On 2/26/21 11:02 AM, Janosch Frank wrote:
On 2/18/21 6:26 PM, Pierre Morel wrote:
Measurement block format 1 is made available by the extended
measurement block facility and is indicated in the SCHIB by
the bit in the PMCW.
...
+void msch_with_wrong_fmt1_mbo(unsigned int schid, uint64_t mb)
+{
+ struct pmcw *pmcw = &schib.pmcw;
+ int cc;
+
+ /* Read the SCHIB for this subchannel */
+ cc = stsch(schid, &schib);
+ if (cc) {> + report(0, "stsch: sch %08x failed with cc=%d", schid, cc);
+ return;
+ }
+
+ /* Update the SCHIB to enable the measurement block */
+ pmcw->flags |= PMCW_MBUE;
+ pmcw->flags2 |= PMCW_MBF1;
+ schib.mbo = mb;
+
+ /* Tell the CSS we want to modify the subchannel */
+ expect_pgm_int();
+ cc = msch(schid, &schib);
+ check_pgm_int_code(PGM_INT_CODE_OPERAND);
Why would you expect a PGM in a library function are PGMs normal for IO
instructions? oO
Is this a test function which should be part of your test file in
s390x/*.c or is it part of the IO library which should:
- Abort if an initialization failed and we can assume that future tests
are now useless
- Return an error so the test can report an error
- Return success
Now it looks clear to me that this test belongs to the tests and not the
lib.
I put it there to avoid exporting the SCHIB, but after all, why not
exporting the SCHIB, we may need access to it from other tests again in
the future.
+}
diff --git a/s390x/css.c b/s390x/css.c
index b65aa89..576df48 100644
--- a/s390x/css.c
+++ b/s390x/css.c
@@ -257,6 +257,58 @@ end:
report_prefix_pop();
}
+/*
+ * test_schm_fmt1:
+ * With measurement block format 1 the mesurement block is
+ * dedicated to a subchannel.
+ */
+static void test_schm_fmt1(void)
+{
+ struct measurement_block_format1 *mb1;
+
+ report_prefix_push("Format 1");
+
+ if (!test_device_sid) {
+ report_skip("No device");
+ goto end;
+ }
...
+ report(start_measure((u64)mb1, 0, true) &&
+ mb1->ssch_rsch_count == SCHM_UPDATE_CNT,
+ "SSCH measured %d", mb1->ssch_rsch_count);
+ report_prefix_pop();
+
+ schm(NULL, 0); /* Stop the measurement */
+ free_io_mem(mb1, sizeof(struct measurement_block_format1));
+end:
+ report_prefix_pop();
+}
+
static struct {
const char *name;
void (*func)(void);
@@ -268,6 +320,7 @@ static struct {
{ "sense (ssch/tsch)", test_sense },
{ "measurement block (schm)", test_schm },
{ "measurement block format0", test_schm_fmt0 },
+ { "measurement block format1", test_schm_fmt1 },
Output will then be:
"measurement block format1: Format 1: Report message"
Wouldn't it make more sense to put the format 0 and 1 tests into
test_schm() so we'd have:
"measurement block (schm): Format 0: Report message" ?
too much prefix push...
rationalizing will make the goto disapear...
Thanks for review
Regards,
Pierre
--
Pierre Morel
IBM Lab Boeblingen
