Message ID | 20240820065623.1140399-1-hari55@linux.ibm.com |
---|---|
Headers | show |
Series | KVM: s390: selftests: Add regression tests for CPU subfunctions | expand |
On Tue Aug 20, 2024 at 8:48 AM CEST, Hariharan Mari wrote: > Introduce new regression tests to verify the ASM inline block in the SORTL > and DFLTCC CPU subfunctions for the s390x architecture. These tests ensure > that future changes to the ASM code are properly validated. > > The test procedure: > > 1. Create a VM and request the KVM_S390_VM_CPU_MACHINE_SUBFUNC attribute > from the KVM_S390_VM_CPU_MODEL group for this VM. This SUBFUNC attribute > contains the results of all CPU subfunction instructions. > 2. For each tested subfunction (SORTL and DFLTCC), execute the > corresponding ASM instruction and capture the result array. > 3. Perform a memory comparison between the results stored in the SUBFUNC > attribute (obtained in step 1) and the ASM instruction results (obtained > in step 2) for each tested subfunction. > > This process ensures that the KVM implementation accurately reflects the > behavior of the actual CPU instructions for the tested subfunctions. > > Suggested-by: Janosch Frank <frankja@linux.ibm.com> > Signed-off-by: Hariharan Mari <hari55@linux.ibm.com> > Reviewed-by: Janosch Frank <frankja@linux.ibm.com> LGTM Reviewed-by: Christoph Schlameuss <schlameuss@linux.ibm.com> > --- > tools/testing/selftests/kvm/Makefile | 1 + > .../selftests/kvm/include/s390x/facility.h | 50 ++++++++ > .../kvm/s390x/cpumodel_subfuncs_test.c | 115 ++++++++++++++++++ > 3 files changed, 166 insertions(+) > create mode 100644 tools/testing/selftests/kvm/include/s390x/facility.h > create mode 100644 tools/testing/selftests/kvm/s390x/cpumodel_subfuncs_test.c [...]
On Tue Aug 20, 2024 at 8:48 AM CEST, Hariharan Mari wrote: > Extend the existing regression test framework for s390x CPU subfunctions > to include tests for the KMCTR (Cipher Message with Counter) KMO > (Cipher Message with Output Feedback), KMF (Cipher Message with Cipher > Feedback) and PCC (Perform Cryptographic Computation) crypto functions. > > The test procedure follows the established pattern. > > Suggested-by: Janosch Frank <frankja@linux.ibm.com> > Signed-off-by: Hariharan Mari <hari55@linux.ibm.com> > Reviewed-by: Janosch Frank <frankja@linux.ibm.com> LGTM Reviewed-by: Christoph Schlameuss <schlameuss@linux.ibm.com> > --- > .../kvm/s390x/cpumodel_subfuncs_test.c | 65 +++++++++++++++++++ > 1 file changed, 65 insertions(+) [...]
On Tue Aug 20, 2024 at 8:48 AM CEST, Hariharan Mari wrote: > Extend the existing regression test framework for s390x CPU subfunctions > to include tests for the Perform Locked Operation (PLO) subfunction > functions. > > PLO was introduced in the very first 64-bit machine generation. > Hence it is assumed PLO is always installed in the Z Arch. > The test procedure follows the established pattern. > > Suggested-by: Janosch Frank <frankja@linux.ibm.com> > Signed-off-by: Hariharan Mari <hari55@linux.ibm.com> > Reviewed-by: Janosch Frank <frankja@linux.ibm.com> LGTM (apart from the accidental whitespace) Reviewed-by: Christoph Schlameuss <schlameuss@linux.ibm.com> > --- > .../kvm/s390x/cpumodel_subfuncs_test.c | 34 +++++++++++++++++++ > 1 file changed, 34 insertions(+) > > diff --git a/tools/testing/selftests/kvm/s390x/cpumodel_subfuncs_test.c b/tools/testing/selftests/kvm/s390x/cpumodel_subfuncs_test.c > index c31f445c6f03..255984a52365 100644 > --- a/tools/testing/selftests/kvm/s390x/cpumodel_subfuncs_test.c > +++ b/tools/testing/selftests/kvm/s390x/cpumodel_subfuncs_test.c > @@ -20,6 +20,8 @@ > > #include "kvm_util.h" > > +#define U8_MAX ((u8)~0U) > + > /** > * Query available CPU subfunctions > */ > @@ -37,6 +39,33 @@ static void get_cpu_machine_subfuntions(struct kvm_vm *vm, > TEST_ASSERT(!r, "Get cpu subfunctions failed r=%d errno=%d", r, errno); > } > > +static inline int plo_test_bit(unsigned char nr) > +{ > + unsigned long function = (unsigned long)nr | 0x100; > + int cc; > + > + asm volatile(" lgr 0,%[function]\n" > + /* Parameter registers are ignored for "test bit" */ > + " plo 0,0,0,0(0)\n" > + " ipm %0\n" > + " srl %0,28\n" > + : "=d" (cc) > + : [function] "d" (function) > + : "cc", "0"); > + return cc == 0; > +} > + > +/* > + * Testing Perform Locked Operation (PLO) CPU subfunction's ASM block > + */ > +static void test_plo_asm_block(u8 (*query)[32]) > +{ > + for (int i = 0; i <= U8_MAX; ++i) { > + if (plo_test_bit(i)) > + (*query)[i >> 3] |= 0x80 >> (i & 7); > + } > +} > + > /* > * Testing Crypto Compute Message Authentication Code (KMAC) CPU subfunction's > * ASM block > @@ -237,6 +266,11 @@ struct testdef { > testfunc_t test; > int facility_bit; > } testlist[] = { > + /* PLO was introduced in the very first 64-bit machine generation. > + * Hence it is assumed PLO is always installed in Z Arch . ^ accidental whitespace > + */ > + { "PLO", cpu_subfunc.plo, sizeof(cpu_subfunc.plo), > + test_plo_asm_block, 1 }, > /* MSA - Facility bit 17 */ > { "KMAC", cpu_subfunc.kmac, sizeof(cpu_subfunc.kmac), > test_kmac_asm_block, 17 },
On Tue, 20 Aug 2024 08:48:37 +0200 Hariharan Mari <hari55@linux.ibm.com> wrote: > Extend the existing regression test framework for s390x CPU subfunctions > to include tests for the Perform Locked Operation (PLO) subfunction > functions. > > PLO was introduced in the very first 64-bit machine generation. > Hence it is assumed PLO is always installed in the Z Arch. > The test procedure follows the established pattern. > > Suggested-by: Janosch Frank <frankja@linux.ibm.com> > Signed-off-by: Hariharan Mari <hari55@linux.ibm.com> > Reviewed-by: Janosch Frank <frankja@linux.ibm.com> > --- > .../kvm/s390x/cpumodel_subfuncs_test.c | 34 +++++++++++++++++++ > 1 file changed, 34 insertions(+) > > diff --git a/tools/testing/selftests/kvm/s390x/cpumodel_subfuncs_test.c b/tools/testing/selftests/kvm/s390x/cpumodel_subfuncs_test.c > index c31f445c6f03..255984a52365 100644 > --- a/tools/testing/selftests/kvm/s390x/cpumodel_subfuncs_test.c > +++ b/tools/testing/selftests/kvm/s390x/cpumodel_subfuncs_test.c > @@ -20,6 +20,8 @@ > > #include "kvm_util.h" > > +#define U8_MAX ((u8)~0U) a more descriptive macro name would be better, maybe something like: #define PLO_MAX_PARAMETER 255 the current macro is not much better than having just a magic number :) > + > /** > * Query available CPU subfunctions > */ > @@ -37,6 +39,33 @@ static void get_cpu_machine_subfuntions(struct kvm_vm *vm, > TEST_ASSERT(!r, "Get cpu subfunctions failed r=%d errno=%d", r, errno); > } > > +static inline int plo_test_bit(unsigned char nr) > +{ > + unsigned long function = (unsigned long)nr | 0x100; I think the (unsigned long) cast is not needed > + int cc; > + > + asm volatile(" lgr 0,%[function]\n" > + /* Parameter registers are ignored for "test bit" */ > + " plo 0,0,0,0(0)\n" > + " ipm %0\n" > + " srl %0,28\n" > + : "=d" (cc) > + : [function] "d" (function) > + : "cc", "0"); > + return cc == 0; > +} > + > +/* > + * Testing Perform Locked Operation (PLO) CPU subfunction's ASM block > + */ > +static void test_plo_asm_block(u8 (*query)[32]) > +{ > + for (int i = 0; i <= U8_MAX; ++i) { > + if (plo_test_bit(i)) > + (*query)[i >> 3] |= 0x80 >> (i & 7); > + } > +} > + > /* > * Testing Crypto Compute Message Authentication Code (KMAC) CPU subfunction's > * ASM block > @@ -237,6 +266,11 @@ struct testdef { > testfunc_t test; > int facility_bit; > } testlist[] = { > + /* PLO was introduced in the very first 64-bit machine generation. multi-line comments should not have text in the opening line > + * Hence it is assumed PLO is always installed in Z Arch . > + */ > + { "PLO", cpu_subfunc.plo, sizeof(cpu_subfunc.plo), > + test_plo_asm_block, 1 }, > /* MSA - Facility bit 17 */ > { "KMAC", cpu_subfunc.kmac, sizeof(cpu_subfunc.kmac), > test_kmac_asm_block, 17 },