Message-ID: <5540adcd-4ba6-53e7-c7fe-b7116e6403ca@quicinc.com>
Date:   Tue, 21 Nov 2023 12:01:15 +0530
From:   Devi Priya <quic_devipriy@...cinc.com>
To:     Christian Marangi <ansuelsmth@...il.com>
CC:     Andy Gross <agross@...nel.org>,
        Bjorn Andersson <andersson@...nel.org>,
        Konrad Dybcio <konrad.dybcio@...aro.org>,
        Michael Turquette <mturquette@...libre.com>,
        Stephen Boyd <sboyd@...nel.org>,
        <linux-arm-msm@...r.kernel.org>, <linux-clk@...r.kernel.org>,
        <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v6 2/3] clk: qcom: clk-rcg2: add support for rcg2 freq
 multi ops



On 11/20/2023 11:44 PM, Christian Marangi wrote:
> On Mon, Nov 20, 2023 at 03:51:50PM +0530, Devi Priya wrote:
>>
>>
>> On 9/16/2023 7:30 PM, Christian Marangi wrote:
>>> Some RCG frequency can be reached by multiple configuration.
>>>
>>> Add clk_rcg2_fm_ops ops to support these special RCG configurations.
>>>
>>> These alternative ops will select the frequency using a CEIL policy.
>>>
>>> When the correct frequency is found, the correct config is selected by
>>> calculating the final rate (by checking the defined parent and values
>>> in the config that is being checked) and deciding based on the one that
>>> is less different than the requested one.
>>>
>>> These check are skipped if there is just on config for the requested
>>> freq.
>>>
>>> qcom_find_freq_multi is added to search the freq with the new struct
>>> freq_multi_tbl.
>>> __clk_rcg2_select_conf is used to select the correct conf by simulating
>>> the final clock.
>>> If a conf can't be found due to parent not reachable, a WARN is printed
>>> and -EINVAL is returned.
>>>
>>> Signed-off-by: Christian Marangi <ansuelsmth@...il.com>
>>> ---
>>>    drivers/clk/qcom/clk-rcg.h  |   1 +
>>>    drivers/clk/qcom/clk-rcg2.c | 167 ++++++++++++++++++++++++++++++++++++
>>>    drivers/clk/qcom/common.c   |  18 ++++
>>>    drivers/clk/qcom/common.h   |   2 +
>>>    4 files changed, 188 insertions(+)
>>>
>>> diff --git a/drivers/clk/qcom/clk-rcg.h b/drivers/clk/qcom/clk-rcg.h
>>> index c81458db6ce4..dc9a77965e68 100644
>>> --- a/drivers/clk/qcom/clk-rcg.h
>>> +++ b/drivers/clk/qcom/clk-rcg.h
>>> @@ -190,6 +190,7 @@ struct clk_rcg2_gfx3d {
>>>    extern const struct clk_ops clk_rcg2_ops;
>>>    extern const struct clk_ops clk_rcg2_floor_ops;
>>> +extern const struct clk_ops clk_rcg2_fm_ops;
>>>    extern const struct clk_ops clk_rcg2_mux_closest_ops;
>>>    extern const struct clk_ops clk_edp_pixel_ops;
>>>    extern const struct clk_ops clk_byte_ops;
>>> diff --git a/drivers/clk/qcom/clk-rcg2.c b/drivers/clk/qcom/clk-rcg2.c
>>> index e22baf3a7112..617e7ff0f6a3 100644
>>> --- a/drivers/clk/qcom/clk-rcg2.c
>>> +++ b/drivers/clk/qcom/clk-rcg2.c
>>> @@ -266,6 +266,116 @@ static int _freq_tbl_determine_rate(struct clk_hw *hw, const struct freq_tbl *f,
>>>    	return 0;
>>>    }
>>> +static const struct freq_conf *
>>> +__clk_rcg2_select_conf(struct clk_hw *hw, const struct freq_multi_tbl *f,
>>> +		       unsigned long req_rate)
>>> +{
>>> +	unsigned long rate_diff, best_rate_diff = ULONG_MAX;
>>> +	const struct freq_conf *conf, *best_conf;
>>> +	struct clk_rcg2 *rcg = to_clk_rcg2(hw);
>>> +	const char *name = clk_hw_get_name(hw);
>>> +	unsigned long parent_rate, rate;
>>> +	struct clk_hw *p;
>>> +	int index, i;
>>> +
>>> +	/* Init best_conf to the first conf */
>>> +	best_conf = f->confs;
>>> +
>>> +	/* Exit early if only one config is defined */
>>> +	if (f->num_confs == 1)
>>> +		goto exit;
>>> +
>>> +	/* Search in each provided config the one that is near the wanted rate */
>>> +	for (i = 0, conf = f->confs; i < f->num_confs; i++, conf++) {
>>> +		index = qcom_find_src_index(hw, rcg->parent_map, conf->src);
>>> +		if (index < 0)
>>> +			continue;
>>> +
>>> +		p = clk_hw_get_parent_by_index(hw, index);
>>> +		if (!p)
>>> +			continue;
>>> +
>>> +		parent_rate =  clk_hw_get_rate(p);
>>> +		rate = calc_rate(parent_rate, conf->n, conf->m, conf->n, conf->pre_div);
>>> +
>>> +		if (rate == req_rate) {
>>> +			best_conf = conf;
>>> +			goto exit;
>>> +		}
>>> +
>>> +		rate_diff = abs(req_rate - rate);
>>> +		if (rate_diff < best_rate_diff) {
>>> +			best_rate_diff = rate_diff;
>>> +			best_conf = conf;
>>> +		}
>>> +	}
>>> +
>>> +	/*
>>> +	 * Very unlikely. Warn if we couldn't find a correct config
>>> +	 * due to parent not found in every config.
>>> +	 */
>>> +	if (unlikely(i == f->num_confs)) {
>>> +		WARN(1, "%s: can't find a configuration for rate %lu.",
>>> +		     name, req_rate);
>>> +		return ERR_PTR(-EINVAL);
>>> +	}
>> Hi Christian,
>>
>> Thanks a lot for the patch!
>> We have incorporated these changes along with the corresponding clock driver
>> changes & tested it on IPQ9574 & IPQ5332 targets.
>>
>> When setting the clk rate for the nss port clocks, for the requested
>> frequency the correct config gets selected and the
>> clk rate is set properly.
>> We see the WARN getting printed for other frequencies (rate * i where
>> i=2 to maxdiv) that is requested by the clk_hw_round_rate function.
>>
>> Upon analysis, we see that the for loop in clk_divider_bestdiv iterates
>> until the maxdiv value and requests (rate*i) via the clk_hw_round_rate
>> API to find the bestdiv and best_parent_rate. For frequencies which are
>> multiples of the requested frequency (rate*i where i=2 to maxdiv), it
>> seems unlikely to see the WARN being printed.
>>
>> Can you please help us understand when the WARN is likely to be printed
>> & Looking forward to your suggestions on how this WARN could
>> be suppressed in the afore mentioned scenario!
>>
> 
> Hi,
> 
> thanks a lot for testing this. Maybe was a small oversight by me.
> 
> I attached an alternative patch. Can you test it and tell me if the WARN
> is still printed? (the WARN must be printed only when the parent is not
> found, I don't think it's your case)
> 
Hi Christian,

WARN does not get printed with the attached patchset.
Thanks much!

Regards,
Devi Priya

Powered by blists - more mailing lists