[U-Boot] [PATCH 0/3] video: Fix issues with edid parsing

[U-Boot] [PATCH 0/3] video: Fix issues with edid parsing

[Jernej Skrabec]

This series fixes some issues with edid parsing, which prevents some
board/monitors combinations to work properly. Those issues were mainly
found by using sunxi board (OrangePi 2) and DVI screen (Waveshare 10
inch, 1024x600). However, improvements should help other platforms too.

Patch 1 initializes and sets flags in timing info.

Patch 2 adds support for distinguishing HDMI and DVI monitor based on
edid data.

Patch 3 use newly added flag in dw_hdmi driver to properly select DVI or
HDMI mode.

Best regards,
Jernej


Jernej Skrabec (3):
  edid: Set timings flags according to edid
  edid: Add HDMI flag to timing info
  video: dw_hdmi: Select HDMI mode only if monitor supports it

 common/edid.c           | 56 +++++++++++++++++++++++++++++++++++++++++++++++++
 drivers/video/dw_hdmi.c | 31 ++++++++++++++-------------
 include/edid.h          | 14 +++++++++++++
 include/fdtdec.h        |  1 +
 4 files changed, 87 insertions(+), 15 deletions(-)

-- 
2.12.2

[U-Boot] [PATCH 1/3] edid: Set timings flags according to edid

[Jernej Skrabec]

Timing flags are never set, so they may contain garbage. Since some
drivers check them, video output may be broken on those drivers.

Initialize them to 0 and check for few common flags.

Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
---

 common/edid.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/common/edid.c b/common/edid.c
index e08e420920..ab7069fdcd 100644
--- a/common/edid.c
+++ b/common/edid.c
@@ -85,6 +85,7 @@ static void decode_timing(u8 *buf, struct display_timing *timing)
 	uint x_mm, y_mm;
 	unsigned int ha, hbl, hso, hspw, hborder;
 	unsigned int va, vbl, vso, vspw, vborder;
+	struct edid_detailed_timing *t = (struct edid_detailed_timing *)buf;
 
 	/* Edid contains pixel clock in terms of 10KHz */
 	set_entry(&timing->pixelclock, (buf[0] + (buf[1] << 8)) * 10000);
@@ -111,6 +112,19 @@ static void decode_timing(u8 *buf, struct display_timing *timing)
 	set_entry(&timing->vback_porch, vbl - vso - vspw);
 	set_entry(&timing->vsync_len, vspw);
 
+	timing->flags = 0;
+	if (EDID_DETAILED_TIMING_FLAG_HSYNC_POLARITY(*t))
+		timing->flags |= DISPLAY_FLAGS_HSYNC_HIGH;
+	else
+		timing->flags |= DISPLAY_FLAGS_HSYNC_LOW;
+	if (EDID_DETAILED_TIMING_FLAG_VSYNC_POLARITY(*t))
+		timing->flags |= DISPLAY_FLAGS_VSYNC_HIGH;
+	else
+		timing->flags |= DISPLAY_FLAGS_VSYNC_LOW;
+
+	if (EDID_DETAILED_TIMING_FLAG_INTERLACED(*t))
+		timing->flags = DISPLAY_FLAGS_INTERLACED;
+
 	debug("Detailed mode clock %u Hz, %d mm x %d mm\n"
 	      "               %04x %04x %04x %04x hborder %x\n"
 	      "               %04x %04x %04x %04x vborder %x\n",
-- 
2.12.2

[U-Boot] [PATCH 2/3] edid: Add HDMI flag to timing info

[Jernej Skrabec]

Some DVI monitors don't show anything in HDMI mode since audio stream
confuses them. To solve this situation, this commit adds HDMI flag in
timing data and sets it accordingly during edid parsing.

First existence of extension block is checked. If it exists and it is
CEA861 extension, then data blocks are checked for presence of HDMI
vendor specific data block. If it is present, HDMI flag is set.

Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
---

 common/edid.c    | 42 ++++++++++++++++++++++++++++++++++++++++++
 include/edid.h   | 14 ++++++++++++++
 include/fdtdec.h |  1 +
 3 files changed, 57 insertions(+)

diff --git a/common/edid.c b/common/edid.c
index ab7069fdcd..19410aa4fc 100644
--- a/common/edid.c
+++ b/common/edid.c
@@ -136,6 +136,39 @@ static void decode_timing(u8 *buf, struct display_timing *timing)
 	      va + vbl, vborder);
 }
 
+/**
+ * Check if HDMI vendor specific data block is present in CEA block
+ * @param info	CEA extension block
+ * @return true if block is found
+ */
+static bool cea_is_hdmi_vsdb_present(struct edid_cea861_info *info)
+{
+	u8 end, i = 0;
+
+	/* check for end of data block */
+	end = info->dtd_offset;
+	if (end == 0)
+		end = 127;
+	if (end < 4 || end > 127)
+		return false;
+	end -= 4;
+
+	while (i < end) {
+		/* Look for vendor specific data block of appropriate size */
+		if ((EDID_CEA861_DB_TYPE(*info, i) == EDID_CEA861_DB_VENDOR) &&
+		    (EDID_CEA861_DB_LEN(*info, i) >= 5)) {
+			u8 *db = &info->data[i + 1];
+			u32 oui = db[0] | (db[1] << 8) | (db[2] << 16);
+
+			if (oui == HDMI_IEEE_OUI)
+				return true;
+		}
+		i += EDID_CEA861_DB_LEN(*info, i) + 1;
+	}
+
+	return false;
+}
+
 int edid_get_timing(u8 *buf, int buf_size, struct display_timing *timing,
 		    int *panel_bits_per_colourp)
 {
@@ -181,6 +214,15 @@ int edid_get_timing(u8 *buf, int buf_size, struct display_timing *timing,
 			((edid->video_input_definition & 0x70) >> 3) + 4;
 	}
 
+	timing->hdmi_monitor = false;
+	if (edid->extension_flag && (buf_size >= EDID_EXT_SIZE)) {
+		struct edid_cea861_info *info =
+			(struct edid_cea861_info *)(buf + sizeof(*edid));
+
+		if (info->extension_tag == EDID_CEA861_EXTENSION_TAG)
+			timing->hdmi_monitor = cea_is_hdmi_vsdb_present(info);
+	}
+
 	return 0;
 }
 
diff --git a/include/edid.h b/include/edid.h
index 8b022fa98a..a9f2f3d3ab 100644
--- a/include/edid.h
+++ b/include/edid.h
@@ -19,6 +19,9 @@
 #define EDID_SIZE	128
 #define EDID_EXT_SIZE	256
 
+/* OUI of HDMI vendor specific data block */
+#define HDMI_IEEE_OUI 0x000c03
+
 #define GET_BIT(_x, _pos) \
 	(((_x) >> (_pos)) & 1)
 #define GET_BITS(_x, _pos_msb, _pos_lsb) \
@@ -234,6 +237,13 @@ struct edid1_info {
 	unsigned char checksum;
 } __attribute__ ((__packed__));
 
+enum edid_cea861_db_types {
+	EDID_CEA861_DB_AUDIO = 0x01,
+	EDID_CEA861_DB_VIDEO = 0x02,
+	EDID_CEA861_DB_VENDOR = 0x03,
+	EDID_CEA861_DB_SPEAKER = 0x04,
+};
+
 struct edid_cea861_info {
 	unsigned char extension_tag;
 #define EDID_CEA861_EXTENSION_TAG	0x02
@@ -251,6 +261,10 @@ struct edid_cea861_info {
 #define EDID_CEA861_DTD_COUNT(_x) \
 	GET_BITS(((_x).dtd_count), 3, 0)
 	unsigned char data[124];
+#define EDID_CEA861_DB_TYPE(_x, offset) \
+	GET_BITS((_x).data[offset], 7, 5)
+#define EDID_CEA861_DB_LEN(_x, offset) \
+	GET_BITS((_x).data[offset], 4, 0)
 } __attribute__ ((__packed__));
 
 /**
diff --git a/include/fdtdec.h b/include/fdtdec.h
index 2134701c54..340766da1b 100644
--- a/include/fdtdec.h
+++ b/include/fdtdec.h
@@ -967,6 +967,7 @@ struct display_timing {
 	struct timing_entry vsync_len;		/* ver. sync len */
 
 	enum display_flags flags;		/* display flags */
+	bool hdmi_monitor;			/* is hdmi monitor? */
 };
 
 /**
-- 
2.12.2

[U-Boot] [PATCH 3/3] video: dw_hdmi: Select HDMI mode only if monitor supports it

[Jernej Skrabec]

Some DVI monitors don't work in HDMI mode. Set it only if edid data
explicitly states that it is supported.

Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
---

 drivers/video/dw_hdmi.c | 31 ++++++++++++++++---------------
 1 file changed, 16 insertions(+), 15 deletions(-)

diff --git a/drivers/video/dw_hdmi.c b/drivers/video/dw_hdmi.c
index 8a53109823..6039d676c5 100644
--- a/drivers/video/dw_hdmi.c
+++ b/drivers/video/dw_hdmi.c
@@ -414,13 +414,9 @@ static void hdmi_av_composer(struct dw_hdmi *hdmi,
 		   HDMI_FC_INVIDCONF_DE_IN_POLARITY_ACTIVE_HIGH :
 		   HDMI_FC_INVIDCONF_DE_IN_POLARITY_ACTIVE_LOW);
 
-	/*
-	 * TODO(sjg at chromium.org>: Need to check for HDMI / DVI
-	 * inv_val |= (edid->hdmi_monitor_detected ?
-	 *	   HDMI_FC_INVIDCONF_DVI_MODEZ_HDMI_MODE :
-	 *	   HDMI_FC_INVIDCONF_DVI_MODEZ_DVI_MODE);
-	 */
-	inv_val |= HDMI_FC_INVIDCONF_DVI_MODEZ_HDMI_MODE;
+	inv_val |= (edid->hdmi_monitor ?
+		   HDMI_FC_INVIDCONF_DVI_MODEZ_HDMI_MODE :
+		   HDMI_FC_INVIDCONF_DVI_MODEZ_DVI_MODE);
 
 	inv_val |= HDMI_FC_INVIDCONF_R_V_BLANK_IN_OSC_ACTIVE_LOW;
 
@@ -459,7 +455,7 @@ static void hdmi_av_composer(struct dw_hdmi *hdmi,
 }
 
 /* hdmi initialization step b.4 */
-static void hdmi_enable_video_path(struct dw_hdmi *hdmi)
+static void hdmi_enable_video_path(struct dw_hdmi *hdmi, bool audio)
 {
 	uint clkdis;
 
@@ -484,8 +480,10 @@ static void hdmi_enable_video_path(struct dw_hdmi *hdmi)
 	clkdis &= ~HDMI_MC_CLKDIS_TMDSCLK_DISABLE;
 	hdmi_write(hdmi, clkdis, HDMI_MC_CLKDIS);
 
-	clkdis &= ~HDMI_MC_CLKDIS_AUDCLK_DISABLE;
-	hdmi_write(hdmi, clkdis, HDMI_MC_CLKDIS);
+	if (audio) {
+		clkdis &= ~HDMI_MC_CLKDIS_AUDCLK_DISABLE;
+		hdmi_write(hdmi, clkdis, HDMI_MC_CLKDIS);
+	}
 }
 
 /* workaround to clear the overflow condition */
@@ -716,7 +714,8 @@ int dw_hdmi_enable(struct dw_hdmi *hdmi, const struct display_timing *edid)
 {
 	int ret;
 
-	debug("hdmi, mode info : clock %d hdis %d vdis %d\n",
+	debug("%s, mode info : clock %d hdis %d vdis %d\n",
+	      edid->hdmi_monitor ? "hdmi" : "dvi",
 	      edid->pixelclock.typ, edid->hactive.typ, edid->vactive.typ);
 
 	hdmi_av_composer(hdmi, edid);
@@ -725,11 +724,13 @@ int dw_hdmi_enable(struct dw_hdmi *hdmi, const struct display_timing *edid)
 	if (ret)
 		return ret;
 
-	hdmi_enable_video_path(hdmi);
+	hdmi_enable_video_path(hdmi, edid->hdmi_monitor);
 
-	hdmi_audio_fifo_reset(hdmi);
-	hdmi_audio_set_format(hdmi);
-	hdmi_audio_set_samplerate(hdmi, edid->pixelclock.typ);
+	if (edid->hdmi_monitor) {
+		hdmi_audio_fifo_reset(hdmi);
+		hdmi_audio_set_format(hdmi);
+		hdmi_audio_set_samplerate(hdmi, edid->pixelclock.typ);
+	}
 
 	hdmi_video_packetize(hdmi);
 	hdmi_video_sample(hdmi);
-- 
2.12.2

[U-Boot] [PATCH 1/3] edid: Set timings flags according to edid

[Simon Glass]

On 29 April 2017 at 06:43, Jernej Skrabec <jernej.skrabec@siol.net> wrote:
> Timing flags are never set, so they may contain garbage. Since some
> drivers check them, video output may be broken on those drivers.
>
> Initialize them to 0 and check for few common flags.
>
> Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
> ---
>
>  common/edid.c | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)

Reviewed-by: Simon Glass <sjg@chromium.org>

[U-Boot] [PATCH 2/3] edid: Add HDMI flag to timing info

[Simon Glass]

On 29 April 2017 at 06:43, Jernej Skrabec <jernej.skrabec@siol.net> wrote:
> Some DVI monitors don't show anything in HDMI mode since audio stream
> confuses them. To solve this situation, this commit adds HDMI flag in
> timing data and sets it accordingly during edid parsing.
>
> First existence of extension block is checked. If it exists and it is
> CEA861 extension, then data blocks are checked for presence of HDMI
> vendor specific data block. If it is present, HDMI flag is set.
>
> Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
> ---
>
>  common/edid.c    | 42 ++++++++++++++++++++++++++++++++++++++++++
>  include/edid.h   | 14 ++++++++++++++
>  include/fdtdec.h |  1 +
>  3 files changed, 57 insertions(+)

Reviewed-by: Simon Glass <sjg@chromium.org>

[U-Boot] [PATCH 3/3] video: dw_hdmi: Select HDMI mode only if monitor supports it

[Simon Glass]

On 29 April 2017 at 06:43, Jernej Skrabec <jernej.skrabec@siol.net> wrote:
> Some DVI monitors don't work in HDMI mode. Set it only if edid data
> explicitly states that it is supported.
>
> Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
> ---
>
>  drivers/video/dw_hdmi.c | 31 ++++++++++++++++---------------
>  1 file changed, 16 insertions(+), 15 deletions(-)
>

Reviewed-by: Simon Glass <sjg@chromium.org>

[U-Boot] [PATCH 1/3] edid: Set timings flags according to edid

[Anatolij Gustschin]

On Sat, 29 Apr 2017 14:43:35 +0200
Jernej Skrabec jernej.skrabec at siol.net wrote:

> Timing flags are never set, so they may contain garbage. Since some
> drivers check them, video output may be broken on those drivers.
> 
> Initialize them to 0 and check for few common flags.
> 
> Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
> ---
> 
>  common/edid.c | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)

applied to u-boot-video/master, thanks!

--
Anatolij

[U-Boot] [PATCH 2/3] edid: Add HDMI flag to timing info

[Anatolij Gustschin]

On Sat, 29 Apr 2017 14:43:36 +0200
Jernej Skrabec jernej.skrabec at siol.net wrote:

> Some DVI monitors don't show anything in HDMI mode since audio stream
> confuses them. To solve this situation, this commit adds HDMI flag in
> timing data and sets it accordingly during edid parsing.
> 
> First existence of extension block is checked. If it exists and it is
> CEA861 extension, then data blocks are checked for presence of HDMI
> vendor specific data block. If it is present, HDMI flag is set.
> 
> Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
> ---
> 
>  common/edid.c    | 42 ++++++++++++++++++++++++++++++++++++++++++
>  include/edid.h   | 14 ++++++++++++++
>  include/fdtdec.h |  1 +
>  3 files changed, 57 insertions(+)

applied to u-boot-video/master, thanks!

--
Anatolij

[U-Boot] [PATCH 3/3] video: dw_hdmi: Select HDMI mode only if monitor supports it

[Anatolij Gustschin]

On Sat, 29 Apr 2017 14:43:37 +0200
Jernej Skrabec jernej.skrabec at siol.net wrote:

> Some DVI monitors don't work in HDMI mode. Set it only if edid data
> explicitly states that it is supported.
> 
> Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
> ---
> 
>  drivers/video/dw_hdmi.c | 31 ++++++++++++++++---------------
>  1 file changed, 16 insertions(+), 15 deletions(-)

applied to u-boot-video/master, thanks!

--
Anatolij