Skip to contentSkip to conversation

drm: add helper extracting SADs from EDID

RFC PATCH Superseded Rafał Miłecki 4 participants 8 messages Apr 7, 2013

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

RFC PATCH 1/1 drm: add helper extracting SADs from EDID

Some devices (ATI/AMD cards) don't want passing ELD struct to the hardware but just require filling specific registers and then hardware/firmware does the rest. In such a cases we need to read info from SAD blocks and put them in the correct registers.

---
 drivers/gpu/drm/drm_edid.c |   55 ++++++++++++++++++++++++++++++++++++++++++++
 include/drm/drm_edid.h     |   24 +++++++++++++++++++
 2 files changed, 79 insertions(+)
drivers/gpu/drm/drm_edid.c +55 -0
index e2acfdb..7c9e799 100644
@@ -2511,6 +2511,61 @@ void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid)
 EXPORT_SYMBOL(drm_edid_to_eld);
 
 /**
+ * drm_edid_to_sad - extracts SADs from EDID
+ * @edid: EDID to parse
+ *
+ * Looks for CEA EDID block and extracts SADs (Short Audio Descriptors) from it.
+ */
+struct cea_sad *drm_edid_to_sad(struct edid *edid)
+{
+	struct cea_sad *sads = NULL;
+	int i, start, end, dbl;
+	u8 *db, *cea;
+
+	cea = drm_find_cea_extension(edid);
+	if (!cea) {
+		DRM_DEBUG_KMS("SAD: no CEA Extension found\n");
+		return NULL;
+	}
+
+	if (cea_revision(cea) < 3) {
+		DRM_DEBUG_KMS("SAD: wrong CEA revision\n");
+		return NULL;
+	}
+
+	if (cea_db_offsets(cea, &start, &end)) {
+		DRM_DEBUG_KMS("SAD: invalid data block offsets\n");
+		return NULL;
+	}
+
+	for_each_cea_db(cea, i, start, end) {
+		db = &cea[i];
+		dbl = cea_db_payload_len(db);
+
+		if (cea_db_tag(db) == AUDIO_BLOCK) {
+			int count = dbl / 3; /* SAD is 3B */
+			int j;
+
+			sads = kzalloc(count + 1 * sizeof(*sads), GFP_KERNEL);
+			if (!sads)
+				return NULL;
+
+			for (j = 0; j < count; j++) {
+				u8 *sad = &db[1 + j * 3];
+
+				sads[j].format = (sad[0] & 0x78) >> 3;
+				sads[j].channels = sad[0] & 0x7;
+				sads[j].freq = sad[1] & 0x7F;
+				sads[j].byte2 = sad[2];
+			}
+		}
+	}
+
+	return sads;
+}
+EXPORT_SYMBOL(drm_edid_to_sad);
+
+/**
  * drm_av_sync_delay - HDMI/DP sink audio-video sync delay in millisecond
  * @connector: connector associated with the HDMI/DP sink
  * @mode: the display mode
include/drm/drm_edid.h +24 -0
index 5da1b4a..b36d052 100644
@@ -244,12 +244,36 @@ struct edid {
 
 #define EDID_PRODUCT_ID(e) ((e)->prod_code[0] | ((e)->prod_code[1] << 8))
 
+#define SAD_FORMAT_LPCM			0x01
+#define SAD_FORMAT_AC3			0x02
+#define SAD_FORMAT_MPEG1		0x03
+#define SAD_FORMAT_MP3			0x04
+#define SAD_FORMAT_MPEG2		0x05
+#define SAD_FORMAT_AAC			0x06
+#define SAD_FORMAT_DTS			0x07
+#define SAD_FORMAT_ATRAC		0x08
+#define SAD_FORMAT_ONE_BIT_AUDIO	0x09
+#define SAD_FORMAT_DOLBY_DIGITAL	0x0a
+#define SAD_FORMAT_DTS_HD		0x0b
+#define SAD_FORMAT_MAT_MLP		0x0c
+#define SAD_FORMAT_DST			0x0d
+#define SAD_FORMAT_WMA_PRO		0x0e
+
+/* Short Audio Descriptor */
+struct cea_sad {
+	u8 format;
+	u8 channels; /* max number of channels - 1 */
+	u8 freq;
+	u8 byte2; /* meaning depends on format */
+};
+
 struct drm_encoder;
 struct drm_connector;
 struct drm_display_mode;
 struct hdmi_avi_infoframe;
 
 void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid);
+struct cea_sad *drm_edid_to_sad(struct edid *edid);
 int drm_av_sync_delay(struct drm_connector *connector,
 		      struct drm_display_mode *mode);
 struct drm_connector *drm_select_eld(struct drm_encoder *encoder,
Signature · 1 line
1.7.10.4
#1 lore raw

Hi,

will there be a followup patch where this function is actually used?

Best regards,
Zoltán Böszörményi

2013-04-07 12:52 keltezéssel, Rafał Miłecki írta:

Quoted 106 lines

Some devices (ATI/AMD cards) don't want passing ELD struct to the hardware but just require filling specific registers and then hardware/firmware does the rest. In such a cases we need to read info from SAD blocks and put them in the correct registers.

---
  drivers/gpu/drm/drm_edid.c |   55 ++++++++++++++++++++++++++++++++++++++++++++
  include/drm/drm_edid.h     |   24 +++++++++++++++++++
  2 files changed, 79 insertions(+)
drivers/gpu/drm/drm_edid.c +55 -0
index e2acfdb..7c9e799 100644
@@ -2511,6 +2511,61 @@ void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid)
  EXPORT_SYMBOL(drm_edid_to_eld);
 
  /**
+ * drm_edid_to_sad - extracts SADs from EDID
+ * @edid: EDID to parse
+ *
+ * Looks for CEA EDID block and extracts SADs (Short Audio Descriptors) from it.
+ */
+struct cea_sad *drm_edid_to_sad(struct edid *edid)
+{
+	struct cea_sad *sads = NULL;
+	int i, start, end, dbl;
+	u8 *db, *cea;
+
+	cea = drm_find_cea_extension(edid);
+	if (!cea) {
+		DRM_DEBUG_KMS("SAD: no CEA Extension found\n");
+		return NULL;
+	}
+
+	if (cea_revision(cea) < 3) {
+		DRM_DEBUG_KMS("SAD: wrong CEA revision\n");
+		return NULL;
+	}
+
+	if (cea_db_offsets(cea, &start, &end)) {
+		DRM_DEBUG_KMS("SAD: invalid data block offsets\n");
+		return NULL;
+	}
+
+	for_each_cea_db(cea, i, start, end) {
+		db = &cea[i];
+		dbl = cea_db_payload_len(db);
+
+		if (cea_db_tag(db) == AUDIO_BLOCK) {
+			int count = dbl / 3; /* SAD is 3B */
+			int j;
+
+			sads = kzalloc(count + 1 * sizeof(*sads), GFP_KERNEL);
+			if (!sads)
+				return NULL;
+
+			for (j = 0; j < count; j++) {
+				u8 *sad = &db[1 + j * 3];
+
+				sads[j].format = (sad[0] & 0x78) >> 3;
+				sads[j].channels = sad[0] & 0x7;
+				sads[j].freq = sad[1] & 0x7F;
+				sads[j].byte2 = sad[2];
+			}
+		}
+	}
+
+	return sads;
+}
+EXPORT_SYMBOL(drm_edid_to_sad);
+
+/**
   * drm_av_sync_delay - HDMI/DP sink audio-video sync delay in millisecond
   * @connector: connector associated with the HDMI/DP sink
   * @mode: the display mode
include/drm/drm_edid.h +23 -0
index 5da1b4a..b36d052 100644
@@ -244,12 +244,36 @@ struct edid {
 
  #define EDID_PRODUCT_ID(e) ((e)->prod_code[0] | ((e)->prod_code[1] << 8))
 
+#define SAD_FORMAT_LPCM			0x01
+#define SAD_FORMAT_AC3			0x02
+#define SAD_FORMAT_MPEG1		0x03
+#define SAD_FORMAT_MP3			0x04
+#define SAD_FORMAT_MPEG2		0x05
+#define SAD_FORMAT_AAC			0x06
+#define SAD_FORMAT_DTS			0x07
+#define SAD_FORMAT_ATRAC		0x08
+#define SAD_FORMAT_ONE_BIT_AUDIO	0x09
+#define SAD_FORMAT_DOLBY_DIGITAL	0x0a
+#define SAD_FORMAT_DTS_HD		0x0b
+#define SAD_FORMAT_MAT_MLP		0x0c
+#define SAD_FORMAT_DST			0x0d
+#define SAD_FORMAT_WMA_PRO		0x0e
+
+/* Short Audio Descriptor */
+struct cea_sad {
+	u8 format;
+	u8 channels; /* max number of channels - 1 */
+	u8 freq;
+	u8 byte2; /* meaning depends on format */
+};
+

include/drm/drm_edid.h struct edid { · line 279

Quoted 10 lines
include/drm/drm_edid.h +1 -0
  struct drm_encoder;
  struct drm_connector;
  struct drm_display_mode;
  struct hdmi_avi_infoframe;
 
  void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid);
+struct cea_sad *drm_edid_to_sad(struct edid *edid);
  int drm_av_sync_delay(struct drm_connector *connector,
  		      struct drm_display_mode *mode);
  struct drm_connector *drm_select_eld(struct drm_encoder *encoder,

_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel

#2 lore raw

2013/4/7 Boszormenyi Zoltan <[email protected]>:

Quoted 1 line

will there be a followup patch where this function is actually used?

Sure :) It's just early patch post, with RFC, to see if I'm doing it the right way.

Signature · 5 lines
Rafał
_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel
#3 lore raw
In reply to Rafał Miłecki

2013/4/7 Rafał Miłecki <[email protected]>:

Quoted 4 lines

Some devices (ATI/AMD cards) don't want passing ELD struct to the hardware but just require filling specific registers and then hardware/firmware does the rest. In such a cases we need to read info from SAD blocks and put them in the correct registers.

If you wish to see how exactly it works on radeon, see for example evergreend.h:
AZ_F0_CODEC_PIN0_CONTROL_AUDIO_DESCRIPTOR0 AZ_F0_CODEC_PIN0_CONTROL_AUDIO_DESCRIPTOR1
& friends.

Signature · 5 lines
Rafał
_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel
#4 lore raw
In reply to Rafał Miłecki

Am Sonntag, den 07.04.2013, 12:52 +0200 schrieb Rafał Miłecki:

Quoted 1 line

Some devices (ATI/AMD cards) don't want passing ELD struct to the

want to pass

Quoted 1 line

hardware but just require filling specific registers and then

at end: »then *the* hardware«

(I think)

Quoted 1 line

hardware/firmware does the rest. In such a cases we need to read info

• »In such cases« (without »a«)
• the info

Quoted 1 line

from SAD blocks and put them in the correct registers.

Maybe mention the function name in the commit message (even summary) too.

Quoted 11 lines
---
 drivers/gpu/drm/drm_edid.c |   55 ++++++++++++++++++++++++++++++++++++++++++++
 include/drm/drm_edid.h     |   24 +++++++++++++++++++
 2 files changed, 79 insertions(+)
drivers/gpu/drm/drm_edid.c +0 -0
index e2acfdb..7c9e799 100644
@@ -2511,6 +2511,61 @@ void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid)
 EXPORT_SYMBOL(drm_edid_to_eld);

drivers/gpu/drm/drm_edid.c void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid) · line 2521

Quoted 10 lines
drivers/gpu/drm/drm_edid.c +8 -0
 
 /**
+ * drm_edid_to_sad - extracts SADs from EDID
+ * @edid: EDID to parse
+ *
+ * Looks for CEA EDID block and extracts SADs (Short Audio Descriptors) from it.
+ */
+struct cea_sad *drm_edid_to_sad(struct edid *edid)
+{
+	struct cea_sad *sads = NULL;

No need to set it NULL, as it gets assigned later on? Looks like in the end there is a corner case, where nothing gets assigned in the for_each_cea_db loop. So the compiler is going to complain.

Quoted 8 lines
drivers/gpu/drm/drm_edid.c +8 -0
+	int i, start, end, dbl;
+	u8 *db, *cea;
+
+	cea = drm_find_cea_extension(edid);
+	if (!cea) {
+		DRM_DEBUG_KMS("SAD: no CEA Extension found\n");
+		return NULL;
+	}

drivers/gpu/drm/drm_edid.c void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid) · line 2539

Quoted 10 lines
drivers/gpu/drm/drm_edid.c +10 -0
+
+	if (cea_revision(cea) < 3) {
+		DRM_DEBUG_KMS("SAD: wrong CEA revision\n");
+		return NULL;
+	}
+
+	if (cea_db_offsets(cea, &start, &end)) {
+		DRM_DEBUG_KMS("SAD: invalid data block offsets\n");
+		return NULL;
+	}

Could the above be put in some helper too:
is_cea_valid_for_version(cea, version)?

drivers/gpu/drm/drm_edid.c void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid) · line 2550

Quoted 10 lines
drivers/gpu/drm/drm_edid.c +10 -0
+	for_each_cea_db(cea, i, start, end) {
+		db = &cea[i];
+		dbl = cea_db_payload_len(db);
+
+		if (cea_db_tag(db) == AUDIO_BLOCK) {
+			int count = dbl / 3; /* SAD is 3B */
+			int j;
+
+			sads = kzalloc(count + 1 * sizeof(*sads), GFP_KERNEL);
+			if (!sads)

Add an error too? »kzallac failed« or something like this?

Quoted 26 lines
drivers/gpu/drm/drm_edid.c +18 -0
+				return NULL;
+
+			for (j = 0; j < count; j++) {
+				u8 *sad = &db[1 + j * 3];
+
+				sads[j].format = (sad[0] & 0x78) >> 3;
+				sads[j].channels = sad[0] & 0x7;
+				sads[j].freq = sad[1] & 0x7F;
+				sads[j].byte2 = sad[2];
+			}
+		}
+	}
+
+	return sads;
+}
+EXPORT_SYMBOL(drm_edid_to_sad);
+
+/**
  * drm_av_sync_delay - HDMI/DP sink audio-video sync delay in millisecond
  * @connector: connector associated with the HDMI/DP sink
  * @mode: the display mode
include/drm/drm_edid.h +0 -0
index 5da1b4a..b36d052 100644

include/drm/drm_edid.h struct edid { · line 248

Quoted 5 lines
include/drm/drm_edid.h +2 -0
@@ -244,12 +244,36 @@ struct edid {
 
 #define EDID_PRODUCT_ID(e) ((e)->prod_code[0] | ((e)->prod_code[1] << 8))
 
+#define SAD_FORMAT_LPCM			0x01
+#define SAD_FORMAT_AC3			0x02

At least in my MUA indentation looks different.

Quoted 21 lines
include/drm/drm_edid.h +21 -0
+#define SAD_FORMAT_MPEG1		0x03
+#define SAD_FORMAT_MP3			0x04
+#define SAD_FORMAT_MPEG2		0x05
+#define SAD_FORMAT_AAC			0x06
+#define SAD_FORMAT_DTS			0x07
+#define SAD_FORMAT_ATRAC		0x08
+#define SAD_FORMAT_ONE_BIT_AUDIO	0x09
+#define SAD_FORMAT_DOLBY_DIGITAL	0x0a
+#define SAD_FORMAT_DTS_HD		0x0b
+#define SAD_FORMAT_MAT_MLP		0x0c
+#define SAD_FORMAT_DST			0x0d
+#define SAD_FORMAT_WMA_PRO		0x0e
+
+/* Short Audio Descriptor */
+struct cea_sad {
+	u8 format;
+	u8 channels; /* max number of channels - 1 */
+	u8 freq;
+	u8 byte2; /* meaning depends on format */
+};
+

include/drm/drm_edid.h struct edid { · line 279

Quoted 10 lines
include/drm/drm_edid.h +1 -0
 struct drm_encoder;
 struct drm_connector;
 struct drm_display_mode;
 struct hdmi_avi_infoframe;
 
 void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid);
+struct cea_sad *drm_edid_to_sad(struct edid *edid);
 int drm_av_sync_delay(struct drm_connector *connector,
 		      struct drm_display_mode *mode);
 struct drm_connector *drm_select_eld(struct drm_encoder *encoder,

Otherwise this looks good.

Thanks,

Paul

_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel

#5 lore raw

Thanks for comments!

2013/4/7 Paul Menzel <[email protected]>:

Quoted 8 lines

Am Sonntag, den 07.04.2013, 12:52 +0200 schrieb Rafał Miłecki:

drivers/gpu/drm/drm_edid.c void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid) · line 2521

Quoted 3 lines
drivers/gpu/drm/drm_edid.c +3 -0
+struct cea_sad *drm_edid_to_sad(struct edid *edid)
+{
+     struct cea_sad *sads = NULL;

No need to set it NULL, as it gets assigned later on? Looks like in the end there is a corner case, where nothing gets assigned in the for_each_cea_db loop. So the compiler is going to complain.

Right, I can drop that. Curious, my gcc didn't complain.

Quoted 10 lines

drivers/gpu/drm/drm_edid.c void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid) · line 2539

Quoted 9 lines
drivers/gpu/drm/drm_edid.c +9 -0
+     if (cea_revision(cea) < 3) {
+             DRM_DEBUG_KMS("SAD: wrong CEA revision\n");
+             return NULL;
+     }
+
+     if (cea_db_offsets(cea, &start, &end)) {
+             DRM_DEBUG_KMS("SAD: invalid data block offsets\n");
+             return NULL;
+     }
Quoted 2 lines

Could the above be put in some helper too:
is_cea_valid_for_version(cea, version)?

Not sure if it's worth a helper. But maybe I'll go and just use a single check? if (cea_revision(cea) < 3 || cea_db_offsets(...))

Quoted 4 lines
Quoted 4 lines
drivers/gpu/drm/drm_edid.c +4 -0
+     for_each_cea_db(cea, i, start, end) {
+             db = &cea[i];
+             dbl = cea_db_payload_len(db);
+
Quoted 8 lines

drivers/gpu/drm/drm_edid.c void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid) · line 2550

Quoted 6 lines
drivers/gpu/drm/drm_edid.c +6 -0
+             if (cea_db_tag(db) == AUDIO_BLOCK) {
+                     int count = dbl / 3; /* SAD is 3B */
+                     int j;
+
+                     sads = kzalloc(count + 1 * sizeof(*sads), GFP_KERNEL);
+                     if (!sads)

Add an error too? »kzallac failed« or something like this?

Allocation failures are aloud enough on the lower level. I was corrected few times already to don't put warnings when alloc fails. If you want me to point to the discussions, I'll have to search though.

Quoted 4 lines

include/drm/drm_edid.h struct edid { · line 248

Quoted 2 lines
include/drm/drm_edid.h +2 -0
+#define SAD_FORMAT_LPCM                      0x01
+#define SAD_FORMAT_AC3                       0x02

At least in my MUA indentation looks different.

It's just a matter of .patch format. Patch file has "+" sign at a beginning of line and \t can look different when viewing patch. Please apply a patch and check drm_edid.h then - it'll be OK.

Signature · 5 lines
Rafał
_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel
#6 lore raw

Am Sonntag, den 07.04.2013, 14:11 +0200 schrieb Rafał Miłecki:

Quoted 12 lines

Thanks for comments!

2013/4/7 Paul Menzel <[email protected]>:

Quoted 8 lines

Am Sonntag, den 07.04.2013, 12:52 +0200 schrieb Rafał Miłecki:

drivers/gpu/drm/drm_edid.c void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid) · line 2521

Quoted 3 lines
drivers/gpu/drm/drm_edid.c +3 -0
+struct cea_sad *drm_edid_to_sad(struct edid *edid)
+{
+     struct cea_sad *sads = NULL;

No need to set it NULL, as it gets assigned later on? Looks like in the end there is a corner case, where nothing gets assigned in the for_each_cea_db loop. So the compiler is going to complain.

Quoted 1 line

Right, I can drop that. Curious, my gcc didn't complain.

Oh, I did not check with GCC. But there is that corner case, if I am not mistaken.

Quoted 10 lines
Quoted 10 lines
Quoted 9 lines
drivers/gpu/drm/drm_edid.c +9 -0
+     if (cea_revision(cea) < 3) {
+             DRM_DEBUG_KMS("SAD: wrong CEA revision\n");
+             return NULL;
+     }
+
+     if (cea_db_offsets(cea, &start, &end)) {
+             DRM_DEBUG_KMS("SAD: invalid data block offsets\n");
+             return NULL;
+     }
Quoted 5 lines
Quoted 2 lines

Could the above be put in some helper too:
is_cea_valid_for_version(cea, version)?

Not sure if it's worth a helper. But maybe I'll go and just use a single check? if (cea_revision(cea) < 3 || cea_db_offsets(...))

I kind of like the useful error messages, you put there. So if no helper, then I’d prefer if you leave it as is.

Quoted 13 lines
Quoted 4 lines
Quoted 4 lines
drivers/gpu/drm/drm_edid.c +4 -0
+     for_each_cea_db(cea, i, start, end) {
+             db = &cea[i];
+             dbl = cea_db_payload_len(db);
+
Quoted 8 lines

drivers/gpu/drm/drm_edid.c void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid) · line 2550

Quoted 6 lines
drivers/gpu/drm/drm_edid.c +6 -0
+             if (cea_db_tag(db) == AUDIO_BLOCK) {
+                     int count = dbl / 3; /* SAD is 3B */
+                     int j;
+
+                     sads = kzalloc(count + 1 * sizeof(*sads), GFP_KERNEL);
+                     if (!sads)

Add an error too? »kzallac failed« or something like this?

Quoted 3 lines

Allocation failures are aloud enough on the lower level. I was corrected few times already to don't put warnings when alloc fails. If you want me to point to the discussions, I'll have to search though.

No need. I believe you.

Quoted 8 lines
Quoted 4 lines

include/drm/drm_edid.h struct edid { · line 248

Quoted 2 lines
include/drm/drm_edid.h +2 -0
+#define SAD_FORMAT_LPCM                      0x01
+#define SAD_FORMAT_AC3                       0x02

At least in my MUA indentation looks different.

It's just a matter of .patch format. Patch file has "+" sign at a beginning of line and \t can look different when viewing patch. Please apply a patch and check drm_edid.h then - it'll be OK.

Thanks. Sorry for not checking that myself.

Thanks,

paul

_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel

#7 lore raw
In reply to Rafał Miłecki
Quoted 14 lines · Rafał Miłecki

Some devices (ATI/AMD cards) don't want passing ELD struct to the hardware but just require filling specific registers and then hardware/firmware does the rest. In such a cases we need to read info from SAD blocks and put them in the correct registers.

---
 drivers/gpu/drm/drm_edid.c |   55 ++++++++++++++++++++++++++++++++++++++++++++
 include/drm/drm_edid.h     |   24 +++++++++++++++++++
 2 files changed, 79 insertions(+)
drivers/gpu/drm/drm_edid.c +0 -0
index e2acfdb..7c9e799 100644

drivers/gpu/drm/drm_edid.c void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid) · line 2517

Quoted 7 lines
drivers/gpu/drm/drm_edid.c +4 -0
@@ -2511,6 +2511,61 @@ void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid)
 EXPORT_SYMBOL(drm_edid_to_eld);
 
 /**
+ * drm_edid_to_sad - extracts SADs from EDID
+ * @edid: EDID to parse
+ *
+ * Looks for CEA EDID block and extracts SADs (Short Audio Descriptors) from it.

The function returns an array of SADs with the zeroed SAD as a terminator. That kind of magic needs clear documentation IMHO.

In fact I think it would be better if you returned the number of SADs too, instead of relying on the zero termination. With a clear "number of SADs" return parameter it's less likely people would misuse this by accident.

Quoted 22 lines
drivers/gpu/drm/drm_edid.c +22 -0
+ */
+struct cea_sad *drm_edid_to_sad(struct edid *edid)
+{
+	struct cea_sad *sads = NULL;
+	int i, start, end, dbl;
+	u8 *db, *cea;
+
+	cea = drm_find_cea_extension(edid);
+	if (!cea) {
+		DRM_DEBUG_KMS("SAD: no CEA Extension found\n");
+		return NULL;
+	}
+
+	if (cea_revision(cea) < 3) {
+		DRM_DEBUG_KMS("SAD: wrong CEA revision\n");
+		return NULL;
+	}
+
+	if (cea_db_offsets(cea, &start, &end)) {
+		DRM_DEBUG_KMS("SAD: invalid data block offsets\n");
+		return NULL;
+	}

drivers/gpu/drm/drm_edid.c void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid) · line 2549

Quoted 10 lines
drivers/gpu/drm/drm_edid.c +10 -0
+
+	for_each_cea_db(cea, i, start, end) {
+		db = &cea[i];
+		dbl = cea_db_payload_len(db);
+
+		if (cea_db_tag(db) == AUDIO_BLOCK) {
+			int count = dbl / 3; /* SAD is 3B */
+			int j;
+
+			sads = kzalloc(count + 1 * sizeof(*sads), GFP_KERNEL);

Missing parens -> miscalculated size. Should really be kcalloc().

drivers/gpu/drm/drm_edid.c void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid) · line 2560

Quoted 11 lines
drivers/gpu/drm/drm_edid.c +11 -0
+			if (!sads)
+				return NULL;
+
+			for (j = 0; j < count; j++) {
+				u8 *sad = &db[1 + j * 3];
+
+				sads[j].format = (sad[0] & 0x78) >> 3;
+				sads[j].channels = sad[0] & 0x7;
+				sads[j].freq = sad[1] & 0x7F;
+				sads[j].byte2 = sad[2];
+			}

If the EDID has multiple AUDIO_BLOCKs this will leak memory. 'break' here would avoid that.

Quoted 59 lines
drivers/gpu/drm/drm_edid.c +8 -0
+		}
+	}
+
+	return sads;
+}
+EXPORT_SYMBOL(drm_edid_to_sad);
+
+/**
  * drm_av_sync_delay - HDMI/DP sink audio-video sync delay in millisecond
  * @connector: connector associated with the HDMI/DP sink
  * @mode: the display mode
include/drm/drm_edid.h +24 -0
index 5da1b4a..b36d052 100644
@@ -244,12 +244,36 @@ struct edid {
 
 #define EDID_PRODUCT_ID(e) ((e)->prod_code[0] | ((e)->prod_code[1] << 8))
 
+#define SAD_FORMAT_LPCM			0x01
+#define SAD_FORMAT_AC3			0x02
+#define SAD_FORMAT_MPEG1		0x03
+#define SAD_FORMAT_MP3			0x04
+#define SAD_FORMAT_MPEG2		0x05
+#define SAD_FORMAT_AAC			0x06
+#define SAD_FORMAT_DTS			0x07
+#define SAD_FORMAT_ATRAC		0x08
+#define SAD_FORMAT_ONE_BIT_AUDIO	0x09
+#define SAD_FORMAT_DOLBY_DIGITAL	0x0a
+#define SAD_FORMAT_DTS_HD		0x0b
+#define SAD_FORMAT_MAT_MLP		0x0c
+#define SAD_FORMAT_DST			0x0d
+#define SAD_FORMAT_WMA_PRO		0x0e
+
+/* Short Audio Descriptor */
+struct cea_sad {
+	u8 format;
+	u8 channels; /* max number of channels - 1 */
+	u8 freq;
+	u8 byte2; /* meaning depends on format */
+};
+
 struct drm_encoder;
 struct drm_connector;
 struct drm_display_mode;
 struct hdmi_avi_infoframe;
 
 void drm_edid_to_eld(struct drm_connector *connector, struct edid *edid);
+struct cea_sad *drm_edid_to_sad(struct edid *edid);
 int drm_av_sync_delay(struct drm_connector *connector,
 		      struct drm_display_mode *mode);
 struct drm_connector *drm_select_eld(struct drm_encoder *encoder,
Signature · 6 lines
1.7.10.4

_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel
Signature · 7 lines
Ville Syrjälä
[email protected]
http://www.sci.fi/~syrjala/
_______________________________________________
dri-devel mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/dri-devel