linux-kernel - Re: [PATCH 4/6] usb: dwc3: Handle fractional reference clocks

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20d904b3-30e9-eb19-3b7f-6d53d8d2806d@seco.com>
Date:   Tue, 18 Jan 2022 18:46:05 -0500
From:   Sean Anderson <sean.anderson@...o.com>
To:     Robert Hancock <robert.hancock@...ian.com>,
        "gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>,
        "linux-usb@...r.kernel.org" <linux-usb@...r.kernel.org>
Cc:     "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "bjagadee@...eaurora.org" <bjagadee@...eaurora.org>,
        "balbi@...nel.org" <balbi@...nel.org>,
        "baruch@...s.co.il" <baruch@...s.co.il>
Subject: Re: [PATCH 4/6] usb: dwc3: Handle fractional reference clocks



On 1/14/22 8:09 PM, Robert Hancock wrote:
> On Fri, 2022-01-14 at 18:39 -0500, Sean Anderson wrote:
>> GUCTL.REFCLKPER can only account for clock frequencies with integer
>> periods. To address this, program REFCLK_FLADJ with the relative error
>> caused by period truncation. The formula given in the register reference
>> has been rearranged to allow calculation based on rate (instead of
>> period), and to allow for fixed-point arithmetic.
>> 
>> Signed-off-by: Sean Anderson <sean.anderson@...o.com>
>> ---
>> 
>>  drivers/usb/dwc3/core.c | 25 +++++++++++++++++++++++--
>>  drivers/usb/dwc3/core.h |  1 +
>>  2 files changed, 24 insertions(+), 2 deletions(-)
>> 
>> diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c
>> index 5214daceda86..48bb3e42cdd0 100644
>> --- a/drivers/usb/dwc3/core.c
>> +++ b/drivers/usb/dwc3/core.c
>> @@ -348,7 +348,7 @@ static void dwc3_frame_length_adjustment(struct dwc3
>> *dwc)
>>  static void dwc3_ref_clk_period(struct dwc3 *dwc)
>>  {
>>  	u32 reg;
>> -	unsigned long rate, period;
>> +	unsigned long fladj, rate, period;
>>  
>>  	if (dwc->ref_clk) {
>>  		rate = clk_get_rate(dwc->ref_clk);
>> @@ -357,16 +357,37 @@ static void dwc3_ref_clk_period(struct dwc3 *dwc)
>>  		period = NSEC_PER_SEC / rate;
>>  	} else if (dwc->ref_clk_per) {
>>  		period = dwc->ref_clk_per;
>> +		rate = NSEC_PER_SEC / period;
>>  	} else {
>>  		return;
>>  	}
>>  
>> +	/*
>> +	 * The calculation below is
>> +	 *
>> +	 * 125000 * (NSEC_PER_SEC / (rate * period) - 1)
>> +	 *
>> +	 * but rearranged for fixed-point arithmetic.
>> +	 *
>> +	 * Note that rate * period ~= NSEC_PER_SECOND, minus the number of
>> +	 * nanoseconds of error caused by the truncation which happened during
>> +	 * the division when calculating rate or period (whichever one was
>> +	 * derived from the other). We first calculate the relative error, then
>> +	 * scale it to units of 0.08%.
>> +	 */
>> +	fladj = div64_u64(125000ULL * NSEC_PER_SEC, (u64)rate * period);
>> +	fladj -= 125000;
>> +
>>  	reg = dwc3_readl(dwc->regs, DWC3_GUCTL);
>>  	reg &= ~DWC3_GUCTL_REFCLKPER_MASK;
>>  	reg |=  FIELD_PREP(DWC3_GUCTL_REFCLKPER_MASK, period);
>>  	dwc3_writel(dwc->regs, DWC3_GUCTL, reg);
>> -}
>>  
>> 
> 
> dwc3_frame_length_adjustment which also writes to the DWC3_GFLADJ register has
> a conditional to skip it if DWC3_VER_IS_PRIOR(DWC3, 250A), not sure if it's
> needed for this field or not?

That's a good question...

The oldest version I found with it is 2.60a [1].

[1] https://www.mouser.com/pdfdocs/silver-celeron-datasheet-vol-2.pdf

--Sean