Re: [kvm-unit-tests PATCH] arm64: add micro test

2018-05-03 Thread Kalra, Ashish
>>On 5/3/2018 9:54 PM, Marc Zyngier wrote: On 03/05/18 12:12, Kalra, Ashish wrote: The reason for the problem you're seeing is probably that the constraints are not quite right. Here, x0 is not simply clobbered, but is actively written to in the middle of the sequence (it is

Re: [kvm-unit-tests PATCH] arm64: add micro test

2018-05-03 Thread Kalra, Ashish
The reason for the problem you're seeing is probably that the constraints are not quite right. Here, x0 is not simply clobbered, but is actively written to in the middle of the sequence (it is at least an early clobber). It is also, I assume, a result from the hypercall, so it cannot simply be

Re: [kvm-unit-tests PATCH] arm64: add micro test

2018-05-01 Thread Kalra, Ashish
Hello Shih-Wei, This exception is probably due to a bug in the hvc_test (hvc call) code, and there is a generic bug in the hvc test code. In the function hvc_test(), local variables c1, c2 get mapped to registers (typically x1 & x2) and those are "clobbered" across the hvc call, hence, c1 gets

Re: [kvm-unit-tests PATCH] arm64: add micro test

2017-12-18 Thread Kalra, Ashish
Hello Shih-Wei, This "micro" test is really useful for micro-architectural performance measurement of KVM/ARM. But, to get more accurate cycle counts, it will be preferable to compute counter read (and isb) overhead and deduct it from the final cycle count, something like this : +static unsigne

Re: [kvm-unit-tests PATCH] arm64: add micro test

2017-12-18 Thread Kalra, Ashish
Hello Shih-Wei, This "micro" test is really useful for micro-architectural performance measurement of KVM/ARM. But, to get more accurate cycle counts, it will be preferable to compute counter read (and isb) overhead and deduct it from the final cycle count, something like this : +static unsigne

Re: [kvm-unit-tests PATCH] arm64: add micro test

2017-12-18 Thread Kalra, Ashish
Hello Shih-Wei, This "micro" test is really useful for micro-architectural performance measurement of KVM/ARM. But, to get slightly more accurate cycle counts, it will be preferable to compute counter read (and isb) overhead and deduct it from the final cycle count, something like this : +

Re: kvm-unit-tests arm ptimer test

2017-08-10 Thread Kalra, Ashish
. Thanks, Ashish On 8/10/2017 3:01 AM, Kalra, Ashish wrote: Hello, I have a general question on the ptimer test for KVM unit tests for the ARM/ARM64 platform, as the guest is only allowed to access to the physical counter and access to the physical timer is not allowed for the guest (CNTHCTL.EL1PCTEN

kvm-unit-tests arm ptimer test

2017-08-09 Thread Kalra, Ashish
Hello, I have a general question on the ptimer test for KVM unit tests for the ARM/ARM64 platform, as the guest is only allowed to access to the physical counter and access to the physical timer is not allowed for the guest (CNTHCTL.EL1PCTEN is set and CNTHCTL.EL1PCEN is clear), so how is thi