Message-ID: <4cfaad4f-8e0f-f550-5052-3a3fe54ec946@arm.com>
Date:   Fri, 14 Jul 2023 11:24:29 +0100
From:   Suzuki K Poulose <suzuki.poulose@....com>
To:     Tao Zhang <quic_taozha@...cinc.com>,
        Mike Leach <mike.leach@...aro.org>
Cc:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Mathieu Poirier <mathieu.poirier@...aro.org>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Konrad Dybcio <konradybcio@...il.com>,
        Rob Herring <robh+dt@...nel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
        Jinlong Mao <quic_jinlmao@...cinc.com>,
        Leo Yan <leo.yan@...aro.org>, coresight@...ts.linaro.org,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
        devicetree@...r.kernel.org,
        Tingwei Zhang <quic_tingweiz@...cinc.com>,
        Yuanfang Zhang <quic_yuanfang@...cinc.com>,
        Trilok Soni <quic_tsoni@...cinc.com>,
        Hao Zhang <quic_hazha@...cinc.com>,
        linux-arm-msm@...r.kernel.org, andersson@...nel.org
Subject: Re: [PATCH v6 09/13] Add nodes for dsb edge control

On 14/07/2023 06:50, Tao Zhang wrote:
> 
> On 7/14/2023 12:37 AM, Suzuki K Poulose wrote:
>> On 13/07/2023 17:13, Tao Zhang wrote:
>>>
>>> On 7/13/2023 5:34 PM, Suzuki K Poulose wrote:
>>>> On 13/07/2023 09:54, Mike Leach wrote:
>>>>> HI Tao,
>>>>>
>>>>> On Wed, 12 Jul 2023 at 14:53, Tao Zhang <quic_taozha@...cinc.com> 
>>>>> wrote:
>>>>>>
>>>>>>
>>>>>> On 6/20/2023 9:41 PM, Suzuki K Poulose wrote:
>>>>>>> On 20/06/2023 09:31, Tao Zhang wrote:
>>>>>>>>
>>>>>>>> On 6/20/2023 3:37 PM, Greg Kroah-Hartman wrote:
>>>>>>>>> On Tue, Jun 20, 2023 at 03:32:37PM +0800, Tao Zhang wrote:
>>>>>>>>>> Add the nodes to set value for DSB edge control and DSB edge
>>>>>>>>>> control mask. Each DSB subunit TPDM has maximum of n(n<16) EDCR
>>>>>>>>>> resgisters to configure edge control. DSB edge detection control
>>>>>>>>>> 00: Rising edge detection
>>>>>>>>>> 01: Falling edge detection
>>>>>>>>>> 10: Rising and falling edge detection (toggle detection)
>>>>>>>>>> And each DSB subunit TPDM has maximum of m(m<8) ECDMR 
>>>>>>>>>> registers to
>>>>>>>>>> configure mask. Eight 32 bit registers providing DSB interface
>>>>>>>>>> edge detection mask control.
>>>>>>>>>>
>>>>>>>>>> Signed-off-by: Tao Zhang <quic_taozha@...cinc.com>
>>>>>>>>>> ---
>>>>>>>>>> .../ABI/testing/sysfs-bus-coresight-devices-tpdm |  32 +++++
>>>>>>>>>>    drivers/hwtracing/coresight/coresight-tpdm.c | 143
>>>>>>>>>> ++++++++++++++++++++-
>>>>>>>>>>    drivers/hwtracing/coresight/coresight-tpdm.h | 22 ++++
>>>>>>>>>>    3 files changed, 196 insertions(+), 1 deletion(-)
>>>>>>>>>>
>>>>>>>>>> diff --git
>>>>>>>>>> a/Documentation/ABI/testing/sysfs-bus-coresight-devices-tpdm
>>>>>>>>>> b/Documentation/ABI/testing/sysfs-bus-coresight-devices-tpdm
>>>>>>>>>> index 2a82cd0..34189e4a 100644
>>>>>>>>>> --- a/Documentation/ABI/testing/sysfs-bus-coresight-devices-tpdm
>>>>>>>>>> +++ b/Documentation/ABI/testing/sysfs-bus-coresight-devices-tpdm
>>>>>>>>>> @@ -60,3 +60,35 @@ Description:
>>>>>>>>>>            Bit[3] : Set to 0 for low performance mode.
>>>>>>>>>>                     Set to 1 for high performance mode.
>>>>>>>>>>            Bit[4:8] : Select byte lane for high performance mode.
>>>>>>>>>> +
>>>>>>>>>> +What: /sys/bus/coresight/devices/<tpdm-name>/dsb_edge_ctrl
>>>>>>>>>> +Date:        March 2023
>>>>>>>>>> +KernelVersion    6.5
>>>>>>>>>> +Contact:    Jinlong Mao (QUIC) <quic_jinlmao@...cinc.com>, Tao
>>>>>>>>>> Zhang (QUIC) <quic_taozha@...cinc.com>
>>>>>>>>>> +Description:
>>>>>>>>>> +        Read/Write a set of the edge control registers of the 
>>>>>>>>>> DSB
>>>>>>>>>> +        in TPDM.
>>>>>>>>>> +
>>>>>>>>>> +        Expected format is the following:
>>>>>>>>>> +        <integer1> <integer2> <integer3>
>>>>>>>>> sysfs is "one value", not 3.  Please never have to parse a 
>>>>>>>>> sysfs file.
>>>>>>>>
>>>>>>>> Do you mean sysfs file can only accept "one value"?
>>>>>>>>
>>>>>>>> I see that more than one value are written to the sysfs file
>>>>>>>> "trigout_attach".
>>>>>>>>
>>>>>>>>>
>>>>>>>>>> +static ssize_t dsb_edge_ctrl_show(struct device *dev,
>>>>>>>>>> +                       struct device_attribute *attr,
>>>>>>>>>> +                       char *buf)
>>>>>>>>>> +{
>>>>>>>>>> +    struct tpdm_drvdata *drvdata = dev_get_drvdata(dev->parent);
>>>>>>>>>> +    ssize_t size = 0;
>>>>>>>>>> +    unsigned long bytes;
>>>>>>>>>> +    int i;
>>>>>>>>>> +
>>>>>>>>>> +    spin_lock(&drvdata->spinlock);
>>>>>>>>>> +    for (i = 0; i < TPDM_DSB_MAX_EDCR; i++) {
>>>>>>>>>> +        bytes = sysfs_emit_at(buf, size,
>>>>>>>>>> +                  "Index:0x%x Val:0x%x\n", i,
>>>>>>>>> Again, no, one value, no "string" needed to parse anything.
>>>>>>>>
>>>>>>>> I also see other sysfs files can be read more than one value in 
>>>>>>>> other
>>>>>>>> drivers.
>>>>>>>>
>>>>>>>> Is this "one value" limitation the usage rule of Linux sysfs 
>>>>>>>> system?
>>>>>>>>
>>>>>>>> Or am I misunderstanding what you mean?
>>>>>>>
>>>>>>> Please fix the other sysfs tunables in the following patches.
>>>>>>
>>>>>> List a new solution for the similar cases below, please see if this
>>>>>> design is reasonable?
>>>>>>
>>>>>> 1. Two SysFS files("dsb_edge_ctrl_idx" and "dsb_edge_ctrl_val") 
>>>>>> will be
>>>>>> created in this case.
>>>>>>
>>>>>> 2. First write to the node "dsb_edge_ctrl_idx" to set the index 
>>>>>> number
>>>>>> of the edge detection.
>>>>>>
>>>>>> 3. Then write to the node "dsb_edge_ctrl_val" to set the value of the
>>>>>> edge detection.
>>>>>>
>>>>>> For example, if we need need to set "Falling edge detection" to 
>>>>>> the edge
>>>>>> detection #220-#222, we can issue the following commands.
>>>>>>
>>>>>> echo 0xdc > tpdm1/dsb_edge_ctrl_idx
>>>>>>
>>>>>> echo 0x1 > tpdm1/dsb_edge_ctrl_val
>>>>>>
>>>>>> echo 0xdd > tpdm1/dsb_edge_ctrl_idx
>>>>>>
>>>>>> echo 0x1 > tpdm1/dsb_edge_ctrl_val
>>>>>>
>>>>>> echo 0xde > tpdm1/dsb_edge_ctrl_idx
>>>>>>
>>>>>> echo 0x1 > tpdm1/dsb_edge_ctrl_val
>>>>>>
>>>>>> If this design is acceptable, we will rewrite other similar nodes 
>>>>>> based
>>>>>> on this solution.
>>>>>>
>>>>>
>>>>> This index / value model is used in the coresight drivers so should be
>>>>> OK - eg etm4 has cntr_idx / cntrldvr / cntr_val / cntr_ctrl, where
>>>>> index selects the counter, and the other val registers are applied to
>>>>> that counter.
>>>>
>>>> True. That model is useful when there are variable number of 
>>>> "counters".
>>>> I guess it doesn't hurt to have a 64bit (or even 32bit) file for each
>>>> EDCR.
>>>>
>>>> e.g, edcr0...edcr15
>>>>
>>>> Given there are only 16 of them, it is fine to keep a file for each.
>>>> This may be grouped under "mgmt" similar to what we have for other
>>>> components. That way, it can be easily hidden by checking for the
>>>> presence of DSB.
>>>
>>> The number of EDCR registers is not fixed. The maximum range is [0:15].
>>>
>>> But the address of the maximum number of the registers will be 
>>> reserved first,
>>>
>>> and can be accessed safely even if the TPDM doesn't have the maximum 
>>> number
>>>
>>> of  EDCR registers.
>>>
>>> And we are not able to dynamically know the number of EDCR registers 
>>> per DSB
>>>
>>> TPDM.
>>>
>>> Can we use our proposal in this case?
>>
>> Please provide a file edcrN for each of the 0 <= N < 16. That way it is
>> easier to avoid locking the index. It doesn't matter how many EDCRs are
>> supported, there is a maximum limit and it is always guaranteed to be
>> write safe, if some are not implemented. Thus it is much easier from a 
>> programming perspective too.
> 
> Hi Suzuki,
> 
> 
> Thanks for the suggestion.
> 
> I'd like to further clarify our proposal below in case I didn't express 
> it clearly before.
> 
> 1. In our design, the users don't need to know the mapping between the 
> number of the edge detection
> 
> and the control bits in EDCRN registers. They only need to focus on the 
> edge detection they need, don't
> 
> need to care about the design of the HW.

Agreed

> 
> 2. For example, if there are two users configure in the same test. One 
> needs to configure edge detection #7
> 
> as "Falling edge detection". The other one needs to configure edge 
> detection #8 as "toggle detection". They will
> 
> issue the following commands to implement it.
> 
> echo 0x7 > tpdm1/dsb_edge_ctrl_idx
> 
> echo 0x1 > tpdm1/dsb_edge_ctrl_val
> 
> echo 0x8 > tpdm1/dsb_edge_ctrl_idx
> 
> echo 0x2 > tpdm1/dsb_edge_ctrl_val
> 
> The value written to edcr0 will be 0x24000 in our proposal.
> 
> But in the solution of "tpdm1/dsb_edge_ctrl/edcrN 0 <= N < 16".
> 
> One user calculates that he needs to write 0x4000 to edcr0.
> 
> echo 0x4000 > tpdm1/dsb_edge_ctrl/edcr0
> 
> The other one calculates that he needs to write 0x20000 to edcr0.
> 
> echo 0x20000 > tpdm1/dsb_edge_ctrl/edcr0
> 
> The last write will overwrite the previous value in this case and 0x20000
> 
> will be written to the edcr0 finally.

The solution of edcrN expects the users follow a Read-Modify-Write.
But given you want to control individual lines separately (which are 256
in number), I am fine with the _idx/value solution.

> 
> 3. Some DSB TPDMs may not have 16 EDCR registers. For example, TPDM2
> 
> may only have 7 EDCR registers. If we still create 16 edcr file at 
> tpdm2/dsb_edge_ctrl,
> 
> this may confuse users.

This is not relevant. The user can't know the maximum number anyway. If 
the user knows TPDM2 has only 7 EDCR, then don't bother about the other
files.

Please go ahead with the _idx /_value

Suzuki

Powered by blists - more mailing lists