omapdss DISPC error: SYNC_LOST_DIGIT

[Lee Jones]

Lots of spurious "omapdss DISPC error: SYNC_LOST_DIGIT" error messages printed out to the console when a monitor is connected via HDMI.

- Kernel -
Description: Ubuntu 10.04 LTS
Release: 10.04
Linux version 2.6.33-900-omap4 (xxxxx@xxxxx) (gcc version 4.4.3 (Ubuntu 4.4.3-4ubuntu5) ) #1~panda SMP PREEMPT Tue Jun 8 08:50:45 EDT 2010

- Board -
Panda OMAP 4

- Monitor-
LG W2261VP 22" LCD Widescreen Full HD Gaming Monitor 1920 x 1080 250cd/m2 10,000:1 2ms VGA/DVI/HDMI

[Bryan Wu (cooloney)]

After grepping the OMAP4 kernel, I found some code was commented in the driver:

drivers/video/omap2/dss/dispc.c
----
        if (errors & DISPC_IRQ_SYNC_LOST_DIGIT) {

        DSSERR("SYNC_LOST_DIGIT\n");
/*commenting below code as with 1080P Decode we see a sync lost digit for
first frame as it takes long time to decode but it later recovers*/
#if 0
                struct omap_overlay_manager *manager = NULL;
                bool enable = false;

                DSSERR("SYNC_LOST_DIGIT, disabling TV\n");

                for (i = 0; i < omap_dss_get_num_overlay_managers(); ++i) {
                        struct omap_overlay_manager *mgr;
                        mgr = omap_dss_get_overlay_manager(i);

                        if (mgr->id == OMAP_DSS_CHANNEL_DIGIT) {
                                manager = mgr;
                                enable = mgr->device->state ==
                                                OMAP_DSS_DISPLAY_ACTIVE;
                                mgr->device->driver->disable(mgr->device);
                                break;
                        }

                }

                if (manager) {
                        struct omap_dss_device *dssdev = manager->device;
                        for (i = 0; i < omap_dss_get_num_overlays(); ++i) {
                                struct omap_overlay *ovl;
                                ovl = omap_dss_get_overlay(i);

                                if (!(ovl->caps & OMAP_DSS_OVL_CAP_DISPC))
                                        continue;

                                if (ovl->id != 0 && ovl->manager == manager)
                                        dispc_enable_plane(ovl->id, 0);
                        }

                        dispc_go(manager->id);
                        mdelay(50);
                        if (enable)
                                dssdev->driver->enable(dssdev);
                }
#endif
        }
----

I am not sure about that.
-Bryan

[Lee Jones (lag)]

I'm sure there is a neater solution than that.

I will test and post my results on here.

[Sebastien JAN (sebjan)]

Not sure about the root cause for that. In the meantime, here is a kernel hack that shall disable this trace and free some cpu (to avoid locking the board because of the huge traces generated). This shall enable using the HDMI output.

--- drivers/video/omap2/dss/dispc.c 2010-06-15 09:57:56.000000000 +0200
+++ drivers/video/omap2/dss/dispc.c 2010-0mc-pdm1.patch6-04 17:04:06.000000000 +0200
@@ -4182,7 +4182,7 @@mc-pdm1.patch
  PIS(VID1_FIFO_UNDERFLOW);
  PIS(VID2_FIFO_UNDERFLOW);
  PIS(SYNC_LOST);
- PIS(SYNC_LOST_DIGIT);
+ /*PIS(SYNC_LOST_DIGIT);*/
 #undef PIS

  printk("\n");
@@ -4431,7 +4431,7 @@
   bool enable = false;
 #endif

- DSSERR("SYNC_LOST_DIGIT\n");
+ /*DSSERR("SYNC_LOST_DIGIT\n");*/
  /*
   DSSERR("SYNC_LOST_DIGIT, disabling TV\n");

[Lee Jones (lag)]

FAO Sebastien,

I already have something similar to allow me access to the system console.

My method uses a 'static int' and only allows the message to appear 10 times.

I'm currently building the very latest ti-omap4 branch to see if the error has regressed.

I'll keep you informed of my progress.

[Lee Jones (lag)]

Bug seems to have disappeared in the latest Linux version 2.6.34-901-omap4 kernel.

I'm going to close this bug until it rears its ugly head again.

[Lee Jones (lag)]

Change of plan.

After some extra experimentation it appears the bug still occurs, but only if an HDMI monitor is plugged in and turned off.

[mythripk (mythripk)]

the reason why the portion of code in dispc.c
drivers/video/omap2/dss/dispc.c
----
        if (errors & DISPC_IRQ_SYNC_LOST_DIGIT) {

        DSSERR("SYNC_LOST_DIGIT\n");
/*commenting below code as with 1080P Decode we see a sync lost digit for
first frame as it takes long time to decode but it later recovers*/
is commented out is that , The sync lost digit can be got for several reasons . One of the reasons is that when IVA-HD decoded video is played on HDMI , then the first frame takes about 1s to decode and results in a SYNC_LOST_DIGIT, which if the code is not commented would disable the overlay manager.

Now coming to the SYNC_LOST_DIGIT as such , It would be helpful if you could elaborate on how you are able to reproduce the issue.I assume the timing set is correct.

[Lee Jones (lag)]

A patch has been put up for RFC at the address below.

http://omappedia.org/wiki/RFCs

This patch allows the kernel to auto detect a monitor's settings and only attempt to turn it on when it's found. This clears the spurious SYNC_LOST_DIGIT messages.

[Lee Jones (lag)]

The fix for this issue has been committed with 2.6.35.

[Adriano Pallavicino (adriano.pallavicino)]

Hi, I've got same issue on my pandaboard rel. A3.
kernel is:
=====================

adriano@pandaboard:~$ uname -r
2.6.38-1208-omap4
=====================

and the error printed out on serial console in a while(1) loop

=====================
[ 730.415527] omapdss DISPC error: SYNC_LOST_DIGIT, disabling TV
=====================

the distro is ubuntu headless 11.04

=====================

adriano@pandaboard:~$ uname -a
Linux pandaboard 2.6.38-1208-omap4 #11-Ubuntu SMP PREEMPT Fri Apr 15 16:34:35 UTC 2011 armv7l armv7l armv7l GNU/Linux
=====================

I hope to wrote on the right place, best regards

Adriano