| 11:33 | vpeter | jnettlet: I will send 2 PR's for Udoo. One for device tree and one for shutdown. The third one for vt1613 codec is in the work (as expecped it doesn't work with this kernel). |
| 11:34 | jnettlet | okay. |
| 11:49 | mk01 | jnettlet: |
| 11:49 | jnettlet | hey |
| 11:49 | mk01 | hey |
| 11:49 | mk01 | tried to put your double timer idea into work |
| 11:49 | mk01 | but can't do it so nicely as it sounds |
| 11:50 | mk01 | due to a logical flow problem I have (and can't find an elegant way for it): |
| 11:51 | mk01 | we want second timer to schedule function at +200ms after plugin |
| 11:51 | mk01 | but before plugin plugout comes first |
| 11:51 | mk01 | to detect plugin we have to reprogram the masks |
| 11:51 | mk01 | this is happening in hotplug worker |
| 11:52 | mk01 | which itself is scheduled with the 1s delay |
| 11:52 | mk01 | so |
| 11:52 | mk01 | "plugin" comes already delayed |
| 11:52 | mk01 | as until we unmask irqs after plugout and change the polarity |
| 11:53 | mk01 | we stand down for this 1s |
| 11:55 | mk01 | maybe this is all you wanted and I just misunderstood, maybe you have an nice idea how to achieve that. |
| 11:55 | mk01 | ANYHOW |
| 11:56 | mk01 | I implemented the second timer and for now (testing) satisfied with the flow it is with 2nd timer now: |
| 11:56 | mk01 | 1. plugout happens |
| 11:56 | mk01 | 2. 1s jutter |
| 11:56 | mk01 | 3. hotplug worker kicks in - we reprogram polarity |
| 11:57 | mk01 | 4. almost immediatelly we get irq |
| 11:57 | mk01 | 5. in irq handler I detect it's plugin and schedule the edidread to +200ms |
| 11:57 | mk01 | 6. at time index 1.2s from plugout edidread happens |
| 11:57 | mk01 | 7.~ 0.8-1s later full plugin event is processed |
| 11:58 | mk01 | I rebooted into kernel with this change one device and basically for past 12+h I had no "no EDID read" problem |
| 12:00 | jnettlet | oh interesting so your problem is that i2c isn't ready to pass data after we bring up the phy. |
| 12:01 | mk01 | basically yes |
| 12:02 | mk01 | and the factor is AVR |
| 12:05 | mk01 | although wait, let me turn dynamic debug on what is the timing of all those functions |
| 12:06 | mk01 | because phy_init is not triggered from cable_connected() |
| 12:08 | mk01 | ok |
| 12:08 | mk01 | phy_init comes from hdmi_setup what is AFTER edidread |
| 12:09 | mk01 | so from perspective the situation is the same as before. edidread happens with phy disabled |
| 12:10 | jnettlet | which is what we want |
| 12:12 | mk01 | ya, just saying it was that way before as well. |
| 12:13 | mk01 | just wanted say basically : I'm fine now and reported back that this timing (or order of steps) is not doing troubles |
| 12:14 | mk01 | (even if not so nice as I understood your idea initially -> 1. plugout 2. plugin (without delays) 3. +200 ms after (2) edidread ....... then all the masked out by jitter events) |
| 12:16 | mk01 | but probably that would work only with direct TV connection. where the 1s jitter nicelly masks all the intermediates and basically plugout + plugin happens only once |
| 12:17 | mk01 | but with AVR this 1s is too short to mask all the switching out |
| 12:22 | jnettlet | yeah, I need to think about this some more. |
| 14:07 | jnettlet | mk01, what resolution did you see the screen stuttering at? |
| 14:11 | mk01 | jnettlet: any . just it needs to be content (with frames) with bottom field first |
| 14:11 | mk01 | (so interlaced) |
| 14:11 | jnettlet | mk01, I am seeing it with progressive as well |
| 14:12 | mk01 | hmmm |
| 14:12 | mk01 | 'any' ? |
| 14:12 | jnettlet | yeah, because my monitor is 1600x900 |
| 14:12 | mk01 | AH |
| 14:12 | mk01 | yes |
| 14:12 | jnettlet | if I play it on 1080p monitor it is fine |
| 14:13 | mk01 | I have only one odd resolution |
| 14:13 | mk01 | and that's 1366x768 |
| 14:13 | mk01 | initially it wasn't playing at all |
| 14:13 | mk01 | sound ok, but black screen |
| 14:13 | mk01 | had to pad the resolution itself |
| 14:13 | mk01 | (on fb1) |
| 14:14 | mk01 | (and I mean resolution, not only virtual x/y params) |
| 14:15 | jnettlet | this was interlaced or progressive content? |
| 14:16 | mk01 | ANY |
| 16:44 | alex818 | anybody here? |
| 16:45 | alex818 | Wicd network manager fails on "validating authentication" |
| 16:45 | alex818 | trying to connect to wifi |
| 17:12 | jnettlet | use connman |
| 18:19 | vpeter | jnettlet: Could you please, please share your current kernel config for openelec? I accidently delete working one and can't build working kernel anymore :( |