Star BBAF during RAW recording crashes ML

Issue #2254 new
Licaon Kter
created an issue

Having * (star) set up for back button auto-focus ( via C.FnIV 1 or 3 on my M1 ) will crash ML if used during a RAW video recording and will require a battery pull for recovery.

See video: https://mega.co.nz/#!1R53GIgL!_GRBQzRX_bIPcHY0t37aK0TVqGta2aJ5CMUugjfUAxs

Build info:

Magic Lantern version : Nightly.2015Apr07.EOSM202
Mercurial changeset   : de7779da4ac8 (unified) tip
Built on 2015-04-06 22:09:09 UTC by jenkins@magiclantern.fm.

I got some 52 CRASHxxx.LOG logs and 2 LOGxxx.LOG, mostly duplicates or differing by 1K in the free memory info, those that were unique otherwise are attached.

/PS: also reported on the forums: http://www.magiclantern.fm/forum/index.php?topic=14945.msg144893#msg144893

Comments (20)

  1. Daniel Fort

    I was able to reproduce this issue following the steps shown on your video. However, I think that the people that are successfully using RAW video on the EOS-M are using Movie crop mode so I tried that and could not reproduce the issue. Note that with Movie crop mode the focus box doesn't show while recording video so that might be related to the problem.

    Here's my crash log using your build:

    ASSERT: !IS_ERROR( TryPostEvent( this->hTaskClass, this, EV_VD_INTERRUPT_EVF, NULL, 0 ) ) at ./Evf/EvfState.c:504, task ? lv:1 mode:3

    Magic Lantern version : Nightly.2015May31.EOSM202 Mercurial changeset : 705d001cb9d5+ (unified) Built on 2015-05-31 09:24:04 UTC by licaon@pillardss. Free Memory : 152K + 3364K

  2. Licaon Kter reporter

    Good to know. Although either crop mode is used more or not on this camera I find questionable, given that focus gets even harder to keep track of.

    Also this does not need my build to repro anyway.

  3. Licaon Kter reporter

    Still an issue

    ASSERT: !IS_ERROR( TryPostEvent( this->hTaskClass, this, EV_VD_INTERRUPT_EVF, NULL, 0 ) )
    at ./Evf/EvfState.c:504, task ?
    lv:1 mode:3
    
    
    Magic Lantern version : Nightly.2016Mar13.EOSM202
    Mercurial changeset   : dd16e5d6fd13+ (unified) tip
    Built on 2016-03-13 09:15:28 UTC by dev@dev.
    Free Memory  : 129K + 3174K
    
  4. Daniel Fort

    The only camera besides the EOSM and 5D3 I currently have to test on is the 700D. I just tested and when back button focusing is set and recording raw--recording stops. The message on the screen is "Frame skipped. Stopping" but no crash, battery pull needed or log created.

  5. Alex

    I guess there's something in the vsync hook interferring with autofocus.

    To confirm, you can try commenting out that hook (the line with MODULE_CBR(CBR_VSYNC)) and see if it still crashes. Of course, it won't record anything.

  6. Licaon Kter reporter

    Yeah

    ASSERT: !IS_ERROR( TryPostEvent( this->hTaskClass, this, EV_VD_INTERRUPT_EVF, NULL, 0 ) )
    at ./Evf/EvfState.c:504, task ?
    lv:1 mode:3
    

    and some say

    ASSERT: !IS_ERROR( TryPostEvent( this->hTaskClass, this, EV_READOUTDONE_INTERRUPT_EVF, NULL, 0 ) )
    at ./Evf/EvfState.c:479, task ?
    lv:1 mode:3
    

    Any reason this keeps on creating error .LOGs seemingly forever?

    Anyway, you need to guide me regarding small hacks? SRM memory?

  7. Alex

    Some more ideas:

    • Does the half-shutter button have to be configured for autofocus (it is by default), or it crashes even if it's configured not to trigger AF? (Canon menu, CFn 5 set to 1)

    • Does the crash occur in photo mode? (try with and without raw-based overlays active) Examples:

      • global draw off -> expecting no crash
      • global draw on, fast zebras, focus peaking, waveform, vectorscope or any combination of these, all others off -> expecting no crash
      • global draw on, raw-based histogram/zebras/spotmeter active (any of these) -> expecting crash
  8. Log in to comment