by Dave Stevenson

[permalink] [raw]

Subject: Re: [PATCH 03/15] drm/vc4: hdmi: Fix register offset with longer CEC messages

Hi Dom & Maxime

On Thu, 10 Dec 2020 at 13:46, Maxime Ripard <[email protected]> wrote:
>
> From: Dom Cobley <[email protected]>
>
> The code prior to 311e305fdb4e ("drm/vc4: hdmi: Implement a register
> layout abstraction") was relying on the fact that the register offset
> was incremented by 4 for each readl call. That worked since the register
> width is 4 bytes.
>
> However, since that commit the HDMI_READ macro is now taking an enum,
> and the offset doesn't increment by 4 but 1 now. Divide the index by 4
> to fix this.
>
> Fixes: 311e305fdb4e ("drm/vc4: hdmi: Implement a register layout abstraction")
> Signed-off-by: Dom Cobley <[email protected]>
> Signed-off-by: Maxime Ripard <[email protected]>

Reviewed-by: Dave Stevenson <[email protected]>

> ---
> drivers/gpu/drm/vc4/vc4_hdmi.c | 17 +++++++++++++++--
> 1 file changed, 15 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/vc4/vc4_hdmi.c b/drivers/gpu/drm/vc4/vc4_hdmi.c
> index 3df1747dd917..28b78ea885ea 100644
> --- a/drivers/gpu/drm/vc4/vc4_hdmi.c
> +++ b/drivers/gpu/drm/vc4/vc4_hdmi.c
> @@ -1434,13 +1434,20 @@ static irqreturn_t vc4_cec_irq_handler_thread(int irq, void *priv)
>
> static void vc4_cec_read_msg(struct vc4_hdmi *vc4_hdmi, u32 cntrl1)
> {
> + struct drm_device *dev = vc4_hdmi->connector.dev;
> struct cec_msg *msg = &vc4_hdmi->cec_rx_msg;
> unsigned int i;
>
> msg->len = 1 + ((cntrl1 & VC4_HDMI_CEC_REC_WRD_CNT_MASK) >>
> VC4_HDMI_CEC_REC_WRD_CNT_SHIFT);
> +
> + if (msg->len > 16) {
> + drm_err(dev, "Attempting to read too much data (%d)\n", msg->len);
> + return;
> + }
> +
> for (i = 0; i < msg->len; i += 4) {
> - u32 val = HDMI_READ(HDMI_CEC_RX_DATA_1 + i);
> + u32 val = HDMI_READ(HDMI_CEC_RX_DATA_1 + (i >> 2));
>
> msg->msg[i] = val & 0xff;
> msg->msg[i + 1] = (val >> 8) & 0xff;
> @@ -1533,11 +1540,17 @@ static int vc4_hdmi_cec_adap_transmit(struct cec_adapter *adap, u8 attempts,
> u32 signal_free_time, struct cec_msg *msg)
> {
> struct vc4_hdmi *vc4_hdmi = cec_get_drvdata(adap);
> + struct drm_device *dev = vc4_hdmi->connector.dev;
> u32 val;
> unsigned int i;
>
> + if (msg->len > 16) {
> + drm_err(dev, "Attempting to transmit too much data (%d)\n", msg->len);
> + return -ENOMEM;
> + }
> +
> for (i = 0; i < msg->len; i += 4)
> - HDMI_WRITE(HDMI_CEC_TX_DATA_1 + i,
> + HDMI_WRITE(HDMI_CEC_TX_DATA_1 + (i >> 2),
> (msg->msg[i]) |
> (msg->msg[i + 1] << 8) |
> (msg->msg[i + 2] << 16) |
> --
> 2.28.0
>

2020-12-17 10:51:52

[permalink] [raw]

Subject: Re: [PATCH 05/15] drm/vc4: hdmi: Restore cec physical address on reconnect

Hi Dave,

Thanks for your review

On Fri, Dec 18, 2020 at 02:45:54PM +0000, Dave Stevenson wrote:
> On Fri, 18 Dec 2020 at 14:21, Dave Stevenson
> <[email protected]> wrote:
> >
> > Hi Maxime & Dom
> >
> > On Thu, 10 Dec 2020 at 13:47, Maxime Ripard <[email protected]> wrote:
> > >
> > > From: Dom Cobley <[email protected]>
> > >
> > > Currently we call cec_phys_addr_invalidate on a hotplug deassert.
> > > That may be due to a TV power cycling, or an AVR being switched
> > > on (and switching edid).
> > >
> > > This makes CEC unusable since our controller wouldn't have a physical
> > > address anymore.
> > >
> > > Set it back up again on the hotplug assert.
> > >
> > > Fixes: 15b4511a4af6 ("drm/vc4: add HDMI CEC support")
> > > Signed-off-by: Dom Cobley <[email protected]>
> > > Signed-off-by: Maxime Ripard <[email protected]>
> > > ---
> > > drivers/gpu/drm/vc4/vc4_hdmi.c | 25 +++++++++++++++++--------
> > > 1 file changed, 17 insertions(+), 8 deletions(-)
> > >
> > > diff --git a/drivers/gpu/drm/vc4/vc4_hdmi.c b/drivers/gpu/drm/vc4/vc4_hdmi.c
> > > index 28b78ea885ea..eff3bac562c6 100644
> > > --- a/drivers/gpu/drm/vc4/vc4_hdmi.c
> > > +++ b/drivers/gpu/drm/vc4/vc4_hdmi.c
> > > @@ -136,20 +136,29 @@ static enum drm_connector_status
> > > vc4_hdmi_connector_detect(struct drm_connector *connector, bool force)
> > > {
> > > struct vc4_hdmi *vc4_hdmi = connector_to_vc4_hdmi(connector);
> > > + bool connected = false;
> > >
> > > if (vc4_hdmi->hpd_gpio) {
> > > if (gpio_get_value_cansleep(vc4_hdmi->hpd_gpio) ^
> > > vc4_hdmi->hpd_active_low)
> > > - return connector_status_connected;
> > > - cec_phys_addr_invalidate(vc4_hdmi->cec_adap);
> > > - return connector_status_disconnected;
> > > - }
> > > -
> > > - if (drm_probe_ddc(vc4_hdmi->ddc))
> > > - return connector_status_connected;
> > > -
> > > + connected = true;
> > > + } else if (drm_probe_ddc(vc4_hdmi->ddc))
> > > + connected = true;
> > > if (HDMI_READ(HDMI_HOTPLUG) & VC4_HDMI_HOTPLUG_CONNECTED)
> >
> > This needs to become an "else if(...".
> > It used to be that all the other paths would return, so were mutually
> > exclusive to this. Now they set a thing and keep going we need to
> > avoid reading the register should there be a HPD gpio or the ddc probe
> > succeeds.
> > Memory says that otherwise Pi3 always reports connected.
> >
> > I fixed this in a downstream patch already -
> > https://github.com/raspberrypi/linux/commit/d345caec1e9b2317b9cd7eb5b92ae453a0d3e98c
> >
> > Otherwise fine.
> >
> > Dave
> >
> > > + connected = true;
> > > + if (connected) {
> > > + if (connector->status != connector_status_connected) {
> > > + struct edid *edid = drm_get_edid(connector, vc4_hdmi->ddc);
> > > +
> > > + if (edid) {
> > > + cec_s_phys_addr_from_edid(vc4_hdmi->cec_adap, edid);
> > > + vc4_hdmi->encoder.hdmi_monitor = drm_detect_hdmi_monitor(edid);
> > > + drm_connector_update_edid_property(connector, edid);
>
> Actually looking at this again in the context of the other changes, do
> we need to call drm_connector_update_edid_property() here?
>
> We've just called drm_get_edid() to get the edid, and that calls
> drm_connector_update_edid_property() as well [1]

Yeah, you're right I'll drop it

> Updating vc4_hdmi->encoder.hdmi_monitor may be necessary. It's
> otherwise done in vc4_hdmi_connector_get_modes, which I sort of expect
> to be called almost immediately by the framework when connector_detect
> returns "connected". I haven't checked if that is guaranteed though.
>
> vc4_hdmi_connector_get_modes also includes a manual call to
> drm_connector_update_edid_property after having just called
> drm_get_edid, so that one feels redundant too.

.get_modes is called in drm_helper_probe_single_connector_modes, which
is usually the helper set in .fill_modes. .fill_modes seems to only be
called when either DRM_IOCTL_MODE_GETCONNECTOR is called, or when the
connector status is forced through sysfs, so it doesn't look like it's
done automatically.

I'm not sure we need to set hdmi_monitor though, it's only used to
configure the display related side, and that can't happen without
get_modes being called.

Maxime

Attachments:

(No filename) (4.51 kB)
signature.asc (235.00 B)
Download all attachments