r14
> +	/* If the function returns, just loop here */
> +0:	j	0

Would it make sense to e.g. load a disabled wait psw instead? Or does
that mess things up elsewhere?

>  
>  pgm_int:
>  	SAVE_REGS

Re: [kvm-unit-tests PATCH v3 6/9] s390x: smp: Loop if secondary cpu returns into cpu setup again

[David Hildenbrand]

On 20.01.20 12:27, Cornelia Huck wrote:
> On Fri, 17 Jan 2020 05:46:37 -0500
> Janosch Frank <frankja@linux.ibm.com> wrote:
> 
>> Up to now a secondary cpu could have returned from the function it was
>> executing and ending up somewhere in cstart64.S. This was mostly
>> circumvented by an endless loop in the function that it executed.
>>
>> Let's add a loop to the end of the cpu setup, so we don't have to rely
>> on added loops in the tests.
>>
>> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
>> ---
>>  s390x/cstart64.S | 2 ++
>>  1 file changed, 2 insertions(+)
>>
>> diff --git a/s390x/cstart64.S b/s390x/cstart64.S
>> index 9af6bb3..5fd8d2f 100644
>> --- a/s390x/cstart64.S
>> +++ b/s390x/cstart64.S
>> @@ -162,6 +162,8 @@ smp_cpu_setup_state:
>>  	/* We should only go once through cpu setup and not for every restart */
>>  	stg	%r14, GEN_LC_RESTART_NEW_PSW + 8
>>  	br	%r14
>> +	/* If the function returns, just loop here */
>> +0:	j	0
> 
> Would it make sense to e.g. load a disabled wait psw instead? Or does
> that mess things up elsewhere?

I think we can keep it like that for now (simpler than loading a PSW).
Overall, I don't care about burning CPU cycles here :)


-- 
Thanks,

David / dhildenb

[kvm-unit-tests PATCH v3 7/9] s390x: smp: Remove unneeded cpu loops

[Janosch Frank]

Now that we have a loop which is executed after we return from the
main function of a secondary cpu, we can remove the surplus loops.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
---
 s390x/smp.c | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/s390x/smp.c b/s390x/smp.c
index 555ed72..c12a3db 100644
--- a/s390x/smp.c
+++ b/s390x/smp.c
@@ -29,15 +29,9 @@ static void wait_for_flag(void)
 	}
 }
 
-static void cpu_loop(void)
-{
-	for (;;) {}
-}
-
 static void test_func(void)
 {
 	testflag = 1;
-	cpu_loop();
 }
 
 static void test_start(void)
@@ -234,7 +228,7 @@ int main(void)
 
 	/* Setting up the cpu to give it a stack and lowcore */
 	psw.mask = extract_psw_mask();
-	psw.addr = (unsigned long)cpu_loop;
+	psw.addr = (unsigned long)test_func;
 	smp_cpu_setup(1, psw);
 	smp_cpu_stop(1);
 
-- 
2.20.1

Re: [kvm-unit-tests PATCH v3 7/9] s390x: smp: Remove unneeded cpu loops

[Cornelia Huck]

On Fri, 17 Jan 2020 05:46:38 -0500
Janosch Frank <frankja@linux.ibm.com> wrote:

> Now that we have a loop which is executed after we return from the
> main function of a secondary cpu, we can remove the surplus loops.
> 
> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
> ---
>  s390x/smp.c | 8 +-------
>  1 file changed, 1 insertion(+), 7 deletions(-)
> 
> diff --git a/s390x/smp.c b/s390x/smp.c
> index 555ed72..c12a3db 100644
> --- a/s390x/smp.c
> +++ b/s390x/smp.c
> @@ -29,15 +29,9 @@ static void wait_for_flag(void)
>  	}
>  }
>  
> -static void cpu_loop(void)
> -{
> -	for (;;) {}
> -}
> -
>  static void test_func(void)
>  {
>  	testflag = 1;
> -	cpu_loop();
>  }
>  
>  static void test_start(void)
> @@ -234,7 +228,7 @@ int main(void)
>  
>  	/* Setting up the cpu to give it a stack and lowcore */
>  	psw.mask = extract_psw_mask();
> -	psw.addr = (unsigned long)cpu_loop;
> +	psw.addr = (unsigned long)test_func;

Before, you did not set testflag here... intended change?

>  	smp_cpu_setup(1, psw);
>  	smp_cpu_stop(1);
>  

Re: [kvm-unit-tests PATCH v3 7/9] s390x: smp: Remove unneeded cpu loops

[Janosch Frank]

[-- Attachment #1.1: Type: text/plain, Size: 1258 bytes --]

On 1/20/20 12:29 PM, Cornelia Huck wrote:
> On Fri, 17 Jan 2020 05:46:38 -0500
> Janosch Frank <frankja@linux.ibm.com> wrote:
> 
>> Now that we have a loop which is executed after we return from the
>> main function of a secondary cpu, we can remove the surplus loops.
>>
>> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
>> ---
>>  s390x/smp.c | 8 +-------
>>  1 file changed, 1 insertion(+), 7 deletions(-)
>>
>> diff --git a/s390x/smp.c b/s390x/smp.c
>> index 555ed72..c12a3db 100644
>> --- a/s390x/smp.c
>> +++ b/s390x/smp.c
>> @@ -29,15 +29,9 @@ static void wait_for_flag(void)
>>  	}
>>  }
>>  
>> -static void cpu_loop(void)
>> -{
>> -	for (;;) {}
>> -}
>> -
>>  static void test_func(void)
>>  {
>>  	testflag = 1;
>> -	cpu_loop();
>>  }
>>  
>>  static void test_start(void)
>> @@ -234,7 +228,7 @@ int main(void)
>>  
>>  	/* Setting up the cpu to give it a stack and lowcore */
>>  	psw.mask = extract_psw_mask();
>> -	psw.addr = (unsigned long)cpu_loop;
>> +	psw.addr = (unsigned long)test_func;
> 
> Before, you did not set testflag here... intended change?

Yes
It is set to 0 before the first test, so it shouldn't matter.

> 
>>  	smp_cpu_setup(1, psw);
>>  	smp_cpu_stop(1);
>>  
> 



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [kvm-unit-tests PATCH v3 7/9] s390x: smp: Remove unneeded cpu loops

[Cornelia Huck]

[-- Attachment #1: Type: text/plain, Size: 1611 bytes --]

On Mon, 20 Jan 2020 15:41:52 +0100
Janosch Frank <frankja@linux.ibm.com> wrote:

> On 1/20/20 12:29 PM, Cornelia Huck wrote:
> > On Fri, 17 Jan 2020 05:46:38 -0500
> > Janosch Frank <frankja@linux.ibm.com> wrote:
> >   
> >> Now that we have a loop which is executed after we return from the
> >> main function of a secondary cpu, we can remove the surplus loops.
> >>
> >> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
> >> ---
> >>  s390x/smp.c | 8 +-------
> >>  1 file changed, 1 insertion(+), 7 deletions(-)
> >>
> >> diff --git a/s390x/smp.c b/s390x/smp.c
> >> index 555ed72..c12a3db 100644
> >> --- a/s390x/smp.c
> >> +++ b/s390x/smp.c
> >> @@ -29,15 +29,9 @@ static void wait_for_flag(void)
> >>  	}
> >>  }
> >>  
> >> -static void cpu_loop(void)
> >> -{
> >> -	for (;;) {}
> >> -}
> >> -
> >>  static void test_func(void)
> >>  {
> >>  	testflag = 1;
> >> -	cpu_loop();
> >>  }
> >>  
> >>  static void test_start(void)
> >> @@ -234,7 +228,7 @@ int main(void)
> >>  
> >>  	/* Setting up the cpu to give it a stack and lowcore */
> >>  	psw.mask = extract_psw_mask();
> >> -	psw.addr = (unsigned long)cpu_loop;
> >> +	psw.addr = (unsigned long)test_func;  
> > 
> > Before, you did not set testflag here... intended change?  
> 
> Yes
> It is set to 0 before the first test, so it shouldn't matter.

Hm... I got a bit lost in all those changes, so I checked your branch
on github, and I don't see it being set to 0 before test_start() is
called?

> 
> >   
> >>  	smp_cpu_setup(1, psw);
> >>  	smp_cpu_stop(1);
> >>    
> >   
> 
> 


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [kvm-unit-tests PATCH v3 7/9] s390x: smp: Remove unneeded cpu loops

[Janosch Frank]

[-- Attachment #1.1: Type: text/plain, Size: 1820 bytes --]

On 1/20/20 5:11 PM, Cornelia Huck wrote:
> On Mon, 20 Jan 2020 15:41:52 +0100
> Janosch Frank <frankja@linux.ibm.com> wrote:
> 
>> On 1/20/20 12:29 PM, Cornelia Huck wrote:
>>> On Fri, 17 Jan 2020 05:46:38 -0500
>>> Janosch Frank <frankja@linux.ibm.com> wrote:
>>>   
>>>> Now that we have a loop which is executed after we return from the
>>>> main function of a secondary cpu, we can remove the surplus loops.
>>>>
>>>> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
>>>> ---
>>>>  s390x/smp.c | 8 +-------
>>>>  1 file changed, 1 insertion(+), 7 deletions(-)
>>>>
>>>> diff --git a/s390x/smp.c b/s390x/smp.c
>>>> index 555ed72..c12a3db 100644
>>>> --- a/s390x/smp.c
>>>> +++ b/s390x/smp.c
>>>> @@ -29,15 +29,9 @@ static void wait_for_flag(void)
>>>>  	}
>>>>  }
>>>>  
>>>> -static void cpu_loop(void)
>>>> -{
>>>> -	for (;;) {}
>>>> -}
>>>> -
>>>>  static void test_func(void)
>>>>  {
>>>>  	testflag = 1;
>>>> -	cpu_loop();
>>>>  }
>>>>  
>>>>  static void test_start(void)
>>>> @@ -234,7 +228,7 @@ int main(void)
>>>>  
>>>>  	/* Setting up the cpu to give it a stack and lowcore */
>>>>  	psw.mask = extract_psw_mask();
>>>> -	psw.addr = (unsigned long)cpu_loop;
>>>> +	psw.addr = (unsigned long)test_func;  
>>>
>>> Before, you did not set testflag here... intended change?  
>>
>> Yes
>> It is set to 0 before the first test, so it shouldn't matter.
> 
> Hm... I got a bit lost in all those changes, so I checked your branch
> on github, and I don't see it being set to 0 before test_start() is
> called?

Well, that's because test_start doesn't care about the flag.
ecall and emcall are the first users, and they set it to 0 before using it.

> 
>>
>>>   
>>>>  	smp_cpu_setup(1, psw);
>>>>  	smp_cpu_stop(1);
>>>>    
>>>   
>>
>>
> 



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [kvm-unit-tests PATCH v3 7/9] s390x: smp: Remove unneeded cpu loops

[Cornelia Huck]

[-- Attachment #1: Type: text/plain, Size: 2130 bytes --]

On Tue, 21 Jan 2020 13:46:51 +0100
Janosch Frank <frankja@linux.ibm.com> wrote:

> On 1/20/20 5:11 PM, Cornelia Huck wrote:
> > On Mon, 20 Jan 2020 15:41:52 +0100
> > Janosch Frank <frankja@linux.ibm.com> wrote:
> >   
> >> On 1/20/20 12:29 PM, Cornelia Huck wrote:  
> >>> On Fri, 17 Jan 2020 05:46:38 -0500
> >>> Janosch Frank <frankja@linux.ibm.com> wrote:
> >>>     
> >>>> Now that we have a loop which is executed after we return from the
> >>>> main function of a secondary cpu, we can remove the surplus loops.
> >>>>
> >>>> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
> >>>> ---
> >>>>  s390x/smp.c | 8 +-------
> >>>>  1 file changed, 1 insertion(+), 7 deletions(-)
> >>>>
> >>>> diff --git a/s390x/smp.c b/s390x/smp.c
> >>>> index 555ed72..c12a3db 100644
> >>>> --- a/s390x/smp.c
> >>>> +++ b/s390x/smp.c
> >>>> @@ -29,15 +29,9 @@ static void wait_for_flag(void)
> >>>>  	}
> >>>>  }
> >>>>  
> >>>> -static void cpu_loop(void)
> >>>> -{
> >>>> -	for (;;) {}
> >>>> -}
> >>>> -
> >>>>  static void test_func(void)
> >>>>  {
> >>>>  	testflag = 1;
> >>>> -	cpu_loop();
> >>>>  }
> >>>>  
> >>>>  static void test_start(void)
> >>>> @@ -234,7 +228,7 @@ int main(void)
> >>>>  
> >>>>  	/* Setting up the cpu to give it a stack and lowcore */
> >>>>  	psw.mask = extract_psw_mask();
> >>>> -	psw.addr = (unsigned long)cpu_loop;
> >>>> +	psw.addr = (unsigned long)test_func;    
> >>>
> >>> Before, you did not set testflag here... intended change?    
> >>
> >> Yes
> >> It is set to 0 before the first test, so it shouldn't matter.  
> > 
> > Hm... I got a bit lost in all those changes, so I checked your branch
> > on github, and I don't see it being set to 0 before test_start() is
> > called?  
> 
> Well, that's because test_start doesn't care about the flag.

But I see a wait_for_flag() in there? What am I missing?

> ecall and emcall are the first users, and they set it to 0 before using it.
> 
> >   
> >>  
> >>>     
> >>>>  	smp_cpu_setup(1, psw);
> >>>>  	smp_cpu_stop(1);
> >>>>      
> >>>     
> >>
> >>  
> >   
> 
> 


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [kvm-unit-tests PATCH v3 7/9] s390x: smp: Remove unneeded cpu loops

[Janosch Frank]

[-- Attachment #1.1: Type: text/plain, Size: 2259 bytes --]

On 1/21/20 1:59 PM, Cornelia Huck wrote:
> On Tue, 21 Jan 2020 13:46:51 +0100
> Janosch Frank <frankja@linux.ibm.com> wrote:
> 
>> On 1/20/20 5:11 PM, Cornelia Huck wrote:
>>> On Mon, 20 Jan 2020 15:41:52 +0100
>>> Janosch Frank <frankja@linux.ibm.com> wrote:
>>>   
>>>> On 1/20/20 12:29 PM, Cornelia Huck wrote:  
>>>>> On Fri, 17 Jan 2020 05:46:38 -0500
>>>>> Janosch Frank <frankja@linux.ibm.com> wrote:
>>>>>     
>>>>>> Now that we have a loop which is executed after we return from the
>>>>>> main function of a secondary cpu, we can remove the surplus loops.
>>>>>>
>>>>>> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
>>>>>> ---
>>>>>>  s390x/smp.c | 8 +-------
>>>>>>  1 file changed, 1 insertion(+), 7 deletions(-)
>>>>>>
>>>>>> diff --git a/s390x/smp.c b/s390x/smp.c
>>>>>> index 555ed72..c12a3db 100644
>>>>>> --- a/s390x/smp.c
>>>>>> +++ b/s390x/smp.c
>>>>>> @@ -29,15 +29,9 @@ static void wait_for_flag(void)
>>>>>>  	}
>>>>>>  }
>>>>>>  
>>>>>> -static void cpu_loop(void)
>>>>>> -{
>>>>>> -	for (;;) {}
>>>>>> -}
>>>>>> -
>>>>>>  static void test_func(void)
>>>>>>  {
>>>>>>  	testflag = 1;
>>>>>> -	cpu_loop();
>>>>>>  }
>>>>>>  
>>>>>>  static void test_start(void)
>>>>>> @@ -234,7 +228,7 @@ int main(void)
>>>>>>  
>>>>>>  	/* Setting up the cpu to give it a stack and lowcore */
>>>>>>  	psw.mask = extract_psw_mask();
>>>>>> -	psw.addr = (unsigned long)cpu_loop;
>>>>>> +	psw.addr = (unsigned long)test_func;    
>>>>>
>>>>> Before, you did not set testflag here... intended change?    
>>>>
>>>> Yes
>>>> It is set to 0 before the first test, so it shouldn't matter.  
>>>
>>> Hm... I got a bit lost in all those changes, so I checked your branch
>>> on github, and I don't see it being set to 0 before test_start() is
>>> called?  
>>
>> Well, that's because test_start doesn't care about the flag.
> 
> But I see a wait_for_flag() in there? What am I missing?
> 
>> ecall and emcall are the first users, and they set it to 0 before using it.

Well, cpu #1 will update tesflag to 1 in ecall() and emcall()

>>
>>>   
>>>>  
>>>>>     
>>>>>>  	smp_cpu_setup(1, psw);
>>>>>>  	smp_cpu_stop(1);
>>>>>>      
>>>>>     
>>>>
>>>>  
>>>   
>>
>>
> 



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[kvm-unit-tests PATCH v3 8/9] s390x: smp: Test all CRs on initial reset

[Janosch Frank]

All CRs are set to 0 and CRs 0 and 14 are set to pre-defined values,
so we also need to test 1-13 and 15 for 0.

And while we're at it, let's also set some values to cr 1, 7 and 13, so
we can actually be sure that they will be zeroed.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
---
 s390x/smp.c | 18 +++++++++++++++++-
 1 file changed, 17 insertions(+), 1 deletion(-)

diff --git a/s390x/smp.c b/s390x/smp.c
index c12a3db..1385488 100644
--- a/s390x/smp.c
+++ b/s390x/smp.c
@@ -169,16 +169,30 @@ static void test_emcall(void)
 	report_prefix_pop();
 }
 
+/* Used to dirty registers of cpu #1 before it is reset */
+static void test_func_initial(void)
+{
+	lctlg(1, 0x42000UL);
+	lctlg(7, 0x43000UL);
+	lctlg(13, 0x44000UL);
+	mb();
+	testflag = 1;
+}
+
 static void test_reset_initial(void)
 {
 	struct cpu_status *status = alloc_pages(0);
+	uint64_t nullp[12] = {};
 	struct psw psw;
 
 	psw.mask = extract_psw_mask();
-	psw.addr = (unsigned long)test_func;
+	psw.addr = (unsigned long)test_func_initial;
 
 	report_prefix_push("reset initial");
+	testflag = 0;
+	mb();
 	smp_cpu_start(1, psw);
+	wait_for_flag();
 
 	sigp_retry(1, SIGP_INITIAL_CPU_RESET, 0, NULL);
 	sigp(1, SIGP_STORE_STATUS_AT_ADDRESS, (uintptr_t)status, NULL);
@@ -189,6 +203,8 @@ static void test_reset_initial(void)
 	report(!status->fpc, "fpc");
 	report(!status->cputm, "cpu timer");
 	report(!status->todpr, "todpr");
+	report(!memcmp(&status->crs[1], nullp, sizeof(status->crs[1]) * 12), "cr1-13 == 0");
+	report(status->crs[15] == 0, "cr15 == 0");
 	report_prefix_pop();
 
 	report_prefix_push("initialized");
-- 
2.20.1

Re: [kvm-unit-tests PATCH v3 8/9] s390x: smp: Test all CRs on initial reset

[Cornelia Huck]

On Fri, 17 Jan 2020 05:46:39 -0500
Janosch Frank <frankja@linux.ibm.com> wrote:

> All CRs are set to 0 and CRs 0 and 14 are set to pre-defined values,
> so we also need to test 1-13 and 15 for 0.
> 
> And while we're at it, let's also set some values to cr 1, 7 and 13, so
> we can actually be sure that they will be zeroed.
> 
> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
> ---
>  s390x/smp.c | 18 +++++++++++++++++-
>  1 file changed, 17 insertions(+), 1 deletion(-)

Reviewed-by: Cornelia Huck <cohuck@redhat.com>

Re: [kvm-unit-tests PATCH v3 8/9] s390x: smp: Test all CRs on initial reset

[David Hildenbrand]

On 17.01.20 11:46, Janosch Frank wrote:
> All CRs are set to 0 and CRs 0 and 14 are set to pre-defined values,
> so we also need to test 1-13 and 15 for 0.
> 
> And while we're at it, let's also set some values to cr 1, 7 and 13, so
> we can actually be sure that they will be zeroed.
> 
> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
> ---
>  s390x/smp.c | 18 +++++++++++++++++-
>  1 file changed, 17 insertions(+), 1 deletion(-)
> 
> diff --git a/s390x/smp.c b/s390x/smp.c
> index c12a3db..1385488 100644
> --- a/s390x/smp.c
> +++ b/s390x/smp.c
> @@ -169,16 +169,30 @@ static void test_emcall(void)
>  	report_prefix_pop();
>  }
>  
> +/* Used to dirty registers of cpu #1 before it is reset */
> +static void test_func_initial(void)
> +{
> +	lctlg(1, 0x42000UL);
> +	lctlg(7, 0x43000UL);
> +	lctlg(13, 0x44000UL);
> +	mb();
> +	testflag = 1;
> +}
> +
>  static void test_reset_initial(void)
>  {
>  	struct cpu_status *status = alloc_pages(0);
> +	uint64_t nullp[12] = {};
>  	struct psw psw;
>  
>  	psw.mask = extract_psw_mask();
> -	psw.addr = (unsigned long)test_func;
> +	psw.addr = (unsigned long)test_func_initial;
>  
>  	report_prefix_push("reset initial");
> +	testflag = 0;
> +	mb();

maybe use a  set_flag() function like

mb();
testflag = val;
mb();

and use it everywhere you set the flag? (e.g., in test_func_initial())

Apart from that

Reviewed-by: David Hildenbrand <david@redhat.com>

>  	smp_cpu_start(1, psw);
> +	wait_for_flag();
>  
>  	sigp_retry(1, SIGP_INITIAL_CPU_RESET, 0, NULL);
>  	sigp(1, SIGP_STORE_STATUS_AT_ADDRESS, (uintptr_t)status, NULL);
> @@ -189,6 +203,8 @@ static void test_reset_initial(void)
>  	report(!status->fpc, "fpc");
>  	report(!status->cputm, "cpu timer");
>  	report(!status->todpr, "todpr");
> +	report(!memcmp(&status->crs[1], nullp, sizeof(status->crs[1]) * 12), "cr1-13 == 0");
> +	report(status->crs[15] == 0, "cr15 == 0");
>  	report_prefix_pop();
>  
>  	report_prefix_push("initialized");
> 


-- 
Thanks,

David / dhildenb

Re: [kvm-unit-tests PATCH v3 8/9] s390x: smp: Test all CRs on initial reset

[Janosch Frank]

[-- Attachment #1.1: Type: text/plain, Size: 2265 bytes --]

On 1/20/20 1:10 PM, David Hildenbrand wrote:
> On 17.01.20 11:46, Janosch Frank wrote:
>> All CRs are set to 0 and CRs 0 and 14 are set to pre-defined values,
>> so we also need to test 1-13 and 15 for 0.
>>
>> And while we're at it, let's also set some values to cr 1, 7 and 13, so
>> we can actually be sure that they will be zeroed.
>>
>> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
>> ---
>>  s390x/smp.c | 18 +++++++++++++++++-
>>  1 file changed, 17 insertions(+), 1 deletion(-)
>>
>> diff --git a/s390x/smp.c b/s390x/smp.c
>> index c12a3db..1385488 100644
>> --- a/s390x/smp.c
>> +++ b/s390x/smp.c
>> @@ -169,16 +169,30 @@ static void test_emcall(void)
>>  	report_prefix_pop();
>>  }
>>  
>> +/* Used to dirty registers of cpu #1 before it is reset */
>> +static void test_func_initial(void)
>> +{
>> +	lctlg(1, 0x42000UL);
>> +	lctlg(7, 0x43000UL);
>> +	lctlg(13, 0x44000UL);
>> +	mb();
>> +	testflag = 1;
>> +}
>> +
>>  static void test_reset_initial(void)
>>  {
>>  	struct cpu_status *status = alloc_pages(0);
>> +	uint64_t nullp[12] = {};
>>  	struct psw psw;
>>  
>>  	psw.mask = extract_psw_mask();
>> -	psw.addr = (unsigned long)test_func;
>> +	psw.addr = (unsigned long)test_func_initial;
>>  
>>  	report_prefix_push("reset initial");
>> +	testflag = 0;
>> +	mb();
> 
> maybe use a  set_flag() function like
> 
> mb();
> testflag = val;
> mb();
> 
> and use it everywhere you set the flag? (e.g., in test_func_initial())

Hmm, so basically a set_test_flag() function in a new patch :)
Ok, I'll have a look.

> 
> Apart from that
> 
> Reviewed-by: David Hildenbrand <david@redhat.com>
> 
>>  	smp_cpu_start(1, psw);
>> +	wait_for_flag();
>>  
>>  	sigp_retry(1, SIGP_INITIAL_CPU_RESET, 0, NULL);
>>  	sigp(1, SIGP_STORE_STATUS_AT_ADDRESS, (uintptr_t)status, NULL);
>> @@ -189,6 +203,8 @@ static void test_reset_initial(void)
>>  	report(!status->fpc, "fpc");
>>  	report(!status->cputm, "cpu timer");
>>  	report(!status->todpr, "todpr");
>> +	report(!memcmp(&status->crs[1], nullp, sizeof(status->crs[1]) * 12), "cr1-13 == 0");
>> +	report(status->crs[15] == 0, "cr15 == 0");
>>  	report_prefix_pop();
>>  
>>  	report_prefix_push("initialized");
>>
> 
> 



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [kvm-unit-tests PATCH v3 8/9] s390x: smp: Test all CRs on initial reset

[David Hildenbrand]

On 20.01.20 15:49, Janosch Frank wrote:
> On 1/20/20 1:10 PM, David Hildenbrand wrote:
>> On 17.01.20 11:46, Janosch Frank wrote:
>>> All CRs are set to 0 and CRs 0 and 14 are set to pre-defined values,
>>> so we also need to test 1-13 and 15 for 0.
>>>
>>> And while we're at it, let's also set some values to cr 1, 7 and 13, so
>>> we can actually be sure that they will be zeroed.
>>>
>>> Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
>>> ---
>>>  s390x/smp.c | 18 +++++++++++++++++-
>>>  1 file changed, 17 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/s390x/smp.c b/s390x/smp.c
>>> index c12a3db..1385488 100644
>>> --- a/s390x/smp.c
>>> +++ b/s390x/smp.c
>>> @@ -169,16 +169,30 @@ static void test_emcall(void)
>>>  	report_prefix_pop();
>>>  }
>>>  
>>> +/* Used to dirty registers of cpu #1 before it is reset */
>>> +static void test_func_initial(void)
>>> +{
>>> +	lctlg(1, 0x42000UL);
>>> +	lctlg(7, 0x43000UL);
>>> +	lctlg(13, 0x44000UL);
>>> +	mb();
>>> +	testflag = 1;
>>> +}
>>> +
>>>  static void test_reset_initial(void)
>>>  {
>>>  	struct cpu_status *status = alloc_pages(0);
>>> +	uint64_t nullp[12] = {};
>>>  	struct psw psw;
>>>  
>>>  	psw.mask = extract_psw_mask();
>>> -	psw.addr = (unsigned long)test_func;
>>> +	psw.addr = (unsigned long)test_func_initial;
>>>  
>>>  	report_prefix_push("reset initial");
>>> +	testflag = 0;
>>> +	mb();
>>
>> maybe use a  set_flag() function like
>>
>> mb();
>> testflag = val;
>> mb();
>>
>> and use it everywhere you set the flag? (e.g., in test_func_initial())
> 
> Hmm, so basically a set_test_flag() function in a new patch :)
> Ok, I'll have a look.

Sorry that this patch set keeps exploding :)


-- 
Thanks,

David / dhildenb

[kvm-unit-tests PATCH v3 9/9] s390x: smp: Dirty fpc before initial reset test

[Janosch Frank]

Let's dirty the fpc, before we test if the initial reset sets it to 0.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
---
 s390x/smp.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/s390x/smp.c b/s390x/smp.c
index 1385488..aedcd65 100644
--- a/s390x/smp.c
+++ b/s390x/smp.c
@@ -172,6 +172,7 @@ static void test_emcall(void)
 /* Used to dirty registers of cpu #1 before it is reset */
 static void test_func_initial(void)
 {
+	asm volatile("sfpc %0" :: "d" (0x11));
 	lctlg(1, 0x42000UL);
 	lctlg(7, 0x43000UL);
 	lctlg(13, 0x44000UL);
-- 
2.20.1