KVM: arm64: Use visibility hook to treat ID regs as RAZ
v2 cover KVM: arm64: Use visibility hook to treat ID regs as RAZ
For reasons unknown, the Arm architecture defines the 64-bit views of the 32-bit ID registers as UNKNOWN [1]. This combines poorly with the fact that KVM unconditionally exposes these registers to userspace, which could throw a wrench in migration between 64-bit only systems.
This series reworks KVM's definition of these registers to RAZ/WI with the goal of providing consistent register values across 64-bit machines.
Patches 1-3 clean up the ID register accessors, taking advantage of the fact that the generic accessors know how to handle RAZ.
Patches 4-6 start switch the handling of potentially nonzero AArch32 ID registers to RAZ/WI. RAZ covers up the architecturally UNKNOWN values, and WI allows for migration off of kernels that may provide garbage. Note that hidden AArch32 ID registers continue to have RAZ behavior with the additional expectation of invariance.
Lastly, patch 7 includes a small test for the issue.
Applies to 6.0-rc3. Tested with KVM selftests under the fast model w/ asymmetric 32 bit support and no 32 bit support whatsoever.
v1 -> v2:
- Collect Reiji's r-b tags (thanks!)
- Call sysreg_visible_as_raz() from read_id_reg() (Reiji)
- Hoist sysreg_user_write_ignore() into kvm_sys_reg_set_user() (Reiji)
Oliver Upton (7):
KVM: arm64: Use visibility hook to treat ID regs as RAZ
KVM: arm64: Remove internal accessor helpers for id regs
KVM: arm64: Drop raz parameter from read_id_reg()
KVM: arm64: Spin off helper for calling visibility hook
KVM: arm64: Add a visibility bit to ignore user writes
KVM: arm64: Treat 32bit ID registers as RAZ/WI on 64bit-only system
KVM: selftests: Add test for RAZ/WI AArch32 ID registers
arch/arm64/kvm/sys_regs.c | 150 +++++++++--------- arch/arm64/kvm/sys_regs.h | 24 ++- tools/testing/selftests/kvm/.gitignore | 1 + tools/testing/selftests/kvm/Makefile | 1 + .../kvm/aarch64/aarch64_only_id_regs.c | 135 ++++++++++++++++ 5 files changed, 225 insertions(+), 86 deletions(-) create mode 100644 tools/testing/selftests/kvm/aarch64/aarch64_only_id_regs.c
base-commit: b90cb1053190353cc30f0fef0ef1f378ccc063c5
Signature · 6 lines
2.37.2.789.g6183377224-goog _______________________________________________ kvmarm mailing list [email protected] https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
PATCH v2 1/7 KVM: arm64: Use visibility hook to treat ID regs as RAZ
The generic id reg accessors already handle RAZ registers by way of the visibility hook. Add a visibility hook that returns REG_RAZ unconditionally and throw out the RAZ specific accessors.
--- arch/arm64/kvm/sys_regs.c | 33 ++++++++++++++------------------- 1 file changed, 14 insertions(+), 19 deletions(-)
arch/arm64/kvm/sys_regs.c +14 -19
@@ -1145,6 +1145,12 @@ static unsigned int id_visibility(const struct kvm_vcpu *vcpu,return 0 ; } +static unsigned int raz_visibility (const struct kvm_vcpu *vcpu, +const struct sys_reg_desc *r) +{ +return REG_RAZ ; +} +/* cpufeature ID register access trap handlers */ static bool __access_id_reg(struct kvm_vcpu *vcpu,
@@ -1168,13 +1174,6 @@ static bool access_id_reg(struct kvm_vcpu *vcpu,return __access_id_reg (vcpu, p, r, raz); } -static bool access_raz_id_reg (struct kvm_vcpu *vcpu, -struct sys_reg_params *p, -const struct sys_reg_desc *r) -{ -return __access_id_reg (vcpu, p, r, true); -} -/* Visibility overrides for SVE-specific control registers */ static unsigned int sve_visibility (const struct kvm_vcpu *vcpu,const struct sys_reg_desc *rd)
@@ -1262,12 +1261,6 @@ static int set_id_reg(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,return __set_id_reg (vcpu, rd, val, raz); } -static int set_raz_id_reg (struct kvm_vcpu *vcpu,const struct sys_reg_desc *rd, -u64 val) -{ -return __set_id_reg (vcpu, rd, val, true); -} -static int get_raz_reg (struct kvm_vcpu *vcpu,const struct sys_reg_desc *rd,u64 *val) {
@@ -1374,9 +1367,10 @@ static unsigned int mte_visibility(const struct kvm_vcpu *vcpu,*/ #define ID_UNALLOCATED (crm, op2) { \ Op0(3), Op1(0), CRn(0), CRm(crm), Op2(op2), \ - .access = access_raz_id_reg, \ - .get_user = get_raz_reg, \ - .set_user = set_raz_id_reg, \ + .access = access_id_reg, \ + .get_user = get_id_reg, \ + .set_user = set_id_reg, \ + .visibility = raz_visibility \ }/*
@@ -1386,9 +1380,10 @@ static unsigned int mte_visibility(const struct kvm_vcpu *vcpu,*/ #define ID_HIDDEN (name) { \ SYS_DESC(SYS_##name), \ - .access = access_raz_id_reg, \ - .get_user = get_raz_reg, \ - .set_user = set_raz_id_reg, \ + .access = access_id_reg, \ + .get_user = get_id_reg, \ + .set_user = set_id_reg, \ + .visibility = raz_visibility, \ }/*
Signature · 6 lines
2.37.2.789.g6183377224-goog _______________________________________________ kvmarm mailing list [email protected] https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
PATCH v2 2/7 KVM: arm64: Remove internal accessor helpers for id regs
In reply to Oliver UptonThe internal accessors are only ever called once. Dump out their contents in the caller.
No functional change intended.
--- arch/arm64/kvm/sys_regs.c | 46 ++++++++++----------------------------- 1 file changed, 12 insertions(+), 34 deletions(-)
arch/arm64/kvm/sys_regs.c +12 -34
@@ -1153,25 +1153,17 @@ static unsigned int raz_visibility(const struct kvm_vcpu *vcpu,/* cpufeature ID register access trap handlers */ -static bool __access_id_reg (struct kvm_vcpu *vcpu, -struct sys_reg_params *p, -const struct sys_reg_desc *r, -bool raz) -{ -if (p->is_write) -return write_to_read_only (vcpu, p, r); - - p->regval =read_id_reg (vcpu, r, raz); -return true; -} -static bool access_id_reg (struct kvm_vcpu *vcpu,struct sys_reg_params *p,const struct sys_reg_desc *r) {bool raz =sysreg_visible_as_raz (vcpu, r); -return __access_id_reg (vcpu, p, r, raz); +if (p->is_write) +return write_to_read_only (vcpu, p, r); + + p->regval =read_id_reg (vcpu, r, raz); +return true; }/* Visibility overrides for SVE-specific control registers */
@@ -1226,31 +1218,13 @@ static int set_id_aa64pfr0_el1(struct kvm_vcpu *vcpu,* are stored, and for set_id_reg() we don't allow the effective value * to be changed. */ -static int __get_id_reg (const struct kvm_vcpu *vcpu, -const struct sys_reg_desc *rd,u64 *val, -bool raz) -{ - *val =read_id_reg (vcpu, rd, raz); -return 0 ; -} - -static int __set_id_reg (const struct kvm_vcpu *vcpu, -const struct sys_reg_desc *rd,u64 val, -bool raz) -{ -/* This is what we mean by invariant: you can't change it. */ -if (val !=read_id_reg (vcpu, rd, raz)) -return -EINVAL ; - -return 0 ; -} -static int get_id_reg (struct kvm_vcpu *vcpu,const struct sys_reg_desc *rd,u64 *val) {bool raz =sysreg_visible_as_raz (vcpu, rd); -return __get_id_reg (vcpu, rd, val, raz); + *val =read_id_reg (vcpu, rd, raz); +return 0 ; } static int set_id_reg(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
@@ -1258,7 +1232,11 @@ static int set_id_reg(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd, {bool raz =sysreg_visible_as_raz (vcpu, rd); -return __set_id_reg (vcpu, rd, val, raz); +/* This is what we mean by invariant: you can't change it. */ +if (val !=read_id_reg (vcpu, rd, raz)) +return -EINVAL ; + +return 0 ; } static int get_raz_reg(struct kvm_vcpu *vcpu, const struct sys_reg_desc *rd,
Signature · 6 lines
2.37.2.789.g6183377224-goog _______________________________________________ kvmarm mailing list [email protected] https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Re: [PATCH v2 2/7] KVM: arm64: Remove internal accessor helpers for id regs
Quoted 7 lines · Oliver Upton
The internal accessors are only ever called once. Dump out their contents in the caller.
No functional change intended.
Signed-off-by Oliver Upton <[email protected]>
_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
PATCH v2 3/7 KVM: arm64: Drop raz parameter from read_id_reg()
In reply to Oliver UptonThere is no longer a need for caller-specified RAZ visibility. Hoist the call to sysreg_visible_as_raz() into read_id_reg() and drop the parameter.
No functional change intended.
--- arch/arm64/kvm/sys_regs.c | 19 ++++++------------- 1 file changed, 6 insertions(+), 13 deletions(-)
arch/arm64/kvm/sys_regs.c +6 -13
@@ -1063,13 +1063,12 @@ static bool access_arch_timer(struct kvm_vcpu *vcpu, }/* Read a sanitised cpufeature ID register by sys_reg_desc */ -static u64 read_id_reg (const struct kvm_vcpu *vcpu, -struct sys_reg_desc const *r,bool raz) +static u64 read_id_reg (const struct kvm_vcpu *vcpu,struct sys_reg_desc const *r) {u32 id =reg_to_encoding (r);u64 val; -if (raz) +if (sysreg_visible_as_raz (vcpu, r))return 0 ; val =read_sanitised_ftr_reg (id);
@@ -1157,12 +1156,10 @@ static bool access_id_reg(struct kvm_vcpu *vcpu, struct sys_reg_params *p, const struct sys_reg_desc *r) { - bool raz = sysreg_visible_as_raz(vcpu, r); -if (p->is_write)return write_to_read_only (vcpu, p, r); - p->regval =read_id_reg (vcpu, r, raz); + p->regval =read_id_reg (vcpu, r);return true; }
@@ -1199,7 +1196,7 @@ static int set_id_aa64pfr0_el1(struct kvm_vcpu *vcpu,return -EINVAL ;/* We can only differ with CSV[23], and anything else is an error */ - val ^=read_id_reg (vcpu, rd, false); + val ^=read_id_reg (vcpu, rd); val &= ~((0xFUL <<ID_AA64PFR0_CSV2_SHIFT ) | (0xFUL <<ID_AA64PFR0_CSV3_SHIFT ));if (val)
@@ -1221,19 +1218,15 @@ static int set_id_aa64pfr0_el1(struct kvm_vcpu *vcpu,static int get_id_reg (struct kvm_vcpu *vcpu,const struct sys_reg_desc *rd,u64 *val) { -bool raz =sysreg_visible_as_raz (vcpu, rd); - - *val =read_id_reg (vcpu, rd, raz); + *val =read_id_reg (vcpu, rd);return 0 ; }static int set_id_reg (struct kvm_vcpu *vcpu,const struct sys_reg_desc *rd,u64 val) { -bool raz =sysreg_visible_as_raz (vcpu, rd); -/* This is what we mean by invariant: you can't change it. */ -if (val !=read_id_reg (vcpu, rd, raz)) +if (val !=read_id_reg (vcpu, rd))return -EINVAL ;return 0 ;
Signature · 6 lines
2.37.2.789.g6183377224-goog _______________________________________________ kvmarm mailing list [email protected] https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Re: [PATCH v2 3/7] KVM: arm64: Drop raz parameter from read_id_reg()
Quoted 9 lines · Oliver Upton
There is no longer a need for caller-specified RAZ visibility. Hoist the call to
sysreg_visible_as_raz()intoread_id_reg()and drop the parameter.No functional change intended.
Suggested-by Reiji Watanabe <[email protected]>Signed-off-by Oliver Upton <[email protected]>
PATCH v2 4/7 KVM: arm64: Spin off helper for calling visibility hook
In reply to Oliver UptonNo functional change intended.
--- arch/arm64/kvm/sys_regs.h | 19 +++++++++++-------- 1 file changed, 11 insertions(+), 8 deletions(-)
arch/arm64/kvm/sys_regs.h +11 -8
@@ -136,22 +136,25 @@ static inline void reset_val(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r__vcpu_sys_reg (vcpu, r->reg) = r->val; } -static inline bool sysreg_hidden (const struct kvm_vcpu *vcpu, -const struct sys_reg_desc *r) +static inline unsigned int sysreg_visibility (const struct kvm_vcpu *vcpu, +const struct sys_reg_desc *r) {if (likely (!r->visibility)) -return false; +return 0 ; -return r->visibility (vcpu, r) ®_HIDDEN ; +return r->visibility (vcpu, r); +} + +static inline bool sysreg_hidden (const struct kvm_vcpu *vcpu, +const struct sys_reg_desc *r) +{ +return sysreg_visibility (vcpu, r) ®_HIDDEN ; }static inline bool sysreg_visible_as_raz (const struct kvm_vcpu *vcpu,const struct sys_reg_desc *r) { -if (likely (!r->visibility)) -return false; - -return r->visibility (vcpu, r) ®_RAZ ; +return sysreg_visibility (vcpu, r) ®_RAZ ; } static inline int cmp_sys_reg(const struct sys_reg_desc *i1,
Signature · 6 lines
2.37.2.789.g6183377224-goog _______________________________________________ kvmarm mailing list [email protected] https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
PATCH v2 5/7 KVM: arm64: Add a visibility bit to ignore user writes
In reply to Oliver UptonWe're about to ignore writes to AArch32 ID registers on AArch64-only systems. Add a bit to indicate a register is handled as write ignore when accessed from userspace.
--- arch/arm64/kvm/sys_regs.c | 3 +++ arch/arm64/kvm/sys_regs.h | 7 +++++++ 2 files changed, 10 insertions(+)
arch/arm64/kvm/sys_regs.c +3 -0
@@ -2775,6 +2775,9 @@ int kvm_sys_reg_set_user(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg,if (!r)return -ENOENT ; +if (sysreg_user_write_ignore (vcpu, r)) +return 0 ; +if (r->set_user) { ret = (r->set_user)(vcpu, r, val); }else {
arch/arm64/kvm/sys_regs.h +7 -0
@@ -86,6 +86,7 @@ struct sys_reg_desc {#define REG_HIDDEN (1 << 0)/* hidden from userspace and guest */ #define REG_RAZ (1 << 1)/* RAZ from userspace and guest */ +#define REG_USER_WI (1 << 2)/* WI from userspace only */ static __printf(2, 3) inline void print_sys_reg_msg(const struct sys_reg_params *p,
@@ -157,6 +158,12 @@ static inline bool sysreg_visible_as_raz(const struct kvm_vcpu *vcpu,return sysreg_visibility (vcpu, r) ®_RAZ ; } +static inline bool sysreg_user_write_ignore (const struct kvm_vcpu *vcpu, +const struct sys_reg_desc *r) +{ +return sysreg_visibility (vcpu, r) ®_USER_WI ; +} +static inline int cmp_sys_reg (const struct sys_reg_desc *i1,const struct sys_reg_desc *i2) {
Signature · 6 lines
2.37.2.789.g6183377224-goog _______________________________________________ kvmarm mailing list [email protected] https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Re: [PATCH v2 5/7] KVM: arm64: Add a visibility bit to ignore user writes
Quoted 6 lines · Oliver Upton
We're about to ignore writes to AArch32 ID registers on AArch64-only systems. Add a bit to indicate a register is handled as write ignore when accessed from userspace.
Signed-off-by Oliver Upton <[email protected]>
_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
PATCH v2 6/7 KVM: arm64: Treat 32bit ID registers as RAZ/WI on 64bit-only system
In reply to Oliver UptonOne of the oddities of the architecture is that the AArch64 views of the AArch32 ID registers are UNKNOWN if AArch32 isn't implemented at any EL. Nonetheless, KVM exposes these registers to userspace for the sake of save/restore. It is possible that the UNKNOWN value could differ between systems, leading to a rejected write from userspace.
Avoid the issue altogether by handling the AArch32 ID registers as RAZ/WI when on an AArch64-only system.
--- arch/arm64/kvm/sys_regs.c | 63 ++++++++++++++++++++++++++------------- 1 file changed, 43 insertions(+), 20 deletions(-)
arch/arm64/kvm/sys_regs.c +43 -20
@@ -1144,6 +1144,20 @@ static unsigned int id_visibility(const struct kvm_vcpu *vcpu,return 0 ; } +static unsigned int aa32_id_visibility (const struct kvm_vcpu *vcpu, +const struct sys_reg_desc *r) +{ +/* +* AArch32 ID registers are UNKNOWN if AArch32 isn't implemented at any +* EL. Promote to RAZ/WI in order to guarantee consistency between +* systems. +*/ +if (!kvm_supports_32bit_el0 ()) +return REG_RAZ |REG_USER_WI ; + +return id_visibility (vcpu, r); +} +static unsigned int raz_visibility (const struct kvm_vcpu *vcpu,const struct sys_reg_desc *r) {
@@ -1331,6 +1345,15 @@ static unsigned int mte_visibility(const struct kvm_vcpu *vcpu, .visibility = id_visibility, \ } +/* sys_reg_desc initialiser for known cpufeature ID registers */ +#define AA32_ID_SANITISED (name) { \ + SYS_DESC(SYS_##name), \ + .access = access_id_reg, \ + .get_user = get_id_reg, \ + .set_user = set_id_reg, \ + .visibility = aa32_id_visibility, \ +} +/* * sys_reg_desc initialiser for architecturally unallocated cpufeature ID * register with encoding Op0=3, Op1=0, CRn=0, CRm=crm, Op2=op2
@@ -1418,33 +1441,33 @@ static const struct sys_reg_desc sys_reg_descs[] = {/* AArch64 mappings of the AArch32 ID registers */ /* CRm=1 */ -ID_SANITISED (ID_PFR0_EL1 ), -ID_SANITISED (ID_PFR1_EL1 ), -ID_SANITISED (ID_DFR0_EL1 ), +AA32_ID_SANITISED (ID_PFR0_EL1 ), +AA32_ID_SANITISED (ID_PFR1_EL1 ), +AA32_ID_SANITISED (ID_DFR0_EL1 ),ID_HIDDEN (ID_AFR0_EL1 ), -ID_SANITISED (ID_MMFR0_EL1 ), -ID_SANITISED (ID_MMFR1_EL1 ), -ID_SANITISED (ID_MMFR2_EL1 ), -ID_SANITISED (ID_MMFR3_EL1 ), +AA32_ID_SANITISED (ID_MMFR0_EL1 ), +AA32_ID_SANITISED (ID_MMFR1_EL1 ), +AA32_ID_SANITISED (ID_MMFR2_EL1 ), +AA32_ID_SANITISED (ID_MMFR3_EL1 ),/* CRm=2 */ -ID_SANITISED (ID_ISAR0_EL1 ), -ID_SANITISED (ID_ISAR1_EL1 ), -ID_SANITISED (ID_ISAR2_EL1 ), -ID_SANITISED (ID_ISAR3_EL1 ), -ID_SANITISED (ID_ISAR4_EL1 ), -ID_SANITISED (ID_ISAR5_EL1 ), -ID_SANITISED (ID_MMFR4_EL1 ), -ID_SANITISED (ID_ISAR6_EL1 ), +AA32_ID_SANITISED (ID_ISAR0_EL1 ), +AA32_ID_SANITISED (ID_ISAR1_EL1 ), +AA32_ID_SANITISED (ID_ISAR2_EL1 ), +AA32_ID_SANITISED (ID_ISAR3_EL1 ), +AA32_ID_SANITISED (ID_ISAR4_EL1 ), +AA32_ID_SANITISED (ID_ISAR5_EL1 ), +AA32_ID_SANITISED (ID_MMFR4_EL1 ), +AA32_ID_SANITISED (ID_ISAR6_EL1 ),/* CRm=3 */ -ID_SANITISED (MVFR0_EL1 ), -ID_SANITISED (MVFR1_EL1 ), -ID_SANITISED (MVFR2_EL1 ), +AA32_ID_SANITISED (MVFR0_EL1 ), +AA32_ID_SANITISED (MVFR1_EL1 ), +AA32_ID_SANITISED (MVFR2_EL1 ),ID_UNALLOCATED (3 ,3 ), -ID_SANITISED (ID_PFR2_EL1 ), +AA32_ID_SANITISED (ID_PFR2_EL1 ),ID_HIDDEN (ID_DFR1_EL1 ), -ID_SANITISED (ID_MMFR5_EL1 ), +AA32_ID_SANITISED (ID_MMFR5_EL1 ),ID_UNALLOCATED (3 ,7 ), /* AArch64 ID registers */
Signature · 6 lines
2.37.2.789.g6183377224-goog _______________________________________________ kvmarm mailing list [email protected] https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Re: [PATCH v2 6/7] KVM: arm64: Treat 32bit ID registers as RAZ/WI on 64bit-only system
Hi Oliver,
Quoted 95 lines · Oliver Upton
One of the oddities of the architecture is that the AArch64 views of the AArch32 ID registers are UNKNOWN if AArch32 isn't implemented at any EL. Nonetheless, KVM exposes these registers to userspace for the sake of save/restore. It is possible that the UNKNOWN value could differ between systems, leading to a rejected write from userspace.
Avoid the issue altogether by handling the AArch32 ID registers as RAZ/WI when on an AArch64-only system.
Signed-off-by Oliver Upton <[email protected]>--- arch/arm64/kvm/sys_regs.c | 63 ++++++++++++++++++++++++++------------- 1 file changed, 43 insertions(+), 20 deletions(-)arch/arm64/kvm/sys_regs.c +38
-15@@ -1144,6 +1144,20 @@ static unsigned int id_visibility(const struct kvm_vcpu *vcpu,return 0 ; } +static unsigned int aa32_id_visibility (const struct kvm_vcpu *vcpu, +const struct sys_reg_desc *r) +{ +/* +* AArch32 ID registers are UNKNOWN if AArch32 isn't implemented at any +* EL. Promote to RAZ/WI in order to guarantee consistency between +* systems. +*/ +if (!kvm_supports_32bit_el0 ()) +return REG_RAZ |REG_USER_WI ; + +return id_visibility (vcpu, r); +} +static unsigned int raz_visibility (const struct kvm_vcpu *vcpu,const struct sys_reg_desc *r) {@@ -1331,6 +1345,15 @@ static unsigned int mte_visibility(const struct kvm_vcpu *vcpu, .visibility = id_visibility, \ } +/* sys_reg_desc initialiser for known cpufeature ID registers */ +#define AA32_ID_SANITISED (name) { \ + SYS_DESC(SYS_##name), \ + .access = access_id_reg, \ + .get_user = get_id_reg, \ + .set_user = set_id_reg, \ + .visibility = aa32_id_visibility, \ +} +/* * sys_reg_desc initialiser for architecturally unallocated cpufeature ID * register with encoding Op0=3, Op1=0, CRn=0, CRm=crm, Op2=op2 @@ -1418,33 +1441,33 @@ static const struct sys_reg_desc sys_reg_descs[] = {/* AArch64 mappings of the AArch32 ID registers */ /* CRm=1 */ -ID_SANITISED (ID_PFR0_EL1 ), -ID_SANITISED (ID_PFR1_EL1 ), -ID_SANITISED (ID_DFR0_EL1 ), +AA32_ID_SANITISED (ID_PFR0_EL1 ), +AA32_ID_SANITISED (ID_PFR1_EL1 ), +AA32_ID_SANITISED (ID_DFR0_EL1 ),ID_HIDDEN (ID_AFR0_EL1 ), -ID_SANITISED (ID_MMFR0_EL1 ), -ID_SANITISED (ID_MMFR1_EL1 ), -ID_SANITISED (ID_MMFR2_EL1 ), -ID_SANITISED (ID_MMFR3_EL1 ), +AA32_ID_SANITISED (ID_MMFR0_EL1 ), +AA32_ID_SANITISED (ID_MMFR1_EL1 ), +AA32_ID_SANITISED (ID_MMFR2_EL1 ), +AA32_ID_SANITISED (ID_MMFR3_EL1 ),/* CRm=2 */ -ID_SANITISED (ID_ISAR0_EL1 ), -ID_SANITISED (ID_ISAR1_EL1 ), -ID_SANITISED (ID_ISAR2_EL1 ), -ID_SANITISED (ID_ISAR3_EL1 ), -ID_SANITISED (ID_ISAR4_EL1 ), -ID_SANITISED (ID_ISAR5_EL1 ), -ID_SANITISED (ID_MMFR4_EL1 ), -ID_SANITISED (ID_ISAR6_EL1 ), +AA32_ID_SANITISED (ID_ISAR0_EL1 ), +AA32_ID_SANITISED (ID_ISAR1_EL1 ), +AA32_ID_SANITISED (ID_ISAR2_EL1 ), +AA32_ID_SANITISED (ID_ISAR3_EL1 ), +AA32_ID_SANITISED (ID_ISAR4_EL1 ), +AA32_ID_SANITISED (ID_ISAR5_EL1 ), +AA32_ID_SANITISED (ID_MMFR4_EL1 ), +AA32_ID_SANITISED (ID_ISAR6_EL1 ), /* CRm=3 */
arch/arm64/kvm/sys_regs.c static const struct sys_reg_desc sys_reg_descs[] = { · line 1469
Quoted 10 lines
arch/arm64/kvm/sys_regs.c +4
-4-ID_SANITISED (MVFR0_EL1 ), -ID_SANITISED (MVFR1_EL1 ), -ID_SANITISED (MVFR2_EL1 ), +AA32_ID_SANITISED (MVFR0_EL1 ), +AA32_ID_SANITISED (MVFR1_EL1 ), +AA32_ID_SANITISED (MVFR2_EL1 ),ID_UNALLOCATED (3 ,3 ), -ID_SANITISED (ID_PFR2_EL1 ), +AA32_ID_SANITISED (ID_PFR2_EL1 ), ID_HIDDEN(ID_DFR1_EL1),
Perhaps it might be better to handle ID_AFR0_EL1 and ID_DFR1_EL1 in the same way as the other AArch32 ID registers for consistency ? (i.e. treat them RAZ/USER_WI instead of RAZ if kvm_supports_32bit_el0()
is false instead of RAZ)
Thank you,
Reiji
Quoted 8 lines
arch/arm64/kvm/sys_regs.c +1
-2-ID_SANITISED (ID_MMFR5_EL1 ), +AA32_ID_SANITISED (ID_MMFR5_EL1 ), ID_UNALLOCATED(3,7),/* AArch64 ID registers */ --2.37.2.789.g6183377224-goog
_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Re: [PATCH v2 6/7] KVM: arm64: Treat 32bit ID registers as RAZ/WI on 64bit-only system
Hi Reiji,
On Tue, Sep 06, 2022 at 09:52:53PM -0700, Reiji Watanabe wrote:
[...]
Quoted 12 lines
arch/arm64/kvm/sys_regs.c static const struct sys_reg_desc sys_reg_descs[] = { · line 1469
Quoted 11 lines
/* CRm=3 */arch/arm64/kvm/sys_regs.c +4
-4-ID_SANITISED (MVFR0_EL1 ), -ID_SANITISED (MVFR1_EL1 ), -ID_SANITISED (MVFR2_EL1 ), +AA32_ID_SANITISED (MVFR0_EL1 ), +AA32_ID_SANITISED (MVFR1_EL1 ), +AA32_ID_SANITISED (MVFR2_EL1 ),ID_UNALLOCATED (3 ,3 ), -ID_SANITISED (ID_PFR2_EL1 ), +AA32_ID_SANITISED (ID_PFR2_EL1 ), ID_HIDDEN(ID_DFR1_EL1),
Quoted 4 lines
Perhaps it might be better to handle
ID_AFR0_EL1andID_DFR1_EL1in the same way as the other AArch32 ID registers for consistency ? (i.e. treat them RAZ/USER_WI instead of RAZ ifkvm_supports_32bit_el0()
is false instead of RAZ)
Thanks for having a look. I stopped short of treating these registers as RAZ/USER_WI since an attempted nonzero write to either of these registers is a userspace bug (KVM always advertised 0).
As the ABI isn't busted for these registers I'd prefer to leave it in place. Having said that, I'm not too strongly motivated in either direction.
--
Thanks,
Oliver
PATCH v2 7/7 KVM: selftests: Add test for RAZ/WI AArch32 ID registers
In reply to Oliver UptonAdd a test to assert that KVM handles the AArch64 views of the AArch32 ID registers as RAZ/WI (writable only from userspace).
--- tools/testing/selftests/kvm/.gitignore | 1 + tools/testing/selftests/kvm/Makefile | 1 + .../kvm/aarch64/aarch64_only_id_regs.c | 135 ++++++++++++++++++ 3 files changed, 137 insertions(+) create mode 100644 tools/testing/selftests/kvm/aarch64/aarch64_only_id_regs.c
tools/testing/selftests/kvm/.gitignore +1 -0
@@ -1,4 +1,5 @@ # SPDX-License-Identifier: GPL-2.0-only +/aarch64/aarch64_only_id_regs /aarch64/arch_timer /aarch64/debug-exceptions /aarch64/get-reg-list
tools/testing/selftests/kvm/Makefile +1 -0
@@ -144,6 +144,7 @@ TEST_GEN_PROGS_x86_64 += system_counter_offset_test # Compiled outputs used by test targets TEST_GEN_PROGS_EXTENDED_x86_64 += x86_64/nx_huge_pages_test +TEST_GEN_PROGS_aarch64 += aarch64/aarch64_only_id_regs TEST_GEN_PROGS_aarch64 += aarch64/arch_timer TEST_GEN_PROGS_aarch64 += aarch64/debug-exceptions TEST_GEN_PROGS_aarch64 += aarch64/get-reg-list
tools/testing/selftests/kvm/aarch64/aarch64_only_id_regs.c new file +135 -0
135 lines not on this page · raw message
Signature · 6 lines
2.37.2.789.g6183377224-goog _______________________________________________ kvmarm mailing list [email protected] https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Re: [PATCH v2 7/7] KVM: selftests: Add test for RAZ/WI AArch32 ID registers
Quoted 113 lines · Oliver Upton
Add a test to assert that KVM handles the AArch64 views of the AArch32 ID registers as RAZ/WI (writable only from userspace).
Signed-off-by Oliver Upton <[email protected]>--- tools/testing/selftests/kvm/.gitignore | 1 + tools/testing/selftests/kvm/Makefile | 1 + .../kvm/aarch64/aarch64_only_id_regs.c | 135 ++++++++++++++++++ 3 files changed, 137 insertions(+) create mode 100644 tools/testing/selftests/kvm/aarch64/aarch64_only_id_regs.ctools/testing/selftests/kvm/.gitignore +1
-0@@ -1,4 +1,5 @@ # SPDX-License-Identifier: GPL-2.0-only +/aarch64/aarch64_only_id_regs /aarch64/arch_timer /aarch64/debug-exceptions /aarch64/get-reg-listtools/testing/selftests/kvm/Makefile +1
-0@@ -144,6 +144,7 @@ TEST_GEN_PROGS_x86_64 += system_counter_offset_test # Compiled outputs used by test targets TEST_GEN_PROGS_EXTENDED_x86_64 += x86_64/nx_huge_pages_test +TEST_GEN_PROGS_aarch64 += aarch64/aarch64_only_id_regs TEST_GEN_PROGS_aarch64 += aarch64/arch_timer TEST_GEN_PROGS_aarch64 += aarch64/debug-exceptions TEST_GEN_PROGS_aarch64 += aarch64/get-reg-listtools/testing/selftests/kvm/aarch64/aarch64_only_id_regs.c new file +74
-0@@ -0,0 +1,135 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* +* aarch64_only_id_regs - Test for ID register behavior on AArch64-only systems +* +* Copyright (c) 2022 Google LLC. +* +* Test that KVM handles the AArch64 views of the AArch32 ID registers as RAZ +* and WI from userspace. +*/ + +#include <stdint.h> + +#include "kvm_util.h" +#include "processor.h" +#include "test_util.h" + +#define BAD_ID_REG_VAL 0x1badc0deul + +#define GUEST_ASSERT_REG_RAZ (reg) GUEST_ASSERT_EQ(read_sysreg_s(reg), 0) + +static void guest_main (void ) +{ +GUEST_ASSERT_REG_RAZ (SYS_ID_PFR0_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_PFR1_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_DFR0_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_MMFR0_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_MMFR1_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_MMFR2_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_MMFR3_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_ISAR0_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_ISAR1_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_ISAR2_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_ISAR3_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_ISAR4_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_ISAR5_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_MMFR4_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_ISAR6_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_MVFR0_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_MVFR1_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_MVFR2_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_PFR2_EL1 ); +GUEST_ASSERT_REG_RAZ (SYS_ID_MMFR5_EL1 ); + +GUEST_DONE (); +} + +static void test_guest_raz (struct kvm_vcpu *vcpu) +{ +struct ucall uc; + +vcpu_run (vcpu); + +switch (get_ucall (vcpu, &uc)) { +case UCALL_ABORT : +REPORT_GUEST_ASSERT (uc); +break ; +case UCALL_DONE : +break ; +default : +TEST_FAIL ("Unexpected ucall: %lu" , uc.cmd); + } +} + +static uint64_t reg_ids[] = { +KVM_ARM64_SYS_REG (SYS_ID_PFR0_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_PFR1_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_DFR0_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_MMFR0_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_MMFR1_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_MMFR2_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_MMFR3_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_ISAR0_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_ISAR1_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_ISAR2_EL1 ),
tools/testing/selftests/kvm/aarch64/aarch64_only_id_regs.c line 84
Quoted 10 lines
tools/testing/selftests/kvm/aarch64/aarch64_only_id_regs.c new file +10
-0+KVM_ARM64_SYS_REG (SYS_ID_ISAR3_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_ISAR4_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_ISAR5_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_MMFR4_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_ISAR6_EL1 ), +KVM_ARM64_SYS_REG (SYS_MVFR0_EL1 ), +KVM_ARM64_SYS_REG (SYS_MVFR1_EL1 ), +KVM_ARM64_SYS_REG (SYS_MVFR2_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_PFR2_EL1 ), + KVM_ARM64_SYS_REG(SYS_ID_MMFR5_EL1),
Hi Oliver,
I see all the hidden and unallocated registers have been filtered out of the test lists. They should also behave as RAZ, right? Maybe we should keep them in the lists here for consistency and to test them as well.
Thanks,
drew
Quoted 54 lines
tools/testing/selftests/kvm/aarch64/aarch64_only_id_regs.c new file +51
-0+}; + +static void test_user_raz_wi (struct kvm_vcpu *vcpu) +{ +int i; + +for (i =0 ; i <ARRAY_SIZE (reg_ids); i++) { +uint64_t reg_id = reg_ids[i]; +uint64_t val; + +vcpu_get_reg (vcpu, reg_id, &val); +ASSERT_EQ (val,0 ); + +/* +* Expect the ioctl to succeed with no effect on the register +* value. +*/ +vcpu_set_reg (vcpu, reg_id,BAD_ID_REG_VAL ); + +vcpu_get_reg (vcpu, reg_id, &val); +ASSERT_EQ (val,0 ); + } +} + +static bool vcpu_aarch64_only (struct kvm_vcpu *vcpu) +{ +uint64_t val, el0; + +vcpu_get_reg (vcpu,KVM_ARM64_SYS_REG (SYS_ID_AA64PFR0_EL1 ), &val); + + el0 = (val &ARM64_FEATURE_MASK (ID_AA64PFR0_EL0 )) >>ID_AA64PFR0_EL0_SHIFT ; +return el0 == ID_AA64PFR0_ELx_64BIT_ONLY; +} + +int main (void ) +{ +struct kvm_vcpu *vcpu; +struct kvm_vm *vm; + + vm =vm_create_with_one_vcpu (&vcpu, guest_main); + +TEST_REQUIRE (vcpu_aarch64_only (vcpu)); + +ucall_init (vm,NULL ); + +test_user_raz_wi (vcpu); +test_guest_raz (vcpu); + +ucall_uninit (vm); +kvm_vm_free (vm); +}Signature · 1 line
2.37.2.789.g6183377224-goog
_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Re: [PATCH v2 7/7] KVM: selftests: Add test for RAZ/WI AArch32 ID registers
Hi Drew,
On Mon, Sep 05, 2022 at 09:31:40AM +0200, Andrew Jones wrote:
[...]
Quoted 24 lines
Quoted 11 lines
tools/testing/selftests/kvm/aarch64/aarch64_only_id_regs.c new file +11
-0+static uint64_t reg_ids[] = { +KVM_ARM64_SYS_REG (SYS_ID_PFR0_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_PFR1_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_DFR0_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_MMFR0_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_MMFR1_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_MMFR2_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_MMFR3_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_ISAR0_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_ISAR1_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_ISAR2_EL1 ),tools/testing/selftests/kvm/aarch64/aarch64_only_id_regs.c line 84
Quoted 10 lines
tools/testing/selftests/kvm/aarch64/aarch64_only_id_regs.c new file +10
-0+KVM_ARM64_SYS_REG (SYS_ID_ISAR3_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_ISAR4_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_ISAR5_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_MMFR4_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_ISAR6_EL1 ), +KVM_ARM64_SYS_REG (SYS_MVFR0_EL1 ), +KVM_ARM64_SYS_REG (SYS_MVFR1_EL1 ), +KVM_ARM64_SYS_REG (SYS_MVFR2_EL1 ), +KVM_ARM64_SYS_REG (SYS_ID_PFR2_EL1 ), + KVM_ARM64_SYS_REG(SYS_ID_MMFR5_EL1),Hi Oliver,
Quoted 3 lines
I see all the hidden and unallocated registers have been filtered out of the test lists. They should also behave as RAZ, right? Maybe we should keep them in the lists here for consistency and to test them as well.
Sure, can do. The reason I only tested these registers is because they have RAZ/WI behavior with this series, whereas the rest are RAZ + invariant. Should be easy enough to cover the whole range, though.
--
Thanks,
Oliver
_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Lol, mess up my own copypasta:
Quoted 2 lines · Oliver Upton
For reasons unknown, the Arm architecture defines the 64-bit views of the 32-bit ID registers as UNKNOWN [1]. This combines poorly with the
^ on AArch64-only systems.
Quoted 15 lines
fact that KVM unconditionally exposes these registers to userspace, which could throw a wrench in migration between 64-bit only systems.
This series reworks KVM's definition of these registers to RAZ/WI with the goal of providing consistent register values across 64-bit machines.
Patches 1-3 clean up the ID register accessors, taking advantage of the fact that the generic accessors know how to handle RAZ.
Patches 4-6 start switch the handling of potentially nonzero AArch32 ID registers to RAZ/WI. RAZ covers up the architecturally UNKNOWN values, and WI allows for migration off of kernels that may provide garbage. Note that hidden AArch32 ID registers continue to have RAZ behavior with the additional expectation of invariance.
Quoted 4 lines
Lastly, patch 7 includes a small test for the issue.
Applies to 6.0-rc3. Tested with KVM selftests under the fast model w/ asymmetric 32 bit support and no 32 bit support whatsoever.
[1]: DDI0487H.a Table D12-2 'Instruction encodings for non-Debug System Register accesses'
v1: https://lore.kernel.org/kvmarm/[email protected]/
--
Thanks,
Oliver
Quoted 23 lines
v1 -> v2:
- Collect Reiji's r-b tags (thanks!)
- Callsysreg_visible_as_raz()fromread_id_reg()(Reiji)
- Hoistsysreg_user_write_ignore()intokvm_sys_reg_set_user()(Reiji)Oliver Upton (7):
KVM: arm64: Use visibility hook to treat ID regs as RAZ
KVM: arm64: Remove internal accessor helpers for id regs
KVM: arm64: Drop raz parameter fromread_id_reg()
KVM: arm64: Spin off helper for calling visibility hook
KVM: arm64: Add a visibility bit to ignore user writes
KVM: arm64: Treat 32bit ID registers as RAZ/WI on 64bit-only system
KVM: selftests: Add test for RAZ/WI AArch32 ID registersarch/arm64/kvm/sys_regs.c | 150 +++++++++--------- arch/arm64/kvm/sys_regs.h | 24 ++- tools/testing/selftests/kvm/.gitignore | 1 + tools/testing/selftests/kvm/Makefile | 1 + .../kvm/aarch64/aarch64_only_id_regs.c | 135 ++++++++++++++++ 5 files changed, 225 insertions(+), 86 deletions(-) create mode 100644 tools/testing/selftests/kvm/aarch64/aarch64_only_id_regs.c
Quoted 4 lines
base-commit:
b90cb1053190353cc30f0fef0ef1f378ccc063c5Signature · 1 line
2.37.2.789.g6183377224-goog
_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm