| 12:41 | mk01_ | jnettlet: thanks for info. btw I'm not confident about mine commit you merged (with the hp_state) reordering |
| 13:36 | jnettlet | mk01_, is it working with your receiver, or do you still need a fix for that? |
| 13:47 | mk01_ | jnettlet: no "no EDID" in the logs. |
| 13:48 | jnettlet | interesting |
| 13:49 | mk01_ | but imx-ipuv3 2400000.ipu: warning: wait for bg sync eof timeout is new (never seen before) - ending with gpu stack dump |
| 13:49 | jnettlet | it is still broken behaviour by the spec. I have always needed to fix this since I originally wrote the patch for the XO-4. It had a micro-hdmi port so was very fiddly with new connections |
| 13:50 | jnettlet | on boot, or after pulling the cable? |
| 13:50 | mk01_ | it looks it comes with HP irqs |
| 13:51 | jnettlet | are you running kodi? |
| 13:51 | mk01_ | yes, it is running in background |
| 13:52 | jnettlet | yeah, we will need to make Kodi aware of hotplug events to fix that. |
| 13:53 | jnettlet | it assumes that your framebuffer is always there no matter what. It can't deal with dynamic changes. |
| 13:53 | mk01_ | no. I'm stopping renderer. with plugout. |
| 13:54 | mk01_ | then enabling it only after new resolutions are red |
| 13:55 | jnettlet | where are you reading plugout from? |
| 13:55 | mk01_ | from udev |
| 13:56 | mk01_ | which is monitoring hdmi |
| 13:56 | jnettlet | yeah, that is going to be too slow |
| 13:56 | mk01_ | 20e0000.hdmi_video |
| 13:56 | mk01_ | no discussion there. |
| 13:57 | jnettlet | we are going to have to notify of the plugout event and then have a delay before we tear everything down |
| 13:57 | jnettlet | but I don't think it is worth putting more time into fbdev hotplugging. This is all fixed with drm. |
| 13:59 | mk01_ | but what is the specific change there ? the power down through the new fb_blank() ? |
| 14:02 | mk01_ | ((and I have to say I'm currently running code with the mentioned commit reverted)) <- it is doing 'a disco' - by reenabling unmasking HP in the middle of the hotplug_worker() run |
| 14:02 | mk01_ | you run DVI, right ? |
| 14:03 | jnettlet | Yes, i test both |
| 14:04 | jnettlet | Yes the fb_blank disables the ipu worker that was allocated. That is of the reasons i added the change saves 80mA of power for headless uses. |
| 14:05 | mk01_ | last time you was thinking loud - about variant to do just one switch |
| 14:05 | mk01_ | (in the order). not on both events |
| 14:09 | mk01_ | jnettlet: what was the idea like ? to reverse the order in plugout(), but keep in plugin() ? |
| 14:09 | jnettlet | yeah. |