[SRU][F][PATCH 1/2] KVM: s390x: fix SCK locking

Fri Feb 17 10:34:50 UTC 2023

On 17-02-2023 10:02, Stefan Bader wrote:
> On 16.02.23 10:54, Roxana Nicolescu wrote:
>> From: Claudio Imbrenda <imbrenda at linux.ibm.com>
>>
>> BugLink: https://bugs.launchpad.net/bugs/1999882
>>
>> [ Upstream commit c0573ba5c5a2244dc02060b1f374d4593c1d20b7 ]
>>
>> When handling the SCK instruction, the kvm lock is taken, even though
>> the vcpu lock is already being held. The normal locking order is kvm
>> lock first and then vcpu lock. This is can (and in some circumstances
>> does) lead to deadlocks.
>>
>> The function kvm_s390_set_tod_clock is called both by the SCK handler
>> and by some IOCTLs to set the clock. The IOCTLs will not hold the vcpu
>> lock, so they can safely take the kvm lock. The SCK handler holds the
>> vcpu lock, but will also somehow need to acquire the kvm lock without
>> relinquishing the vcpu lock.
>>
>> The solution is to factor out the code to set the clock, and provide
>> two wrappers. One is called like the original function and does the
>> locking, the other is called kvm_s390_try_set_tod_clock and uses
>> trylock to try to acquire the kvm lock. This new wrapper is then used
>> in the SCK handler. If locking fails, -EAGAIN is returned, which is
>> eventually propagated to userspace, thus also freeing the vcpu lock and
>> allowing for forward progress.
>>
>> This is not the most efficient or elegant way to solve this issue, but
>> the SCK instruction is deprecated and its performance is not critical.
>>
>> The goal of this patch is just to provide a simple but correct way to
>> fix the bug.
>>
>> Fixes: 6a3f95a6b04c ("KVM: s390: Intercept SCK instruction")
>> Signed-off-by: Claudio Imbrenda <imbrenda at linux.ibm.com>
>> Reviewed-by: Christian Borntraeger <borntraeger at linux.ibm.com>
>> Reviewed-by: Janis Schoetterl-Glausch <scgl at linux.ibm.com>
>> Link: 
>> https://lore.kernel.org/r/20220301143340.111129-1-imbrenda@linux.ibm.com
>> Cc: stable at vger.kernel.org
>> Signed-off-by: Christian Borntraeger <borntraeger at linux.ibm.com>
>> Stable-dep-of: 6973091d1b50 ("KVM: s390: pv: don't allow userspace to 
>> set the clock under PV")
>> Signed-off-by: Sasha Levin <sashal at kernel.org>
>> (backported from commit c0573ba5c5a2244dc02060b1f374d4593c1d20b7)
>
> It would be helpful if you described here what had to be adjusted. 
> That helps reviewers to know what to check. Something like
>
> [roxanan: context adjustments ... or ... adjusted arguments for...]
>
>> Signed-off-by: Roxana Nicolescu <roxana.nicolescu at canonical.com>
>> ---
>>   arch/s390/kvm/kvm-s390.c | 19 ++++++++++++++++---
>>   arch/s390/kvm/kvm-s390.h |  4 ++--
>>   arch/s390/kvm/priv.c     | 15 ++++++++++++++-
>>   3 files changed, 32 insertions(+), 6 deletions(-)
>>
>> diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
>> index 1fb8b8a9fa0f..0f1b0dde0de3 100644
>> --- a/arch/s390/kvm/kvm-s390.c
>> +++ b/arch/s390/kvm/kvm-s390.c
>> @@ -3940,14 +3940,12 @@ static int kvm_s390_handle_requests(struct 
>> kvm_vcpu *vcpu)
>>       return 0;
>>   }
>>   -void kvm_s390_set_tod_clock(struct kvm *kvm,
>> -                const struct kvm_s390_vm_tod_clock *gtod)
>> +static void __kvm_s390_set_tod_clock(struct kvm *kvm, const struct 
>> kvm_s390_vm_tod_clock *gtod)
>>   {
>>       struct kvm_vcpu *vcpu;
>>       struct kvm_s390_tod_clock_ext htod;
>>       int i;
>>   -    mutex_lock(&kvm->lock);
>>       preempt_disable();
>>         get_tod_clock_ext((char *)&htod);
>> @@ -3968,7 +3966,22 @@ void kvm_s390_set_tod_clock(struct kvm *kvm,
>>         kvm_s390_vcpu_unblock_all(kvm);
>>       preempt_enable();
>> +}
>> +
>> +void kvm_s390_set_tod_clock(struct kvm *kvm, const struct 
>> kvm_s390_vm_tod_clock *gtod)
>> +{
>> +    mutex_lock(&kvm->lock);
>> +    __kvm_s390_set_tod_clock(kvm, gtod);
>> +    mutex_unlock(&kvm->lock);
>> +}
>> +
>> +int kvm_s390_try_set_tod_clock(struct kvm *kvm, const struct 
>> kvm_s390_vm_tod_clock *gtod)
>> +{
>> +    if (!mutex_trylock(&kvm->lock))
>> +        return 0;
>> +    __kvm_s390_set_tod_clock(kvm, gtod);
>>       mutex_unlock(&kvm->lock);
>> +    return 1;
>>   }
>>     /**
>> diff --git a/arch/s390/kvm/kvm-s390.h b/arch/s390/kvm/kvm-s390.h
>> index 817d213693e0..c94c1e29eeca 100644
>> --- a/arch/s390/kvm/kvm-s390.h
>> +++ b/arch/s390/kvm/kvm-s390.h
>> @@ -335,8 +335,8 @@ int kvm_s390_handle_sigp(struct kvm_vcpu *vcpu);
>>   int kvm_s390_handle_sigp_pei(struct kvm_vcpu *vcpu);
>>     /* implemented in kvm-s390.c */
>> -void kvm_s390_set_tod_clock(struct kvm *kvm,
>> -                const struct kvm_s390_vm_tod_clock *gtod);
>> +void kvm_s390_set_tod_clock(struct kvm *kvm, const struct 
>> kvm_s390_vm_tod_clock *gtod);
>> +int kvm_s390_try_set_tod_clock(struct kvm *kvm, const struct 
>> kvm_s390_vm_tod_clock *gtod);
>>   long kvm_arch_fault_in_page(struct kvm_vcpu *vcpu, gpa_t gpa, int 
>> writable);
>>   int kvm_s390_store_status_unloaded(struct kvm_vcpu *vcpu, unsigned 
>> long addr);
>>   int kvm_s390_vcpu_store_status(struct kvm_vcpu *vcpu, unsigned long 
>> addr);
>> diff --git a/arch/s390/kvm/priv.c b/arch/s390/kvm/priv.c
>> index 840b383ba756..8f03992d25cc 100644
>> --- a/arch/s390/kvm/priv.c
>> +++ b/arch/s390/kvm/priv.c
>> @@ -103,7 +103,20 @@ static int handle_set_clock(struct kvm_vcpu *vcpu)
>>           return kvm_s390_inject_prog_cond(vcpu, rc);
>>         VCPU_EVENT(vcpu, 3, "SCK: setting guest TOD to 0x%llx", 
>> gtod.tod);
>> -    kvm_s390_set_tod_clock(vcpu->kvm, &gtod);
>> +    /*
>> +     * To set the TOD clock the kvm lock must be taken, but the vcpu 
>> lock
>> +     * is already held in handle_set_clock. The usual lock order is the
>> +     * opposite.  As SCK is deprecated and should not be used in 
>> several
>> +     * cases, for example when the multiple epoch facility or TOD clock
>> +     * steering facility is installed (see Principles of Operation),  a
>> +     * slow path can be used.  If the lock can not be taken via 
>> try_lock,
>> +     * the instruction will be retried via -EAGAIN at a later point in
>> +     * time.
>> +     */
>> +    if (!kvm_s390_try_set_tod_clock(vcpu->kvm, &gtod)) {
>> +        kvm_s390_retry_instr(vcpu);
>> +        return -EAGAIN;
>> +    }
>>         kvm_s390_set_psw_cc(vcpu, 0);
>>       return 0;
>
Thanks Stefan for your review. I just submitted a v2.

Roxana