• State
  • Locked Locked
  • Replies 3 replies
  • Subscribers 99 subscribers
  • Views 315 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DM368 ipipe_enable lockup

Pete Eberlein
Intellectual 580 points

Using 2.6.37+ kernel on DM368.  Using custom hardware with HDMI video decoder and VPFE in continuous mode, I see a lockup when switching the video source from 1080i to 720p.  The location where the lockup occurs is drivers/media/video/davinci/dm365_ipipe.c: ipipe_enable() in this loop:

    if (en && param->rsz_common.source == IPIPE_DATA) {
        /* wait for IPIPE to become inactive */
        do {
            ret = regr_ip(IPIPE_SRC_EN);
        }while (ret);

Anyone know how or why the resizer is getting stuck in the enabled state?  I've verified that it was previously disabled using en=0.

  • 0
    Expert 1690 points

    Pete,

    I was tracking down a lockup with our hardware and tracked down the lockup to the ipipe_enable call.  Did you ever make any headway here?  Sometimes it can take 20 or 30 minutes before it happens.

    I should point out that I am enabling / disabling RSZB every other frame, since I don't always need the second output and I very much need to conserve DDR bandwidth since I am using a DM365.

    Mark

  • 0 in reply to mdeneen
    Intellectual 580 points

    Mark,

    I was never able to figure out what was causing the lockup, but for now I've been using a counter limit in the dm365_ipipe.c like this:

            /* wait for IPIPE to become inactive */
            do {
                ret = regr_ip(IPIPE_SRC_EN);
                if (ret) schedule();
            }while (ret && ++count < 10000);

    This prevents it from locking up completely, and I haven't observed any adverse effects.  I too am disabling RSZB every other frame, while using continuous mode.  I've made some changes to vpfe_resizer as well to enable RSZ_B and allow all bufs to be dequeued (previously it would never allow the last one to be dequeued)

  • 0 in reply to Pete Eberlein
    Expert 1690 points

    Pete,

    Thank you for the update.  I am using a multi-channel decoder, so I am unable to use continuous mode. However, this was similar to the approach I was going to take and I'm glad to hear that you've not seen any adverse effects so far.

    Mark