| 07:57 | warped-rudi> | jnettlet[m]: Which version of the Vivante blobs is supposed to be used with your latest 4.9 imx kernel? |
| 07:59 | jnettlet[m]> | warped-rudi: https://www.nxp.com/lgfiles/NMG/MAD/YOCTO/6.2.2.p0-aarch32.bin https://www.nxp.com/lgfiles/NMG/MAD/YOCTO/6.2.2.p0-aarch32.bin |
| 08:00 | jnettlet[m]> | still sorting out the serdev backport for ti bluetooth support. Should be done this AM |
| 08:00 | jnettlet[m]> | warped-rudi: Note I have a global Kconfig option that needs to be set that builds the NXP pieces that conflict with mainline drivers |
| 08:01 | jnettlet[m]> | I also have about a dozen older patches that I am still testing to see if it is worth bringing them forward into the new codebase. |
| 08:02 | jnettlet[m]> | we will be posting Debian Stretch images this weekend as well |
| 08:02 | warped-rudi> | Ok, I'll try that. EGL fails to initialize here using 5.0.11p4... |
| 08:06 | jnettlet[m]> | yes, as is standard Vivante has completely broken API between driver revisions |
| 08:10 | warped-rudi> | Unfortunately, the link you privied seems to be dead |
| 08:10 | warped-rudi> | Server says "'page not found' |
| 08:11 | jnettlet[m]> | oh sorry that is missing the prefix. one second |
| 08:12 | jnettlet[m]> | https://www.nxp.com/lgfiles/NMG/MAD/YOCTO/imx-gpu-viv-6.2.2.p0-aarch32.bin |
| 08:12 | jnettlet[m]> | I still have no idea why NXP hides this stuff away |
| 08:14 | warped-rudi> | Yeah, it was always tricky to locate that stuff. Anyway, downloading it now. Thanks! |
| 08:14 | jnettlet[m]> | here is the g2d library if you want that as well. https://www.nxp.com/lgfiles/NMG/MAD/YOCTO/imx-gpu-g2d-6.2.2.p0.bin |
| 08:18 | warped-rudi> | Yes. ISTR that I had to look at the Yocto recipes in order to find out which files are present on the NXP server. |
| 08:19 | jnettlet[m]> | fyi I have found that the buildroot recipes are more straight forward and easy to piece together |
| 08:44 | jnettlet[m]> | warped-rudi: bluetooth is fixed. pushing the patches. |
| 08:48 | warped-rudi> | Great. I don't have a v1.5 SOM. So even though I included support for it into geexbox, I never tested it. |
| 09:21 | jnettlet[m]> | warped-rudi: well why didn't you ask? |
| 09:28 | warped-rudi> | I think I did ask Rabeeh while back. But maybe he was too busy and I didn't want to appear geedy. |
| 09:29 | warped-rudi> | s/geedy/greedy/ |
| 09:29 | warped-rudi> | BTW, anyone of you guys show up at openEmbeeded in Nuremberg? |
| 09:31 | warped-rudi> | Arrhg, I of course I mean Embedded World. I guess I'll be there on march 1 |
| 09:33 | jnettlet[m]> | yes we have our own booth. I think 5 or 6 of us will be there |
| 09:34 | jnettlet[m]> | definitely come stop by and say hi. I will bring a 1.5 som with me just for you. |
| 09:39 | warped-rudi> | O.K, that a good reason for a visit. Looking forward to meet you! |
| 09:40 | jnettlet[m]> | definitely. Always good to say hi in meat space |
| 09:40 | jnettlet[m]> | warped-rudi: I am going to one last git force push now and then after that I will keep the git tree consistent |
| 09:41 | jnettlet[m]> | topi`: I have backported the rs485 patches in this 4.9 tree if you want to test with that. |
| 10:24 | warped-rudi> | jnettlet[m]: With the 6.2.2.p0 blobs, Kodi seems to work \o/ However, PCM audio produces noise or is distorted. |
| 10:25 | warped-rudi> | DTS and AC3 passthrough works. |
| 10:27 | jnettlet[m]> | warped-rudi: is the default audio just being over-driven? |
| 10:32 | warped-rudi> | No, it sounds like total garbage is sent out. However, even DTS-MA passthrough is O.K. |
| 10:38 | jnettlet[m]> | warped-rudi: what settings 48k? |
| 10:43 | warped-rudi> | My CDs are archived in form of 44.1k FLAC files. |
| 10:47 | jnettlet[m]> | can you test aplay to see if anything 48k works. I don't have analog headphones right now |
| 10:48 | warped-rudi> | I'm talking about HDMI audio. And it looks like it's not related to teh sample rate. |
| 10:51 | jnettlet[m]> | oh, I had tested HDMI audio and it was working |
| 10:51 | jnettlet[m]> | oh you may need a new sdma firmware |
| 10:53 | warped-rudi> | I'm using in-kernel sdma-fw and it reports version 3.3. Did you try 24bit when you tested HDMI out? |
| 10:56 | jnettlet[m]> | yes |
| 11:00 | warped-rudi> | Hmm, could it be that the alignment (i.e. upper vs. lower 24bits in a 32bit word) has changed? |
| 11:02 | jnettlet[m]> | actually just reproduced here also with a different clip. |
| 11:07 | warped-rudi> | It really looks like 16bit samples play O.K. Test two *.wav files using 'aplay'. |
| 11:08 | warped-rudi> | However, I'm pretty sure that Kodi will convert to a format that uses the lower 24 bits of a 32bit entity. |
| 11:09 | warped-rudi> | This is consistent with ALSA, but different from what is used in Windows. |
| 11:10 | jnettlet[m]> | looked over the merge of your patchset and everything looks clean and coherent |
| 11:13 | warped-rudi> | I will see if I can find something out later today or tomorrow. |
| 11:13 | jnettlet[m]> | thanks. let me know. I will also take a more in depth look |
| 15:16 | warped-rudi> | jnettlet[m]: I appears that my suspicion was correct! pcm_lib seems to have switched to MSB alignment for 24bit data. |
| 15:17 | warped-rudi> | If I shift 24bit pcm_data to the right by 8 bits in hdmi_dma_add_frame_info(), I get correct output. |
| 15:18 | warped-rudi> | So there is nothing wrong with our patches. It's an upstream behaviour change. |
| 15:42 | jnettlet[m]> | warped-rudi: simple enough. You want to send a patch? |
| 15:46 | warped-rudi> | Will do so in the evening. Have to leave now. |
| 22:02 | Ke> | on this boot mbin seems stable again |