[U-Boot] [PATCH] x86: queensbay: switche P state to the highest freq

[Christian Gmeiner <christian.gmeiner@gmail.com>]

2018-04-10 11:11 GMT+02:00 Bin Meng <bmeng.cn@gmail.com>:
> Hi Christian,
>
> On Tue, Apr 10, 2018 at 5:09 PM, Christian Gmeiner
> <christian.gmeiner@gmail.com> wrote:
>> Hi Bin,
>>
>> 2018-04-10 11:07 GMT+02:00 Bin Meng <bmeng.cn@gmail.com>:
>>> Hi Christian,
>>>
>>> On Tue, Apr 10, 2018 at 4:47 PM, Christian Gmeiner
>>> <christian.gmeiner@gmail.com> wrote:
>>>> Hi Bin
>>>>
>>>> Thanks for you quick review!
>>>>
>>>>>
>>>>> On Tue, Apr 10, 2018 at 4:01 PM, Christian Gmeiner
>>>>> <christian.gmeiner@gmail.com> wrote:
>>>>>> Fixes performance issue when running vx5/vx7 images.
>>>>>
>>>>> Could you elaborate more on what performance issue you are seeing?
>>>>>
>>>>
>>>> The cache performance was bad without this change. We found this problem
>>>> when switching from bldk to u-boot.
>>>>
>>>> old: https://imagebin.ca/v/3xssw7stbY6A
>>>> new: https://imagebin.ca/v/3xstHMCC9fmJ
>>>
>>> Thank you for the benchmark data. Then please consider describing your
>>> benchmark test suite and on what configuration the data is worse
>>> without this patch (eg: sequential 128-bit read) in the commit
>>> message. This will help other people understand the patch and the
>>> commit in the future.
>>>
>>
>> Will do.
>>
>>>>
>>>>> nits: better to spell out "VxWorks"
>>>>
>>>> ok
>>>>
>>>>>
>>>>>>
>>>>>> Signed-off-by: Christian Gmeiner <christian.gmeiner@gmail.com>
>>>>>> ---
>>>>>>  arch/x86/cpu/queensbay/Makefile |  2 +-
>>>>>>  arch/x86/cpu/queensbay/cpu.c    | 61 +++++++++++++++++++++++++++++++++++++++++
>>>>>>  2 files changed, 62 insertions(+), 1 deletion(-)
>>>>>>  create mode 100644 arch/x86/cpu/queensbay/cpu.c
>>>>>>
>>>>>> diff --git a/arch/x86/cpu/queensbay/Makefile b/arch/x86/cpu/queensbay/Makefile
>>>>>> index c0681995bd..3dd23465d4 100644
>>>>>> --- a/arch/x86/cpu/queensbay/Makefile
>>>>>> +++ b/arch/x86/cpu/queensbay/Makefile
>>>>>> @@ -5,4 +5,4 @@
>>>>>>  #
>>>>>>
>>>>>>  obj-y += fsp_configs.o irq.o
>>>>>> -obj-y += tnc.o
>>>>>> +obj-y += tnc.o cpu.o
>>>>>> diff --git a/arch/x86/cpu/queensbay/cpu.c b/arch/x86/cpu/queensbay/cpu.c
>>>>>> new file mode 100644
>>>>>> index 0000000000..e98e4ac8ef
>>>>>> --- /dev/null
>>>>>> +++ b/arch/x86/cpu/queensbay/cpu.c
>>>>>> @@ -0,0 +1,61 @@
>>>>>> +/*
>>>>>> + * Copyright (C) 2016, Bachmann electronic GmbH
>>>>>
>>>>> nits: 2018?
>>>>>
>>>>
>>>> This change was made some months ago :) But yeah.. can change it.
>>>>
>>>>>> + *
>>>>>> + * SPDX-License-Identifier:    GPL-2.0+
>>>>>> + */
>>>>>> +
>>>>>> +#include <common.h>
>>>>>> +#include <cpu.h>
>>>>>> +#include <dm.h>
>>>>>> +#include <asm/cpu.h>
>>>>>> +#include <asm/cpu_x86.h>
>>>>>> +#include <asm/msr.h>
>>>>>> +
>>>>>> +static void set_max_freq(void)
>>>>>> +{
>>>>>> +       msr_t msr;
>>>>>> +
>>>>>> +       msr = msr_read(MSR_IA32_MISC_ENABLES);
>>>>>> +       msr.lo |= (1 << 16);
>>>>>
>>>>> Please use macro instead of magic numbers
>>>>
>>>> ok
>>>>
>>>>>
>>>>>> +       msr_write(MSR_IA32_MISC_ENABLES, msr);
>>>>>> +
>>>>>> +       msr = msr_read(MSR_IA32_PERF_CTL);
>>>>>> +       msr.lo = 0x101f;
>>>>>
>>>>> ditto
>>>>
>>>> ok

I *think* I have read this msr running under a 'good' bios so I am not
really sure
what the meaning of the bits are. I tried a quick google and did not found any
E6xx datasheet containing the description. Is it okay to just use the
magic value here?

-- 
greets
--
Christian Gmeiner, MSc

https://christian-gmeiner.info