KVM: arm64: Preserve PMCR immutable values across reset
PATCH v3 1/1 KVM: arm64: Preserve PMCR immutable values across reset
We allow user space to set the PMCR register to any value. However, when time comes for a vcpu reset (for example on PSCI online), PMCR is reset to the hardware capabilities.
I would like to explicitly expose different PMU capabilities (number of supported event counters) to the guest than hardware supports. Ideally across vcpu resets.
So this patch adopts the reset path to only populate the immutable PMCR register bits from hardware when they were not initialized previously. This effectively means that on a normal reset, only the guest settable fields are reset, while on vcpu creation the register gets populated from hardware like before.
With this in place and a change in user space to invoke SET_ONE_REG on the PMCR for every vcpu, I can reliably set the PMU event counter number to arbitrary values.
--- arch/arm64/kvm/sys_regs.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)
arch/arm64/kvm/sys_regs.c +8 -1
@@ -663,7 +663,14 @@ static void reset_pmcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r) {u64 pmcr, val; - pmcr =read_sysreg (pmcr_el0); +/* +* If we already received PMCR from a previous ONE_REG call, +* maintain its immutable flags +*/ + pmcr =__vcpu_sys_reg (vcpu, r->reg); +if (!__vcpu_sys_reg (vcpu, r->reg)) + pmcr =read_sysreg (pmcr_el0); +/* * Writable bits of PMCR_EL0 (ARMV8_PMU_PMCR_MASK) are reset to UNKNOWN * except PMCR.E resetting to zero.
Signature · 19 lines
2.16.4 Amazon Development Center Germany GmbH Krausenstr. 38 10117 Berlin Geschaeftsfuehrung: Christian Schlaeger, Jonathan Weiss Eingetragen am Amtsgericht Charlottenburg unter HRB 149173 B Sitz: Berlin Ust-ID: DE 289 237 879 _______________________________________________ kvmarm mailing list [email protected] https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Quoted 46 lines · Alexander Graf
We allow user space to set the PMCR register to any value. However, when time comes for a vcpu reset (for example on PSCI online), PMCR is reset to the hardware capabilities.
I would like to explicitly expose different PMU capabilities (number of supported event counters) to the guest than hardware supports. Ideally across vcpu resets.
So this patch adopts the reset path to only populate the immutable PMCR register bits from hardware when they were not initialized previously. This effectively means that on a normal reset, only the guest settable fields are reset, while on vcpu creation the register gets populated from hardware like before.
With this in place and a change in user space to invoke
SET_ONE_REGon the PMCR for every vcpu, I can reliably set the PMU event counter number to arbitrary values.Signed-off-by Alexander Graf <[email protected]>--- arch/arm64/kvm/sys_regs.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)arch/arm64/kvm/sys_regs.c +8
-1@@ -663,7 +663,14 @@ static void reset_pmcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r) {u64 pmcr, val; - pmcr =read_sysreg (pmcr_el0); +/* +* If we already received PMCR from a previous ONE_REG call, +* maintain its immutable flags +*/ + pmcr =__vcpu_sys_reg (vcpu, r->reg); +if (!__vcpu_sys_reg (vcpu, r->reg)) + pmcr =read_sysreg (pmcr_el0); +/* * Writable bits of PMCR_EL0 (ARMV8_PMU_PMCR_MASK) are reset to UNKNOWN * except PMCR.E resetting to zero. Signature · 1 line
2.16.4
Aha, a much simpler patch than I expected. With this approach we don't need a get_user() function, or to use 'val', but don't we still want to add sanity checks with a set_user() function? At least to ensure immutable flags match and that PMCR_EL0.N isn't too big?
Silently changing the user's input, which I see we also do for e.g. MPIDR, isn't super user friendly.
Thanks,
drew
_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Quoted 45 lines · Andrew Jones
Quoted 43 lines · Alexander Graf
We allow user space to set the PMCR register to any value. However, when time comes for a vcpu reset (for example on PSCI online), PMCR is reset to the hardware capabilities.
I would like to explicitly expose different PMU capabilities (number of supported event counters) to the guest than hardware supports. Ideally across vcpu resets.
So this patch adopts the reset path to only populate the immutable PMCR register bits from hardware when they were not initialized previously. This effectively means that on a normal reset, only the guest settable fields are reset, while on vcpu creation the register gets populated from hardware like before.
With this in place and a change in user space to invoke
SET_ONE_REGon the PMCR for every vcpu, I can reliably set the PMU event counter number to arbitrary values.Signed-off-by Alexander Graf <[email protected]>--- arch/arm64/kvm/sys_regs.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)arch/arm64/kvm/sys_regs.c +8
-1@@ -663,7 +663,14 @@ static void reset_pmcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r) {u64 pmcr, val; - pmcr =read_sysreg (pmcr_el0); +/* +* If we already received PMCR from a previous ONE_REG call, +* maintain its immutable flags +*/ + pmcr =__vcpu_sys_reg (vcpu, r->reg); +if (!__vcpu_sys_reg (vcpu, r->reg)) + pmcr =read_sysreg (pmcr_el0); +/* * Writable bits of PMCR_EL0 (ARMV8_PMU_PMCR_MASK) are reset to UNKNOWN * except PMCR.E resetting to zero.
Quoted 8 lines
Quoted 3 lines
--
2.16.4Aha, a much simpler patch than I expected. With this approach we don't need a
get_user()function, or to use 'val', but don't we still want to add sanity checks with aset_user()function? At least to ensure immutable flags match and that PMCR_EL0.N isn't too big?
We don't check for any flags today, so in a way adding checks would be ABI breakage.
And as Marc pointed out, all of the counters are basically virtual through perf. So if you report 31 counters, you end up spawning 31 perf counters which get multiplexed, so it would work (albeit not be terribly accurate).
That leaves identification bits as something we can check for. But do we really have to? What's the worst thing that can happen? KVM user space can shoot themselves in the foot. Well, they can also set PC to an invalid value. If you do bad things you get bad results :). As long as it's not a security risk, I'm not sure the benefits of checking outweigh the risks.
Quoted 2 lines
Silently changing the user's input, which I see we also do for e.g. MPIDR, isn't super user friendly.
Yes :).
Alex
Amazon Development Center Germany GmbH
Krausenstr. 38
10117 Berlin
Geschaeftsfuehrung: Christian Schlaeger, Jonathan Weiss Eingetragen am Amtsgericht Charlottenburg unter HRB 149173 B
Sitz: Berlin
Ust-ID: DE 289 237 879
_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Quoted 64 lines · Alexander Graf
Quoted 45 lines · Andrew Jones
Quoted 43 lines · Alexander Graf
We allow user space to set the PMCR register to any value. However, when time comes for a vcpu reset (for example on PSCI online), PMCR is reset to the hardware capabilities.
I would like to explicitly expose different PMU capabilities (number of supported event counters) to the guest than hardware supports. Ideally across vcpu resets.
So this patch adopts the reset path to only populate the immutable PMCR register bits from hardware when they were not initialized previously. This effectively means that on a normal reset, only the guest settable fields are reset, while on vcpu creation the register gets populated from hardware like before.
With this in place and a change in user space to invoke
SET_ONE_REGon the PMCR for every vcpu, I can reliably set the PMU event counter number to arbitrary values.Signed-off-by Alexander Graf <[email protected]>--- arch/arm64/kvm/sys_regs.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)arch/arm64/kvm/sys_regs.c +8
-1@@ -663,7 +663,14 @@ static void reset_pmcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r) {u64 pmcr, val; - pmcr =read_sysreg (pmcr_el0); +/* +* If we already received PMCR from a previous ONE_REG call, +* maintain its immutable flags +*/ + pmcr =__vcpu_sys_reg (vcpu, r->reg); +if (!__vcpu_sys_reg (vcpu, r->reg)) + pmcr =read_sysreg (pmcr_el0); +/* * Writable bits of PMCR_EL0 (ARMV8_PMU_PMCR_MASK) are reset to UNKNOWN * except PMCR.E resetting to zero. Quoted 8 lines
Quoted 3 lines
--
2.16.4Aha, a much simpler patch than I expected. With this approach we don't need a
get_user()function, or to use 'val', but don't we still want to add sanity checks with aset_user()function? At least to ensure immutable flags match and that PMCR_EL0.N isn't too big?We don't check for any flags today, so in a way adding checks would be ABI breakage.
And as Marc pointed out, all of the counters are basically virtual through perf. So if you report 31 counters, you end up spawning 31 perf counters which get multiplexed, so it would work (albeit not be terribly accurate).
Quoted 5 lines
That leaves identification bits as something we can check for. But do we really have to? What's the worst thing that can happen? KVM user space can shoot themselves in the foot. Well, they can also set PC to an invalid value. If you do bad things you get bad results :). As long as it's not a security risk, I'm not sure the benefits of checking outweigh the risks.
That's a reasonable justification.
Thanks,
drew
Quoted 21 lines
Quoted 3 lines
Silently changing the user's input, which I see we also do for e.g. MPIDR, isn't super user friendly.
Yes :).
Alex
Amazon Development Center Germany GmbH
Krausenstr. 38
10117 Berlin
Geschaeftsfuehrung: Christian Schlaeger, Jonathan Weiss Eingetragen am Amtsgericht Charlottenburg unter HRB 149173 B
Sitz: Berlin
Ust-ID: DE 289 237 879
Quoted 46 lines · Alexander Graf
We allow user space to set the PMCR register to any value. However, when time comes for a vcpu reset (for example on PSCI online), PMCR is reset to the hardware capabilities.
I would like to explicitly expose different PMU capabilities (number of supported event counters) to the guest than hardware supports. Ideally across vcpu resets.
So this patch adopts the reset path to only populate the immutable PMCR register bits from hardware when they were not initialized previously. This effectively means that on a normal reset, only the guest settable fields are reset, while on vcpu creation the register gets populated from hardware like before.
With this in place and a change in user space to invoke
SET_ONE_REGon the PMCR for every vcpu, I can reliably set the PMU event counter number to arbitrary values.Signed-off-by Alexander Graf <[email protected]>--- arch/arm64/kvm/sys_regs.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)arch/arm64/kvm/sys_regs.c +8
-1@@ -663,7 +663,14 @@ static void reset_pmcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r) {u64 pmcr, val; - pmcr =read_sysreg (pmcr_el0); +/* +* If we already received PMCR from a previous ONE_REG call, +* maintain its immutable flags +*/ + pmcr =__vcpu_sys_reg (vcpu, r->reg); +if (!__vcpu_sys_reg (vcpu, r->reg)) + pmcr =read_sysreg (pmcr_el0); +/* * Writable bits of PMCR_EL0 (ARMV8_PMU_PMCR_MASK) are reset to UNKNOWN * except PMCR.E resetting to zero. Signature · 1 line
2.16.4
_______________________________________________
kvmarm mailing list
[email protected]
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
Marc Zyngier
maintainerQuoted 43 lines · Alexander Graf
We allow user space to set the PMCR register to any value. However, when time comes for a vcpu reset (for example on PSCI online), PMCR is reset to the hardware capabilities.
I would like to explicitly expose different PMU capabilities (number of supported event counters) to the guest than hardware supports. Ideally across vcpu resets.
So this patch adopts the reset path to only populate the immutable PMCR register bits from hardware when they were not initialized previously. This effectively means that on a normal reset, only the guest settable fields are reset, while on vcpu creation the register gets populated from hardware like before.
With this in place and a change in user space to invoke
SET_ONE_REGon the PMCR for every vcpu, I can reliably set the PMU event counter number to arbitrary values.Signed-off-by Alexander Graf <[email protected]>--- arch/arm64/kvm/sys_regs.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)arch/arm64/kvm/sys_regs.c +8
-1@@ -663,7 +663,14 @@ static void reset_pmcr(struct kvm_vcpu *vcpu, const struct sys_reg_desc *r) {u64 pmcr, val; - pmcr =read_sysreg (pmcr_el0); +/* +* If we already received PMCR from a previous ONE_REG call, +* maintain its immutable flags +*/ + pmcr =__vcpu_sys_reg (vcpu, r->reg); +if (!__vcpu_sys_reg (vcpu, r->reg)) + pmcr =read_sysreg (pmcr_el0); +/* * Writable bits of PMCR_EL0 (ARMV8_PMU_PMCR_MASK) are reset to
Quoted 2 lines
UNKNOWN
* except PMCR.E resetting to zero.
I'm afraid you may need a bit more than just this hack. At the moment, although we can write junk into the shadow copy of PMCR_EL0, the reset will make that behave correctly. With this patch, the junk sticks and gets exposed to the guest.
You need at least a .set_user callback to the handling of PMCR_EL0 so that the value stored is legal, follows the architectural behaviour, and matches the host capabilities.
Thanks,
M.
Signature · 1 line
Jazz is not dead. It just smells funny...