diff mbox series

[-next,1/2] media: aspeed: refine hsync/vsync polarity setting logic

Message ID 20190910190756.31432-2-jae.hyun.yoo@linux.intel.com (mailing list archive)
State New, archived
Headers show
Series media: aspeed: refine mode detection flow | expand

Commit Message

Jae Hyun Yoo Sept. 10, 2019, 7:07 p.m. UTC
This commit refines hsync/vsync polarity setting logic by making
also clearing register bits possible based on probed sync state
accordingly.

Signed-off-by: Jae Hyun Yoo <jae.hyun.yoo@linux.intel.com>
---
 drivers/media/platform/aspeed-video.c | 43 +++++++++++++--------------
 1 file changed, 21 insertions(+), 22 deletions(-)

Comments

Andrew Jeffery Sept. 12, 2019, 5:33 a.m. UTC | #1
On Wed, 11 Sep 2019, at 04:37, Jae Hyun Yoo wrote:
> This commit refines hsync/vsync polarity setting logic by making
> also clearing register bits possible based on probed sync state
> accordingly.

That was tough to parse, but I think I understand. Trying to rephrase:

Enable clearing of hsync/vsync plarity bits based on probed sync state.

What was the issue that drove the change? Do you know why it was done
the way it was prior to this patch?

Andrew
Jae Hyun Yoo Sept. 12, 2019, 5:04 p.m. UTC | #2
Hi Andrew,

On 9/11/2019 10:33 PM, Andrew Jeffery wrote:
> On Wed, 11 Sep 2019, at 04:37, Jae Hyun Yoo wrote:
>> This commit refines hsync/vsync polarity setting logic by making
>> also clearing register bits possible based on probed sync state
>> accordingly.
> 
> That was tough to parse, but I think I understand. Trying to rephrase:
> 
> Enable clearing of hsync/vsync plarity bits based on probed sync state.

Correct.

> What was the issue that drove the change? Do you know why it was done
> the way it was prior to this patch?

Because this driver detects weird resolutions sometimes. Investigated
that it uses
     aspeed_video_update(video, VE_CTRL, 0, ctrl);
so only setting of polarity bits is available. Means that clearing of
the bits isn't available so it can't set back the polarities to normal.

To fix the issue, this patch makes it use
     aspeed_video_write(video, VE_CTRL, ctrl);
instead of above one with adding bit masking code changes.

Thanks,
Jae
diff mbox series

Patch

diff --git a/drivers/media/platform/aspeed-video.c b/drivers/media/platform/aspeed-video.c
index eb12f3793062..8f77079da55a 100644
--- a/drivers/media/platform/aspeed-video.c
+++ b/drivers/media/platform/aspeed-video.c
@@ -614,7 +614,7 @@  static void aspeed_video_check_and_set_polarity(struct aspeed_video *video)
 	int i;
 	int hsync_counter = 0;
 	int vsync_counter = 0;
-	u32 sts;
+	u32 sts, ctrl;
 
 	for (i = 0; i < NUM_POLARITY_CHECKS; ++i) {
 		sts = aspeed_video_read(video, VE_MODE_DETECT_STATUS);
@@ -629,30 +629,29 @@  static void aspeed_video_check_and_set_polarity(struct aspeed_video *video)
 			hsync_counter++;
 	}
 
-	if (hsync_counter < 0 || vsync_counter < 0) {
-		u32 ctrl = 0;
+	ctrl = aspeed_video_read(video, VE_CTRL);
 
-		if (hsync_counter < 0) {
-			ctrl = VE_CTRL_HSYNC_POL;
-			video->detected_timings.polarities &=
-				~V4L2_DV_HSYNC_POS_POL;
-		} else {
-			video->detected_timings.polarities |=
-				V4L2_DV_HSYNC_POS_POL;
-		}
-
-		if (vsync_counter < 0) {
-			ctrl = VE_CTRL_VSYNC_POL;
-			video->detected_timings.polarities &=
-				~V4L2_DV_VSYNC_POS_POL;
-		} else {
-			video->detected_timings.polarities |=
-				V4L2_DV_VSYNC_POS_POL;
-		}
+	if (hsync_counter < 0) {
+		ctrl |= VE_CTRL_HSYNC_POL;
+		video->detected_timings.polarities &=
+			~V4L2_DV_HSYNC_POS_POL;
+	} else {
+		ctrl &= ~VE_CTRL_HSYNC_POL;
+		video->detected_timings.polarities |=
+			V4L2_DV_HSYNC_POS_POL;
+	}
 
-		if (ctrl)
-			aspeed_video_update(video, VE_CTRL, 0, ctrl);
+	if (vsync_counter < 0) {
+		ctrl |= VE_CTRL_VSYNC_POL;
+		video->detected_timings.polarities &=
+			~V4L2_DV_VSYNC_POS_POL;
+	} else {
+		ctrl &= ~VE_CTRL_VSYNC_POL;
+		video->detected_timings.polarities |=
+			V4L2_DV_VSYNC_POS_POL;
 	}
+
+	aspeed_video_write(video, VE_CTRL, ctrl);
 }
 
 static bool aspeed_video_alloc_buf(struct aspeed_video *video,