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] [thread-next>] [day] [month] [year] [list]
Date:   Wed, 11 Mar 2020 19:39:15 +0900
From:   Tomasz Figa <tfiga@...omium.org>
To:     Sakari Ailus <sakari.ailus@...ux.intel.com>
Cc:     Linux Media Mailing List <linux-media@...r.kernel.org>,
        Shunqian Zheng <zhengsq@...k-chips.com>,
        Mauro Carvalho Chehab <mchehab@...nel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Dongchun Zhu <dongchun.zhu@...iatek.com>
Subject: Re: [PATCH] media: i2c: ov5695: Fix power on and off sequences

Hi Sakari,

On Sat, Dec 21, 2019 at 12:19 AM Sakari Ailus
<sakari.ailus@...ux.intel.com> wrote:
>
> Hi Tomasz,
>
> On Fri, Dec 20, 2019 at 10:08:00PM +0900, Tomasz Figa wrote:
> > From: Dongchun Zhu <dongchun.zhu@...iatek.com>
> >
> > From the measured hardware signal, OV5695 reset pin goes high for a
> > short period of time during boot-up. From the sensor specification, the
> > reset pin is active low and the DT binding defines the pin as active
> > low, which means that the values set by the driver are inverted and thus
> > the value requested in probe ends up high.
> >
> > Fix it by changing probe to request the reset GPIO initialized to high,
> > which makes the initial state of the physical signal low.
> >
> > In addition, DOVDD rising must occur before DVDD rising from spec., but
> > regulator_bulk_enable() API enables all the regulators asynchronously.
> > Use an explicit loops of regulator_enable() instead.
> >
> > For power off sequence, it is required that DVDD falls first. Given the
> > bulk API does not give any guarantee about the order of regulators,
> > change the driver to use regulator_disable() instead.
> >
> > The sensor also requires a delay between reset high and first I2C
> > transaction, which was assumed to be 8192 XVCLK cycles, but 1ms is
> > recommended by the vendor. Fix this as well.
> >
> > Signed-off-by: Dongchun Zhu <dongchun.zhu@...iatek.com>
> > Signed-off-by: Tomasz Figa <tfiga@...omium.org>
> > ---
> >  drivers/media/i2c/ov5695.c | 41 +++++++++++++++++++++-----------------
> >  1 file changed, 23 insertions(+), 18 deletions(-)
> >
> > diff --git a/drivers/media/i2c/ov5695.c b/drivers/media/i2c/ov5695.c
> > index d6cd15bb699ac..8d0cc3893fcfc 100644
> > --- a/drivers/media/i2c/ov5695.c
> > +++ b/drivers/media/i2c/ov5695.c
> > @@ -971,16 +971,9 @@ static int ov5695_s_stream(struct v4l2_subdev *sd, int on)
> >       return ret;
> >  }
> >
> > -/* Calculate the delay in us by clock rate and clock cycles */
> > -static inline u32 ov5695_cal_delay(u32 cycles)
> > -{
> > -     return DIV_ROUND_UP(cycles, OV5695_XVCLK_FREQ / 1000 / 1000);
> > -}
> > -
> >  static int __ov5695_power_on(struct ov5695 *ov5695)
> >  {
> > -     int ret;
> > -     u32 delay_us;
> > +     int i, ret;
> >       struct device *dev = &ov5695->client->dev;
> >
> >       ret = clk_prepare_enable(ov5695->xvclk);
> > @@ -991,21 +984,24 @@ static int __ov5695_power_on(struct ov5695 *ov5695)
> >
> >       gpiod_set_value_cansleep(ov5695->reset_gpio, 1);
> >
> > -     ret = regulator_bulk_enable(OV5695_NUM_SUPPLIES, ov5695->supplies);
> > -     if (ret < 0) {
> > -             dev_err(dev, "Failed to enable regulators\n");
> > -             goto disable_clk;
> > +     for (i = 0; i < OV5695_NUM_SUPPLIES; i++) {
> > +             ret = regulator_enable(ov5695->supplies[i].consumer);
>
> The regulator voltage takes some time before it settles. If the hardware
> requires a particular order, then presumably there should be a small delay
> to ensure that. 1 ms should be plenty.

The regulator API guarantees that when regulator_enable() returns, the
voltage is stable. Regulator ramp up delays can be also configured via
DT to take care for per-platform variability.

>
> I also think it'd be necessary to add a comment here explaining the
> requirements for enabling regulators, as otherwise I expect someone to
> "fix" this sooner or later.

True. Let me add a comment.

>
> Same for powering off.
>

Same as above.

Best regards,
Tomasz

Powered by blists - more mailing lists