AnsweredAssumed Answered

Deadlock in 100.8

Question asked by mprickard on Jul 1, 2020
Latest reply on Aug 16, 2020 by MBranscomb-esristaff

I asked a question recently about deadlocks in 100.7.

Michael Branscomb answered my question and I upgraded to 100.8

That upgrade made my experience much better, but as our testers have used it more, deadlocks are still occurring.

The callstacks are a little different than what I had seen in 100.7.

I have attached a txt file with the callstack details.

In the minidump I got from him, there are 45 threads waiting at mtx_do_lock.

The summary of threads is:

  • Esri_runtimecore::Geo_views::Map_view::set_viewpoint_
  • Esri_runtimecore::Map_renderer::Canvas_drawable::highlight_color
    • from addlayer
  • Esri_runtimecore::Map_renderer::Map::copy_display_properties(void)
    • from CoreGeoView.Pulse
  • Esri_runtimecore::Geo_views::Geo_view::update_attribution_(void)
    • There are a lot of threads with this same callstack.


The second attachment, "100.8.hang2.txt" is an example when I was debugging.

The summary is:

  • Esri_runtimecore::Geo_views::Map_view::set_view_insets_
    • on the main/dispatcher thread
  • Esri_runtimecore::Geo_views::Geo_view::update_attribution_
    • 2 threads with this callstack this time.
  • Esri_runtimecore::Map_renderer::Canvas_drawable::highlight_color
    • from addlayer
  • Esri_runtimecore::Map_renderer::Map::copy_display_properties
    • from CoreGeoView.Pulse

Any ideas?