diff mbox series

[PATCHv2,13/22] drm/bridge: tc358767: use more reliable seq when finishing LT

Message ID 20190326103146.24795-14-tomi.valkeinen@ti.com (mailing list archive)
State New, archived
Headers show
Series drm/bridge: tc358767: DP support | expand

Commit Message

Tomi Valkeinen March 26, 2019, 10:31 a.m. UTC
At the end of the link training, two steps have to be taken: 1)
tc358767's LT mode is disabled by a write to DP0_SRCCTRL, and 2) Remove
LT flag in DPCD 0x102.

Toshiba's documentation tells to first write the DPCD, then modify
DP0_SRCCTRL. In my testing this often causes issues, and the link
disconnects right after those steps.

If I reverse the steps, it works every time. There's a chance that this
is DP sink specific, though, but as my testing shows this sequence to be
much more reliable, let's change it.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
---
 drivers/gpu/drm/bridge/tc358767.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

Comments

Andrzej Hajda April 15, 2019, 8:51 a.m. UTC | #1
On 26.03.2019 11:31, Tomi Valkeinen wrote:
> At the end of the link training, two steps have to be taken: 1)
> tc358767's LT mode is disabled by a write to DP0_SRCCTRL, and 2) Remove
> LT flag in DPCD 0x102.
>
> Toshiba's documentation tells to first write the DPCD, then modify
> DP0_SRCCTRL. In my testing this often causes issues, and the link
> disconnects right after those steps.
>
> If I reverse the steps, it works every time. There's a chance that this
> is DP sink specific, though, but as my testing shows this sequence to be
> much more reliable, let's change it.
>
> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>

 --
Regards
Andrzej
Laurent Pinchart April 20, 2019, 9:44 p.m. UTC | #2
Hi Tomi,

Thank you for the patch.

On Tue, Mar 26, 2019 at 12:31:37PM +0200, Tomi Valkeinen wrote:
> At the end of the link training, two steps have to be taken: 1)
> tc358767's LT mode is disabled by a write to DP0_SRCCTRL, and 2) Remove
> LT flag in DPCD 0x102.
> 
> Toshiba's documentation tells to first write the DPCD, then modify
> DP0_SRCCTRL. In my testing this often causes issues, and the link
> disconnects right after those steps.
> 
> If I reverse the steps, it works every time. There's a chance that this
> is DP sink specific, though, but as my testing shows this sequence to be
> much more reliable, let's change it.

As this departs from the documentation I think it deserves a comment in
the code.

> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
> ---
>  drivers/gpu/drm/bridge/tc358767.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/bridge/tc358767.c b/drivers/gpu/drm/bridge/tc358767.c
> index 2a6c0c0d47a6..700e161015af 100644
> --- a/drivers/gpu/drm/bridge/tc358767.c
> +++ b/drivers/gpu/drm/bridge/tc358767.c
> @@ -958,6 +958,9 @@ static int tc_main_link_enable(struct tc_data *tc)
>  	if (ret)
>  		goto err;
>  
> +	/* Clear Training Pattern, set AutoCorrect Mode = 1 */
> +	tc_write(DP0_SRCCTRL, tc_srcctrl(tc) | DP0_SRCCTRL_AUTOCORRECT);
> +
>  	/* Clear DPCD 0x102 */
>  	/* Note: Can Not use DP0_SNKLTCTRL (0x06E4) short cut */
>  	tmp[0] = tc->link.scrambler_dis ? DP_LINK_SCRAMBLING_DISABLE : 0x00;
> @@ -965,9 +968,6 @@ static int tc_main_link_enable(struct tc_data *tc)
>  	if (ret < 0)
>  		goto err_dpcd_write;
>  
> -	/* Clear Training Pattern, set AutoCorrect Mode = 1 */
> -	tc_write(DP0_SRCCTRL, tc_srcctrl(tc) | DP0_SRCCTRL_AUTOCORRECT);
> -
>  	/* Wait */
>  	timeout = 100;
>  	do {
Tomi Valkeinen May 3, 2019, 11:04 a.m. UTC | #3
On 21/04/2019 00:44, Laurent Pinchart wrote:
> Hi Tomi,
> 
> Thank you for the patch.
> 
> On Tue, Mar 26, 2019 at 12:31:37PM +0200, Tomi Valkeinen wrote:
>> At the end of the link training, two steps have to be taken: 1)
>> tc358767's LT mode is disabled by a write to DP0_SRCCTRL, and 2) Remove
>> LT flag in DPCD 0x102.
>>
>> Toshiba's documentation tells to first write the DPCD, then modify
>> DP0_SRCCTRL. In my testing this often causes issues, and the link
>> disconnects right after those steps.
>>
>> If I reverse the steps, it works every time. There's a chance that this
>> is DP sink specific, though, but as my testing shows this sequence to be
>> much more reliable, let's change it.
> 
> As this departs from the documentation I think it deserves a comment in
> the code.

Yep. It's not exactly documentation, but just an excel sheet that they
use for testing. The actual documentation doesn't talk about this.

I'll add a comment.

 Tomi
diff mbox series

Patch

diff --git a/drivers/gpu/drm/bridge/tc358767.c b/drivers/gpu/drm/bridge/tc358767.c
index 2a6c0c0d47a6..700e161015af 100644
--- a/drivers/gpu/drm/bridge/tc358767.c
+++ b/drivers/gpu/drm/bridge/tc358767.c
@@ -958,6 +958,9 @@  static int tc_main_link_enable(struct tc_data *tc)
 	if (ret)
 		goto err;
 
+	/* Clear Training Pattern, set AutoCorrect Mode = 1 */
+	tc_write(DP0_SRCCTRL, tc_srcctrl(tc) | DP0_SRCCTRL_AUTOCORRECT);
+
 	/* Clear DPCD 0x102 */
 	/* Note: Can Not use DP0_SNKLTCTRL (0x06E4) short cut */
 	tmp[0] = tc->link.scrambler_dis ? DP_LINK_SCRAMBLING_DISABLE : 0x00;
@@ -965,9 +968,6 @@  static int tc_main_link_enable(struct tc_data *tc)
 	if (ret < 0)
 		goto err_dpcd_write;
 
-	/* Clear Training Pattern, set AutoCorrect Mode = 1 */
-	tc_write(DP0_SRCCTRL, tc_srcctrl(tc) | DP0_SRCCTRL_AUTOCORRECT);
-
 	/* Wait */
 	timeout = 100;
 	do {