Re: [PATCH 5/6] KVM: selftests: memslot_perf_test: Consolidate memory sizes

From: Gavin Shan <gshan@redhat.com>
To: Sean Christopherson <seanjc@google.com>
Cc: "Maciej S. Szmigiero" <mail@maciej.szmigiero.name>,
	kvmarm@lists.cs.columbia.edu, kvmarm@lists.linux.dev,
	kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
	ajones@ventanamicro.com, pbonzini@redhat.com, maz@kernel.org,
	shuah@kernel.org, oliver.upton@linux.dev, peterx@redhat.com,
	ricarkol@google.com, zhenyzha@redhat.com, shan.gavin@gmail.com
Subject: Re: [PATCH 5/6] KVM: selftests: memslot_perf_test: Consolidate memory sizes
Date: Tue, 18 Oct 2022 07:39:24 +0800	[thread overview]
Message-ID: <c1442e8f-e86e-abee-fb4e-6f4a95697d17@redhat.com> (raw)
Message-ID: <20221017233924.eZ1QwTrV3Fv6lsoDl7iMlsm-LsZY0uZwIIpZ509QPXA@z> (raw)
In-Reply-To: <Y03l+kwGxNW1Icp/@google.com>

On 10/18/22 7:32 AM, Sean Christopherson wrote:
> On Tue, Oct 18, 2022, Gavin Shan wrote:
>> On 10/18/22 6:56 AM, Maciej S. Szmigiero wrote:
>>> On 18.10.2022 00:51, Gavin Shan wrote:
>>>> On 10/18/22 6:08 AM, Sean Christopherson wrote:
>>>>> On Mon, Oct 17, 2022, Maciej S. Szmigiero wrote:
>>>>>>> +#define MEM_EXTRA_SIZE        0x10000
>>>>>>
>>>>>> Also, an expression like "(64 << 10)" is more readable than a "1"
>>>>>> with a tail of zeroes (it's easy to add one zero too many or be one
>>>>>> zero short).
>>>>>
>>>>> +1 to not open coding raw numbers.
>>>>>
>>>>> I think it's high time KVM selftests add #defines for the common sizes, e.g. SIZE_4KB,
>>>>> 16KB, 64K, 2MB, 1GB, etc...
>>>>>
>>>>> Alternatively (or in addition), just #define 1KB, 1MB, 1GB, and 1TB, and then do
>>>>> math off of those.
>>>>>
>>>>
>>>> Ok. I will have one separate patch to define those sizes in kvm_util_base.h,
>>>> right after '#define NSEC_PER_SEC 1000000000L'. Sean, could you let me know
>>>> if it looks good to you?
>>>>
>>>>       #define KB         (1UL << 10)
>>>>       #define MB         (1UL << 20)
>>>>       #define GB         (1UL << 30)
>>>>       #define TB         (1UL << 40)
> 
> Any objection to prefixing these with SIZE_ as well?  IMO it's worth burning the
> extra five characters to make it all but impossible to misinterpret code.
> 

'SIZE_' prefix works for me either.

>>>>       /* Base page and huge page size */
>>>>       #define SIZE_4KB   (  4 * KB)
>>>>       #define SIZE_16KB  ( 16 * KB)
>>>>       #define SIZE_64KB  ( 64 * KB)
>>>>       #define SIZE_2MB   (  2 * MB)
>>>>       #define SIZE_32MB  ( 32 * MB)
>>>>       #define SIZE_512MB (512 * MB)
>>>>       #define SIZE_1GB   (  1 * GB)
>>>>       #define SIZE_16GB  ( 16 * GB)
>>>
>>> FYI, QEMU uses KiB, MiB, GiB, etc., see [1].
>>>
>>
>> Right. I checked QEMU's definitions and it makes sense to use
>> KiB, MiB, GiB, TiB. I don't think we need PiB and EiB because
>> our tests don't use that large memory.
> 
> Ha!  I had typed out KiB, etc... but then thought, "nah, I'm being silly".  KiB
> and friends work for me.
> 

Thanks for your confirm, Sean.

Thanks,
Gavin