Skip to contentSkip to conversation

MST DSC support in drm-mst

PATCH Superseded David Francis 4 participants 15 messages Aug 22, 2019 – Aug 26, 2019

Subsystems (2)

v4 of 9 Superseded Latest is v9 →

lore
Share & tools
lore
#0 lore raw

v4 cover MST DSC support in drm-mst

Add necessary support for MST DSC.
(Display Stream COmpression over Multi-Stream Transport)

v4: Split patchset and rebase onto drm-tip

David Francis (5):
drm/dp-mst: Add PBN calculation for DSC modes drm/dp-mst: Parse FEC capability on MST ports drm/dp-mst: Add MST support to DP DPCD R/W functions drm/dp-mst: Fill branch->num_ports
drm/dp-mst: Add helpers for querying and enabling MST DSC

 drivers/gpu/drm/drm_dp_aux_dev.c      |  12 +-
 drivers/gpu/drm/drm_dp_helper.c       |  10 +-
 drivers/gpu/drm/drm_dp_mst_topology.c | 243 ++++++++++++++++++++++++++
 include/drm/drm_dp_mst_helper.h       |   8 +-
 4 files changed, 260 insertions(+), 13 deletions(-)
Signature · 6 lines
2.17.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#1 lore raw

PATCH v4 1/5 drm/dp-mst: Add PBN calculation for DSC modes

With DSC, bpp can be a multiple of 1/16, so drm_dp_calc_pbn_mode is insufficient.

Add drm_dp_calc_pbn_mode_dsc, a function which is the same as drm_dp_calc_pbn_mode, but the bpp is in units of 1/16.

---
 drivers/gpu/drm/drm_dp_mst_topology.c | 43 +++++++++++++++++++++++++++
 include/drm/drm_dp_mst_helper.h       |  2 +-
 2 files changed, 44 insertions(+), 1 deletion(-)
drivers/gpu/drm/drm_dp_mst_topology.c +43 -0
index 82add736e17d..8e2e731c35c5 100644
@@ -3591,6 +3591,49 @@ static int test_calc_pbn_mode(void)
 	return 0;
 }
 
+/**
+ * drm_dp_calc_pbn_mode_dsc() - Calculate the PBN for a mode with DSC enabled.
+ * @clock: dot clock for the mode
+ * @dsc_bpp: dsc bits per pixel x16 (e.g. dsc_bpp = 136 is 8.5 bpp)
+ *
+ * This uses the formula in the spec to calculate the PBN value for a mode,
+ * given that the mode is using DSC
+ * Returns:
+ * PBN required for this mode
+ */
+int drm_dp_calc_pbn_mode_dsc(int clock, int dsc_bpp)
+{
+	u64 kbps;
+	s64 peak_kbps;
+	u32 numerator;
+	u32 denominator;
+
+	kbps = clock * dsc_bpp;
+
+	/*
+	 * margin 5300ppm + 300ppm ~ 0.6% as per spec, factor is 1.006
+	 * The unit of 54/64Mbytes/sec is an arbitrary unit chosen based on
+	 * common multiplier to render an integer PBN for all link rate/lane
+	 * counts combinations
+	 * calculate
+	 * peak_kbps *= (1/16) bppx16 to bpp
+	 * peak_kbps *= (1006/1000)
+	 * peak_kbps *= (64/54)
+	 * peak_kbps *= 8    convert to bytes
+	 *
+	 * Divide numerator and denominator by 16 to avoid overflow
+	 */
+
+	numerator = 64 * 1006 / 16;
+	denominator = 54 * 8 * 1000 * 1000;
+
+	kbps *= numerator;
+	peak_kbps = drm_fixp_from_fraction(kbps, denominator);
+
+	return drm_fixp2int_ceil(peak_kbps);
+}
+EXPORT_SYMBOL(drm_dp_calc_pbn_mode_dsc);
+
 /* we want to kick the TX after we've ack the up/down IRQs. */
 static void drm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr)
 {
include/drm/drm_dp_mst_helper.h +1 -1
index 2ba6253ea6d3..ddb518f2157a 100644
@@ -611,7 +611,7 @@ struct edid *drm_dp_mst_get_edid(struct drm_connector *connector, struct drm_dp_
 
 
 int drm_dp_calc_pbn_mode(int clock, int bpp);
-
+int drm_dp_calc_pbn_mode_dsc(int clock, int dsc_bpp);
 
 bool drm_dp_mst_allocate_vcpi(struct drm_dp_mst_topology_mgr *mgr,
 			      struct drm_dp_mst_port *port, int pbn, int slots);
Signature · 6 lines
2.17.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#2 lore raw

Re: [PATCH v4 1/5] drm/dp-mst: Add PBN calculation for DSC modes

Agh, sorry to go back on this - I'm noticing a lot more changes we should have now that I've been able apply this locally to my tree. See below:

Quoted 56 lines · David Francis

With DSC, bpp can be a multiple of 1/16, so drm_dp_calc_pbn_mode is insufficient.

Add drm_dp_calc_pbn_mode_dsc, a function which is the same as drm_dp_calc_pbn_mode, but the bpp is in units of 1/16.

---
 drivers/gpu/drm/drm_dp_mst_topology.c | 43 +++++++++++++++++++++++++++
 include/drm/drm_dp_mst_helper.h       |  2 +-
 2 files changed, 44 insertions(+), 1 deletion(-)
drivers/gpu/drm/drm_dp_mst_topology.c +2 -0
index 82add736e17d..8e2e731c35c5 100644
@@ -3591,6 +3591,49 @@ static int test_calc_pbn_mode(void)
 	return 0;
 }
 
+/**
+ * drm_dp_calc_pbn_mode_dsc() - Calculate the PBN for a mode with DSC

enabled.
+ * @clock: dot clock for the mode
+ * @dsc_bpp: dsc bits per pixel x16 (e.g. dsc_bpp = 136 is 8.5 bpp)
+ *
+ * This uses the formula in the spec to calculate the PBN value for a mode,
+ * given that the mode is using DSC
+ * Returns:
+ * PBN required for this mode
+ */

drivers/gpu/drm/drm_dp_mst_topology.c +22 -0
+int drm_dp_calc_pbn_mode_dsc(int clock, int dsc_bpp)
+{
+	u64 kbps;
+	s64 peak_kbps;
+	u32 numerator;
+	u32 denominator;
+
+	kbps = clock * dsc_bpp;
+
+	/*
+	 * margin 5300ppm + 300ppm ~ 0.6% as per spec, factor is 1.006
+	 * The unit of 54/64Mbytes/sec is an arbitrary unit chosen based on
+	 * common multiplier to render an integer PBN for all link rate/lane
+	 * counts combinations
+	 * calculate
+	 * peak_kbps *= (1/16) bppx16 to bpp
+	 * peak_kbps *= (1006/1000)
+	 * peak_kbps *= (64/54)
+	 * peak_kbps *= 8    convert to bytes
+	 *
+	 * Divide numerator and denominator by 16 to avoid overflow
+	 */

drivers/gpu/drm/drm_dp_mst_topology.c static int test_calc_pbn_mode(void) · line 3635

Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10 -0
+
+	numerator = 64 * 1006 / 16;
+	denominator = 54 * 8 * 1000 * 1000;
+
+	kbps *= numerator;
+	peak_kbps = drm_fixp_from_fraction(kbps, denominator);
+
+	return drm_fixp2int_ceil(peak_kbps);
+}
+EXPORT_SYMBOL(drm_dp_calc_pbn_mode_dsc);

I didn't notice until now that the only difference between the two functions is

numerator = 64 * 1006 / 16;

So it doesn't seem like it's worth having two seperate functions for this, so let's just get rid of drm_dp_calc_pbn_mode_dsc() and combine it with drm_dp_calc_pbn_mode() by adding a bool parameter to specify whether or not we're calculating for dsc

Quoted 20 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1 -0
+
 /* we want to kick the TX after we've ack the up/down IRQs. */
 static void drm_dp_mst_kick_tx(struct drm_dp_mst_topology_mgr *mgr)
 {
include/drm/drm_dp_mst_helper.h +1 -1
index 2ba6253ea6d3..ddb518f2157a 100644
@@ -611,7 +611,7 @@ struct edid *drm_dp_mst_get_edid(struct drm_connector
 *connector, struct drm_dp_
 
 
 int drm_dp_calc_pbn_mode(int clock, int bpp);
-
+int drm_dp_calc_pbn_mode_dsc(int clock, int dsc_bpp);
 
 bool drm_dp_mst_allocate_vcpi(struct drm_dp_mst_topology_mgr *mgr,

struct drm_dp_mst_port *port, int pbn, int
slots);

Signature · 7 lines
Cheers,
	Lyude Paul

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#3 lore raw

PATCH v4 2/5 drm/dp-mst: Parse FEC capability on MST ports

In reply to David Francis

As of DP1.4, ENUM_PATH_RESOURCES returns a bit indicating if FEC can be supported up to that point in the MST network.

The bit is the first byte of the ENUM_PATH_RESOURCES ack reply, bottom-most bit (refer to section 2.11.9.4 of DP standard, v1.4)

That value is needed for FEC and DSC support

Store it on drm_dp_mst_port

---
 drivers/gpu/drm/drm_dp_mst_topology.c | 2 ++
 include/drm/drm_dp_mst_helper.h       | 3 +++
 2 files changed, 5 insertions(+)
drivers/gpu/drm/drm_dp_mst_topology.c +2 -0
index 8e2e731c35c5..50a044718439 100644
@@ -553,6 +553,7 @@ static bool drm_dp_sideband_parse_enum_path_resources_ack(struct drm_dp_sideband
 {
 	int idx = 1;
 	repmsg->u.path_resources.port_number = (raw->msg[idx] >> 4) & 0xf;
+	repmsg->u.path_resources.fec_capable = raw->msg[idx] & 0x1;
 	idx++;
 	if (idx > raw->curlen)
 		goto fail_len;
@@ -2183,6 +2184,7 @@ static int drm_dp_send_enum_path_resources(struct drm_dp_mst_topology_mgr *mgr,
 			DRM_DEBUG_KMS("enum path resources %d: %d %d\n", txmsg->reply.u.path_resources.port_number, txmsg->reply.u.path_resources.full_payload_bw_number,
 			       txmsg->reply.u.path_resources.avail_payload_bw_number);
 			port->available_pbn = txmsg->reply.u.path_resources.avail_payload_bw_number;
+			port->fec_capable = txmsg->reply.u.path_resources.fec_capable;
 		}
 	}
 
include/drm/drm_dp_mst_helper.h +3 -0
index ddb518f2157a..fa973773a4a7 100644
@@ -108,6 +108,8 @@ struct drm_dp_mst_port {
 	 * audio-capable.
 	 */
 	bool has_audio;
+
+	bool fec_capable;
 };
 
 /**
@@ -312,6 +314,7 @@ struct drm_dp_port_number_req {
 
 struct drm_dp_enum_path_resources_ack_reply {
 	u8 port_number;
+	bool fec_capable;
 	u16 full_payload_bw_number;
 	u16 avail_payload_bw_number;
 };
Signature · 6 lines
2.17.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#4 lore raw

PATCH v4 3/5 drm/dp-mst: Add MST support to DP DPCD R/W functions

In reply to David Francis

Instead of having drm_dp_dpcd_read/write and drm_dp_mst_dpcd_read/write as entry points into the aux code, have drm_dp_dpcd_read/write handle both.

This means that DRM drivers can make MST DPCD read/writes.

v2: Fix spacing

---
 drivers/gpu/drm/drm_dp_aux_dev.c | 12 ++----------
 drivers/gpu/drm/drm_dp_helper.c  | 10 ++++++++--
 2 files changed, 10 insertions(+), 12 deletions(-)
drivers/gpu/drm/drm_dp_aux_dev.c +2 -10
index 0cfb386754c3..418cad4f649a 100644
@@ -163,11 +163,7 @@ static ssize_t auxdev_read_iter(struct kiocb *iocb, struct iov_iter *to)
 			break;
 		}
 
-		if (aux_dev->aux->is_remote)
-			res = drm_dp_mst_dpcd_read(aux_dev->aux, pos, buf,
-						   todo);
-		else
-			res = drm_dp_dpcd_read(aux_dev->aux, pos, buf, todo);
+		res = drm_dp_dpcd_read(aux_dev->aux, pos, buf, todo);
 
 		if (res <= 0)
 			break;
@@ -215,11 +211,7 @@ static ssize_t auxdev_write_iter(struct kiocb *iocb, struct iov_iter *from)
 			break;
 		}
 
-		if (aux_dev->aux->is_remote)
-			res = drm_dp_mst_dpcd_write(aux_dev->aux, pos, buf,
-						    todo);
-		else
-			res = drm_dp_dpcd_write(aux_dev->aux, pos, buf, todo);
+		res = drm_dp_mst_dpcd_write(aux_dev->aux, pos, buf, todo);
 
 		if (res <= 0)
 			break;
drivers/gpu/drm/drm_dp_helper.c +8 -2
index ffc68d305afe..b1599760b6a3 100644
@@ -30,6 +30,7 @@
 #include <linux/seq_file.h>
 
 #include <drm/drm_dp_helper.h>
+#include <drm/drm_dp_mst_helper.h>
 #include <drm/drm_print.h>
 #include <drm/drm_vblank.h>
 
@@ -251,7 +252,7 @@ static int drm_dp_dpcd_access(struct drm_dp_aux *aux, u8 request,
 
 /**
  * drm_dp_dpcd_read() - read a series of bytes from the DPCD
- * @aux: DisplayPort AUX channel
+ * @aux: DisplayPort AUX channel (SST or MST)
  * @offset: address of the (first) register to read
  * @buffer: buffer to store the register values
  * @size: number of bytes in @buffer
@@ -268,6 +269,8 @@ ssize_t drm_dp_dpcd_read(struct drm_dp_aux *aux, unsigned int offset,
 {
 	int ret;
 
+	if (aux->is_remote)
+		return drm_dp_mst_dpcd_read(aux, offset, buffer, size);
 	/*
 	 * HP ZR24w corrupts the first DPCD access after entering power save
 	 * mode. Eg. on a read, the entire buffer will be filled with the same
@@ -296,7 +299,7 @@ EXPORT_SYMBOL(drm_dp_dpcd_read);
 
 /**
  * drm_dp_dpcd_write() - write a series of bytes to the DPCD
- * @aux: DisplayPort AUX channel
+ * @aux: DisplayPort AUX channel (SST or MST)
  * @offset: address of the (first) register to write
  * @buffer: buffer containing the values to write
  * @size: number of bytes in @buffer
@@ -313,6 +316,9 @@ ssize_t drm_dp_dpcd_write(struct drm_dp_aux *aux, unsigned int offset,
 {
 	int ret;
 
+	if (aux->is_remote)
+		return drm_dp_mst_dpcd_write(aux, offset, buffer, size);
+
 	ret = drm_dp_dpcd_access(aux, DP_AUX_NATIVE_WRITE, offset, buffer,
 				 size);
 	drm_dp_dump_access(aux, DP_AUX_NATIVE_WRITE, offset, buffer, ret);
Signature · 6 lines
2.17.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
Reviewed-by on v4 3/5
#5 lore raw

Re: [PATCH v4 3/5] drm/dp-mst: Add MST support to DP DPCD R/W functions

Mhhh, one last nitpick

Quoted 73 lines · David Francis

Instead of having drm_dp_dpcd_read/write and drm_dp_mst_dpcd_read/write as entry points into the aux code, have drm_dp_dpcd_read/write handle both.

This means that DRM drivers can make MST DPCD read/writes.

v2: Fix spacing

---
 drivers/gpu/drm/drm_dp_aux_dev.c | 12 ++----------
 drivers/gpu/drm/drm_dp_helper.c  | 10 ++++++++--
 2 files changed, 10 insertions(+), 12 deletions(-)
drivers/gpu/drm/drm_dp_aux_dev.c +1 -5
index 0cfb386754c3..418cad4f649a 100644
@@ -163,11 +163,7 @@ static ssize_t auxdev_read_iter(struct kiocb *iocb,
 struct iov_iter *to)
 			break;
 		}
 
-		if (aux_dev->aux->is_remote)
-			res = drm_dp_mst_dpcd_read(aux_dev->aux, pos, buf,
-						   todo);
-		else
-			res = drm_dp_dpcd_read(aux_dev->aux, pos, buf, todo);
+		res = drm_dp_dpcd_read(aux_dev->aux, pos, buf, todo);
 
 		if (res <= 0)
drivers/gpu/drm/drm_dp_aux_dev.c +1 -5
 			break;
@@ -215,11 +211,7 @@ static ssize_t auxdev_write_iter(struct kiocb *iocb,
 struct iov_iter *from)
 			break;
 		}
 
-		if (aux_dev->aux->is_remote)
-			res = drm_dp_mst_dpcd_write(aux_dev->aux, pos, buf,
-						    todo);
-		else
-			res = drm_dp_dpcd_write(aux_dev->aux, pos, buf, todo);
+		res = drm_dp_mst_dpcd_write(aux_dev->aux, pos, buf, todo);
 
 		if (res <= 0)
drivers/gpu/drm/drm_dp_aux_dev.c +0 -0
 			break;
drivers/gpu/drm/drm_dp_helper.c +2 -1
index ffc68d305afe..b1599760b6a3 100644
@@ -30,6 +30,7 @@
 #include <linux/seq_file.h>
 
 #include <drm/drm_dp_helper.h>
+#include <drm/drm_dp_mst_helper.h>
 #include <drm/drm_print.h>
 #include <drm/drm_vblank.h>
 
@@ -251,7 +252,7 @@ static int drm_dp_dpcd_access(struct drm_dp_aux *aux, u8
 request,
 
 /**
  * drm_dp_dpcd_read() - read a series of bytes from the DPCD
- * @aux: DisplayPort AUX channel
+ * @aux: DisplayPort AUX channel (SST or MST)
  * @offset: address of the (first) register to read
  * @buffer: buffer to store the register values

* @size: number of bytes in @buffer

drivers/gpu/drm/drm_dp_helper.c ssize_t drm_dp_dpcd_read(struct drm_dp_aux *aux, · line 274

Quoted 6 lines
drivers/gpu/drm/drm_dp_helper.c +2 -0
@@ -268,6 +269,8 @@ ssize_t drm_dp_dpcd_read(struct drm_dp_aux *aux,
 unsigned int offset,
 {
 	int ret;
 
+	if (aux->is_remote)
+		return drm_dp_mst_dpcd_read(aux, offset, buffer, size);

Let's actually replace this with

if (aux->is_remote)

	ret = drm_dp_mst_dpcd_read(aux, offset, buffer, size);
else
	ret = drm_dp_dpcd_access(aux, DP_AUX_NATIVE_WRITE, offset, buffer,
	                         size);

That way we still call drm_dp_dump_access() so that we can log the DPCD read to the kernel log.

Let's also add a space down here ⬇...

Quoted 13 lines
drivers/gpu/drm/drm_dp_helper.c +0 -0
 	/*
 	 * HP ZR24w corrupts the first DPCD access after entering power save
drivers/gpu/drm/drm_dp_helper.c +1 -1
 	 * mode. Eg. on a read, the entire buffer will be filled with the same
@@ -296,7 +299,7 @@ EXPORT_SYMBOL(drm_dp_dpcd_read);
 
 /**
  * drm_dp_dpcd_write() - write a series of bytes to the DPCD
- * @aux: DisplayPort AUX channel
+ * @aux: DisplayPort AUX channel (SST or MST)
  * @offset: address of the (first) register to write
  * @buffer: buffer containing the values to write
  * @size: number of bytes in @buffer

drivers/gpu/drm/drm_dp_helper.c ssize_t drm_dp_dpcd_write(struct drm_dp_aux *aux, · line 322

Quoted 7 lines
drivers/gpu/drm/drm_dp_helper.c +3 -0
@@ -313,6 +316,9 @@ ssize_t drm_dp_dpcd_write(struct drm_dp_aux *aux,
 unsigned int offset,
 {
 	int ret;
 
+	if (aux->is_remote)
+		return drm_dp_mst_dpcd_write(aux, offset, buffer, size);
+

...and do the same thing with drm_dp_dpcd_write() that we did with drm_dp_dpcd_read()

With those changes, this patch is:

Quoted 3 lines
drivers/gpu/drm/drm_dp_helper.c +0 -0
 	ret = drm_dp_dpcd_access(aux, DP_AUX_NATIVE_WRITE, offset, buffer,
 				 size);

drm_dp_dump_access(aux, DP_AUX_NATIVE_WRITE, offset, buffer, ret);

Signature · 7 lines
Cheers,
	Lyude Paul

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#6 lore raw

PATCH v4 4/5 drm/dp-mst: Fill branch->num_ports

In reply to David Francis

This field on drm_dp_mst_branch was never filled

Initialize it to zero when the list of ports is created. When a port is added to the list, increment num_ports, and when a port is removed from the list, decrement num_ports.

v2: remember to decrement on port removal

---
 drivers/gpu/drm/drm_dp_mst_topology.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/gpu/drm/drm_dp_mst_topology.c +3 -0
index 50a044718439..af4b5cec7c84 100644
@@ -919,6 +919,7 @@ static struct drm_dp_mst_branch *drm_dp_add_mst_branch_device(u8 lct, u8 *rad)
 	INIT_LIST_HEAD(&mstb->ports);
 	kref_init(&mstb->topology_kref);
 	kref_init(&mstb->malloc_kref);
+	mstb->num_ports = 0;
 	return mstb;
 }
 
@@ -1669,6 +1670,7 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,
 		mutex_lock(&mstb->mgr->lock);
 		drm_dp_mst_topology_get_port(port);
 		list_add(&port->next, &mstb->ports);
+		mstb->num_ports++;
 		mutex_unlock(&mstb->mgr->lock);
 	}
 
@@ -1703,6 +1705,7 @@ static void drm_dp_add_port(struct drm_dp_mst_branch *mstb,
 			/* remove it from the port list */
 			mutex_lock(&mstb->mgr->lock);
 			list_del(&port->next);
+			mstb->num_ports--;
 			mutex_unlock(&mstb->mgr->lock);
 			/* drop port list reference */
 			drm_dp_mst_topology_put_port(port);
Signature · 6 lines
2.17.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#7 lore raw

Re: [PATCH v4 4/5] drm/dp-mst: Fill branch->num_ports

Quoted 20 lines · David Francis

This field on drm_dp_mst_branch was never filled

Initialize it to zero when the list of ports is created. When a port is added to the list, increment num_ports, and when a port is removed from the list, decrement num_ports.

v2: remember to decrement on port removal

---
 drivers/gpu/drm/drm_dp_mst_topology.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/gpu/drm/drm_dp_mst_topology.c +0 -0
index 50a044718439..af4b5cec7c84 100644

drivers/gpu/drm/drm_dp_mst_topology.c static struct drm_dp_mst_branch · line 923

Quoted 5 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1 -0
@@ -919,6 +919,7 @@ static struct drm_dp_mst_branch
 *drm_dp_add_mst_branch_device(u8 lct, u8 *rad)
 	INIT_LIST_HEAD(&mstb->ports);
 	kref_init(&mstb->topology_kref);
 	kref_init(&mstb->malloc_kref);
+	mstb->num_ports = 0;

Just realized as well, we can drop the mstb->num_ports = 0; here, since we're allocating the struct with kzalloc()

Quoted 21 lines
drivers/gpu/drm/drm_dp_mst_topology.c +2 -0
 	return mstb;
 }
@@ -1669,6 +1670,7 @@ static void drm_dp_add_port(struct drm_dp_mst_branch
 *mstb,
 		mutex_lock(&mstb->mgr->lock);
 		drm_dp_mst_topology_get_port(port);
 		list_add(&port->next, &mstb->ports);
+		mstb->num_ports++;
 		mutex_unlock(&mstb->mgr->lock);
 	}
@@ -1703,6 +1705,7 @@ static void drm_dp_add_port(struct drm_dp_mst_branch
 *mstb,
 			/* remove it from the port list */
 			mutex_lock(&mstb->mgr->lock);
 			list_del(&port->next);
+			mstb->num_ports--;
 			mutex_unlock(&mstb->mgr->lock);
 			/* drop port list reference */

drm_dp_mst_topology_put_port(port);

Signature · 7 lines
Cheers,
	Lyude Paul

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#8 lore raw

PATCH v4 5/5 drm/dp-mst: Add helpers for querying and enabling MST DSC

In reply to David Francis

Add drm_dp_mst_dsc_caps_for_port and drm_dp_mst_dsc_enable, two helper functions for MST DSC

The former, given a port, returns the raw DPCD DSC caps off that port.

The latter, given a port, enables or disables DSC on that port.

In both cases, the port given as input should be a leaf of the MST tree with an attached display.

The logic for this is somewhat complicated, as DSC can be enabled in 4 different ways.

Case 1: DP-to-DP peer device
if the branch immediately upstream has
- PDT = DP_PEER_DEVICE_DP_MST_BRANCHING (2)
- DPCD rev. >= DP 1.4
- Exactly one input and one output
- The output has PDT = DP_PEER_DEVICE_SST_SINK (3)

In this case, DSC could be possible either on the endpoint or the peer device. Prefer the endpoint, which is possible if
- The endpoint has DP_DSC_DECOMPRESSION_IS_SUPPORTED bit set
- The endpoint has DP_FEC_CAPABLE bit set
- The peer device has DSC_PASSTHROUGH_CAPABILITY bit set (from DP v2.0)

Otherwise, use the peer device

Case 2: DP-to-HDMI peer device
If the output port has
- PDT = DP_PEER_DEVICE_DP_LEGACY_CONV (4)
- DPCD rev. >= DP 1.4
- LDPS = true
- MCS = false

In this case, DSC can only be attempted on the peer device (the output port)

Case 3: Virtual DP Sink (Internal Display Panel)
If the output port has
- DPCD rev. >= DP 1.4
- port_num >= 8

In this case, DSC can only be attempted on the peer device (the output port)

Case 4: Synaptix Workaround
If the output has
- link DPCD rev. >= DP 1.4
- link branch_dev_id = 0x90CC24 (Synaptix)
- There is exactly one branch device between the link and output

In this case, DSC can be attempted, but only using the *link* aux device's caps. This is a quirk.

Cc Lyude Paul <[email protected]>
Cc Wenjing Liu <[email protected]>
Cc Nikola Cornij <[email protected]>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 195 ++++++++++++++++++++++++++
 include/drm/drm_dp_mst_helper.h       |   3 +
 2 files changed, 198 insertions(+)
drivers/gpu/drm/drm_dp_mst_topology.c +195 -0
index af4b5cec7c84..00ddc54af65b 100644
@@ -4186,3 +4186,198 @@ static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux)
 {
 	i2c_del_adapter(&aux->ddc);
 }
+
+/**
+ * drm_dp_mst_is_virtual_dpcd() - Is the given port a virtual DPCD device?
+ * @port: The port to check
+ *
+ * Returns:
+ * true if the port is a virtual DPCD peer device, false otherwise
+ */
+static bool drm_dp_mst_is_virtual_dpcd(struct drm_dp_mst_port *port)
+{
+	struct drm_dp_mst_port *downstream_port;
+
+	if (!port)
+		return false;
+
+	/* Virtual DP Sink (Internal Display Panel) */
+	if (port->port_num >= 8 && port->dpcd_rev >= DP_DPCD_REV_14)
+		return true;
+
+	/* DP-to-HDMI Protocol Converter */
+	if (port->pdt == DP_PEER_DEVICE_DP_LEGACY_CONV &&
+			!port->mcs &&
+			port->ldps &&
+			port->dpcd_rev >= DP_DPCD_REV_14)
+		return true;
+
+	/* DP-to-DP */
+	if (port->pdt == DP_PEER_DEVICE_MST_BRANCHING &&
+			port->mstb &&
+			port->dpcd_rev >= DP_DPCD_REV_14 &&
+			port->mstb->num_ports == 2) {
+		list_for_each_entry(downstream_port, &port->mstb->ports, next) {
+			if (!downstream_port->input &&
+				downstream_port->pdt == DP_PEER_DEVICE_SST_SINK)
+				return true;
+		}
+	}
+
+	return false;
+}
+
+/**
+ * drm_dp_mst_is_virtual_dpcd() - Test for Synaptix DSC quirk
+ * @port: The port to check
+ *
+ * Some Synaptix MST hubs support DSC even though they do not support virtual
+ * DPCD. This is a quirk.
+ *
+ * Returns:
+ * true if the Synaptix workaround is required, false otherwise
+ */
+static bool drm_dp_mst_dsc_synaptix_workaround(struct drm_dp_mst_port *port)
+{
+	u8 data[3] = { 0 };
+	u32 dev_id;
+	struct drm_dp_aux *phys_aux;
+
+	/* The hub must be directly connected to the connector */
+	if (port->mgr->mst_primary != port->parent)
+		return false;
+
+	phys_aux = port->mgr->aux;
+	if (drm_dp_dpcd_read(phys_aux, DP_BRANCH_OUI, data, 3) < 0)
+		return false;
+	dev_id = (data[0] << 16) & (data[1] << 8) & data[3];
+	/* Synaptix device ID */
+	if (dev_id != 0x90CC24)
+		return false;
+
+	if (drm_dp_dpcd_read(phys_aux, DP_DPCD_REV, data, 1) < 0)
+		return false;
+	/* Must be DPCD rev. 1.4 or later */
+	if (data[0] < DP_DPCD_REV_14)
+		return false;
+
+	if (drm_dp_dpcd_read(&port->aux,
+			DP_DOWNSTREAMPORT_PRESENT, data, 1) < 0)
+		return false;
+	/* Must not be a VGA converter */
+	if ((data[0] & 7) == 3)
+		return false;
+
+	return true;
+}
+
+/**
+ * drm_dp_mst_dsc_aux_for_port() - Find the correct aux for DSC
+ * @port: The port to check. A leaf of the MST tree with an attached display.
+ *
+ * Depending on the situation, DSC may be enabled via the endpoint aux,
+ * the immediately upstream aux, or the connector's physical aux.
+ *
+ * Returns:
+ * NULL if DSC cannot be enabled on this port, otherwise the aux device
+ */
+struct drm_dp_aux *drm_dp_mst_dsc_aux_for_port(struct drm_dp_mst_port *port)
+{
+	u8 upstream_dsc = 0;
+	u8 endpoint_dsc = 0;
+	u8 endpoint_fec = 0;
+	struct drm_dp_mst_port *immediate_upstream_port;
+	struct drm_dp_mst_port *fec_port;
+
+	if (port && port->parent)
+		immediate_upstream_port = port->parent->port_parent;
+	else
+		immediate_upstream_port = NULL;
+
+	fec_port = immediate_upstream_port;
+	while (fec_port) {
+		if (!fec_port->fec_capable)
+			return NULL;
+
+		fec_port = fec_port->parent->port_parent;
+	}
+
+	if (immediate_upstream_port) {
+		if (drm_dp_dpcd_read(&immediate_upstream_port->aux,
+				DP_DSC_SUPPORT, &upstream_dsc, 1) < 0)
+			return NULL;
+	}
+
+	if (drm_dp_dpcd_read(&port->aux,
+			DP_DSC_SUPPORT, &endpoint_dsc, 1) < 0)
+		return NULL;
+	if (drm_dp_dpcd_read(&port->aux,
+			DP_FEC_CAPABILITY, &endpoint_fec, 1) < 0)
+		return NULL;
+
+	/* Enpoint decompression with DP-to-DP peer device */
+	if (drm_dp_mst_is_virtual_dpcd(immediate_upstream_port)
+			&& (upstream_dsc & 0x2) /* DSC passthrough */
+			&& (endpoint_fec & DP_FEC_CAPABLE)
+			&& (endpoint_dsc & DP_DSC_DECOMPRESSION_IS_SUPPORTED))
+		return &port->aux;
+
+	/* Virtual DPCD decompression with DP-to-DP peer device */
+	if (drm_dp_mst_is_virtual_dpcd(immediate_upstream_port))
+		return &immediate_upstream_port->aux;
+
+	/* Virtual DPCD decompression with DP-to-HDMI or Virtual DP Sink */
+	if (drm_dp_mst_is_virtual_dpcd(port))
+		return &port->aux;
+
+	/* Synaptix workaround */
+	if (drm_dp_mst_dsc_synaptix_workaround(port))
+		return port->mgr->aux;
+
+	return NULL;
+}
+
+/**
+ * drm_dp_mst_dsc_aux_for_port() - Retrieve the DSC capability registers
+ * @port: The port to check. A leaf of the MST tree with an attached display.
+ * @caps: Output.  A pointer to an array at least 16 bytes long
+ *
+ * Reads the DSC capability registers (DSC_SUPPORT through
+ * BITS_PER_PIXEL_INCREMENT) and store them in the given pointer. Use
+ * the correct aux for DSC on the given port.
+ *
+ * Returns:
+ * The number of bytes read on success, or a negative error code on failure
+ */
+int drm_dp_mst_dsc_caps_for_port(struct drm_dp_mst_port *port, u8 *caps)
+{
+	struct drm_dp_aux *aux = drm_dp_mst_dsc_aux_for_port(port);
+
+	if (!aux)
+		return -EINVAL;
+
+	return drm_dp_dpcd_read(aux, DP_DSC_SUPPORT, caps, 16);
+}
+EXPORT_SYMBOL(drm_dp_mst_dsc_caps_for_port);
+
+/**
+ * drm_dp_mst_dsc_aux_for_port() - Enable DSC on an MST endpoint
+ * @port: The port to check. A leaf of the MST tree with an attached display.
+ * @enable: true for turn on DSC, false for turn off DSC
+ *
+ * Writes DP_DSC_ENABLE on the correct aux for the given port.
+ *
+ * Returns:
+ * The number of bytes written on success, or a negative error code on failure
+ */
+int drm_dp_mst_dsc_enable(struct drm_dp_mst_port *port, bool enable)
+{
+	struct drm_dp_aux *aux = drm_dp_mst_dsc_aux_for_port(port);
+	u8 enable_dsc = enable ? 1 : 0;
+
+	if (!aux)
+		return -EINVAL;
+
+	return drm_dp_dpcd_write(aux, DP_DSC_ENABLE, &enable_dsc, 1);
+}
+EXPORT_SYMBOL(drm_dp_mst_dsc_enable);
include/drm/drm_dp_mst_helper.h +3 -0
index fa973773a4a7..0f70dc8dfbeb 100644
@@ -674,6 +674,9 @@ int __must_check drm_dp_mst_atomic_check(struct drm_atomic_state *state);
 void drm_dp_mst_get_port_malloc(struct drm_dp_mst_port *port);
 void drm_dp_mst_put_port_malloc(struct drm_dp_mst_port *port);
 
+int drm_dp_mst_dsc_caps_for_port(struct drm_dp_mst_port *port, u8 *caps);
+int drm_dp_mst_dsc_enable(struct drm_dp_mst_port *port, bool enable);
+
 extern const struct drm_private_state_funcs drm_dp_mst_topology_state_funcs;
 
 /**
Signature · 6 lines
2.17.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#9 lore raw

Re: [PATCH v4 5/5] drm/dp-mst: Add helpers for querying and enabling MST DSC

Quoted 71 lines · David Francis

Add drm_dp_mst_dsc_caps_for_port and drm_dp_mst_dsc_enable, two helper functions for MST DSC

The former, given a port, returns the raw DPCD DSC caps off that port.

The latter, given a port, enables or disables DSC on that port.

In both cases, the port given as input should be a leaf of the MST tree with an attached display.

The logic for this is somewhat complicated, as DSC can be enabled in 4 different ways.

Case 1: DP-to-DP peer device
if the branch immediately upstream has
- PDT = DP_PEER_DEVICE_DP_MST_BRANCHING (2)
- DPCD rev. >= DP 1.4
- Exactly one input and one output
- The output has PDT = DP_PEER_DEVICE_SST_SINK (3)

In this case, DSC could be possible either on the endpoint or the peer device. Prefer the endpoint, which is possible if
- The endpoint has DP_DSC_DECOMPRESSION_IS_SUPPORTED bit set
- The endpoint has DP_FEC_CAPABLE bit set
- The peer device has DSC_PASSTHROUGH_CAPABILITY bit set (from DP v2.0)

Otherwise, use the peer device

Case 2: DP-to-HDMI peer device
If the output port has
- PDT = DP_PEER_DEVICE_DP_LEGACY_CONV (4)
- DPCD rev. >= DP 1.4
- LDPS = true
- MCS = false

In this case, DSC can only be attempted on the peer device (the output port)

Case 3: Virtual DP Sink (Internal Display Panel)
If the output port has
- DPCD rev. >= DP 1.4
- port_num >= 8

In this case, DSC can only be attempted on the peer device (the output port)

Case 4: Synaptix Workaround
If the output has
- link DPCD rev. >= DP 1.4
- link branch_dev_id = 0x90CC24 (Synaptix)
- There is exactly one branch device between the link and output

In this case, DSC can be attempted, but only using the *link* aux device's caps. This is a quirk.

Cc Lyude Paul <[email protected]>
Cc Wenjing Liu <[email protected]>
Cc Nikola Cornij <[email protected]>
---
 drivers/gpu/drm/drm_dp_mst_topology.c | 195 ++++++++++++++++++++++++++
 include/drm/drm_dp_mst_helper.h       |   3 +
 2 files changed, 198 insertions(+)
drivers/gpu/drm/drm_dp_mst_topology.c +0 -0
index af4b5cec7c84..00ddc54af65b 100644

drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_mst_unregister_i2c_bus(struct · line 4193

Quoted 9 lines
drivers/gpu/drm/drm_dp_mst_topology.c +5 -0
@@ -4186,3 +4186,198 @@ static void drm_dp_mst_unregister_i2c_bus(struct
 drm_dp_aux *aux)
 {
 	i2c_del_adapter(&aux->ddc);
 }
+
+/**
+ * drm_dp_mst_is_virtual_dpcd() - Is the given port a virtual DPCD device?
+ * @port: The port to check
+ *

Mind elaborating a bit here in the kernel documentation on what a virtual DPCD device is?

drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux) · line 4202

Quoted 9 lines

+ * Returns:
+ * true if the port is a virtual DPCD peer device, false otherwise
+ */

drivers/gpu/drm/drm_dp_mst_topology.c +6 -0
+static bool drm_dp_mst_is_virtual_dpcd(struct drm_dp_mst_port *port)
+{
+	struct drm_dp_mst_port *downstream_port;
+
+	if (!port)
+		return false;

I'd drop this, and just move the if (!port) check into drm_dp_mst_dsc_aux_for_port(). Makes things a little more explicit.

drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux) · line 4213

Quoted 11 lines
drivers/gpu/drm/drm_dp_mst_topology.c +11 -0
+
+	/* Virtual DP Sink (Internal Display Panel) */
+	if (port->port_num >= 8 && port->dpcd_rev >= DP_DPCD_REV_14)
+		return true;
+
+	/* DP-to-HDMI Protocol Converter */
+	if (port->pdt == DP_PEER_DEVICE_DP_LEGACY_CONV &&
+			!port->mcs &&
+			port->ldps &&
+			port->dpcd_rev >= DP_DPCD_REV_14)
+		return true;

Please fix the indenting here, most DRM code (I may have mistakenly pointed you at the kernel style guidelines, but I realized they don't explicitly mention this other then the rule of "your code should look like the code around it") written with line continuations starting on the same column as the beginning paranthesis

if (some_very_long_variable == another_very_long_variable_wow ||

    every_identifier_here_is_long > but_what_if_we_have_a_sub_conditional ||
    (then_we_end_up_formatting_it_like_this &&
     starting_at_the_beginning_paranthesis)) {
	same_for_function_calls("Which also require very long lines\n",
	                        foo, bar, baz);
}

drivers/gpu/drm/drm_dp_mst_topology.c

Quoted 13 lines
drivers/gpu/drm/drm_dp_mst_topology.c +7 -0
+
+	/* DP-to-DP */
+	if (port->pdt == DP_PEER_DEVICE_MST_BRANCHING &&
+			port->mstb &&
+			port->dpcd_rev >= DP_DPCD_REV_14 &&
+			port->mstb->num_ports == 2) {
+		list_for_each_entry(downstream_port, &port->mstb->ports, next)

{

drivers/gpu/drm/drm_dp_mst_topology.c +2 -0
+			if (!downstream_port->input &&
+				downstream_port->pdt ==

DP_PEER_DEVICE_SST_SINK)

drivers/gpu/drm/drm_dp_mst_topology.c +2 -0
+				return true;
+		}

This isn't safe, anything that iterates downwards (so either mstb->ports, or port->mstb) in the MST topology needs to happen under &mgr->lock, since otherwise the port list can change under us. Note there's probably some places this isn't followed, but I'm in the process of fixing that:

https://patchwork.freedesktop.org/patch/318683/?series=63847&rev=1

drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux) · line 4231

Quoted 7 lines
drivers/gpu/drm/drm_dp_mst_topology.c +7 -0
+	}
+
+	return false;
+}
+
+/**
+ * drm_dp_mst_is_virtual_dpcd() - Test for Synaptix DSC quirk

I think you messed up a bunch of your kerneldoc comments by accident ^

Quoted 17 lines

+ * @port: The port to check
+ *
+ * Some Synaptix MST hubs support DSC even though they do not support virtual
+ * DPCD. This is a quirk.
+ *
+ * Returns:
+ * true if the Synaptix workaround is required, false otherwise
+ */
+static bool drm_dp_mst_dsc_synaptix_workaround(struct drm_dp_mst_port *port)

drivers/gpu/drm/drm_dp_mst_topology.c +6 -0
+{
+	u8 data[3] = { 0 };
+	u32 dev_id;
+	struct drm_dp_aux *phys_aux;
+
+	/* The hub must be directly connected to the connector */

drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux) · line 4256

Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10 -0
+	if (port->mgr->mst_primary != port->parent)
+		return false;
+
+	phys_aux = port->mgr->aux;
+	if (drm_dp_dpcd_read(phys_aux, DP_BRANCH_OUI, data, 3) < 0)
+		return false;
+	dev_id = (data[0] << 16) & (data[1] << 8) & data[3];
+	/* Synaptix device ID */
+	if (dev_id != 0x90CC24)
+		return false;

More things I'm noticing now that I've got this applied locally. I originally suggested using only the dpcd_quirks list here, but it appears I didn't read this function closely enough and missed the fact you have to read from multiple aux ports here. Huh.

So, first off we don't need to read the ident by hand here, we should just use the helper in drm_dp_read_desc() and add a quirk for this into dpcd_quirk_list[]. Maybe something like DP_DPCD_QUIRK_DSC_WITHOUT_VIRTUAL_DPCD. This also ives us the benefit of having this information printed out into the kernel debug log.

Plus, you accidentally go out of bounds when you check data[3] anyway ;)

drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux) · line 4262

Quoted 6 lines
drivers/gpu/drm/drm_dp_mst_topology.c +6 -0
+
+	if (drm_dp_dpcd_read(phys_aux, DP_DPCD_REV, data, 1) < 0)
+		return false;
+	/* Must be DPCD rev. 1.4 or later */
+	if (data[0] < DP_DPCD_REV_14)
+		return false;

No need to reread the DPCD here, just use the cached dpcd in port->mgr->edid

drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux) · line 4265

Quoted 3 lines
drivers/gpu/drm/drm_dp_mst_topology.c +3 -0
+
+	if (drm_dp_dpcd_read(&port->aux,
+			DP_DOWNSTREAMPORT_PRESENT, data, 1) < 0)

More indenting to fix

drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux) · line 4268

Quoted 3 lines
drivers/gpu/drm/drm_dp_mst_topology.c +3 -0
+		return false;
+	/* Must not be a VGA converter */
+	if ((data[0] & 7) == 3)

Let's just use the proper macros from include/drm/drm_dp_helper.h here

Quoted 109 lines
drivers/gpu/drm/drm_dp_mst_topology.c +8 -0
+		return false;
+
+	return true;
+}
+
+/**
+ * drm_dp_mst_dsc_aux_for_port() - Find the correct aux for DSC
+ * @port: The port to check. A leaf of the MST tree with an attached

display.
+ *
+ * Depending on the situation, DSC may be enabled via the endpoint aux,
+ * the immediately upstream aux, or the connector's physical aux.
+ *
+ * Returns:
+ * NULL if DSC cannot be enabled on this port, otherwise the aux device
+ */
+struct drm_dp_aux *drm_dp_mst_dsc_aux_for_port(struct drm_dp_mst_port *port)

drivers/gpu/drm/drm_dp_mst_topology.c +58 -0
+{
+	u8 upstream_dsc = 0;
+	u8 endpoint_dsc = 0;
+	u8 endpoint_fec = 0;
+	struct drm_dp_mst_port *immediate_upstream_port;
+	struct drm_dp_mst_port *fec_port;
+
+	if (port && port->parent)
+		immediate_upstream_port = port->parent->port_parent;
+	else
+		immediate_upstream_port = NULL;
+
+	fec_port = immediate_upstream_port;
+	while (fec_port) {
+		if (!fec_port->fec_capable)
+			return NULL;
+
+		fec_port = fec_port->parent->port_parent;
+	}
+
+	if (immediate_upstream_port) {
+		if (drm_dp_dpcd_read(&immediate_upstream_port->aux,
+				DP_DSC_SUPPORT, &upstream_dsc, 1) < 0)
+			return NULL;
+	}
+
+	if (drm_dp_dpcd_read(&port->aux,
+			DP_DSC_SUPPORT, &endpoint_dsc, 1) < 0)
+		return NULL;
+	if (drm_dp_dpcd_read(&port->aux,
+			DP_FEC_CAPABILITY, &endpoint_fec, 1) < 0)
+		return NULL;
+
+	/* Enpoint decompression with DP-to-DP peer device */
+	if (drm_dp_mst_is_virtual_dpcd(immediate_upstream_port)
+			&& (upstream_dsc & 0x2) /* DSC passthrough */
+			&& (endpoint_fec & DP_FEC_CAPABLE)
+			&& (endpoint_dsc & DP_DSC_DECOMPRESSION_IS_SUPPORTED))
+		return &port->aux;
+
+	/* Virtual DPCD decompression with DP-to-DP peer device */
+	if (drm_dp_mst_is_virtual_dpcd(immediate_upstream_port))
+		return &immediate_upstream_port->aux;
+
+	/* Virtual DPCD decompression with DP-to-HDMI or Virtual DP Sink */
+	if (drm_dp_mst_is_virtual_dpcd(port))
+		return &port->aux;
+
+	/* Synaptix workaround */
+	if (drm_dp_mst_dsc_synaptix_workaround(port))
+		return port->mgr->aux;
+
+	return NULL;
+}
+
+/**
+ * drm_dp_mst_dsc_aux_for_port() - Retrieve the DSC capability registers
+ * @port: The port to check. A leaf of the MST tree with an attached

display.
+ * @caps: Output. A pointer to an array at least 16 bytes long
+ *
+ * Reads the DSC capability registers (DSC_SUPPORT through
+ * BITS_PER_PIXEL_INCREMENT) and store them in the given pointer. Use
+ * the correct aux for DSC on the given port.
+ *
+ * Returns:
+ * The number of bytes read on success, or a negative error code on failure
+ */

drivers/gpu/drm/drm_dp_mst_topology.c +14 -0
+int drm_dp_mst_dsc_caps_for_port(struct drm_dp_mst_port *port, u8 *caps)
+{
+	struct drm_dp_aux *aux = drm_dp_mst_dsc_aux_for_port(port);
+
+	if (!aux)
+		return -EINVAL;
+
+	return drm_dp_dpcd_read(aux, DP_DSC_SUPPORT, caps, 16);
+}
+EXPORT_SYMBOL(drm_dp_mst_dsc_caps_for_port);
+
+/**
+ * drm_dp_mst_dsc_aux_for_port() - Enable DSC on an MST endpoint
+ * @port: The port to check. A leaf of the MST tree with an attached

display.
+ * @enable: true for turn on DSC, false for turn off DSC
+ *
+ * Writes DP_DSC_ENABLE on the correct aux for the given port.
+ *
+ * Returns:
+ * The number of bytes written on success, or a negative error code on failure
+ */

drivers/gpu/drm/drm_dp_mst_topology.c static void drm_dp_mst_unregister_i2c_bus(struct drm_dp_aux *aux) · line 4382

Quoted 10 lines
drivers/gpu/drm/drm_dp_mst_topology.c +10 -0
+int drm_dp_mst_dsc_enable(struct drm_dp_mst_port *port, bool enable)
+{
+	struct drm_dp_aux *aux = drm_dp_mst_dsc_aux_for_port(port);
+	u8 enable_dsc = enable ? 1 : 0;
+
+	if (!aux)
+		return -EINVAL;
+
+	return drm_dp_dpcd_write(aux, DP_DSC_ENABLE, &enable_dsc, 1);
+}

Why not just make it so that drm_dp_mst_dsc_caps() also returns the appropriate drm_dp_aux device to use, and then pass the pointer to drm_dp_mst_dsc_enable() instead of passing it the drm_dp_mst_port*? That way we won't have to probe for these caps twice.

Quoted 18 lines
drivers/gpu/drm/drm_dp_mst_topology.c +1 -0
+EXPORT_SYMBOL(drm_dp_mst_dsc_enable);
include/drm/drm_dp_mst_helper.h +3 -0
index fa973773a4a7..0f70dc8dfbeb 100644
@@ -674,6 +674,9 @@ int __must_check drm_dp_mst_atomic_check(struct
 drm_atomic_state *state);
 void drm_dp_mst_get_port_malloc(struct drm_dp_mst_port *port);
 void drm_dp_mst_put_port_malloc(struct drm_dp_mst_port *port);
 
+int drm_dp_mst_dsc_caps_for_port(struct drm_dp_mst_port *port, u8 *caps);
+int drm_dp_mst_dsc_enable(struct drm_dp_mst_port *port, bool enable);
+
 extern const struct drm_private_state_funcs
 drm_dp_mst_topology_state_funcs;

/**

Signature · 7 lines
Cheers,
	Lyude Paul

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#10 lore raw
In reply to David Francis

Still in the process of reviewing this, but one minor change that should be done on all of the patches (which I didn't notice before, whoops):

s:drm/dp-mst:drm/dp_mst:g

Quoted 18 lines · David Francis

Add necessary support for MST DSC.
(Display Stream COmpression over Multi-Stream Transport)

v4: Split patchset and rebase onto drm-tip

David Francis (5):
drm/dp-mst: Add PBN calculation for DSC modes drm/dp-mst: Parse FEC capability on MST ports drm/dp-mst: Add MST support to DP DPCD R/W functions drm/dp-mst: Fill branch->num_ports
drm/dp-mst: Add helpers for querying and enabling MST DSC

 drivers/gpu/drm/drm_dp_aux_dev.c      |  12 +-
 drivers/gpu/drm/drm_dp_helper.c       |  10 +-
 drivers/gpu/drm/drm_dp_mst_topology.c | 243 ++++++++++++++++++++++++++
 include/drm/drm_dp_mst_helper.h       |   8 +-
 4 files changed, 260 insertions(+), 13 deletions(-)
Signature · 7 lines
Cheers,
	Lyude Paul

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#11 lore raw
In reply to David Francis

OK-done reviewing, but there's some stuff missing here. The PBN bandwidth
checks in https://patchwork.freedesktop.org/patch/325604/?series=65423&rev=3
need to be moved into drm_dp_mst_atomic_check(), along with moving amdgpu over to using drm_dp_mst_atomic_check(). Doing so will also give us PBN bandwidth checks in both i915 and nouveau as well, and keeps the bandwidth calculation where it should be.

Additionally, you still need to move the code here into an MST atomic helper or drm_dp_mst_atomic_check() as well:

https://patchwork.freedesktop.org/patch/325611/?series=65423&rev=3

Unless I'm mistaken, adding each CRTC which has a connector whose PBN requires recalculation into the atomic state is something every DRM driver is going to need to do. And, you can do this more easily by adding PBN information into drm_dp_mst_topology_state. Yes-it's a lot of locks, but since every connector in an MST topology is sharing the bandwidth on the same link it's kind of expected. I already know I'm going to have to do basically the same thing with every driver once I've got the time to actually implement fallback link rate retraining with MST topologies.

If you need help figuring out how to structure this in a way that works for all drivers, I'm willing to help and I'm sure Manasi is as well.

Quoted 18 lines · David Francis

Add necessary support for MST DSC.
(Display Stream COmpression over Multi-Stream Transport)

v4: Split patchset and rebase onto drm-tip

David Francis (5):
drm/dp-mst: Add PBN calculation for DSC modes drm/dp-mst: Parse FEC capability on MST ports drm/dp-mst: Add MST support to DP DPCD R/W functions drm/dp-mst: Fill branch->num_ports
drm/dp-mst: Add helpers for querying and enabling MST DSC

 drivers/gpu/drm/drm_dp_aux_dev.c      |  12 +-
 drivers/gpu/drm/drm_dp_helper.c       |  10 +-
 drivers/gpu/drm/drm_dp_mst_topology.c | 243 ++++++++++++++++++++++++++
 include/drm/drm_dp_mst_helper.h       |   8 +-
 4 files changed, 260 insertions(+), 13 deletions(-)
Signature · 7 lines
Cheers,
	Lyude Paul

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#12 lore raw

Adding DSC functionality to drm_dp_mst_atomic_check() is a good idea. However, until amdgpu switches over to that system, I wouldn't be able to test those changes. Making that switch is on our TODO list, and it would fix a number of problems with our current MST implementation, but it's going to be a major rewrite.

MST DSC hardware is already on the market. It would be expedient to merge the patches we need for Navi support sooner and update drm_dp_mst_atomic_check when we're able to test it.

David Francis

________________________________________
From: Lyude Paul <[email protected]>
Sent: August 22, 2019 8:03 PM
To: Francis, David; [email protected]; Manasi Navare
Subject: Re: [PATCH v4 0/5] MST DSC support in drm-mst

OK-done reviewing, but there's some stuff missing here. The PBN bandwidth
checks in https://patchwork.freedesktop.org/patch/325604/?series=65423&rev=3
need to be moved into drm_dp_mst_atomic_check(), along with moving amdgpu over to using drm_dp_mst_atomic_check(). Doing so will also give us PBN bandwidth checks in both i915 and nouveau as well, and keeps the bandwidth calculation where it should be.

Additionally, you still need to move the code here into an MST atomic helper or drm_dp_mst_atomic_check() as well:

https://patchwork.freedesktop.org/patch/325611/?series=65423&rev=3

Unless I'm mistaken, adding each CRTC which has a connector whose PBN requires recalculation into the atomic state is something every DRM driver is going to need to do. And, you can do this more easily by adding PBN information into drm_dp_mst_topology_state. Yes-it's a lot of locks, but since every connector in an MST topology is sharing the bandwidth on the same link it's kind of expected. I already know I'm going to have to do basically the same thing with every driver once I've got the time to actually implement fallback link rate retraining with MST topologies.

If you need help figuring out how to structure this in a way that works for all drivers, I'm willing to help and I'm sure Manasi is as well.

Quoted 12 lines · David Francis

Add necessary support for MST DSC.
(Display Stream COmpression over Multi-Stream Transport)

v4: Split patchset and rebase onto drm-tip

David Francis (5):
drm/dp-mst: Add PBN calculation for DSC modes drm/dp-mst: Parse FEC capability on MST ports drm/dp-mst: Add MST support to DP DPCD R/W functions drm/dp-mst: Fill branch->num_ports
drm/dp-mst: Add helpers for querying and enabling MST DSC

Quoted 6 lines
 drivers/gpu/drm/drm_dp_aux_dev.c      |  12 +-
 drivers/gpu/drm/drm_dp_helper.c       |  10 +-
 drivers/gpu/drm/drm_dp_mst_topology.c | 243 ++++++++++++++++++++++++++
 include/drm/drm_dp_mst_helper.h       |   8 +-
 4 files changed, 260 insertions(+), 13 deletions(-)

--
Cheers,
Lyude Paul

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

maintainer
#13 lore raw
Quoted 10 lines · David

Adding DSC functionality to drm_dp_mst_atomic_check() is a good idea. However, until amdgpu switches over to that system, I wouldn't be able to test those changes. Making that switch is on our TODO list, and it would fix a number of problems with our current MST implementation, but it's going to be a major rewrite.

MST DSC hardware is already on the market. It would be expedient to merge the patches we need for Navi support sooner and update drm_dp_mst_atomic_check when we're able to test it.

Is there any commitment to rewriting it, a timeline or anything?

The problem with this situation is there is always new hardware coming onto the market, and there is always pressure to support all the features of that new hardware, and the pressure always comes like this and being expedient. However I've found that a lot of the time the required refactor or work is never done, because the time is being allocated now to the next GPU that is coming on the market, and nobody ever cares enough to clean up their technical debt.

How come the needs for MST DSC support wasn't identified earlier, blocked on refactoring of the code to use common code, and then that task made a higher priority?

I'm sorta inclined to say no we shouldn't be merging any driver specific code here, because this is the only point we can push pressure on to refactor the MST implementation, which I guess otherwise we'll just keep avoiding until Lyude ends up doing it for you.

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

#14 lore raw
Quoted 22 lines · Dave Airlie
Quoted 10 lines · David

Adding DSC functionality to drm_dp_mst_atomic_check() is a good idea. However, until amdgpu switches over to that system, I wouldn't be able to test those changes. Making that switch is on our TODO list, and it would fix a number of problems with our current MST implementation, but it's going to be a major rewrite.

MST DSC hardware is already on the market. It would be expedient to merge the patches we need for Navi support sooner and update drm_dp_mst_atomic_check when we're able to test it.

Is there any commitment to rewriting it, a timeline or anything?

The problem with this situation is there is always new hardware coming onto the market, and there is always pressure to support all the features of that new hardware, and the pressure always comes like this and being expedient. However I've found that a lot of the time the required refactor or work is never done, because the time is being allocated now to the next GPU that is coming on the market, and nobody ever cares enough to clean up their technical debt.

Quoted 4 lines

How come the needs for MST DSC support wasn't identified earlier, blocked on refactoring of the code to use common code, and then that task made a higher priority?

drm_dp_mst_atomic_check was introduced by Lyude back in January with
https://patchwork.freedesktop.org/patch/276405/ as part of
https://patchwork.freedesktop.org/series/54031/

At the time Lyude updated i915 and nouveau to use these helpers. amdgpu wasn't updated.

Quoted 6 lines

I'm sorta inclined to say no we shouldn't be merging any driver specific code here, because this is the only point we can push pressure on to refactor the MST implementation, which I guess otherwise we'll just keep avoiding until Lyude ends up doing it for you.

That's fair. I agree that the refactor should be done and I understand where you're coming from. Since David is heading back to school in less than a week I was inclined to see if we can push back a little so he can get his change in. Other than that I don't mind holding off on the merge unless the refactor is done.

Adding Mikita who'll pick up DSC stuff from David and will iterate on these patches if necessary and look at the MST refactor.

Thanks for keeping us honest.

Harry

Quoted 6 lines

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

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