[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <jq3tsnri2ehb37ueb66he5lwifxe4kq676jzcpa4fpk66sol6w@buxjhhwga73z>
Date: Mon, 9 Jun 2025 19:57:45 +0300
From: Dmitry Baryshkov <dmitry.baryshkov@....qualcomm.com>
To: Yongbang Shi <shiyongbang@...wei.com>
Cc: xinliang.liu@...aro.org, tiantao6@...ilicon.com,
maarten.lankhorst@...ux.intel.com, mripard@...nel.org,
tzimmermann@...e.de, airlied@...il.com, daniel@...ll.ch,
kong.kongxinwei@...ilicon.com, liangjian010@...wei.com,
chenjianmin@...wei.com, lidongming5@...wei.com, libaihan@...wei.com,
shenjian15@...wei.com, shaojijie@...wei.com,
jani.nikula@...ux.intel.com, dri-devel@...ts.freedesktop.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH drm-dp 02/10] drm/hisilicon/hibmc: fix dp probabilistical
detect errors after HPD irq
On Mon, Jun 09, 2025 at 10:47:50PM +0800, Yongbang Shi wrote:
>
> > On Fri, May 30, 2025 at 05:54:24PM +0800, Yongbang Shi wrote:
> > > From: Baihan Li <libaihan@...wei.com>
> > >
> > > The debouncing when HPD pulled out still remains sometimes, 200ms still can
> > > not ensure helper_detect() is correct. So add a flag to hold the sink
> > > status, and changed detect_ctx() functions by using flag to check status.
> > >
> > > Fixes: 3c7623fb5bb6 ("drm/hisilicon/hibmc: Enable this hot plug detect of irq feature")
> > > Signed-off-by: Baihan Li <libaihan@...wei.com>
> > > ---
> > > drivers/gpu/drm/hisilicon/hibmc/dp/dp_hw.h | 1 +
> > > .../gpu/drm/hisilicon/hibmc/hibmc_drm_dp.c | 38 +++++++++++++------
> > > 2 files changed, 28 insertions(+), 11 deletions(-)
> > >
> > > diff --git a/drivers/gpu/drm/hisilicon/hibmc/dp/dp_hw.h b/drivers/gpu/drm/hisilicon/hibmc/dp/dp_hw.h
> > > index 665f5b166dfb..68867475508c 100644
> > > --- a/drivers/gpu/drm/hisilicon/hibmc/dp/dp_hw.h
> > > +++ b/drivers/gpu/drm/hisilicon/hibmc/dp/dp_hw.h
> > > @@ -50,6 +50,7 @@ struct hibmc_dp {
> > > struct drm_dp_aux aux;
> > > struct hibmc_dp_cbar_cfg cfg;
> > > u32 irq_status;
> > > + int hpd_status;
> > > };
> > > int hibmc_dp_hw_init(struct hibmc_dp *dp);
> > > diff --git a/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_dp.c b/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_dp.c
> > > index d06832e62e96..191fb434baa7 100644
> > > --- a/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_dp.c
> > > +++ b/drivers/gpu/drm/hisilicon/hibmc/hibmc_drm_dp.c
> > > @@ -13,7 +13,8 @@
> > > #include "hibmc_drm_drv.h"
> > > #include "dp/dp_hw.h"
> > > -#define DP_MASKED_SINK_HPD_PLUG_INT BIT(2)
> > > +#define HIBMC_DP_MASKED_SINK_HPD_PLUG_INT BIT(2)
> > > +#define HIBMC_DP_MASKED_SINK_HPD_UNPLUG_INT BIT(3)
> > > static int hibmc_dp_connector_get_modes(struct drm_connector *connector)
> > > {
> > > @@ -34,9 +35,12 @@ static int hibmc_dp_connector_get_modes(struct drm_connector *connector)
> > > static int hibmc_dp_detect(struct drm_connector *connector,
> > > struct drm_modeset_acquire_ctx *ctx, bool force)
> > > {
> > > - mdelay(200);
> > > + struct hibmc_dp *dp = to_hibmc_dp(connector);
> > > - return drm_connector_helper_detect_from_ddc(connector, ctx, force);
> > > + if (dp->hpd_status)
> > > + return connector_status_connected;
> > > + else
> > > + return connector_status_disconnected;
> > > }
> > > static const struct drm_connector_helper_funcs hibmc_dp_conn_helper_funcs = {
> > > @@ -115,22 +119,34 @@ irqreturn_t hibmc_dp_hpd_isr(int irq, void *arg)
> > > {
> > > struct drm_device *dev = (struct drm_device *)arg;
> > > struct hibmc_drm_private *priv = to_hibmc_drm_private(dev);
> > > + struct hibmc_dp *dp = &priv->dp;
> > > int idx;
> > > if (!drm_dev_enter(dev, &idx))
> > > return -ENODEV;
> > > - if (priv->dp.irq_status & DP_MASKED_SINK_HPD_PLUG_INT) {
> > > - drm_dbg_dp(&priv->dev, "HPD IN isr occur!\n");
> > > - hibmc_dp_hpd_cfg(&priv->dp);
> > > + if (dp->hpd_status) { /* only check unplug int when the last status is HPD in */
> > I think this way you'll ignore HPD short pulses. Could you possibly
> > clarify whether it is the case or not?
>
> We actually doesn't enable short HPD here, this feature just used in our electrical tests.
I don't think HPD pulse needs to be enabled specially. It is documented
as IRQ that Source needs to respond to by rereading DPCD Link / Sink
Status bits and acting accordingly.
--
With best wishes
Dmitry
Powered by blists - more mailing lists