| Message ID | 20210701153853.33063-1-borntraeger@de.ibm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [PATCH/RFC] KVM: selftests: introduce P44V64 for z196 and EC12 | expand |
On 01.07.21 17:38, Christian Borntraeger wrote: > Older machines likes z196 and zEC12 do only support 44 bits of physical > addresses. Make this the default and check via IBC if we are on a later > machine. We then add P47V64 as an additional model. > > Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com> > Fixes: 1bc603af73dd ("KVM: selftests: introduce P47V64 for s390x") [...] > +#ifdef __s390x__ > + { > + int kvm_fd, vm_fd; > + struct kvm_s390_vm_cpu_processor info; > + > + kvm_fd = open_kvm_dev_path_or_exit(); > + vm_fd = ioctl(kvm_fd, KVM_CREATE_VM, 0); > + kvm_device_access(vm_fd, KVM_S390_VM_CPU_MODEL, > + KVM_S390_VM_CPU_PROCESSOR, &info, false); Can we always assume to run on a kernel where this won't fail? > + close(vm_fd); > + close(kvm_fd); > + /* Starting with z13 we have 47bits of physical address */ This matches the definition in the QEMU cpu models. > + if (info.ibc >= 0x30) > + guest_mode_append(VM_MODE_P47V64_4K, true, true); > + } > +#endif In general, LGTM
On 05.07.21 11:53, David Hildenbrand wrote: > On 01.07.21 17:38, Christian Borntraeger wrote: >> Older machines likes z196 and zEC12 do only support 44 bits of physical >> addresses. Make this the default and check via IBC if we are on a later >> machine. We then add P47V64 as an additional model. >> >> Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com> >> Fixes: 1bc603af73dd ("KVM: selftests: introduce P47V64 for s390x") > > [...] > >> +#ifdef __s390x__ >> + { >> + int kvm_fd, vm_fd; >> + struct kvm_s390_vm_cpu_processor info; >> + >> + kvm_fd = open_kvm_dev_path_or_exit(); >> + vm_fd = ioctl(kvm_fd, KVM_CREATE_VM, 0); >> + kvm_device_access(vm_fd, KVM_S390_VM_CPU_MODEL, >> + KVM_S390_VM_CPU_PROCESSOR, &info, false); > > Can we always assume to run on a kernel where this won't fail? As far as I can tell, the selftests are bundled with a given kernel (and there it should not fail). I guess most selftests will fail with a 3.x kernel and we do not care? > >> + close(vm_fd); >> + close(kvm_fd); >> + /* Starting with z13 we have 47bits of physical address */ > > This matches the definition in the QEMU cpu models. > >> + if (info.ibc >= 0x30) >> + guest_mode_append(VM_MODE_P47V64_4K, true, true); >> + } >> +#endif > > > In general, LGTM > >
On 05.07.21 11:59, Christian Borntraeger wrote: > > > On 05.07.21 11:53, David Hildenbrand wrote: >> On 01.07.21 17:38, Christian Borntraeger wrote: >>> Older machines likes z196 and zEC12 do only support 44 bits of physical >>> addresses. Make this the default and check via IBC if we are on a later >>> machine. We then add P47V64 as an additional model. >>> >>> Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com> >>> Fixes: 1bc603af73dd ("KVM: selftests: introduce P47V64 for s390x") >> >> [...] >> >>> +#ifdef __s390x__ >>> + { >>> + int kvm_fd, vm_fd; >>> + struct kvm_s390_vm_cpu_processor info; >>> + >>> + kvm_fd = open_kvm_dev_path_or_exit(); >>> + vm_fd = ioctl(kvm_fd, KVM_CREATE_VM, 0); >>> + kvm_device_access(vm_fd, KVM_S390_VM_CPU_MODEL, >>> + KVM_S390_VM_CPU_PROCESSOR, &info, false); >> >> Can we always assume to run on a kernel where this won't fail? > > As far as I can tell, the selftests are bundled with a given kernel (and > there it should not fail). I guess most selftests will fail with a 3.x > kernel and we do not care? Fair enough Reviewed-by: David Hildenbrand <david@redhat.com>
On 01/07/2021 17.38, Christian Borntraeger wrote: > Older machines likes z196 and zEC12 do only support 44 bits of physical > addresses. Make this the default and check via IBC if we are on a later > machine. We then add P47V64 as an additional model. > > Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com> > Fixes: 1bc603af73dd ("KVM: selftests: introduce P47V64 for s390x") > --- > tools/testing/selftests/kvm/include/kvm_util.h | 3 ++- > tools/testing/selftests/kvm/lib/guest_modes.c | 16 ++++++++++++++++ > tools/testing/selftests/kvm/lib/kvm_util.c | 5 +++++ > 3 files changed, 23 insertions(+), 1 deletion(-) > > diff --git a/tools/testing/selftests/kvm/include/kvm_util.h b/tools/testing/selftests/kvm/include/kvm_util.h > index 35739567189e..74d73532fce9 100644 > --- a/tools/testing/selftests/kvm/include/kvm_util.h > +++ b/tools/testing/selftests/kvm/include/kvm_util.h > @@ -44,6 +44,7 @@ enum vm_guest_mode { > VM_MODE_P40V48_64K, > VM_MODE_PXXV48_4K, /* For 48bits VA but ANY bits PA */ > VM_MODE_P47V64_4K, > + VM_MODE_P44V64_4K, > NUM_VM_MODES, > }; > > @@ -61,7 +62,7 @@ enum vm_guest_mode { > > #elif defined(__s390x__) > > -#define VM_MODE_DEFAULT VM_MODE_P47V64_4K > +#define VM_MODE_DEFAULT VM_MODE_P44V64_4K > #define MIN_PAGE_SHIFT 12U > #define ptes_per_page(page_size) ((page_size) / 16) > > diff --git a/tools/testing/selftests/kvm/lib/guest_modes.c b/tools/testing/selftests/kvm/lib/guest_modes.c > index 25bff307c71f..c330f414ef96 100644 > --- a/tools/testing/selftests/kvm/lib/guest_modes.c > +++ b/tools/testing/selftests/kvm/lib/guest_modes.c > @@ -22,6 +22,22 @@ void guest_modes_append_default(void) > } > } > #endif > +#ifdef __s390x__ > + { > + int kvm_fd, vm_fd; > + struct kvm_s390_vm_cpu_processor info; > + > + kvm_fd = open_kvm_dev_path_or_exit(); > + vm_fd = ioctl(kvm_fd, KVM_CREATE_VM, 0); > + kvm_device_access(vm_fd, KVM_S390_VM_CPU_MODEL, > + KVM_S390_VM_CPU_PROCESSOR, &info, false); > + close(vm_fd); > + close(kvm_fd); > + /* Starting with z13 we have 47bits of physical address */ > + if (info.ibc >= 0x30) > + guest_mode_append(VM_MODE_P47V64_4K, true, true); Wouldn't it make more sense to check the processor number in /proc/cpuinfo? ... well, I guess both ways of checking have their advantages and disadvantages, so anyway: Reviewed-by: Thomas Huth <thuth@redhat.com>
On Thu, Jul 01 2021, Christian Borntraeger <borntraeger@de.ibm.com> wrote: > Older machines likes z196 and zEC12 do only support 44 bits of physical s/likes/like/ > addresses. Make this the default and check via IBC if we are on a later > machine. We then add P47V64 as an additional model. > > Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com> > Fixes: 1bc603af73dd ("KVM: selftests: introduce P47V64 for s390x") > --- > tools/testing/selftests/kvm/include/kvm_util.h | 3 ++- > tools/testing/selftests/kvm/lib/guest_modes.c | 16 ++++++++++++++++ > tools/testing/selftests/kvm/lib/kvm_util.c | 5 +++++ > 3 files changed, 23 insertions(+), 1 deletion(-) > Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Paolo, since you have not yet pulled my queue for 5.14. Shall I add the two selftest patches and send a new pull request? On 01.07.21 17:38, Christian Borntraeger wrote: > Older machines likes z196 and zEC12 do only support 44 bits of physical > addresses. Make this the default and check via IBC if we are on a later > machine. We then add P47V64 as an additional model. > > Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com> > Fixes: 1bc603af73dd ("KVM: selftests: introduce P47V64 for s390x") > --- > tools/testing/selftests/kvm/include/kvm_util.h | 3 ++- > tools/testing/selftests/kvm/lib/guest_modes.c | 16 ++++++++++++++++ > tools/testing/selftests/kvm/lib/kvm_util.c | 5 +++++ > 3 files changed, 23 insertions(+), 1 deletion(-) > > diff --git a/tools/testing/selftests/kvm/include/kvm_util.h b/tools/testing/selftests/kvm/include/kvm_util.h > index 35739567189e..74d73532fce9 100644 > --- a/tools/testing/selftests/kvm/include/kvm_util.h > +++ b/tools/testing/selftests/kvm/include/kvm_util.h > @@ -44,6 +44,7 @@ enum vm_guest_mode { > VM_MODE_P40V48_64K, > VM_MODE_PXXV48_4K, /* For 48bits VA but ANY bits PA */ > VM_MODE_P47V64_4K, > + VM_MODE_P44V64_4K, > NUM_VM_MODES, > }; > > @@ -61,7 +62,7 @@ enum vm_guest_mode { > > #elif defined(__s390x__) > > -#define VM_MODE_DEFAULT VM_MODE_P47V64_4K > +#define VM_MODE_DEFAULT VM_MODE_P44V64_4K > #define MIN_PAGE_SHIFT 12U > #define ptes_per_page(page_size) ((page_size) / 16) > > diff --git a/tools/testing/selftests/kvm/lib/guest_modes.c b/tools/testing/selftests/kvm/lib/guest_modes.c > index 25bff307c71f..c330f414ef96 100644 > --- a/tools/testing/selftests/kvm/lib/guest_modes.c > +++ b/tools/testing/selftests/kvm/lib/guest_modes.c > @@ -22,6 +22,22 @@ void guest_modes_append_default(void) > } > } > #endif > +#ifdef __s390x__ > + { > + int kvm_fd, vm_fd; > + struct kvm_s390_vm_cpu_processor info; > + > + kvm_fd = open_kvm_dev_path_or_exit(); > + vm_fd = ioctl(kvm_fd, KVM_CREATE_VM, 0); > + kvm_device_access(vm_fd, KVM_S390_VM_CPU_MODEL, > + KVM_S390_VM_CPU_PROCESSOR, &info, false); > + close(vm_fd); > + close(kvm_fd); > + /* Starting with z13 we have 47bits of physical address */ > + if (info.ibc >= 0x30) > + guest_mode_append(VM_MODE_P47V64_4K, true, true); > + } > +#endif > } > > void for_each_guest_mode(void (*func)(enum vm_guest_mode, void *), void *arg) > diff --git a/tools/testing/selftests/kvm/lib/kvm_util.c b/tools/testing/selftests/kvm/lib/kvm_util.c > index a2b732cf96ea..8606000c439e 100644 > --- a/tools/testing/selftests/kvm/lib/kvm_util.c > +++ b/tools/testing/selftests/kvm/lib/kvm_util.c > @@ -176,6 +176,7 @@ const char *vm_guest_mode_string(uint32_t i) > [VM_MODE_P40V48_64K] = "PA-bits:40, VA-bits:48, 64K pages", > [VM_MODE_PXXV48_4K] = "PA-bits:ANY, VA-bits:48, 4K pages", > [VM_MODE_P47V64_4K] = "PA-bits:47, VA-bits:64, 4K pages", > + [VM_MODE_P44V64_4K] = "PA-bits:44, VA-bits:64, 4K pages", > }; > _Static_assert(sizeof(strings)/sizeof(char *) == NUM_VM_MODES, > "Missing new mode strings?"); > @@ -194,6 +195,7 @@ const struct vm_guest_mode_params vm_guest_mode_params[] = { > { 40, 48, 0x10000, 16 }, > { 0, 0, 0x1000, 12 }, > { 47, 64, 0x1000, 12 }, > + { 44, 64, 0x1000, 12 }, > }; > _Static_assert(sizeof(vm_guest_mode_params)/sizeof(struct vm_guest_mode_params) == NUM_VM_MODES, > "Missing new mode params?"); > @@ -282,6 +284,9 @@ struct kvm_vm *vm_create(enum vm_guest_mode mode, uint64_t phy_pages, int perm) > case VM_MODE_P47V64_4K: > vm->pgtable_levels = 5; > break; > + case VM_MODE_P44V64_4K: > + vm->pgtable_levels = 5; > + break; > default: > TEST_FAIL("Unknown guest mode, mode: 0x%x", mode); > } >
On 06.07.21 09:40, Christian Borntraeger wrote: > Paolo, > > since you have not yet pulled my queue for 5.14. Shall I add the two selftest patches and send a new > pull request? Hmm, I cant put it on top of the next queue since I would need to rebase. So lets do the original pull request and I will do another one on top of kvm/master for the 2 selftest patches. > > On 01.07.21 17:38, Christian Borntraeger wrote: >> Older machines likes z196 and zEC12 do only support 44 bits of physical >> addresses. Make this the default and check via IBC if we are on a later >> machine. We then add P47V64 as an additional model. >> >> Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com> >> Fixes: 1bc603af73dd ("KVM: selftests: introduce P47V64 for s390x") >> --- >> tools/testing/selftests/kvm/include/kvm_util.h | 3 ++- >> tools/testing/selftests/kvm/lib/guest_modes.c | 16 ++++++++++++++++ >> tools/testing/selftests/kvm/lib/kvm_util.c | 5 +++++ >> 3 files changed, 23 insertions(+), 1 deletion(-) >> >> diff --git a/tools/testing/selftests/kvm/include/kvm_util.h b/tools/testing/selftests/kvm/include/kvm_util.h >> index 35739567189e..74d73532fce9 100644 >> --- a/tools/testing/selftests/kvm/include/kvm_util.h >> +++ b/tools/testing/selftests/kvm/include/kvm_util.h >> @@ -44,6 +44,7 @@ enum vm_guest_mode { >> VM_MODE_P40V48_64K, >> VM_MODE_PXXV48_4K, /* For 48bits VA but ANY bits PA */ >> VM_MODE_P47V64_4K, >> + VM_MODE_P44V64_4K, >> NUM_VM_MODES, >> }; >> @@ -61,7 +62,7 @@ enum vm_guest_mode { >> #elif defined(__s390x__) >> -#define VM_MODE_DEFAULT VM_MODE_P47V64_4K >> +#define VM_MODE_DEFAULT VM_MODE_P44V64_4K >> #define MIN_PAGE_SHIFT 12U >> #define ptes_per_page(page_size) ((page_size) / 16) >> diff --git a/tools/testing/selftests/kvm/lib/guest_modes.c b/tools/testing/selftests/kvm/lib/guest_modes.c >> index 25bff307c71f..c330f414ef96 100644 >> --- a/tools/testing/selftests/kvm/lib/guest_modes.c >> +++ b/tools/testing/selftests/kvm/lib/guest_modes.c >> @@ -22,6 +22,22 @@ void guest_modes_append_default(void) >> } >> } >> #endif >> +#ifdef __s390x__ >> + { >> + int kvm_fd, vm_fd; >> + struct kvm_s390_vm_cpu_processor info; >> + >> + kvm_fd = open_kvm_dev_path_or_exit(); >> + vm_fd = ioctl(kvm_fd, KVM_CREATE_VM, 0); >> + kvm_device_access(vm_fd, KVM_S390_VM_CPU_MODEL, >> + KVM_S390_VM_CPU_PROCESSOR, &info, false); >> + close(vm_fd); >> + close(kvm_fd); >> + /* Starting with z13 we have 47bits of physical address */ >> + if (info.ibc >= 0x30) >> + guest_mode_append(VM_MODE_P47V64_4K, true, true); >> + } >> +#endif >> } >> void for_each_guest_mode(void (*func)(enum vm_guest_mode, void *), void *arg) >> diff --git a/tools/testing/selftests/kvm/lib/kvm_util.c b/tools/testing/selftests/kvm/lib/kvm_util.c >> index a2b732cf96ea..8606000c439e 100644 >> --- a/tools/testing/selftests/kvm/lib/kvm_util.c >> +++ b/tools/testing/selftests/kvm/lib/kvm_util.c >> @@ -176,6 +176,7 @@ const char *vm_guest_mode_string(uint32_t i) >> [VM_MODE_P40V48_64K] = "PA-bits:40, VA-bits:48, 64K pages", >> [VM_MODE_PXXV48_4K] = "PA-bits:ANY, VA-bits:48, 4K pages", >> [VM_MODE_P47V64_4K] = "PA-bits:47, VA-bits:64, 4K pages", >> + [VM_MODE_P44V64_4K] = "PA-bits:44, VA-bits:64, 4K pages", >> }; >> _Static_assert(sizeof(strings)/sizeof(char *) == NUM_VM_MODES, >> "Missing new mode strings?"); >> @@ -194,6 +195,7 @@ const struct vm_guest_mode_params vm_guest_mode_params[] = { >> { 40, 48, 0x10000, 16 }, >> { 0, 0, 0x1000, 12 }, >> { 47, 64, 0x1000, 12 }, >> + { 44, 64, 0x1000, 12 }, >> }; >> _Static_assert(sizeof(vm_guest_mode_params)/sizeof(struct vm_guest_mode_params) == NUM_VM_MODES, >> "Missing new mode params?"); >> @@ -282,6 +284,9 @@ struct kvm_vm *vm_create(enum vm_guest_mode mode, uint64_t phy_pages, int perm) >> case VM_MODE_P47V64_4K: >> vm->pgtable_levels = 5; >> break; >> + case VM_MODE_P44V64_4K: >> + vm->pgtable_levels = 5; >> + break; >> default: >> TEST_FAIL("Unknown guest mode, mode: 0x%x", mode); >> } >>
On 06.07.21 09:45, Christian Borntraeger wrote: > On 06.07.21 09:40, Christian Borntraeger wrote: >> Paolo, >> >> since you have not yet pulled my queue for 5.14. Shall I add the two selftest patches and send a new >> pull request? > > Hmm, I cant put it on top of the next queue since I would need to rebase. > So lets do the original pull request and I will do another one > on top of kvm/master for the 2 selftest patches. And I just realized that you did already pull the s390x and sent it to Linus. Pull request for the selftest patches will follow soon.
diff --git a/tools/testing/selftests/kvm/include/kvm_util.h b/tools/testing/selftests/kvm/include/kvm_util.h index 35739567189e..74d73532fce9 100644 --- a/tools/testing/selftests/kvm/include/kvm_util.h +++ b/tools/testing/selftests/kvm/include/kvm_util.h @@ -44,6 +44,7 @@ enum vm_guest_mode { VM_MODE_P40V48_64K, VM_MODE_PXXV48_4K, /* For 48bits VA but ANY bits PA */ VM_MODE_P47V64_4K, + VM_MODE_P44V64_4K, NUM_VM_MODES, }; @@ -61,7 +62,7 @@ enum vm_guest_mode { #elif defined(__s390x__) -#define VM_MODE_DEFAULT VM_MODE_P47V64_4K +#define VM_MODE_DEFAULT VM_MODE_P44V64_4K #define MIN_PAGE_SHIFT 12U #define ptes_per_page(page_size) ((page_size) / 16) diff --git a/tools/testing/selftests/kvm/lib/guest_modes.c b/tools/testing/selftests/kvm/lib/guest_modes.c index 25bff307c71f..c330f414ef96 100644 --- a/tools/testing/selftests/kvm/lib/guest_modes.c +++ b/tools/testing/selftests/kvm/lib/guest_modes.c @@ -22,6 +22,22 @@ void guest_modes_append_default(void) } } #endif +#ifdef __s390x__ + { + int kvm_fd, vm_fd; + struct kvm_s390_vm_cpu_processor info; + + kvm_fd = open_kvm_dev_path_or_exit(); + vm_fd = ioctl(kvm_fd, KVM_CREATE_VM, 0); + kvm_device_access(vm_fd, KVM_S390_VM_CPU_MODEL, + KVM_S390_VM_CPU_PROCESSOR, &info, false); + close(vm_fd); + close(kvm_fd); + /* Starting with z13 we have 47bits of physical address */ + if (info.ibc >= 0x30) + guest_mode_append(VM_MODE_P47V64_4K, true, true); + } +#endif } void for_each_guest_mode(void (*func)(enum vm_guest_mode, void *), void *arg) diff --git a/tools/testing/selftests/kvm/lib/kvm_util.c b/tools/testing/selftests/kvm/lib/kvm_util.c index a2b732cf96ea..8606000c439e 100644 --- a/tools/testing/selftests/kvm/lib/kvm_util.c +++ b/tools/testing/selftests/kvm/lib/kvm_util.c @@ -176,6 +176,7 @@ const char *vm_guest_mode_string(uint32_t i) [VM_MODE_P40V48_64K] = "PA-bits:40, VA-bits:48, 64K pages", [VM_MODE_PXXV48_4K] = "PA-bits:ANY, VA-bits:48, 4K pages", [VM_MODE_P47V64_4K] = "PA-bits:47, VA-bits:64, 4K pages", + [VM_MODE_P44V64_4K] = "PA-bits:44, VA-bits:64, 4K pages", }; _Static_assert(sizeof(strings)/sizeof(char *) == NUM_VM_MODES, "Missing new mode strings?"); @@ -194,6 +195,7 @@ const struct vm_guest_mode_params vm_guest_mode_params[] = { { 40, 48, 0x10000, 16 }, { 0, 0, 0x1000, 12 }, { 47, 64, 0x1000, 12 }, + { 44, 64, 0x1000, 12 }, }; _Static_assert(sizeof(vm_guest_mode_params)/sizeof(struct vm_guest_mode_params) == NUM_VM_MODES, "Missing new mode params?"); @@ -282,6 +284,9 @@ struct kvm_vm *vm_create(enum vm_guest_mode mode, uint64_t phy_pages, int perm) case VM_MODE_P47V64_4K: vm->pgtable_levels = 5; break; + case VM_MODE_P44V64_4K: + vm->pgtable_levels = 5; + break; default: TEST_FAIL("Unknown guest mode, mode: 0x%x", mode); }
Older machines likes z196 and zEC12 do only support 44 bits of physical addresses. Make this the default and check via IBC if we are on a later machine. We then add P47V64 as an additional model. Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com> Fixes: 1bc603af73dd ("KVM: selftests: introduce P47V64 for s390x") --- tools/testing/selftests/kvm/include/kvm_util.h | 3 ++- tools/testing/selftests/kvm/lib/guest_modes.c | 16 ++++++++++++++++ tools/testing/selftests/kvm/lib/kvm_util.c | 5 +++++ 3 files changed, 23 insertions(+), 1 deletion(-)