ASoC/tda998x: Fix reporting of nonexistent capture streams

Merge series from Mark Brown <broonie@kernel.org>:

The recently added pcm-test selftest has pointed out that systems with
the tda998x driver end up advertising that they support capture when in
reality as far as I can see the tda998x devices are transmit only.  The
DAIs registered through hdmi-codec are bidirectional, meaning that for
I2S systems when combined with a typical bidrectional CPU DAI the
overall capability of the PCM is bidirectional.  In most cases the I2S
links will clock OK but no useful audio will be returned which isn't so
bad but we should still not advertise the useless capability, and some
systems may notice problems for example due to pinmux management.

This is happening due to the hdmi-codec helpers not providing any
mechanism for indicating unidirectional audio so add one and use it in
the tda998x driver.  It is likely other hdmi-codec users are also
affected but I don't have those systems to hand.

Mark Brown (2):
  ASoC: hdmi-codec: Allow playback and capture to be disabled
  drm: tda99x: Don't advertise non-existent capture support

 drivers/gpu/drm/i2c/tda998x_drv.c |  2 ++
 include/sound/hdmi-codec.h        |  4 ++++
 sound/soc/codecs/hdmi-codec.c     | 30 +++++++++++++++++++++++++-----
 3 files changed, 31 insertions(+), 5 deletions(-)

base-commit: f0c4d9fc9cc9462659728d168387191387e903cc
--
2.30.2

1 files changed, 20 insertions, 4 deletions

diff --git a/drivers/platform/surface/aggregator/ssh_packet_layer.c b/drivers/platform/surface/aggregator/ssh_packet_layer.c
index 6748fe4ac5d5..def8d7ac541f 100644
--- a/drivers/platform/surface/aggregator/ssh_packet_layer.c
+++ b/drivers/platform/surface/aggregator/ssh_packet_layer.c
@@ -1596,16 +1596,32 @@ static void ssh_ptl_timeout_reap(struct work_struct *work)
 		ssh_ptl_tx_wakeup_packet(ptl);
 }
 
-static bool ssh_ptl_rx_retransmit_check(struct ssh_ptl *ptl, u8 seq)
+static bool ssh_ptl_rx_retransmit_check(struct ssh_ptl *ptl, const struct ssh_frame *frame)
 {
 	int i;
 
 	/*
+	 * Ignore unsequenced packets. On some devices (notably Surface Pro 9),
+	 * unsequenced events will always be sent with SEQ=0x00. Attempting to
+	 * detect retransmission would thus just block all events.
+	 *
+	 * While sequence numbers would also allow detection of retransmitted
+	 * packets in unsequenced communication, they have only ever been used
+	 * to cover edge-cases in sequenced transmission. In particular, the
+	 * only instance of packets being retransmitted (that we are aware of)
+	 * is due to an ACK timeout. As this does not happen in unsequenced
+	 * communication, skip the retransmission check for those packets
+	 * entirely.
+	 */
+	if (frame->type == SSH_FRAME_TYPE_DATA_NSQ)
+		return false;
+
+	/*
 	 * Check if SEQ has been seen recently (i.e. packet was
 	 * re-transmitted and we should ignore it).
 	 */
 	for (i = 0; i < ARRAY_SIZE(ptl->rx.blocked.seqs); i++) {
-		if (likely(ptl->rx.blocked.seqs[i] != seq))
+		if (likely(ptl->rx.blocked.seqs[i] != frame->seq))
 			continue;
 
 		ptl_dbg(ptl, "ptl: ignoring repeated data packet\n");
@@ -1613,7 +1629,7 @@ static bool ssh_ptl_rx_retransmit_check(struct ssh_ptl *ptl, u8 seq)
 	}
 
 	/* Update list of blocked sequence IDs. */
-	ptl->rx.blocked.seqs[ptl->rx.blocked.offset] = seq;
+	ptl->rx.blocked.seqs[ptl->rx.blocked.offset] = frame->seq;
 	ptl->rx.blocked.offset = (ptl->rx.blocked.offset + 1)
 				  % ARRAY_SIZE(ptl->rx.blocked.seqs);
 
@@ -1624,7 +1640,7 @@ static void ssh_ptl_rx_dataframe(struct ssh_ptl *ptl,
 				 const struct ssh_frame *frame,
 				 const struct ssam_span *payload)
 {
-	if (ssh_ptl_rx_retransmit_check(ptl, frame->seq))
+	if (ssh_ptl_rx_retransmit_check(ptl, frame))
 		return;
 
 	ptl->ops.data_received(ptl, payload);