s390x: Add exit time test
v2 cover s390x: Add exit time test
v1->v2:
---
* add missing cc clobber, fix constraints for get_clock_us() (thanks Thomas)
* avoid array and use pointer to const char* (thanks Thomas)
* add comment why testing nop makes sense (thanks Thomas)
* rework constraints and clobbers (thanks Thomas)
Sometimes, it is useful to measure the exit time of certain instructions to e.g. identify performance regressions in instructions emulated by the hypervisor.
This series adds a test which executes some instructions and measures their execution time. Since their execution time depends a lot on the environment at hand, all tests are reported as PASS currently.
The point of this series is not so much the instructions which have been chosen here (but your ideas are welcome), but rather the general question whether it makes sense to have a test like this in kvm-unit-tests.
Nico Boehr (2):
lib/s390x: time: add wrapper for stckf
s390x: add exittime tests
lib/s390x/asm/time.h | 11 +- s390x/Makefile | 1 + s390x/exittime.c | 255 +++++++++++++++++++++++++++++++++++++++++++ s390x/unittests.cfg | 4 + 4 files changed, 270 insertions(+), 1 deletion(-) create mode 100644 s390x/exittime.c
Signature · 1 line
2.36.1
PATCH v2 1/2 lib/s390x: time: add wrapper for stckf
Upcoming changes will do performance measurements of instructions. Since stck is designed to return unique values even on concurrent calls, it is unsuited for performance measurements. stckf should be used in this case.
While touching that code, also add a missing cc clobber in get_clock_us() and avoid the memory clobber by moving the clock value to the output operands.
Hence, add a nice wrapper for stckf to the time library.
--- lib/s390x/asm/time.h | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-)
lib/s390x/asm/time.h +10 -1
@@ -14,11 +14,20 @@#define STCK_SHIFT_US (63 - 51)#define STCK_MAX ((1UL << 52) - 1) +static inline uint64_t get_clock_fast (void ) +{ +uint64_t clk; + + asmvolatile (" stckf %0 " :"=Q" (clk) : :"cc" ); + +return clk; +} +static inline uint64_t get_clock_us (void ) {uint64_t clk; - asmvolatile (" stck %0 " : :"Q" (clk) :"memory" ); + asmvolatile (" stck %0 " :"=Q" (clk) : :"cc" );return clk >>STCK_SHIFT_US ; }
Signature · 1 line
2.36.1
Re: [kvm-unit-tests PATCH v2 1/2] lib/s390x: time: add wrapper for stckf
Quoted 29 lines · On Thu, 1 Sep 2022 17:09:55 +0200 Nico Boehr <[email protected]> wrote:
Upcoming changes will do performance measurements of instructions. Since stck is designed to return unique values even on concurrent calls, it is unsuited for performance measurements. stckf should be used in this case.
While touching that code, also add a missing cc clobber in
get_clock_us()and avoid the memory clobber by moving the clock value to the output operands.Hence, add a nice wrapper for stckf to the time library.
Signed-off-by Nico Boehr <[email protected]>--- lib/s390x/asm/time.h | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-)lib/s390x/asm/time.h +5
-0@@ -14,11 +14,20 @@#define STCK_SHIFT_US (63 - 51)#define STCK_MAX ((1UL << 52) - 1) +static inline uint64_t get_clock_fast (void ) +{ +uint64_t clk; + + asmvolatile (" stckf %0 " :"=Q" (clk) : :"cc" );
lib/s390x/asm/time.h line 30
Quoted 10 lines
lib/s390x/asm/time.h +5
-1+ +return clk; +} +static inline uint64_t get_clock_us (void ) {uint64_t clk; - asmvolatile (" stck %0 " : :"Q" (clk) :"memory" ); + asmvolatile (" stck %0 " :"=Q" (clk) : :"cc" );
this fix is not needed if you use the wrapper for stck from your other patch series
Quoted 3 lines
lib/s390x/asm/time.h +0
-0return clk >>STCK_SHIFT_US ; }
PATCH v2 2/2 s390x: add exittime tests
In reply to Nico BoehrAdd a test to measure the execution time of several instructions. This can be helpful in finding performance regressions in hypervisor code.
All tests are currently reported as PASS, since the baseline for their execution time depends on the respective environment and since needs to be determined on a case-by-case basis.
--- s390x/Makefile | 1 + s390x/exittime.c | 255 ++++++++++++++++++++++++++++++++++++++++++++ s390x/unittests.cfg | 4 + 3 files changed, 260 insertions(+) create mode 100644 s390x/exittime.c
s390x/Makefile +1 -0
@@ -34,6 +34,7 @@ tests += $(TEST_DIR)/migration.elf tests += $(TEST_DIR)/pv-attest.elf tests += $(TEST_DIR)/migration-cmm.elf tests += $(TEST_DIR)/migration-skey.elf +tests += $(TEST_DIR)/exittime.elf pv-tests += $(TEST_DIR)/pv-diags.elf
s390x/exittime.c new file +255 -0
255 lines not on this page · raw message
s390x/unittests.cfg +4 -0
@@ -185,3 +185,7 @@ groups = migration [migration-skey] file = migration-skey.elf groups = migration + +[exittime] +file = exittime.elf +smp = 2
Signature · 1 line
2.36.1
Re: [kvm-unit-tests PATCH v2 2/2] s390x: add exittime tests
Quoted 60 lines · On Thu, 1 Sep 2022 17:09:56 +0200 Nico Boehr <[email protected]> wrote:
Add a test to measure the execution time of several instructions. This can be helpful in finding performance regressions in hypervisor code.
All tests are currently reported as PASS, since the baseline for their execution time depends on the respective environment and since needs to be determined on a case-by-case basis.
Signed-off-by Nico Boehr <[email protected]>--- s390x/Makefile | 1 + s390x/exittime.c | 255 ++++++++++++++++++++++++++++++++++++++++++++ s390x/unittests.cfg | 4 + 3 files changed, 260 insertions(+) create mode 100644 s390x/exittime.cs390x/Makefile +1
-0@@ -34,6 +34,7 @@ tests += $(TEST_DIR)/migration.elf tests += $(TEST_DIR)/pv-attest.elf tests += $(TEST_DIR)/migration-cmm.elf tests += $(TEST_DIR)/migration-skey.elf +tests += $(TEST_DIR)/exittime.elf pv-tests += $(TEST_DIR)/pv-diags.elfs390x/exittime.c new file +27
-0@@ -0,0 +1,255 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* +* Measure run time of various instructions. Can be used to find runtime +* regressions of instructions which cause exits. +* +* Copyright IBM Corp. 2022 +* +* Authors: +* Nico Boehr <[email protected]> +*/ +#include <libcflat.h> +#include <smp.h> +#include <sclp.h> +#include <asm/time.h> +#include <asm/sigp.h> +#include <asm/interrupt.h> +#include <asm/page.h> + +char pagebuf[PAGE_SIZE] __attribute__((__aligned__(PAGE_SIZE))); + +static void test_sigp_sense_running (long destcpu) +{ +smp_sigp (destcpu,SIGP_SENSE_RUNNING ,0 ,NULL ); +} + +static void test_nop (long ignore) +{
s390x/exittime.c line 37
Quoted 10 lines
s390x/exittime.c new file +10
-0+/* nops don't trap into the hypervisor, so let's test them for reference */ + asmvolatile ("nop" : : :"memory" ); +} + +static void test_diag9c (long destcpu) +{ + asmvolatile ("diag %[destcpu],0,0x9c" + : + : [destcpu]"d" (destcpu) + :
here you leave some unused :
Quoted 11 lines
s390x/exittime.c new file +11
-0+ ); +} + +static long setup_get_this_cpuaddr (long ignore) +{ +return stap (); +} + +static void test_diag44 (long ignore) +{ + asmvolatile ("diag 0,0,0x44" );
s390x/exittime.c line 58
Quoted 10 lines
s390x/exittime.c new file +10
-0+} + +static void test_stnsm (long ignore) +{ +int out; + + asmvolatile ( +"stnsm %[out],0xff" + : [out]"=Q" (out) + :
here too, but you are using only 2
Quoted 59 lines
s390x/exittime.c new file +59
-0+ ); +} + +static void test_stosm (long ignore) +{ +int out; + + asmvolatile ( +"stosm %[out],0" + : [out]"=Q" (out) + : + ); +} + +static long setup_ssm (long ignore) +{ +long system_mask =0 ; + + asmvolatile ( +"stosm %[system_mask],0" + : [system_mask]"=Q" (system_mask) + : + : + ); + +return system_mask; +} + +static void test_ssm (long old_system_mask) +{ + asmvolatile ( +"ssm %[old_system_mask]" + : + : [old_system_mask]"Q" (old_system_mask) + : + ); +} + +static long setup_lctl4 (long ignore) +{ +long ctl4_orig =0 ; + + asmvolatile ( +"stctg 4,4,%[ctl4_orig]" + : [ctl4_orig]"=S" (ctl4_orig) + : + : + ); + +return ctl4_orig; +} + +static void test_lctl4 (long ctl4_orig) +{ + asmvolatile ( +"lctlg 4,4,%[ctl4_orig]" + : + : [ctl4_orig]"S" (ctl4_orig) + :
s390x/exittime.c line 127
Quoted 10 lines
s390x/exittime.c new file +10
-0+ ); +} + +static void test_stpx (long ignore) +{ +unsigned int prefix; + + asmvolatile ( +"stpx %[prefix]" + : [prefix] "=Q" (prefix)
here you are only using the : you actually need
Quoted 20 lines
s390x/exittime.c new file +20
-0+ ); +} + +static void test_stfl (long ignore) +{ + asmvolatile ( +"stfl 0" : : :"memory" + ); +} + +static void test_epsw (long ignore) +{ +long r1, r2; + + asmvolatile ( +"epsw %[r1], %[r2]" + : [r1]"=d" (r1), [r2]"=d" (r2) + : + : + );
s390x/exittime.c line 157
Quoted 10 lines
s390x/exittime.c new file +10
-0+} + +static void test_illegal (long ignore) +{ +expect_pgm_int (); + asmvolatile ( +".word 0" + : + : + :
here none are needed
s390x/exittime.c line 160
Quoted 3 lines
s390x/exittime.c new file +3
-0+ ); +clear_pgm_int (); +}
decide how you want to do for the : and then do it uniformly.
either you always put all three : (except when none are needed), or you always only put as many as needed, without empty trailing ones.
Quoted 79 lines
s390x/exittime.c new file +79
-0+ +static long setup_servc (long arg) +{ +memset (pagebuf,0 ,PAGE_SIZE ); +return arg; +} + +static void test_servc (long ignore) +{ +SCCB *sccb = (SCCB *) pagebuf; + + sccb->h.length =8 ; +servc (0 , (unsigned long ) sccb); +} + +static void test_stsi (long fc) +{ +stsi (pagebuf, fc,2 ,2 ); +} + +struct test { +const char *name; +/* +* When non-null, will be called once before running the test loop. +* Its return value will be given as argument to testfunc. +*/ +long (*setupfunc)(long arg); +void (*testfunc)(long arg); +long arg; +long iters; +}const exittime_tests[] = { + {"nop" ,NULL , test_nop,0 ,200000 }, + {"sigp sense running(0)" ,NULL , test_sigp_sense_running,0 ,20000 }, + {"sigp sense running(1)" ,NULL , test_sigp_sense_running,1 ,20000 }, + {"diag9c(self)" , setup_get_this_cpuaddr, test_diag9c,0 ,2000 }, + {"diag9c(0)" ,NULL , test_diag9c,0 ,2000 }, + {"diag9c(1)" ,NULL , test_diag9c,1 ,2000 }, + {"diag44" ,NULL , test_diag44,0 ,2000 }, + {"stnsm" ,NULL , test_stnsm,0 ,200000 }, + {"stosm" ,NULL , test_stosm,0 ,200000 }, + {"ssm" , setup_ssm, test_ssm,0 ,200000 }, + {"lctl4" , setup_lctl4, test_lctl4,0 ,20000 }, + {"stpx" ,NULL , test_stpx,0 ,2000 }, + {"stfl" ,NULL , test_stfl,0 ,2000 }, + {"epsw" ,NULL , test_epsw,0 ,20000 }, + {"illegal" ,NULL , test_illegal,0 ,2000 }, + {"servc" , setup_servc, test_servc,0 ,2000 }, + {"stsi122" ,NULL , test_stsi,1 ,200 }, + {"stsi222" ,NULL , test_stsi,2 ,200 }, + {"stsi322" ,NULL , test_stsi,3 ,200 }, +}; + +static uint64_t tod_to_us (uint64_t tod) +{ +return tod >>STCK_SHIFT_US ; +} + +int main (void ) +{ +int i, j, k, testfunc_arg; +const int outer_iters =100 ; +struct test const *current_test; +uint64_t start, end, elapsed, worst, best, total; + +report_prefix_push ("exittime" ); +report_pass ("reporting total/best/worst of %d outer iterations" , outer_iters); + +for (i =0 ; i <ARRAY_SIZE (exittime_tests); i++) { + current_test = &exittime_tests[i]; + total =0 ; + worst =0 ; + best =-1 ; +report_prefix_pushf ("%s" , current_test->name); + + testfunc_arg = current_test->arg; +if (current_test->setupfunc) + testfunc_arg = current_test->setupfunc (testfunc_arg); + +for (j =0 ; j < outer_iters; j++) {
s390x/exittime.c line 249
Quoted 10 lines
s390x/exittime.c new file +10
-0+ start =get_clock_fast (); +for (k =0 ; k < current_test->iters; k++) + current_test->testfunc (testfunc_arg); + end =get_clock_fast (); + elapsed = end - start; + best =MIN (best, elapsed); + worst =MAX (worst, elapsed); + total += elapsed; + } +report_pass ("iters/total/best/worst %lu/%lu/%lu/%lu us" , current_test->iters,tod_to_us (total),tod_to_us (best),tod_to_us (worst));
perhaps it would not be a bad idea to also print the average and the standard deviation (σ)
Quoted 18 lines
s390x/exittime.c new file +6
-0+report_prefix_pop (); + } + +report_prefix_pop (); +return report_summary (); +}s390x/unittests.cfg +4
-0@@ -185,3 +185,7 @@ groups = migration [migration-skey] file = migration-skey.elf groups = migration + +[exittime] +file = exittime.elf +smp = 2
Quoted 12 lines · On Thu, 1 Sep 2022 17:09:54 +0200 Nico Boehr <[email protected]> wrote:
v1->v2:
---
* add missing cc clobber, fix constraints forget_clock_us()(thanks Thomas)
* avoid array and use pointer to const char* (thanks Thomas)
* add comment why testing nop makes sense (thanks Thomas)
* rework constraints and clobbers (thanks Thomas)Sometimes, it is useful to measure the exit time of certain instructions to e.g. identify performance regressions in instructions emulated by the hypervisor.
Quoted 8 lines
This series adds a test which executes some instructions and measures their execution time. Since their execution time depends a lot on the environment at hand, all tests are reported as PASS currently.
The point of this series is not so much the instructions which have been chosen here (but your ideas are welcome), but rather the general question whether it makes sense to have a test like this in kvm-unit-tests.
perhaps it makes sense to merge this patch series with your other one, "s390x: Add migration test for guest TOD clock"
they both concern timing and reshuffling around and/or fixing timing functions
Quoted 12 lines
Nico Boehr (2):
lib/s390x: time: add wrapper for stckf
s390x: add exittime testslib/s390x/asm/time.h | 11 +- s390x/Makefile | 1 + s390x/exittime.c | 255 +++++++++++++++++++++++++++++++++++++++++++ s390x/unittests.cfg | 4 + 4 files changed, 270 insertions(+), 1 deletion(-) create mode 100644 s390x/exittime.c
Claudio,
Quoting Claudio Imbrenda (2022-10-06 13:18:56)
Quoted 5 lines
perhaps it makes sense to merge this patch series with your other one, "s390x: Add migration test for guest TOD clock"
they both concern timing and reshuffling around and/or fixing timing functions
Since the TOD migration test is useful for regression testing my PV clock fix[1], I would like to have this in ASAP. The exittime test, though, can wait for a bit. If it's OK for you, I would like to keep these things seperate and rather (re-)base the exittime test on the TOD migration test.