| Message ID | 20221213165405.2953539-2-scgl@linux.ibm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | KVM: s390: Extend MEM_OP ioctl by storage key checked cmpxchg | expand |
On 13/12/2022 17.53, Janis Schoetterl-Glausch wrote: > User space can use the MEM_OP ioctl to make storage key checked reads > and writes to the guest, however, it has no way of performing atomic, > key checked, accesses to the guest. > Extend the MEM_OP ioctl in order to allow for this, by adding a cmpxchg > mode. For now, support this mode for absolute accesses only. > > This mode can be use, for example, to set the device-state-change > indicator and the adapter-local-summary indicator atomically. > > Signed-off-by: Janis Schoetterl-Glausch <scgl@linux.ibm.com> > --- > include/uapi/linux/kvm.h | 7 +++ > arch/s390/kvm/gaccess.h | 3 ++ > arch/s390/kvm/gaccess.c | 102 +++++++++++++++++++++++++++++++++++++++ > arch/s390/kvm/kvm-s390.c | 39 ++++++++++++++- > 4 files changed, 149 insertions(+), 2 deletions(-) > > diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h > index 0d5d4419139a..f106db1af5ee 100644 > --- a/include/uapi/linux/kvm.h > +++ b/include/uapi/linux/kvm.h > @@ -588,6 +588,8 @@ struct kvm_s390_mem_op { > struct { > __u8 ar; /* the access register number */ > __u8 key; /* access key, ignored if flag unset */ > + __u8 pad1[6]; /* ignored */ > + __u64 old_addr; /* ignored if flag unset */ > }; > __u32 sida_offset; /* offset into the sida */ > __u8 reserved[32]; /* ignored */ > @@ -604,6 +606,11 @@ struct kvm_s390_mem_op { > #define KVM_S390_MEMOP_F_CHECK_ONLY (1ULL << 0) > #define KVM_S390_MEMOP_F_INJECT_EXCEPTION (1ULL << 1) > #define KVM_S390_MEMOP_F_SKEY_PROTECTION (1ULL << 2) > +#define KVM_S390_MEMOP_F_CMPXCHG (1ULL << 3) > +/* flags specifying extension support */ > +#define KVM_S390_MEMOP_EXTENSION_CAP_CMPXCHG 0x2 > +/* Non program exception return codes (pgm codes are 16 bit) */ > +#define KVM_S390_MEMOP_R_NO_XCHG (1 << 16) > > /* for KVM_INTERRUPT */ > struct kvm_interrupt { > diff --git a/arch/s390/kvm/gaccess.h b/arch/s390/kvm/gaccess.h > index 9408d6cc8e2c..92a3b9fb31ec 100644 > --- a/arch/s390/kvm/gaccess.h > +++ b/arch/s390/kvm/gaccess.h > @@ -206,6 +206,9 @@ int access_guest_with_key(struct kvm_vcpu *vcpu, unsigned long ga, u8 ar, > int access_guest_real(struct kvm_vcpu *vcpu, unsigned long gra, > void *data, unsigned long len, enum gacc_mode mode); > > +int cmpxchg_guest_abs_with_key(struct kvm *kvm, gpa_t gpa, int len, > + __uint128_t *old, __uint128_t new, u8 access_key); > + > /** > * write_guest_with_key - copy data from kernel space to guest space > * @vcpu: virtual cpu > diff --git a/arch/s390/kvm/gaccess.c b/arch/s390/kvm/gaccess.c > index 0243b6e38d36..6165e761a637 100644 > --- a/arch/s390/kvm/gaccess.c > +++ b/arch/s390/kvm/gaccess.c > @@ -1161,6 +1161,108 @@ int access_guest_real(struct kvm_vcpu *vcpu, unsigned long gra, > return rc; > } > > +/** > + * cmpxchg_guest_abs_with_key() - Perform cmpxchg on guest absolute address. > + * @kvm: Virtual machine instance. > + * @gpa: Absolute guest address of the location to be changed. > + * @len: Operand length of the cmpxchg, required: 1 <= len <= 16. Providing a > + * non power of two will result in failure. > + * @old_addr: Pointer to old value. If the location at @gpa contains this value, the > + * exchange will succeed. After calling cmpxchg_guest_abs_with_key() *@old > + * contains the value at @gpa before the attempt to exchange the value. > + * @new: The value to place at @gpa. > + * @access_key: The access key to use for the guest access. > + * > + * Atomically exchange the value at @gpa by @new, if it contains *@old. > + * Honors storage keys. > + * > + * Return: * 0: successful exchange > + * * 1: exchange unsuccessful > + * * a program interruption code indicating the reason cmpxchg could > + * not be attempted > + * * -EINVAL: address misaligned or len not power of two > + * * -EAGAIN: transient failure (len 1 or 2) > + * * -EOPNOTSUPP: read-only memslot (should never occur) > + */ > +int cmpxchg_guest_abs_with_key(struct kvm *kvm, gpa_t gpa, int len, > + __uint128_t *old_addr, __uint128_t new, > + u8 access_key) > +{ > + gfn_t gfn = gpa >> PAGE_SHIFT; > + struct kvm_memory_slot *slot = gfn_to_memslot(kvm, gfn); > + bool writable; > + hva_t hva; > + int ret; > + > + if (!IS_ALIGNED(gpa, len)) > + return -EINVAL; > + > + hva = gfn_to_hva_memslot_prot(slot, gfn, &writable); > + if (kvm_is_error_hva(hva)) > + return PGM_ADDRESSING; > + /* > + * Check if it's a read-only memslot, even though that cannot occur > + * since those are unsupported. > + * Don't try to actually handle that case. > + */ > + if (!writable) > + return -EOPNOTSUPP; > + > + hva += offset_in_page(gpa); > + switch (len) { > + case 1: { > + u8 old; > + > + ret = cmpxchg_user_key((u8 *)hva, &old, *old_addr, new, access_key); > + ret = ret < 0 ? ret : old != *old_addr; > + *old_addr = old; > + break; > + } > + case 2: { > + u16 old; > + > + ret = cmpxchg_user_key((u16 *)hva, &old, *old_addr, new, access_key); > + ret = ret < 0 ? ret : old != *old_addr; > + *old_addr = old; > + break; > + } > + case 4: { > + u32 old; > + > + ret = cmpxchg_user_key((u32 *)hva, &old, *old_addr, new, access_key); > + ret = ret < 0 ? ret : old != *old_addr; > + *old_addr = old; > + break; > + } > + case 8: { > + u64 old; > + > + ret = cmpxchg_user_key((u64 *)hva, &old, *old_addr, new, access_key); > + ret = ret < 0 ? ret : old != *old_addr; > + *old_addr = old; > + break; > + } > + case 16: { > + __uint128_t old; > + > + ret = cmpxchg_user_key((__uint128_t *)hva, &old, *old_addr, new, access_key); > + ret = ret < 0 ? ret : old != *old_addr; > + *old_addr = old; > + break; > + } > + default: > + return -EINVAL; > + } > + mark_page_dirty_in_slot(kvm, slot, gfn); > + /* > + * Assume that the fault is caused by protection, either key protection > + * or user page write protection. > + */ > + if (ret == -EFAULT) > + ret = PGM_PROTECTION; > + return ret; > +} > + > /** > * guest_translate_address_with_key - translate guest logical into guest absolute address > * @vcpu: virtual cpu > diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c > index 45d4b8182b07..47bcf2cb4345 100644 > --- a/arch/s390/kvm/kvm-s390.c > +++ b/arch/s390/kvm/kvm-s390.c > @@ -576,7 +576,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) > case KVM_CAP_S390_VCPU_RESETS: > case KVM_CAP_SET_GUEST_DEBUG: > case KVM_CAP_S390_DIAG318: > - case KVM_CAP_S390_MEM_OP_EXTENSION: > r = 1; > break; > case KVM_CAP_SET_GUEST_DEBUG2: > @@ -590,6 +589,14 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) > case KVM_CAP_S390_MEM_OP: > r = MEM_OP_MAX_SIZE; > break; > + case KVM_CAP_S390_MEM_OP_EXTENSION: > + /* > + * Flag bits indicating which extensions are supported. > + * The first extension doesn't use a flag, but pretend it does, > + * this way that can be changed in the future. > + */ > + r = KVM_S390_MEMOP_EXTENSION_CAP_CMPXCHG | 1; > + break; > case KVM_CAP_NR_VCPUS: > case KVM_CAP_MAX_VCPUS: > case KVM_CAP_MAX_VCPU_ID: > @@ -2714,12 +2721,19 @@ static bool access_key_invalid(u8 access_key) > static int kvm_s390_vm_mem_op(struct kvm *kvm, struct kvm_s390_mem_op *mop) > { > void __user *uaddr = (void __user *)mop->buf; > + void __user *old_addr = (void __user *)mop->old_addr; > + union { > + __uint128_t quad; > + char raw[sizeof(__uint128_t)]; > + } old = { .quad = 0}, new = { .quad = 0 }; > + unsigned int off_in_quad = sizeof(new) - mop->size; > u64 supported_flags; > void *tmpbuf = NULL; > int r, srcu_idx; > > supported_flags = KVM_S390_MEMOP_F_SKEY_PROTECTION > - | KVM_S390_MEMOP_F_CHECK_ONLY; > + | KVM_S390_MEMOP_F_CHECK_ONLY > + | KVM_S390_MEMOP_F_CMPXCHG; > if (mop->flags & ~supported_flags || !mop->size) > return -EINVAL; > if (mop->size > MEM_OP_MAX_SIZE) > @@ -2741,6 +2755,19 @@ static int kvm_s390_vm_mem_op(struct kvm *kvm, struct kvm_s390_mem_op *mop) > } else { > mop->key = 0; > } > + if (mop->flags & KVM_S390_MEMOP_F_CMPXCHG) { > + /* > + * This validates off_in_quad. Checking that size is a power > + * of two is not necessary, as cmpxchg_guest_abs_with_key > + * takes care of that > + */ > + if (mop->size > sizeof(new)) > + return -EINVAL; I'd maybe add a check for mop->op == KVM_S390_MEMOP_ABSOLUTE_WRITE here, since calling the _READ function with the F_CMPXCHG flag set does not make too much sense. Anyway, patch looks good to me, so with or without that additional check: Reviewed-by: Thomas Huth <thuth@redhat.com> > + if (copy_from_user(&new.raw[off_in_quad], uaddr, mop->size)) > + return -EFAULT; > + if (copy_from_user(&old.raw[off_in_quad], old_addr, mop->size)) > + return -EFAULT; > + } > if (!(mop->flags & KVM_S390_MEMOP_F_CHECK_ONLY)) { > tmpbuf = vmalloc(mop->size); > if (!tmpbuf) > @@ -2771,6 +2798,14 @@ static int kvm_s390_vm_mem_op(struct kvm *kvm, struct kvm_s390_mem_op *mop) > case KVM_S390_MEMOP_ABSOLUTE_WRITE: { > if (mop->flags & KVM_S390_MEMOP_F_CHECK_ONLY) { > r = check_gpa_range(kvm, mop->gaddr, mop->size, GACC_STORE, mop->key); > + } else if (mop->flags & KVM_S390_MEMOP_F_CMPXCHG) { > + r = cmpxchg_guest_abs_with_key(kvm, mop->gaddr, mop->size, > + &old.quad, new.quad, mop->key); > + if (r == 1) { > + r = KVM_S390_MEMOP_R_NO_XCHG; > + if (copy_to_user(old_addr, &old.raw[off_in_quad], mop->size)) > + r = -EFAULT; > + } > } else { > if (copy_from_user(tmpbuf, uaddr, mop->size)) { > r = -EFAULT;
On Wed, 2022-12-14 at 10:19 +0100, Thomas Huth wrote: > On 13/12/2022 17.53, Janis Schoetterl-Glausch wrote: > > User space can use the MEM_OP ioctl to make storage key checked reads > > and writes to the guest, however, it has no way of performing atomic, > > key checked, accesses to the guest. > > Extend the MEM_OP ioctl in order to allow for this, by adding a cmpxchg > > mode. For now, support this mode for absolute accesses only. > > > > This mode can be use, for example, to set the device-state-change > > indicator and the adapter-local-summary indicator atomically. > > > > Signed-off-by: Janis Schoetterl-Glausch <scgl@linux.ibm.com> > > --- > > include/uapi/linux/kvm.h | 7 +++ > > arch/s390/kvm/gaccess.h | 3 ++ > > arch/s390/kvm/gaccess.c | 102 +++++++++++++++++++++++++++++++++++++++ > > arch/s390/kvm/kvm-s390.c | 39 ++++++++++++++- > > 4 files changed, 149 insertions(+), 2 deletions(-) > > [...] > > > > @@ -2714,12 +2721,19 @@ static bool access_key_invalid(u8 access_key) > > static int kvm_s390_vm_mem_op(struct kvm *kvm, struct kvm_s390_mem_op *mop) > > { > > void __user *uaddr = (void __user *)mop->buf; > > + void __user *old_addr = (void __user *)mop->old_addr; > > + union { > > + __uint128_t quad; > > + char raw[sizeof(__uint128_t)]; > > + } old = { .quad = 0}, new = { .quad = 0 }; > > + unsigned int off_in_quad = sizeof(new) - mop->size; > > u64 supported_flags; > > void *tmpbuf = NULL; > > int r, srcu_idx; > > > > supported_flags = KVM_S390_MEMOP_F_SKEY_PROTECTION > > - | KVM_S390_MEMOP_F_CHECK_ONLY; > > + | KVM_S390_MEMOP_F_CHECK_ONLY > > + | KVM_S390_MEMOP_F_CMPXCHG; > > if (mop->flags & ~supported_flags || !mop->size) > > return -EINVAL; > > if (mop->size > MEM_OP_MAX_SIZE) > > @@ -2741,6 +2755,19 @@ static int kvm_s390_vm_mem_op(struct kvm *kvm, struct kvm_s390_mem_op *mop) > > } else { > > mop->key = 0; > > } > > + if (mop->flags & KVM_S390_MEMOP_F_CMPXCHG) { > > + /* > > + * This validates off_in_quad. Checking that size is a power > > + * of two is not necessary, as cmpxchg_guest_abs_with_key > > + * takes care of that > > + */ > > + if (mop->size > sizeof(new)) > > + return -EINVAL; > > I'd maybe add a check for mop->op == KVM_S390_MEMOP_ABSOLUTE_WRITE here, > since calling the _READ function with the F_CMPXCHG flag set does not make > too much sense. Good point. > > Anyway, patch looks good to me, so with or without that additional check: > Reviewed-by: Thomas Huth <thuth@redhat.com> Thanks! > > > + if (copy_from_user(&new.raw[off_in_quad], uaddr, mop->size)) > > + return -EFAULT; > > + if (copy_from_user(&old.raw[off_in_quad], old_addr, mop->size)) > > + return -EFAULT; > > + } > > if (!(mop->flags & KVM_S390_MEMOP_F_CHECK_ONLY)) { > > tmpbuf = vmalloc(mop->size); > > if (!tmpbuf) > > @@ -2771,6 +2798,14 @@ static int kvm_s390_vm_mem_op(struct kvm *kvm, struct kvm_s390_mem_op *mop) > > case KVM_S390_MEMOP_ABSOLUTE_WRITE: { > > if (mop->flags & KVM_S390_MEMOP_F_CHECK_ONLY) { > > r = check_gpa_range(kvm, mop->gaddr, mop->size, GACC_STORE, mop->key); > > + } else if (mop->flags & KVM_S390_MEMOP_F_CMPXCHG) { > > + r = cmpxchg_guest_abs_with_key(kvm, mop->gaddr, mop->size, > > + &old.quad, new.quad, mop->key); > > + if (r == 1) { > > + r = KVM_S390_MEMOP_R_NO_XCHG; > > + if (copy_to_user(old_addr, &old.raw[off_in_quad], mop->size)) > > + r = -EFAULT; > > + } > > } else { > > if (copy_from_user(tmpbuf, uaddr, mop->size)) { > > r = -EFAULT; >
On Wed, 2022-12-14 at 10:23 +0800, kernel test robot wrote: > Hi Janis, > > Thank you for the patch! Yet something to improve: > > [auto build test ERROR on 739ad2e4e15b585a0eaf98b7bdee62b2dd9588c9] > > url: https://github.com/intel-lab-lkp/linux/commits/Janis-Schoetterl-Glausch/KVM-s390-Extend-MEM_OP-ioctl-by-storage-key-checked-cmpxchg/20221214-005540 > base: 739ad2e4e15b585a0eaf98b7bdee62b2dd9588c9 > patch link: https://lore.kernel.org/r/20221213165405.2953539-2-scgl%40linux.ibm.com > patch subject: [PATCH v4 1/9] KVM: s390: Extend MEM_OP ioctl by storage key checked cmpxchg > config: s390-randconfig-r004-20221213 > compiler: s390-linux-gcc (GCC) 12.1.0 > reproduce (this is a W=1 build): > wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross > chmod +x ~/bin/make.cross > # https://github.com/intel-lab-lkp/linux/commit/4e0991bd47ba30c7588e042da7a84d84b9f84056 > git remote add linux-review https://github.com/intel-lab-lkp/linux > git fetch --no-tags linux-review Janis-Schoetterl-Glausch/KVM-s390-Extend-MEM_OP-ioctl-by-storage-key-checked-cmpxchg/20221214-005540 > git checkout 4e0991bd47ba30c7588e042da7a84d84b9f84056 > # save the config file > mkdir build_dir && cp config build_dir/.config > COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.1.0 make.cross W=1 O=build_dir ARCH=s390 SHELL=/bin/bash > > If you fix the issue, kindly add following tag where applicable > > Reported-by: kernel test robot <lkp@intel.com> > > All errors (new ones prefixed by >>): > > s390-linux-ld: arch/s390/kvm/gaccess.o: in function `__cmpxchg_user_key': > > > arch/s390/include/asm/uaccess.h:410: undefined reference to `__ashlti3' > > > s390-linux-ld: arch/s390/include/asm/uaccess.h:411: undefined reference to `__ashlti3' > s390-linux-ld: arch/s390/include/asm/uaccess.h:458: undefined reference to `__ashlti3' > s390-linux-ld: arch/s390/include/asm/uaccess.h:459: undefined reference to `__ashlti3' > > > vim +410 arch/s390/include/asm/uaccess.h > > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 396 > 4148575abe1e14 Heiko Carstens 2022-11-02 397 static __always_inline int __cmpxchg_user_key(unsigned long address, void *uval, > 4148575abe1e14 Heiko Carstens 2022-11-02 398 __uint128_t old, __uint128_t new, > 4148575abe1e14 Heiko Carstens 2022-11-02 399 unsigned long key, int size) > 4148575abe1e14 Heiko Carstens 2022-11-02 400 { > 4148575abe1e14 Heiko Carstens 2022-11-02 401 int rc = 0; > 4148575abe1e14 Heiko Carstens 2022-11-02 402 > 4148575abe1e14 Heiko Carstens 2022-11-02 403 switch (size) { > 4148575abe1e14 Heiko Carstens 2022-11-02 404 case 1: { > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 405 unsigned int prev, shift, mask, _old, _new; > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 406 unsigned long count; > 4148575abe1e14 Heiko Carstens 2022-11-02 407 > 4148575abe1e14 Heiko Carstens 2022-11-02 408 shift = (3 ^ (address & 3)) << 3; > 4148575abe1e14 Heiko Carstens 2022-11-02 409 address ^= address & 3; > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 @410 _old = (old & 0xff) << shift; > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 @411 _new = (new & 0xff) << shift; Not sure what it is in this config that causes gcc to emit this symbol instead of a shift instruction, but casting old/new to 32 bit fixes the error. > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 412 mask = ~(0xff << shift); > 4148575abe1e14 Heiko Carstens 2022-11-02 413 asm volatile( > 4148575abe1e14 Heiko Carstens 2022-11-02 414 " spka 0(%[key])\n" > 4148575abe1e14 Heiko Carstens 2022-11-02 415 " sacf 256\n" > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 416 " llill %[count],%[max_loops]\n" > 4148575abe1e14 Heiko Carstens 2022-11-02 417 "0: l %[prev],%[address]\n" > 4148575abe1e14 Heiko Carstens 2022-11-02 418 "1: nr %[prev],%[mask]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 419 " xilf %[mask],0xffffffff\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 420 " or %[new],%[prev]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 421 " or %[prev],%[tmp]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 422 "2: lr %[tmp],%[prev]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 423 "3: cs %[prev],%[new],%[address]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 424 "4: jnl 5f\n" > 4148575abe1e14 Heiko Carstens 2022-11-02 425 " xr %[tmp],%[prev]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 426 " xr %[new],%[tmp]\n" > 4148575abe1e14 Heiko Carstens 2022-11-02 427 " nr %[tmp],%[mask]\n" > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 428 " jnz 5f\n" > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 429 " brct %[count],2b\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 430 "5: sacf 768\n" > 4148575abe1e14 Heiko Carstens 2022-11-02 431 " spka %[default_key]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 432 EX_TABLE_UA_LOAD_REG(0b, 5b, %[rc], %[prev]) > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 433 EX_TABLE_UA_LOAD_REG(1b, 5b, %[rc], %[prev]) > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 434 EX_TABLE_UA_LOAD_REG(3b, 5b, %[rc], %[prev]) > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 435 EX_TABLE_UA_LOAD_REG(4b, 5b, %[rc], %[prev]) > 4148575abe1e14 Heiko Carstens 2022-11-02 436 : [rc] "+&d" (rc), > 4148575abe1e14 Heiko Carstens 2022-11-02 437 [prev] "=&d" (prev), > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 438 [address] "+Q" (*(int *)address), > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 439 [tmp] "+&d" (_old), > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 440 [new] "+&d" (_new), > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 441 [mask] "+&d" (mask), > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 442 [count] "=a" (count) > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 443 : [key] "%[count]" (key << 4), > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 444 [default_key] "J" (PAGE_DEFAULT_KEY), > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 445 [max_loops] "J" (CMPXCHG_USER_KEY_MAX_LOOPS) > 4148575abe1e14 Heiko Carstens 2022-11-02 446 : "memory", "cc"); > 4148575abe1e14 Heiko Carstens 2022-11-02 447 *(unsigned char *)uval = prev >> shift; > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 448 if (!count) > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 449 rc = -EAGAIN; > 4148575abe1e14 Heiko Carstens 2022-11-02 450 return rc; > 4148575abe1e14 Heiko Carstens 2022-11-02 451 } > 4148575abe1e14 Heiko Carstens 2022-11-02 452 case 2: { > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 453 unsigned int prev, shift, mask, _old, _new; > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 454 unsigned long count; > 4148575abe1e14 Heiko Carstens 2022-11-02 455 > 4148575abe1e14 Heiko Carstens 2022-11-02 456 shift = (2 ^ (address & 2)) << 3; > 4148575abe1e14 Heiko Carstens 2022-11-02 457 address ^= address & 2; > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 458 _old = (old & 0xffff) << shift; > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 459 _new = (new & 0xffff) << shift; Same here. > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 460 mask = ~(0xffff << shift); > 4148575abe1e14 Heiko Carstens 2022-11-02 461 asm volatile( > 4148575abe1e14 Heiko Carstens 2022-11-02 462 " spka 0(%[key])\n" > 4148575abe1e14 Heiko Carstens 2022-11-02 463 " sacf 256\n" > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 464 " llill %[count],%[max_loops]\n" > 4148575abe1e14 Heiko Carstens 2022-11-02 465 "0: l %[prev],%[address]\n" > 4148575abe1e14 Heiko Carstens 2022-11-02 466 "1: nr %[prev],%[mask]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 467 " xilf %[mask],0xffffffff\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 468 " or %[new],%[prev]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 469 " or %[prev],%[tmp]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 470 "2: lr %[tmp],%[prev]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 471 "3: cs %[prev],%[new],%[address]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 472 "4: jnl 5f\n" > 4148575abe1e14 Heiko Carstens 2022-11-02 473 " xr %[tmp],%[prev]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 474 " xr %[new],%[tmp]\n" > 4148575abe1e14 Heiko Carstens 2022-11-02 475 " nr %[tmp],%[mask]\n" > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 476 " jnz 5f\n" > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 477 " brct %[count],2b\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 478 "5: sacf 768\n" > 4148575abe1e14 Heiko Carstens 2022-11-02 479 " spka %[default_key]\n" > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 480 EX_TABLE_UA_LOAD_REG(0b, 5b, %[rc], %[prev]) > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 481 EX_TABLE_UA_LOAD_REG(1b, 5b, %[rc], %[prev]) > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 482 EX_TABLE_UA_LOAD_REG(3b, 5b, %[rc], %[prev]) > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 483 EX_TABLE_UA_LOAD_REG(4b, 5b, %[rc], %[prev]) > 4148575abe1e14 Heiko Carstens 2022-11-02 484 : [rc] "+&d" (rc), > 4148575abe1e14 Heiko Carstens 2022-11-02 485 [prev] "=&d" (prev), > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 486 [address] "+Q" (*(int *)address), > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 487 [tmp] "+&d" (_old), > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 488 [new] "+&d" (_new), > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 489 [mask] "+&d" (mask), > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 490 [count] "=a" (count) > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 491 : [key] "%[count]" (key << 4), > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 492 [default_key] "J" (PAGE_DEFAULT_KEY), > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 493 [max_loops] "J" (CMPXCHG_USER_KEY_MAX_LOOPS) > 4148575abe1e14 Heiko Carstens 2022-11-02 494 : "memory", "cc"); > 4148575abe1e14 Heiko Carstens 2022-11-02 495 *(unsigned short *)uval = prev >> shift; > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 496 if (!count) > 739ad2e4e15b58 Janis Schoetterl-Glausch 2022-11-17 497 rc = -EAGAIN; > 4148575abe1e14 Heiko Carstens 2022-11-02 498 return rc; [...]
On Mon, Dec 19, 2022 at 10:24:33PM +0100, Janis Schoetterl-Glausch wrote: > On Wed, 2022-12-14 at 10:23 +0800, kernel test robot wrote: > > Hi Janis, > > > > Thank you for the patch! Yet something to improve: > > All errors (new ones prefixed by >>): > > > > s390-linux-ld: arch/s390/kvm/gaccess.o: in function `__cmpxchg_user_key': > > > > arch/s390/include/asm/uaccess.h:410: undefined reference to `__ashlti3' > > > > s390-linux-ld: arch/s390/include/asm/uaccess.h:411: undefined reference to `__ashlti3' > > s390-linux-ld: arch/s390/include/asm/uaccess.h:458: undefined reference to `__ashlti3' > > s390-linux-ld: arch/s390/include/asm/uaccess.h:459: undefined reference to `__ashlti3' > > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 @410 _old = (old & 0xff) << shift; > > 51098f0eb22e2f Janis Schoetterl-Glausch 2022-11-16 @411 _new = (new & 0xff) << shift; > > Not sure what it is in this config that causes gcc to emit this > symbol instead of a shift instruction, but casting old/new to 32 bit > fixes the error. Right.. now we have the same fun with 128 bit arithmetics that we had with 64 bit arithmetics on 32 bit. I really missed that :) Fixed the way you proposed it: https://git.kernel.org/pub/scm/linux/kernel/git/s390/linux.git/commit/?h=cmpxchg_user_key&id=b33d59fb37ddcb6ee65d4fa23cc3d58793d13c5b
diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h index 0d5d4419139a..f106db1af5ee 100644 --- a/include/uapi/linux/kvm.h +++ b/include/uapi/linux/kvm.h @@ -588,6 +588,8 @@ struct kvm_s390_mem_op { struct { __u8 ar; /* the access register number */ __u8 key; /* access key, ignored if flag unset */ + __u8 pad1[6]; /* ignored */ + __u64 old_addr; /* ignored if flag unset */ }; __u32 sida_offset; /* offset into the sida */ __u8 reserved[32]; /* ignored */ @@ -604,6 +606,11 @@ struct kvm_s390_mem_op { #define KVM_S390_MEMOP_F_CHECK_ONLY (1ULL << 0) #define KVM_S390_MEMOP_F_INJECT_EXCEPTION (1ULL << 1) #define KVM_S390_MEMOP_F_SKEY_PROTECTION (1ULL << 2) +#define KVM_S390_MEMOP_F_CMPXCHG (1ULL << 3) +/* flags specifying extension support */ +#define KVM_S390_MEMOP_EXTENSION_CAP_CMPXCHG 0x2 +/* Non program exception return codes (pgm codes are 16 bit) */ +#define KVM_S390_MEMOP_R_NO_XCHG (1 << 16) /* for KVM_INTERRUPT */ struct kvm_interrupt { diff --git a/arch/s390/kvm/gaccess.h b/arch/s390/kvm/gaccess.h index 9408d6cc8e2c..92a3b9fb31ec 100644 --- a/arch/s390/kvm/gaccess.h +++ b/arch/s390/kvm/gaccess.h @@ -206,6 +206,9 @@ int access_guest_with_key(struct kvm_vcpu *vcpu, unsigned long ga, u8 ar, int access_guest_real(struct kvm_vcpu *vcpu, unsigned long gra, void *data, unsigned long len, enum gacc_mode mode); +int cmpxchg_guest_abs_with_key(struct kvm *kvm, gpa_t gpa, int len, + __uint128_t *old, __uint128_t new, u8 access_key); + /** * write_guest_with_key - copy data from kernel space to guest space * @vcpu: virtual cpu diff --git a/arch/s390/kvm/gaccess.c b/arch/s390/kvm/gaccess.c index 0243b6e38d36..6165e761a637 100644 --- a/arch/s390/kvm/gaccess.c +++ b/arch/s390/kvm/gaccess.c @@ -1161,6 +1161,108 @@ int access_guest_real(struct kvm_vcpu *vcpu, unsigned long gra, return rc; } +/** + * cmpxchg_guest_abs_with_key() - Perform cmpxchg on guest absolute address. + * @kvm: Virtual machine instance. + * @gpa: Absolute guest address of the location to be changed. + * @len: Operand length of the cmpxchg, required: 1 <= len <= 16. Providing a + * non power of two will result in failure. + * @old_addr: Pointer to old value. If the location at @gpa contains this value, the + * exchange will succeed. After calling cmpxchg_guest_abs_with_key() *@old + * contains the value at @gpa before the attempt to exchange the value. + * @new: The value to place at @gpa. + * @access_key: The access key to use for the guest access. + * + * Atomically exchange the value at @gpa by @new, if it contains *@old. + * Honors storage keys. + * + * Return: * 0: successful exchange + * * 1: exchange unsuccessful + * * a program interruption code indicating the reason cmpxchg could + * not be attempted + * * -EINVAL: address misaligned or len not power of two + * * -EAGAIN: transient failure (len 1 or 2) + * * -EOPNOTSUPP: read-only memslot (should never occur) + */ +int cmpxchg_guest_abs_with_key(struct kvm *kvm, gpa_t gpa, int len, + __uint128_t *old_addr, __uint128_t new, + u8 access_key) +{ + gfn_t gfn = gpa >> PAGE_SHIFT; + struct kvm_memory_slot *slot = gfn_to_memslot(kvm, gfn); + bool writable; + hva_t hva; + int ret; + + if (!IS_ALIGNED(gpa, len)) + return -EINVAL; + + hva = gfn_to_hva_memslot_prot(slot, gfn, &writable); + if (kvm_is_error_hva(hva)) + return PGM_ADDRESSING; + /* + * Check if it's a read-only memslot, even though that cannot occur + * since those are unsupported. + * Don't try to actually handle that case. + */ + if (!writable) + return -EOPNOTSUPP; + + hva += offset_in_page(gpa); + switch (len) { + case 1: { + u8 old; + + ret = cmpxchg_user_key((u8 *)hva, &old, *old_addr, new, access_key); + ret = ret < 0 ? ret : old != *old_addr; + *old_addr = old; + break; + } + case 2: { + u16 old; + + ret = cmpxchg_user_key((u16 *)hva, &old, *old_addr, new, access_key); + ret = ret < 0 ? ret : old != *old_addr; + *old_addr = old; + break; + } + case 4: { + u32 old; + + ret = cmpxchg_user_key((u32 *)hva, &old, *old_addr, new, access_key); + ret = ret < 0 ? ret : old != *old_addr; + *old_addr = old; + break; + } + case 8: { + u64 old; + + ret = cmpxchg_user_key((u64 *)hva, &old, *old_addr, new, access_key); + ret = ret < 0 ? ret : old != *old_addr; + *old_addr = old; + break; + } + case 16: { + __uint128_t old; + + ret = cmpxchg_user_key((__uint128_t *)hva, &old, *old_addr, new, access_key); + ret = ret < 0 ? ret : old != *old_addr; + *old_addr = old; + break; + } + default: + return -EINVAL; + } + mark_page_dirty_in_slot(kvm, slot, gfn); + /* + * Assume that the fault is caused by protection, either key protection + * or user page write protection. + */ + if (ret == -EFAULT) + ret = PGM_PROTECTION; + return ret; +} + /** * guest_translate_address_with_key - translate guest logical into guest absolute address * @vcpu: virtual cpu diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c index 45d4b8182b07..47bcf2cb4345 100644 --- a/arch/s390/kvm/kvm-s390.c +++ b/arch/s390/kvm/kvm-s390.c @@ -576,7 +576,6 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) case KVM_CAP_S390_VCPU_RESETS: case KVM_CAP_SET_GUEST_DEBUG: case KVM_CAP_S390_DIAG318: - case KVM_CAP_S390_MEM_OP_EXTENSION: r = 1; break; case KVM_CAP_SET_GUEST_DEBUG2: @@ -590,6 +589,14 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) case KVM_CAP_S390_MEM_OP: r = MEM_OP_MAX_SIZE; break; + case KVM_CAP_S390_MEM_OP_EXTENSION: + /* + * Flag bits indicating which extensions are supported. + * The first extension doesn't use a flag, but pretend it does, + * this way that can be changed in the future. + */ + r = KVM_S390_MEMOP_EXTENSION_CAP_CMPXCHG | 1; + break; case KVM_CAP_NR_VCPUS: case KVM_CAP_MAX_VCPUS: case KVM_CAP_MAX_VCPU_ID: @@ -2714,12 +2721,19 @@ static bool access_key_invalid(u8 access_key) static int kvm_s390_vm_mem_op(struct kvm *kvm, struct kvm_s390_mem_op *mop) { void __user *uaddr = (void __user *)mop->buf; + void __user *old_addr = (void __user *)mop->old_addr; + union { + __uint128_t quad; + char raw[sizeof(__uint128_t)]; + } old = { .quad = 0}, new = { .quad = 0 }; + unsigned int off_in_quad = sizeof(new) - mop->size; u64 supported_flags; void *tmpbuf = NULL; int r, srcu_idx; supported_flags = KVM_S390_MEMOP_F_SKEY_PROTECTION - | KVM_S390_MEMOP_F_CHECK_ONLY; + | KVM_S390_MEMOP_F_CHECK_ONLY + | KVM_S390_MEMOP_F_CMPXCHG; if (mop->flags & ~supported_flags || !mop->size) return -EINVAL; if (mop->size > MEM_OP_MAX_SIZE) @@ -2741,6 +2755,19 @@ static int kvm_s390_vm_mem_op(struct kvm *kvm, struct kvm_s390_mem_op *mop) } else { mop->key = 0; } + if (mop->flags & KVM_S390_MEMOP_F_CMPXCHG) { + /* + * This validates off_in_quad. Checking that size is a power + * of two is not necessary, as cmpxchg_guest_abs_with_key + * takes care of that + */ + if (mop->size > sizeof(new)) + return -EINVAL; + if (copy_from_user(&new.raw[off_in_quad], uaddr, mop->size)) + return -EFAULT; + if (copy_from_user(&old.raw[off_in_quad], old_addr, mop->size)) + return -EFAULT; + } if (!(mop->flags & KVM_S390_MEMOP_F_CHECK_ONLY)) { tmpbuf = vmalloc(mop->size); if (!tmpbuf) @@ -2771,6 +2798,14 @@ static int kvm_s390_vm_mem_op(struct kvm *kvm, struct kvm_s390_mem_op *mop) case KVM_S390_MEMOP_ABSOLUTE_WRITE: { if (mop->flags & KVM_S390_MEMOP_F_CHECK_ONLY) { r = check_gpa_range(kvm, mop->gaddr, mop->size, GACC_STORE, mop->key); + } else if (mop->flags & KVM_S390_MEMOP_F_CMPXCHG) { + r = cmpxchg_guest_abs_with_key(kvm, mop->gaddr, mop->size, + &old.quad, new.quad, mop->key); + if (r == 1) { + r = KVM_S390_MEMOP_R_NO_XCHG; + if (copy_to_user(old_addr, &old.raw[off_in_quad], mop->size)) + r = -EFAULT; + } } else { if (copy_from_user(tmpbuf, uaddr, mop->size)) { r = -EFAULT;
User space can use the MEM_OP ioctl to make storage key checked reads and writes to the guest, however, it has no way of performing atomic, key checked, accesses to the guest. Extend the MEM_OP ioctl in order to allow for this, by adding a cmpxchg mode. For now, support this mode for absolute accesses only. This mode can be use, for example, to set the device-state-change indicator and the adapter-local-summary indicator atomically. Signed-off-by: Janis Schoetterl-Glausch <scgl@linux.ibm.com> --- include/uapi/linux/kvm.h | 7 +++ arch/s390/kvm/gaccess.h | 3 ++ arch/s390/kvm/gaccess.c | 102 +++++++++++++++++++++++++++++++++++++++ arch/s390/kvm/kvm-s390.c | 39 ++++++++++++++- 4 files changed, 149 insertions(+), 2 deletions(-)